diff options
Diffstat (limited to 'src')
42 files changed, 13271 insertions, 4781 deletions
diff --git a/src/args_parser.c b/src/args_parser.c new file mode 100644 index 0000000..a678691 --- /dev/null +++ b/src/args_parser.c @@ -0,0 +1,924 @@ +#include "../include/args_parser.h" +#include "../include/defs.h" +#include "../include/egl.h" +#include "../include/window/window.h" + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <inttypes.h> +#include <limits.h> +#include <assert.h> +#include <libgen.h> +#include <sys/stat.h> + +#ifndef GSR_VERSION +#define GSR_VERSION "unknown" +#endif + +static const ArgEnum video_codec_enums[] = { + { .name = "auto", .value = GSR_VIDEO_CODEC_AUTO }, + { .name = "h264", .value = GSR_VIDEO_CODEC_H264 }, + { .name = "h265", .value = GSR_VIDEO_CODEC_HEVC }, + { .name = "hevc", .value = GSR_VIDEO_CODEC_HEVC }, + { .name = "hevc_hdr", .value = GSR_VIDEO_CODEC_HEVC_HDR }, + { .name = "hevc_10bit", .value = GSR_VIDEO_CODEC_HEVC_10BIT }, + { .name = "av1", .value = GSR_VIDEO_CODEC_AV1 }, + { .name = "av1_hdr", .value = GSR_VIDEO_CODEC_AV1_HDR }, + { .name = "av1_10bit", .value = GSR_VIDEO_CODEC_AV1_10BIT }, + { .name = "vp8", .value = GSR_VIDEO_CODEC_VP8 }, + { .name = "vp9", .value = GSR_VIDEO_CODEC_VP9 }, +}; + +static const ArgEnum audio_codec_enums[] = { + { .name = "opus", .value = GSR_AUDIO_CODEC_OPUS }, + { .name = "aac", .value = GSR_AUDIO_CODEC_AAC }, + { .name = "flac", .value = GSR_AUDIO_CODEC_FLAC }, +}; + +static const ArgEnum video_encoder_enums[] = { + { .name = "gpu", .value = GSR_VIDEO_ENCODER_HW_GPU }, + { .name = "cpu", .value = GSR_VIDEO_ENCODER_HW_CPU }, +}; + +static const ArgEnum pixel_format_enums[] = { + { .name = "yuv420", .value = GSR_PIXEL_FORMAT_YUV420 }, + { .name = "yuv444", .value = GSR_PIXEL_FORMAT_YUV444 }, +}; + +static const ArgEnum framerate_mode_enums[] = { + { .name = "vfr", .value = GSR_FRAMERATE_MODE_VARIABLE }, + { .name = "cfr", .value = GSR_FRAMERATE_MODE_CONSTANT }, + { .name = "content", .value = GSR_FRAMERATE_MODE_CONTENT }, +}; + +static const ArgEnum bitrate_mode_enums[] = { + { .name = "auto", .value = GSR_BITRATE_MODE_AUTO }, + { .name = "qp", .value = GSR_BITRATE_MODE_QP }, + { .name = "cbr", .value = GSR_BITRATE_MODE_CBR }, + { .name = "vbr", .value = GSR_BITRATE_MODE_VBR }, +}; + +static const ArgEnum color_range_enums[] = { + { .name = "limited", .value = GSR_COLOR_RANGE_LIMITED }, + { .name = "full", .value = GSR_COLOR_RANGE_FULL }, +}; + +static const ArgEnum tune_enums[] = { + { .name = "performance", .value = GSR_TUNE_PERFORMANCE }, + { .name = "quality", .value = GSR_TUNE_QUALITY }, +}; + +static const ArgEnum replay_storage_enums[] = { + { .name = "ram", .value = GSR_REPLAY_STORAGE_RAM }, + { .name = "disk", .value = GSR_REPLAY_STORAGE_DISK }, +}; + +static void arg_deinit(Arg *arg) { + if(arg->values) { + free(arg->values); + arg->values = NULL; + } +} + +static bool arg_append_value(Arg *arg, const char *value) { + if(arg->num_values + 1 >= arg->capacity_num_values) { + const int new_capacity_num_values = arg->capacity_num_values == 0 ? 4 : arg->capacity_num_values*2; + void *new_data = realloc(arg->values, new_capacity_num_values * sizeof(const char*)); + if(!new_data) + return false; + + arg->values = new_data; + arg->capacity_num_values = new_capacity_num_values; + } + + arg->values[arg->num_values] = value; + ++arg->num_values; + return true; +} + +static bool arg_get_enum_value_by_name(const Arg *arg, const char *name, int *enum_value) { + assert(arg->type == ARG_TYPE_ENUM); + assert(arg->enum_values); + for(int i = 0; i < arg->num_enum_values; ++i) { + if(strcmp(arg->enum_values[i].name, name) == 0) { + *enum_value = arg->enum_values[i].value; + return true; + } + } + return false; +} + +static void arg_print_expected_enum_names(const Arg *arg) { + assert(arg->type == ARG_TYPE_ENUM); + assert(arg->enum_values); + for(int i = 0; i < arg->num_enum_values; ++i) { + if(i > 0) { + if(i == arg->num_enum_values -1) + fprintf(stderr, " or "); + else + fprintf(stderr, ", "); + } + fprintf(stderr, "'%s'", arg->enum_values[i].name); + } +} + +static Arg* args_get_by_key(Arg *args, int num_args, const char *key) { + for(int i = 0; i < num_args; ++i) { + if(strcmp(args[i].key, key) == 0) + return &args[i]; + } + return NULL; +} + +static const char* args_get_value_by_key(Arg *args, int num_args, const char *key) { + for(int i = 0; i < num_args; ++i) { + if(strcmp(args[i].key, key) == 0) { + if(args[i].num_values == 0) + return NULL; + else + return args[i].values[0]; + } + } + return NULL; +} + +static bool args_get_boolean_by_key(Arg *args, int num_args, const char *key, bool default_value) { + Arg *arg = args_get_by_key(args, num_args, key); + assert(arg); + if(arg->num_values == 0) { + return default_value; + } else { + assert(arg->type == ARG_TYPE_BOOLEAN); + return arg->typed_value.boolean; + } +} + +static int args_get_enum_by_key(Arg *args, int num_args, const char *key, int default_value) { + Arg *arg = args_get_by_key(args, num_args, key); + assert(arg); + if(arg->num_values == 0) { + return default_value; + } else { + assert(arg->type == ARG_TYPE_ENUM); + return arg->typed_value.enum_value; + } +} + +static int64_t args_get_i64_by_key(Arg *args, int num_args, const char *key, int64_t default_value) { + Arg *arg = args_get_by_key(args, num_args, key); + assert(arg); + if(arg->num_values == 0) { + return default_value; + } else { + assert(arg->type == ARG_TYPE_I64); + return arg->typed_value.i64_value; + } +} + +static double args_get_double_by_key(Arg *args, int num_args, const char *key, double default_value) { + Arg *arg = args_get_by_key(args, num_args, key); + assert(arg); + if(arg->num_values == 0) { + return default_value; + } else { + assert(arg->type == ARG_TYPE_DOUBLE); + return arg->typed_value.d_value; + } +} + +static void usage_header() { + const bool inside_flatpak = getenv("FLATPAK_ID") != NULL; + const char *program_name = inside_flatpak ? "flatpak run --command=gpu-screen-recorder com.dec05eba.gpu_screen_recorder" : "gpu-screen-recorder"; + printf("usage: %s -w <window_id|monitor|focused|portal|region> [-c <container_format>] [-s WxH] [-region WxH+X+Y] [-f <fps>] [-a <audio_input>] [-q <quality>] [-r <replay_buffer_size_sec>] [-replay-storage ram|disk] [-restart-replay-on-save yes|no] [-k h264|hevc|av1|vp8|vp9|hevc_hdr|av1_hdr|hevc_10bit|av1_10bit] [-ac aac|opus|flac] [-ab <bitrate>] [-oc yes|no] [-fm cfr|vfr|content] [-bm auto|qp|vbr|cbr] [-cr limited|full] [-tune performance|quality] [-df yes|no] [-sc <script_path>] [-cursor yes|no] [-keyint <value>] [-restore-portal-session yes|no] [-portal-session-token-filepath filepath] [-encoder gpu|cpu] [-o <output_file>] [-ro <output_directory>] [--list-capture-options [card_path]] [--list-audio-devices] [--list-application-audio] [-v yes|no] [-gl-debug yes|no] [--version] [-h|--help]\n", program_name); + fflush(stdout); +} + +// TODO: Update with portal info +static void usage_full() { + const bool inside_flatpak = getenv("FLATPAK_ID") != NULL; + const char *program_name = inside_flatpak ? "flatpak run --command=gpu-screen-recorder com.dec05eba.gpu_screen_recorder" : "gpu-screen-recorder"; + usage_header(); + printf("\n"); + printf("OPTIONS:\n"); + printf(" -w Window id to record, a display (monitor name), \"screen\", \"screen-direct\", \"focused\", \"portal\" or \"region\".\n"); + printf(" If this is \"portal\" then xdg desktop screencast portal with PipeWire will be used. Portal option is only available on Wayland.\n"); + printf(" If you select to save the session (token) in the desktop portal capture popup then the session will be saved for the next time you use \"portal\",\n"); + printf(" but the session will be ignored unless you run GPU Screen Recorder with the '-restore-portal-session yes' option.\n"); + printf(" If this is \"region\" then the region specified by the -region option is recorded.\n"); + printf(" If this is \"screen\" then the first monitor found is recorded.\n"); + printf(" \"screen-direct\" can only be used on Nvidia X11, to allow recording without breaking VRR (G-SYNC). This also records all of your monitors.\n"); + printf(" Using this \"screen-direct\" option is not recommended unless you use VRR (G-SYNC) as there are Nvidia driver issues that can cause your system or games to freeze/crash.\n"); + printf(" The \"screen-direct\" option is not needed on AMD, Intel nor Nvidia on Wayland as VRR works properly in those cases.\n"); + printf(" Run GPU Screen Recorder with the --list-capture-options option to list valid values for this option.\n"); + printf("\n"); + printf(" -c Container format for output file, for example mp4, or flv. Only required if no output file is specified or if recording in replay buffer mode.\n"); + printf(" If an output file is specified and -c is not used then the container format is determined from the output filename extension.\n"); + printf(" Only containers that support h264, hevc, av1, vp8 or vp9 are supported, which means that only mp4, mkv, flv, webm (and some others) are supported.\n"); + printf("\n"); + printf(" -s The output resolution limit of the video in the format WxH, for example 1920x1080. If this is 0x0 then the original resolution is used. Optional, except when -w is \"focused\".\n"); + printf(" Note: the captured content is scaled to this size. The output resolution might not be exactly as specified by this option. The original aspect ratio is respected so the resolution will match that.\n"); + printf(" The video encoder might also need to add padding, which will result in black bars on the sides of the video. This is especially an issue on AMD.\n"); + printf("\n"); + printf(" -region\n"); + printf(" The region to capture, only to be used with -w region. This is in format WxH+X+Y, which is compatible with tools such as slop (X11) and slurp (kde plasma, wlroots and hyprland).\n"); + printf(" The region can be inside any monitor. If width and height are 0 (for example 0x0+500+500) then the entire monitor that the region is inside in will be recorded.\n"); + printf(" Note: currently the region can't span multiple monitors.\n"); + printf("\n"); + printf(" -f Frame rate to record at. Recording will only capture frames at this target frame rate.\n"); + printf(" For constant frame rate mode this option is the frame rate every frame will be captured at and if the capture frame rate is below this target frame rate then the frames will be duplicated.\n"); + printf(" For variable frame rate mode this option is the max frame rate and if the capture frame rate is below this target frame rate then frames will not be duplicated.\n"); + printf(" Content frame rate is similar to variable frame rate mode, except the frame rate will match the frame rate of the captured content when possible, but not capturing above the frame rate set in this -f option.\n"); + printf(" Optional, set to 60 by default.\n"); + printf("\n"); + printf(" -a Audio device or application to record from (pulse audio device). Can be specified multiple times. Each time this is specified a new audio track is added for the specified audio device or application.\n"); + printf(" The audio device can also be \"default_output\" in which case the default output device is used, or \"default_input\" in which case the default input device is used.\n"); + printf(" Multiple audio sources can be merged into one audio track by using \"|\" as a separator into one -a argument, for example: -a \"default_output|default_input\".\n"); + printf(" The audio name can also be prefixed with \"device:\", for example: -a \"device:default_output\".\n"); + printf(" To record audio from an application then prefix the audio name with \"app:\", for example: -a \"app:Brave\". The application name is case-insensitive.\n"); + printf(" To record audio from all applications except the provided ones prefix the audio name with \"app-inverse:\", for example: -a \"app-inverse:Brave\".\n"); + printf(" \"app:\" and \"app-inverse:\" can't be mixed in one audio track.\n"); + printf(" One audio track can contain both audio devices and application audio, for example: -a \"default_output|device:alsa_output.pci-0000_00_1b.0.analog-stereo.monitor|app:Brave\".\n"); + printf(" Recording application audio is only possible when the sound server on the system is PipeWire.\n"); + printf(" If the audio name is an empty string then the argument is ignored.\n"); + printf(" Optional, no audio track is added by default.\n"); + printf(" Run GPU Screen Recorder with the --list-audio-devices option to list valid audio device names.\n"); + printf(" Run GPU Screen Recorder with the --list-application-audio option to list valid application names. It's possible to use an application name that is not listed in --list-application-audio,\n"); + printf(" for example when trying to record audio from an application that hasn't started yet.\n"); + printf("\n"); + printf(" -q Video quality. Should be either 'medium', 'high', 'very_high' or 'ultra' when using '-bm qp' or '-bm vbr' options, and '-bm qp' is the default option used.\n"); + printf(" 'high' is the recommended option when live streaming or when you have a slower harddrive.\n"); + printf(" When using '-bm cbr' option then this is option is instead used to specify the video bitrate in kbps.\n"); + printf(" Optional when using '-bm qp' or '-bm vbr' options, set to 'very_high' be default.\n"); + printf(" Required when using '-bm cbr' option.\n"); + printf("\n"); + printf(" -r Replay buffer time in seconds. If this is set, then only the last seconds as set by this option will be stored\n"); + printf(" and the video will only be saved when the gpu-screen-recorder is closed. This feature is similar to Nvidia's instant replay feature This option has be between 5 and 1200.\n"); + printf(" Note that the video data is stored in RAM (unless -replay-storage disk is used), so don't use too long replay buffer time and use constant bitrate option (-bm cbr) to prevent RAM usage from going too high in busy scenes.\n"); + printf(" Optional, disabled by default.\n"); + printf("\n"); + printf(" -replay-storage\n"); + printf(" Specify where temporary replay is stored. Should be either 'ram' or 'disk'. If set to 'disk' then replay data is stored in temporary files in the same directory as -o.\n"); + printf(" Preferably avoid setting this to 'disk' unless -o is set to a HDD, as constant writes to a SSD can reduce the life-time of the SSD.\n"); + printf(" Optional, set to 'ram' by default.\n"); + printf("\n"); + printf(" -restart-replay-on-save\n"); + printf(" Restart replay on save. For example if this is set to 'no' and replay time (-r) is set to 60 seconds and a replay is saved once then the first replay video is 60 seconds long\n"); + printf(" and if a replay is saved 10 seconds later then the second replay video will also be 60 seconds long and contain 50 seconds of the previous video as well.\n"); + printf(" If this is set to 'yes' then after a replay is saved the replay buffer data is cleared and the second replay will start from that point onward.\n"); + printf(" The replay is only restarted when saving a full replay (SIGUSR1 signal)\n"); + printf(" Optional, set to 'no' by default.\n"); + printf("\n"); + printf(" -k Video codec to use. Should be either 'auto', 'h264', 'hevc', 'av1', 'vp8', 'vp9', 'hevc_hdr', 'av1_hdr', 'hevc_10bit' or 'av1_10bit'.\n"); + printf(" Optional, set to 'auto' by default which defaults to 'h264'. Forcefully set to 'h264' if the file container type is 'flv'.\n"); + printf(" 'hevc_hdr' and 'av1_hdr' option is not available on X11 nor when using the portal capture option.\n"); + printf(" 'hevc_10bit' and 'av1_10bit' options allow you to select 10 bit color depth which can reduce banding and improve quality in darker areas, but not all video players support 10 bit color depth\n"); + printf(" and if you upload the video to a website the website might reduce 10 bit to 8 bit.\n"); + printf(" Note that when using 'hevc_hdr' or 'av1_hdr' the color depth is also 10 bits.\n"); + printf("\n"); + printf(" -ac Audio codec to use. Should be either 'aac', 'opus' or 'flac'. Optional, set to 'opus' for .mp4/.mkv files, otherwise set to 'aac'.\n"); + printf(" 'opus' and 'flac' is only supported by .mp4/.mkv files. 'opus' is recommended for best performance and smallest audio size.\n"); + printf(" Flac audio codec is option is disable at the moment because of a temporary issue.\n"); + printf("\n"); + printf(" -ab Audio bitrate in kbps. If this is set to 0 then it's the same as if it's absent, in which case the bitrate is determined automatically depending on the audio codec.\n"); + printf(" Optional, by default the bitrate is 128kbps for opus and flac and 160kbps for aac.\n"); + printf("\n"); + printf(" -oc Overclock memory transfer rate to the maximum performance level. This only applies to NVIDIA on X11 and exists to overcome a bug in NVIDIA driver where performance level\n"); + printf(" is dropped when you record a game. Only needed if you are recording a game that is bottlenecked by GPU. The same issue exists on Wayland but overclocking is not possible on Wayland.\n"); + printf(" Works only if your have \"Coolbits\" set to \"12\" in NVIDIA X settings, see README for more information. Note! use at your own risk! Optional, disabled by default.\n"); + printf("\n"); + printf(" -fm Framerate mode. Should be either 'cfr' (constant frame rate), 'vfr' (variable frame rate) or 'content'. Optional, set to 'vfr' by default.\n"); + printf(" 'vfr' is recommended for recording for less issue with very high system load but some applications such as video editors may not support it properly.\n"); + printf(" 'content' is currently only supported on X11 or when using portal capture option. The 'content' option matches the recording frame rate to the captured content.\n"); + printf("\n"); + printf(" -bm Bitrate mode. Should be either 'auto', 'qp' (constant quality), 'vbr' (variable bitrate) or 'cbr' (constant bitrate). Optional, set to 'auto' by default which defaults to 'qp' on all devices\n"); + printf(" except steam deck that has broken drivers and doesn't support qp.\n"); + printf(" Note: 'vbr' option is not supported when using '-encoder cpu' option.\n"); + printf("\n"); + printf(" -cr Color range. Should be either 'limited' (aka mpeg) or 'full' (aka jpeg). Optional, set to 'limited' by default.\n"); + printf(" Limited color range means that colors are in range 16-235 (4112-60395 for hdr) while full color range means that colors are in range 0-255 (0-65535 for hdr).\n"); + printf(" Note that some buggy video players (such as vlc) are unable to correctly display videos in full color range and when upload the video to websites the website\n"); + printf(" might re-encoder the video to make the video limited color range.\n"); + printf("\n"); + printf(" -tune\n"); + printf(" Tune for performance or quality. Should be either 'performance' or 'quality'. At the moment this option only has an effect on Nvidia where setting this to quality\n"); + printf(" sets options such as preset, multipass and b frames. Optional, set to 'performance' by default.\n"); + printf("\n"); + printf(" -df Organise replays in folders based on the current date.\n"); + printf("\n"); + printf(" -sc Run a script on the saved video file (asynchronously). The first argument to the script is the filepath to the saved video file and the second argument is the recording type (either \"regular\" or \"replay\").\n"); + printf(" Not applicable for live streams.\n"); + printf("\n"); + printf(" -cursor\n"); + printf(" Record cursor. Optional, set to 'yes' by default.\n"); + printf("\n"); + printf(" -keyint\n"); + printf(" Specifies the keyframe interval in seconds, the max amount of time to wait to generate a keyframe. Keyframes can be generated more often than this.\n"); + printf(" This also affects seeking in the video and may affect how the replay video is cut. If this is set to 10 for example then you can only seek in 10-second chunks in the video.\n"); + printf(" Setting this to a higher value reduces the video file size if you are ok with the previously described downside. This option is expected to be a floating point number.\n"); + printf(" By default this value is set to 2.0.\n"); + printf("\n"); + printf(" -restore-portal-session\n"); + printf(" If GPU Screen Recorder should use the same capture option as the last time. Using this option removes the popup asking what you want to record the next time you record with '-w portal'\n"); + printf(" if you selected the option to save session (token) in the desktop portal screencast popup.\n"); + printf(" This option may not have any effect on your Wayland compositor and your systems desktop portal needs to support ScreenCast version 5 or later. Optional, set to 'no' by default.\n"); + printf("\n"); + printf(" -portal-session-token-filepath\n"); + printf(" This option is used together with -restore-portal-session option to specify the file path to save/restore the portal session token to/from.\n"); + printf(" This can be used to remember different portal capture options depending on different recording option (such as recording/replay).\n"); + printf(" Optional, set to \"$XDG_CONFIG_HOME/gpu-screen-recorder/restore_token\" by default ($XDG_CONFIG_HOME defaults to \"$HOME/.config\").\n"); + printf(" Note: the directory to the portal session token file is created automatically if it doesn't exist.\n"); + printf("\n"); + printf(" -encoder\n"); + printf(" Which device should be used for video encoding. Should either be 'gpu' or 'cpu'. 'cpu' option currently only work with h264 codec option (-k).\n"); + printf(" Optional, set to 'gpu' by default.\n"); + printf("\n"); + printf(" --info\n"); + printf(" List info about the system. Lists the following information (prints them to stdout and exits):\n"); + printf(" Supported video codecs (h264, h264_software, hevc, hevc_hdr, hevc_10bit, av1, av1_hdr, av1_10bit, vp8, vp9) and image codecs (jpeg, png) (if supported).\n"); + printf(" Supported capture options (window, focused, screen, monitors and portal, if supported by the system).\n"); + printf(" If opengl initialization fails then the program exits with 22, if no usable drm device is found then it exits with 23. On success it exits with 0.\n"); + printf("\n"); + printf(" --list-capture-options\n"); + printf(" List available capture options. Lists capture options in the following format (prints them to stdout and exits):\n"); + printf(" <option>\n"); + printf(" <monitor_name>|<resolution>\n"); + printf(" For example:\n"); + printf(" window\n"); + printf(" DP-1|1920x1080\n"); + printf(" The <option> and <monitor_name> is the name that can be passed to GPU Screen Recorder with the -w option.\n"); + printf(" --list-capture-options optionally accepts a card path (\"/dev/dri/cardN\") which can improve the performance of running this command.\n"); + printf("\n"); + printf(" --list-audio-devices\n"); + printf(" List audio devices. Lists audio devices in the following format (prints them to stdout and exits):\n"); + printf(" <audio_device_name>|<audio_device_name_in_human_readable_format>\n"); + printf(" For example:\n"); + printf(" bluez_input.88:C9:E8:66:A2:27|WH-1000XM4\n"); + printf(" alsa_output.pci-0000_0c_00.4.iec958-stereo|Monitor of Starship/Matisse HD Audio Controller Digital Stereo (IEC958)\n"); + printf(" The <audio_device_name> is the name that can be passed to GPU Screen Recorder with the -a option.\n"); + printf("\n"); + printf(" --list-application-audio\n"); + printf(" Lists applications that you can record from (prints them to stdout and exits), for example:\n"); + printf(" firefox\n"); + printf(" csgo\n"); + printf(" These names are the application audio names that can be passed to GPU Screen Recorder with the -a option.\n"); + printf("\n"); + printf(" --version\n"); + printf(" Print version (%s) and exit\n", GSR_VERSION); + printf("\n"); + //fprintf(stderr, " -pixfmt The pixel format to use for the output video. yuv420 is the most common format and is best supported, but the color is compressed, so colors can look washed out and certain colors of text can look bad. Use yuv444 for no color compression, but the video may not work everywhere and it may not work with hardware video decoding. Optional, set to 'yuv420' by default\n"); + printf(" -o The output file path. If omitted then the encoded data is sent to stdout. Required in replay mode (when using -r).\n"); + printf(" In replay mode this has to be a directory instead of a file.\n"); + printf(" Note: the directory to the file is created automatically if it doesn't already exist.\n"); + printf("\n"); + printf(" -ro The output directory for regular recordings in replay/streaming mode. Required to start recording in replay/streaming mode.\n"); + printf(" Note: the directory to the file is created automatically if it doesn't already exist.\n"); + printf("\n"); + printf(" -v Prints fps and damage info once per second. Optional, set to 'yes' by default.\n"); + printf("\n"); + printf(" -gl-debug\n"); + printf(" Print opengl debug output. Optional, set to 'no' by default.\n"); + printf("\n"); + printf(" -h, --help\n"); + printf(" Show this help.\n"); + printf("\n"); + printf("NOTES:\n"); + printf(" Send signal SIGINT to gpu-screen-recorder (Ctrl+C, or pkill -SIGINT -f gpu-screen-recorder) to stop and save the recording. When in replay mode this stops recording without saving.\n"); + printf(" Send signal SIGUSR2 to gpu-screen-recorder (pkill -SIGUSR2 -f gpu-screen-recorder) to pause/unpause recording. Only applicable when recording (not streaming nor replay).\n"); + printf(" Send signal SIGUSR1 to gpu-screen-recorder (pkill -SIGUSR1 -f gpu-screen-recorder) to save a replay (when in replay mode).\n"); + printf(" Send signal SIGRTMIN+1 to gpu-screen-recorder (pkill -SIGRTMIN+1 -f gpu-screen-recorder) to save a replay of the last 10 seconds (when in replay mode).\n"); + printf(" Send signal SIGRTMIN+2 to gpu-screen-recorder (pkill -SIGRTMIN+2 -f gpu-screen-recorder) to save a replay of the last 30 seconds (when in replay mode).\n"); + printf(" Send signal SIGRTMIN+3 to gpu-screen-recorder (pkill -SIGRTMIN+3 -f gpu-screen-recorder) to save a replay of the last 60 seconds (when in replay mode).\n"); + printf(" Send signal SIGRTMIN+4 to gpu-screen-recorder (pkill -SIGRTMIN+4 -f gpu-screen-recorder) to save a replay of the last 5 minutes (when in replay mode).\n"); + printf(" Send signal SIGRTMIN+5 to gpu-screen-recorder (pkill -SIGRTMIN+5 -f gpu-screen-recorder) to save a replay of the last 10 minutes (when in replay mode).\n"); + printf(" Send signal SIGRTMIN+6 to gpu-screen-recorder (pkill -SIGRTMIN+6 -f gpu-screen-recorder) to save a replay of the last 30 minutes (when in replay mode).\n"); + printf(" Send signal SIGRTMIN to gpu-screen-recorder (pkill -SIGRTMIN -f gpu-screen-recorder) to start/stop recording a regular video when in replay/streaming mode.\n"); + printf("\n"); + printf("EXAMPLES:\n"); + printf(" %s -w screen -f 60 -a default_output -o video.mp4\n", program_name); + printf(" %s -w screen -f 60 -a default_output -a default_input -o video.mp4\n", program_name); + printf(" %s -w screen -f 60 -a \"default_output|default_input\" -o video.mp4\n", program_name); + printf(" %s -w screen -f 60 -a default_output -c mkv -r 60 -o \"$HOME/Videos\"\n", program_name); + printf(" %s -w screen -f 60 -a default_output -c mkv -r 1800 -replay-storage disk -bm cbr -q 40000 -o \"$HOME/Videos\"\n", program_name); + printf(" %s -w screen -f 60 -a default_output -c mkv -sc script.sh -r 60 -o \"$HOME/Videos\"\n", program_name); + printf(" %s -w portal -f 60 -a default_output -restore-portal-session yes -o video.mp4\n", program_name); + printf(" %s -w screen -f 60 -a default_output -bm cbr -q 15000 -o video.mp4\n", program_name); + printf(" %s -w screen -f 60 -a \"app:firefox|app:csgo\" -o video.mp4\n", program_name); + printf(" %s -w screen -f 60 -a \"app-inverse:firefox|app-inverse:csgo\" -o video.mp4\n", program_name); + printf(" %s -w screen -f 60 -a \"default_input|app-inverse:Brave\" -o video.mp4\n", program_name); + printf(" %s -w screen -o image.jpg\n", program_name); + printf(" %s -w screen -q medium -o image.jpg\n", program_name); + printf(" %s -w region -region 640x480+100+100 -o video.mp4\n", program_name); + printf(" %s -w region -region $(slop) -o video.mp4\n", program_name); + printf(" %s -w region -region $(slurp -f \"%%wx%%h+%%x+%%y\") -o video.mp4\n", program_name); + //fprintf(stderr, " gpu-screen-recorder -w screen -f 60 -q ultra -pixfmt yuv444 -o video.mp4\n"); + fflush(stdout); +} + +static void usage() { + usage_header(); +} + +// TODO: Does this match all livestreaming cases? +static bool is_livestream_path(const char *str) { + const int len = strlen(str); + if((len >= 7 && memcmp(str, "http://", 7) == 0) || (len >= 8 && memcmp(str, "https://", 8) == 0)) + return true; + else if((len >= 7 && memcmp(str, "rtmp://", 7) == 0) || (len >= 8 && memcmp(str, "rtmps://", 8) == 0)) + return true; + else if((len >= 7 && memcmp(str, "rtsp://", 7) == 0)) + return true; + else if((len >= 6 && memcmp(str, "srt://", 6) == 0)) + return true; + else if((len >= 6 && memcmp(str, "tcp://", 6) == 0)) + return true; + else if((len >= 6 && memcmp(str, "udp://", 6) == 0)) + return true; + else + return false; +} + +static bool args_parser_set_values(args_parser *self) { + self->video_encoder = (gsr_video_encoder_hardware)args_get_enum_by_key(self->args, NUM_ARGS, "-encoder", GSR_VIDEO_ENCODER_HW_GPU); + self->pixel_format = (gsr_pixel_format)args_get_enum_by_key(self->args, NUM_ARGS, "-pixfmt", GSR_PIXEL_FORMAT_YUV420); + self->framerate_mode = (gsr_framerate_mode)args_get_enum_by_key(self->args, NUM_ARGS, "-fm", GSR_FRAMERATE_MODE_VARIABLE); + self->color_range = (gsr_color_range)args_get_enum_by_key(self->args, NUM_ARGS, "-cr", GSR_COLOR_RANGE_LIMITED); + self->tune = (gsr_tune)args_get_enum_by_key(self->args, NUM_ARGS, "-tune", GSR_TUNE_PERFORMANCE); + self->video_codec = (gsr_video_codec)args_get_enum_by_key(self->args, NUM_ARGS, "-k", GSR_VIDEO_CODEC_AUTO); + self->audio_codec = (gsr_audio_codec)args_get_enum_by_key(self->args, NUM_ARGS, "-ac", GSR_AUDIO_CODEC_OPUS); + self->bitrate_mode = (gsr_bitrate_mode)args_get_enum_by_key(self->args, NUM_ARGS, "-bm", GSR_BITRATE_MODE_AUTO); + self->replay_storage = (gsr_replay_storage)args_get_enum_by_key(self->args, NUM_ARGS, "-replay-storage", GSR_REPLAY_STORAGE_RAM); + + const char *window = args_get_value_by_key(self->args, NUM_ARGS, "-w"); + snprintf(self->window, sizeof(self->window), "%s", window); + self->verbose = args_get_boolean_by_key(self->args, NUM_ARGS, "-v", true); + self->gl_debug = args_get_boolean_by_key(self->args, NUM_ARGS, "-gl-debug", false); + self->record_cursor = args_get_boolean_by_key(self->args, NUM_ARGS, "-cursor", true); + self->date_folders = args_get_boolean_by_key(self->args, NUM_ARGS, "-df", false); + self->restore_portal_session = args_get_boolean_by_key(self->args, NUM_ARGS, "-restore-portal-session", false); + self->restart_replay_on_save = args_get_boolean_by_key(self->args, NUM_ARGS, "-restart-replay-on-save", false); + self->overclock = args_get_boolean_by_key(self->args, NUM_ARGS, "-oc", false); + + self->audio_bitrate = args_get_i64_by_key(self->args, NUM_ARGS, "-ab", 0); + self->audio_bitrate *= 1000LL; + + self->keyint = args_get_double_by_key(self->args, NUM_ARGS, "-keyint", 2.0); + + if(self->audio_codec == GSR_AUDIO_CODEC_FLAC) { + fprintf(stderr, "Warning: flac audio codec is temporary disabled, using opus audio codec instead\n"); + self->audio_codec = GSR_AUDIO_CODEC_OPUS; + } + + self->portal_session_token_filepath = args_get_value_by_key(self->args, NUM_ARGS, "-portal-session-token-filepath"); + if(self->portal_session_token_filepath) { + int len = strlen(self->portal_session_token_filepath); + if(len > 0 && self->portal_session_token_filepath[len - 1] == '/') { + fprintf(stderr, "Error: -portal-session-token-filepath should be a path to a file but it ends with a /: %s\n", self->portal_session_token_filepath); + return false; + } + } + + self->recording_saved_script = args_get_value_by_key(self->args, NUM_ARGS, "-sc"); + if(self->recording_saved_script) { + struct stat buf; + if(stat(self->recording_saved_script, &buf) == -1 || !S_ISREG(buf.st_mode)) { + fprintf(stderr, "Error: Script \"%s\" either doesn't exist or it's not a file\n", self->recording_saved_script); + usage(); + return false; + } + + if(!(buf.st_mode & S_IXUSR)) { + fprintf(stderr, "Error: Script \"%s\" is not executable\n", self->recording_saved_script); + usage(); + return false; + } + } + + const char *quality_str = args_get_value_by_key(self->args, NUM_ARGS, "-q"); + self->video_quality = GSR_VIDEO_QUALITY_VERY_HIGH; + self->video_bitrate = 0; + + if(self->bitrate_mode == GSR_BITRATE_MODE_CBR) { + if(!quality_str) { + fprintf(stderr, "Error: option '-q' is required when using '-bm cbr' option\n"); + usage(); + return false; + } + + if(sscanf(quality_str, "%" PRIi64, &self->video_bitrate) != 1) { + fprintf(stderr, "Error: -q argument \"%s\" is not an integer value. When using '-bm cbr' option '-q' is expected to be an integer value\n", quality_str); + usage(); + return false; + } + + if(self->video_bitrate < 0) { + fprintf(stderr, "Error: -q is expected to be 0 or larger, got %" PRIi64 "\n", self->video_bitrate); + usage(); + return false; + } + + self->video_bitrate *= 1000LL; + } else { + if(!quality_str) + quality_str = "very_high"; + + if(strcmp(quality_str, "medium") == 0) { + self->video_quality = GSR_VIDEO_QUALITY_MEDIUM; + } else if(strcmp(quality_str, "high") == 0) { + self->video_quality = GSR_VIDEO_QUALITY_HIGH; + } else if(strcmp(quality_str, "very_high") == 0) { + self->video_quality = GSR_VIDEO_QUALITY_VERY_HIGH; + } else if(strcmp(quality_str, "ultra") == 0) { + self->video_quality = GSR_VIDEO_QUALITY_ULTRA; + } else { + fprintf(stderr, "Error: -q should either be 'medium', 'high', 'very_high' or 'ultra', got: '%s'\n", quality_str); + usage(); + return false; + } + } + + const char *output_resolution_str = args_get_value_by_key(self->args, NUM_ARGS, "-s"); + if(!output_resolution_str && strcmp(self->window, "focused") == 0) { + fprintf(stderr, "Error: option -s is required when using '-w focused' option\n"); + usage(); + return false; + } + + self->output_resolution = (vec2i){0, 0}; + if(output_resolution_str) { + if(sscanf(output_resolution_str, "%dx%d", &self->output_resolution.x, &self->output_resolution.y) != 2) { + fprintf(stderr, "Error: invalid value for option -s '%s', expected a value in format WxH\n", output_resolution_str); + usage(); + return false; + } + + if(self->output_resolution.x < 0 || self->output_resolution.y < 0) { + fprintf(stderr, "Error: invalid value for option -s '%s', expected width and height to be greater or equal to 0\n", output_resolution_str); + usage(); + return false; + } + } + + self->region_size = (vec2i){0, 0}; + self->region_position = (vec2i){0, 0}; + const char *region_str = args_get_value_by_key(self->args, NUM_ARGS, "-region"); + if(region_str) { + if(strcmp(self->window, "region") != 0) { + fprintf(stderr, "Error: option -region can only be used when option '-w region' is used\n"); + usage(); + return false; + } + + if(sscanf(region_str, "%dx%d+%d+%d", &self->region_size.x, &self->region_size.y, &self->region_position.x, &self->region_position.y) != 4) { + fprintf(stderr, "Error: invalid value for option -region '%s', expected a value in format WxH+X+Y\n", region_str); + usage(); + return false; + } + + if(self->region_size.x < 0 || self->region_size.y < 0 || self->region_position.x < 0 || self->region_position.y < 0) { + fprintf(stderr, "Error: invalid value for option -region '%s', expected width, height, x and y to be greater or equal to 0\n", region_str); + usage(); + return false; + } + } else { + if(strcmp(self->window, "region") == 0) { + fprintf(stderr, "Error: option -region is required when '-w region' is used\n"); + usage(); + return false; + } + } + + self->fps = args_get_i64_by_key(self->args, NUM_ARGS, "-f", 60); + self->replay_buffer_size_secs = args_get_i64_by_key(self->args, NUM_ARGS, "-r", -1); + if(self->replay_buffer_size_secs != -1) + self->replay_buffer_size_secs += (int64_t)(self->keyint + 0.5); // Add a few seconds to account of lost packets because of non-keyframe packets skipped + + self->container_format = args_get_value_by_key(self->args, NUM_ARGS, "-c"); + if(self->container_format && strcmp(self->container_format, "mkv") == 0) + self->container_format = "matroska"; + + const bool is_replaying = self->replay_buffer_size_secs != -1; + self->is_livestream = false; + self->filename = args_get_value_by_key(self->args, NUM_ARGS, "-o"); + if(self->filename) { + self->is_livestream = is_livestream_path(self->filename); + if(self->is_livestream) { + if(is_replaying) { + fprintf(stderr, "Error: replay mode is not applicable to live streaming\n"); + return false; + } + } else { + if(!is_replaying) { + char directory_buf[PATH_MAX]; + snprintf(directory_buf, sizeof(directory_buf), "%s", self->filename); + char *directory = dirname(directory_buf); + if(strcmp(directory, ".") != 0 && strcmp(directory, "/") != 0) { + if(create_directory_recursive(directory) != 0) { + fprintf(stderr, "Error: failed to create directory for output file: %s\n", self->filename); + return false; + } + } + } else { + if(!self->container_format) { + fprintf(stderr, "Error: option -c is required when using option -r\n"); + usage(); + return false; + } + + struct stat buf; + if(stat(self->filename, &buf) != -1 && !S_ISDIR(buf.st_mode)) { + fprintf(stderr, "Error: File \"%s\" exists but it's not a directory\n", self->filename); + usage(); + return false; + } + } + } + } else { + if(!is_replaying) { + self->filename = "/dev/stdout"; + } else { + fprintf(stderr, "Error: Option -o is required when using option -r\n"); + usage(); + return false; + } + + if(!self->container_format) { + fprintf(stderr, "Error: option -c is required when not using option -o\n"); + usage(); + return false; + } + } + + self->is_output_piped = strcmp(self->filename, "/dev/stdout") == 0; + self->low_latency_recording = self->is_livestream || self->is_output_piped; + + self->replay_recording_directory = args_get_value_by_key(self->args, NUM_ARGS, "-ro"); + + const bool is_portal_capture = strcmp(self->window, "portal") == 0; + if(!self->restore_portal_session && is_portal_capture) + fprintf(stderr, "Info: option '-w portal' was used without '-restore-portal-session yes'. The previous screencast session will be ignored\n"); + + if(self->is_livestream && self->recording_saved_script) { + fprintf(stderr, "Warning: live stream detected, -sc script is ignored\n"); + self->recording_saved_script = NULL; + } + + return true; +} + +bool args_parser_parse(args_parser *self, int argc, char **argv, const args_handlers *arg_handlers, void *userdata) { + assert(arg_handlers); + memset(self, 0, sizeof(*self)); + + if(argc <= 1) { + usage_full(); + return false; + } + + if(argc == 2 && (strcmp(argv[1], "-h") == 0 || strcmp(argv[1], "--help") == 0)) { + usage_full(); + return false; + } + + if(argc == 2 && strcmp(argv[1], "--info") == 0) { + arg_handlers->info(userdata); + return true; + } + + if(argc == 2 && strcmp(argv[1], "--list-audio-devices") == 0) { + arg_handlers->list_audio_devices(userdata); + return true; + } + + if(argc == 2 && strcmp(argv[1], "--list-application-audio") == 0) { + arg_handlers->list_application_audio(userdata); + return true; + } + + if(strcmp(argv[1], "--list-capture-options") == 0) { + if(argc == 2) { + arg_handlers->list_capture_options(NULL, userdata); + return true; + } else if(argc == 3 || argc == 4) { + const char *card_path = argv[2]; + arg_handlers->list_capture_options(card_path, userdata); + return true; + } else { + fprintf(stderr, "Error: expected --list-capture-options to be called with either no extra arguments or 1 extra argument (card path)\n"); + return false; + } + } + + if(argc == 2 && strcmp(argv[1], "--version") == 0) { + arg_handlers->version(userdata); + return true; + } + + int arg_index = 0; + self->args[arg_index++] = (Arg){ .key = "-w", .optional = false, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-c", .optional = true, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-f", .optional = true, .list = false, .type = ARG_TYPE_I64, .integer_value_min = 1, .integer_value_max = 1000 }; + self->args[arg_index++] = (Arg){ .key = "-s", .optional = true, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-region", .optional = true, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-a", .optional = true, .list = true, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-q", .optional = true, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-o", .optional = true, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-ro", .optional = true, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-r", .optional = true, .list = false, .type = ARG_TYPE_I64, .integer_value_min = 2, .integer_value_max = 86400 }; + self->args[arg_index++] = (Arg){ .key = "-restart-replay-on-save", .optional = true, .list = false, .type = ARG_TYPE_BOOLEAN }; + self->args[arg_index++] = (Arg){ .key = "-k", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = video_codec_enums, .num_enum_values = sizeof(video_codec_enums)/sizeof(ArgEnum) }; + self->args[arg_index++] = (Arg){ .key = "-ac", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = audio_codec_enums, .num_enum_values = sizeof(audio_codec_enums)/sizeof(ArgEnum) }; + self->args[arg_index++] = (Arg){ .key = "-ab", .optional = true, .list = false, .type = ARG_TYPE_I64, .integer_value_min = 0, .integer_value_max = 50000 }; + self->args[arg_index++] = (Arg){ .key = "-oc", .optional = true, .list = false, .type = ARG_TYPE_BOOLEAN }; + self->args[arg_index++] = (Arg){ .key = "-fm", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = framerate_mode_enums, .num_enum_values = sizeof(framerate_mode_enums)/sizeof(ArgEnum) }; + self->args[arg_index++] = (Arg){ .key = "-bm", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = bitrate_mode_enums, .num_enum_values = sizeof(bitrate_mode_enums)/sizeof(ArgEnum) }; + self->args[arg_index++] = (Arg){ .key = "-pixfmt", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = pixel_format_enums, .num_enum_values = sizeof(pixel_format_enums)/sizeof(ArgEnum) }; + self->args[arg_index++] = (Arg){ .key = "-v", .optional = true, .list = false, .type = ARG_TYPE_BOOLEAN }; + self->args[arg_index++] = (Arg){ .key = "-gl-debug", .optional = true, .list = false, .type = ARG_TYPE_BOOLEAN }; + self->args[arg_index++] = (Arg){ .key = "-df", .optional = true, .list = false, .type = ARG_TYPE_BOOLEAN }; + self->args[arg_index++] = (Arg){ .key = "-sc", .optional = true, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-cr", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = color_range_enums, .num_enum_values = sizeof(color_range_enums)/sizeof(ArgEnum) }; + self->args[arg_index++] = (Arg){ .key = "-tune", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = tune_enums, .num_enum_values = sizeof(tune_enums)/sizeof(ArgEnum) }; + self->args[arg_index++] = (Arg){ .key = "-cursor", .optional = true, .list = false, .type = ARG_TYPE_BOOLEAN }; + self->args[arg_index++] = (Arg){ .key = "-keyint", .optional = true, .list = false, .type = ARG_TYPE_DOUBLE, .integer_value_min = 0, .integer_value_max = 500 }; + self->args[arg_index++] = (Arg){ .key = "-restore-portal-session", .optional = true, .list = false, .type = ARG_TYPE_BOOLEAN }; + self->args[arg_index++] = (Arg){ .key = "-portal-session-token-filepath", .optional = true, .list = false, .type = ARG_TYPE_STRING }; + self->args[arg_index++] = (Arg){ .key = "-encoder", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = video_encoder_enums, .num_enum_values = sizeof(video_encoder_enums)/sizeof(ArgEnum) }; + self->args[arg_index++] = (Arg){ .key = "-replay-storage", .optional = true, .list = false, .type = ARG_TYPE_ENUM, .enum_values = replay_storage_enums, .num_enum_values = sizeof(replay_storage_enums)/sizeof(ArgEnum) }; + assert(arg_index == NUM_ARGS); + + for(int i = 1; i < argc; i += 2) { + const char *arg_name = argv[i]; + Arg *arg = args_get_by_key(self->args, NUM_ARGS, arg_name); + if(!arg) { + fprintf(stderr, "Error: invalid argument '%s'\n", arg_name); + usage(); + return false; + } + + if(arg->num_values > 0 && !arg->list) { + fprintf(stderr, "Error: expected argument '%s' to only be specified once\n", arg_name); + usage(); + return false; + } + + if(i + 1 >= argc) { + fprintf(stderr, "Error: missing value for argument '%s'\n", arg_name); + usage(); + return false; + } + + const char *arg_value = argv[i + 1]; + switch(arg->type) { + case ARG_TYPE_STRING: { + break; + } + case ARG_TYPE_BOOLEAN: { + if(strcmp(arg_value, "yes") == 0) { + arg->typed_value.boolean = true; + } else if(strcmp(arg_value, "no") == 0) { + arg->typed_value.boolean = false; + } else { + fprintf(stderr, "Error: %s should either be 'yes' or 'no', got: '%s'\n", arg_name, arg_value); + usage(); + return false; + } + break; + } + case ARG_TYPE_ENUM: { + if(!arg_get_enum_value_by_name(arg, arg_value, &arg->typed_value.enum_value)) { + fprintf(stderr, "Error: %s should either be ", arg_name); + arg_print_expected_enum_names(arg); + fprintf(stderr, ", got: '%s'\n", arg_value); + usage(); + return false; + } + break; + } + case ARG_TYPE_I64: { + if(sscanf(arg_value, "%" PRIi64, &arg->typed_value.i64_value) != 1) { + fprintf(stderr, "Error: %s argument \"%s\" is not an integer\n", arg_name, arg_value); + usage(); + return false; + } + + if(arg->typed_value.i64_value < arg->integer_value_min) { + fprintf(stderr, "Error: %s argument is expected to be larger than %" PRIi64 ", got %" PRIi64 "\n", arg_name, arg->integer_value_min, arg->typed_value.i64_value); + usage(); + return false; + } + + if(arg->typed_value.i64_value > arg->integer_value_max) { + fprintf(stderr, "Error: %s argument is expected to be less than %" PRIi64 ", got %" PRIi64 "\n", arg_name, arg->integer_value_max, arg->typed_value.i64_value); + usage(); + return false; + } + break; + } + case ARG_TYPE_DOUBLE: { + if(sscanf(arg_value, "%lf", &arg->typed_value.d_value) != 1) { + fprintf(stderr, "Error: %s argument \"%s\" is not an floating-point number\n", arg_name, arg_value); + usage(); + return false; + } + + if(arg->typed_value.d_value < arg->integer_value_min) { + fprintf(stderr, "Error: %s argument is expected to be larger than %" PRIi64 ", got %lf\n", arg_name, arg->integer_value_min, arg->typed_value.d_value); + usage(); + return false; + } + + if(arg->typed_value.d_value > arg->integer_value_max) { + fprintf(stderr, "Error: %s argument is expected to be less than %" PRIi64 ", got %lf\n", arg_name, arg->integer_value_max, arg->typed_value.d_value); + usage(); + return false; + } + break; + } + } + + if(!arg_append_value(arg, arg_value)) { + fprintf(stderr, "Error: failed to append argument, out of memory\n"); + return false; + } + } + + for(int i = 0; i < NUM_ARGS; ++i) { + const Arg *arg = &self->args[i]; + if(!arg->optional && arg->num_values == 0) { + fprintf(stderr, "Error: missing argument '%s'\n", arg->key); + usage(); + return false; + } + } + + return args_parser_set_values(self); +} + +void args_parser_deinit(args_parser *self) { + for(int i = 0; i < NUM_ARGS; ++i) { + arg_deinit(&self->args[i]); + } +} + +bool args_parser_validate_with_gl_info(args_parser *self, gsr_egl *egl) { + const bool wayland = gsr_window_get_display_server(egl->window) == GSR_DISPLAY_SERVER_WAYLAND; + + if(self->bitrate_mode == (gsr_bitrate_mode)GSR_BITRATE_MODE_AUTO) { + // QP is broken on steam deck, see https://github.com/ValveSoftware/SteamOS/issues/1609 + self->bitrate_mode = egl->gpu_info.is_steam_deck ? GSR_BITRATE_MODE_VBR : GSR_BITRATE_MODE_QP; + } + + if(egl->gpu_info.is_steam_deck && self->bitrate_mode == GSR_BITRATE_MODE_QP) { + fprintf(stderr, "Warning: qp bitrate mode is not supported on Steam Deck because of Steam Deck driver bugs. Using vbr instead\n"); + self->bitrate_mode = GSR_BITRATE_MODE_VBR; + } + + if(self->video_encoder == GSR_VIDEO_ENCODER_HW_CPU && self->bitrate_mode == GSR_BITRATE_MODE_VBR) { + fprintf(stderr, "Warning: bitrate mode has been forcefully set to qp because software encoding option doesn't support vbr option\n"); + self->bitrate_mode = GSR_BITRATE_MODE_QP; + } + + if(egl->gpu_info.vendor != GSR_GPU_VENDOR_NVIDIA && self->overclock) { + fprintf(stderr, "Info: overclock option has no effect on amd/intel, ignoring option\n"); + self->overclock = false; + } + + if(egl->gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA && self->overclock && wayland) { + fprintf(stderr, "Info: overclocking is not possible on nvidia on wayland, ignoring option\n"); + self->overclock = false; + } + + if(egl->gpu_info.is_steam_deck) { + fprintf(stderr, "Warning: steam deck has multiple driver issues. One of them has been reported here: https://github.com/ValveSoftware/SteamOS/issues/1609\n" + "If you have issues with GPU Screen Recorder on steam deck that you don't have on a desktop computer then report the issue to Valve and/or AMD.\n"); + } + + self->very_old_gpu = false; + if(egl->gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA && egl->gpu_info.gpu_version != 0 && egl->gpu_info.gpu_version < 900) { + fprintf(stderr, "Info: your gpu appears to be very old (older than maxwell architecture). Switching to lower preset\n"); + self->very_old_gpu = true; + } + + if(video_codec_is_hdr(self->video_codec) && !wayland) { + fprintf(stderr, "Error: hdr video codec option %s is not available on X11\n", video_codec_to_string(self->video_codec)); + usage(); + return false; + } + + const bool is_portal_capture = strcmp(self->window, "portal") == 0; + if(video_codec_is_hdr(self->video_codec) && is_portal_capture) { + fprintf(stderr, "Warning: portal capture option doesn't support hdr yet (PipeWire doesn't support hdr), the video will be tonemapped from hdr to sdr\n"); + self->video_codec = hdr_video_codec_to_sdr_video_codec(self->video_codec); + } + + return true; +} + +void args_parser_print_usage(void) { + usage(); +} + +Arg* args_parser_get_arg(args_parser *self, const char *arg_name) { + return args_get_by_key(self->args, NUM_ARGS, arg_name); +} diff --git a/src/capture/capture.c b/src/capture/capture.c index eea0d1d..bc95300 100644 --- a/src/capture/capture.c +++ b/src/capture/capture.c @@ -1,59 +1,53 @@ #include "../../include/capture/capture.h" -#include <stdio.h> +#include <assert.h> -int gsr_capture_start(gsr_capture *cap, AVCodecContext *video_codec_context) { - if(cap->started) - return -1; - - int res = cap->start(cap, video_codec_context); +int gsr_capture_start(gsr_capture *cap, gsr_capture_metadata *capture_metadata) { + assert(!cap->started); + int res = cap->start(cap, capture_metadata); if(res == 0) cap->started = true; return res; } -void gsr_capture_tick(gsr_capture *cap, AVCodecContext *video_codec_context, AVFrame **frame) { - if(!cap->started) { - fprintf(stderr, "gsr error: gsp_capture_tick failed: the gsr capture has not been started\n"); - return; - } - +void gsr_capture_tick(gsr_capture *cap) { + assert(cap->started); if(cap->tick) - cap->tick(cap, video_codec_context, frame); + cap->tick(cap); } -bool gsr_capture_should_stop(gsr_capture *cap, bool *err) { - if(!cap->started) { - fprintf(stderr, "gsr error: gsr_capture_should_stop failed: the gsr capture has not been started\n"); - return false; - } +void gsr_capture_on_event(gsr_capture *cap, gsr_egl *egl) { + if(cap->on_event) + cap->on_event(cap, egl); +} - if(!cap->should_stop) +bool gsr_capture_should_stop(gsr_capture *cap, bool *err) { + assert(cap->started); + if(cap->should_stop) + return cap->should_stop(cap, err); + else return false; - - return cap->should_stop(cap, err); } -int gsr_capture_capture(gsr_capture *cap, AVFrame *frame) { - if(!cap->started) { - fprintf(stderr, "gsr error: gsr_capture_capture failed: the gsr capture has not been started\n"); - return -1; - } - return cap->capture(cap, frame); +int gsr_capture_capture(gsr_capture *cap, gsr_capture_metadata *capture_metadata, gsr_color_conversion *color_conversion) { + assert(cap->started); + return cap->capture(cap, capture_metadata, color_conversion); } -void gsr_capture_end(gsr_capture *cap, AVFrame *frame) { - if(!cap->started) { - fprintf(stderr, "gsr error: gsr_capture_end failed: the gsr capture has not been started\n"); - return; - } - - if(!cap->capture_end) - return; +bool gsr_capture_uses_external_image(gsr_capture *cap) { + if(cap->uses_external_image) + return cap->uses_external_image(cap); + else + return false; +} - cap->capture_end(cap, frame); +bool gsr_capture_set_hdr_metadata(gsr_capture *cap, AVMasteringDisplayMetadata *mastering_display_metadata, AVContentLightMetadata *light_metadata) { + if(cap->set_hdr_metadata) + return cap->set_hdr_metadata(cap, mastering_display_metadata, light_metadata); + else + return false; } -void gsr_capture_destroy(gsr_capture *cap, AVCodecContext *video_codec_context) { - cap->destroy(cap, video_codec_context); +void gsr_capture_destroy(gsr_capture *cap) { + cap->destroy(cap); } diff --git a/src/capture/kms.c b/src/capture/kms.c new file mode 100644 index 0000000..18858f2 --- /dev/null +++ b/src/capture/kms.c @@ -0,0 +1,778 @@ +#include "../../include/capture/kms.h" +#include "../../include/utils.h" +#include "../../include/color_conversion.h" +#include "../../include/cursor.h" +#include "../../include/window/window.h" +#include "../../kms/client/kms_client.h" + +#include <stdlib.h> +#include <string.h> +#include <stdio.h> +#include <unistd.h> +#include <fcntl.h> + +#include <xf86drm.h> +#include <libdrm/drm_fourcc.h> + +#include <libavutil/mastering_display_metadata.h> + +#define FIND_CRTC_BY_NAME_TIMEOUT_SECONDS 2.0 + +#define HDMI_STATIC_METADATA_TYPE1 0 +#define HDMI_EOTF_SMPTE_ST2084 2 + +#define MAX_CONNECTOR_IDS 32 + +typedef struct { + uint32_t connector_ids[MAX_CONNECTOR_IDS]; + int num_connector_ids; +} MonitorId; + +typedef struct { + gsr_capture_kms_params params; + + gsr_kms_client kms_client; + gsr_kms_response kms_response; + + vec2i capture_pos; + vec2i capture_size; + MonitorId monitor_id; + + gsr_monitor_rotation monitor_rotation; + + unsigned int input_texture_id; + unsigned int external_input_texture_id; + unsigned int cursor_texture_id; + + bool no_modifiers_fallback; + bool external_texture_fallback; + + struct hdr_output_metadata hdr_metadata; + bool hdr_metadata_set; + + bool is_x11; + gsr_cursor x11_cursor; + + //int drm_fd; + //uint64_t prev_sequence; + //bool damaged; + + vec2i prev_target_pos; + vec2i prev_plane_size; + + double last_time_monitor_check; +} gsr_capture_kms; + +static void gsr_capture_kms_cleanup_kms_fds(gsr_capture_kms *self) { + for(int i = 0; i < self->kms_response.num_items; ++i) { + for(int j = 0; j < self->kms_response.items[i].num_dma_bufs; ++j) { + gsr_kms_response_dma_buf *dma_buf = &self->kms_response.items[i].dma_buf[j]; + if(dma_buf->fd > 0) { + close(dma_buf->fd); + dma_buf->fd = -1; + } + } + self->kms_response.items[i].num_dma_bufs = 0; + } + self->kms_response.num_items = 0; +} + +static void gsr_capture_kms_stop(gsr_capture_kms *self) { + if(self->input_texture_id) { + self->params.egl->glDeleteTextures(1, &self->input_texture_id); + self->input_texture_id = 0; + } + + if(self->external_input_texture_id) { + self->params.egl->glDeleteTextures(1, &self->external_input_texture_id); + self->external_input_texture_id = 0; + } + + if(self->cursor_texture_id) { + self->params.egl->glDeleteTextures(1, &self->cursor_texture_id); + self->cursor_texture_id = 0; + } + + // if(self->drm_fd > 0) { + // close(self->drm_fd); + // self->drm_fd = -1; + // } + + gsr_capture_kms_cleanup_kms_fds(self); + gsr_kms_client_deinit(&self->kms_client); + gsr_cursor_deinit(&self->x11_cursor); +} + +static int max_int(int a, int b) { + return a > b ? a : b; +} + +static void gsr_capture_kms_create_input_texture_ids(gsr_capture_kms *self) { + const float border_color[4] = { 0.0f, 0.0f, 0.0f, 0.0f }; + + self->params.egl->glGenTextures(1, &self->input_texture_id); + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->input_texture_id); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, border_color); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + + self->params.egl->glGenTextures(1, &self->external_input_texture_id); + self->params.egl->glBindTexture(GL_TEXTURE_EXTERNAL_OES, self->external_input_texture_id); + self->params.egl->glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameterfv(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_BORDER_COLOR, border_color); + self->params.egl->glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + self->params.egl->glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + self->params.egl->glBindTexture(GL_TEXTURE_EXTERNAL_OES, 0); + + const bool cursor_texture_id_is_external = self->params.egl->gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA; + const int cursor_texture_id_target = cursor_texture_id_is_external ? GL_TEXTURE_EXTERNAL_OES : GL_TEXTURE_2D; + + self->params.egl->glGenTextures(1, &self->cursor_texture_id); + self->params.egl->glBindTexture(cursor_texture_id_target, self->cursor_texture_id); + self->params.egl->glTexParameteri(cursor_texture_id_target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameteri(cursor_texture_id_target, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameterfv(cursor_texture_id_target, GL_TEXTURE_BORDER_COLOR, border_color); + self->params.egl->glTexParameteri(cursor_texture_id_target, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + self->params.egl->glTexParameteri(cursor_texture_id_target, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + self->params.egl->glBindTexture(cursor_texture_id_target, 0); +} + +/* TODO: On monitor reconfiguration, find monitor x, y, width and height again. Do the same for nvfbc. */ + +typedef struct { + MonitorId *monitor_id; + const char *monitor_to_capture; + int monitor_to_capture_len; + int num_monitors; +} MonitorCallbackUserdata; + +static void monitor_callback(const gsr_monitor *monitor, void *userdata) { + MonitorCallbackUserdata *monitor_callback_userdata = userdata; + ++monitor_callback_userdata->num_monitors; + + if(monitor_callback_userdata->monitor_to_capture_len != monitor->name_len || memcmp(monitor_callback_userdata->monitor_to_capture, monitor->name, monitor->name_len) != 0) + return; + + if(monitor_callback_userdata->monitor_id->num_connector_ids < MAX_CONNECTOR_IDS) { + monitor_callback_userdata->monitor_id->connector_ids[monitor_callback_userdata->monitor_id->num_connector_ids] = monitor->connector_id; + ++monitor_callback_userdata->monitor_id->num_connector_ids; + } + + if(monitor_callback_userdata->monitor_id->num_connector_ids == MAX_CONNECTOR_IDS) + fprintf(stderr, "gsr warning: reached max connector ids\n"); +} + +static vec2i rotate_capture_size_if_rotated(gsr_capture_kms *self, vec2i capture_size) { + if(self->monitor_rotation == GSR_MONITOR_ROT_90 || self->monitor_rotation == GSR_MONITOR_ROT_270) { + int tmp_x = capture_size.x; + capture_size.x = capture_size.y; + capture_size.y = tmp_x; + } + return capture_size; +} + +static int gsr_capture_kms_start(gsr_capture *cap, gsr_capture_metadata *capture_metadata) { + gsr_capture_kms *self = cap->priv; + + gsr_capture_kms_create_input_texture_ids(self); + + gsr_monitor monitor; + self->monitor_id.num_connector_ids = 0; + + int kms_init_res = gsr_kms_client_init(&self->kms_client, self->params.egl->card_path); + if(kms_init_res != 0) + return kms_init_res; + + self->is_x11 = gsr_window_get_display_server(self->params.egl->window) == GSR_DISPLAY_SERVER_X11; + const gsr_connection_type connection_type = self->is_x11 ? GSR_CONNECTION_X11 : GSR_CONNECTION_DRM; + if(self->is_x11) { + Display *display = gsr_window_get_display(self->params.egl->window); + gsr_cursor_init(&self->x11_cursor, self->params.egl, display); + } + + MonitorCallbackUserdata monitor_callback_userdata = { + &self->monitor_id, + self->params.display_to_capture, strlen(self->params.display_to_capture), + 0, + }; + for_each_active_monitor_output(self->params.egl->window, self->params.egl->card_path, connection_type, monitor_callback, &monitor_callback_userdata); + + if(!get_monitor_by_name(self->params.egl, connection_type, self->params.display_to_capture, &monitor)) { + fprintf(stderr, "gsr error: gsr_capture_kms_start: failed to find monitor by name \"%s\"\n", self->params.display_to_capture); + gsr_capture_kms_stop(self); + return -1; + } + + monitor.name = self->params.display_to_capture; + vec2i monitor_position = {0, 0}; + drm_monitor_get_display_server_data(self->params.egl->window, &monitor, &self->monitor_rotation, &monitor_position); + + self->capture_pos = monitor.pos; + /* Monitor size is already rotated on x11 when the monitor is rotated, no need to apply it ourselves */ + if(self->is_x11) + self->capture_size = monitor.size; + else + self->capture_size = rotate_capture_size_if_rotated(self, monitor.size); + + if(self->params.output_resolution.x > 0 && self->params.output_resolution.y > 0) { + self->params.output_resolution = scale_keep_aspect_ratio(self->capture_size, self->params.output_resolution); + capture_metadata->width = self->params.output_resolution.x; + capture_metadata->height = self->params.output_resolution.y; + } else if(self->params.region_size.x > 0 && self->params.region_size.y > 0) { + capture_metadata->width = self->params.region_size.x; + capture_metadata->height = self->params.region_size.y; + } else { + capture_metadata->width = self->capture_size.x; + capture_metadata->height = self->capture_size.y; + } + + self->last_time_monitor_check = clock_get_monotonic_seconds(); + return 0; +} + +static void gsr_capture_kms_on_event(gsr_capture *cap, gsr_egl *egl) { + gsr_capture_kms *self = cap->priv; + if(!self->is_x11) + return; + + XEvent *xev = gsr_window_get_event_data(egl->window); + gsr_cursor_on_event(&self->x11_cursor, xev); +} + +// TODO: This is disabled for now because we want to be able to record at a framerate higher than the monitor framerate +// static void gsr_capture_kms_tick(gsr_capture *cap) { +// gsr_capture_kms *self = cap->priv; + +// if(self->drm_fd <= 0) +// self->drm_fd = open(self->params.egl->card_path, O_RDONLY); + +// if(self->drm_fd <= 0) +// return; + +// uint64_t sequence = 0; +// uint64_t ns = 0; +// if(drmCrtcGetSequence(self->drm_fd, 79, &sequence, &ns) != 0) +// return; + +// if(sequence != self->prev_sequence) { +// self->prev_sequence = sequence; +// self->damaged = true; +// } +// } + +static gsr_kms_response_item* find_drm_by_connector_id(gsr_kms_response *kms_response, uint32_t connector_id) { + for(int i = 0; i < kms_response->num_items; ++i) { + if(kms_response->items[i].connector_id == connector_id && !kms_response->items[i].is_cursor) + return &kms_response->items[i]; + } + return NULL; +} + +static gsr_kms_response_item* find_largest_drm(gsr_kms_response *kms_response) { + if(kms_response->num_items == 0) + return NULL; + + int64_t largest_size = 0; + gsr_kms_response_item *largest_drm = &kms_response->items[0]; + for(int i = 0; i < kms_response->num_items; ++i) { + const int64_t size = (int64_t)kms_response->items[i].width * (int64_t)kms_response->items[i].height; + if(size > largest_size && !kms_response->items[i].is_cursor) { + largest_size = size; + largest_drm = &kms_response->items[i]; + } + } + return largest_drm; +} + +static gsr_kms_response_item* find_cursor_drm(gsr_kms_response *kms_response, uint32_t connector_id) { + gsr_kms_response_item *cursor_drm = NULL; + for(int i = 0; i < kms_response->num_items; ++i) { + if(kms_response->items[i].is_cursor) { + cursor_drm = &kms_response->items[i]; + if(kms_response->items[i].connector_id == connector_id) + break; + } + } + return cursor_drm; +} + +static bool hdr_metadata_is_supported_format(const struct hdr_output_metadata *hdr_metadata) { + return hdr_metadata->metadata_type == HDMI_STATIC_METADATA_TYPE1 && + hdr_metadata->hdmi_metadata_type1.metadata_type == HDMI_STATIC_METADATA_TYPE1 && + hdr_metadata->hdmi_metadata_type1.eotf == HDMI_EOTF_SMPTE_ST2084; +} + +// TODO: Check if this hdr data can be changed after the call to av_packet_side_data_add +static void gsr_kms_set_hdr_metadata(gsr_capture_kms *self, const gsr_kms_response_item *drm_fd) { + if(self->hdr_metadata_set) + return; + + self->hdr_metadata_set = true; + self->hdr_metadata = drm_fd->hdr_metadata; +} + +static vec2i swap_vec2i(vec2i value) { + int tmp = value.x; + value.x = value.y; + value.y = tmp; + return value; +} + +static EGLImage gsr_capture_kms_create_egl_image(gsr_capture_kms *self, const gsr_kms_response_item *drm_fd, const int *fds, const uint32_t *offsets, const uint32_t *pitches, const uint64_t *modifiers, bool use_modifiers) { + intptr_t img_attr[44]; + setup_dma_buf_attrs(img_attr, drm_fd->pixel_format, drm_fd->width, drm_fd->height, fds, offsets, pitches, modifiers, drm_fd->num_dma_bufs, use_modifiers); + while(self->params.egl->eglGetError() != EGL_SUCCESS){} + EGLImage image = self->params.egl->eglCreateImage(self->params.egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr); + if(!image || self->params.egl->eglGetError() != EGL_SUCCESS) { + if(image) + self->params.egl->eglDestroyImage(self->params.egl->egl_display, image); + return NULL; + } + return image; +} + +static EGLImage gsr_capture_kms_create_egl_image_with_fallback(gsr_capture_kms *self, const gsr_kms_response_item *drm_fd) { + // TODO: This causes a crash sometimes on steam deck, why? is it a driver bug? a vaapi pure version doesn't cause a crash. + // Even ffmpeg kmsgrab causes this crash. The error is: + // amdgpu: Failed to allocate a buffer: + // amdgpu: size : 28508160 bytes + // amdgpu: alignment : 2097152 bytes + // amdgpu: domains : 4 + // amdgpu: flags : 4 + // amdgpu: Failed to allocate a buffer: + // amdgpu: size : 28508160 bytes + // amdgpu: alignment : 2097152 bytes + // amdgpu: domains : 4 + // amdgpu: flags : 4 + // EE ../jupiter-mesa/src/gallium/drivers/radeonsi/radeon_vcn_enc.c:516 radeon_create_encoder UVD - Can't create CPB buffer. + // [hevc_vaapi @ 0x55ea72b09840] Failed to upload encode parameters: 2 (resource allocation failed). + // [hevc_vaapi @ 0x55ea72b09840] Encode failed: -5. + // Error: avcodec_send_frame failed, error: Input/output error + // Assertion pic->display_order == pic->encode_order failed at libavcodec/vaapi_encode_h265.c:765 + // kms server info: kms client shutdown, shutting down the server + + int fds[GSR_KMS_MAX_DMA_BUFS]; + uint32_t offsets[GSR_KMS_MAX_DMA_BUFS]; + uint32_t pitches[GSR_KMS_MAX_DMA_BUFS]; + uint64_t modifiers[GSR_KMS_MAX_DMA_BUFS]; + + for(int i = 0; i < drm_fd->num_dma_bufs; ++i) { + fds[i] = drm_fd->dma_buf[i].fd; + offsets[i] = drm_fd->dma_buf[i].offset; + pitches[i] = drm_fd->dma_buf[i].pitch; + modifiers[i] = drm_fd->modifier; + } + + EGLImage image = NULL; + if(self->no_modifiers_fallback) { + image = gsr_capture_kms_create_egl_image(self, drm_fd, fds, offsets, pitches, modifiers, false); + } else { + image = gsr_capture_kms_create_egl_image(self, drm_fd, fds, offsets, pitches, modifiers, true); + if(!image) { + fprintf(stderr, "gsr error: gsr_capture_kms_create_egl_image_with_fallback: failed to create egl image with modifiers, trying without modifiers\n"); + self->no_modifiers_fallback = true; + image = gsr_capture_kms_create_egl_image(self, drm_fd, fds, offsets, pitches, modifiers, false); + } + } + return image; +} + +static bool gsr_capture_kms_bind_image_to_texture(gsr_capture_kms *self, EGLImage image, unsigned int texture_id, bool external_texture) { + const int texture_target = external_texture ? GL_TEXTURE_EXTERNAL_OES : GL_TEXTURE_2D; + while(self->params.egl->glGetError() != 0){} + self->params.egl->glBindTexture(texture_target, texture_id); + self->params.egl->glEGLImageTargetTexture2DOES(texture_target, image); + const bool success = self->params.egl->glGetError() == 0; + self->params.egl->glBindTexture(texture_target, 0); + return success; +} + +static void gsr_capture_kms_bind_image_to_input_texture_with_fallback(gsr_capture_kms *self, EGLImage image) { + if(self->external_texture_fallback) { + gsr_capture_kms_bind_image_to_texture(self, image, self->external_input_texture_id, true); + } else { + if(!gsr_capture_kms_bind_image_to_texture(self, image, self->input_texture_id, false)) { + fprintf(stderr, "gsr error: gsr_capture_kms_capture: failed to bind image to texture, trying with external texture\n"); + self->external_texture_fallback = true; + gsr_capture_kms_bind_image_to_texture(self, image, self->external_input_texture_id, true); + } + } +} + +static gsr_kms_response_item* find_monitor_drm(gsr_capture_kms *self, bool *capture_is_combined_plane) { + *capture_is_combined_plane = false; + gsr_kms_response_item *drm_fd = NULL; + + for(int i = 0; i < self->monitor_id.num_connector_ids; ++i) { + drm_fd = find_drm_by_connector_id(&self->kms_response, self->monitor_id.connector_ids[i]); + if(drm_fd) + break; + } + + // Will never happen on wayland unless the target monitor has been disconnected + if(!drm_fd && self->is_x11) { + drm_fd = find_largest_drm(&self->kms_response); + *capture_is_combined_plane = true; + } + + return drm_fd; +} + +static gsr_kms_response_item* find_cursor_drm_if_on_monitor(gsr_capture_kms *self, uint32_t monitor_connector_id, bool capture_is_combined_plane) { + gsr_kms_response_item *cursor_drm_fd = find_cursor_drm(&self->kms_response, monitor_connector_id); + if(!capture_is_combined_plane && cursor_drm_fd && cursor_drm_fd->connector_id != monitor_connector_id) + cursor_drm_fd = NULL; + return cursor_drm_fd; +} + +static void render_drm_cursor(gsr_capture_kms *self, gsr_color_conversion *color_conversion, const gsr_kms_response_item *cursor_drm_fd, vec2i target_pos, vec2i output_size, vec2i framebuffer_size) { + const vec2d scale = { + self->capture_size.x == 0 ? 0 : (double)output_size.x / (double)self->capture_size.x, + self->capture_size.y == 0 ? 0 : (double)output_size.y / (double)self->capture_size.y + }; + + const bool cursor_texture_id_is_external = self->params.egl->gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA; + const vec2i cursor_size = {cursor_drm_fd->width, cursor_drm_fd->height}; + + vec2i cursor_pos = {cursor_drm_fd->x, cursor_drm_fd->y}; + switch(self->monitor_rotation) { + case GSR_MONITOR_ROT_0: + break; + case GSR_MONITOR_ROT_90: + cursor_pos = swap_vec2i(cursor_pos); + cursor_pos.x = framebuffer_size.x - cursor_pos.x; + // TODO: Remove this horrible hack + cursor_pos.x -= cursor_size.x; + break; + case GSR_MONITOR_ROT_180: + cursor_pos.x = framebuffer_size.x - cursor_pos.x; + cursor_pos.y = framebuffer_size.y - cursor_pos.y; + // TODO: Remove this horrible hack + cursor_pos.x -= cursor_size.x; + cursor_pos.y -= cursor_size.y; + break; + case GSR_MONITOR_ROT_270: + cursor_pos = swap_vec2i(cursor_pos); + cursor_pos.y = framebuffer_size.y - cursor_pos.y; + // TODO: Remove this horrible hack + cursor_pos.y -= cursor_size.y; + break; + } + + cursor_pos.x -= self->params.region_position.x; + cursor_pos.y -= self->params.region_position.y; + + cursor_pos.x *= scale.x; + cursor_pos.y *= scale.y; + + cursor_pos.x += target_pos.x; + cursor_pos.y += target_pos.y; + + int fds[GSR_KMS_MAX_DMA_BUFS]; + uint32_t offsets[GSR_KMS_MAX_DMA_BUFS]; + uint32_t pitches[GSR_KMS_MAX_DMA_BUFS]; + uint64_t modifiers[GSR_KMS_MAX_DMA_BUFS]; + + for(int i = 0; i < cursor_drm_fd->num_dma_bufs; ++i) { + fds[i] = cursor_drm_fd->dma_buf[i].fd; + offsets[i] = cursor_drm_fd->dma_buf[i].offset; + pitches[i] = cursor_drm_fd->dma_buf[i].pitch; + modifiers[i] = cursor_drm_fd->modifier; + } + + intptr_t img_attr_cursor[44]; + setup_dma_buf_attrs(img_attr_cursor, cursor_drm_fd->pixel_format, cursor_drm_fd->width, cursor_drm_fd->height, + fds, offsets, pitches, modifiers, cursor_drm_fd->num_dma_bufs, true); + + EGLImage cursor_image = self->params.egl->eglCreateImage(self->params.egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr_cursor); + const int target = cursor_texture_id_is_external ? GL_TEXTURE_EXTERNAL_OES : GL_TEXTURE_2D; + self->params.egl->glBindTexture(target, self->cursor_texture_id); + self->params.egl->glEGLImageTargetTexture2DOES(target, cursor_image); + self->params.egl->glBindTexture(target, 0); + + if(cursor_image) + self->params.egl->eglDestroyImage(self->params.egl->egl_display, cursor_image); + + self->params.egl->glEnable(GL_SCISSOR_TEST); + self->params.egl->glScissor(target_pos.x, target_pos.y, output_size.x, output_size.y); + + gsr_color_conversion_draw(color_conversion, self->cursor_texture_id, + cursor_pos, (vec2i){cursor_size.x * scale.x, cursor_size.y * scale.y}, + (vec2i){0, 0}, cursor_size, cursor_size, + gsr_monitor_rotation_to_rotation(self->monitor_rotation), GSR_SOURCE_COLOR_RGB, cursor_texture_id_is_external, true); + + self->params.egl->glDisable(GL_SCISSOR_TEST); +} + +static void render_x11_cursor(gsr_capture_kms *self, gsr_color_conversion *color_conversion, vec2i capture_pos, vec2i target_pos, vec2i output_size) { + if(!self->x11_cursor.visible) + return; + + const vec2d scale = { + self->capture_size.x == 0 ? 0 : (double)output_size.x / (double)self->capture_size.x, + self->capture_size.y == 0 ? 0 : (double)output_size.y / (double)self->capture_size.y + }; + + Display *display = gsr_window_get_display(self->params.egl->window); + gsr_cursor_tick(&self->x11_cursor, DefaultRootWindow(display)); + + const vec2i cursor_pos = { + target_pos.x + (self->x11_cursor.position.x - self->x11_cursor.hotspot.x - capture_pos.x) * scale.x, + target_pos.y + (self->x11_cursor.position.y - self->x11_cursor.hotspot.y - capture_pos.y) * scale.y + }; + + self->params.egl->glEnable(GL_SCISSOR_TEST); + self->params.egl->glScissor(target_pos.x, target_pos.y, output_size.x, output_size.y); + + gsr_color_conversion_draw(color_conversion, self->x11_cursor.texture_id, + cursor_pos, (vec2i){self->x11_cursor.size.x * scale.x, self->x11_cursor.size.y * scale.y}, + (vec2i){0, 0}, self->x11_cursor.size, self->x11_cursor.size, + GSR_ROT_0, GSR_SOURCE_COLOR_RGB, false, true); + + self->params.egl->glDisable(GL_SCISSOR_TEST); +} + +static void gsr_capture_kms_update_capture_size_change(gsr_capture_kms *self, gsr_color_conversion *color_conversion, vec2i target_pos, const gsr_kms_response_item *drm_fd) { + if(target_pos.x != self->prev_target_pos.x || target_pos.y != self->prev_target_pos.y || drm_fd->src_w != self->prev_plane_size.x || drm_fd->src_h != self->prev_plane_size.y) { + self->prev_target_pos = target_pos; + self->prev_plane_size = self->capture_size; + gsr_color_conversion_clear(color_conversion); + } +} + +static void gsr_capture_kms_update_connector_ids(gsr_capture_kms *self) { + const double now = clock_get_monotonic_seconds(); + if(now - self->last_time_monitor_check < FIND_CRTC_BY_NAME_TIMEOUT_SECONDS) + return; + + self->last_time_monitor_check = now; + /* TODO: Assume for now that there is only 1 framebuffer for all monitors and it doesn't change */ + if(self->is_x11) + return; + + self->monitor_id.num_connector_ids = 0; + const gsr_connection_type connection_type = self->is_x11 ? GSR_CONNECTION_X11 : GSR_CONNECTION_DRM; + // MonitorCallbackUserdata monitor_callback_userdata = { + // &self->monitor_id, + // self->params.display_to_capture, strlen(self->params.display_to_capture), + // 0, + // }; + // for_each_active_monitor_output(self->params.egl->window, self->params.egl->card_path, connection_type, monitor_callback, &monitor_callback_userdata); + + gsr_monitor monitor; + if(!get_monitor_by_name(self->params.egl, connection_type, self->params.display_to_capture, &monitor)) { + fprintf(stderr, "gsr error: gsr_capture_kms_update_connector_ids: failed to find monitor by name \"%s\"\n", self->params.display_to_capture); + return; + } + + self->monitor_id.num_connector_ids = 1; + self->monitor_id.connector_ids[0] = monitor.connector_id; + + monitor.name = self->params.display_to_capture; + vec2i monitor_position = {0, 0}; + // TODO: This is cached. We need it updated. + drm_monitor_get_display_server_data(self->params.egl->window, &monitor, &self->monitor_rotation, &monitor_position); + + self->capture_pos = monitor.pos; + /* Monitor size is already rotated on x11 when the monitor is rotated, no need to apply it ourselves */ + if(self->is_x11) + self->capture_size = monitor.size; + else + self->capture_size = rotate_capture_size_if_rotated(self, monitor.size); +} + +static int gsr_capture_kms_capture(gsr_capture *cap, gsr_capture_metadata *capture_metadata, gsr_color_conversion *color_conversion) { + gsr_capture_kms *self = cap->priv; + + gsr_capture_kms_cleanup_kms_fds(self); + + if(gsr_kms_client_get_kms(&self->kms_client, &self->kms_response) != 0) { + fprintf(stderr, "gsr error: gsr_capture_kms_capture: failed to get kms, error: %d (%s)\n", self->kms_response.result, self->kms_response.err_msg); + return -1; + } + + if(self->kms_response.num_items == 0) { + static bool error_shown = false; + if(!error_shown) { + error_shown = true; + fprintf(stderr, "gsr error: no drm found, capture will fail\n"); + } + return -1; + } + + gsr_capture_kms_update_connector_ids(self); + + bool capture_is_combined_plane = false; + const gsr_kms_response_item *drm_fd = find_monitor_drm(self, &capture_is_combined_plane); + if(!drm_fd) { + gsr_capture_kms_cleanup_kms_fds(self); + return -1; + } + + if(drm_fd->has_hdr_metadata && self->params.hdr && hdr_metadata_is_supported_format(&drm_fd->hdr_metadata)) + gsr_kms_set_hdr_metadata(self, drm_fd); + + self->capture_size = rotate_capture_size_if_rotated(self, (vec2i){ drm_fd->src_w, drm_fd->src_h }); + const vec2i original_frame_size = self->capture_size; + if(self->params.region_size.x > 0 && self->params.region_size.y > 0) + self->capture_size = self->params.region_size; + + const bool is_scaled = self->params.output_resolution.x > 0 && self->params.output_resolution.y > 0; + vec2i output_size = is_scaled ? self->params.output_resolution : self->capture_size; + output_size = scale_keep_aspect_ratio(self->capture_size, output_size); + + const vec2i target_pos = { max_int(0, capture_metadata->width / 2 - output_size.x / 2), max_int(0, capture_metadata->height / 2 - output_size.y / 2) }; + gsr_capture_kms_update_capture_size_change(self, color_conversion, target_pos, drm_fd); + + vec2i capture_pos = self->capture_pos; + if(!capture_is_combined_plane) + capture_pos = (vec2i){drm_fd->x, drm_fd->y}; + + capture_pos.x += self->params.region_position.x; + capture_pos.y += self->params.region_position.y; + + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); + + EGLImage image = gsr_capture_kms_create_egl_image_with_fallback(self, drm_fd); + if(image) { + gsr_capture_kms_bind_image_to_input_texture_with_fallback(self, image); + self->params.egl->eglDestroyImage(self->params.egl->egl_display, image); + } + + gsr_color_conversion_draw(color_conversion, self->external_texture_fallback ? self->external_input_texture_id : self->input_texture_id, + target_pos, output_size, + capture_pos, self->capture_size, original_frame_size, + gsr_monitor_rotation_to_rotation(self->monitor_rotation), GSR_SOURCE_COLOR_RGB, self->external_texture_fallback, false); + + if(self->params.record_cursor) { + gsr_kms_response_item *cursor_drm_fd = find_cursor_drm_if_on_monitor(self, drm_fd->connector_id, capture_is_combined_plane); + // The cursor is handled by x11 on x11 instead of using the cursor drm plane because on prime systems with a dedicated nvidia gpu + // the cursor plane is not available when the cursor is on the monitor controlled by the nvidia device. + // TODO: This doesn't work properly with software cursor on x11 since it will draw the x11 cursor on top of the cursor already in the framebuffer. + // Detect if software cursor is used on x11 somehow. + if(self->is_x11) { + vec2i cursor_monitor_offset = self->capture_pos; + cursor_monitor_offset.x += self->params.region_position.x; + cursor_monitor_offset.y += self->params.region_position.y; + render_x11_cursor(self, color_conversion, cursor_monitor_offset, target_pos, output_size); + } else if(cursor_drm_fd) { + const vec2i framebuffer_size = rotate_capture_size_if_rotated(self, (vec2i){ drm_fd->src_w, drm_fd->src_h }); + render_drm_cursor(self, color_conversion, cursor_drm_fd, target_pos, output_size, framebuffer_size); + } + } + + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); + + gsr_capture_kms_cleanup_kms_fds(self); + + return 0; +} + +static bool gsr_capture_kms_should_stop(gsr_capture *cap, bool *err) { + (void)cap; + if(err) + *err = false; + return false; +} + +static bool gsr_capture_kms_uses_external_image(gsr_capture *cap) { + (void)cap; + return true; +} + +static bool gsr_capture_kms_set_hdr_metadata(gsr_capture *cap, AVMasteringDisplayMetadata *mastering_display_metadata, AVContentLightMetadata *light_metadata) { + gsr_capture_kms *self = cap->priv; + + if(!self->hdr_metadata_set) + return false; + + light_metadata->MaxCLL = self->hdr_metadata.hdmi_metadata_type1.max_cll; + light_metadata->MaxFALL = self->hdr_metadata.hdmi_metadata_type1.max_fall; + + for(int i = 0; i < 3; ++i) { + mastering_display_metadata->display_primaries[i][0] = av_make_q(self->hdr_metadata.hdmi_metadata_type1.display_primaries[i].x, 50000); + mastering_display_metadata->display_primaries[i][1] = av_make_q(self->hdr_metadata.hdmi_metadata_type1.display_primaries[i].y, 50000); + } + + mastering_display_metadata->white_point[0] = av_make_q(self->hdr_metadata.hdmi_metadata_type1.white_point.x, 50000); + mastering_display_metadata->white_point[1] = av_make_q(self->hdr_metadata.hdmi_metadata_type1.white_point.y, 50000); + + mastering_display_metadata->min_luminance = av_make_q(self->hdr_metadata.hdmi_metadata_type1.min_display_mastering_luminance, 10000); + mastering_display_metadata->max_luminance = av_make_q(self->hdr_metadata.hdmi_metadata_type1.max_display_mastering_luminance, 1); + + mastering_display_metadata->has_primaries = true; + mastering_display_metadata->has_luminance = true; + + return true; +} + +// static bool gsr_capture_kms_is_damaged(gsr_capture *cap) { +// gsr_capture_kms *self = cap->priv; +// return self->damaged; +// } + +// static void gsr_capture_kms_clear_damage(gsr_capture *cap) { +// gsr_capture_kms *self = cap->priv; +// self->damaged = false; +// } + +static void gsr_capture_kms_destroy(gsr_capture *cap) { + gsr_capture_kms *self = cap->priv; + if(cap->priv) { + gsr_capture_kms_stop(self); + free((void*)self->params.display_to_capture); + self->params.display_to_capture = NULL; + free(cap->priv); + cap->priv = NULL; + } + free(cap); +} + +gsr_capture* gsr_capture_kms_create(const gsr_capture_kms_params *params) { + if(!params) { + fprintf(stderr, "gsr error: gsr_capture_kms_create params is NULL\n"); + return NULL; + } + + gsr_capture *cap = calloc(1, sizeof(gsr_capture)); + if(!cap) + return NULL; + + gsr_capture_kms *cap_kms = calloc(1, sizeof(gsr_capture_kms)); + if(!cap_kms) { + free(cap); + return NULL; + } + + const char *display_to_capture = strdup(params->display_to_capture); + if(!display_to_capture) { + free(cap); + free(cap_kms); + return NULL; + } + + cap_kms->params = *params; + cap_kms->params.display_to_capture = display_to_capture; + + *cap = (gsr_capture) { + .start = gsr_capture_kms_start, + .on_event = gsr_capture_kms_on_event, + //.tick = gsr_capture_kms_tick, + .should_stop = gsr_capture_kms_should_stop, + .capture = gsr_capture_kms_capture, + .uses_external_image = gsr_capture_kms_uses_external_image, + .set_hdr_metadata = gsr_capture_kms_set_hdr_metadata, + //.is_damaged = gsr_capture_kms_is_damaged, + //.clear_damage = gsr_capture_kms_clear_damage, + .destroy = gsr_capture_kms_destroy, + .priv = cap_kms + }; + + return cap; +} diff --git a/src/capture/kms_cuda.c b/src/capture/kms_cuda.c deleted file mode 100644 index d93d603..0000000 --- a/src/capture/kms_cuda.c +++ /dev/null @@ -1,615 +0,0 @@ -#include "../../include/capture/kms_cuda.h" -#include "../../kms/client/kms_client.h" -#include "../../include/utils.h" -#include "../../include/color_conversion.h" -#include "../../include/cuda.h" -#include <stdlib.h> -#include <stdio.h> -#include <unistd.h> -#include <assert.h> -#include <libavutil/hwcontext.h> -#include <libavutil/hwcontext_cuda.h> -#include <libavutil/frame.h> -#include <libavcodec/avcodec.h> - -/* - TODO: Use dummy pool for cuda buffer so we can create our own cuda buffers from pixel buffer objects - and copy the input textures to the pixel buffer objects. Use sw_format NV12 as well. Then this is - similar to kms_vaapi. This allows us to remove one extra texture and texture copy. -*/ -/* TODO: Support cursor plane capture when nvidia supports cursor plane */ - -#define MAX_CONNECTOR_IDS 32 - -typedef struct { - uint32_t connector_ids[MAX_CONNECTOR_IDS]; - int num_connector_ids; -} MonitorId; - -typedef struct { - gsr_capture_kms_cuda_params params; - XEvent xev; - - bool should_stop; - bool stop_is_error; - bool created_hw_frame; - - gsr_cuda cuda; - - gsr_kms_client kms_client; - gsr_kms_response kms_response; - gsr_kms_response_fd wayland_kms_data; - bool using_wayland_capture; - - vec2i capture_pos; - vec2i capture_size; - MonitorId monitor_id; - - CUgraphicsResource cuda_graphics_resource; - CUarray mapped_array; - - unsigned int input_texture; - unsigned int target_texture; - gsr_color_conversion color_conversion; -} gsr_capture_kms_cuda; - -static int max_int(int a, int b) { - return a > b ? a : b; -} - -static void gsr_capture_kms_cuda_stop(gsr_capture *cap, AVCodecContext *video_codec_context); - -static bool cuda_create_codec_context(gsr_capture_kms_cuda *cap_kms, AVCodecContext *video_codec_context) { - CUcontext old_ctx; - cap_kms->cuda.cuCtxPushCurrent_v2(cap_kms->cuda.cu_ctx); - - AVBufferRef *device_ctx = av_hwdevice_ctx_alloc(AV_HWDEVICE_TYPE_CUDA); - if(!device_ctx) { - fprintf(stderr, "Error: Failed to create hardware device context\n"); - cap_kms->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - AVHWDeviceContext *hw_device_context = (AVHWDeviceContext*)device_ctx->data; - AVCUDADeviceContext *cuda_device_context = (AVCUDADeviceContext*)hw_device_context->hwctx; - cuda_device_context->cuda_ctx = cap_kms->cuda.cu_ctx; - if(av_hwdevice_ctx_init(device_ctx) < 0) { - fprintf(stderr, "Error: Failed to create hardware device context\n"); - av_buffer_unref(&device_ctx); - cap_kms->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - AVBufferRef *frame_context = av_hwframe_ctx_alloc(device_ctx); - if(!frame_context) { - fprintf(stderr, "Error: Failed to create hwframe context\n"); - av_buffer_unref(&device_ctx); - cap_kms->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - AVHWFramesContext *hw_frame_context = - (AVHWFramesContext *)frame_context->data; - hw_frame_context->width = video_codec_context->width; - hw_frame_context->height = video_codec_context->height; - hw_frame_context->sw_format = AV_PIX_FMT_BGR0; - hw_frame_context->format = video_codec_context->pix_fmt; - hw_frame_context->device_ref = device_ctx; - hw_frame_context->device_ctx = (AVHWDeviceContext*)device_ctx->data; - - hw_frame_context->initial_pool_size = 1; - - if (av_hwframe_ctx_init(frame_context) < 0) { - fprintf(stderr, "Error: Failed to initialize hardware frame context " - "(note: ffmpeg version needs to be > 4.0)\n"); - av_buffer_unref(&device_ctx); - //av_buffer_unref(&frame_context); - cap_kms->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - video_codec_context->hw_device_ctx = av_buffer_ref(device_ctx); - video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); - return true; -} - -// TODO: On monitor reconfiguration, find monitor x, y, width and height again. Do the same for nvfbc. - -typedef struct { - gsr_capture_kms_cuda *cap_kms; - const char *monitor_to_capture; - int monitor_to_capture_len; - int num_monitors; -} MonitorCallbackUserdata; - -static void monitor_callback(const gsr_monitor *monitor, void *userdata) { - MonitorCallbackUserdata *monitor_callback_userdata = userdata; - ++monitor_callback_userdata->num_monitors; - - if(monitor_callback_userdata->monitor_to_capture_len != monitor->name_len || memcmp(monitor_callback_userdata->monitor_to_capture, monitor->name, monitor->name_len) != 0) - return; - - if(monitor_callback_userdata->cap_kms->monitor_id.num_connector_ids < MAX_CONNECTOR_IDS) { - monitor_callback_userdata->cap_kms->monitor_id.connector_ids[monitor_callback_userdata->cap_kms->monitor_id.num_connector_ids] = monitor->connector_id; - ++monitor_callback_userdata->cap_kms->monitor_id.num_connector_ids; - } - - if(monitor_callback_userdata->cap_kms->monitor_id.num_connector_ids == MAX_CONNECTOR_IDS) - fprintf(stderr, "gsr warning: reached max connector ids\n"); -} - -static int gsr_capture_kms_cuda_start(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_kms_cuda *cap_kms = cap->priv; - - gsr_monitor monitor; - cap_kms->monitor_id.num_connector_ids = 0; - if(gsr_egl_start_capture(cap_kms->params.egl, cap_kms->params.display_to_capture)) { - if(!get_monitor_by_name(cap_kms->params.egl, GSR_CONNECTION_WAYLAND, cap_kms->params.display_to_capture, &monitor)) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_start: failed to find monitor by name \"%s\"\n", cap_kms->params.display_to_capture); - gsr_capture_kms_cuda_stop(cap, video_codec_context); - return -1; - } - cap_kms->using_wayland_capture = true; - } else { - int kms_init_res = gsr_kms_client_init(&cap_kms->kms_client, cap_kms->params.card_path); - if(kms_init_res != 0) { - gsr_capture_kms_cuda_stop(cap, video_codec_context); - return kms_init_res; - } - - MonitorCallbackUserdata monitor_callback_userdata = { - cap_kms, - cap_kms->params.display_to_capture, strlen(cap_kms->params.display_to_capture), - 0 - }; - for_each_active_monitor_output((void*)cap_kms->params.card_path, GSR_CONNECTION_DRM, monitor_callback, &monitor_callback_userdata); - - if(!get_monitor_by_name((void*)cap_kms->params.card_path, GSR_CONNECTION_DRM, cap_kms->params.display_to_capture, &monitor)) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_start: failed to find monitor by name \"%s\"\n", cap_kms->params.display_to_capture); - gsr_capture_kms_cuda_stop(cap, video_codec_context); - return -1; - } - } - - cap_kms->capture_pos = monitor.pos; - cap_kms->capture_size = monitor.size; - - video_codec_context->width = max_int(2, cap_kms->capture_size.x & ~1); - video_codec_context->height = max_int(2, cap_kms->capture_size.y & ~1); - - /* Disable vsync */ - cap_kms->params.egl->eglSwapInterval(cap_kms->params.egl->egl_display, 0); - - // TODO: overclocking is not supported on wayland... - if(!gsr_cuda_load(&cap_kms->cuda, NULL, false)) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_start: failed to load cuda\n"); - gsr_capture_kms_cuda_stop(cap, video_codec_context); - return -1; - } - - if(!cuda_create_codec_context(cap_kms, video_codec_context)) { - gsr_capture_kms_cuda_stop(cap, video_codec_context); - return -1; - } - - return 0; -} - -static unsigned int gl_create_texture(gsr_capture_kms_cuda *cap_kms, int width, int height) { - unsigned int texture_id = 0; - cap_kms->params.egl->glGenTextures(1, &texture_id); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, texture_id); - cap_kms->params.egl->glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, width, height, 0, GL_RGB, GL_UNSIGNED_BYTE, NULL); - - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); - - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - return texture_id; -} - -static bool cuda_register_opengl_texture(gsr_capture_kms_cuda *cap_kms) { - CUresult res; - CUcontext old_ctx; - res = cap_kms->cuda.cuCtxPushCurrent_v2(cap_kms->cuda.cu_ctx); - // TODO: Use cuGraphicsEGLRegisterImage instead with the window egl image (dont use window_texture). - // That removes the need for an extra texture and texture copy - res = cap_kms->cuda.cuGraphicsGLRegisterImage( - &cap_kms->cuda_graphics_resource, cap_kms->target_texture, GL_TEXTURE_2D, - CU_GRAPHICS_REGISTER_FLAGS_READ_ONLY); - if (res != CUDA_SUCCESS) { - const char *err_str = "unknown"; - cap_kms->cuda.cuGetErrorString(res, &err_str); - fprintf(stderr, "gsr error: cuda_register_opengl_texture: cuGraphicsGLRegisterImage failed, error: %s, texture " "id: %u\n", err_str, cap_kms->target_texture); - res = cap_kms->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - res = cap_kms->cuda.cuGraphicsResourceSetMapFlags(cap_kms->cuda_graphics_resource, CU_GRAPHICS_MAP_RESOURCE_FLAGS_READ_ONLY); - res = cap_kms->cuda.cuGraphicsMapResources(1, &cap_kms->cuda_graphics_resource, 0); - - res = cap_kms->cuda.cuGraphicsSubResourceGetMappedArray(&cap_kms->mapped_array, cap_kms->cuda_graphics_resource, 0, 0); - res = cap_kms->cuda.cuCtxPopCurrent_v2(&old_ctx); - return true; -} - -static void gsr_capture_kms_cuda_tick(gsr_capture *cap, AVCodecContext *video_codec_context, AVFrame **frame) { - gsr_capture_kms_cuda *cap_kms = cap->priv; - - // TODO: - cap_kms->params.egl->glClear(GL_COLOR_BUFFER_BIT); - - if(!cap_kms->created_hw_frame) { - cap_kms->created_hw_frame = true; - - av_frame_free(frame); - *frame = av_frame_alloc(); - if(!frame) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_tick: failed to allocate frame\n"); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - (*frame)->format = video_codec_context->pix_fmt; - (*frame)->width = video_codec_context->width; - (*frame)->height = video_codec_context->height; - (*frame)->color_range = video_codec_context->color_range; - (*frame)->color_primaries = video_codec_context->color_primaries; - (*frame)->color_trc = video_codec_context->color_trc; - (*frame)->colorspace = video_codec_context->colorspace; - (*frame)->chroma_location = video_codec_context->chroma_sample_location; - - if(av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, *frame, 0) < 0) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_tick: av_hwframe_get_buffer failed\n"); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - - cap_kms->params.egl->glGenTextures(1, &cap_kms->input_texture); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, cap_kms->input_texture); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - cap_kms->target_texture = gl_create_texture(cap_kms, video_codec_context->width, video_codec_context->height); - if(cap_kms->target_texture == 0) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_tick: failed to create opengl texture\n"); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - - if(!cuda_register_opengl_texture(cap_kms)) { - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - - gsr_color_conversion_params color_conversion_params = {0}; - color_conversion_params.egl = cap_kms->params.egl; - color_conversion_params.source_color = GSR_SOURCE_COLOR_RGB; - color_conversion_params.destination_color = GSR_DESTINATION_COLOR_BGR; - - color_conversion_params.destination_textures[0] = cap_kms->target_texture; - color_conversion_params.num_destination_textures = 1; - - if(gsr_color_conversion_init(&cap_kms->color_conversion, &color_conversion_params) != 0) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_tick: failed to create color conversion\n"); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - } -} - -static bool gsr_capture_kms_cuda_should_stop(gsr_capture *cap, bool *err) { - gsr_capture_kms_cuda *cap_kms = cap->priv; - if(cap_kms->should_stop) { - if(err) - *err = cap_kms->stop_is_error; - return true; - } - - if(err) - *err = false; - return false; -} - -/* Prefer non combined planes */ -static gsr_kms_response_fd* find_drm_by_connector_id(gsr_kms_response *kms_response, uint32_t connector_id) { - int index_combined = -1; - for(int i = 0; i < kms_response->num_fds; ++i) { - if(kms_response->fds[i].connector_id == connector_id && !kms_response->fds[i].is_cursor) { - if(kms_response->fds[i].is_combined_plane) - index_combined = i; - else - return &kms_response->fds[i]; - } - } - - if(index_combined != -1) - return &kms_response->fds[index_combined]; - else - return NULL; -} - -static gsr_kms_response_fd* find_first_combined_drm(gsr_kms_response *kms_response) { - for(int i = 0; i < kms_response->num_fds; ++i) { - if(kms_response->fds[i].is_combined_plane && !kms_response->fds[i].is_cursor) - return &kms_response->fds[i]; - } - return NULL; -} - -static gsr_kms_response_fd* find_largest_drm(gsr_kms_response *kms_response) { - if(kms_response->num_fds == 0) - return NULL; - - int64_t largest_size = 0; - gsr_kms_response_fd *largest_drm = &kms_response->fds[0]; - for(int i = 0; i < kms_response->num_fds; ++i) { - const int64_t size = (int64_t)kms_response->fds[i].width * (int64_t)kms_response->fds[i].height; - if(size > largest_size && !kms_response->fds[i].is_cursor) { - largest_size = size; - largest_drm = &kms_response->fds[i]; - } - } - return largest_drm; -} - -static void gsr_capture_kms_unload_cuda_graphics(gsr_capture_kms_cuda *cap_kms) { - if(cap_kms->cuda.cu_ctx) { - CUcontext old_ctx; - cap_kms->cuda.cuCtxPushCurrent_v2(cap_kms->cuda.cu_ctx); - - if(cap_kms->cuda_graphics_resource) { - cap_kms->cuda.cuGraphicsUnmapResources(1, &cap_kms->cuda_graphics_resource, 0); - cap_kms->cuda.cuGraphicsUnregisterResource(cap_kms->cuda_graphics_resource); - cap_kms->cuda_graphics_resource = 0; - } - - cap_kms->cuda.cuCtxPopCurrent_v2(&old_ctx); - } -} - -static gsr_kms_response_fd* find_cursor_drm(gsr_kms_response *kms_response) { - for(int i = 0; i < kms_response->num_fds; ++i) { - if(kms_response->fds[i].is_cursor) - return &kms_response->fds[i]; - } - return NULL; -} - -static int gsr_capture_kms_cuda_capture(gsr_capture *cap, AVFrame *frame) { - (void)frame; - gsr_capture_kms_cuda *cap_kms = cap->priv; - - for(int i = 0; i < cap_kms->kms_response.num_fds; ++i) { - if(cap_kms->kms_response.fds[i].fd > 0) - close(cap_kms->kms_response.fds[i].fd); - cap_kms->kms_response.fds[i].fd = 0; - } - cap_kms->kms_response.num_fds = 0; - - gsr_kms_response_fd *drm_fd = NULL; - gsr_kms_response_fd *cursor_drm_fd = NULL; - bool capture_is_combined_plane = false; - if(cap_kms->using_wayland_capture) { - gsr_egl_update(cap_kms->params.egl); - cap_kms->wayland_kms_data.fd = cap_kms->params.egl->fd; - cap_kms->wayland_kms_data.width = cap_kms->params.egl->width; - cap_kms->wayland_kms_data.height = cap_kms->params.egl->height; - cap_kms->wayland_kms_data.pitch = cap_kms->params.egl->pitch; - cap_kms->wayland_kms_data.offset = cap_kms->params.egl->offset; - cap_kms->wayland_kms_data.pixel_format = cap_kms->params.egl->pixel_format; - cap_kms->wayland_kms_data.modifier = cap_kms->params.egl->modifier; - cap_kms->wayland_kms_data.connector_id = 0; - cap_kms->wayland_kms_data.is_combined_plane = false; - cap_kms->wayland_kms_data.is_cursor = false; - cap_kms->wayland_kms_data.x = cap_kms->wayland_kms_data.x; // TODO: Use these - cap_kms->wayland_kms_data.y = cap_kms->wayland_kms_data.y; - cap_kms->wayland_kms_data.src_w = cap_kms->wayland_kms_data.width; - cap_kms->wayland_kms_data.src_h = cap_kms->wayland_kms_data.height; - - cap_kms->capture_pos.x = cap_kms->wayland_kms_data.x; - cap_kms->capture_pos.y = cap_kms->wayland_kms_data.y; - - if(cap_kms->wayland_kms_data.fd <= 0) - return -1; - - drm_fd = &cap_kms->wayland_kms_data; - } else { - if(gsr_kms_client_get_kms(&cap_kms->kms_client, &cap_kms->kms_response) != 0) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_capture: failed to get kms, error: %d (%s)\n", cap_kms->kms_response.result, cap_kms->kms_response.err_msg); - return -1; - } - - if(cap_kms->kms_response.num_fds == 0) { - static bool error_shown = false; - if(!error_shown) { - error_shown = true; - fprintf(stderr, "gsr error: no drm found, capture will fail\n"); - } - return -1; - } - - for(int i = 0; i < cap_kms->monitor_id.num_connector_ids; ++i) { - drm_fd = find_drm_by_connector_id(&cap_kms->kms_response, cap_kms->monitor_id.connector_ids[i]); - if(drm_fd) - break; - } - - // Will never happen on wayland unless the target monitor has been disconnected - if(!drm_fd) { - drm_fd = find_first_combined_drm(&cap_kms->kms_response); - if(!drm_fd) - drm_fd = find_largest_drm(&cap_kms->kms_response); - capture_is_combined_plane = true; - } - - cursor_drm_fd = find_cursor_drm(&cap_kms->kms_response); - } - - if(!drm_fd) - return -1; - - if(!capture_is_combined_plane && cursor_drm_fd && cursor_drm_fd->connector_id != drm_fd->connector_id) - cursor_drm_fd = NULL; - - const intptr_t img_attr[] = { - //EGL_IMAGE_PRESERVED_KHR, EGL_TRUE, - EGL_LINUX_DRM_FOURCC_EXT, drm_fd->pixel_format,//cap_kms->params.egl->pixel_format, ARGB8888 - EGL_WIDTH, drm_fd->width,//cap_kms->params.egl->width, - EGL_HEIGHT, drm_fd->height,//cap_kms->params.egl->height, - EGL_DMA_BUF_PLANE0_FD_EXT, drm_fd->fd,//cap_kms->params.egl->fd, - EGL_DMA_BUF_PLANE0_OFFSET_EXT, drm_fd->offset,//cap_kms->params.egl->offset, - EGL_DMA_BUF_PLANE0_PITCH_EXT, drm_fd->pitch,//cap_kms->params.egl->pitch, - EGL_DMA_BUF_PLANE0_MODIFIER_LO_EXT, drm_fd->modifier & 0xFFFFFFFFULL,//cap_kms->params.egl->modifier & 0xFFFFFFFFULL, - EGL_DMA_BUF_PLANE0_MODIFIER_HI_EXT, drm_fd->modifier >> 32ULL,//cap_kms->params.egl->modifier >> 32ULL, - EGL_NONE - }; - - EGLImage image = cap_kms->params.egl->eglCreateImage(cap_kms->params.egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, cap_kms->input_texture); - cap_kms->params.egl->glEGLImageTargetTexture2DOES(GL_TEXTURE_2D, image); - cap_kms->params.egl->eglDestroyImage(cap_kms->params.egl->egl_display, image); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - vec2i capture_pos = cap_kms->capture_pos; - if(!capture_is_combined_plane) - capture_pos = (vec2i){drm_fd->x, drm_fd->y}; - - gsr_color_conversion_draw(&cap_kms->color_conversion, cap_kms->input_texture, - (vec2i){0, 0}, cap_kms->capture_size, - capture_pos, cap_kms->capture_size, - 0.0f); - - cap_kms->params.egl->eglSwapBuffers(cap_kms->params.egl->egl_display, cap_kms->params.egl->egl_surface); - - frame->linesize[0] = frame->width * 4; - - CUDA_MEMCPY2D memcpy_struct; - memcpy_struct.srcXInBytes = 0; - memcpy_struct.srcY = 0; - memcpy_struct.srcMemoryType = CU_MEMORYTYPE_ARRAY; - - memcpy_struct.dstXInBytes = 0; - memcpy_struct.dstY = 0; - memcpy_struct.dstMemoryType = CU_MEMORYTYPE_DEVICE; - - memcpy_struct.srcArray = cap_kms->mapped_array; - memcpy_struct.srcPitch = frame->linesize[0]; - memcpy_struct.dstDevice = (CUdeviceptr)frame->data[0]; - memcpy_struct.dstPitch = frame->linesize[0]; - memcpy_struct.WidthInBytes = frame->width * 4; - memcpy_struct.Height = frame->height; - cap_kms->cuda.cuMemcpy2D_v2(&memcpy_struct); - - return 0; -} - -static void gsr_capture_kms_cuda_capture_end(gsr_capture *cap, AVFrame *frame) { - (void)frame; - gsr_capture_kms_cuda *cap_kms = cap->priv; - - gsr_egl_cleanup_frame(cap_kms->params.egl); - - for(int i = 0; i < cap_kms->kms_response.num_fds; ++i) { - if(cap_kms->kms_response.fds[i].fd > 0) - close(cap_kms->kms_response.fds[i].fd); - cap_kms->kms_response.fds[i].fd = 0; - } - cap_kms->kms_response.num_fds = 0; -} - -static void gsr_capture_kms_cuda_stop(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_kms_cuda *cap_kms = cap->priv; - - gsr_color_conversion_deinit(&cap_kms->color_conversion); - - gsr_capture_kms_unload_cuda_graphics(cap_kms); - - if(cap_kms->params.egl->egl_context) { - if(cap_kms->input_texture) { - cap_kms->params.egl->glDeleteTextures(1, &cap_kms->input_texture); - cap_kms->input_texture = 0; - } - - if(cap_kms->target_texture) { - cap_kms->params.egl->glDeleteTextures(1, &cap_kms->target_texture); - cap_kms->target_texture = 0; - } - } - - for(int i = 0; i < cap_kms->kms_response.num_fds; ++i) { - if(cap_kms->kms_response.fds[i].fd > 0) - close(cap_kms->kms_response.fds[i].fd); - cap_kms->kms_response.fds[i].fd = 0; - } - cap_kms->kms_response.num_fds = 0; - - if(video_codec_context->hw_device_ctx) - av_buffer_unref(&video_codec_context->hw_device_ctx); - if(video_codec_context->hw_frames_ctx) - av_buffer_unref(&video_codec_context->hw_frames_ctx); - - gsr_cuda_unload(&cap_kms->cuda); - gsr_kms_client_deinit(&cap_kms->kms_client); -} - -static void gsr_capture_kms_cuda_destroy(gsr_capture *cap, AVCodecContext *video_codec_context) { - (void)video_codec_context; - gsr_capture_kms_cuda *cap_kms = cap->priv; - if(cap->priv) { - gsr_capture_kms_cuda_stop(cap, video_codec_context); - free((void*)cap_kms->params.display_to_capture); - cap_kms->params.display_to_capture = NULL; - free(cap->priv); - cap->priv = NULL; - } - free(cap); -} - -gsr_capture* gsr_capture_kms_cuda_create(const gsr_capture_kms_cuda_params *params) { - if(!params) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_create params is NULL\n"); - return NULL; - } - - gsr_capture *cap = calloc(1, sizeof(gsr_capture)); - if(!cap) - return NULL; - - gsr_capture_kms_cuda *cap_kms = calloc(1, sizeof(gsr_capture_kms_cuda)); - if(!cap_kms) { - free(cap); - return NULL; - } - - const char *display_to_capture = strdup(params->display_to_capture); - if(!display_to_capture) { - free(cap); - free(cap_kms); - return NULL; - } - - cap_kms->params = *params; - cap_kms->params.display_to_capture = display_to_capture; - - *cap = (gsr_capture) { - .start = gsr_capture_kms_cuda_start, - .tick = gsr_capture_kms_cuda_tick, - .should_stop = gsr_capture_kms_cuda_should_stop, - .capture = gsr_capture_kms_cuda_capture, - .capture_end = gsr_capture_kms_cuda_capture_end, - .destroy = gsr_capture_kms_cuda_destroy, - .priv = cap_kms - }; - - return cap; -} diff --git a/src/capture/kms_vaapi.c b/src/capture/kms_vaapi.c deleted file mode 100644 index 681f345..0000000 --- a/src/capture/kms_vaapi.c +++ /dev/null @@ -1,661 +0,0 @@ -#include "../../include/capture/kms_vaapi.h" -#include "../../kms/client/kms_client.h" -#include "../../include/utils.h" -#include "../../include/color_conversion.h" -#include <stdlib.h> -#include <stdio.h> -#include <unistd.h> -#include <assert.h> -#include <libavutil/hwcontext.h> -#include <libavutil/hwcontext_vaapi.h> -#include <libavutil/frame.h> -#include <libavcodec/avcodec.h> -#include <va/va.h> -#include <va/va_drmcommon.h> - -#define MAX_CONNECTOR_IDS 32 - -typedef struct { - uint32_t connector_ids[MAX_CONNECTOR_IDS]; - int num_connector_ids; -} MonitorId; - -typedef struct { - gsr_capture_kms_vaapi_params params; - XEvent xev; - - bool should_stop; - bool stop_is_error; - bool created_hw_frame; - - gsr_kms_client kms_client; - gsr_kms_response kms_response; - gsr_kms_response_fd wayland_kms_data; - bool using_wayland_capture; - - vec2i capture_pos; - vec2i capture_size; - MonitorId monitor_id; - - VADisplay va_dpy; - VADRMPRIMESurfaceDescriptor prime; - - unsigned int input_texture; - unsigned int target_textures[2]; - unsigned int cursor_texture; - - gsr_color_conversion color_conversion; -} gsr_capture_kms_vaapi; - -static int max_int(int a, int b) { - return a > b ? a : b; -} - -static void gsr_capture_kms_vaapi_stop(gsr_capture *cap, AVCodecContext *video_codec_context); - -static bool drm_create_codec_context(gsr_capture_kms_vaapi *cap_kms, AVCodecContext *video_codec_context) { - char render_path[128]; - if(!gsr_card_path_get_render_path(cap_kms->params.card_path, render_path)) { - fprintf(stderr, "gsr error: failed to get /dev/dri/renderDXXX file from %s\n", cap_kms->params.card_path); - return false; - } - - AVBufferRef *device_ctx; - if(av_hwdevice_ctx_create(&device_ctx, AV_HWDEVICE_TYPE_VAAPI, render_path, NULL, 0) < 0) { - fprintf(stderr, "Error: Failed to create hardware device context\n"); - return false; - } - - AVBufferRef *frame_context = av_hwframe_ctx_alloc(device_ctx); - if(!frame_context) { - fprintf(stderr, "Error: Failed to create hwframe context\n"); - av_buffer_unref(&device_ctx); - return false; - } - - AVHWFramesContext *hw_frame_context = - (AVHWFramesContext *)frame_context->data; - hw_frame_context->width = video_codec_context->width; - hw_frame_context->height = video_codec_context->height; - hw_frame_context->sw_format = AV_PIX_FMT_NV12;//AV_PIX_FMT_0RGB32;//AV_PIX_FMT_YUV420P;//AV_PIX_FMT_0RGB32;//AV_PIX_FMT_NV12; - hw_frame_context->format = video_codec_context->pix_fmt; - hw_frame_context->device_ref = device_ctx; - hw_frame_context->device_ctx = (AVHWDeviceContext*)device_ctx->data; - - hw_frame_context->initial_pool_size = 1; // TODO: (and in other places) - - AVVAAPIDeviceContext *vactx =((AVHWDeviceContext*)device_ctx->data)->hwctx; - cap_kms->va_dpy = vactx->display; - - if (av_hwframe_ctx_init(frame_context) < 0) { - fprintf(stderr, "Error: Failed to initialize hardware frame context " - "(note: ffmpeg version needs to be > 4.0)\n"); - av_buffer_unref(&device_ctx); - //av_buffer_unref(&frame_context); - return false; - } - - video_codec_context->hw_device_ctx = av_buffer_ref(device_ctx); - video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); - return true; -} - -#define DRM_FORMAT_MOD_INVALID 72057594037927935 - -// TODO: On monitor reconfiguration, find monitor x, y, width and height again. Do the same for nvfbc. - -typedef struct { - gsr_capture_kms_vaapi *cap_kms; - const char *monitor_to_capture; - int monitor_to_capture_len; - int num_monitors; -} MonitorCallbackUserdata; - -static void monitor_callback(const gsr_monitor *monitor, void *userdata) { - (void)monitor; - MonitorCallbackUserdata *monitor_callback_userdata = userdata; - ++monitor_callback_userdata->num_monitors; - - if(monitor_callback_userdata->monitor_to_capture_len != monitor->name_len || memcmp(monitor_callback_userdata->monitor_to_capture, monitor->name, monitor->name_len) != 0) - return; - - if(monitor_callback_userdata->cap_kms->monitor_id.num_connector_ids < MAX_CONNECTOR_IDS) { - monitor_callback_userdata->cap_kms->monitor_id.connector_ids[monitor_callback_userdata->cap_kms->monitor_id.num_connector_ids] = monitor->connector_id; - ++monitor_callback_userdata->cap_kms->monitor_id.num_connector_ids; - } - - if(monitor_callback_userdata->cap_kms->monitor_id.num_connector_ids == MAX_CONNECTOR_IDS) - fprintf(stderr, "gsr warning: reached max connector ids\n"); -} - -static int gsr_capture_kms_vaapi_start(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_kms_vaapi *cap_kms = cap->priv; - - gsr_monitor monitor; - cap_kms->monitor_id.num_connector_ids = 0; - if(gsr_egl_start_capture(cap_kms->params.egl, cap_kms->params.display_to_capture)) { - if(!get_monitor_by_name(cap_kms->params.egl, GSR_CONNECTION_WAYLAND, cap_kms->params.display_to_capture, &monitor)) { - fprintf(stderr, "gsr error: gsr_capture_kms_cuda_start: failed to find monitor by name \"%s\"\n", cap_kms->params.display_to_capture); - gsr_capture_kms_vaapi_stop(cap, video_codec_context); - return -1; - } - cap_kms->using_wayland_capture = true; - } else { - int kms_init_res = gsr_kms_client_init(&cap_kms->kms_client, cap_kms->params.card_path); - if(kms_init_res != 0) { - gsr_capture_kms_vaapi_stop(cap, video_codec_context); - return kms_init_res; - } - - MonitorCallbackUserdata monitor_callback_userdata = { - cap_kms, - cap_kms->params.display_to_capture, strlen(cap_kms->params.display_to_capture), - 0, - }; - for_each_active_monitor_output((void*)cap_kms->params.card_path, GSR_CONNECTION_DRM, monitor_callback, &monitor_callback_userdata); - - if(!get_monitor_by_name((void*)cap_kms->params.card_path, GSR_CONNECTION_DRM, cap_kms->params.display_to_capture, &monitor)) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_start: failed to find monitor by name \"%s\"\n", cap_kms->params.display_to_capture); - gsr_capture_kms_vaapi_stop(cap, video_codec_context); - return -1; - } - } - - cap_kms->capture_pos = monitor.pos; - cap_kms->capture_size = monitor.size; - - /* Disable vsync */ - cap_kms->params.egl->eglSwapInterval(cap_kms->params.egl->egl_display, 0); - - video_codec_context->width = max_int(2, even_number_ceil(cap_kms->capture_size.x)); - video_codec_context->height = max_int(2, even_number_ceil(cap_kms->capture_size.y)); - - if(!drm_create_codec_context(cap_kms, video_codec_context)) { - gsr_capture_kms_vaapi_stop(cap, video_codec_context); - return -1; - } - - return 0; -} - -static uint32_t fourcc(uint32_t a, uint32_t b, uint32_t c, uint32_t d) { - return (d << 24) | (c << 16) | (b << 8) | a; -} - -#define FOURCC_NV12 842094158 - -static void gsr_capture_kms_vaapi_tick(gsr_capture *cap, AVCodecContext *video_codec_context, AVFrame **frame) { - gsr_capture_kms_vaapi *cap_kms = cap->priv; - - // TODO: - cap_kms->params.egl->glClear(GL_COLOR_BUFFER_BIT); - - if(!cap_kms->created_hw_frame) { - cap_kms->created_hw_frame = true; - - av_frame_free(frame); - *frame = av_frame_alloc(); - if(!frame) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_tick: failed to allocate frame\n"); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - (*frame)->format = video_codec_context->pix_fmt; - (*frame)->width = video_codec_context->width; - (*frame)->height = video_codec_context->height; - (*frame)->color_range = video_codec_context->color_range; - (*frame)->color_primaries = video_codec_context->color_primaries; - (*frame)->color_trc = video_codec_context->color_trc; - (*frame)->colorspace = video_codec_context->colorspace; - (*frame)->chroma_location = video_codec_context->chroma_sample_location; - - int res = av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, *frame, 0); - if(res < 0) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_tick: av_hwframe_get_buffer failed: %d\n", res); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - - VASurfaceID target_surface_id = (uintptr_t)(*frame)->data[3]; - - VAStatus va_status = vaExportSurfaceHandle(cap_kms->va_dpy, target_surface_id, VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME_2, VA_EXPORT_SURFACE_READ_WRITE | VA_EXPORT_SURFACE_SEPARATE_LAYERS, &cap_kms->prime); - if(va_status != VA_STATUS_SUCCESS) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_tick: vaExportSurfaceHandle failed, error: %d\n", va_status); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - vaSyncSurface(cap_kms->va_dpy, target_surface_id); - - cap_kms->params.egl->glGenTextures(1, &cap_kms->input_texture); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, cap_kms->input_texture); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - cap_kms->params.egl->glGenTextures(1, &cap_kms->cursor_texture); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, cap_kms->cursor_texture); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - if(cap_kms->prime.fourcc == FOURCC_NV12) { - cap_kms->params.egl->glGenTextures(2, cap_kms->target_textures); - for(int i = 0; i < 2; ++i) { - const uint32_t formats[2] = { fourcc('R', '8', ' ', ' '), fourcc('G', 'R', '8', '8') }; - const int layer = i; - const int plane = 0; - - const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size - //const uint64_t modifier = cap_kms->prime.objects[cap_kms->prime.layers[layer].object_index[plane]].drm_format_modifier; - - const intptr_t img_attr[] = { - EGL_LINUX_DRM_FOURCC_EXT, formats[i], - EGL_WIDTH, cap_kms->prime.width / div[i], - EGL_HEIGHT, cap_kms->prime.height / div[i], - EGL_DMA_BUF_PLANE0_FD_EXT, cap_kms->prime.objects[cap_kms->prime.layers[layer].object_index[plane]].fd, - EGL_DMA_BUF_PLANE0_OFFSET_EXT, cap_kms->prime.layers[layer].offset[plane], - EGL_DMA_BUF_PLANE0_PITCH_EXT, cap_kms->prime.layers[layer].pitch[plane], - // TODO: - //EGL_DMA_BUF_PLANE0_MODIFIER_LO_EXT, modifier & 0xFFFFFFFFULL, - //EGL_DMA_BUF_PLANE0_MODIFIER_HI_EXT, modifier >> 32ULL, - EGL_NONE - }; - - while(cap_kms->params.egl->eglGetError() != EGL_SUCCESS){} - EGLImage image = cap_kms->params.egl->eglCreateImage(cap_kms->params.egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr); - if(!image) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_tick: failed to create egl image from drm fd for output drm fd, error: %d\n", cap_kms->params.egl->eglGetError()); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, cap_kms->target_textures[i]); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); - cap_kms->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); - - while(cap_kms->params.egl->glGetError()) {} - while(cap_kms->params.egl->eglGetError() != EGL_SUCCESS){} - cap_kms->params.egl->glEGLImageTargetTexture2DOES(GL_TEXTURE_2D, image); - if(cap_kms->params.egl->glGetError() != 0 || cap_kms->params.egl->eglGetError() != EGL_SUCCESS) { - // TODO: Get the error properly - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_tick: failed to bind egl image to gl texture, error: %d\n", cap_kms->params.egl->eglGetError()); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - cap_kms->params.egl->eglDestroyImage(cap_kms->params.egl->egl_display, image); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - return; - } - - cap_kms->params.egl->eglDestroyImage(cap_kms->params.egl->egl_display, image); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - } - - gsr_color_conversion_params color_conversion_params = {0}; - color_conversion_params.egl = cap_kms->params.egl; - color_conversion_params.source_color = GSR_SOURCE_COLOR_RGB; - color_conversion_params.destination_color = GSR_DESTINATION_COLOR_NV12; - - color_conversion_params.destination_textures[0] = cap_kms->target_textures[0]; - color_conversion_params.destination_textures[1] = cap_kms->target_textures[1]; - color_conversion_params.num_destination_textures = 2; - - if(gsr_color_conversion_init(&cap_kms->color_conversion, &color_conversion_params) != 0) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_tick: failed to create color conversion\n"); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - } else { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_tick: unexpected fourcc %u for output drm fd, expected nv12\n", cap_kms->prime.fourcc); - cap_kms->should_stop = true; - cap_kms->stop_is_error = true; - return; - } - } -} - -static bool gsr_capture_kms_vaapi_should_stop(gsr_capture *cap, bool *err) { - gsr_capture_kms_vaapi *cap_kms = cap->priv; - if(cap_kms->should_stop) { - if(err) - *err = cap_kms->stop_is_error; - return true; - } - - if(err) - *err = false; - return false; -} - -/* Prefer non combined planes */ -static gsr_kms_response_fd* find_drm_by_connector_id(gsr_kms_response *kms_response, uint32_t connector_id) { - int index_combined = -1; - for(int i = 0; i < kms_response->num_fds; ++i) { - if(kms_response->fds[i].connector_id == connector_id && !kms_response->fds[i].is_cursor) { - if(kms_response->fds[i].is_combined_plane) - index_combined = i; - else - return &kms_response->fds[i]; - } - } - - if(index_combined != -1) - return &kms_response->fds[index_combined]; - else - return NULL; -} - -static gsr_kms_response_fd* find_first_combined_drm(gsr_kms_response *kms_response) { - for(int i = 0; i < kms_response->num_fds; ++i) { - if(kms_response->fds[i].is_combined_plane && !kms_response->fds[i].is_cursor) - return &kms_response->fds[i]; - } - return NULL; -} - -static gsr_kms_response_fd* find_largest_drm(gsr_kms_response *kms_response) { - if(kms_response->num_fds == 0) - return NULL; - - int64_t largest_size = 0; - gsr_kms_response_fd *largest_drm = &kms_response->fds[0]; - for(int i = 0; i < kms_response->num_fds; ++i) { - const int64_t size = (int64_t)kms_response->fds[i].width * (int64_t)kms_response->fds[i].height; - if(size > largest_size && !kms_response->fds[i].is_cursor) { - largest_size = size; - largest_drm = &kms_response->fds[i]; - } - } - return largest_drm; -} - -static gsr_kms_response_fd* find_cursor_drm(gsr_kms_response *kms_response) { - for(int i = 0; i < kms_response->num_fds; ++i) { - if(kms_response->fds[i].is_cursor) - return &kms_response->fds[i]; - } - return NULL; -} - -static int gsr_capture_kms_vaapi_capture(gsr_capture *cap, AVFrame *frame) { - (void)frame; - gsr_capture_kms_vaapi *cap_kms = cap->priv; - - for(int i = 0; i < cap_kms->kms_response.num_fds; ++i) { - if(cap_kms->kms_response.fds[i].fd > 0) - close(cap_kms->kms_response.fds[i].fd); - cap_kms->kms_response.fds[i].fd = 0; - } - cap_kms->kms_response.num_fds = 0; - - gsr_kms_response_fd *drm_fd = NULL; - gsr_kms_response_fd *cursor_drm_fd = NULL; - bool capture_is_combined_plane = false; - if(cap_kms->using_wayland_capture) { - gsr_egl_update(cap_kms->params.egl); - cap_kms->wayland_kms_data.fd = cap_kms->params.egl->fd; - cap_kms->wayland_kms_data.width = cap_kms->params.egl->width; - cap_kms->wayland_kms_data.height = cap_kms->params.egl->height; - cap_kms->wayland_kms_data.pitch = cap_kms->params.egl->pitch; - cap_kms->wayland_kms_data.offset = cap_kms->params.egl->offset; - cap_kms->wayland_kms_data.pixel_format = cap_kms->params.egl->pixel_format; - cap_kms->wayland_kms_data.modifier = cap_kms->params.egl->modifier; - cap_kms->wayland_kms_data.connector_id = 0; - cap_kms->wayland_kms_data.is_combined_plane = false; - cap_kms->wayland_kms_data.is_cursor = false; - cap_kms->wayland_kms_data.x = cap_kms->wayland_kms_data.x; // TODO: Use these - cap_kms->wayland_kms_data.y = cap_kms->wayland_kms_data.y; - cap_kms->wayland_kms_data.src_w = cap_kms->wayland_kms_data.width; - cap_kms->wayland_kms_data.src_h = cap_kms->wayland_kms_data.height; - - cap_kms->capture_pos.x = cap_kms->wayland_kms_data.x; - cap_kms->capture_pos.y = cap_kms->wayland_kms_data.y; - - if(cap_kms->wayland_kms_data.fd <= 0) - return -1; - - drm_fd = &cap_kms->wayland_kms_data; - } else { - if(gsr_kms_client_get_kms(&cap_kms->kms_client, &cap_kms->kms_response) != 0) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_capture: failed to get kms, error: %d (%s)\n", cap_kms->kms_response.result, cap_kms->kms_response.err_msg); - return -1; - } - - if(cap_kms->kms_response.num_fds == 0) { - static bool error_shown = false; - if(!error_shown) { - error_shown = true; - fprintf(stderr, "gsr error: no drm found, capture will fail\n"); - } - return -1; - } - - for(int i = 0; i < cap_kms->monitor_id.num_connector_ids; ++i) { - drm_fd = find_drm_by_connector_id(&cap_kms->kms_response, cap_kms->monitor_id.connector_ids[i]); - if(drm_fd) - break; - } - - // Will never happen on wayland unless the target monitor has been disconnected - if(!drm_fd) { - drm_fd = find_first_combined_drm(&cap_kms->kms_response); - if(!drm_fd) - drm_fd = find_largest_drm(&cap_kms->kms_response); - capture_is_combined_plane = true; - } - - cursor_drm_fd = find_cursor_drm(&cap_kms->kms_response); - } - - if(!drm_fd) - return -1; - - if(!capture_is_combined_plane && cursor_drm_fd && cursor_drm_fd->connector_id != drm_fd->connector_id) - cursor_drm_fd = NULL; - - // TODO: This causes a crash sometimes on steam deck, why? is it a driver bug? a vaapi pure version doesn't cause a crash. - // Even ffmpeg kmsgrab causes this crash. The error is: - // amdgpu: Failed to allocate a buffer: - // amdgpu: size : 28508160 bytes - // amdgpu: alignment : 2097152 bytes - // amdgpu: domains : 4 - // amdgpu: flags : 4 - // amdgpu: Failed to allocate a buffer: - // amdgpu: size : 28508160 bytes - // amdgpu: alignment : 2097152 bytes - // amdgpu: domains : 4 - // amdgpu: flags : 4 - // EE ../jupiter-mesa/src/gallium/drivers/radeonsi/radeon_vcn_enc.c:516 radeon_create_encoder UVD - Can't create CPB buffer. - // [hevc_vaapi @ 0x55ea72b09840] Failed to upload encode parameters: 2 (resource allocation failed). - // [hevc_vaapi @ 0x55ea72b09840] Encode failed: -5. - // Error: avcodec_send_frame failed, error: Input/output error - // Assertion pic->display_order == pic->encode_order failed at libavcodec/vaapi_encode_h265.c:765 - // kms server info: kms client shutdown, shutting down the server - const intptr_t img_attr[] = { - EGL_LINUX_DRM_FOURCC_EXT, drm_fd->pixel_format, - EGL_WIDTH, drm_fd->width, - EGL_HEIGHT, drm_fd->height, - EGL_DMA_BUF_PLANE0_FD_EXT, drm_fd->fd, - EGL_DMA_BUF_PLANE0_OFFSET_EXT, drm_fd->offset, - EGL_DMA_BUF_PLANE0_PITCH_EXT, drm_fd->pitch, - // TODO: - //EGL_DMA_BUF_PLANE0_MODIFIER_LO_EXT, drm_fd->modifier & 0xFFFFFFFFULL, - //EGL_DMA_BUF_PLANE0_MODIFIER_HI_EXT, drm_fd->modifier >> 32ULL, - EGL_NONE - }; - - EGLImage image = cap_kms->params.egl->eglCreateImage(cap_kms->params.egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, cap_kms->input_texture); - cap_kms->params.egl->glEGLImageTargetTexture2DOES(GL_TEXTURE_2D, image); - cap_kms->params.egl->eglDestroyImage(cap_kms->params.egl->egl_display, image); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - vec2i capture_pos = cap_kms->capture_pos; - - if(cap_kms->using_wayland_capture) { - gsr_color_conversion_draw(&cap_kms->color_conversion, cap_kms->input_texture, - (vec2i){0, 0}, cap_kms->capture_size, - (vec2i){0, 0}, cap_kms->capture_size, - 0.0f); - } else { - if(!capture_is_combined_plane) - capture_pos = (vec2i){drm_fd->x, drm_fd->y}; - - float texture_rotation = 0.0f; - - gsr_color_conversion_draw(&cap_kms->color_conversion, cap_kms->input_texture, - (vec2i){0, 0}, cap_kms->capture_size, - capture_pos, cap_kms->capture_size, - texture_rotation); - - if(cursor_drm_fd) { - const intptr_t img_attr_cursor[] = { - EGL_LINUX_DRM_FOURCC_EXT, cursor_drm_fd->pixel_format, - EGL_WIDTH, cursor_drm_fd->width, - EGL_HEIGHT, cursor_drm_fd->height, - EGL_DMA_BUF_PLANE0_FD_EXT, cursor_drm_fd->fd, - EGL_DMA_BUF_PLANE0_OFFSET_EXT, cursor_drm_fd->offset, - EGL_DMA_BUF_PLANE0_PITCH_EXT, cursor_drm_fd->pitch, - EGL_DMA_BUF_PLANE0_MODIFIER_LO_EXT, cursor_drm_fd->modifier & 0xFFFFFFFFULL, - EGL_DMA_BUF_PLANE0_MODIFIER_HI_EXT, cursor_drm_fd->modifier >> 32ULL, - EGL_NONE - }; - - EGLImage cursor_image = cap_kms->params.egl->eglCreateImage(cap_kms->params.egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr_cursor); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, cap_kms->cursor_texture); - cap_kms->params.egl->glEGLImageTargetTexture2DOES(GL_TEXTURE_2D, cursor_image); - cap_kms->params.egl->eglDestroyImage(cap_kms->params.egl->egl_display, cursor_image); - cap_kms->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - vec2i cursor_size = {cursor_drm_fd->width, cursor_drm_fd->height}; - gsr_color_conversion_draw(&cap_kms->color_conversion, cap_kms->cursor_texture, - (vec2i){cursor_drm_fd->x, cursor_drm_fd->y}, cursor_size, - (vec2i){0, 0}, cursor_size, - 0.0f); - } - } - - cap_kms->params.egl->eglSwapBuffers(cap_kms->params.egl->egl_display, cap_kms->params.egl->egl_surface); - - return 0; -} - -static void gsr_capture_kms_vaapi_capture_end(gsr_capture *cap, AVFrame *frame) { - (void)frame; - gsr_capture_kms_vaapi *cap_kms = cap->priv; - - gsr_egl_cleanup_frame(cap_kms->params.egl); - - for(int i = 0; i < cap_kms->kms_response.num_fds; ++i) { - if(cap_kms->kms_response.fds[i].fd > 0) - close(cap_kms->kms_response.fds[i].fd); - cap_kms->kms_response.fds[i].fd = 0; - } - cap_kms->kms_response.num_fds = 0; -} - -static void gsr_capture_kms_vaapi_stop(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_kms_vaapi *cap_kms = cap->priv; - - gsr_color_conversion_deinit(&cap_kms->color_conversion); - - for(uint32_t i = 0; i < cap_kms->prime.num_objects; ++i) { - if(cap_kms->prime.objects[i].fd > 0) { - close(cap_kms->prime.objects[i].fd); - cap_kms->prime.objects[i].fd = 0; - } - } - - if(cap_kms->params.egl->egl_context) { - if(cap_kms->input_texture) { - cap_kms->params.egl->glDeleteTextures(1, &cap_kms->input_texture); - cap_kms->input_texture = 0; - } - - if(cap_kms->cursor_texture) { - cap_kms->params.egl->glDeleteTextures(1, &cap_kms->cursor_texture); - cap_kms->cursor_texture = 0; - } - - cap_kms->params.egl->glDeleteTextures(2, cap_kms->target_textures); - cap_kms->target_textures[0] = 0; - cap_kms->target_textures[1] = 0; - } - - for(int i = 0; i < cap_kms->kms_response.num_fds; ++i) { - if(cap_kms->kms_response.fds[i].fd > 0) - close(cap_kms->kms_response.fds[i].fd); - cap_kms->kms_response.fds[i].fd = 0; - } - cap_kms->kms_response.num_fds = 0; - - if(video_codec_context->hw_device_ctx) - av_buffer_unref(&video_codec_context->hw_device_ctx); - if(video_codec_context->hw_frames_ctx) - av_buffer_unref(&video_codec_context->hw_frames_ctx); - - gsr_kms_client_deinit(&cap_kms->kms_client); -} - -static void gsr_capture_kms_vaapi_destroy(gsr_capture *cap, AVCodecContext *video_codec_context) { - (void)video_codec_context; - gsr_capture_kms_vaapi *cap_kms = cap->priv; - if(cap->priv) { - gsr_capture_kms_vaapi_stop(cap, video_codec_context); - free((void*)cap_kms->params.display_to_capture); - cap_kms->params.display_to_capture = NULL; - free(cap->priv); - cap->priv = NULL; - } - free(cap); -} - -gsr_capture* gsr_capture_kms_vaapi_create(const gsr_capture_kms_vaapi_params *params) { - if(!params) { - fprintf(stderr, "gsr error: gsr_capture_kms_vaapi_create params is NULL\n"); - return NULL; - } - - gsr_capture *cap = calloc(1, sizeof(gsr_capture)); - if(!cap) - return NULL; - - gsr_capture_kms_vaapi *cap_kms = calloc(1, sizeof(gsr_capture_kms_vaapi)); - if(!cap_kms) { - free(cap); - return NULL; - } - - const char *display_to_capture = strdup(params->display_to_capture); - if(!display_to_capture) { - /* TODO XCloseDisplay */ - free(cap); - free(cap_kms); - return NULL; - } - - cap_kms->params = *params; - cap_kms->params.display_to_capture = display_to_capture; - - *cap = (gsr_capture) { - .start = gsr_capture_kms_vaapi_start, - .tick = gsr_capture_kms_vaapi_tick, - .should_stop = gsr_capture_kms_vaapi_should_stop, - .capture = gsr_capture_kms_vaapi_capture, - .capture_end = gsr_capture_kms_vaapi_capture_end, - .destroy = gsr_capture_kms_vaapi_destroy, - .priv = cap_kms - }; - - return cap; -} diff --git a/src/capture/nvfbc.c b/src/capture/nvfbc.c index 5b62310..b92bd41 100644 --- a/src/capture/nvfbc.c +++ b/src/capture/nvfbc.c @@ -1,16 +1,18 @@ #include "../../include/capture/nvfbc.h" #include "../../external/NvFBC.h" -#include "../../include/cuda.h" +#include "../../include/egl.h" +#include "../../include/utils.h" +#include "../../include/color_conversion.h" +#include "../../include/window/window.h" + #include <dlfcn.h> #include <stdlib.h> #include <string.h> #include <stdio.h> +#include <math.h> +#include <assert.h> + #include <X11/Xlib.h> -#include <libavutil/hwcontext.h> -#include <libavutil/hwcontext_cuda.h> -#include <libavutil/frame.h> -#include <libavutil/version.h> -#include <libavcodec/avcodec.h> typedef struct { gsr_capture_nvfbc_params params; @@ -22,16 +24,16 @@ typedef struct { bool fbc_handle_created; bool capture_session_created; - gsr_cuda cuda; - bool frame_initialized; -} gsr_capture_nvfbc; + NVFBC_TOGL_SETUP_PARAMS setup_params; -#if defined(_WIN64) || defined(__LP64__) -typedef unsigned long long CUdeviceptr_v2; -#else -typedef unsigned int CUdeviceptr_v2; -#endif -typedef CUdeviceptr_v2 CUdeviceptr; + bool supports_direct_cursor; + uint32_t width, height; + NVFBC_TRACKING_TYPE tracking_type; + uint32_t output_id; + uint32_t tracking_width, tracking_height; + bool nvfbc_needs_recreate; + double nvfbc_dead_start; +} gsr_capture_nvfbc; static int max_int(int a, int b) { return a > b ? a : b; @@ -100,7 +102,7 @@ static void set_func_ptr(void **dst, void *src) { } static bool gsr_capture_nvfbc_load_library(gsr_capture *cap) { - gsr_capture_nvfbc *cap_nvfbc = cap->priv; + gsr_capture_nvfbc *self = cap->priv; dlerror(); /* clear */ void *lib = dlopen("libnvidia-fbc.so.1", RTLD_LAZY); @@ -109,157 +111,84 @@ static bool gsr_capture_nvfbc_load_library(gsr_capture *cap) { return false; } - set_func_ptr((void**)&cap_nvfbc->nv_fbc_create_instance, dlsym(lib, "NvFBCCreateInstance")); - if(!cap_nvfbc->nv_fbc_create_instance) { + set_func_ptr((void**)&self->nv_fbc_create_instance, dlsym(lib, "NvFBCCreateInstance")); + if(!self->nv_fbc_create_instance) { fprintf(stderr, "gsr error: unable to resolve symbol 'NvFBCCreateInstance'\n"); dlclose(lib); return false; } - memset(&cap_nvfbc->nv_fbc_function_list, 0, sizeof(cap_nvfbc->nv_fbc_function_list)); - cap_nvfbc->nv_fbc_function_list.dwVersion = NVFBC_VERSION; - NVFBCSTATUS status = cap_nvfbc->nv_fbc_create_instance(&cap_nvfbc->nv_fbc_function_list); + memset(&self->nv_fbc_function_list, 0, sizeof(self->nv_fbc_function_list)); + self->nv_fbc_function_list.dwVersion = NVFBC_VERSION; + NVFBCSTATUS status = self->nv_fbc_create_instance(&self->nv_fbc_function_list); if(status != NVFBC_SUCCESS) { fprintf(stderr, "gsr error: failed to create NvFBC instance (status: %d)\n", status); dlclose(lib); return false; } - cap_nvfbc->library = lib; + self->library = lib; return true; } -#if LIBAVUTIL_VERSION_MAJOR < 57 -static AVBufferRef* dummy_hw_frame_init(int size) { - return av_buffer_alloc(size); -} -#else -static AVBufferRef* dummy_hw_frame_init(size_t size) { - return av_buffer_alloc(size); -} -#endif - -static bool ffmpeg_create_cuda_contexts(gsr_capture_nvfbc *cap_nvfbc, AVCodecContext *video_codec_context) { - AVBufferRef *device_ctx = av_hwdevice_ctx_alloc(AV_HWDEVICE_TYPE_CUDA); - if(!device_ctx) { - fprintf(stderr, "gsr error: cuda_create_codec_context failed: failed to create hardware device context\n"); - return false; - } - - AVHWDeviceContext *hw_device_context = (AVHWDeviceContext*)device_ctx->data; - AVCUDADeviceContext *cuda_device_context = (AVCUDADeviceContext*)hw_device_context->hwctx; - cuda_device_context->cuda_ctx = cap_nvfbc->cuda.cu_ctx; - if(av_hwdevice_ctx_init(device_ctx) < 0) { - fprintf(stderr, "gsr error: cuda_create_codec_context failed: failed to create hardware device context\n"); - av_buffer_unref(&device_ctx); - return false; - } - - AVBufferRef *frame_context = av_hwframe_ctx_alloc(device_ctx); - if(!frame_context) { - fprintf(stderr, "gsr error: cuda_create_codec_context failed: failed to create hwframe context\n"); - av_buffer_unref(&device_ctx); - return false; - } - - AVHWFramesContext *hw_frame_context = (AVHWFramesContext*)frame_context->data; - hw_frame_context->width = video_codec_context->width; - hw_frame_context->height = video_codec_context->height; - hw_frame_context->sw_format = AV_PIX_FMT_BGR0; - hw_frame_context->format = video_codec_context->pix_fmt; - hw_frame_context->device_ref = device_ctx; - hw_frame_context->device_ctx = (AVHWDeviceContext*)device_ctx->data; - - hw_frame_context->pool = av_buffer_pool_init(1, dummy_hw_frame_init); - hw_frame_context->initial_pool_size = 1; - - if (av_hwframe_ctx_init(frame_context) < 0) { - fprintf(stderr, "gsr error: cuda_create_codec_context failed: failed to initialize hardware frame context " - "(note: ffmpeg version needs to be > 4.0)\n"); - av_buffer_unref(&device_ctx); - //av_buffer_unref(&frame_context); - return false; +static void gsr_capture_nvfbc_destroy_session(gsr_capture_nvfbc *self) { + if(self->fbc_handle_created && self->capture_session_created) { + NVFBC_DESTROY_CAPTURE_SESSION_PARAMS destroy_capture_params; + memset(&destroy_capture_params, 0, sizeof(destroy_capture_params)); + destroy_capture_params.dwVersion = NVFBC_DESTROY_CAPTURE_SESSION_PARAMS_VER; + self->nv_fbc_function_list.nvFBCDestroyCaptureSession(self->nv_fbc_handle, &destroy_capture_params); + self->capture_session_created = false; } - - video_codec_context->hw_device_ctx = av_buffer_ref(device_ctx); - video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); - return true; } -static int gsr_capture_nvfbc_start(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_nvfbc *cap_nvfbc = cap->priv; - if(!gsr_cuda_load(&cap_nvfbc->cuda, cap_nvfbc->params.dpy, cap_nvfbc->params.overclock)) - return -1; - - if(!gsr_capture_nvfbc_load_library(cap)) { - gsr_cuda_unload(&cap_nvfbc->cuda); - return -1; +static void gsr_capture_nvfbc_destroy_handle(gsr_capture_nvfbc *self) { + if(self->fbc_handle_created) { + NVFBC_DESTROY_HANDLE_PARAMS destroy_params; + memset(&destroy_params, 0, sizeof(destroy_params)); + destroy_params.dwVersion = NVFBC_DESTROY_HANDLE_PARAMS_VER; + self->nv_fbc_function_list.nvFBCDestroyHandle(self->nv_fbc_handle, &destroy_params); + self->fbc_handle_created = false; + self->nv_fbc_handle = 0; } +} - const uint32_t x = max_int(cap_nvfbc->params.pos.x, 0); - const uint32_t y = max_int(cap_nvfbc->params.pos.y, 0); - const uint32_t width = max_int(cap_nvfbc->params.size.x, 0); - const uint32_t height = max_int(cap_nvfbc->params.size.y, 0); - - const bool capture_region = (x > 0 || y > 0 || width > 0 || height > 0); - - bool supports_direct_cursor = false; - bool direct_capture = cap_nvfbc->params.direct_capture; - int driver_major_version = 0; - int driver_minor_version = 0; - if(direct_capture && get_driver_version(&driver_major_version, &driver_minor_version)) { - fprintf(stderr, "Info: detected nvidia version: %d.%d\n", driver_major_version, driver_minor_version); - - // TODO: - if(version_at_least(driver_major_version, driver_minor_version, 515, 57) && version_less_than(driver_major_version, driver_minor_version, 520, 56)) { - direct_capture = false; - fprintf(stderr, "Warning: \"screen-direct\" has temporary been disabled as it causes stuttering with driver versions >= 515.57 and < 520.56. Please update your driver if possible. Capturing \"screen\" instead.\n"); - } - - // TODO: - // Cursor capture disabled because moving the cursor doesn't update capture rate to monitor hz and instead captures at 10-30 hz - /* - if(direct_capture) { - if(version_at_least(driver_major_version, driver_minor_version, 515, 57)) - supports_direct_cursor = true; - else - fprintf(stderr, "Info: capturing \"screen-direct\" but driver version appears to be less than 515.57. Disabling capture of cursor. Please update your driver if you want to capture your cursor or record \"screen\" instead.\n"); - } - */ - } +static void gsr_capture_nvfbc_destroy_session_and_handle(gsr_capture_nvfbc *self) { + gsr_capture_nvfbc_destroy_session(self); + gsr_capture_nvfbc_destroy_handle(self); +} +static int gsr_capture_nvfbc_setup_handle(gsr_capture_nvfbc *self) { NVFBCSTATUS status; - NVFBC_TRACKING_TYPE tracking_type; - uint32_t output_id = 0; - cap_nvfbc->fbc_handle_created = false; - cap_nvfbc->capture_session_created = false; NVFBC_CREATE_HANDLE_PARAMS create_params; memset(&create_params, 0, sizeof(create_params)); create_params.dwVersion = NVFBC_CREATE_HANDLE_PARAMS_VER; + create_params.bExternallyManagedContext = NVFBC_TRUE; + create_params.glxCtx = self->params.egl->glx_context; + create_params.glxFBConfig = self->params.egl->glx_fb_config; - status = cap_nvfbc->nv_fbc_function_list.nvFBCCreateHandle(&cap_nvfbc->nv_fbc_handle, &create_params); + status = self->nv_fbc_function_list.nvFBCCreateHandle(&self->nv_fbc_handle, &create_params); if(status != NVFBC_SUCCESS) { // Reverse engineering for interoperability const uint8_t enable_key[] = { 0xac, 0x10, 0xc9, 0x2e, 0xa5, 0xe6, 0x87, 0x4f, 0x8f, 0x4b, 0xf4, 0x61, 0xf8, 0x56, 0x27, 0xe9 }; create_params.privateData = enable_key; create_params.privateDataSize = 16; - status = cap_nvfbc->nv_fbc_function_list.nvFBCCreateHandle(&cap_nvfbc->nv_fbc_handle, &create_params); + status = self->nv_fbc_function_list.nvFBCCreateHandle(&self->nv_fbc_handle, &create_params); if(status != NVFBC_SUCCESS) { - fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: %s\n", cap_nvfbc->nv_fbc_function_list.nvFBCGetLastErrorStr(cap_nvfbc->nv_fbc_handle)); + fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: %s\n", self->nv_fbc_function_list.nvFBCGetLastErrorStr(self->nv_fbc_handle)); goto error_cleanup; } } - cap_nvfbc->fbc_handle_created = true; + self->fbc_handle_created = true; NVFBC_GET_STATUS_PARAMS status_params; memset(&status_params, 0, sizeof(status_params)); status_params.dwVersion = NVFBC_GET_STATUS_PARAMS_VER; - status = cap_nvfbc->nv_fbc_function_list.nvFBCGetStatus(cap_nvfbc->nv_fbc_handle, &status_params); + status = self->nv_fbc_function_list.nvFBCGetStatus(self->nv_fbc_handle, &status_params); if(status != NVFBC_SUCCESS) { - fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: %s\n", cap_nvfbc->nv_fbc_function_list.nvFBCGetLastErrorStr(cap_nvfbc->nv_fbc_handle)); + fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: %s\n", self->nv_fbc_function_list.nvFBCGetLastErrorStr(self->nv_fbc_handle)); goto error_cleanup; } @@ -268,10 +197,13 @@ static int gsr_capture_nvfbc_start(gsr_capture *cap, AVCodecContext *video_codec goto error_cleanup; } - uint32_t tracking_width = XWidthOfScreen(DefaultScreenOfDisplay(cap_nvfbc->params.dpy)); - uint32_t tracking_height = XHeightOfScreen(DefaultScreenOfDisplay(cap_nvfbc->params.dpy)); - tracking_type = strcmp(cap_nvfbc->params.display_to_capture, "screen") == 0 ? NVFBC_TRACKING_SCREEN : NVFBC_TRACKING_OUTPUT; - if(tracking_type == NVFBC_TRACKING_OUTPUT) { + assert(gsr_window_get_display_server(self->params.egl->window) == GSR_DISPLAY_SERVER_X11); + Display *display = gsr_window_get_display(self->params.egl->window); + + self->tracking_width = XWidthOfScreen(DefaultScreenOfDisplay(display)); + self->tracking_height = XHeightOfScreen(DefaultScreenOfDisplay(display)); + self->tracking_type = strcmp(self->params.display_to_capture, "screen") == 0 ? NVFBC_TRACKING_SCREEN : NVFBC_TRACKING_OUTPUT; + if(self->tracking_type == NVFBC_TRACKING_OUTPUT) { if(!status_params.bXRandRAvailable) { fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: the xrandr extension is not available\n"); goto error_cleanup; @@ -282,176 +214,201 @@ static int gsr_capture_nvfbc_start(gsr_capture *cap, AVCodecContext *video_codec goto error_cleanup; } - output_id = get_output_id_from_display_name(status_params.outputs, status_params.dwOutputNum, cap_nvfbc->params.display_to_capture, &tracking_width, &tracking_height); - if(output_id == 0) { - fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: display '%s' not found\n", cap_nvfbc->params.display_to_capture); + self->output_id = get_output_id_from_display_name(status_params.outputs, status_params.dwOutputNum, self->params.display_to_capture, &self->tracking_width, &self->tracking_height); + if(self->output_id == 0) { + fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: display '%s' not found\n", self->params.display_to_capture); goto error_cleanup; } } + self->width = self->tracking_width; + self->height = self->tracking_height; + return 0; + + error_cleanup: + gsr_capture_nvfbc_destroy_session_and_handle(self); + return -1; +} + +static int gsr_capture_nvfbc_setup_session(gsr_capture_nvfbc *self) { NVFBC_CREATE_CAPTURE_SESSION_PARAMS create_capture_params; memset(&create_capture_params, 0, sizeof(create_capture_params)); create_capture_params.dwVersion = NVFBC_CREATE_CAPTURE_SESSION_PARAMS_VER; - create_capture_params.eCaptureType = NVFBC_CAPTURE_SHARED_CUDA; - create_capture_params.bWithCursor = (!direct_capture || supports_direct_cursor) ? NVFBC_TRUE : NVFBC_FALSE; - if(capture_region) - create_capture_params.captureBox = (NVFBC_BOX){ x, y, width, height }; - create_capture_params.eTrackingType = tracking_type; - create_capture_params.dwSamplingRateMs = 1000u / ((uint32_t)cap_nvfbc->params.fps + 1); - create_capture_params.bAllowDirectCapture = direct_capture ? NVFBC_TRUE : NVFBC_FALSE; - create_capture_params.bPushModel = direct_capture ? NVFBC_TRUE : NVFBC_FALSE; - //create_capture_params.bDisableAutoModesetRecovery = true; // TODO: - if(tracking_type == NVFBC_TRACKING_OUTPUT) - create_capture_params.dwOutputId = output_id; - - status = cap_nvfbc->nv_fbc_function_list.nvFBCCreateCaptureSession(cap_nvfbc->nv_fbc_handle, &create_capture_params); + create_capture_params.eCaptureType = NVFBC_CAPTURE_TO_GL; + create_capture_params.bWithCursor = (!self->params.direct_capture || self->supports_direct_cursor) ? NVFBC_TRUE : NVFBC_FALSE; + if(!self->params.record_cursor) + create_capture_params.bWithCursor = false; + create_capture_params.eTrackingType = self->tracking_type; + create_capture_params.dwSamplingRateMs = (uint32_t)ceilf(1000.0f / (float)self->params.fps); + create_capture_params.bAllowDirectCapture = self->params.direct_capture ? NVFBC_TRUE : NVFBC_FALSE; + create_capture_params.bPushModel = self->params.direct_capture ? NVFBC_TRUE : NVFBC_FALSE; + create_capture_params.bDisableAutoModesetRecovery = true; + if(self->tracking_type == NVFBC_TRACKING_OUTPUT) + create_capture_params.dwOutputId = self->output_id; + + NVFBCSTATUS status = self->nv_fbc_function_list.nvFBCCreateCaptureSession(self->nv_fbc_handle, &create_capture_params); if(status != NVFBC_SUCCESS) { - fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: %s\n", cap_nvfbc->nv_fbc_function_list.nvFBCGetLastErrorStr(cap_nvfbc->nv_fbc_handle)); - goto error_cleanup; + fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: %s\n", self->nv_fbc_function_list.nvFBCGetLastErrorStr(self->nv_fbc_handle)); + return -1; } - cap_nvfbc->capture_session_created = true; + self->capture_session_created = true; - NVFBC_TOCUDA_SETUP_PARAMS setup_params; - memset(&setup_params, 0, sizeof(setup_params)); - setup_params.dwVersion = NVFBC_TOCUDA_SETUP_PARAMS_VER; - setup_params.eBufferFormat = NVFBC_BUFFER_FORMAT_BGRA; + memset(&self->setup_params, 0, sizeof(self->setup_params)); + self->setup_params.dwVersion = NVFBC_TOGL_SETUP_PARAMS_VER; + self->setup_params.eBufferFormat = NVFBC_BUFFER_FORMAT_BGRA; - status = cap_nvfbc->nv_fbc_function_list.nvFBCToCudaSetUp(cap_nvfbc->nv_fbc_handle, &setup_params); + status = self->nv_fbc_function_list.nvFBCToGLSetUp(self->nv_fbc_handle, &self->setup_params); if(status != NVFBC_SUCCESS) { - fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: %s\n", cap_nvfbc->nv_fbc_function_list.nvFBCGetLastErrorStr(cap_nvfbc->nv_fbc_handle)); - goto error_cleanup; + fprintf(stderr, "gsr error: gsr_capture_nvfbc_start failed: %s\n", self->nv_fbc_function_list.nvFBCGetLastErrorStr(self->nv_fbc_handle)); + gsr_capture_nvfbc_destroy_session(self); + return -1; } - if(capture_region) { - video_codec_context->width = width & ~1; - video_codec_context->height = height & ~1; - } else { - video_codec_context->width = tracking_width & ~1; - video_codec_context->height = tracking_height & ~1; + return 0; +} + +static void gsr_capture_nvfbc_stop(gsr_capture_nvfbc *self) { + gsr_capture_nvfbc_destroy_session_and_handle(self); + if(self->library) { + dlclose(self->library); + self->library = NULL; + } + if(self->params.display_to_capture) { + free((void*)self->params.display_to_capture); + self->params.display_to_capture = NULL; } +} - if(!ffmpeg_create_cuda_contexts(cap_nvfbc, video_codec_context)) - goto error_cleanup; +static int gsr_capture_nvfbc_start(gsr_capture *cap, gsr_capture_metadata *capture_metadata) { + gsr_capture_nvfbc *self = cap->priv; - return 0; + if(!gsr_capture_nvfbc_load_library(cap)) + return -1; - error_cleanup: - if(cap_nvfbc->fbc_handle_created) { - if(cap_nvfbc->capture_session_created) { - NVFBC_DESTROY_CAPTURE_SESSION_PARAMS destroy_capture_params; - memset(&destroy_capture_params, 0, sizeof(destroy_capture_params)); - destroy_capture_params.dwVersion = NVFBC_DESTROY_CAPTURE_SESSION_PARAMS_VER; - cap_nvfbc->nv_fbc_function_list.nvFBCDestroyCaptureSession(cap_nvfbc->nv_fbc_handle, &destroy_capture_params); - cap_nvfbc->capture_session_created = false; + self->supports_direct_cursor = false; + int driver_major_version = 0; + int driver_minor_version = 0; + if(self->params.direct_capture && get_driver_version(&driver_major_version, &driver_minor_version)) { + fprintf(stderr, "Info: detected nvidia version: %d.%d\n", driver_major_version, driver_minor_version); + + // TODO: + if(version_at_least(driver_major_version, driver_minor_version, 515, 57) && version_less_than(driver_major_version, driver_minor_version, 520, 56)) { + self->params.direct_capture = false; + fprintf(stderr, "Warning: \"screen-direct\" has temporary been disabled as it causes stuttering with driver versions >= 515.57 and < 520.56. Please update your driver if possible. Capturing \"screen\" instead.\n"); } - NVFBC_DESTROY_HANDLE_PARAMS destroy_params; - memset(&destroy_params, 0, sizeof(destroy_params)); - destroy_params.dwVersion = NVFBC_DESTROY_HANDLE_PARAMS_VER; - cap_nvfbc->nv_fbc_function_list.nvFBCDestroyHandle(cap_nvfbc->nv_fbc_handle, &destroy_params); - cap_nvfbc->fbc_handle_created = false; + // TODO: + // Cursor capture disabled because moving the cursor doesn't update capture rate to monitor hz and instead captures at 10-30 hz + /* + if(direct_capture) { + if(version_at_least(driver_major_version, driver_minor_version, 515, 57)) + self->supports_direct_cursor = true; + else + fprintf(stderr, "Info: capturing \"screen-direct\" but driver version appears to be less than 515.57. Disabling capture of cursor. Please update your driver if you want to capture your cursor or record \"screen\" instead.\n"); + } + */ } - if(video_codec_context->hw_device_ctx) - av_buffer_unref(&video_codec_context->hw_device_ctx); - if(video_codec_context->hw_frames_ctx) - av_buffer_unref(&video_codec_context->hw_frames_ctx); - - gsr_cuda_unload(&cap_nvfbc->cuda); - return -1; -} + if(gsr_capture_nvfbc_setup_handle(self) != 0) { + goto error_cleanup; + } -static void gsr_capture_nvfbc_destroy_session(gsr_capture *cap) { - gsr_capture_nvfbc *cap_nvfbc = cap->priv; + if(gsr_capture_nvfbc_setup_session(self) != 0) { + goto error_cleanup; + } - if(cap_nvfbc->fbc_handle_created) { - if(cap_nvfbc->capture_session_created) { - NVFBC_DESTROY_CAPTURE_SESSION_PARAMS destroy_capture_params; - memset(&destroy_capture_params, 0, sizeof(destroy_capture_params)); - destroy_capture_params.dwVersion = NVFBC_DESTROY_CAPTURE_SESSION_PARAMS_VER; - cap_nvfbc->nv_fbc_function_list.nvFBCDestroyCaptureSession(cap_nvfbc->nv_fbc_handle, &destroy_capture_params); - cap_nvfbc->capture_session_created = false; - } + capture_metadata->width = self->tracking_width; + capture_metadata->height = self->tracking_height; - NVFBC_DESTROY_HANDLE_PARAMS destroy_params; - memset(&destroy_params, 0, sizeof(destroy_params)); - destroy_params.dwVersion = NVFBC_DESTROY_HANDLE_PARAMS_VER; - cap_nvfbc->nv_fbc_function_list.nvFBCDestroyHandle(cap_nvfbc->nv_fbc_handle, &destroy_params); - cap_nvfbc->fbc_handle_created = false; + if(self->params.output_resolution.x > 0 && self->params.output_resolution.y > 0) { + self->params.output_resolution = scale_keep_aspect_ratio((vec2i){capture_metadata->width, capture_metadata->height}, self->params.output_resolution); + capture_metadata->width = self->params.output_resolution.x; + capture_metadata->height = self->params.output_resolution.y; + } else if(self->params.region_size.x > 0 && self->params.region_size.y > 0) { + capture_metadata->width = self->params.region_size.x; + capture_metadata->height = self->params.region_size.y; } - cap_nvfbc->nv_fbc_handle = 0; + return 0; + + error_cleanup: + gsr_capture_nvfbc_stop(self); + return -1; } -static void gsr_capture_nvfbc_tick(gsr_capture *cap, AVCodecContext *video_codec_context, AVFrame **frame) { - gsr_capture_nvfbc *cap_nvfbc = cap->priv; - if(!cap_nvfbc->frame_initialized && video_codec_context->hw_frames_ctx) { - cap_nvfbc->frame_initialized = true; - (*frame)->hw_frames_ctx = video_codec_context->hw_frames_ctx; - (*frame)->buf[0] = av_buffer_pool_get(((AVHWFramesContext*)video_codec_context->hw_frames_ctx->data)->pool); - (*frame)->extended_data = (*frame)->data; - (*frame)->color_range = video_codec_context->color_range; - (*frame)->color_primaries = video_codec_context->color_primaries; - (*frame)->color_trc = video_codec_context->color_trc; - (*frame)->colorspace = video_codec_context->colorspace; - (*frame)->chroma_location = video_codec_context->chroma_sample_location; +static int gsr_capture_nvfbc_capture(gsr_capture *cap, gsr_capture_metadata *capture_metadata, gsr_color_conversion *color_conversion) { + gsr_capture_nvfbc *self = cap->priv; + + const double nvfbc_recreate_retry_time_seconds = 1.0; + if(self->nvfbc_needs_recreate) { + const double now = clock_get_monotonic_seconds(); + if(now - self->nvfbc_dead_start >= nvfbc_recreate_retry_time_seconds) { + self->nvfbc_dead_start = now; + gsr_capture_nvfbc_destroy_session_and_handle(self); + + if(gsr_capture_nvfbc_setup_handle(self) != 0) { + fprintf(stderr, "gsr error: gsr_capture_nvfbc_capture failed to recreate nvfbc handle, trying again in %f second(s)\n", nvfbc_recreate_retry_time_seconds); + return -1; + } + + if(gsr_capture_nvfbc_setup_session(self) != 0) { + fprintf(stderr, "gsr error: gsr_capture_nvfbc_capture failed to recreate nvfbc session, trying again in %f second(s)\n", nvfbc_recreate_retry_time_seconds); + return -1; + } + + self->nvfbc_needs_recreate = false; + } else { + return 0; + } } -} -static int gsr_capture_nvfbc_capture(gsr_capture *cap, AVFrame *frame) { - gsr_capture_nvfbc *cap_nvfbc = cap->priv; + vec2i frame_size = (vec2i){self->width, self->height}; + const vec2i original_frame_size = frame_size; + if(self->params.region_size.x > 0 && self->params.region_size.y > 0) + frame_size = self->params.region_size; + + const bool is_scaled = self->params.output_resolution.x > 0 && self->params.output_resolution.y > 0; + vec2i output_size = is_scaled ? self->params.output_resolution : frame_size; + output_size = scale_keep_aspect_ratio(frame_size, output_size); - CUdeviceptr cu_device_ptr = 0; + const vec2i target_pos = { max_int(0, capture_metadata->width / 2 - output_size.x / 2), max_int(0, capture_metadata->height / 2 - output_size.y / 2) }; NVFBC_FRAME_GRAB_INFO frame_info; memset(&frame_info, 0, sizeof(frame_info)); - NVFBC_TOCUDA_GRAB_FRAME_PARAMS grab_params; + NVFBC_TOGL_GRAB_FRAME_PARAMS grab_params; memset(&grab_params, 0, sizeof(grab_params)); - grab_params.dwVersion = NVFBC_TOCUDA_GRAB_FRAME_PARAMS_VER; - grab_params.dwFlags = NVFBC_TOCUDA_GRAB_FLAGS_NOWAIT;/* | NVFBC_TOCUDA_GRAB_FLAGS_FORCE_REFRESH;*/ + grab_params.dwVersion = NVFBC_TOGL_GRAB_FRAME_PARAMS_VER; + grab_params.dwFlags = NVFBC_TOGL_GRAB_FLAGS_NOWAIT | NVFBC_TOGL_GRAB_FLAGS_FORCE_REFRESH; // TODO: Remove NVFBC_TOGL_GRAB_FLAGS_FORCE_REFRESH grab_params.pFrameGrabInfo = &frame_info; - grab_params.pCUDADeviceBuffer = &cu_device_ptr; grab_params.dwTimeoutMs = 0; - NVFBCSTATUS status = cap_nvfbc->nv_fbc_function_list.nvFBCToCudaGrabFrame(cap_nvfbc->nv_fbc_handle, &grab_params); + NVFBCSTATUS status = self->nv_fbc_function_list.nvFBCToGLGrabFrame(self->nv_fbc_handle, &grab_params); if(status != NVFBC_SUCCESS) { - fprintf(stderr, "gsr error: gsr_capture_nvfbc_capture failed: %s\n", cap_nvfbc->nv_fbc_function_list.nvFBCGetLastErrorStr(cap_nvfbc->nv_fbc_handle)); - return -1; + fprintf(stderr, "gsr error: gsr_capture_nvfbc_capture failed: %s (%d), recreating session after %f second(s)\n", self->nv_fbc_function_list.nvFBCGetLastErrorStr(self->nv_fbc_handle), status, nvfbc_recreate_retry_time_seconds); + self->nvfbc_needs_recreate = true; + self->nvfbc_dead_start = clock_get_monotonic_seconds(); + return 0; } - /* - *byte_size = frame_info.dwByteSize; + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); - TODO: Check bIsNewFrame - TODO: Check dwWidth and dwHeight and update size in video output in ffmpeg. This can happen when xrandr is used to change monitor resolution - */ + gsr_color_conversion_draw(color_conversion, self->setup_params.dwTextures[grab_params.dwTextureIndex], + target_pos, (vec2i){output_size.x, output_size.y}, + self->params.region_position, frame_size, original_frame_size, + GSR_ROT_0, GSR_SOURCE_COLOR_BGR, false, false); + + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); - frame->data[0] = (uint8_t*)cu_device_ptr; - //frame->data[1] = (uint8_t*)cu_device_ptr; - //frame->data[2] = (uint8_t*)cu_device_ptr; - frame->linesize[0] = frame->width * 4; - // TODO: Use these when outputting yuv444 by changing nvfbc color to YUV444P and sw_format to YUV444P - //frame->linesize[1] = frame->width * 1; - //frame->linesize[2] = frame->width * 1; return 0; } -static void gsr_capture_nvfbc_destroy(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_nvfbc *cap_nvfbc = cap->priv; - gsr_capture_nvfbc_destroy_session(cap); - if(video_codec_context->hw_device_ctx) - av_buffer_unref(&video_codec_context->hw_device_ctx); - if(video_codec_context->hw_frames_ctx) - av_buffer_unref(&video_codec_context->hw_frames_ctx); - if(cap_nvfbc) { - gsr_cuda_unload(&cap_nvfbc->cuda); - dlclose(cap_nvfbc->library); - free((void*)cap_nvfbc->params.display_to_capture); - cap_nvfbc->params.display_to_capture = NULL; - free(cap->priv); - cap->priv = NULL; - } +static void gsr_capture_nvfbc_destroy(gsr_capture *cap) { + gsr_capture_nvfbc *self = cap->priv; + gsr_capture_nvfbc_stop(self); + free(cap->priv); free(cap); } @@ -486,13 +443,13 @@ gsr_capture* gsr_capture_nvfbc_create(const gsr_capture_nvfbc_params *params) { cap_nvfbc->params = *params; cap_nvfbc->params.display_to_capture = display_to_capture; cap_nvfbc->params.fps = max_int(cap_nvfbc->params.fps, 1); - + *cap = (gsr_capture) { .start = gsr_capture_nvfbc_start, - .tick = gsr_capture_nvfbc_tick, + .tick = NULL, .should_stop = NULL, .capture = gsr_capture_nvfbc_capture, - .capture_end = NULL, + .uses_external_image = NULL, .destroy = gsr_capture_nvfbc_destroy, .priv = cap_nvfbc }; diff --git a/src/capture/portal.c b/src/capture/portal.c new file mode 100644 index 0000000..a5e62af --- /dev/null +++ b/src/capture/portal.c @@ -0,0 +1,430 @@ +#include "../../include/capture/portal.h" +#include "../../include/color_conversion.h" +#include "../../include/egl.h" +#include "../../include/utils.h" +#include "../../dbus/client/dbus_client.h" +#include "../../include/pipewire_video.h" + +#include <stdlib.h> +#include <stdio.h> +#include <unistd.h> +#include <limits.h> +#include <assert.h> + +typedef struct { + gsr_capture_portal_params params; + + gsr_texture_map texture_map; + + gsr_dbus_client dbus_client; + char session_handle[128]; + + gsr_pipewire_video pipewire; + vec2i capture_size; + gsr_pipewire_video_dmabuf_data dmabuf_data[GSR_PIPEWIRE_VIDEO_DMABUF_MAX_PLANES]; + int num_dmabuf_data; +} gsr_capture_portal; + +static void gsr_capture_portal_cleanup_plane_fds(gsr_capture_portal *self) { + for(int i = 0; i < self->num_dmabuf_data; ++i) { + if(self->dmabuf_data[i].fd > 0) { + close(self->dmabuf_data[i].fd); + self->dmabuf_data[i].fd = 0; + } + } + self->num_dmabuf_data = 0; +} + +static void gsr_capture_portal_stop(gsr_capture_portal *self) { + if(self->texture_map.texture_id) { + self->params.egl->glDeleteTextures(1, &self->texture_map.texture_id); + self->texture_map.texture_id = 0; + } + + if(self->texture_map.external_texture_id) { + self->params.egl->glDeleteTextures(1, &self->texture_map.external_texture_id); + self->texture_map.external_texture_id = 0; + } + + if(self->texture_map.cursor_texture_id) { + self->params.egl->glDeleteTextures(1, &self->texture_map.cursor_texture_id); + self->texture_map.cursor_texture_id = 0; + } + + gsr_capture_portal_cleanup_plane_fds(self); + gsr_pipewire_video_deinit(&self->pipewire); + gsr_dbus_client_deinit(&self->dbus_client); +} + +static void gsr_capture_portal_create_input_textures(gsr_capture_portal *self) { + const float border_color[4] = { 0.0f, 0.0f, 0.0f, 0.0f }; + + self->params.egl->glGenTextures(1, &self->texture_map.texture_id); + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->texture_map.texture_id); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, border_color); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + + self->params.egl->glGenTextures(1, &self->texture_map.external_texture_id); + self->params.egl->glBindTexture(GL_TEXTURE_EXTERNAL_OES, self->texture_map.external_texture_id); + self->params.egl->glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameterfv(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_BORDER_COLOR, border_color); + self->params.egl->glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + self->params.egl->glTexParameteri(GL_TEXTURE_EXTERNAL_OES, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + self->params.egl->glBindTexture(GL_TEXTURE_EXTERNAL_OES, 0); + + self->params.egl->glGenTextures(1, &self->texture_map.cursor_texture_id); + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->texture_map.cursor_texture_id); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, border_color); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); +} + +static void get_default_gpu_screen_recorder_restore_token_path(char *buffer, size_t buffer_size) { + const char *xdg_config_home = getenv("XDG_CONFIG_HOME"); + if(xdg_config_home) { + snprintf(buffer, buffer_size, "%s/gpu-screen-recorder/restore_token", xdg_config_home); + } else { + const char *home = getenv("HOME"); + if(!home) + home = "/tmp"; + snprintf(buffer, buffer_size, "%s/.config/gpu-screen-recorder/restore_token", home); + } +} + +static bool create_directory_to_file(const char *filepath) { + char dir[PATH_MAX]; + dir[0] = '\0'; + + const char *split = strrchr(filepath, '/'); + if(!split) /* Assuming it's the current directory (for example if filepath is "restore_token"), which doesn't need to be created */ + return true; + + snprintf(dir, sizeof(dir), "%.*s", (int)(split - filepath), filepath); + if(create_directory_recursive(dir) != 0) { + fprintf(stderr, "gsr warning: gsr_capture_portal_save_restore_token: failed to create directory (%s) for restore token\n", dir); + return false; + } + return true; +} + +static void gsr_capture_portal_save_restore_token(const char *restore_token, const char *portal_session_token_filepath) { + char restore_token_path[PATH_MAX]; + restore_token_path[0] = '\0'; + if(portal_session_token_filepath) + snprintf(restore_token_path, sizeof(restore_token_path), "%s", portal_session_token_filepath); + else + get_default_gpu_screen_recorder_restore_token_path(restore_token_path, sizeof(restore_token_path)); + + if(!create_directory_to_file(restore_token_path)) + return; + + FILE *f = fopen(restore_token_path, "wb"); + if(!f) { + fprintf(stderr, "gsr warning: gsr_capture_portal_save_restore_token: failed to create restore token file (%s)\n", restore_token_path); + return; + } + + const int restore_token_len = strlen(restore_token); + if((long)fwrite(restore_token, 1, restore_token_len, f) != restore_token_len) { + fprintf(stderr, "gsr warning: gsr_capture_portal_save_restore_token: failed to write restore token to file (%s)\n", restore_token_path); + fclose(f); + return; + } + + fprintf(stderr, "gsr info: gsr_capture_portal_save_restore_token: saved restore token to cache (%s)\n", restore_token); + fclose(f); +} + +static void gsr_capture_portal_get_restore_token_from_cache(char *buffer, size_t buffer_size, const char *portal_session_token_filepath) { + assert(buffer_size > 0); + buffer[0] = '\0'; + + char restore_token_path[PATH_MAX]; + restore_token_path[0] = '\0'; + if(portal_session_token_filepath) + snprintf(restore_token_path, sizeof(restore_token_path), "%s", portal_session_token_filepath); + else + get_default_gpu_screen_recorder_restore_token_path(restore_token_path, sizeof(restore_token_path)); + + FILE *f = fopen(restore_token_path, "rb"); + if(!f) { + fprintf(stderr, "gsr info: gsr_capture_portal_get_restore_token_from_cache: no restore token found in cache or failed to load (%s)\n", restore_token_path); + return; + } + + fseek(f, 0, SEEK_END); + long file_size = ftell(f); + fseek(f, 0, SEEK_SET); + + if(file_size > 0 && file_size < 1024 && file_size < (long)buffer_size && (long)fread(buffer, 1, file_size, f) != file_size) { + buffer[0] = '\0'; + fprintf(stderr, "gsr warning: gsr_capture_portal_get_restore_token_from_cache: failed to read restore token (%s)\n", restore_token_path); + fclose(f); + return; + } + + if(file_size > 0 && file_size < (long)buffer_size) + buffer[file_size] = '\0'; + + fprintf(stderr, "gsr info: gsr_capture_portal_get_restore_token_from_cache: read cached restore token (%s)\n", buffer); + fclose(f); +} + +static int gsr_capture_portal_setup_dbus(gsr_capture_portal *self, int *pipewire_fd, uint32_t *pipewire_node) { + *pipewire_fd = 0; + *pipewire_node = 0; + int response_status = 0; + + char restore_token[1024]; + restore_token[0] = '\0'; + if(self->params.restore_portal_session) + gsr_capture_portal_get_restore_token_from_cache(restore_token, sizeof(restore_token), self->params.portal_session_token_filepath); + + if(!gsr_dbus_client_init(&self->dbus_client, restore_token)) + return -1; + + fprintf(stderr, "gsr info: gsr_capture_portal_setup_dbus: CreateSession\n"); + response_status = gsr_dbus_client_screencast_create_session(&self->dbus_client, self->session_handle, sizeof(self->session_handle)); + if(response_status != 0) { + fprintf(stderr, "gsr error: gsr_capture_portal_setup_dbus: CreateSession failed\n"); + return response_status; + } + + fprintf(stderr, "gsr info: gsr_capture_portal_setup_dbus: SelectSources\n"); + response_status = gsr_dbus_client_screencast_select_sources(&self->dbus_client, self->session_handle, GSR_PORTAL_CAPTURE_TYPE_ALL, self->params.record_cursor ? GSR_PORTAL_CURSOR_MODE_EMBEDDED : GSR_PORTAL_CURSOR_MODE_HIDDEN); + if(response_status != 0) { + fprintf(stderr, "gsr error: gsr_capture_portal_setup_dbus: SelectSources failed\n"); + return response_status; + } + + fprintf(stderr, "gsr info: gsr_capture_portal_setup_dbus: Start\n"); + response_status = gsr_dbus_client_screencast_start(&self->dbus_client, self->session_handle, pipewire_node); + if(response_status != 0) { + fprintf(stderr, "gsr error: gsr_capture_portal_setup_dbus: Start failed\n"); + return response_status; + } + + const char *screencast_restore_token = gsr_dbus_client_screencast_get_restore_token(&self->dbus_client); + if(screencast_restore_token) + gsr_capture_portal_save_restore_token(screencast_restore_token, self->params.portal_session_token_filepath); + + fprintf(stderr, "gsr info: gsr_capture_portal_setup_dbus: OpenPipeWireRemote\n"); + if(!gsr_dbus_client_screencast_open_pipewire_remote(&self->dbus_client, self->session_handle, pipewire_fd)) { + fprintf(stderr, "gsr error: gsr_capture_portal_setup_dbus: OpenPipeWireRemote failed\n"); + return -1; + } + + fprintf(stderr, "gsr info: gsr_capture_portal_setup_dbus: desktop portal setup finished\n"); + return 0; +} + +static bool gsr_capture_portal_get_frame_dimensions(gsr_capture_portal *self) { + gsr_pipewire_video_region region = {0, 0, 0, 0}; + gsr_pipewire_video_region cursor_region = {0, 0, 0, 0}; + fprintf(stderr, "gsr info: gsr_capture_portal_start: waiting for pipewire negotiation\n"); + + const double start_time = clock_get_monotonic_seconds(); + while(clock_get_monotonic_seconds() - start_time < 5.0) { + bool uses_external_image = false; + uint32_t fourcc = 0; + uint64_t modifiers = 0; + if(gsr_pipewire_video_map_texture(&self->pipewire, self->texture_map, ®ion, &cursor_region, self->dmabuf_data, &self->num_dmabuf_data, &fourcc, &modifiers, &uses_external_image)) { + gsr_capture_portal_cleanup_plane_fds(self); + self->capture_size.x = region.width; + self->capture_size.y = region.height; + fprintf(stderr, "gsr info: gsr_capture_portal_start: pipewire negotiation finished\n"); + return true; + } + usleep(30 * 1000); /* 30 milliseconds */ + } + + fprintf(stderr, "gsr info: gsr_capture_portal_start: timed out waiting for pipewire negotiation (5 seconds)\n"); + return false; +} + +static int gsr_capture_portal_start(gsr_capture *cap, gsr_capture_metadata *capture_metadata) { + gsr_capture_portal *self = cap->priv; + + gsr_capture_portal_create_input_textures(self); + + int pipewire_fd = 0; + uint32_t pipewire_node = 0; + const int response_status = gsr_capture_portal_setup_dbus(self, &pipewire_fd, &pipewire_node); + if(response_status != 0) { + gsr_capture_portal_stop(self); + // Response status values: + // 0: Success, the request is carried out + // 1: The user cancelled the interaction + // 2: The user interaction was ended in some other way + // Response status value 2 happens usually if there was some kind of error in the desktop portal on the system + if(response_status == 2) { + fprintf(stderr, "gsr error: gsr_capture_portal_start: desktop portal capture failed. Either you Wayland compositor doesn't support desktop portal capture or it's incorrectly setup on your system\n"); + return 50; + } else if(response_status == 1) { + fprintf(stderr, "gsr error: gsr_capture_portal_start: desktop portal capture failed. It seems like desktop portal capture was canceled by the user.\n"); + return 60; + } else { + return -1; + } + } + + fprintf(stderr, "gsr info: gsr_capture_portal_start: setting up pipewire\n"); + /* TODO: support hdr when pipewire supports it */ + /* gsr_pipewire closes the pipewire fd, even on failure */ + if(!gsr_pipewire_video_init(&self->pipewire, pipewire_fd, pipewire_node, capture_metadata->fps, self->params.record_cursor, self->params.egl)) { + fprintf(stderr, "gsr error: gsr_capture_portal_start: failed to setup pipewire with fd: %d, node: %" PRIu32 "\n", pipewire_fd, pipewire_node); + gsr_capture_portal_stop(self); + return -1; + } + fprintf(stderr, "gsr info: gsr_capture_portal_start: pipewire setup finished\n"); + + if(!gsr_capture_portal_get_frame_dimensions(self)) { + gsr_capture_portal_stop(self); + return -1; + } + + if(self->params.output_resolution.x == 0 && self->params.output_resolution.y == 0) { + capture_metadata->width = self->capture_size.x; + capture_metadata->height = self->capture_size.y; + } else { + self->params.output_resolution = scale_keep_aspect_ratio(self->capture_size, self->params.output_resolution); + capture_metadata->width = self->params.output_resolution.x; + capture_metadata->height = self->params.output_resolution.y; + } + + return 0; +} + +static int max_int(int a, int b) { + return a > b ? a : b; +} + +static int gsr_capture_portal_capture(gsr_capture *cap, gsr_capture_metadata *capture_metadata, gsr_color_conversion *color_conversion) { + (void)color_conversion; + gsr_capture_portal *self = cap->priv; + + /* TODO: Handle formats other than RGB(a) */ + gsr_pipewire_video_region region = {0, 0, 0, 0}; + gsr_pipewire_video_region cursor_region = {0, 0, 0, 0}; + uint32_t pipewire_fourcc = 0; + uint64_t pipewire_modifiers = 0; + bool using_external_image = false; + if(gsr_pipewire_video_map_texture(&self->pipewire, self->texture_map, ®ion, &cursor_region, self->dmabuf_data, &self->num_dmabuf_data, &pipewire_fourcc, &pipewire_modifiers, &using_external_image)) { + if(region.width != self->capture_size.x || region.height != self->capture_size.y) { + self->capture_size.x = region.width; + self->capture_size.y = region.height; + gsr_color_conversion_clear(color_conversion); + } + } else { + return 0; + } + + const bool is_scaled = self->params.output_resolution.x > 0 && self->params.output_resolution.y > 0; + vec2i output_size = is_scaled ? self->params.output_resolution : self->capture_size; + output_size = scale_keep_aspect_ratio(self->capture_size, output_size); + + const vec2i target_pos = { max_int(0, capture_metadata->width / 2 - output_size.x / 2), max_int(0, capture_metadata->height / 2 - output_size.y / 2) }; + + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); + + // TODO: Handle region crop + + gsr_color_conversion_draw(color_conversion, using_external_image ? self->texture_map.external_texture_id : self->texture_map.texture_id, + target_pos, output_size, + (vec2i){region.x, region.y}, self->capture_size, self->capture_size, + GSR_ROT_0, GSR_SOURCE_COLOR_RGB, using_external_image, false); + + if(self->params.record_cursor && self->texture_map.cursor_texture_id > 0 && cursor_region.width > 0) { + const vec2d scale = { + self->capture_size.x == 0 ? 0 : (double)output_size.x / (double)self->capture_size.x, + self->capture_size.y == 0 ? 0 : (double)output_size.y / (double)self->capture_size.y + }; + + const vec2i cursor_pos = { + target_pos.x + (cursor_region.x * scale.x), + target_pos.y + (cursor_region.y * scale.y) + }; + + self->params.egl->glEnable(GL_SCISSOR_TEST); + self->params.egl->glScissor(target_pos.x, target_pos.y, output_size.x, output_size.y); + gsr_color_conversion_draw(color_conversion, self->texture_map.cursor_texture_id, + (vec2i){cursor_pos.x, cursor_pos.y}, (vec2i){cursor_region.width * scale.x, cursor_region.height * scale.y}, + (vec2i){0, 0}, (vec2i){cursor_region.width, cursor_region.height}, (vec2i){cursor_region.width, cursor_region.height}, + GSR_ROT_0, GSR_SOURCE_COLOR_RGB, false, true); + self->params.egl->glDisable(GL_SCISSOR_TEST); + } + + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); + + gsr_capture_portal_cleanup_plane_fds(self); + + return 0; +} + +static bool gsr_capture_portal_uses_external_image(gsr_capture *cap) { + (void)cap; + return true; +} + +static bool gsr_capture_portal_is_damaged(gsr_capture *cap) { + gsr_capture_portal *self = cap->priv; + return gsr_pipewire_video_is_damaged(&self->pipewire); +} + +static void gsr_capture_portal_clear_damage(gsr_capture *cap) { + gsr_capture_portal *self = cap->priv; + gsr_pipewire_video_clear_damage(&self->pipewire); +} + +static void gsr_capture_portal_destroy(gsr_capture *cap) { + gsr_capture_portal *self = cap->priv; + if(cap->priv) { + gsr_capture_portal_stop(self); + free(cap->priv); + cap->priv = NULL; + } + free(cap); +} + +gsr_capture* gsr_capture_portal_create(const gsr_capture_portal_params *params) { + if(!params) { + fprintf(stderr, "gsr error: gsr_capture_portal_create params is NULL\n"); + return NULL; + } + + gsr_capture *cap = calloc(1, sizeof(gsr_capture)); + if(!cap) + return NULL; + + gsr_capture_portal *cap_portal = calloc(1, sizeof(gsr_capture_portal)); + if(!cap_portal) { + free(cap); + return NULL; + } + + cap_portal->params = *params; + + *cap = (gsr_capture) { + .start = gsr_capture_portal_start, + .tick = NULL, + .should_stop = NULL, + .capture = gsr_capture_portal_capture, + .uses_external_image = gsr_capture_portal_uses_external_image, + .is_damaged = gsr_capture_portal_is_damaged, + .clear_damage = gsr_capture_portal_clear_damage, + .destroy = gsr_capture_portal_destroy, + .priv = cap_portal + }; + + return cap; +} diff --git a/src/capture/xcomposite.c b/src/capture/xcomposite.c new file mode 100644 index 0000000..db41f63 --- /dev/null +++ b/src/capture/xcomposite.c @@ -0,0 +1,338 @@ +#include "../../include/capture/xcomposite.h" +#include "../../include/window_texture.h" +#include "../../include/utils.h" +#include "../../include/cursor.h" +#include "../../include/color_conversion.h" +#include "../../include/window/window.h" + +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <assert.h> + +#include <X11/Xlib.h> + +typedef struct { + gsr_capture_xcomposite_params params; + Display *display; + + bool should_stop; + bool stop_is_error; + bool window_resized; + bool follow_focused_initialized; + bool init_new_window; + + Window window; + vec2i window_size; + vec2i texture_size; + double window_resize_timer; + + WindowTexture window_texture; + + Atom net_active_window_atom; + + gsr_cursor cursor; + + bool clear_background; +} gsr_capture_xcomposite; + +static void gsr_capture_xcomposite_stop(gsr_capture_xcomposite *self) { + window_texture_deinit(&self->window_texture); + gsr_cursor_deinit(&self->cursor); +} + +static int max_int(int a, int b) { + return a > b ? a : b; +} + +static Window get_focused_window(Display *display, Atom net_active_window_atom) { + Atom type; + int format = 0; + unsigned long num_items = 0; + unsigned long bytes_after = 0; + unsigned char *properties = NULL; + if(XGetWindowProperty(display, DefaultRootWindow(display), net_active_window_atom, 0, 1024, False, AnyPropertyType, &type, &format, &num_items, &bytes_after, &properties) == Success && properties) { + Window focused_window = *(unsigned long*)properties; + XFree(properties); + return focused_window; + } + return None; +} + +static int gsr_capture_xcomposite_start(gsr_capture *cap, gsr_capture_metadata *capture_metadata) { + gsr_capture_xcomposite *self = cap->priv; + + if(self->params.follow_focused) { + self->net_active_window_atom = XInternAtom(self->display, "_NET_ACTIVE_WINDOW", False); + if(!self->net_active_window_atom) { + fprintf(stderr, "gsr error: gsr_capture_xcomposite_start failed: failed to get _NET_ACTIVE_WINDOW atom\n"); + return -1; + } + self->window = get_focused_window(self->display, self->net_active_window_atom); + } else { + self->window = self->params.window; + } + + /* TODO: Do these in tick, and allow error if follow_focused */ + + XWindowAttributes attr; + if(!XGetWindowAttributes(self->display, self->window, &attr) && !self->params.follow_focused) { + fprintf(stderr, "gsr error: gsr_capture_xcomposite_start failed: invalid window id: %lu\n", self->window); + return -1; + } + + self->window_size.x = max_int(attr.width, 0); + self->window_size.y = max_int(attr.height, 0); + + if(self->params.follow_focused) + XSelectInput(self->display, DefaultRootWindow(self->display), PropertyChangeMask); + + // TODO: Get select and add these on top of it and then restore at the end. Also do the same in other xcomposite + XSelectInput(self->display, self->window, StructureNotifyMask | ExposureMask); + + if(window_texture_init(&self->window_texture, self->display, self->window, self->params.egl) != 0 && !self->params.follow_focused) { + fprintf(stderr, "gsr error: gsr_capture_xcomposite_start: failed to get window texture for window %ld\n", (long)self->window); + return -1; + } + + if(gsr_cursor_init(&self->cursor, self->params.egl, self->display) != 0) { + gsr_capture_xcomposite_stop(self); + return -1; + } + + self->texture_size.x = 0; + self->texture_size.y = 0; + + self->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&self->window_texture)); + self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &self->texture_size.x); + self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &self->texture_size.y); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + + if(self->params.output_resolution.x == 0 && self->params.output_resolution.y == 0) { + capture_metadata->width = self->texture_size.x; + capture_metadata->height = self->texture_size.y; + } else { + capture_metadata->width = self->params.output_resolution.x; + capture_metadata->height = self->params.output_resolution.y; + } + + self->window_resize_timer = clock_get_monotonic_seconds(); + return 0; +} + +static void gsr_capture_xcomposite_tick(gsr_capture *cap) { + gsr_capture_xcomposite *self = cap->priv; + + if(self->params.follow_focused && !self->follow_focused_initialized) { + self->init_new_window = true; + } + + if(self->init_new_window) { + self->init_new_window = false; + Window focused_window = get_focused_window(self->display, self->net_active_window_atom); + if(focused_window != self->window || !self->follow_focused_initialized) { + self->follow_focused_initialized = true; + XSelectInput(self->display, self->window, 0); + self->window = focused_window; + XSelectInput(self->display, self->window, StructureNotifyMask | ExposureMask); + + XWindowAttributes attr; + attr.width = 0; + attr.height = 0; + if(!XGetWindowAttributes(self->display, self->window, &attr)) + fprintf(stderr, "gsr error: gsr_capture_xcomposite_tick failed: invalid window id: %lu\n", self->window); + + self->window_size.x = max_int(attr.width, 0); + self->window_size.y = max_int(attr.height, 0); + + window_texture_deinit(&self->window_texture); + window_texture_init(&self->window_texture, self->display, self->window, self->params.egl); // TODO: Do not do the below window_texture_on_resize after this + + self->texture_size.x = 0; + self->texture_size.y = 0; + + self->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&self->window_texture)); + self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &self->texture_size.x); + self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &self->texture_size.y); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + + self->window_resized = false; + self->clear_background = true; + } + } + + const double window_resize_timeout = 1.0; // 1 second + if(self->window_resized && clock_get_monotonic_seconds() - self->window_resize_timer >= window_resize_timeout) { + self->window_resized = false; + + if(window_texture_on_resize(&self->window_texture) != 0) { + fprintf(stderr, "gsr error: gsr_capture_xcomposite_tick: window_texture_on_resize failed\n"); + //self->should_stop = true; + //self->stop_is_error = true; + return; + } + + self->texture_size.x = 0; + self->texture_size.y = 0; + + self->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&self->window_texture)); + self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &self->texture_size.x); + self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &self->texture_size.y); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + + self->clear_background = true; + } +} + +static void gsr_capture_xcomposite_on_event(gsr_capture *cap, gsr_egl *egl) { + gsr_capture_xcomposite *self = cap->priv; + XEvent *xev = gsr_window_get_event_data(egl->window); + switch(xev->type) { + case DestroyNotify: { + /* Window died (when not following focused window), so we stop recording */ + if(!self->params.follow_focused && xev->xdestroywindow.window == self->window) { + self->should_stop = true; + self->stop_is_error = false; + } + break; + } + case Expose: { + /* Requires window texture recreate */ + if(xev->xexpose.count == 0 && xev->xexpose.window == self->window) { + self->window_resize_timer = clock_get_monotonic_seconds(); + self->window_resized = true; + } + break; + } + case ConfigureNotify: { + /* Window resized */ + if(xev->xconfigure.window == self->window && (xev->xconfigure.width != self->window_size.x || xev->xconfigure.height != self->window_size.y)) { + self->window_size.x = max_int(xev->xconfigure.width, 0); + self->window_size.y = max_int(xev->xconfigure.height, 0); + self->window_resize_timer = clock_get_monotonic_seconds(); + self->window_resized = true; + } + break; + } + case PropertyNotify: { + /* Focused window changed */ + if(self->params.follow_focused && xev->xproperty.atom == self->net_active_window_atom) { + self->init_new_window = true; + } + break; + } + } + + gsr_cursor_on_event(&self->cursor, xev); +} + +static bool gsr_capture_xcomposite_should_stop(gsr_capture *cap, bool *err) { + gsr_capture_xcomposite *self = cap->priv; + if(self->should_stop) { + if(err) + *err = self->stop_is_error; + return true; + } + + if(err) + *err = false; + return false; +} + +static int gsr_capture_xcomposite_capture(gsr_capture *cap, gsr_capture_metadata *capture_metdata, gsr_color_conversion *color_conversion) { + gsr_capture_xcomposite *self = cap->priv; + + if(self->clear_background) { + self->clear_background = false; + gsr_color_conversion_clear(color_conversion); + } + + const bool is_scaled = self->params.output_resolution.x > 0 && self->params.output_resolution.y > 0; + vec2i output_size = is_scaled ? self->params.output_resolution : self->texture_size; + output_size = scale_keep_aspect_ratio(self->texture_size, output_size); + + const vec2i target_pos = { max_int(0, capture_metdata->width / 2 - output_size.x / 2), max_int(0, capture_metdata->height / 2 - output_size.y / 2) }; + + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); + + gsr_color_conversion_draw(color_conversion, window_texture_get_opengl_texture_id(&self->window_texture), + target_pos, output_size, + (vec2i){0, 0}, self->texture_size, self->texture_size, + GSR_ROT_0, GSR_SOURCE_COLOR_RGB, false, false); + + if(self->params.record_cursor && self->cursor.visible) { + const vec2d scale = { + self->texture_size.x == 0 ? 0 : (double)output_size.x / (double)self->texture_size.x, + self->texture_size.y == 0 ? 0 : (double)output_size.y / (double)self->texture_size.y + }; + + gsr_cursor_tick(&self->cursor, self->window); + + const vec2i cursor_pos = { + target_pos.x + (self->cursor.position.x - self->cursor.hotspot.x) * scale.x, + target_pos.y + (self->cursor.position.y - self->cursor.hotspot.y) * scale.y + }; + + if(cursor_pos.x < target_pos.x || cursor_pos.x + self->cursor.size.x > target_pos.x + output_size.x || cursor_pos.y < target_pos.y || cursor_pos.y + self->cursor.size.y > target_pos.y + output_size.y) + self->clear_background = true; + + gsr_color_conversion_draw(color_conversion, self->cursor.texture_id, + cursor_pos, (vec2i){self->cursor.size.x * scale.x, self->cursor.size.y * scale.y}, + (vec2i){0, 0}, self->cursor.size, self->cursor.size, + GSR_ROT_0, GSR_SOURCE_COLOR_RGB, false, true); + } + + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); + + return 0; +} + +static uint64_t gsr_capture_xcomposite_get_window_id(gsr_capture *cap) { + gsr_capture_xcomposite *self = cap->priv; + return self->window; +} + +static void gsr_capture_xcomposite_destroy(gsr_capture *cap) { + if(cap->priv) { + gsr_capture_xcomposite_stop(cap->priv); + free(cap->priv); + cap->priv = NULL; + } + free(cap); +} + +gsr_capture* gsr_capture_xcomposite_create(const gsr_capture_xcomposite_params *params) { + if(!params) { + fprintf(stderr, "gsr error: gsr_capture_xcomposite_create params is NULL\n"); + return NULL; + } + + gsr_capture *cap = calloc(1, sizeof(gsr_capture)); + if(!cap) + return NULL; + + gsr_capture_xcomposite *cap_xcomp = calloc(1, sizeof(gsr_capture_xcomposite)); + if(!cap_xcomp) { + free(cap); + return NULL; + } + + cap_xcomp->params = *params; + cap_xcomp->display = gsr_window_get_display(params->egl->window); + + *cap = (gsr_capture) { + .start = gsr_capture_xcomposite_start, + .on_event = gsr_capture_xcomposite_on_event, + .tick = gsr_capture_xcomposite_tick, + .should_stop = gsr_capture_xcomposite_should_stop, + .capture = gsr_capture_xcomposite_capture, + .uses_external_image = NULL, + .get_window_id = gsr_capture_xcomposite_get_window_id, + .destroy = gsr_capture_xcomposite_destroy, + .priv = cap_xcomp + }; + + return cap; +} diff --git a/src/capture/xcomposite_cuda.c b/src/capture/xcomposite_cuda.c deleted file mode 100644 index 7e65efa..0000000 --- a/src/capture/xcomposite_cuda.c +++ /dev/null @@ -1,512 +0,0 @@ -#include "../../include/capture/xcomposite_cuda.h" -#include "../../include/cuda.h" -#include "../../include/window_texture.h" -#include "../../include/utils.h" -#include <libavutil/hwcontext.h> -#include <libavutil/hwcontext_cuda.h> -#include <libavutil/frame.h> -#include <libavcodec/avcodec.h> - -typedef struct { - gsr_capture_xcomposite_cuda_params params; - XEvent xev; - - bool should_stop; - bool stop_is_error; - bool window_resized; - bool created_hw_frame; - bool follow_focused_initialized; - double window_resize_timer; - - vec2i window_size; - - unsigned int target_texture_id; - vec2i texture_size; - Window window; - WindowTexture window_texture; - Atom net_active_window_atom; - - CUgraphicsResource cuda_graphics_resource; - CUarray mapped_array; - - gsr_cuda cuda; -} gsr_capture_xcomposite_cuda; - -static int max_int(int a, int b) { - return a > b ? a : b; -} - -static int min_int(int a, int b) { - return a < b ? a : b; -} - -static Window get_focused_window(Display *display, Atom net_active_window_atom) { - Atom type; - int format = 0; - unsigned long num_items = 0; - unsigned long bytes_after = 0; - unsigned char *properties = NULL; - if(XGetWindowProperty(display, DefaultRootWindow(display), net_active_window_atom, 0, 1024, False, AnyPropertyType, &type, &format, &num_items, &bytes_after, &properties) == Success && properties) { - Window focused_window = *(unsigned long*)properties; - XFree(properties); - return focused_window; - } - return None; -} - -static void gsr_capture_xcomposite_cuda_stop(gsr_capture *cap, AVCodecContext *video_codec_context); - -static bool cuda_register_opengl_texture(gsr_capture_xcomposite_cuda *cap_xcomp) { - CUresult res; - CUcontext old_ctx; - res = cap_xcomp->cuda.cuCtxPushCurrent_v2(cap_xcomp->cuda.cu_ctx); - // TODO: Use cuGraphicsEGLRegisterImage instead with the window egl image (dont use window_texture). - // That removes the need for an extra texture and texture copy - res = cap_xcomp->cuda.cuGraphicsGLRegisterImage( - &cap_xcomp->cuda_graphics_resource, cap_xcomp->target_texture_id, GL_TEXTURE_2D, - CU_GRAPHICS_REGISTER_FLAGS_READ_ONLY); - if (res != CUDA_SUCCESS) { - const char *err_str = "unknown"; - cap_xcomp->cuda.cuGetErrorString(res, &err_str); - fprintf(stderr, "gsr error: cuda_register_opengl_texture: cuGraphicsGLRegisterImage failed, error: %s, texture " "id: %u\n", err_str, cap_xcomp->target_texture_id); - res = cap_xcomp->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - res = cap_xcomp->cuda.cuGraphicsResourceSetMapFlags(cap_xcomp->cuda_graphics_resource, CU_GRAPHICS_MAP_RESOURCE_FLAGS_READ_ONLY); - res = cap_xcomp->cuda.cuGraphicsMapResources(1, &cap_xcomp->cuda_graphics_resource, 0); - - res = cap_xcomp->cuda.cuGraphicsSubResourceGetMappedArray(&cap_xcomp->mapped_array, cap_xcomp->cuda_graphics_resource, 0, 0); - res = cap_xcomp->cuda.cuCtxPopCurrent_v2(&old_ctx); - return true; -} - -static bool cuda_create_codec_context(gsr_capture_xcomposite_cuda *cap_xcomp, AVCodecContext *video_codec_context) { - CUcontext old_ctx; - cap_xcomp->cuda.cuCtxPushCurrent_v2(cap_xcomp->cuda.cu_ctx); - - AVBufferRef *device_ctx = av_hwdevice_ctx_alloc(AV_HWDEVICE_TYPE_CUDA); - if(!device_ctx) { - fprintf(stderr, "Error: Failed to create hardware device context\n"); - cap_xcomp->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - AVHWDeviceContext *hw_device_context = (AVHWDeviceContext*)device_ctx->data; - AVCUDADeviceContext *cuda_device_context = (AVCUDADeviceContext*)hw_device_context->hwctx; - cuda_device_context->cuda_ctx = cap_xcomp->cuda.cu_ctx; - if(av_hwdevice_ctx_init(device_ctx) < 0) { - fprintf(stderr, "Error: Failed to create hardware device context\n"); - av_buffer_unref(&device_ctx); - cap_xcomp->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - AVBufferRef *frame_context = av_hwframe_ctx_alloc(device_ctx); - if(!frame_context) { - fprintf(stderr, "Error: Failed to create hwframe context\n"); - av_buffer_unref(&device_ctx); - cap_xcomp->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - AVHWFramesContext *hw_frame_context = - (AVHWFramesContext *)frame_context->data; - hw_frame_context->width = video_codec_context->width; - hw_frame_context->height = video_codec_context->height; - hw_frame_context->sw_format = AV_PIX_FMT_BGR0; - hw_frame_context->format = video_codec_context->pix_fmt; - hw_frame_context->device_ref = device_ctx; - hw_frame_context->device_ctx = (AVHWDeviceContext*)device_ctx->data; - - hw_frame_context->initial_pool_size = 1; - - if (av_hwframe_ctx_init(frame_context) < 0) { - fprintf(stderr, "Error: Failed to initialize hardware frame context " - "(note: ffmpeg version needs to be > 4.0)\n"); - av_buffer_unref(&device_ctx); - //av_buffer_unref(&frame_context); - cap_xcomp->cuda.cuCtxPopCurrent_v2(&old_ctx); - return false; - } - - video_codec_context->hw_device_ctx = av_buffer_ref(device_ctx); - video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); - return true; -} - -static unsigned int gl_create_texture(gsr_capture_xcomposite_cuda *cap_xcomp, int width, int height) { - unsigned int texture_id = 0; - cap_xcomp->params.egl->glGenTextures(1, &texture_id); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, texture_id); - cap_xcomp->params.egl->glTexImage2D(GL_TEXTURE_2D, 0, GL_RGB, width, height, 0, GL_RGB, GL_UNSIGNED_BYTE, NULL); - - cap_xcomp->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); - cap_xcomp->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); - cap_xcomp->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); - cap_xcomp->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); - - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - return texture_id; -} - -static int gsr_capture_xcomposite_cuda_start(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_xcomposite_cuda *cap_xcomp = cap->priv; - - if(cap_xcomp->params.follow_focused) { - cap_xcomp->net_active_window_atom = XInternAtom(cap_xcomp->params.dpy, "_NET_ACTIVE_WINDOW", False); - if(!cap_xcomp->net_active_window_atom) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_start failed: failed to get _NET_ACTIVE_WINDOW atom\n"); - return -1; - } - cap_xcomp->window = get_focused_window(cap_xcomp->params.dpy, cap_xcomp->net_active_window_atom); - } else { - cap_xcomp->window = cap_xcomp->params.window; - } - - /* TODO: Do these in tick, and allow error if follow_focused */ - - XWindowAttributes attr; - attr.width = 0; - attr.height = 0; - if(!XGetWindowAttributes(cap_xcomp->params.dpy, cap_xcomp->window, &attr) && !cap_xcomp->params.follow_focused) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_start failed: invalid window id: %lu\n", cap_xcomp->window); - return -1; - } - - cap_xcomp->window_size.x = max_int(attr.width, 0); - cap_xcomp->window_size.y = max_int(attr.height, 0); - - if(cap_xcomp->params.follow_focused) - XSelectInput(cap_xcomp->params.dpy, DefaultRootWindow(cap_xcomp->params.dpy), PropertyChangeMask); - - XSelectInput(cap_xcomp->params.dpy, cap_xcomp->window, StructureNotifyMask | ExposureMask); - - cap_xcomp->params.egl->eglSwapInterval(cap_xcomp->params.egl->egl_display, 0); - if(window_texture_init(&cap_xcomp->window_texture, cap_xcomp->params.dpy, cap_xcomp->window, cap_xcomp->params.egl) != 0 && !cap_xcomp->params.follow_focused) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_start: failed to get window texture for window %ld\n", cap_xcomp->window); - return -1; - } - - cap_xcomp->texture_size.x = 0; - cap_xcomp->texture_size.y = 0; - - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&cap_xcomp->window_texture)); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &cap_xcomp->texture_size.x); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &cap_xcomp->texture_size.y); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - cap_xcomp->texture_size.x = max_int(2, cap_xcomp->texture_size.x & ~1); - cap_xcomp->texture_size.y = max_int(2, cap_xcomp->texture_size.y & ~1); - - video_codec_context->width = cap_xcomp->texture_size.x; - video_codec_context->height = cap_xcomp->texture_size.y; - - if(cap_xcomp->params.region_size.x > 0 && cap_xcomp->params.region_size.y > 0) { - video_codec_context->width = max_int(2, cap_xcomp->params.region_size.x & ~1); - video_codec_context->height = max_int(2, cap_xcomp->params.region_size.y & ~1); - } - - cap_xcomp->target_texture_id = gl_create_texture(cap_xcomp, video_codec_context->width, video_codec_context->height); - if(cap_xcomp->target_texture_id == 0) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_start: failed to create opengl texture\n"); - gsr_capture_xcomposite_cuda_stop(cap, video_codec_context); - return -1; - } - - if(!gsr_cuda_load(&cap_xcomp->cuda, cap_xcomp->params.dpy, cap_xcomp->params.overclock)) { - gsr_capture_xcomposite_cuda_stop(cap, video_codec_context); - return -1; - } - - if(!cuda_create_codec_context(cap_xcomp, video_codec_context)) { - gsr_capture_xcomposite_cuda_stop(cap, video_codec_context); - return -1; - } - - if(!cuda_register_opengl_texture(cap_xcomp)) { - gsr_capture_xcomposite_cuda_stop(cap, video_codec_context); - return -1; - } - - cap_xcomp->window_resize_timer = clock_get_monotonic_seconds(); - return 0; -} - -static void gsr_capture_xcomposite_cuda_stop(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_xcomposite_cuda *cap_xcomp = cap->priv; - - if(cap_xcomp->cuda.cu_ctx) { - CUcontext old_ctx; - cap_xcomp->cuda.cuCtxPushCurrent_v2(cap_xcomp->cuda.cu_ctx); - - if(cap_xcomp->cuda_graphics_resource) { - cap_xcomp->cuda.cuGraphicsUnmapResources(1, &cap_xcomp->cuda_graphics_resource, 0); - cap_xcomp->cuda.cuGraphicsUnregisterResource(cap_xcomp->cuda_graphics_resource); - } - - cap_xcomp->cuda.cuCtxPopCurrent_v2(&old_ctx); - } - - window_texture_deinit(&cap_xcomp->window_texture); - - if(cap_xcomp->target_texture_id) { - cap_xcomp->params.egl->glDeleteTextures(1, &cap_xcomp->target_texture_id); - cap_xcomp->target_texture_id = 0; - } - - if(video_codec_context->hw_device_ctx) - av_buffer_unref(&video_codec_context->hw_device_ctx); - if(video_codec_context->hw_frames_ctx) - av_buffer_unref(&video_codec_context->hw_frames_ctx); - - gsr_cuda_unload(&cap_xcomp->cuda); - - if(cap_xcomp->params.dpy) { - // TODO: This causes a crash, why? maybe some other library dlclose xlib and that also happened to unload this??? - //XCloseDisplay(cap_xcomp->dpy); - cap_xcomp->params.dpy = NULL; - } -} - -static void gsr_capture_xcomposite_cuda_tick(gsr_capture *cap, AVCodecContext *video_codec_context, AVFrame **frame) { - gsr_capture_xcomposite_cuda *cap_xcomp = cap->priv; - - cap_xcomp->params.egl->glClear(GL_COLOR_BUFFER_BIT); - - bool init_new_window = false; - while(XPending(cap_xcomp->params.dpy)) { - XNextEvent(cap_xcomp->params.dpy, &cap_xcomp->xev); - - switch(cap_xcomp->xev.type) { - case DestroyNotify: { - /* Window died (when not following focused window), so we stop recording */ - if(!cap_xcomp->params.follow_focused && cap_xcomp->xev.xdestroywindow.window == cap_xcomp->window) { - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = false; - } - break; - } - case Expose: { - /* Requires window texture recreate */ - if(cap_xcomp->xev.xexpose.count == 0 && cap_xcomp->xev.xexpose.window == cap_xcomp->window) { - cap_xcomp->window_resize_timer = clock_get_monotonic_seconds(); - cap_xcomp->window_resized = true; - } - break; - } - case ConfigureNotify: { - /* Window resized */ - if(cap_xcomp->xev.xconfigure.window == cap_xcomp->window && (cap_xcomp->xev.xconfigure.width != cap_xcomp->window_size.x || cap_xcomp->xev.xconfigure.height != cap_xcomp->window_size.y)) { - cap_xcomp->window_size.x = max_int(cap_xcomp->xev.xconfigure.width, 0); - cap_xcomp->window_size.y = max_int(cap_xcomp->xev.xconfigure.height, 0); - cap_xcomp->window_resize_timer = clock_get_monotonic_seconds(); - cap_xcomp->window_resized = true; - } - break; - } - case PropertyNotify: { - /* Focused window changed */ - if(cap_xcomp->params.follow_focused && cap_xcomp->xev.xproperty.atom == cap_xcomp->net_active_window_atom) { - init_new_window = true; - } - break; - } - } - } - - if(cap_xcomp->params.follow_focused && !cap_xcomp->follow_focused_initialized) { - init_new_window = true; - } - - if(init_new_window) { - Window focused_window = get_focused_window(cap_xcomp->params.dpy, cap_xcomp->net_active_window_atom); - if(focused_window != cap_xcomp->window || !cap_xcomp->follow_focused_initialized) { - cap_xcomp->follow_focused_initialized = true; - XSelectInput(cap_xcomp->params.dpy, cap_xcomp->window, 0); - cap_xcomp->window = focused_window; - XSelectInput(cap_xcomp->params.dpy, cap_xcomp->window, StructureNotifyMask | ExposureMask); - - XWindowAttributes attr; - attr.width = 0; - attr.height = 0; - if(!XGetWindowAttributes(cap_xcomp->params.dpy, cap_xcomp->window, &attr)) - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_tick failed: invalid window id: %lu\n", cap_xcomp->window); - - cap_xcomp->window_size.x = max_int(attr.width, 0); - cap_xcomp->window_size.y = max_int(attr.height, 0); - cap_xcomp->window_resized = true; - - window_texture_deinit(&cap_xcomp->window_texture); - window_texture_init(&cap_xcomp->window_texture, cap_xcomp->params.dpy, cap_xcomp->window, cap_xcomp->params.egl); // TODO: Do not do the below window_texture_on_resize after this - - cap_xcomp->texture_size.x = 0; - cap_xcomp->texture_size.y = 0; - - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&cap_xcomp->window_texture)); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &cap_xcomp->texture_size.x); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &cap_xcomp->texture_size.y); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - cap_xcomp->texture_size.x = min_int(video_codec_context->width, max_int(2, cap_xcomp->texture_size.x & ~1)); - cap_xcomp->texture_size.y = min_int(video_codec_context->height, max_int(2, cap_xcomp->texture_size.y & ~1)); - } - } - - const double window_resize_timeout = 1.0; // 1 second - if(!cap_xcomp->created_hw_frame || (cap_xcomp->window_resized && clock_get_monotonic_seconds() - cap_xcomp->window_resize_timer >= window_resize_timeout)) { - cap_xcomp->window_resized = false; - if(window_texture_on_resize(&cap_xcomp->window_texture) != 0) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_tick: window_texture_on_resize failed\n"); - //cap_xcomp->should_stop = true; - //cap_xcomp->stop_is_error = true; - return; - } - - cap_xcomp->texture_size.x = 0; - cap_xcomp->texture_size.y = 0; - - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&cap_xcomp->window_texture)); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &cap_xcomp->texture_size.x); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &cap_xcomp->texture_size.y); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - cap_xcomp->texture_size.x = min_int(video_codec_context->width, max_int(2, cap_xcomp->texture_size.x & ~1)); - cap_xcomp->texture_size.y = min_int(video_codec_context->height, max_int(2, cap_xcomp->texture_size.y & ~1)); - - if(!cap_xcomp->created_hw_frame) { - cap_xcomp->created_hw_frame = true; - av_frame_free(frame); - *frame = av_frame_alloc(); - if(!frame) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_tick: failed to allocate frame\n"); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - return; - } - (*frame)->format = video_codec_context->pix_fmt; - (*frame)->width = video_codec_context->width; - (*frame)->height = video_codec_context->height; - (*frame)->color_range = video_codec_context->color_range; - (*frame)->color_primaries = video_codec_context->color_primaries; - (*frame)->color_trc = video_codec_context->color_trc; - (*frame)->colorspace = video_codec_context->colorspace; - (*frame)->chroma_location = video_codec_context->chroma_sample_location; - - if(av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, *frame, 0) < 0) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_tick: av_hwframe_get_buffer failed\n"); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - return; - } - } - - // Clear texture with black background because the source texture (window_texture_get_opengl_texture_id(&cap_xcomp->window_texture)) - // might be smaller than cap_xcomp->target_texture_id - cap_xcomp->params.egl->glClearTexImage(cap_xcomp->target_texture_id, 0, GL_RGB, GL_UNSIGNED_BYTE, NULL); - } -} - -static bool gsr_capture_xcomposite_cuda_should_stop(gsr_capture *cap, bool *err) { - gsr_capture_xcomposite_cuda *cap_xcomp = cap->priv; - if(cap_xcomp->should_stop) { - if(err) - *err = cap_xcomp->stop_is_error; - return true; - } - - if(err) - *err = false; - return false; -} - -static int gsr_capture_xcomposite_cuda_capture(gsr_capture *cap, AVFrame *frame) { - gsr_capture_xcomposite_cuda *cap_xcomp = cap->priv; - - vec2i source_pos = { 0, 0 }; - vec2i source_size = cap_xcomp->texture_size; - - if(cap_xcomp->window_texture.texture_id != 0) { - while(cap_xcomp->params.egl->glGetError()) {} - /* TODO: Remove this copy, which is only possible by using nvenc directly and encoding window_pixmap.target_texture_id */ - cap_xcomp->params.egl->glCopyImageSubData( - window_texture_get_opengl_texture_id(&cap_xcomp->window_texture), GL_TEXTURE_2D, 0, source_pos.x, source_pos.y, 0, - cap_xcomp->target_texture_id, GL_TEXTURE_2D, 0, 0, 0, 0, - source_size.x, source_size.y, 1); - unsigned int err = cap_xcomp->params.egl->glGetError(); - if(err != 0) { - static bool error_shown = false; - if(!error_shown) { - error_shown = true; - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_capture: glCopyImageSubData failed, gl error: %d\n", err); - } - } - } - cap_xcomp->params.egl->eglSwapBuffers(cap_xcomp->params.egl->egl_display, cap_xcomp->params.egl->egl_surface); - - frame->linesize[0] = frame->width * 4; - //frame->linesize[0] = frame->width * 1; - //frame->linesize[1] = frame->width * 1; - //frame->linesize[2] = frame->width * 1; - - CUDA_MEMCPY2D memcpy_struct; - memcpy_struct.srcXInBytes = 0; - memcpy_struct.srcY = 0; - memcpy_struct.srcMemoryType = CU_MEMORYTYPE_ARRAY; - - memcpy_struct.dstXInBytes = 0; - memcpy_struct.dstY = 0; - memcpy_struct.dstMemoryType = CU_MEMORYTYPE_DEVICE; - - memcpy_struct.srcArray = cap_xcomp->mapped_array; - memcpy_struct.dstDevice = (CUdeviceptr)frame->data[0]; - memcpy_struct.dstPitch = frame->linesize[0]; - memcpy_struct.WidthInBytes = frame->width * 4;//frame->width * 1; - memcpy_struct.Height = frame->height; - cap_xcomp->cuda.cuMemcpy2D_v2(&memcpy_struct); - - //frame->data[1] = frame->data[0]; - //frame->data[2] = frame->data[0]; - - return 0; -} - -static void gsr_capture_xcomposite_cuda_destroy(gsr_capture *cap, AVCodecContext *video_codec_context) { - if(cap->priv) { - gsr_capture_xcomposite_cuda_stop(cap, video_codec_context); - free(cap->priv); - cap->priv = NULL; - } - free(cap); -} - -gsr_capture* gsr_capture_xcomposite_cuda_create(const gsr_capture_xcomposite_cuda_params *params) { - if(!params) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_cuda_create params is NULL\n"); - return NULL; - } - - gsr_capture *cap = calloc(1, sizeof(gsr_capture)); - if(!cap) - return NULL; - - gsr_capture_xcomposite_cuda *cap_xcomp = calloc(1, sizeof(gsr_capture_xcomposite_cuda)); - if(!cap_xcomp) { - free(cap); - return NULL; - } - - cap_xcomp->params = *params; - - *cap = (gsr_capture) { - .start = gsr_capture_xcomposite_cuda_start, - .tick = gsr_capture_xcomposite_cuda_tick, - .should_stop = gsr_capture_xcomposite_cuda_should_stop, - .capture = gsr_capture_xcomposite_cuda_capture, - .capture_end = NULL, - .destroy = gsr_capture_xcomposite_cuda_destroy, - .priv = cap_xcomp - }; - - return cap; -} diff --git a/src/capture/xcomposite_vaapi.c b/src/capture/xcomposite_vaapi.c deleted file mode 100644 index 2bc21a7..0000000 --- a/src/capture/xcomposite_vaapi.c +++ /dev/null @@ -1,517 +0,0 @@ -#include "../../include/capture/xcomposite_vaapi.h" -#include "../../include/window_texture.h" -#include "../../include/utils.h" -#include "../../include/color_conversion.h" -#include <stdlib.h> -#include <stdio.h> -#include <unistd.h> -#include <assert.h> -#include <X11/Xlib.h> -#include <libavutil/hwcontext.h> -#include <libavutil/hwcontext_vaapi.h> -#include <libavutil/frame.h> -#include <libavcodec/avcodec.h> -#include <va/va.h> -#include <va/va_drmcommon.h> - -typedef struct { - gsr_capture_xcomposite_vaapi_params params; - XEvent xev; - - bool should_stop; - bool stop_is_error; - bool window_resized; - bool created_hw_frame; - bool follow_focused_initialized; - - Window window; - vec2i window_size; - vec2i texture_size; - double window_resize_timer; - - WindowTexture window_texture; - - VADisplay va_dpy; - VADRMPRIMESurfaceDescriptor prime; - - unsigned int target_textures[2]; - - gsr_color_conversion color_conversion; - - Atom net_active_window_atom; -} gsr_capture_xcomposite_vaapi; - -static int max_int(int a, int b) { - return a > b ? a : b; -} - -static int min_int(int a, int b) { - return a < b ? a : b; -} - -static void gsr_capture_xcomposite_vaapi_stop(gsr_capture *cap, AVCodecContext *video_codec_context); - -static Window get_focused_window(Display *display, Atom net_active_window_atom) { - Atom type; - int format = 0; - unsigned long num_items = 0; - unsigned long bytes_after = 0; - unsigned char *properties = NULL; - if(XGetWindowProperty(display, DefaultRootWindow(display), net_active_window_atom, 0, 1024, False, AnyPropertyType, &type, &format, &num_items, &bytes_after, &properties) == Success && properties) { - Window focused_window = *(unsigned long*)properties; - XFree(properties); - return focused_window; - } - return None; -} - -static bool drm_create_codec_context(gsr_capture_xcomposite_vaapi *cap_xcomp, AVCodecContext *video_codec_context) { - char render_path[128]; - if(!gsr_card_path_get_render_path(cap_xcomp->params.card_path, render_path)) { - fprintf(stderr, "gsr error: failed to get /dev/dri/renderDXXX file from %s\n", cap_xcomp->params.card_path); - return false; - } - - AVBufferRef *device_ctx; - if(av_hwdevice_ctx_create(&device_ctx, AV_HWDEVICE_TYPE_VAAPI, render_path, NULL, 0) < 0) { - fprintf(stderr, "Error: Failed to create hardware device context\n"); - return false; - } - - AVBufferRef *frame_context = av_hwframe_ctx_alloc(device_ctx); - if(!frame_context) { - fprintf(stderr, "Error: Failed to create hwframe context\n"); - av_buffer_unref(&device_ctx); - return false; - } - - AVHWFramesContext *hw_frame_context = - (AVHWFramesContext *)frame_context->data; - hw_frame_context->width = video_codec_context->width; - hw_frame_context->height = video_codec_context->height; - hw_frame_context->sw_format = AV_PIX_FMT_NV12;//AV_PIX_FMT_0RGB32;//AV_PIX_FMT_YUV420P;//AV_PIX_FMT_0RGB32;//AV_PIX_FMT_NV12; - hw_frame_context->format = video_codec_context->pix_fmt; - hw_frame_context->device_ref = device_ctx; - hw_frame_context->device_ctx = (AVHWDeviceContext*)device_ctx->data; - - hw_frame_context->initial_pool_size = 1; - - AVVAAPIDeviceContext *vactx =((AVHWDeviceContext*)device_ctx->data)->hwctx; - cap_xcomp->va_dpy = vactx->display; - - if (av_hwframe_ctx_init(frame_context) < 0) { - fprintf(stderr, "Error: Failed to initialize hardware frame context " - "(note: ffmpeg version needs to be > 4.0)\n"); - av_buffer_unref(&device_ctx); - //av_buffer_unref(&frame_context); - return false; - } - - video_codec_context->hw_device_ctx = av_buffer_ref(device_ctx); - video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); - return true; -} - -#define DRM_FORMAT_MOD_INVALID 72057594037927935 - -static int gsr_capture_xcomposite_vaapi_start(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_xcomposite_vaapi *cap_xcomp = cap->priv; - - if(cap_xcomp->params.follow_focused) { - cap_xcomp->net_active_window_atom = XInternAtom(cap_xcomp->params.dpy, "_NET_ACTIVE_WINDOW", False); - if(!cap_xcomp->net_active_window_atom) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_start failed: failed to get _NET_ACTIVE_WINDOW atom\n"); - return -1; - } - cap_xcomp->window = get_focused_window(cap_xcomp->params.dpy, cap_xcomp->net_active_window_atom); - } else { - cap_xcomp->window = cap_xcomp->params.window; - } - - /* TODO: Do these in tick, and allow error if follow_focused */ - - XWindowAttributes attr; - if(!XGetWindowAttributes(cap_xcomp->params.dpy, cap_xcomp->params.window, &attr) && !cap_xcomp->params.follow_focused) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_start failed: invalid window id: %lu\n", cap_xcomp->params.window); - return -1; - } - - cap_xcomp->window_size.x = max_int(attr.width, 0); - cap_xcomp->window_size.y = max_int(attr.height, 0); - - if(cap_xcomp->params.follow_focused) - XSelectInput(cap_xcomp->params.dpy, DefaultRootWindow(cap_xcomp->params.dpy), PropertyChangeMask); - - // TODO: Get select and add these on top of it and then restore at the end. Also do the same in other xcomposite - XSelectInput(cap_xcomp->params.dpy, cap_xcomp->params.window, StructureNotifyMask | ExposureMask); - - if(!cap_xcomp->params.egl->eglExportDMABUFImageQueryMESA) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_start: could not find eglExportDMABUFImageQueryMESA\n"); - return -1; - } - - if(!cap_xcomp->params.egl->eglExportDMABUFImageMESA) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_start: could not find eglExportDMABUFImageMESA\n"); - return -1; - } - - /* Disable vsync */ - cap_xcomp->params.egl->eglSwapInterval(cap_xcomp->params.egl->egl_display, 0); - if(window_texture_init(&cap_xcomp->window_texture, cap_xcomp->params.dpy, cap_xcomp->params.window, cap_xcomp->params.egl) != 0 && !cap_xcomp->params.follow_focused) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_start: failed to get window texture for window %ld\n", cap_xcomp->params.window); - return -1; - } - - cap_xcomp->texture_size.x = 0; - cap_xcomp->texture_size.y = 0; - - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&cap_xcomp->window_texture)); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &cap_xcomp->texture_size.x); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &cap_xcomp->texture_size.y); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - cap_xcomp->texture_size.x = max_int(2, even_number_ceil(cap_xcomp->texture_size.x)); - cap_xcomp->texture_size.y = max_int(2, even_number_ceil(cap_xcomp->texture_size.y)); - - video_codec_context->width = cap_xcomp->texture_size.x; - video_codec_context->height = cap_xcomp->texture_size.y; - - if(cap_xcomp->params.region_size.x > 0 && cap_xcomp->params.region_size.y > 0) { - video_codec_context->width = max_int(2, even_number_ceil(cap_xcomp->params.region_size.x)); - video_codec_context->height = max_int(2, even_number_ceil(cap_xcomp->params.region_size.y)); - } - - if(!drm_create_codec_context(cap_xcomp, video_codec_context)) { - gsr_capture_xcomposite_vaapi_stop(cap, video_codec_context); - return -1; - } - - cap_xcomp->window_resize_timer = clock_get_monotonic_seconds(); - return 0; -} - -static uint32_t fourcc(uint32_t a, uint32_t b, uint32_t c, uint32_t d) { - return (d << 24) | (c << 16) | (b << 8) | a; -} - -#define FOURCC_NV12 842094158 - -static void gsr_capture_xcomposite_vaapi_tick(gsr_capture *cap, AVCodecContext *video_codec_context, AVFrame **frame) { - gsr_capture_xcomposite_vaapi *cap_xcomp = cap->priv; - - // TODO: - cap_xcomp->params.egl->glClear(GL_COLOR_BUFFER_BIT); - - bool init_new_window = false; - while(XPending(cap_xcomp->params.dpy)) { - XNextEvent(cap_xcomp->params.dpy, &cap_xcomp->xev); - - switch(cap_xcomp->xev.type) { - case DestroyNotify: { - /* Window died (when not following focused window), so we stop recording */ - if(!cap_xcomp->params.follow_focused && cap_xcomp->xev.xdestroywindow.window == cap_xcomp->window) { - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = false; - } - break; - } - case Expose: { - /* Requires window texture recreate */ - if(cap_xcomp->xev.xexpose.count == 0 && cap_xcomp->xev.xexpose.window == cap_xcomp->window) { - cap_xcomp->window_resize_timer = clock_get_monotonic_seconds(); - cap_xcomp->window_resized = true; - } - break; - } - case ConfigureNotify: { - /* Window resized */ - if(cap_xcomp->xev.xconfigure.window == cap_xcomp->window && (cap_xcomp->xev.xconfigure.width != cap_xcomp->window_size.x || cap_xcomp->xev.xconfigure.height != cap_xcomp->window_size.y)) { - cap_xcomp->window_size.x = max_int(cap_xcomp->xev.xconfigure.width, 0); - cap_xcomp->window_size.y = max_int(cap_xcomp->xev.xconfigure.height, 0); - cap_xcomp->window_resize_timer = clock_get_monotonic_seconds(); - cap_xcomp->window_resized = true; - } - break; - } - case PropertyNotify: { - /* Focused window changed */ - if(cap_xcomp->params.follow_focused && cap_xcomp->xev.xproperty.atom == cap_xcomp->net_active_window_atom) { - init_new_window = true; - } - break; - } - } - } - - if(cap_xcomp->params.follow_focused && !cap_xcomp->follow_focused_initialized) { - init_new_window = true; - } - - if(init_new_window) { - Window focused_window = get_focused_window(cap_xcomp->params.dpy, cap_xcomp->net_active_window_atom); - if(focused_window != cap_xcomp->window || !cap_xcomp->follow_focused_initialized) { - cap_xcomp->follow_focused_initialized = true; - XSelectInput(cap_xcomp->params.dpy, cap_xcomp->window, 0); - cap_xcomp->window = focused_window; - XSelectInput(cap_xcomp->params.dpy, cap_xcomp->window, StructureNotifyMask | ExposureMask); - - XWindowAttributes attr; - attr.width = 0; - attr.height = 0; - if(!XGetWindowAttributes(cap_xcomp->params.dpy, cap_xcomp->window, &attr)) - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick failed: invalid window id: %lu\n", cap_xcomp->window); - - cap_xcomp->window_size.x = max_int(attr.width, 0); - cap_xcomp->window_size.y = max_int(attr.height, 0); - cap_xcomp->window_resized = true; - - window_texture_deinit(&cap_xcomp->window_texture); - window_texture_init(&cap_xcomp->window_texture, cap_xcomp->params.dpy, cap_xcomp->window, cap_xcomp->params.egl); // TODO: Do not do the below window_texture_on_resize after this - - cap_xcomp->texture_size.x = 0; - cap_xcomp->texture_size.y = 0; - - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&cap_xcomp->window_texture)); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &cap_xcomp->texture_size.x); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &cap_xcomp->texture_size.y); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - cap_xcomp->texture_size.x = min_int(video_codec_context->width, max_int(2, even_number_ceil(cap_xcomp->texture_size.x))); - cap_xcomp->texture_size.y = min_int(video_codec_context->height, max_int(2, even_number_ceil(cap_xcomp->texture_size.y))); - } - } - - const double window_resize_timeout = 1.0; // 1 second - if(!cap_xcomp->created_hw_frame || (cap_xcomp->window_resized && clock_get_monotonic_seconds() - cap_xcomp->window_resize_timer >= window_resize_timeout)) { - cap_xcomp->window_resized = false; - - if(window_texture_on_resize(&cap_xcomp->window_texture) != 0) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick: window_texture_on_resize failed\n"); - //cap_xcomp->should_stop = true; - //cap_xcomp->stop_is_error = true; - return; - } - - cap_xcomp->texture_size.x = 0; - cap_xcomp->texture_size.y = 0; - - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, window_texture_get_opengl_texture_id(&cap_xcomp->window_texture)); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &cap_xcomp->texture_size.x); - cap_xcomp->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &cap_xcomp->texture_size.y); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - - cap_xcomp->texture_size.x = min_int(video_codec_context->width, max_int(2, even_number_ceil(cap_xcomp->texture_size.x))); - cap_xcomp->texture_size.y = min_int(video_codec_context->height, max_int(2, even_number_ceil(cap_xcomp->texture_size.y))); - - if(!cap_xcomp->created_hw_frame) { - cap_xcomp->created_hw_frame = true; - av_frame_free(frame); - *frame = av_frame_alloc(); - if(!frame) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick: failed to allocate frame\n"); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - return; - } - (*frame)->format = video_codec_context->pix_fmt; - (*frame)->width = video_codec_context->width; - (*frame)->height = video_codec_context->height; - (*frame)->color_range = video_codec_context->color_range; - (*frame)->color_primaries = video_codec_context->color_primaries; - (*frame)->color_trc = video_codec_context->color_trc; - (*frame)->colorspace = video_codec_context->colorspace; - (*frame)->chroma_location = video_codec_context->chroma_sample_location; - - int res = av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, *frame, 0); - if(res < 0) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick: av_hwframe_get_buffer failed: %d\n", res); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - return; - } - - VASurfaceID target_surface_id = (uintptr_t)(*frame)->data[3]; - - VAStatus va_status = vaExportSurfaceHandle(cap_xcomp->va_dpy, target_surface_id, VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME_2, VA_EXPORT_SURFACE_READ_WRITE | VA_EXPORT_SURFACE_SEPARATE_LAYERS, &cap_xcomp->prime); - if(va_status != VA_STATUS_SUCCESS) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick: vaExportSurfaceHandle failed, error: %d\n", va_status); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - return; - } - vaSyncSurface(cap_xcomp->va_dpy, target_surface_id); - - if(cap_xcomp->prime.fourcc == FOURCC_NV12) { - cap_xcomp->params.egl->glGenTextures(2, cap_xcomp->target_textures); - for(int i = 0; i < 2; ++i) { - const uint32_t formats[2] = { fourcc('R', '8', ' ', ' '), fourcc('G', 'R', '8', '8') }; - const int layer = i; - const int plane = 0; - - const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size - //const uint64_t modifier = cap_kms->prime.objects[cap_kms->prime.layers[layer].object_index[plane]].drm_format_modifier; - - const intptr_t img_attr[] = { - EGL_LINUX_DRM_FOURCC_EXT, formats[i], - EGL_WIDTH, cap_xcomp->prime.width / div[i], - EGL_HEIGHT, cap_xcomp->prime.height / div[i], - EGL_DMA_BUF_PLANE0_FD_EXT, cap_xcomp->prime.objects[cap_xcomp->prime.layers[layer].object_index[plane]].fd, - EGL_DMA_BUF_PLANE0_OFFSET_EXT, cap_xcomp->prime.layers[layer].offset[plane], - EGL_DMA_BUF_PLANE0_PITCH_EXT, cap_xcomp->prime.layers[layer].pitch[plane], - // TODO: - //EGL_DMA_BUF_PLANE0_MODIFIER_LO_EXT, modifier & 0xFFFFFFFFULL, - //EGL_DMA_BUF_PLANE0_MODIFIER_HI_EXT, modifier >> 32ULL, - EGL_NONE - }; - - while(cap_xcomp->params.egl->eglGetError() != EGL_SUCCESS){} - EGLImage image = cap_xcomp->params.egl->eglCreateImage(cap_xcomp->params.egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr); - if(!image) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick: failed to create egl image from drm fd for output drm fd, error: %d\n", cap_xcomp->params.egl->eglGetError()); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - return; - } - - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, cap_xcomp->target_textures[i]); - cap_xcomp->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); - cap_xcomp->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); - cap_xcomp->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); - cap_xcomp->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); - - while(cap_xcomp->params.egl->glGetError()) {} - while(cap_xcomp->params.egl->eglGetError() != EGL_SUCCESS){} - cap_xcomp->params.egl->glEGLImageTargetTexture2DOES(GL_TEXTURE_2D, image); - if(cap_xcomp->params.egl->glGetError() != 0 || cap_xcomp->params.egl->eglGetError() != EGL_SUCCESS) { - // TODO: Get the error properly - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick: failed to bind egl image to gl texture, error: %d\n", cap_xcomp->params.egl->eglGetError()); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - cap_xcomp->params.egl->eglDestroyImage(cap_xcomp->params.egl->egl_display, image); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - return; - } - - cap_xcomp->params.egl->eglDestroyImage(cap_xcomp->params.egl->egl_display, image); - cap_xcomp->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - } - - gsr_color_conversion_params color_conversion_params = {0}; - color_conversion_params.egl = cap_xcomp->params.egl; - color_conversion_params.source_color = GSR_SOURCE_COLOR_RGB; - color_conversion_params.destination_color = GSR_DESTINATION_COLOR_NV12; - - color_conversion_params.destination_textures[0] = cap_xcomp->target_textures[0]; - color_conversion_params.destination_textures[1] = cap_xcomp->target_textures[1]; - color_conversion_params.num_destination_textures = 2; - - if(gsr_color_conversion_init(&cap_xcomp->color_conversion, &color_conversion_params) != 0) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick: failed to create color conversion\n"); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - return; - } - } else { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_tick: unexpected fourcc %u for output drm fd, expected nv12\n", cap_xcomp->prime.fourcc); - cap_xcomp->should_stop = true; - cap_xcomp->stop_is_error = true; - return; - } - } - } -} - -static bool gsr_capture_xcomposite_vaapi_should_stop(gsr_capture *cap, bool *err) { - gsr_capture_xcomposite_vaapi *cap_xcomp = cap->priv; - if(cap_xcomp->should_stop) { - if(err) - *err = cap_xcomp->stop_is_error; - return true; - } - - if(err) - *err = false; - return false; -} - -static int gsr_capture_xcomposite_vaapi_capture(gsr_capture *cap, AVFrame *frame) { - (void)frame; - gsr_capture_xcomposite_vaapi *cap_xcomp = cap->priv; - - float texture_rotation = 0.0f; - gsr_color_conversion_draw(&cap_xcomp->color_conversion, window_texture_get_opengl_texture_id(&cap_xcomp->window_texture), - (vec2i){0, 0}, cap_xcomp->texture_size, - (vec2i){0, 0}, cap_xcomp->texture_size, - texture_rotation); - - cap_xcomp->params.egl->eglSwapBuffers(cap_xcomp->params.egl->egl_display, cap_xcomp->params.egl->egl_surface); - - return 0; -} - -static void gsr_capture_xcomposite_vaapi_stop(gsr_capture *cap, AVCodecContext *video_codec_context) { - gsr_capture_xcomposite_vaapi *cap_xcomp = cap->priv; - - gsr_color_conversion_deinit(&cap_xcomp->color_conversion); - - for(uint32_t i = 0; i < cap_xcomp->prime.num_objects; ++i) { - if(cap_xcomp->prime.objects[i].fd > 0) { - close(cap_xcomp->prime.objects[i].fd); - cap_xcomp->prime.objects[i].fd = 0; - } - } - - if(cap_xcomp->params.egl->egl_context) { - cap_xcomp->params.egl->glDeleteTextures(2, cap_xcomp->target_textures); - cap_xcomp->target_textures[0] = 0; - cap_xcomp->target_textures[1] = 0; - } - - window_texture_deinit(&cap_xcomp->window_texture); - - if(video_codec_context->hw_device_ctx) - av_buffer_unref(&video_codec_context->hw_device_ctx); - if(video_codec_context->hw_frames_ctx) - av_buffer_unref(&video_codec_context->hw_frames_ctx); -} - -static void gsr_capture_xcomposite_vaapi_destroy(gsr_capture *cap, AVCodecContext *video_codec_context) { - (void)video_codec_context; - if(cap->priv) { - gsr_capture_xcomposite_vaapi_stop(cap, video_codec_context); - free(cap->priv); - cap->priv = NULL; - } - free(cap); -} - -gsr_capture* gsr_capture_xcomposite_vaapi_create(const gsr_capture_xcomposite_vaapi_params *params) { - if(!params) { - fprintf(stderr, "gsr error: gsr_capture_xcomposite_vaapi_create params is NULL\n"); - return NULL; - } - - gsr_capture *cap = calloc(1, sizeof(gsr_capture)); - if(!cap) - return NULL; - - gsr_capture_xcomposite_vaapi *cap_xcomp = calloc(1, sizeof(gsr_capture_xcomposite_vaapi)); - if(!cap_xcomp) { - free(cap); - return NULL; - } - - cap_xcomp->params = *params; - - *cap = (gsr_capture) { - .start = gsr_capture_xcomposite_vaapi_start, - .tick = gsr_capture_xcomposite_vaapi_tick, - .should_stop = gsr_capture_xcomposite_vaapi_should_stop, - .capture = gsr_capture_xcomposite_vaapi_capture, - .capture_end = NULL, - .destroy = gsr_capture_xcomposite_vaapi_destroy, - .priv = cap_xcomp - }; - - return cap; -} diff --git a/src/capture/ximage.c b/src/capture/ximage.c new file mode 100644 index 0000000..9b02907 --- /dev/null +++ b/src/capture/ximage.c @@ -0,0 +1,247 @@ +#include "../../include/capture/ximage.h" +#include "../../include/utils.h" +#include "../../include/cursor.h" +#include "../../include/color_conversion.h" +#include "../../include/window/window.h" + +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <assert.h> + +#include <X11/Xlib.h> + +/* TODO: update when monitors are reconfigured */ + +typedef struct { + gsr_capture_ximage_params params; + Display *display; + gsr_cursor cursor; + gsr_monitor monitor; + vec2i capture_pos; + vec2i capture_size; + unsigned int texture_id; + Window root_window; +} gsr_capture_ximage; + +static void gsr_capture_ximage_stop(gsr_capture_ximage *self) { + gsr_cursor_deinit(&self->cursor); + if(self->texture_id) { + self->params.egl->glDeleteTextures(1, &self->texture_id); + self->texture_id = 0; + } +} + +static int max_int(int a, int b) { + return a > b ? a : b; +} + +static int gsr_capture_ximage_start(gsr_capture *cap, gsr_capture_metadata *capture_metadata) { + gsr_capture_ximage *self = cap->priv; + self->root_window = DefaultRootWindow(self->display); + + if(gsr_cursor_init(&self->cursor, self->params.egl, self->display) != 0) { + gsr_capture_ximage_stop(self); + return -1; + } + + if(!get_monitor_by_name(self->params.egl, GSR_CONNECTION_X11, self->params.display_to_capture, &self->monitor)) { + fprintf(stderr, "gsr error: gsr_capture_ximage_start: failed to find monitor by name \"%s\"\n", self->params.display_to_capture); + gsr_capture_ximage_stop(self); + return -1; + } + + self->capture_pos = self->monitor.pos; + self->capture_size = self->monitor.size; + + if(self->params.region_size.x > 0 && self->params.region_size.y > 0) + self->capture_size = self->params.region_size; + + if(self->params.output_resolution.x > 0 && self->params.output_resolution.y > 0) { + self->params.output_resolution = scale_keep_aspect_ratio(self->capture_size, self->params.output_resolution); + capture_metadata->width = self->params.output_resolution.x; + capture_metadata->height = self->params.output_resolution.y; + } else if(self->params.region_size.x > 0 && self->params.region_size.y > 0) { + capture_metadata->width = self->params.region_size.x; + capture_metadata->height = self->params.region_size.y; + } else { + capture_metadata->width = self->capture_size.x; + capture_metadata->height = self->capture_size.y; + } + + self->texture_id = gl_create_texture(self->params.egl, self->capture_size.x, self->capture_size.y, GL_RGB8, GL_RGB, GL_LINEAR); + if(self->texture_id == 0) { + fprintf(stderr, "gsr error: gsr_capture_ximage_start: failed to create texture\n"); + gsr_capture_ximage_stop(self); + return -1; + } + + return 0; +} + +static void gsr_capture_ximage_on_event(gsr_capture *cap, gsr_egl *egl) { + gsr_capture_ximage *self = cap->priv; + XEvent *xev = gsr_window_get_event_data(egl->window); + gsr_cursor_on_event(&self->cursor, xev); +} + +static bool gsr_capture_ximage_upload_to_texture(gsr_capture_ximage *self, int x, int y, int width, int height) { + const int max_width = XWidthOfScreen(DefaultScreenOfDisplay(self->display)); + const int max_height = XHeightOfScreen(DefaultScreenOfDisplay(self->display)); + + if(x < 0) + x = 0; + else if(x >= max_width) + x = max_width - 1; + + if(y < 0) + y = 0; + else if(y >= max_height) + y = max_height - 1; + + if(width < 0) + width = 0; + else if(x + width >= max_width) + width = max_width - x; + + if(height < 0) + height = 0; + else if(y + height >= max_height) + height = max_height - y; + + XImage *image = XGetImage(self->display, self->root_window, x, y, width, height, AllPlanes, ZPixmap); + if(!image) { + fprintf(stderr, "gsr error: gsr_capture_ximage_upload_to_texture: XGetImage failed\n"); + return false; + } + + bool success = false; + uint8_t *image_data = malloc(image->width * image->height * 3); + if(!image_data) { + fprintf(stderr, "gsr error: gsr_capture_ximage_upload_to_texture: failed to allocate image data\n"); + goto done; + } + + for(int y = 0; y < image->height; ++y) { + for(int x = 0; x < image->width; ++x) { + unsigned long pixel = XGetPixel(image, x, y); + unsigned char red = (pixel & image->red_mask) >> 16; + unsigned char green = (pixel & image->green_mask) >> 8; + unsigned char blue = pixel & image->blue_mask; + + const size_t texture_data_index = (x + y * image->width) * 3; + image_data[texture_data_index + 0] = red; + image_data[texture_data_index + 1] = green; + image_data[texture_data_index + 2] = blue; + } + } + + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->texture_id); + self->params.egl->glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, image->width, image->height, GL_RGB, GL_UNSIGNED_BYTE, image_data); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + success = true; + + done: + free(image_data); + XDestroyImage(image); + return success; +} + +static int gsr_capture_ximage_capture(gsr_capture *cap, gsr_capture_metadata *capture_metdata, gsr_color_conversion *color_conversion) { + gsr_capture_ximage *self = cap->priv; + + const bool is_scaled = self->params.output_resolution.x > 0 && self->params.output_resolution.y > 0; + vec2i output_size = is_scaled ? self->params.output_resolution : self->capture_size; + output_size = scale_keep_aspect_ratio(self->capture_size, output_size); + + const vec2i target_pos = { max_int(0, capture_metdata->width / 2 - output_size.x / 2), max_int(0, capture_metdata->height / 2 - output_size.y / 2) }; + gsr_capture_ximage_upload_to_texture(self, self->capture_pos.x + self->params.region_position.x, self->capture_pos.y + self->params.region_position.y, self->capture_size.x, self->capture_size.y); + + gsr_color_conversion_draw(color_conversion, self->texture_id, + target_pos, output_size, + (vec2i){0, 0}, self->capture_size, self->capture_size, + GSR_ROT_0, GSR_SOURCE_COLOR_RGB, false, false); + + if(self->params.record_cursor && self->cursor.visible) { + const vec2d scale = { + self->capture_size.x == 0 ? 0 : (double)output_size.x / (double)self->capture_size.x, + self->capture_size.y == 0 ? 0 : (double)output_size.y / (double)self->capture_size.y + }; + + gsr_cursor_tick(&self->cursor, self->root_window); + + const vec2i cursor_pos = { + target_pos.x + (self->cursor.position.x - self->cursor.hotspot.x) * scale.x - self->capture_pos.x - self->params.region_position.x, + target_pos.y + (self->cursor.position.y - self->cursor.hotspot.y) * scale.y - self->capture_pos.y - self->params.region_position.y + }; + + self->params.egl->glEnable(GL_SCISSOR_TEST); + self->params.egl->glScissor(target_pos.x, target_pos.y, output_size.x, output_size.y); + + gsr_color_conversion_draw(color_conversion, self->cursor.texture_id, + cursor_pos, (vec2i){self->cursor.size.x * scale.x, self->cursor.size.y * scale.y}, + (vec2i){0, 0}, self->cursor.size, self->cursor.size, + GSR_ROT_0, GSR_SOURCE_COLOR_RGB, false, true); + + self->params.egl->glDisable(GL_SCISSOR_TEST); + } + + self->params.egl->glFlush(); + self->params.egl->glFinish(); + + return 0; +} + +static void gsr_capture_ximage_destroy(gsr_capture *cap) { + gsr_capture_ximage *self = cap->priv; + if(cap->priv) { + gsr_capture_ximage_stop(self); + free((void*)self->params.display_to_capture); + self->params.display_to_capture = NULL; + free(self); + cap->priv = NULL; + } + free(cap); +} + +gsr_capture* gsr_capture_ximage_create(const gsr_capture_ximage_params *params) { + if(!params) { + fprintf(stderr, "gsr error: gsr_capture_ximage_create params is NULL\n"); + return NULL; + } + + gsr_capture *cap = calloc(1, sizeof(gsr_capture)); + if(!cap) + return NULL; + + gsr_capture_ximage *cap_ximage = calloc(1, sizeof(gsr_capture_ximage)); + if(!cap_ximage) { + free(cap); + return NULL; + } + + const char *display_to_capture = strdup(params->display_to_capture); + if(!display_to_capture) { + free(cap); + free(cap_ximage); + return NULL; + } + + cap_ximage->params = *params; + cap_ximage->display = gsr_window_get_display(params->egl->window); + cap_ximage->params.display_to_capture = display_to_capture; + + *cap = (gsr_capture) { + .start = gsr_capture_ximage_start, + .on_event = gsr_capture_ximage_on_event, + .tick = NULL, + .should_stop = NULL, + .capture = gsr_capture_ximage_capture, + .uses_external_image = NULL, + .get_window_id = NULL, + .destroy = gsr_capture_ximage_destroy, + .priv = cap_ximage + }; + + return cap; +} diff --git a/src/codec_query/nvenc.c b/src/codec_query/nvenc.c new file mode 100644 index 0000000..0501851 --- /dev/null +++ b/src/codec_query/nvenc.c @@ -0,0 +1,235 @@ +#include "../../include/codec_query/nvenc.h" +#include "../../include/cuda.h" +#include "../../external/nvEncodeAPI.h" + +#include <dlfcn.h> +#include <stdio.h> +#include <string.h> + +static void* open_nvenc_library(void) { + dlerror(); /* clear */ + void *lib = dlopen("libnvidia-encode.so.1", RTLD_LAZY); + if(!lib) { + lib = dlopen("libnvidia-encode.so", RTLD_LAZY); + if(!lib) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc failed: failed to load libnvidia-encode.so/libnvidia-encode.so.1, error: %s\n", dlerror()); + return NULL; + } + } + return lib; +} + +static bool profile_is_h264(const GUID *profile_guid) { + const GUID *h264_guids[] = { + &NV_ENC_H264_PROFILE_BASELINE_GUID, + &NV_ENC_H264_PROFILE_MAIN_GUID, + &NV_ENC_H264_PROFILE_HIGH_GUID, + &NV_ENC_H264_PROFILE_PROGRESSIVE_HIGH_GUID, + &NV_ENC_H264_PROFILE_CONSTRAINED_HIGH_GUID + }; + + for(int i = 0; i < 5; ++i) { + if(memcmp(profile_guid, h264_guids[i], sizeof(GUID)) == 0) + return true; + } + + return false; +} + +static bool profile_is_hevc(const GUID *profile_guid) { + const GUID *h264_guids[] = { + &NV_ENC_HEVC_PROFILE_MAIN_GUID, + }; + + for(int i = 0; i < 1; ++i) { + if(memcmp(profile_guid, h264_guids[i], sizeof(GUID)) == 0) + return true; + } + + return false; +} + +static bool profile_is_hevc_10bit(const GUID *profile_guid) { + const GUID *h264_guids[] = { + &NV_ENC_HEVC_PROFILE_MAIN10_GUID, + }; + + for(int i = 0; i < 1; ++i) { + if(memcmp(profile_guid, h264_guids[i], sizeof(GUID)) == 0) + return true; + } + + return false; +} + +static bool profile_is_av1(const GUID *profile_guid) { + const GUID *h264_guids[] = { + &NV_ENC_AV1_PROFILE_MAIN_GUID, + }; + + for(int i = 0; i < 1; ++i) { + if(memcmp(profile_guid, h264_guids[i], sizeof(GUID)) == 0) + return true; + } + + return false; +} + +static bool encoder_get_supported_profiles(const NV_ENCODE_API_FUNCTION_LIST *function_list, void *nvenc_encoder, const GUID *encoder_guid, gsr_supported_video_codecs *supported_video_codecs) { + bool success = false; + GUID *profile_guids = NULL; + + uint32_t profile_guid_count = 0; + if(function_list->nvEncGetEncodeProfileGUIDCount(nvenc_encoder, *encoder_guid, &profile_guid_count) != NV_ENC_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: nvEncGetEncodeProfileGUIDCount failed, error: %s\n", function_list->nvEncGetLastErrorString(nvenc_encoder)); + goto fail; + } + + if(profile_guid_count == 0) + goto fail; + + profile_guids = calloc(profile_guid_count, sizeof(GUID)); + if(!profile_guids) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: failed to allocate %d guids\n", (int)profile_guid_count); + goto fail; + } + + if(function_list->nvEncGetEncodeProfileGUIDs(nvenc_encoder, *encoder_guid, profile_guids, profile_guid_count, &profile_guid_count) != NV_ENC_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: nvEncGetEncodeProfileGUIDs failed, error: %s\n", function_list->nvEncGetLastErrorString(nvenc_encoder)); + goto fail; + } + + for(uint32_t i = 0; i < profile_guid_count; ++i) { + if(profile_is_h264(&profile_guids[i])) { + supported_video_codecs->h264 = (gsr_supported_video_codec){ true, false }; + } else if(profile_is_hevc(&profile_guids[i])) { + supported_video_codecs->hevc = (gsr_supported_video_codec){ true, false }; + } else if(profile_is_hevc_10bit(&profile_guids[i])) { + supported_video_codecs->hevc_hdr = (gsr_supported_video_codec){ true, false }; + supported_video_codecs->hevc_10bit = (gsr_supported_video_codec){ true, false }; + } else if(profile_is_av1(&profile_guids[i])) { + supported_video_codecs->av1 = (gsr_supported_video_codec){ true, false }; + supported_video_codecs->av1_hdr = (gsr_supported_video_codec){ true, false }; + supported_video_codecs->av1_10bit = (gsr_supported_video_codec){ true, false }; + } + } + + success = true; + fail: + + if(profile_guids) + free(profile_guids); + + return success; +} + +static bool get_supported_video_codecs(const NV_ENCODE_API_FUNCTION_LIST *function_list, void *nvenc_encoder, gsr_supported_video_codecs *supported_video_codecs) { + bool success = false; + GUID *encoder_guids = NULL; + *supported_video_codecs = (gsr_supported_video_codecs){0}; + + uint32_t encode_guid_count = 0; + if(function_list->nvEncGetEncodeGUIDCount(nvenc_encoder, &encode_guid_count) != NV_ENC_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: nvEncGetEncodeGUIDCount failed, error: %s\n", function_list->nvEncGetLastErrorString(nvenc_encoder)); + goto fail; + } + + if(encode_guid_count == 0) + goto fail; + + encoder_guids = calloc(encode_guid_count, sizeof(GUID)); + if(!encoder_guids) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: failed to allocate %d guids\n", (int)encode_guid_count); + goto fail; + } + + if(function_list->nvEncGetEncodeGUIDs(nvenc_encoder, encoder_guids, encode_guid_count, &encode_guid_count) != NV_ENC_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: nvEncGetEncodeGUIDs failed, error: %s\n", function_list->nvEncGetLastErrorString(nvenc_encoder)); + goto fail; + } + + for(uint32_t i = 0; i < encode_guid_count; ++i) { + encoder_get_supported_profiles(function_list, nvenc_encoder, &encoder_guids[i], supported_video_codecs); + } + + success = true; + fail: + + if(encoder_guids) + free(encoder_guids); + + return success; +} + +#define NVENCAPI_VERSION_470 (11 | (1 << 24)) +#define NVENCAPI_STRUCT_VERSION_470(ver) ((uint32_t)NVENCAPI_VERSION_470 | ((ver)<<16) | (0x7 << 28)) + +bool gsr_get_supported_video_codecs_nvenc(gsr_supported_video_codecs *video_codecs, bool cleanup) { + memset(video_codecs, 0, sizeof(*video_codecs)); + + bool success = false; + void *nvenc_lib = NULL; + void *nvenc_encoder = NULL; + gsr_cuda cuda; + memset(&cuda, 0, sizeof(cuda)); + + if(!gsr_cuda_load(&cuda, NULL, false)) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: failed to load cuda\n"); + goto done; + } + + nvenc_lib = open_nvenc_library(); + if(!nvenc_lib) + goto done; + + typedef NVENCSTATUS NVENCAPI (*FUNC_NvEncodeAPICreateInstance)(NV_ENCODE_API_FUNCTION_LIST *functionList); + FUNC_NvEncodeAPICreateInstance nvEncodeAPICreateInstance = (FUNC_NvEncodeAPICreateInstance)dlsym(nvenc_lib, "NvEncodeAPICreateInstance"); + if(!nvEncodeAPICreateInstance) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: failed to find NvEncodeAPICreateInstance in libnvidia-encode.so\n"); + goto done; + } + + NV_ENCODE_API_FUNCTION_LIST function_list; + memset(&function_list, 0, sizeof(function_list)); + function_list.version = NVENCAPI_STRUCT_VERSION(2); + if(nvEncodeAPICreateInstance(&function_list) != NV_ENC_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: nvEncodeAPICreateInstance failed\n"); + goto done; + } + + NV_ENC_OPEN_ENCODE_SESSION_EX_PARAMS params; + memset(¶ms, 0, sizeof(params)); + params.version = NVENCAPI_STRUCT_VERSION(1); + params.deviceType = NV_ENC_DEVICE_TYPE_CUDA; + params.device = cuda.cu_ctx; + params.apiVersion = NVENCAPI_VERSION; + if(function_list.nvEncOpenEncodeSessionEx(¶ms, &nvenc_encoder) != NV_ENC_SUCCESS) { + // Old nvidia gpus dont support the new nvenc api (which is required for av1). + // In such cases fallback to old api version if possible and try again. + function_list.version = NVENCAPI_STRUCT_VERSION_470(2); + if(nvEncodeAPICreateInstance(&function_list) != NV_ENC_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: nvEncodeAPICreateInstance (retry) failed\n"); + goto done; + } + + params.version = NVENCAPI_STRUCT_VERSION_470(1); + params.apiVersion = NVENCAPI_VERSION_470; + if(function_list.nvEncOpenEncodeSessionEx(¶ms, &nvenc_encoder) != NV_ENC_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_nvenc: nvEncOpenEncodeSessionEx (retry) failed\n"); + goto done; + } + } + + success = get_supported_video_codecs(&function_list, nvenc_encoder, video_codecs); + + done: + if(cleanup) { + if(nvenc_encoder) + function_list.nvEncDestroyEncoder(nvenc_encoder); + if(nvenc_lib) + dlclose(nvenc_lib); + gsr_cuda_unload(&cuda); + } + + return success; +} diff --git a/src/codec_query/vaapi.c b/src/codec_query/vaapi.c new file mode 100644 index 0000000..8930a6c --- /dev/null +++ b/src/codec_query/vaapi.c @@ -0,0 +1,203 @@ +#include "../../include/codec_query/vaapi.h" +#include "../../include/utils.h" + +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <fcntl.h> +#include <unistd.h> + +#include <va/va.h> +#include <va/va_drm.h> + +static bool profile_is_h264(VAProfile profile) { + switch(profile) { + case 5: // VAProfileH264Baseline + case VAProfileH264Main: + case VAProfileH264High: + case VAProfileH264ConstrainedBaseline: + return true; + default: + return false; + } +} + +static bool profile_is_hevc_8bit(VAProfile profile) { + switch(profile) { + case VAProfileHEVCMain: + return true; + default: + return false; + } +} + +static bool profile_is_hevc_10bit(VAProfile profile) { + switch(profile) { + case VAProfileHEVCMain10: + //case VAProfileHEVCMain12: + //case VAProfileHEVCMain422_10: + //case VAProfileHEVCMain422_12: + //case VAProfileHEVCMain444: + //case VAProfileHEVCMain444_10: + //case VAProfileHEVCMain444_12: + return true; + default: + return false; + } +} + +static bool profile_is_av1(VAProfile profile) { + switch(profile) { + case VAProfileAV1Profile0: + case VAProfileAV1Profile1: + return true; + default: + return false; + } +} + +static bool profile_is_vp8(VAProfile profile) { + switch(profile) { + case VAProfileVP8Version0_3: + return true; + default: + return false; + } +} + +static bool profile_is_vp9(VAProfile profile) { + switch(profile) { + case VAProfileVP9Profile0: + case VAProfileVP9Profile1: + case VAProfileVP9Profile2: + case VAProfileVP9Profile3: + return true; + default: + return false; + } +} + +static bool profile_supports_video_encoding(VADisplay va_dpy, VAProfile profile, bool *low_power) { + *low_power = false; + int num_entrypoints = vaMaxNumEntrypoints(va_dpy); + if(num_entrypoints <= 0) + return false; + + VAEntrypoint *entrypoint_list = calloc(num_entrypoints, sizeof(VAEntrypoint)); + if(!entrypoint_list) + return false; + + bool supports_encoding = false; + bool supports_low_power_encoding = false; + if(vaQueryConfigEntrypoints(va_dpy, profile, entrypoint_list, &num_entrypoints) == VA_STATUS_SUCCESS) { + for(int i = 0; i < num_entrypoints; ++i) { + if(entrypoint_list[i] == VAEntrypointEncSlice) + supports_encoding = true; + else if(entrypoint_list[i] == VAEntrypointEncSliceLP) + supports_low_power_encoding = true; + } + } + + if(!supports_encoding && supports_low_power_encoding) + *low_power = true; + + free(entrypoint_list); + return supports_encoding || supports_low_power_encoding; +} + +static bool get_supported_video_codecs(VADisplay va_dpy, gsr_supported_video_codecs *video_codecs, bool cleanup) { + *video_codecs = (gsr_supported_video_codecs){0}; + bool success = false; + VAProfile *profile_list = NULL; + + vaSetInfoCallback(va_dpy, NULL, NULL); + + int va_major = 0; + int va_minor = 0; + if(vaInitialize(va_dpy, &va_major, &va_minor) != VA_STATUS_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vaapi: vaInitialize failed\n"); + return false; + } + + int num_profiles = vaMaxNumProfiles(va_dpy); + if(num_profiles <= 0) + goto fail; + + profile_list = calloc(num_profiles, sizeof(VAProfile)); + if(!profile_list || vaQueryConfigProfiles(va_dpy, profile_list, &num_profiles) != VA_STATUS_SUCCESS) + goto fail; + + for(int i = 0; i < num_profiles; ++i) { + bool low_power = false; + if(profile_is_h264(profile_list[i])) { + if(profile_supports_video_encoding(va_dpy, profile_list[i], &low_power)) { + video_codecs->h264 = (gsr_supported_video_codec){ true, low_power }; + } + } else if(profile_is_hevc_8bit(profile_list[i])) { + if(profile_supports_video_encoding(va_dpy, profile_list[i], &low_power)) + video_codecs->hevc = (gsr_supported_video_codec){ true, low_power }; + } else if(profile_is_hevc_10bit(profile_list[i])) { + if(profile_supports_video_encoding(va_dpy, profile_list[i], &low_power)) { + video_codecs->hevc_hdr = (gsr_supported_video_codec){ true, low_power }; + video_codecs->hevc_10bit = (gsr_supported_video_codec){ true, low_power }; + } + } else if(profile_is_av1(profile_list[i])) { + if(profile_supports_video_encoding(va_dpy, profile_list[i], &low_power)) { + video_codecs->av1 = (gsr_supported_video_codec){ true, low_power }; + video_codecs->av1_hdr = (gsr_supported_video_codec){ true, low_power }; + video_codecs->av1_10bit = (gsr_supported_video_codec){ true, low_power }; + } + } else if(profile_is_vp8(profile_list[i])) { + if(profile_supports_video_encoding(va_dpy, profile_list[i], &low_power)) + video_codecs->vp8 = (gsr_supported_video_codec){ true, low_power }; + } else if(profile_is_vp9(profile_list[i])) { + if(profile_supports_video_encoding(va_dpy, profile_list[i], &low_power)) + video_codecs->vp9 = (gsr_supported_video_codec){ true, low_power }; + } + } + + success = true; + fail: + if(profile_list) + free(profile_list); + + if(cleanup) + vaTerminate(va_dpy); + + return success; +} + +bool gsr_get_supported_video_codecs_vaapi(gsr_supported_video_codecs *video_codecs, const char *card_path, bool cleanup) { + memset(video_codecs, 0, sizeof(*video_codecs)); + bool success = false; + int drm_fd = -1; + + char render_path[128]; + if(!gsr_card_path_get_render_path(card_path, render_path)) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vaapi: failed to get /dev/dri/renderDXXX file from %s\n", card_path); + goto done; + } + + drm_fd = open(render_path, O_RDWR); + if(drm_fd == -1) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vaapi: failed to open device %s\n", render_path); + goto done; + } + + VADisplay va_dpy = vaGetDisplayDRM(drm_fd); + if(va_dpy) { + if(!get_supported_video_codecs(va_dpy, video_codecs, cleanup)) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vaapi: failed to query supported video codecs for device %s\n", render_path); + goto done; + } + success = true; + } + + done: + if(cleanup) { + if(drm_fd > 0) + close(drm_fd); + } + + return success; +} diff --git a/src/codec_query/vulkan.c b/src/codec_query/vulkan.c new file mode 100644 index 0000000..15dd98b --- /dev/null +++ b/src/codec_query/vulkan.c @@ -0,0 +1,156 @@ +#include "../../include/codec_query/vulkan.h" + +#include <stdio.h> +#include <string.h> +#include <stdlib.h> +#include <xf86drm.h> +#define VK_NO_PROTOTYPES +//#include <vulkan/vulkan.h> + +#define MAX_PHYSICAL_DEVICES 32 + +static const char *required_device_extensions[] = { + "VK_KHR_external_memory_fd", + "VK_KHR_external_semaphore_fd", + "VK_KHR_video_encode_queue", + "VK_KHR_video_queue", + "VK_KHR_video_maintenance1", + "VK_EXT_external_memory_dma_buf", + "VK_EXT_external_memory_host", + "VK_EXT_image_drm_format_modifier" +}; +static int num_required_device_extensions = 8; + +bool gsr_get_supported_video_codecs_vulkan(gsr_supported_video_codecs *video_codecs, const char *card_path, bool cleanup) { + memset(video_codecs, 0, sizeof(*video_codecs)); +#if 0 + bool success = false; + VkInstance instance = NULL; + VkPhysicalDevice physical_devices[MAX_PHYSICAL_DEVICES]; + VkDevice device = NULL; + VkExtensionProperties *device_extensions = NULL; + + const VkApplicationInfo app_info = { + .sType = VK_STRUCTURE_TYPE_APPLICATION_INFO, + .pApplicationName = "GPU Screen Recorder", + .applicationVersion = VK_MAKE_VERSION(1, 0, 0), + .pEngineName = "GPU Screen Recorder", + .engineVersion = VK_MAKE_VERSION(1, 0, 0), + .apiVersion = VK_API_VERSION_1_3, + }; + + const VkInstanceCreateInfo instance_create_info = { + .sType = VK_STRUCTURE_TYPE_INSTANCE_CREATE_INFO, + .pApplicationInfo = &app_info + }; + + if(vkCreateInstance(&instance_create_info, NULL, &instance) != VK_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: vkCreateInstance failed\n"); + goto done; + } + + uint32_t num_devices = 0; + if(vkEnumeratePhysicalDevices(instance, &num_devices, NULL) != VK_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: vkEnumeratePhysicalDevices (query num devices) failed\n"); + goto done; + } + + if(num_devices == 0) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: no vulkan capable device found\n"); + goto done; + } + + if(num_devices > MAX_PHYSICAL_DEVICES) + num_devices = MAX_PHYSICAL_DEVICES; + + if(vkEnumeratePhysicalDevices(instance, &num_devices, physical_devices) != VK_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: vkEnumeratePhysicalDevices (get data) failed\n"); + goto done; + } + + VkPhysicalDevice physical_device = NULL; + char device_card_path[128]; + for(uint32_t i = 0; i < num_devices; ++i) { + VkPhysicalDeviceDrmPropertiesEXT device_drm_properties = { + .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DRM_PROPERTIES_EXT + }; + + VkPhysicalDeviceProperties2 device_properties = { + .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROPERTIES_2, + .pNext = &device_drm_properties + }; + vkGetPhysicalDeviceProperties2(physical_devices[i], &device_properties); + + if(!device_drm_properties.hasPrimary) + continue; + + snprintf(device_card_path, sizeof(device_card_path), DRM_DEV_NAME, DRM_DIR_NAME, (int)device_drm_properties.primaryMinor); + if(strcmp(device_card_path, card_path) == 0) { + physical_device = physical_devices[i]; + break; + } + } + + if(!physical_device) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: failed to find a vulkan device that matches opengl device %s\n", card_path); + goto done; + } + + const VkDeviceCreateInfo device_create_info = { + .sType = VK_STRUCTURE_TYPE_DEVICE_CREATE_INFO, + .enabledExtensionCount = num_required_device_extensions, + .ppEnabledExtensionNames = required_device_extensions + }; + + if(vkCreateDevice(physical_device, &device_create_info, NULL, &device) != VK_SUCCESS) { + //fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: vkCreateDevice failed. Device %s likely doesn't support vulkan video encoding\n", card_path); + goto done; + } + + uint32_t num_device_extensions = 0; + if(vkEnumerateDeviceExtensionProperties(physical_device, NULL, &num_device_extensions, NULL) != VK_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: vkEnumerateDeviceExtensionProperties (query num device extensions) failed\n"); + goto done; + } + + device_extensions = calloc(num_device_extensions, sizeof(VkExtensionProperties)); + if(!device_extensions) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: failed to allocate %d device extensions\n", num_device_extensions); + goto done; + } + + if(vkEnumerateDeviceExtensionProperties(physical_device, NULL, &num_device_extensions, device_extensions) != VK_SUCCESS) { + fprintf(stderr, "gsr error: gsr_get_supported_video_codecs_vulkan: vkEnumerateDeviceExtensionProperties (get data) failed\n"); + goto done; + } + + for(uint32_t i = 0; i < num_device_extensions; ++i) { + if(strcmp(device_extensions[i].extensionName, "VK_KHR_video_encode_h264") == 0) { + video_codecs->h264 = true; + } else if(strcmp(device_extensions[i].extensionName, "VK_KHR_video_encode_h265") == 0) { + // TODO: Verify if 10bit and hdr are actually supported + video_codecs->hevc = true; + video_codecs->hevc_10bit = true; + video_codecs->hevc_hdr = true; + } + } + + success = true; + + done: + if(cleanup) { + if(device) + vkDestroyDevice(device, NULL); + if(instance) + vkDestroyInstance(instance, NULL); + } + if(device_extensions) + free(device_extensions); + return success; +#else + // TODO: Low power query + video_codecs->h264 = (gsr_supported_video_codec){ true, false }; + video_codecs->hevc = (gsr_supported_video_codec){ true, false }; + return true; +#endif +} diff --git a/src/color_conversion.c b/src/color_conversion.c index 821ae52..88dc398 100644 --- a/src/color_conversion.c +++ b/src/color_conversion.c @@ -1,144 +1,417 @@ #include "../include/color_conversion.h" +#include "../include/egl.h" #include <stdio.h> -#include <string.h> #include <math.h> +#include <string.h> #include <assert.h> -#define MAX_SHADERS 2 -#define MAX_FRAMEBUFFERS 2 +#define COMPUTE_SHADER_INDEX_Y 0 +#define COMPUTE_SHADER_INDEX_UV 1 +#define COMPUTE_SHADER_INDEX_Y_EXTERNAL 2 +#define COMPUTE_SHADER_INDEX_UV_EXTERNAL 3 +#define COMPUTE_SHADER_INDEX_RGB 4 +#define COMPUTE_SHADER_INDEX_RGB_EXTERNAL 5 +#define COMPUTE_SHADER_INDEX_Y_BLEND 6 +#define COMPUTE_SHADER_INDEX_UV_BLEND 7 +#define COMPUTE_SHADER_INDEX_Y_EXTERNAL_BLEND 8 +#define COMPUTE_SHADER_INDEX_UV_EXTERNAL_BLEND 9 +#define COMPUTE_SHADER_INDEX_RGB_BLEND 10 +#define COMPUTE_SHADER_INDEX_RGB_EXTERNAL_BLEND 11 + +#define GRAPHICS_SHADER_INDEX_Y 0 +#define GRAPHICS_SHADER_INDEX_UV 1 +#define GRAPHICS_SHADER_INDEX_Y_EXTERNAL 2 +#define GRAPHICS_SHADER_INDEX_UV_EXTERNAL 3 +#define GRAPHICS_SHADER_INDEX_RGB 4 +#define GRAPHICS_SHADER_INDEX_RGB_EXTERNAL 5 + +/* https://en.wikipedia.org/wiki/YCbCr, see study/color_space_transform_matrix.png */ + +/* ITU-R BT2020, full */ +/* https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.2020-2-201510-I!!PDF-E.pdf */ +#define RGB_TO_P010_FULL "const mat4 RGBtoYUV = mat4(0.262700, -0.139630, 0.500000, 0.000000,\n" \ + " 0.678000, -0.360370, -0.459786, 0.000000,\n" \ + " 0.059300, 0.500000, -0.040214, 0.000000,\n" \ + " 0.000000, 0.500000, 0.500000, 1.000000);\n" + +/* ITU-R BT2020, limited (full multiplied by (235-16)/255, adding 16/255 to luma) */ +#define RGB_TO_P010_LIMITED "const mat4 RGBtoYUV = mat4(0.225613, -0.119918, 0.429412, 0.000000,\n" \ + " 0.582282, -0.309494, -0.394875, 0.000000,\n" \ + " 0.050928, 0.429412, -0.034537, 0.000000,\n" \ + " 0.062745, 0.500000, 0.500000, 1.000000);\n" + +/* ITU-R BT709, full, custom values: 0.2110 0.7110 0.0710 */ +/* https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.709-6-201506-I!!PDF-E.pdf */ +#define RGB_TO_NV12_FULL "const mat4 RGBtoYUV = mat4(0.211000, -0.113563, 0.500000, 0.000000,\n" \ + " 0.711000, -0.382670, -0.450570, 0.000000,\n" \ + " 0.071000, 0.500000, -0.044994, 0.000000,\n" \ + " 0.000000, 0.500000, 0.500000, 1.000000);\n" + +/* ITU-R BT709, limited, custom values: 0.2100 0.7100 0.0700 (full multiplied by (235-16)/255, adding 16/255 to luma) */ +#define RGB_TO_NV12_LIMITED "const mat4 RGBtoYUV = mat4(0.180353, -0.096964, 0.429412, 0.000000,\n" \ + " 0.609765, -0.327830, -0.385927, 0.000000,\n" \ + " 0.060118, 0.429412, -0.038049, 0.000000,\n" \ + " 0.062745, 0.500000, 0.500000, 1.000000);\n" + +static int max_int(int a, int b) { + return a > b ? a : b; +} -static float abs_f(float v) { - return v >= 0.0f ? v : -v; +static const char* color_format_range_get_transform_matrix(gsr_destination_color color_format, gsr_color_range color_range) { + switch(color_format) { + case GSR_DESTINATION_COLOR_NV12: { + switch(color_range) { + case GSR_COLOR_RANGE_LIMITED: + return RGB_TO_NV12_LIMITED; + case GSR_COLOR_RANGE_FULL: + return RGB_TO_NV12_FULL; + } + break; + } + case GSR_DESTINATION_COLOR_P010: { + switch(color_range) { + case GSR_COLOR_RANGE_LIMITED: + return RGB_TO_P010_LIMITED; + case GSR_COLOR_RANGE_FULL: + return RGB_TO_P010_FULL; + } + break; + } + case GSR_DESTINATION_COLOR_RGB8: + return ""; + default: + return NULL; + } + return NULL; } -#define ROTATE_Z "mat4 rotate_z(in float angle) {\n" \ - " return mat4(cos(angle), -sin(angle), 0.0, 0.0,\n" \ - " sin(angle), cos(angle), 0.0, 0.0,\n" \ - " 0.0, 0.0, 1.0, 0.0,\n" \ - " 0.0, 0.0, 0.0, 1.0);\n" \ - "}\n" +static void get_compute_shader_header(char *header, size_t header_size, bool external_texture) { + if(external_texture) { + snprintf(header, header_size, + "#version 310 es\n" + "#extension GL_ARB_compute_shader: enable\n" + "#extension GL_OES_EGL_image_external : enable\n" + "#extension GL_OES_EGL_image_external_essl3 : require\n" + "layout(binding = 0) uniform highp samplerExternalOES img_input;\n" + "layout(binding = 1) uniform highp sampler2D img_background;\n"); + } else { + snprintf(header, header_size, + "#version 420\n" + "#extension GL_ARB_compute_shader: enable\n" + "layout(binding = 0) uniform highp sampler2D img_input;\n" + "layout(binding = 1) uniform highp sampler2D img_background;\n"); + } +} -/* BT709 limited */ -#define RGB_TO_YUV "const mat4 RGBtoYUV = mat4(0.1826, -0.1006, 0.4392, 0.0,\n" \ - " 0.6142, -0.3386, -0.3989, 0.0,\n" \ - " 0.0620, 0.4392, -0.0403, 0.0,\n" \ - " 0.0625, 0.5000, 0.5000, 1.0);" +static int load_compute_shader_y(gsr_shader *shader, gsr_egl *egl, gsr_color_compute_uniforms *uniforms, int max_local_size_dim, gsr_destination_color color_format, gsr_color_range color_range, bool external_texture, bool alpha_blending) { + const char *color_transform_matrix = color_format_range_get_transform_matrix(color_format, color_range); + + char header[512]; + get_compute_shader_header(header, sizeof(header), external_texture); + + char compute_shader[2048]; + snprintf(compute_shader, sizeof(compute_shader), + "%s" + "layout (local_size_x = %d, local_size_y = %d, local_size_z = 1) in;\n" + "precision highp float;\n" + "uniform ivec2 source_position;\n" + "uniform ivec2 target_position;\n" + "uniform vec2 scale;\n" + "uniform mat2 rotation_matrix;\n" + "layout(rgba8, binding = 0) writeonly uniform highp image2D img_output;\n" + "%s" + "void main() {\n" + " ivec2 texel_coord = ivec2(gl_GlobalInvocationID.xy);\n" + " ivec2 size = ivec2(vec2(textureSize(img_input, 0)) * scale + 0.5);\n" + " ivec2 size_shift = size >> 1;\n" // size/2 + " ivec2 output_size = textureSize(img_background, 0);\n" + " vec2 rotated_texel_coord = vec2(texel_coord - source_position - size_shift) * rotation_matrix + vec2(size_shift) + 0.5;\n" + " vec2 output_texel_coord = vec2(texel_coord - source_position + target_position) + 0.5;\n" + " vec4 source_color = texture(img_input, rotated_texel_coord/vec2(size));\n" + " vec4 source_color_yuv = RGBtoYUV * vec4(source_color.rgb, 1.0);\n" + " vec4 output_color_yuv = %s;\n" + " float y_color = mix(output_color_yuv.r, source_color_yuv.r, source_color.a);\n" + " imageStore(img_output, texel_coord + target_position, vec4(y_color, 1.0, 1.0, 1.0));\n" + "}\n", header, max_local_size_dim, max_local_size_dim, color_transform_matrix, + alpha_blending ? "texture(img_background, output_texel_coord/vec2(output_size))" : "source_color_yuv"); + + if(gsr_shader_init(shader, egl, NULL, NULL, compute_shader) != 0) + return -1; -static int load_shader_bgr(gsr_shader *shader, gsr_egl *egl, int *rotation_uniform) { - char vertex_shader[2048]; - snprintf(vertex_shader, sizeof(vertex_shader), - "#version 300 es \n" - "in vec2 pos; \n" - "in vec2 texcoords; \n" - "out vec2 texcoords_out; \n" - "uniform float rotation; \n" - ROTATE_Z - "void main() \n" - "{ \n" - " texcoords_out = texcoords; \n" - " gl_Position = vec4(pos.x, pos.y, 0.0, 1.0) * rotate_z(rotation); \n" - "} \n"); + uniforms->source_position = egl->glGetUniformLocation(shader->program_id, "source_position"); + uniforms->target_position = egl->glGetUniformLocation(shader->program_id, "target_position"); + uniforms->rotation_matrix = egl->glGetUniformLocation(shader->program_id, "rotation_matrix"); + uniforms->scale = egl->glGetUniformLocation(shader->program_id, "scale"); + return 0; +} - char fragment_shader[] = - "#version 300 es \n" - "precision mediump float; \n" - "in vec2 texcoords_out; \n" - "uniform sampler2D tex1; \n" - "out vec4 FragColor; \n" - "void main() \n" - "{ \n" - " FragColor = texture(tex1, texcoords_out).bgra; \n" - "} \n"; - - if(gsr_shader_init(shader, egl, vertex_shader, fragment_shader) != 0) +static int load_compute_shader_uv(gsr_shader *shader, gsr_egl *egl, gsr_color_compute_uniforms *uniforms, int max_local_size_dim, gsr_destination_color color_format, gsr_color_range color_range, bool external_texture, bool alpha_blending) { + const char *color_transform_matrix = color_format_range_get_transform_matrix(color_format, color_range); + + char header[512]; + get_compute_shader_header(header, sizeof(header), external_texture); + + char compute_shader[2048]; + snprintf(compute_shader, sizeof(compute_shader), + "%s" + "layout (local_size_x = %d, local_size_y = %d, local_size_z = 1) in;\n" + "precision highp float;\n" + "uniform ivec2 source_position;\n" + "uniform ivec2 target_position;\n" + "uniform vec2 scale;\n" + "uniform mat2 rotation_matrix;\n" + "layout(rgba8, binding = 0) writeonly uniform highp image2D img_output;\n" + "%s" + "void main() {\n" + " ivec2 texel_coord = ivec2(gl_GlobalInvocationID.xy);\n" + " ivec2 size = ivec2(vec2(textureSize(img_input, 0)) * scale + 0.5);\n" + " ivec2 size_shift = size >> 2;\n" // size/4 + " ivec2 output_size = textureSize(img_background, 0);\n" + " vec2 rotated_texel_coord = vec2(texel_coord - source_position - size_shift) * rotation_matrix + vec2(size_shift) + 0.5;\n" + " vec2 output_texel_coord = vec2(texel_coord - source_position + target_position) + 0.5;\n" + " vec4 source_color = texture(img_input, rotated_texel_coord/vec2(size>>1));\n" // size/2 + " vec4 source_color_yuv = RGBtoYUV * vec4(source_color.rgb, 1.0);\n" + " vec4 output_color_yuv = %s;\n" + " vec2 uv_color = mix(output_color_yuv.rg, source_color_yuv.gb, source_color.a);\n" + " imageStore(img_output, texel_coord + target_position, vec4(uv_color, 1.0, 1.0));\n" + "}\n", header, max_local_size_dim, max_local_size_dim, color_transform_matrix, + alpha_blending ? "texture(img_background, output_texel_coord/vec2(output_size))" : "source_color_yuv"); + + if(gsr_shader_init(shader, egl, NULL, NULL, compute_shader) != 0) return -1; - gsr_shader_bind_attribute_location(shader, "pos", 0); - gsr_shader_bind_attribute_location(shader, "texcoords", 1); - *rotation_uniform = egl->glGetUniformLocation(shader->program_id, "rotation"); + uniforms->source_position = egl->glGetUniformLocation(shader->program_id, "source_position"); + uniforms->target_position = egl->glGetUniformLocation(shader->program_id, "target_position"); + uniforms->rotation_matrix = egl->glGetUniformLocation(shader->program_id, "rotation_matrix"); + uniforms->scale = egl->glGetUniformLocation(shader->program_id, "scale"); + return 0; +} + +static int load_compute_shader_rgb(gsr_shader *shader, gsr_egl *egl, gsr_color_compute_uniforms *uniforms, int max_local_size_dim, bool external_texture, bool alpha_blending) { + char header[512]; + get_compute_shader_header(header, sizeof(header), external_texture); + + char compute_shader[2048]; + snprintf(compute_shader, sizeof(compute_shader), + "%s" + "layout (local_size_x = %d, local_size_y = %d, local_size_z = 1) in;\n" + "uniform ivec2 source_position;\n" + "uniform ivec2 target_position;\n" + "uniform vec2 scale;\n" + "uniform mat2 rotation_matrix;\n" + "layout(rgba8, binding = 0) writeonly uniform highp image2D img_output;\n" + "void main() {\n" + " ivec2 texel_coord = ivec2(gl_GlobalInvocationID.xy);\n" + " ivec2 size = ivec2(vec2(textureSize(img_input, 0)) * scale + 0.5);\n" + " ivec2 size_shift = size >> 1;\n" // size/2 + " ivec2 output_size = textureSize(img_background, 0);\n" + " vec2 rotated_texel_coord = vec2(texel_coord - source_position - size_shift) * rotation_matrix + vec2(size_shift) + 0.5;\n" + " vec2 output_texel_coord = vec2(texel_coord - source_position + target_position) + 0.5;\n" + " vec4 source_color = texture(img_input, rotated_texel_coord/vec2(size));\n" + " vec4 output_color = %s;\n" + " vec3 color = mix(output_color.rgb, source_color.rgb, source_color.a);\n" + " imageStore(img_output, texel_coord + target_position, vec4(color, 1.0));\n" + "}\n", header, max_local_size_dim, max_local_size_dim, + alpha_blending ? "texture(img_background, output_texel_coord/vec2(output_size))" : "source_color"); + + if(gsr_shader_init(shader, egl, NULL, NULL, compute_shader) != 0) + return -1; + + uniforms->source_position = egl->glGetUniformLocation(shader->program_id, "source_position"); + uniforms->target_position = egl->glGetUniformLocation(shader->program_id, "target_position"); + uniforms->rotation_matrix = egl->glGetUniformLocation(shader->program_id, "rotation_matrix"); + uniforms->scale = egl->glGetUniformLocation(shader->program_id, "scale"); return 0; } -static int load_shader_y(gsr_shader *shader, gsr_egl *egl, int *rotation_uniform) { +static int load_graphics_shader_y(gsr_shader *shader, gsr_egl *egl, gsr_color_graphics_uniforms *uniforms, gsr_destination_color color_format, gsr_color_range color_range, bool external_texture) { + const char *color_transform_matrix = color_format_range_get_transform_matrix(color_format, color_range); + char vertex_shader[2048]; snprintf(vertex_shader, sizeof(vertex_shader), "#version 300 es \n" "in vec2 pos; \n" "in vec2 texcoords; \n" "out vec2 texcoords_out; \n" + "uniform vec2 offset; \n" "uniform float rotation; \n" - ROTATE_Z + "uniform mat2 rotation_matrix; \n" "void main() \n" "{ \n" - " texcoords_out = texcoords; \n" - " gl_Position = vec4(pos.x, pos.y, 0.0, 1.0) * rotate_z(rotation); \n" + " texcoords_out = vec2(texcoords.x - 0.5, texcoords.y - 0.5) * rotation_matrix + vec2(0.5, 0.5); \n" + " gl_Position = vec4(offset.x, offset.y, 0.0, 0.0) + vec4(pos.x, pos.y, 0.0, 1.0); \n" "} \n"); - char fragment_shader[] = - "#version 300 es \n" - "precision mediump float; \n" - "in vec2 texcoords_out; \n" - "uniform sampler2D tex1; \n" - "out vec4 FragColor; \n" - RGB_TO_YUV - "void main() \n" - "{ \n" - " vec4 pixel = texture(tex1, texcoords_out); \n" - " FragColor.x = (RGBtoYUV * vec4(pixel.rgb, 1.0)).x; \n" - " FragColor.w = pixel.a; \n" - "} \n"; - - if(gsr_shader_init(shader, egl, vertex_shader, fragment_shader) != 0) + const char *main_code = + main_code = + " vec4 pixel = texture(tex1, texcoords_out); \n" + " FragColor.x = (RGBtoYUV * vec4(pixel.rgb, 1.0)).x; \n" + " FragColor.w = pixel.a; \n"; + + char fragment_shader[2048]; + if(external_texture) { + snprintf(fragment_shader, sizeof(fragment_shader), + "#version 300 es \n" + "#extension GL_OES_EGL_image_external : enable \n" + "#extension GL_OES_EGL_image_external_essl3 : require \n" + "precision highp float; \n" + "in vec2 texcoords_out; \n" + "uniform samplerExternalOES tex1; \n" + "out vec4 FragColor; \n" + "%s" + "void main() \n" + "{ \n" + "%s" + "} \n", color_transform_matrix, main_code); + } else { + snprintf(fragment_shader, sizeof(fragment_shader), + "#version 300 es \n" + "precision highp float; \n" + "in vec2 texcoords_out; \n" + "uniform sampler2D tex1; \n" + "out vec4 FragColor; \n" + "%s" + "void main() \n" + "{ \n" + "%s" + "} \n", color_transform_matrix, main_code); + } + + if(gsr_shader_init(shader, egl, vertex_shader, fragment_shader, NULL) != 0) return -1; gsr_shader_bind_attribute_location(shader, "pos", 0); gsr_shader_bind_attribute_location(shader, "texcoords", 1); - *rotation_uniform = egl->glGetUniformLocation(shader->program_id, "rotation"); + uniforms->offset = egl->glGetUniformLocation(shader->program_id, "offset"); + uniforms->rotation_matrix = egl->glGetUniformLocation(shader->program_id, "rotation_matrix"); return 0; } -static unsigned int load_shader_uv(gsr_shader *shader, gsr_egl *egl, int *rotation_uniform) { +static unsigned int load_graphics_shader_uv(gsr_shader *shader, gsr_egl *egl, gsr_color_graphics_uniforms *uniforms, gsr_destination_color color_format, gsr_color_range color_range, bool external_texture) { + const char *color_transform_matrix = color_format_range_get_transform_matrix(color_format, color_range); + char vertex_shader[2048]; snprintf(vertex_shader, sizeof(vertex_shader), "#version 300 es \n" "in vec2 pos; \n" "in vec2 texcoords; \n" "out vec2 texcoords_out; \n" + "uniform vec2 offset; \n" "uniform float rotation; \n" - ROTATE_Z + "uniform mat2 rotation_matrix; \n" "void main() \n" "{ \n" - " texcoords_out = texcoords; \n" - " gl_Position = vec4(pos.x, pos.y, 0.0, 1.0) * rotate_z(rotation) * vec4(0.5, 0.5, 1.0, 1.0) - vec4(0.5, 0.5, 0.0, 0.0); \n" + " texcoords_out = vec2(texcoords.x - 0.5, texcoords.y - 0.5) * rotation_matrix + vec2(0.5, 0.5); \n" + " gl_Position = (vec4(offset.x, offset.y, 0.0, 0.0) + vec4(pos.x, pos.y, 0.0, 1.0)) * vec4(0.5, 0.5, 1.0, 1.0) - vec4(0.5, 0.5, 0.0, 0.0); \n" "} \n"); - char fragment_shader[] = - "#version 300 es \n" - "precision mediump float; \n" - "in vec2 texcoords_out; \n" - "uniform sampler2D tex1; \n" - "out vec4 FragColor; \n" - RGB_TO_YUV - "void main() \n" - "{ \n" - " vec4 pixel = texture(tex1, texcoords_out); \n" - " FragColor.xy = (RGBtoYUV * vec4(pixel.rgb, 1.0)).yz; \n" - " FragColor.w = pixel.a; \n" - "} \n"; - - if(gsr_shader_init(shader, egl, vertex_shader, fragment_shader) != 0) + const char *main_code = + main_code = + " vec4 pixel = texture(tex1, texcoords_out); \n" + " FragColor.xy = (RGBtoYUV * vec4(pixel.rgb, 1.0)).yz; \n" + " FragColor.w = pixel.a; \n"; + + char fragment_shader[2048]; + if(external_texture) { + snprintf(fragment_shader, sizeof(fragment_shader), + "#version 300 es \n" + "#extension GL_OES_EGL_image_external : enable \n" + "#extension GL_OES_EGL_image_external_essl3 : require \n" + "precision highp float; \n" + "in vec2 texcoords_out; \n" + "uniform samplerExternalOES tex1; \n" + "out vec4 FragColor; \n" + "%s" + "void main() \n" + "{ \n" + "%s" + "} \n", color_transform_matrix, main_code); + } else { + snprintf(fragment_shader, sizeof(fragment_shader), + "#version 300 es \n" + "precision highp float; \n" + "in vec2 texcoords_out; \n" + "uniform sampler2D tex1; \n" + "out vec4 FragColor; \n" + "%s" + "void main() \n" + "{ \n" + "%s" + "} \n", color_transform_matrix, main_code); + } + + if(gsr_shader_init(shader, egl, vertex_shader, fragment_shader, NULL) != 0) return -1; gsr_shader_bind_attribute_location(shader, "pos", 0); gsr_shader_bind_attribute_location(shader, "texcoords", 1); - *rotation_uniform = egl->glGetUniformLocation(shader->program_id, "rotation"); + uniforms->offset = egl->glGetUniformLocation(shader->program_id, "offset"); + uniforms->rotation_matrix = egl->glGetUniformLocation(shader->program_id, "rotation_matrix"); + return 0; +} + +static unsigned int load_graphics_shader_rgb(gsr_shader *shader, gsr_egl *egl, gsr_color_graphics_uniforms *uniforms, bool external_texture) { + char vertex_shader[2048]; + snprintf(vertex_shader, sizeof(vertex_shader), + "#version 300 es \n" + "in vec2 pos; \n" + "in vec2 texcoords; \n" + "out vec2 texcoords_out; \n" + "uniform vec2 offset; \n" + "uniform float rotation; \n" + "uniform mat2 rotation_matrix; \n" + "void main() \n" + "{ \n" + " texcoords_out = vec2(texcoords.x - 0.5, texcoords.y - 0.5) * rotation_matrix + vec2(0.5, 0.5); \n" + " gl_Position = vec4(offset.x, offset.y, 0.0, 0.0) + vec4(pos.x, pos.y, 0.0, 1.0); \n" + "} \n"); + + const char *main_code = + main_code = + " vec4 pixel = texture(tex1, texcoords_out); \n" + " FragColor = pixel; \n"; + + char fragment_shader[2048]; + if(external_texture) { + snprintf(fragment_shader, sizeof(fragment_shader), + "#version 300 es \n" + "#extension GL_OES_EGL_image_external : enable \n" + "#extension GL_OES_EGL_image_external_essl3 : require \n" + "precision highp float; \n" + "in vec2 texcoords_out; \n" + "uniform samplerExternalOES tex1; \n" + "out vec4 FragColor; \n" + "void main() \n" + "{ \n" + "%s" + "} \n", main_code); + } else { + snprintf(fragment_shader, sizeof(fragment_shader), + "#version 300 es \n" + "precision highp float; \n" + "in vec2 texcoords_out; \n" + "uniform sampler2D tex1; \n" + "out vec4 FragColor; \n" + "void main() \n" + "{ \n" + "%s" + "} \n", main_code); + } + + if(gsr_shader_init(shader, egl, vertex_shader, fragment_shader, NULL) != 0) + return -1; + + gsr_shader_bind_attribute_location(shader, "pos", 0); + gsr_shader_bind_attribute_location(shader, "texcoords", 1); + uniforms->offset = egl->glGetUniformLocation(shader->program_id, "offset"); + uniforms->rotation_matrix = egl->glGetUniformLocation(shader->program_id, "rotation_matrix"); return 0; } static int load_framebuffers(gsr_color_conversion *self) { /* TODO: Only generate the necessary amount of framebuffers (self->params.num_destination_textures) */ const unsigned int draw_buffer = GL_COLOR_ATTACHMENT0; - self->params.egl->glGenFramebuffers(MAX_FRAMEBUFFERS, self->framebuffers); + self->params.egl->glGenFramebuffers(GSR_COLOR_CONVERSION_MAX_FRAMEBUFFERS, self->framebuffers); self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, self->framebuffers[0]); self->params.egl->glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, self->params.destination_textures[0], 0); @@ -172,7 +445,7 @@ static int create_vertices(gsr_color_conversion *self) { self->params.egl->glGenBuffers(1, &self->vertex_buffer_object_id); self->params.egl->glBindBuffer(GL_ARRAY_BUFFER, self->vertex_buffer_object_id); - self->params.egl->glBufferData(GL_ARRAY_BUFFER, 24 * sizeof(float), NULL, GL_STREAM_DRAW); + self->params.egl->glBufferData(GL_ARRAY_BUFFER, 24 * sizeof(float), NULL, GL_DYNAMIC_DRAW); self->params.egl->glEnableVertexAttribArray(0); self->params.egl->glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, 4 * sizeof(float), (void*)0); @@ -184,42 +457,182 @@ static int create_vertices(gsr_color_conversion *self) { return 0; } +static bool gsr_color_conversion_load_compute_shaders(gsr_color_conversion *self) { + switch(self->params.destination_color) { + case GSR_DESTINATION_COLOR_NV12: + case GSR_DESTINATION_COLOR_P010: { + if(load_compute_shader_y(&self->compute_shaders[COMPUTE_SHADER_INDEX_Y], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_Y], self->max_local_size_dim, self->params.destination_color, self->params.color_range, false, false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y compute shader\n"); + return false; + } + + if(load_compute_shader_uv(&self->compute_shaders[COMPUTE_SHADER_INDEX_UV], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_UV], self->max_local_size_dim, self->params.destination_color, self->params.color_range, false, false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load UV compute shader\n"); + return false; + } + + if(load_compute_shader_y(&self->compute_shaders[COMPUTE_SHADER_INDEX_Y_BLEND], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_Y_BLEND], self->max_local_size_dim, self->params.destination_color, self->params.color_range, false, true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y compute shader\n"); + return false; + } + + if(load_compute_shader_uv(&self->compute_shaders[COMPUTE_SHADER_INDEX_UV_BLEND], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_UV_BLEND], self->max_local_size_dim, self->params.destination_color, self->params.color_range, false, true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load UV compute shader\n"); + return false; + } + break; + } + case GSR_DESTINATION_COLOR_RGB8: { + if(load_compute_shader_rgb(&self->compute_shaders[COMPUTE_SHADER_INDEX_RGB], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_RGB], self->max_local_size_dim, false, false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y compute shader\n"); + return false; + } + + if(load_compute_shader_rgb(&self->compute_shaders[COMPUTE_SHADER_INDEX_RGB_BLEND], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_RGB_BLEND], self->max_local_size_dim, false, true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y compute shader\n"); + return false; + } + break; + } + } + return true; +} + +static bool gsr_color_conversion_load_external_compute_shaders(gsr_color_conversion *self) { + switch(self->params.destination_color) { + case GSR_DESTINATION_COLOR_NV12: + case GSR_DESTINATION_COLOR_P010: { + if(load_compute_shader_y(&self->compute_shaders[COMPUTE_SHADER_INDEX_Y_EXTERNAL], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_Y_EXTERNAL], self->max_local_size_dim, self->params.destination_color, self->params.color_range, true, false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y compute shader\n"); + return false; + } + + if(load_compute_shader_uv(&self->compute_shaders[COMPUTE_SHADER_INDEX_UV_EXTERNAL], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_UV_EXTERNAL], self->max_local_size_dim, self->params.destination_color, self->params.color_range, true, false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load UV compute shader\n"); + return false; + } + + if(load_compute_shader_y(&self->compute_shaders[COMPUTE_SHADER_INDEX_Y_EXTERNAL_BLEND], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_Y_EXTERNAL_BLEND], self->max_local_size_dim, self->params.destination_color, self->params.color_range, true, true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y compute shader\n"); + return false; + } + + if(load_compute_shader_uv(&self->compute_shaders[COMPUTE_SHADER_INDEX_UV_EXTERNAL_BLEND], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_UV_EXTERNAL_BLEND], self->max_local_size_dim, self->params.destination_color, self->params.color_range, true, true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load UV compute shader\n"); + return false; + } + break; + } + case GSR_DESTINATION_COLOR_RGB8: { + if(load_compute_shader_rgb(&self->compute_shaders[COMPUTE_SHADER_INDEX_RGB_EXTERNAL], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_RGB_EXTERNAL], self->max_local_size_dim, true, false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y compute shader\n"); + return false; + } + + if(load_compute_shader_rgb(&self->compute_shaders[COMPUTE_SHADER_INDEX_RGB_EXTERNAL_BLEND], self->params.egl, &self->compute_uniforms[COMPUTE_SHADER_INDEX_RGB_EXTERNAL_BLEND], self->max_local_size_dim, true, true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y compute shader\n"); + return false; + } + break; + } + } + return true; +} + +static bool gsr_color_conversion_load_graphics_shaders(gsr_color_conversion *self) { + switch(self->params.destination_color) { + case GSR_DESTINATION_COLOR_NV12: + case GSR_DESTINATION_COLOR_P010: { + if(load_graphics_shader_y(&self->graphics_shaders[GRAPHICS_SHADER_INDEX_Y], self->params.egl, &self->graphics_uniforms[GRAPHICS_SHADER_INDEX_Y], self->params.destination_color, self->params.color_range, false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y graphics shader\n"); + return false; + } + + if(load_graphics_shader_uv(&self->graphics_shaders[GRAPHICS_SHADER_INDEX_UV], self->params.egl, &self->graphics_uniforms[GRAPHICS_SHADER_INDEX_UV], self->params.destination_color, self->params.color_range, false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load UV graphics shader\n"); + return false; + } + break; + } + case GSR_DESTINATION_COLOR_RGB8: { + if(load_graphics_shader_rgb(&self->graphics_shaders[GRAPHICS_SHADER_INDEX_RGB], self->params.egl, &self->graphics_uniforms[GRAPHICS_SHADER_INDEX_RGB], false) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y graphics shader\n"); + return false; + } + break; + } + } + return true; +} + +static bool gsr_color_conversion_load_external_graphics_shaders(gsr_color_conversion *self) { + switch(self->params.destination_color) { + case GSR_DESTINATION_COLOR_NV12: + case GSR_DESTINATION_COLOR_P010: { + if(load_graphics_shader_y(&self->graphics_shaders[GRAPHICS_SHADER_INDEX_Y_EXTERNAL], self->params.egl, &self->graphics_uniforms[GRAPHICS_SHADER_INDEX_Y_EXTERNAL], self->params.destination_color, self->params.color_range, true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y graphics shader\n"); + return false; + } + + if(load_graphics_shader_uv(&self->graphics_shaders[GRAPHICS_SHADER_INDEX_UV_EXTERNAL], self->params.egl, &self->graphics_uniforms[GRAPHICS_SHADER_INDEX_UV_EXTERNAL], self->params.destination_color, self->params.color_range, true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load UV graphics shader\n"); + return false; + } + break; + } + case GSR_DESTINATION_COLOR_RGB8: { + if(load_graphics_shader_rgb(&self->graphics_shaders[GRAPHICS_SHADER_INDEX_RGB_EXTERNAL], self->params.egl, &self->graphics_uniforms[GRAPHICS_SHADER_INDEX_RGB_EXTERNAL], true) != 0) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y graphics shader\n"); + return false; + } + break; + } + } + return true; +} + int gsr_color_conversion_init(gsr_color_conversion *self, const gsr_color_conversion_params *params) { assert(params); assert(params->egl); memset(self, 0, sizeof(*self)); self->params.egl = params->egl; self->params = *params; - - switch(params->destination_color) { - case GSR_DESTINATION_COLOR_BGR: { - if(self->params.num_destination_textures != 1) { - fprintf(stderr, "gsr error: gsr_color_conversion_init: expected 1 destination texture for destination color BGR, got %d destination texture(s)\n", self->params.num_destination_textures); - return -1; - } - - if(load_shader_bgr(&self->shaders[0], self->params.egl, &self->rotation_uniforms[0]) != 0) { - fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load bgr shader\n"); + + int max_compute_work_group_invocations = 256; + self->params.egl->glGetIntegerv(GL_MAX_COMPUTE_FIXED_GROUP_INVOCATIONS, &max_compute_work_group_invocations); + self->max_local_size_dim = sqrt(max_compute_work_group_invocations); + + switch(self->params.destination_color) { + case GSR_DESTINATION_COLOR_NV12: + case GSR_DESTINATION_COLOR_P010: { + if(self->params.num_destination_textures != 2) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: expected 2 destination textures for destination color NV12/P010, got %d destination texture(s)\n", self->params.num_destination_textures); goto err; } break; } - case GSR_DESTINATION_COLOR_NV12: { - if(self->params.num_destination_textures != 2) { - fprintf(stderr, "gsr error: gsr_color_conversion_init: expected 2 destination textures for destination color NV12, got %d destination texture(s)\n", self->params.num_destination_textures); - return -1; - } - - if(load_shader_y(&self->shaders[0], self->params.egl, &self->rotation_uniforms[0]) != 0) { - fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load Y shader\n"); + case GSR_DESTINATION_COLOR_RGB8: { + if(self->params.num_destination_textures != 1) { + fprintf(stderr, "gsr error: gsr_color_conversion_init: expected 1 destination textures for destination color RGB8, got %d destination texture(s)\n", self->params.num_destination_textures); goto err; } + break; + } + } - if(load_shader_uv(&self->shaders[1], self->params.egl, &self->rotation_uniforms[1]) != 0) { - fprintf(stderr, "gsr error: gsr_color_conversion_init: failed to load UV shader\n"); + if(!gsr_color_conversion_load_compute_shaders(self)) { + self->compute_shaders_failed_to_load = true; + fprintf(stderr, "gsr info: failed to load one or more compute shaders, run gpu-screen-recorder with the '-gl-debug yes' option to see why. Falling back to slower graphics shader instead\n"); + if(!gsr_color_conversion_load_graphics_shaders(self)) + goto err; + } + + if(self->params.load_external_image_shader) { + if(!gsr_color_conversion_load_external_compute_shaders(self)) { + self->external_compute_shaders_failed_to_load = true; + fprintf(stderr, "gsr info: failed to load one or more external compute shaders, run gpu-screen-recorder with the '-gl-debug yes' option to see why. Falling back to slower graphics shader instead\n"); + if(!gsr_color_conversion_load_external_graphics_shaders(self)) goto err; - } - break; } } @@ -250,97 +663,341 @@ void gsr_color_conversion_deinit(gsr_color_conversion *self) { self->vertex_array_object_id = 0; } - self->params.egl->glDeleteFramebuffers(MAX_FRAMEBUFFERS, self->framebuffers); - for(int i = 0; i < MAX_FRAMEBUFFERS; ++i) { + self->params.egl->glDeleteFramebuffers(GSR_COLOR_CONVERSION_MAX_FRAMEBUFFERS, self->framebuffers); + for(int i = 0; i < GSR_COLOR_CONVERSION_MAX_FRAMEBUFFERS; ++i) { self->framebuffers[i] = 0; } - for(int i = 0; i < MAX_SHADERS; ++i) { - gsr_shader_deinit(&self->shaders[i]); + for(int i = 0; i < GSR_COLOR_CONVERSION_MAX_COMPUTE_SHADERS; ++i) { + gsr_shader_deinit(&self->compute_shaders[i]); + } + + for(int i = 0; i < GSR_COLOR_CONVERSION_MAX_GRAPHICS_SHADERS; ++i) { + gsr_shader_deinit(&self->graphics_shaders[i]); } self->params.egl = NULL; } -/* |source_pos| is in pixel coordinates and |source_size| */ -int gsr_color_conversion_draw(gsr_color_conversion *self, unsigned int texture_id, vec2i source_pos, vec2i source_size, vec2i texture_pos, vec2i texture_size, float rotation) { +static void gsr_color_conversion_apply_rotation(gsr_rotation rotation, float rotation_matrix[2][2]) { + /* + rotation_matrix[0][0] = cos(angle); + rotation_matrix[0][1] = -sin(angle); + rotation_matrix[1][0] = sin(angle); + rotation_matrix[1][1] = cos(angle); + The manual matrix code below is the same as this code above, but without floating-point errors. + This is done to remove any blurring caused by these floating-point errors. + */ + switch(rotation) { + case GSR_ROT_0: + rotation_matrix[0][0] = 1.0f; + rotation_matrix[0][1] = 0.0f; + rotation_matrix[1][0] = 0.0f; + rotation_matrix[1][1] = 1.0f; + break; + case GSR_ROT_90: + rotation_matrix[0][0] = 0.0f; + rotation_matrix[0][1] = -1.0f; + rotation_matrix[1][0] = 1.0f; + rotation_matrix[1][1] = 0.0f; + break; + case GSR_ROT_180: + rotation_matrix[0][0] = -1.0f; + rotation_matrix[0][1] = 0.0f; + rotation_matrix[1][0] = 0.0f; + rotation_matrix[1][1] = -1.0f; + break; + case GSR_ROT_270: + rotation_matrix[0][0] = 0.0f; + rotation_matrix[0][1] = 1.0f; + rotation_matrix[1][0] = -1.0f; + rotation_matrix[1][1] = 0.0f; + break; + } +} + +static void gsr_color_conversion_swizzle_texture_source(gsr_color_conversion *self, gsr_source_color source_color) { + if(source_color == GSR_SOURCE_COLOR_BGR) { + const int swizzle_mask[] = { GL_BLUE, GL_GREEN, GL_RED, 1 }; + self->params.egl->glTexParameteriv(GL_TEXTURE_2D, GL_TEXTURE_SWIZZLE_RGBA, swizzle_mask); + } +} + +static void gsr_color_conversion_swizzle_reset(gsr_color_conversion *self, gsr_source_color source_color) { + if(source_color == GSR_SOURCE_COLOR_BGR) { + const int swizzle_mask[] = { GL_RED, GL_GREEN, GL_BLUE, GL_ALPHA }; + self->params.egl->glTexParameteriv(GL_TEXTURE_2D, GL_TEXTURE_SWIZZLE_RGBA, swizzle_mask); + } +} + +typedef enum { + GSR_COLOR_COMP_Y, + GSR_COLOR_COMP_UV, + GSR_COLOR_COMP_RGB +} gsr_color_component; + +static int color_component_get_destination_texture_index(gsr_color_component color_component) { + switch(color_component) { + case GSR_COLOR_COMP_Y: return 0; + case GSR_COLOR_COMP_UV: return 1; + case GSR_COLOR_COMP_RGB: return 0; + } + assert(false); + return 0; +} + +static unsigned int color_component_get_color_format(gsr_color_component color_component, bool use_16bit_colors) { + switch(color_component) { + case GSR_COLOR_COMP_Y: return use_16bit_colors ? GL_R16 : GL_R8; + case GSR_COLOR_COMP_UV: return use_16bit_colors ? GL_RG16 : GL_RG8; + case GSR_COLOR_COMP_RGB: return GL_RGBA8; // TODO: 16-bit color support + } + assert(false); + return GL_RGBA8; +} + +static int color_component_get_COMPUTE_SHADER_INDEX(gsr_color_component color_component, bool external_texture, bool alpha_blending) { + switch(color_component) { + case GSR_COLOR_COMP_Y: { + if(external_texture) + return alpha_blending ? COMPUTE_SHADER_INDEX_Y_EXTERNAL_BLEND : COMPUTE_SHADER_INDEX_Y_EXTERNAL; + else + return alpha_blending ? COMPUTE_SHADER_INDEX_Y_BLEND : COMPUTE_SHADER_INDEX_Y; + } + case GSR_COLOR_COMP_UV: { + if(external_texture) + return alpha_blending ? COMPUTE_SHADER_INDEX_UV_EXTERNAL_BLEND : COMPUTE_SHADER_INDEX_UV_EXTERNAL; + else + return alpha_blending ? COMPUTE_SHADER_INDEX_UV_BLEND : COMPUTE_SHADER_INDEX_UV; + } + case GSR_COLOR_COMP_RGB: { + if(external_texture) + return alpha_blending ? COMPUTE_SHADER_INDEX_RGB_EXTERNAL_BLEND : COMPUTE_SHADER_INDEX_RGB_EXTERNAL; + else + return alpha_blending ? COMPUTE_SHADER_INDEX_RGB_BLEND : COMPUTE_SHADER_INDEX_RGB; + } + } + assert(false); + return COMPUTE_SHADER_INDEX_RGB; +} + +static void gsr_color_conversion_dispatch_compute_shader(gsr_color_conversion *self, bool external_texture, bool alpha_blending, float rotation_matrix[2][2], vec2i source_position, vec2i destination_pos, vec2i destination_size, vec2f scale, bool use_16bit_colors, gsr_color_component color_component) { + const int compute_shader_index = color_component_get_COMPUTE_SHADER_INDEX(color_component, external_texture, alpha_blending); + const int destination_texture_index = color_component_get_destination_texture_index(color_component); + const unsigned int color_format = color_component_get_color_format(color_component, use_16bit_colors); + + self->params.egl->glActiveTexture(GL_TEXTURE1); + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->params.destination_textures[destination_texture_index]); + self->params.egl->glActiveTexture(GL_TEXTURE0); + + gsr_color_compute_uniforms *uniform = &self->compute_uniforms[compute_shader_index]; + gsr_shader_use(&self->compute_shaders[compute_shader_index]); + self->params.egl->glUniformMatrix2fv(uniform->rotation_matrix, 1, GL_TRUE, (const float*)rotation_matrix); + self->params.egl->glUniform2i(uniform->source_position, source_position.x, source_position.y); + self->params.egl->glUniform2i(uniform->target_position, destination_pos.x, destination_pos.y); + self->params.egl->glUniform2f(uniform->scale, scale.x, scale.y); + self->params.egl->glBindImageTexture(0, self->params.destination_textures[destination_texture_index], 0, GL_FALSE, 0, GL_WRITE_ONLY, color_format); + const double num_groups_x = ceil((double)destination_size.x/(double)self->max_local_size_dim); + const double num_groups_y = ceil((double)destination_size.y/(double)self->max_local_size_dim); + self->params.egl->glDispatchCompute(max_int(1, num_groups_x), max_int(1, num_groups_y), 1); +} + +static void gsr_color_conversion_draw_graphics(gsr_color_conversion *self, unsigned int texture_id, bool external_texture, float rotation_matrix[2][2], vec2i source_position, vec2i source_size, vec2i destination_pos, vec2i texture_size, vec2f scale, gsr_source_color source_color) { /* TODO: Do not call this every frame? */ vec2i dest_texture_size = {0, 0}; self->params.egl->glBindTexture(GL_TEXTURE_2D, self->params.destination_textures[0]); self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &dest_texture_size.x); self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &dest_texture_size.y); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); - /* TODO: Do not call this every frame? */ - vec2i source_texture_size = {0, 0}; - self->params.egl->glBindTexture(GL_TEXTURE_2D, texture_id); - self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &source_texture_size.x); - self->params.egl->glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &source_texture_size.y); - - if(abs_f(M_PI * 0.5f - rotation) <= 0.001f || abs_f(M_PI * 1.5f - rotation) <= 0.001f) { - float tmp = source_texture_size.x; - source_texture_size.x = source_texture_size.y; - source_texture_size.y = tmp; - } + const int texture_target = external_texture ? GL_TEXTURE_EXTERNAL_OES : GL_TEXTURE_2D; + + self->params.egl->glBindTexture(texture_target, texture_id); + gsr_color_conversion_swizzle_texture_source(self, source_color); const vec2f pos_norm = { - ((float)source_pos.x / (dest_texture_size.x == 0 ? 1.0f : (float)dest_texture_size.x)) * 2.0f, - ((float)source_pos.y / (dest_texture_size.y == 0 ? 1.0f : (float)dest_texture_size.y)) * 2.0f, + ((float)destination_pos.x / (dest_texture_size.x == 0 ? 1.0f : (float)dest_texture_size.x)) * 2.0f, + ((float)destination_pos.y / (dest_texture_size.y == 0 ? 1.0f : (float)dest_texture_size.y)) * 2.0f, }; const vec2f size_norm = { - ((float)source_size.x / (dest_texture_size.x == 0 ? 1.0f : (float)dest_texture_size.x)) * 2.0f, - ((float)source_size.y / (dest_texture_size.y == 0 ? 1.0f : (float)dest_texture_size.y)) * 2.0f, + ((float)source_size.x / (dest_texture_size.x == 0 ? 1.0f : (float)dest_texture_size.x)) * 2.0f * scale.x, + ((float)source_size.y / (dest_texture_size.y == 0 ? 1.0f : (float)dest_texture_size.y)) * 2.0f * scale.y, }; const vec2f texture_pos_norm = { - (float)texture_pos.x / (source_texture_size.x == 0 ? 1.0f : (float)source_texture_size.x), - (float)texture_pos.y / (source_texture_size.y == 0 ? 1.0f : (float)source_texture_size.y), + (float)source_position.x / (texture_size.x == 0 ? 1.0f : (float)texture_size.x), + (float)source_position.y / (texture_size.y == 0 ? 1.0f : (float)texture_size.y), }; const vec2f texture_size_norm = { - (float)texture_size.x / (source_texture_size.x == 0 ? 1.0f : (float)source_texture_size.x), - (float)texture_size.y / (source_texture_size.y == 0 ? 1.0f : (float)source_texture_size.y), + (float)source_size.x / (texture_size.x == 0 ? 1.0f : (float)texture_size.x), + (float)source_size.y / (texture_size.y == 0 ? 1.0f : (float)texture_size.y), }; const float vertices[] = { - -1.0f + pos_norm.x, -1.0f + pos_norm.y + size_norm.y, texture_pos_norm.x, texture_pos_norm.y + texture_size_norm.y, - -1.0f + pos_norm.x, -1.0f + pos_norm.y, texture_pos_norm.x, texture_pos_norm.y, - -1.0f + pos_norm.x + size_norm.x, -1.0f + pos_norm.y, texture_pos_norm.x + texture_size_norm.x, texture_pos_norm.y, + -1.0f + 0.0f, -1.0f + 0.0f + size_norm.y, texture_pos_norm.x, texture_pos_norm.y + texture_size_norm.y, + -1.0f + 0.0f, -1.0f + 0.0f, texture_pos_norm.x, texture_pos_norm.y, + -1.0f + 0.0f + size_norm.x, -1.0f + 0.0f, texture_pos_norm.x + texture_size_norm.x, texture_pos_norm.y, - -1.0f + pos_norm.x, -1.0f + pos_norm.y + size_norm.y, texture_pos_norm.x, texture_pos_norm.y + texture_size_norm.y, - -1.0f + pos_norm.x + size_norm.x, -1.0f + pos_norm.y, texture_pos_norm.x + texture_size_norm.x, texture_pos_norm.y, - -1.0f + pos_norm.x + size_norm.x, -1.0f + pos_norm.y + size_norm.y, texture_pos_norm.x + texture_size_norm.x, texture_pos_norm.y + texture_size_norm.y + -1.0f + 0.0f, -1.0f + 0.0f + size_norm.y, texture_pos_norm.x, texture_pos_norm.y + texture_size_norm.y, + -1.0f + 0.0f + size_norm.x, -1.0f + 0.0f, texture_pos_norm.x + texture_size_norm.x, texture_pos_norm.y, + -1.0f + 0.0f + size_norm.x, -1.0f + 0.0f + size_norm.y, texture_pos_norm.x + texture_size_norm.x, texture_pos_norm.y + texture_size_norm.y }; self->params.egl->glBindVertexArray(self->vertex_array_object_id); self->params.egl->glViewport(0, 0, dest_texture_size.x, dest_texture_size.y); - self->params.egl->glBindTexture(GL_TEXTURE_2D, texture_id); /* TODO: this, also cleanup */ //self->params.egl->glBindBuffer(GL_ARRAY_BUFFER, self->vertex_buffer_object_id); self->params.egl->glBufferSubData(GL_ARRAY_BUFFER, 0, 24 * sizeof(float), vertices); - { - self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, self->framebuffers[0]); - //cap_xcomp->params.egl->glClear(GL_COLOR_BUFFER_BIT); // TODO: Do this in a separate clear_ function. We want to do that when using multiple drm to create the final image (multiple monitors for example) + switch(self->params.destination_color) { + case GSR_DESTINATION_COLOR_NV12: + case GSR_DESTINATION_COLOR_P010: { + self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, self->framebuffers[0]); + //cap_xcomp->params.egl->glClear(GL_COLOR_BUFFER_BIT); // TODO: Do this in a separate clear_ function. We want to do that when using multiple drm to create the final image (multiple monitors for example) + + int shader_index = external_texture ? GRAPHICS_SHADER_INDEX_Y_EXTERNAL : GRAPHICS_SHADER_INDEX_Y; + gsr_shader_use(&self->graphics_shaders[shader_index]); + self->params.egl->glUniformMatrix2fv(self->graphics_uniforms[shader_index].rotation_matrix, 1, GL_TRUE, (const float*)rotation_matrix); + self->params.egl->glUniform2f(self->graphics_uniforms[shader_index].offset, pos_norm.x, pos_norm.y); + self->params.egl->glDrawArrays(GL_TRIANGLES, 0, 6); + + if(self->params.num_destination_textures > 1) { + self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, self->framebuffers[1]); + //cap_xcomp->params.egl->glClear(GL_COLOR_BUFFER_BIT); + + shader_index = external_texture ? GRAPHICS_SHADER_INDEX_UV_EXTERNAL : GRAPHICS_SHADER_INDEX_UV; + gsr_shader_use(&self->graphics_shaders[shader_index]); + self->params.egl->glUniformMatrix2fv(self->graphics_uniforms[shader_index].rotation_matrix, 1, GL_TRUE, (const float*)rotation_matrix); + self->params.egl->glUniform2f(self->graphics_uniforms[shader_index].offset, pos_norm.x, pos_norm.y); + self->params.egl->glDrawArrays(GL_TRIANGLES, 0, 6); + } + break; + } + case GSR_DESTINATION_COLOR_RGB8: { + self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, self->framebuffers[0]); + //cap_xcomp->params.egl->glClear(GL_COLOR_BUFFER_BIT); // TODO: Do this in a separate clear_ function. We want to do that when using multiple drm to create the final image (multiple monitors for example) + + const int shader_index = external_texture ? GRAPHICS_SHADER_INDEX_RGB_EXTERNAL : GRAPHICS_SHADER_INDEX_RGB; + gsr_shader_use(&self->graphics_shaders[shader_index]); + self->params.egl->glUniformMatrix2fv(self->graphics_uniforms[shader_index].rotation_matrix, 1, GL_TRUE, (const float*)rotation_matrix); + self->params.egl->glUniform2f(self->graphics_uniforms[shader_index].offset, pos_norm.x, pos_norm.y); + self->params.egl->glDrawArrays(GL_TRIANGLES, 0, 6); + break; + } + } + + self->params.egl->glBindVertexArray(0); + self->params.egl->glUseProgram(0); + gsr_color_conversion_swizzle_reset(self, source_color); + self->params.egl->glBindTexture(texture_target, 0); + self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, 0); +} + +void gsr_color_conversion_draw(gsr_color_conversion *self, unsigned int texture_id, vec2i destination_pos, vec2i destination_size, vec2i source_pos, vec2i source_size, vec2i texture_size, gsr_rotation rotation, gsr_source_color source_color, bool external_texture, bool alpha_blending) { + assert(!external_texture || self->params.load_external_image_shader); + if(external_texture && !self->params.load_external_image_shader) { + fprintf(stderr, "gsr error: gsr_color_conversion_draw: external texture not loaded\n"); + return; + } - gsr_shader_use(&self->shaders[0]); - self->params.egl->glUniform1f(self->rotation_uniforms[0], rotation); - self->params.egl->glDrawArrays(GL_TRIANGLES, 0, 6); + vec2f scale = {0.0f, 0.0f}; + if(source_size.x > 0 && source_size.y > 0) + scale = (vec2f){ (double)destination_size.x/(double)source_size.x, (double)destination_size.y/(double)source_size.y }; + + vec2i source_position = {0, 0}; + float rotation_matrix[2][2] = {{0, 0}, {0, 0}}; + gsr_color_conversion_apply_rotation(rotation, rotation_matrix); + + const int texture_target = external_texture ? GL_TEXTURE_EXTERNAL_OES : GL_TEXTURE_2D; + self->params.egl->glBindTexture(texture_target, texture_id); + gsr_color_conversion_swizzle_texture_source(self, source_color); + + const bool use_graphics_shader = external_texture ? self->external_compute_shaders_failed_to_load : self->compute_shaders_failed_to_load; + if(use_graphics_shader) { + source_position.x += source_pos.x; + source_position.y += source_pos.y; + gsr_color_conversion_draw_graphics(self, texture_id, external_texture, rotation_matrix, source_position, source_size, destination_pos, texture_size, scale, source_color); + // TODO: Is glFlush and glFinish needed here for graphics garbage? + } else { + switch(rotation) { + case GSR_ROT_0: + break; + case GSR_ROT_90: + source_position.x += (((double)texture_size.x*0.5 - (double)texture_size.y*0.5) * scale.x); + source_position.y += (((double)texture_size.y*0.5 - (double)texture_size.x*0.5) * scale.y); + break; + case GSR_ROT_180: + break; + case GSR_ROT_270: + source_position.x += (((double)texture_size.x*0.5 - (double)texture_size.y*0.5) * scale.x); + source_position.y += (((double)texture_size.y*0.5 - (double)texture_size.x*0.5) * scale.y); + break; + } + source_position.x -= (source_pos.x * scale.x + 0.5); + source_position.y -= (source_pos.y * scale.y + 0.5); + + switch(self->params.destination_color) { + case GSR_DESTINATION_COLOR_NV12: + case GSR_DESTINATION_COLOR_P010: { + const bool use_16bit_colors = self->params.destination_color == GSR_DESTINATION_COLOR_P010; + gsr_color_conversion_dispatch_compute_shader(self, external_texture, alpha_blending, rotation_matrix, source_position, destination_pos, destination_size, scale, use_16bit_colors, GSR_COLOR_COMP_Y); + gsr_color_conversion_dispatch_compute_shader(self, external_texture, alpha_blending, rotation_matrix, (vec2i){source_position.x/2, source_position.y/2}, + (vec2i){destination_pos.x/2, destination_pos.y/2}, (vec2i){destination_size.x/2, destination_size.y/2}, scale, use_16bit_colors, GSR_COLOR_COMP_UV); + break; + } + case GSR_DESTINATION_COLOR_RGB8: { + gsr_color_conversion_dispatch_compute_shader(self, external_texture, alpha_blending, rotation_matrix, source_position, destination_pos, destination_size, scale, false, GSR_COLOR_COMP_RGB); + break; + } + } } + // TODO: Use the minimal barrier required + self->params.egl->glMemoryBarrier(GL_ALL_BARRIER_BITS); // GL_SHADER_IMAGE_ACCESS_BARRIER_BIT + self->params.egl->glUseProgram(0); + + gsr_color_conversion_swizzle_reset(self, source_color); + self->params.egl->glBindTexture(texture_target, 0); +} + +void gsr_color_conversion_clear(gsr_color_conversion *self) { + float color1[4] = {0.0f, 0.0f, 0.0f, 1.0f}; + float color2[4] = {0.0f, 0.0f, 0.0f, 1.0f}; + + switch(self->params.destination_color) { + case GSR_DESTINATION_COLOR_NV12: + case GSR_DESTINATION_COLOR_P010: { + color2[0] = 0.5f; + color2[1] = 0.5f; + color2[2] = 0.0f; + color2[3] = 1.0f; + break; + } + case GSR_DESTINATION_COLOR_RGB8: { + color2[0] = 0.0f; + color2[1] = 0.0f; + color2[2] = 0.0f; + color2[3] = 1.0f; + break; + } + } + + self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, self->framebuffers[0]); + self->params.egl->glClearColor(color1[0], color1[1], color1[2], color1[3]); + self->params.egl->glClear(GL_COLOR_BUFFER_BIT); + if(self->params.num_destination_textures > 1) { self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, self->framebuffers[1]); - //cap_xcomp->params.egl->glClear(GL_COLOR_BUFFER_BIT); - - gsr_shader_use(&self->shaders[1]); - self->params.egl->glUniform1f(self->rotation_uniforms[1], rotation); - self->params.egl->glDrawArrays(GL_TRIANGLES, 0, 6); + self->params.egl->glClearColor(color2[0], color2[1], color2[2], color2[3]); + self->params.egl->glClear(GL_COLOR_BUFFER_BIT); } - self->params.egl->glBindVertexArray(0); - gsr_shader_use_none(&self->shaders[0]); - self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); self->params.egl->glBindFramebuffer(GL_FRAMEBUFFER, 0); - return 0; +} + +gsr_rotation gsr_monitor_rotation_to_rotation(gsr_monitor_rotation monitor_rotation) { + return (gsr_rotation)monitor_rotation; } @@ -19,7 +19,7 @@ bool gsr_cuda_load(gsr_cuda *self, Display *display, bool do_overclock) { } } - dlsym_assign required_dlsym[] = { + const dlsym_assign required_dlsym[] = { { (void**)&self->cuInit, "cuInit" }, { (void**)&self->cuDeviceGetCount, "cuDeviceGetCount" }, { (void**)&self->cuDeviceGet, "cuDeviceGet" }, @@ -28,8 +28,9 @@ bool gsr_cuda_load(gsr_cuda *self, Display *display, bool do_overclock) { { (void**)&self->cuCtxPushCurrent_v2, "cuCtxPushCurrent_v2" }, { (void**)&self->cuCtxPopCurrent_v2, "cuCtxPopCurrent_v2" }, { (void**)&self->cuGetErrorString, "cuGetErrorString" }, - { (void**)&self->cuMemsetD8_v2, "cuMemsetD8_v2" }, { (void**)&self->cuMemcpy2D_v2, "cuMemcpy2D_v2" }, + { (void**)&self->cuMemcpy2DAsync_v2, "cuMemcpy2DAsync_v2" }, + { (void**)&self->cuStreamSynchronize, "cuStreamSynchronize" }, { (void**)&self->cuGraphicsGLRegisterImage, "cuGraphicsGLRegisterImage" }, { (void**)&self->cuGraphicsEGLRegisterImage, "cuGraphicsEGLRegisterImage" }, @@ -81,12 +82,13 @@ bool gsr_cuda_load(gsr_cuda *self, Display *display, bool do_overclock) { goto fail; } - if(self->do_overclock) { - assert(display); + if(self->do_overclock && display) { if(gsr_overclock_load(&self->overclock, display)) gsr_overclock_start(&self->overclock); else fprintf(stderr, "gsr warning: gsr_cuda_load: failed to load xnvctrl, failed to overclock memory transfer rate\n"); + } else if(self->do_overclock && !display) { + fprintf(stderr, "gsr warning: gsr_cuda_load: overclocking enabled but no X server is running. Overclocking has been disabled\n"); } self->library = lib; diff --git a/src/cursor.c b/src/cursor.c new file mode 100644 index 0000000..40532f8 --- /dev/null +++ b/src/cursor.c @@ -0,0 +1,142 @@ +#include "../include/cursor.h" + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <assert.h> + +#include <X11/extensions/Xfixes.h> + +// TODO: Test cursor visibility with XFixesHideCursor + +static bool gsr_cursor_set_from_x11_cursor_image(gsr_cursor *self, XFixesCursorImage *x11_cursor_image, bool *visible) { + uint8_t *cursor_data = NULL; + uint8_t *out = NULL; + *visible = false; + + if(!x11_cursor_image) + goto err; + + if(!x11_cursor_image->pixels) + goto err; + + self->hotspot.x = x11_cursor_image->xhot; + self->hotspot.y = x11_cursor_image->yhot; + self->egl->glBindTexture(GL_TEXTURE_2D, self->texture_id); + + self->size.x = x11_cursor_image->width; + self->size.y = x11_cursor_image->height; + const unsigned long *pixels = x11_cursor_image->pixels; + cursor_data = malloc(self->size.x * self->size.y * 4); + if(!cursor_data) + goto err; + out = cursor_data; + /* Un-premultiply alpha */ + for(int y = 0; y < self->size.y; ++y) { + for(int x = 0; x < self->size.x; ++x) { + uint32_t pixel = *pixels++; + uint8_t *in = (uint8_t*)&pixel; + uint8_t alpha = in[3]; + if(alpha == 0) { + alpha = 1; + } else { + *visible = true; + } + + out[0] = (float)in[2] * 255.0/(float)alpha; + out[1] = (float)in[1] * 255.0/(float)alpha; + out[2] = (float)in[0] * 255.0/(float)alpha; + out[3] = in[3]; + out += 4; + in += 4; + } + } + + // TODO: glTextureSubImage2D if same size + self->egl->glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, self->size.x, self->size.y, 0, GL_RGBA, GL_UNSIGNED_BYTE, cursor_data); + free(cursor_data); + + const float border_color[4] = { 0.0f, 0.0f, 0.0f, 0.0f }; + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->egl->glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, border_color); + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + + self->egl->glBindTexture(GL_TEXTURE_2D, 0); + XFree(x11_cursor_image); + return true; + + err: + self->egl->glBindTexture(GL_TEXTURE_2D, 0); + if(x11_cursor_image) + XFree(x11_cursor_image); + return false; +} + +int gsr_cursor_init(gsr_cursor *self, gsr_egl *egl, Display *display) { + int x_fixes_error_base = 0; + + assert(egl); + assert(display); + memset(self, 0, sizeof(*self)); + self->egl = egl; + self->display = display; + + self->x_fixes_event_base = 0; + if(!XFixesQueryExtension(self->display, &self->x_fixes_event_base, &x_fixes_error_base)) { + fprintf(stderr, "gsr error: gsr_cursor_init: your X11 server is missing the XFixes extension\n"); + gsr_cursor_deinit(self); + return -1; + } + + self->egl->glGenTextures(1, &self->texture_id); + + XFixesSelectCursorInput(self->display, DefaultRootWindow(self->display), XFixesDisplayCursorNotifyMask); + gsr_cursor_set_from_x11_cursor_image(self, XFixesGetCursorImage(self->display), &self->visible); + self->cursor_image_set = true; + + return 0; +} + +void gsr_cursor_deinit(gsr_cursor *self) { + if(!self->egl) + return; + + if(self->texture_id) { + self->egl->glDeleteTextures(1, &self->texture_id); + self->texture_id = 0; + } + + if(self->display) + XFixesSelectCursorInput(self->display, DefaultRootWindow(self->display), 0); + + self->display = NULL; + self->egl = NULL; +} + +bool gsr_cursor_on_event(gsr_cursor *self, XEvent *xev) { + bool updated = false; + + if(xev->type == self->x_fixes_event_base + XFixesCursorNotify) { + XFixesCursorNotifyEvent *cursor_notify_event = (XFixesCursorNotifyEvent*)xev; + if(cursor_notify_event->subtype == XFixesDisplayCursorNotify && cursor_notify_event->window == DefaultRootWindow(self->display)) { + self->cursor_image_set = false; + } + } + + if(!self->cursor_image_set) { + self->cursor_image_set = true; + gsr_cursor_set_from_x11_cursor_image(self, XFixesGetCursorImage(self->display), &self->visible); + updated = true; + } + + return updated; +} + +void gsr_cursor_tick(gsr_cursor *self, Window relative_to) { + Window dummy_window; + int dummy_i; + unsigned int dummy_u; + XQueryPointer(self->display, relative_to, &dummy_window, &dummy_window, &dummy_i, &dummy_i, &self->position.x, &self->position.y, &dummy_u); +} diff --git a/src/damage.c b/src/damage.c new file mode 100644 index 0000000..25a2225 --- /dev/null +++ b/src/damage.c @@ -0,0 +1,327 @@ +#include "../include/damage.h" +#include "../include/utils.h" +#include "../include/window/window.h" + +#include <stdio.h> +#include <string.h> +#include <X11/extensions/Xdamage.h> +#include <X11/extensions/Xrandr.h> + +typedef struct { + vec2i pos; + vec2i size; +} gsr_rectangle; + +static bool rectangles_intersect(gsr_rectangle rect1, gsr_rectangle rect2) { + return rect1.pos.x < rect2.pos.x + rect2.size.x && rect1.pos.x + rect1.size.x > rect2.pos.x && + rect1.pos.y < rect2.pos.y + rect2.size.y && rect1.pos.y + rect1.size.y > rect2.pos.y; +} + +static bool xrandr_is_supported(Display *display) { + int major_version = 0; + int minor_version = 0; + if(!XRRQueryVersion(display, &major_version, &minor_version)) + return false; + + return major_version > 1 || (major_version == 1 && minor_version >= 2); +} + +bool gsr_damage_init(gsr_damage *self, gsr_egl *egl, bool track_cursor) { + memset(self, 0, sizeof(*self)); + self->egl = egl; + self->track_cursor = track_cursor; + + if(gsr_window_get_display_server(egl->window) != GSR_DISPLAY_SERVER_X11) { + fprintf(stderr, "gsr warning: gsr_damage_init: damage tracking is not supported on wayland\n"); + return false; + } + self->display = gsr_window_get_display(egl->window); + + if(!XDamageQueryExtension(self->display, &self->damage_event, &self->damage_error)) { + fprintf(stderr, "gsr warning: gsr_damage_init: XDamage is not supported by your X11 server\n"); + gsr_damage_deinit(self); + return false; + } + + if(!XRRQueryExtension(self->display, &self->randr_event, &self->randr_error)) { + fprintf(stderr, "gsr warning: gsr_damage_init: XRandr is not supported by your X11 server\n"); + gsr_damage_deinit(self); + return false; + } + + if(!xrandr_is_supported(self->display)) { + fprintf(stderr, "gsr warning: gsr_damage_init: your X11 randr version is too old\n"); + gsr_damage_deinit(self); + return false; + } + + if(self->track_cursor) + self->track_cursor = gsr_cursor_init(&self->cursor, self->egl, self->display) == 0; + + XRRSelectInput(self->display, DefaultRootWindow(self->display), RRScreenChangeNotifyMask | RRCrtcChangeNotifyMask | RROutputChangeNotifyMask); + + self->damaged = true; + return true; +} + +void gsr_damage_deinit(gsr_damage *self) { + if(self->damage) { + XDamageDestroy(self->display, self->damage); + self->damage = None; + } + + gsr_cursor_deinit(&self->cursor); + + self->damage_event = 0; + self->damage_error = 0; + + self->randr_event = 0; + self->randr_error = 0; +} + +bool gsr_damage_set_target_window(gsr_damage *self, uint64_t window) { + if(self->damage_event == 0) + return false; + + if(window == self->window) + return true; + + if(self->damage) { + XDamageDestroy(self->display, self->damage); + self->damage = None; + } + + if(self->window) + XSelectInput(self->display, self->window, 0); + + self->window = window; + XSelectInput(self->display, self->window, StructureNotifyMask | ExposureMask); + + XWindowAttributes win_attr; + win_attr.x = 0; + win_attr.y = 0; + win_attr.width = 0; + win_attr.height = 0; + if(!XGetWindowAttributes(self->display, self->window, &win_attr)) + fprintf(stderr, "gsr warning: gsr_damage_set_target_window failed: failed to get window attributes: %ld\n", (long)self->window); + + //self->window_pos.x = win_attr.x; + //self->window_pos.y = win_attr.y; + + self->window_size.x = win_attr.width; + self->window_size.y = win_attr.height; + + self->damage = XDamageCreate(self->display, window, XDamageReportNonEmpty); + if(self->damage) { + XDamageSubtract(self->display, self->damage, None, None); + self->damaged = true; + self->track_type = GSR_DAMAGE_TRACK_WINDOW; + return true; + } else { + fprintf(stderr, "gsr warning: gsr_damage_set_target_window: XDamageCreate failed\n"); + self->track_type = GSR_DAMAGE_TRACK_NONE; + return false; + } +} + +bool gsr_damage_set_target_monitor(gsr_damage *self, const char *monitor_name) { + if(self->damage_event == 0) + return false; + + if(strcmp(self->monitor_name, monitor_name) == 0) + return true; + + if(self->damage) { + XDamageDestroy(self->display, self->damage); + self->damage = None; + } + + memset(&self->monitor, 0, sizeof(self->monitor)); + if(strcmp(monitor_name, "screen-direct") != 0 && strcmp(monitor_name, "screen-direct-force") != 0) { + if(!get_monitor_by_name(self->egl, GSR_CONNECTION_X11, monitor_name, &self->monitor)) + fprintf(stderr, "gsr warning: gsr_damage_set_target_monitor: failed to find monitor: %s\n", monitor_name); + } + + if(self->window) + XSelectInput(self->display, self->window, 0); + + self->window = DefaultRootWindow(self->display); + self->damage = XDamageCreate(self->display, self->window, XDamageReportNonEmpty); + if(self->damage) { + XDamageSubtract(self->display, self->damage, None, None); + self->damaged = true; + snprintf(self->monitor_name, sizeof(self->monitor_name), "%s", monitor_name); + self->track_type = GSR_DAMAGE_TRACK_MONITOR; + return true; + } else { + fprintf(stderr, "gsr warning: gsr_damage_set_target_monitor: XDamageCreate failed\n"); + self->track_type = GSR_DAMAGE_TRACK_NONE; + return false; + } +} + +static void gsr_damage_on_crtc_change(gsr_damage *self, XEvent *xev) { + const XRRCrtcChangeNotifyEvent *rr_crtc_change_event = (XRRCrtcChangeNotifyEvent*)xev; + if(rr_crtc_change_event->crtc == 0 || self->monitor.monitor_identifier == 0) + return; + + if(rr_crtc_change_event->crtc != self->monitor.monitor_identifier) + return; + + if(rr_crtc_change_event->width == 0 || rr_crtc_change_event->height == 0) + return; + + if(rr_crtc_change_event->x != self->monitor.pos.x || rr_crtc_change_event->y != self->monitor.pos.y || + (int)rr_crtc_change_event->width != self->monitor.size.x || (int)rr_crtc_change_event->height != self->monitor.size.y) { + self->monitor.pos.x = rr_crtc_change_event->x; + self->monitor.pos.y = rr_crtc_change_event->y; + + self->monitor.size.x = rr_crtc_change_event->width; + self->monitor.size.y = rr_crtc_change_event->height; + } +} + +static void gsr_damage_on_output_change(gsr_damage *self, XEvent *xev) { + const XRROutputChangeNotifyEvent *rr_output_change_event = (XRROutputChangeNotifyEvent*)xev; + if(!rr_output_change_event->output || self->monitor.monitor_identifier == 0) + return; + + XRRScreenResources *screen_res = XRRGetScreenResources(self->display, DefaultRootWindow(self->display)); + if(!screen_res) + return; + + // TODO: What about scaled output? look at for_each_active_monitor_output_x11_not_cached + XRROutputInfo *out_info = XRRGetOutputInfo(self->display, screen_res, rr_output_change_event->output); + if(out_info && out_info->crtc && out_info->crtc == self->monitor.monitor_identifier) { + XRRCrtcInfo *crtc_info = XRRGetCrtcInfo(self->display, screen_res, out_info->crtc); + if(crtc_info && (crtc_info->x != self->monitor.pos.x || crtc_info->y != self->monitor.pos.y || + (int)crtc_info->width != self->monitor.size.x || (int)crtc_info->height != self->monitor.size.y)) + { + self->monitor.pos.x = crtc_info->x; + self->monitor.pos.y = crtc_info->y; + + self->monitor.size.x = crtc_info->width; + self->monitor.size.y = crtc_info->height; + } + + if(crtc_info) + XRRFreeCrtcInfo(crtc_info); + } + + if(out_info) + XRRFreeOutputInfo(out_info); + + XRRFreeScreenResources(screen_res); +} + +static void gsr_damage_on_randr_event(gsr_damage *self, XEvent *xev) { + const XRRNotifyEvent *rr_event = (XRRNotifyEvent*)xev; + switch(rr_event->subtype) { + case RRNotify_CrtcChange: + gsr_damage_on_crtc_change(self, xev); + break; + case RRNotify_OutputChange: + gsr_damage_on_output_change(self, xev); + break; + } +} + +static void gsr_damage_on_damage_event(gsr_damage *self, XEvent *xev) { + const XDamageNotifyEvent *de = (XDamageNotifyEvent*)xev; + XserverRegion region = XFixesCreateRegion(self->display, NULL, 0); + /* Subtract all the damage, repairing the window */ + XDamageSubtract(self->display, de->damage, None, region); + + if(self->track_type == GSR_DAMAGE_TRACK_WINDOW || (self->track_type == GSR_DAMAGE_TRACK_MONITOR && self->monitor.connector_id == 0)) { + self->damaged = true; + } else { + int num_rectangles = 0; + XRectangle *rectangles = XFixesFetchRegion(self->display, region, &num_rectangles); + if(rectangles) { + const gsr_rectangle monitor_region = { self->monitor.pos, self->monitor.size }; + for(int i = 0; i < num_rectangles; ++i) { + const gsr_rectangle damage_region = { (vec2i){rectangles[i].x, rectangles[i].y}, (vec2i){rectangles[i].width, rectangles[i].height} }; + self->damaged = rectangles_intersect(monitor_region, damage_region); + if(self->damaged) + break; + } + XFree(rectangles); + } + } + + XFixesDestroyRegion(self->display, region); + XFlush(self->display); +} + +static void gsr_damage_on_tick_cursor(gsr_damage *self) { + vec2i prev_cursor_pos = self->cursor.position; + gsr_cursor_tick(&self->cursor, self->window); + if(self->cursor.position.x != prev_cursor_pos.x || self->cursor.position.y != prev_cursor_pos.y) { + const gsr_rectangle cursor_region = { self->cursor.position, self->cursor.size }; + switch(self->track_type) { + case GSR_DAMAGE_TRACK_NONE: { + self->damaged = true; + break; + } + case GSR_DAMAGE_TRACK_WINDOW: { + const gsr_rectangle window_region = { (vec2i){0, 0}, self->window_size }; + self->damaged = self->window_size.x == 0 || rectangles_intersect(window_region, cursor_region); + break; + } + case GSR_DAMAGE_TRACK_MONITOR: { + const gsr_rectangle monitor_region = { self->monitor.pos, self->monitor.size }; + self->damaged = self->monitor.monitor_identifier == 0 || rectangles_intersect(monitor_region, cursor_region); + break; + } + } + } +} + +static void gsr_damage_on_window_configure_notify(gsr_damage *self, XEvent *xev) { + if(xev->xconfigure.window != self->window) + return; + + //self->window_pos.x = xev->xconfigure.x; + //self->window_pos.y = xev->xconfigure.y; + + self->window_size.x = xev->xconfigure.width; + self->window_size.y = xev->xconfigure.height; +} + +void gsr_damage_on_event(gsr_damage *self, XEvent *xev) { + if(self->damage_event == 0 || self->track_type == GSR_DAMAGE_TRACK_NONE) + return; + + if(self->track_type == GSR_DAMAGE_TRACK_WINDOW && xev->type == ConfigureNotify) + gsr_damage_on_window_configure_notify(self, xev); + + if(self->randr_event) { + if(xev->type == self->randr_event + RRScreenChangeNotify) + XRRUpdateConfiguration(xev); + + if(xev->type == self->randr_event + RRNotify) + gsr_damage_on_randr_event(self, xev); + } + + if(self->damage_event && xev->type == self->damage_event + XDamageNotify) + gsr_damage_on_damage_event(self, xev); + + if(self->track_cursor) + gsr_cursor_on_event(&self->cursor, xev); +} + +void gsr_damage_tick(gsr_damage *self) { + if(self->damage_event == 0 || self->track_type == GSR_DAMAGE_TRACK_NONE) + return; + + if(self->track_cursor && self->cursor.visible && !self->damaged) + gsr_damage_on_tick_cursor(self); +} + +bool gsr_damage_is_damaged(gsr_damage *self) { + return self->damage_event == 0 || !self->damage || self->damaged || self->track_type == GSR_DAMAGE_TRACK_NONE; +} + +void gsr_damage_clear(gsr_damage *self) { + self->damaged = false; +} diff --git a/src/defs.c b/src/defs.c new file mode 100644 index 0000000..319d21b --- /dev/null +++ b/src/defs.c @@ -0,0 +1,100 @@ +#include "../include/defs.h" +#include <assert.h> + +bool video_codec_is_hdr(gsr_video_codec video_codec) { + // TODO: Vulkan + switch(video_codec) { + case GSR_VIDEO_CODEC_HEVC_HDR: + case GSR_VIDEO_CODEC_AV1_HDR: + return true; + default: + return false; + } +} + +gsr_video_codec hdr_video_codec_to_sdr_video_codec(gsr_video_codec video_codec) { + // TODO: Vulkan + switch(video_codec) { + case GSR_VIDEO_CODEC_HEVC_HDR: + return GSR_VIDEO_CODEC_HEVC; + case GSR_VIDEO_CODEC_AV1_HDR: + return GSR_VIDEO_CODEC_AV1; + default: + return video_codec; + } +} + +gsr_color_depth video_codec_to_bit_depth(gsr_video_codec video_codec) { + // TODO: 10-bit Vulkan + switch(video_codec) { + case GSR_VIDEO_CODEC_HEVC_HDR: + case GSR_VIDEO_CODEC_HEVC_10BIT: + case GSR_VIDEO_CODEC_AV1_HDR: + case GSR_VIDEO_CODEC_AV1_10BIT: + return GSR_COLOR_DEPTH_10_BITS; + default: + return GSR_COLOR_DEPTH_8_BITS; + } +} + +const char* video_codec_to_string(gsr_video_codec video_codec) { + switch(video_codec) { + case GSR_VIDEO_CODEC_H264: return "h264"; + case GSR_VIDEO_CODEC_HEVC: return "hevc"; + case GSR_VIDEO_CODEC_HEVC_HDR: return "hevc_hdr"; + case GSR_VIDEO_CODEC_HEVC_10BIT: return "hevc_10bit"; + case GSR_VIDEO_CODEC_AV1: return "av1"; + case GSR_VIDEO_CODEC_AV1_HDR: return "av1_hdr"; + case GSR_VIDEO_CODEC_AV1_10BIT: return "av1_10bit"; + case GSR_VIDEO_CODEC_VP8: return "vp8"; + case GSR_VIDEO_CODEC_VP9: return "vp9"; + case GSR_VIDEO_CODEC_H264_VULKAN: return "h264_vulkan"; + case GSR_VIDEO_CODEC_HEVC_VULKAN: return "hevc_vulkan"; + } + return ""; +} + +// bool video_codec_is_hevc(gsr_video_codec video_codec) { +// // TODO: 10-bit vulkan +// switch(video_codec) { +// case GSR_VIDEO_CODEC_HEVC: +// case GSR_VIDEO_CODEC_HEVC_HDR: +// case GSR_VIDEO_CODEC_HEVC_10BIT: +// case GSR_VIDEO_CODEC_HEVC_VULKAN: +// return true; +// default: +// return false; +// } +// } + +bool video_codec_is_av1(gsr_video_codec video_codec) { + // TODO: Vulkan + switch(video_codec) { + case GSR_VIDEO_CODEC_AV1: + case GSR_VIDEO_CODEC_AV1_HDR: + case GSR_VIDEO_CODEC_AV1_10BIT: + return true; + default: + return false; + } +} + +bool video_codec_is_vulkan(gsr_video_codec video_codec) { + switch(video_codec) { + case GSR_VIDEO_CODEC_H264_VULKAN: + case GSR_VIDEO_CODEC_HEVC_VULKAN: + return true; + default: + return false; + } +} + +const char* audio_codec_get_name(gsr_audio_codec audio_codec) { + switch(audio_codec) { + case GSR_AUDIO_CODEC_AAC: return "aac"; + case GSR_AUDIO_CODEC_OPUS: return "opus"; + case GSR_AUDIO_CODEC_FLAC: return "flac"; + } + assert(false); + return ""; +} @@ -1,217 +1,50 @@ #include "../include/egl.h" +#include "../include/window/window.h" #include "../include/library_loader.h" +#include "../include/utils.h" + #include <string.h> #include <stdio.h> #include <stdlib.h> #include <dlfcn.h> #include <assert.h> - -#include <wayland-client.h> -#include <wayland-egl.h> -#include "../external/wlr-export-dmabuf-unstable-v1-client-protocol.h" #include <unistd.h> #include <sys/capability.h> -// Move this shit to a separate wayland file, and have a separate file for x11. - -static void output_handle_geometry(void *data, struct wl_output *wl_output, - int32_t x, int32_t y, int32_t phys_width, int32_t phys_height, - int32_t subpixel, const char *make, const char *model, - int32_t transform) { - (void)wl_output; - (void)phys_width; - (void)phys_height; - (void)subpixel; - (void)make; - (void)model; - (void)transform; - gsr_wayland_output *gsr_output = data; - gsr_output->pos.x = x; - gsr_output->pos.y = y; -} - -static void output_handle_mode(void *data, struct wl_output *wl_output, uint32_t flags, int32_t width, int32_t height, int32_t refresh) { - (void)wl_output; - (void)flags; - (void)refresh; - gsr_wayland_output *gsr_output = data; - gsr_output->size.x = width; - gsr_output->size.y = height; -} - -static void output_handle_done(void *data, struct wl_output *wl_output) { - (void)data; - (void)wl_output; -} - -static void output_handle_scale(void* data, struct wl_output *wl_output, int32_t factor) { - (void)data; - (void)wl_output; - (void)factor; -} - -static void output_handle_name(void *data, struct wl_output *wl_output, const char *name) { - (void)wl_output; - gsr_wayland_output *gsr_output = data; - if(gsr_output->name) { - free(gsr_output->name); - gsr_output->name = NULL; - } - gsr_output->name = strdup(name); -} - -static void output_handle_description(void *data, struct wl_output *wl_output, const char *description) { - (void)data; - (void)wl_output; - (void)description; -} - -static const struct wl_output_listener output_listener = { - .geometry = output_handle_geometry, - .mode = output_handle_mode, - .done = output_handle_done, - .scale = output_handle_scale, - .name = output_handle_name, - .description = output_handle_description, -}; - -static void registry_add_object(void *data, struct wl_registry *registry, uint32_t name, const char *interface, uint32_t version) { - (void)version; - gsr_egl *egl = data; - if (strcmp(interface, "wl_compositor") == 0) { - if(egl->wayland.compositor) { - wl_compositor_destroy(egl->wayland.compositor); - egl->wayland.compositor = NULL; - } - egl->wayland.compositor = wl_registry_bind(registry, name, &wl_compositor_interface, 1); - } else if(strcmp(interface, wl_output_interface.name) == 0) { - if(version < 4) { - fprintf(stderr, "gsr warning: wl output interface version is < 4, expected >= 4 to capture a monitor. Using KMS capture instead\n"); - return; - } - - if(egl->wayland.num_outputs == GSR_MAX_OUTPUTS) { - fprintf(stderr, "gsr warning: reached maximum outputs (32), ignoring output %u\n", name); - return; - } - - gsr_wayland_output *gsr_output = &egl->wayland.outputs[egl->wayland.num_outputs]; - egl->wayland.num_outputs++; - *gsr_output = (gsr_wayland_output) { - .wl_name = name, - .output = wl_registry_bind(registry, name, &wl_output_interface, 4), - .pos = { .x = 0, .y = 0 }, - .size = { .x = 0, .y = 0 }, - .name = NULL, - }; - wl_output_add_listener(gsr_output->output, &output_listener, gsr_output); - } else if(strcmp(interface, zwlr_export_dmabuf_manager_v1_interface.name) == 0) { - if(egl->wayland.export_manager) { - zwlr_export_dmabuf_manager_v1_destroy(egl->wayland.export_manager); - egl->wayland.export_manager = NULL; - } - egl->wayland.export_manager = wl_registry_bind(registry, name, &zwlr_export_dmabuf_manager_v1_interface, 1); - } -} - -static void registry_remove_object(void *data, struct wl_registry *registry, uint32_t name) { - (void)data; - (void)registry; - (void)name; -} - -static struct wl_registry_listener registry_listener = { - .global = registry_add_object, - .global_remove = registry_remove_object, -}; - -static void frame_capture_output(gsr_egl *egl); - -static void frame_start(void *data, struct zwlr_export_dmabuf_frame_v1 *frame, - uint32_t width, uint32_t height, uint32_t offset_x, uint32_t offset_y, - uint32_t buffer_flags, uint32_t flags, uint32_t format, - uint32_t mod_high, uint32_t mod_low, uint32_t num_objects) { - (void)buffer_flags; - (void)flags; - (void)num_objects; - gsr_egl *egl = data; - //fprintf(stderr, "frame start %p, width: %u, height: %u, offset x: %u, offset y: %u, format: %u, num objects: %u\n", (void*)frame, width, height, offset_x, offset_y, format, num_objects); - egl->x = offset_x; - egl->y = offset_y; - egl->width = width; - egl->height = height; - egl->pixel_format = format; - egl->modifier = ((uint64_t)mod_high << 32) | (uint64_t)mod_low; - egl->wayland.current_frame = frame; -} - -static void frame_object(void *data, struct zwlr_export_dmabuf_frame_v1 *frame, - uint32_t index, int32_t fd, uint32_t size, uint32_t offset, - uint32_t stride, uint32_t plane_index) { - // TODO: What if we get multiple objects? then we get multiple fd per frame - (void)frame; - (void)index; - (void)size; - (void)plane_index; - gsr_egl *egl = data; - if(egl->fd > 0) { - close(egl->fd); - egl->fd = 0; - } - egl->fd = fd; - egl->pitch = stride; - egl->offset = offset; - //fprintf(stderr, "new frame %p, fd: %d, index: %u, size: %u, offset: %u, stride: %u, plane_index: %u\n", (void*)frame, fd, index, size, offset, stride, plane_index); -} - -static void frame_ready(void *data, struct zwlr_export_dmabuf_frame_v1 *frame, uint32_t tv_sec_hi, uint32_t tv_sec_lo, uint32_t tv_nsec) { - (void)frame; - (void)tv_sec_hi; - (void)tv_sec_lo; - (void)tv_nsec; - frame_capture_output(data); -} - -static void frame_cancel(void *data, struct zwlr_export_dmabuf_frame_v1 *frame, uint32_t reason) { - (void)frame; - (void)reason; - frame_capture_output(data); -} - -static const struct zwlr_export_dmabuf_frame_v1_listener frame_listener = { - .frame = frame_start, - .object = frame_object, - .ready = frame_ready, - .cancel = frame_cancel, -}; - -static void frame_capture_output(gsr_egl *egl) { - assert(egl->wayland.output_to_capture); - bool with_cursor = true; - - if(egl->wayland.frame_callback) { - zwlr_export_dmabuf_frame_v1_destroy(egl->wayland.frame_callback); - egl->wayland.frame_callback = NULL; - } - - egl->wayland.frame_callback = zwlr_export_dmabuf_manager_v1_capture_output(egl->wayland.export_manager, with_cursor, egl->wayland.output_to_capture->output); - zwlr_export_dmabuf_frame_v1_add_listener(egl->wayland.frame_callback, &frame_listener, egl); -} - -static gsr_wayland_output* get_wayland_output_by_name(gsr_egl *egl, const char *name) { - assert(name); - for(int i = 0; i < egl->wayland.num_outputs; ++i) { - if(egl->wayland.outputs[i].name && strcmp(egl->wayland.outputs[i].name, name) == 0) - return &egl->wayland.outputs[i]; - } - return NULL; -} +// TODO: rename gsr_egl to something else since this includes both egl and glx and in the future maybe vulkan too + +#define GLX_DRAWABLE_TYPE 0x8010 +#define GLX_RENDER_TYPE 0x8011 +#define GLX_RGBA_BIT 0x00000001 +#define GLX_WINDOW_BIT 0x00000001 +#define GLX_PIXMAP_BIT 0x00000002 +#define GLX_BIND_TO_TEXTURE_RGBA_EXT 0x20D1 +#define GLX_BIND_TO_TEXTURE_TARGETS_EXT 0x20D3 +#define GLX_TEXTURE_2D_BIT_EXT 0x00000002 +#define GLX_DOUBLEBUFFER 5 +#define GLX_RED_SIZE 8 +#define GLX_GREEN_SIZE 9 +#define GLX_BLUE_SIZE 10 +#define GLX_ALPHA_SIZE 11 +#define GLX_DEPTH_SIZE 12 +#define GLX_RGBA_TYPE 0x8014 + +#define GLX_CONTEXT_PRIORITY_LEVEL_EXT 0x3100 +#define GLX_CONTEXT_PRIORITY_HIGH_EXT 0x3101 +#define GLX_CONTEXT_PRIORITY_MEDIUM_EXT 0x3102 +#define GLX_CONTEXT_PRIORITY_LOW_EXT 0x3103 static void reset_cap_nice(void) { cap_t caps = cap_get_proc(); if(!caps) return; + cap_flag_value_t cap_sys_nice_value = CAP_CLEAR; + cap_get_flag(caps, CAP_SYS_NICE, CAP_EFFECTIVE, &cap_sys_nice_value); + if(cap_sys_nice_value == CAP_CLEAR) { + fprintf(stderr, "gsr warning: cap_sys_nice capability is missing on the gpu-screen-recorder binary, performance might be affected. If you are using the flatpak version of gpu-screen-recorder then the only fix is to use a non-flatpak version of gpu-screen-recorder\n"); + } + const cap_value_t cap_to_remove = CAP_SYS_NICE; cap_set_flag(caps, CAP_EFFECTIVE, 1, &cap_to_remove, CAP_CLEAR); cap_set_flag(caps, CAP_PERMITTED, 1, &cap_to_remove, CAP_CLEAR); @@ -220,54 +53,27 @@ static void reset_cap_nice(void) { } // TODO: Create egl context without surface (in other words, x11/wayland agnostic, doesn't require x11/wayland dependency) -static bool gsr_egl_create_window(gsr_egl *self, bool wayland) { +static bool gsr_egl_create_window(gsr_egl *self) { EGLConfig ecfg; int32_t num_config = 0; + // TODO: Use EGL_OPENGL_ES_BIT as amd requires that for external texture, but that breaks software encoding const int32_t attr[] = { EGL_BUFFER_SIZE, 24, EGL_RENDERABLE_TYPE, EGL_OPENGL_BIT, - EGL_NONE + EGL_NONE, EGL_NONE }; const int32_t ctxattr[] = { EGL_CONTEXT_CLIENT_VERSION, 2, EGL_CONTEXT_PRIORITY_LEVEL_IMG, EGL_CONTEXT_PRIORITY_HIGH_IMG, /* requires cap_sys_nice, ignored otherwise */ - EGL_NONE + EGL_NONE, EGL_NONE }; - if(wayland) { - self->wayland.dpy = wl_display_connect(NULL); - if(!self->wayland.dpy) { - fprintf(stderr, "gsr error: gsr_egl_create_window failed: wl_display_connect failed\n"); - goto fail; - } - - self->wayland.registry = wl_display_get_registry(self->wayland.dpy); // TODO: Error checking - wl_registry_add_listener(self->wayland.registry, ®istry_listener, self); // TODO: Error checking - - // Fetch globals - wl_display_roundtrip(self->wayland.dpy); - - // Fetch wl_output - wl_display_roundtrip(self->wayland.dpy); - - if(!self->wayland.compositor) { - fprintf(stderr, "gsr error: gsr_gl_create_window failed: failed to find compositor\n"); - goto fail; - } - } else { - self->x11.window = XCreateWindow(self->x11.dpy, DefaultRootWindow(self->x11.dpy), 0, 0, 16, 16, 0, CopyFromParent, InputOutput, CopyFromParent, 0, NULL); - - if(!self->x11.window) { - fprintf(stderr, "gsr error: gsr_gl_create_window failed: failed to create gl window\n"); - goto fail; - } - } - + // TODO: Use EGL_OPENGL_ES_API as amd requires that for external texture, but that breaks software encoding self->eglBindAPI(EGL_OPENGL_API); - self->egl_display = self->eglGetDisplay(self->wayland.dpy ? (EGLNativeDisplayType)self->wayland.dpy : (EGLNativeDisplayType)self->x11.dpy); + self->egl_display = self->eglGetDisplay((EGLNativeDisplayType)gsr_window_get_display(self->window)); if(!self->egl_display) { fprintf(stderr, "gsr error: gsr_egl_create_window failed: eglGetDisplay failed\n"); goto fail; @@ -277,33 +83,26 @@ static bool gsr_egl_create_window(gsr_egl *self, bool wayland) { fprintf(stderr, "gsr error: gsr_egl_create_window failed: eglInitialize failed\n"); goto fail; } - + if(!self->eglChooseConfig(self->egl_display, attr, &ecfg, 1, &num_config) || num_config != 1) { fprintf(stderr, "gsr error: gsr_egl_create_window failed: failed to find a matching config\n"); goto fail; } - + self->egl_context = self->eglCreateContext(self->egl_display, ecfg, NULL, ctxattr); if(!self->egl_context) { fprintf(stderr, "gsr error: gsr_egl_create_window failed: failed to create egl context\n"); goto fail; } - if(wayland) { - self->wayland.surface = wl_compositor_create_surface(self->wayland.compositor); - self->wayland.window = wl_egl_window_create(self->wayland.surface, 16, 16); - self->egl_surface = self->eglCreateWindowSurface(self->egl_display, ecfg, (EGLNativeWindowType)self->wayland.window, NULL); - } else { - self->egl_surface = self->eglCreateWindowSurface(self->egl_display, ecfg, (EGLNativeWindowType)self->x11.window, NULL); - } - + self->egl_surface = self->eglCreateWindowSurface(self->egl_display, ecfg, (EGLNativeWindowType)gsr_window_get_window(self->window), NULL); if(!self->egl_surface) { fprintf(stderr, "gsr error: gsr_egl_create_window failed: failed to create window surface\n"); goto fail; } if(!self->eglMakeCurrent(self->egl_display, self->egl_surface, self->egl_surface, self->egl_context)) { - fprintf(stderr, "gsr error: gsr_egl_create_window failed: failed to make context current\n"); + fprintf(stderr, "gsr error: gsr_egl_create_window failed: failed to make egl context current\n"); goto fail; } @@ -316,8 +115,86 @@ static bool gsr_egl_create_window(gsr_egl *self, bool wayland) { return false; } +static GLXFBConfig glx_fb_config_choose(gsr_egl *self, Display *display) { + const int glx_visual_attribs[] = { + GLX_RENDER_TYPE, GLX_RGBA_BIT, + GLX_DRAWABLE_TYPE, GLX_WINDOW_BIT, + // TODO: + //GLX_BIND_TO_TEXTURE_RGBA_EXT, 1, + //GLX_BIND_TO_TEXTURE_TARGETS_EXT, GLX_TEXTURE_2D_BIT_EXT, + GLX_DOUBLEBUFFER, True, + GLX_RED_SIZE, 8, + GLX_GREEN_SIZE, 8, + GLX_BLUE_SIZE, 8, + GLX_ALPHA_SIZE, 0, + GLX_DEPTH_SIZE, 0, + None, None + }; + + // TODO: Cleanup + int c = 0; + GLXFBConfig *fb_configs = self->glXChooseFBConfig(display, DefaultScreen(display), glx_visual_attribs, &c); + if(c == 0 || !fb_configs) + return NULL; + + return fb_configs[0]; +} + +static bool gsr_egl_switch_to_glx_context(gsr_egl *self) { + // TODO: Cleanup + assert(gsr_window_get_display_server(self->window) == GSR_DISPLAY_SERVER_X11); + Display *display = gsr_window_get_display(self->window); + const Window window = (Window)gsr_window_get_window(self->window); + + if(self->egl_context) { + self->eglMakeCurrent(self->egl_display, NULL, NULL, NULL); + self->eglDestroyContext(self->egl_display, self->egl_context); + self->egl_context = NULL; + } + + if(self->egl_surface) { + self->eglDestroySurface(self->egl_display, self->egl_surface); + self->egl_surface = NULL; + } + + if(self->egl_display) { + self->eglTerminate(self->egl_display); + self->egl_display = NULL; + } + + self->glx_fb_config = glx_fb_config_choose(self, display); + if(!self->glx_fb_config) { + fprintf(stderr, "gsr error: gsr_egl_create_window failed: failed to find a suitable fb config\n"); + goto fail; + } + + // TODO: + //self->glx_context = self->glXCreateContextAttribsARB(display, self->glx_fb_config, NULL, True, context_attrib_list); + self->glx_context = self->glXCreateNewContext(display, self->glx_fb_config, GLX_RGBA_TYPE, NULL, True); + if(!self->glx_context) { + fprintf(stderr, "gsr error: gsr_egl_create_window failed: failed to create glx context\n"); + goto fail; + } + + if(!self->glXMakeContextCurrent(display, window, window, self->glx_context)) { + fprintf(stderr, "gsr error: gsr_egl_create_window failed: failed to make glx context current\n"); + goto fail; + } + + return true; + + fail: + if(self->glx_context) { + self->glXMakeContextCurrent(display, None, None, NULL); + self->glXDestroyContext(display, self->glx_context); + self->glx_context = NULL; + self->glx_fb_config = NULL; + } + return false; +} + static bool gsr_egl_load_egl(gsr_egl *self, void *library) { - dlsym_assign required_dlsym[] = { + const dlsym_assign required_dlsym[] = { { (void**)&self->eglGetError, "eglGetError" }, { (void**)&self->eglGetDisplay, "eglGetDisplay" }, { (void**)&self->eglInitialize, "eglInitialize" }, @@ -350,6 +227,27 @@ static bool gsr_egl_proc_load_egl(gsr_egl *self) { self->eglExportDMABUFImageQueryMESA = (FUNC_eglExportDMABUFImageQueryMESA)self->eglGetProcAddress("eglExportDMABUFImageQueryMESA"); self->eglExportDMABUFImageMESA = (FUNC_eglExportDMABUFImageMESA)self->eglGetProcAddress("eglExportDMABUFImageMESA"); self->glEGLImageTargetTexture2DOES = (FUNC_glEGLImageTargetTexture2DOES)self->eglGetProcAddress("glEGLImageTargetTexture2DOES"); + self->eglQueryDisplayAttribEXT = (FUNC_eglQueryDisplayAttribEXT)self->eglGetProcAddress("eglQueryDisplayAttribEXT"); + self->eglQueryDeviceStringEXT = (FUNC_eglQueryDeviceStringEXT)self->eglGetProcAddress("eglQueryDeviceStringEXT"); + self->eglQueryDmaBufModifiersEXT = (FUNC_eglQueryDmaBufModifiersEXT)self->eglGetProcAddress("eglQueryDmaBufModifiersEXT"); + + self->glCreateMemoryObjectsEXT = (FUNC_glCreateMemoryObjectsEXT)self->eglGetProcAddress("glCreateMemoryObjectsEXT"); + self->glImportMemoryFdEXT = (FUNC_glImportMemoryFdEXT)self->eglGetProcAddress("glImportMemoryFdEXT"); + self->glIsMemoryObjectEXT = (FUNC_glIsMemoryObjectEXT)self->eglGetProcAddress("glIsMemoryObjectEXT"); + self->glTexStorageMem2DEXT = (FUNC_glTexStorageMem2DEXT)self->eglGetProcAddress("glTexStorageMem2DEXT"); + self->glBufferStorageMemEXT = (FUNC_glBufferStorageMemEXT)self->eglGetProcAddress("glBufferStorageMemEXT"); + self->glNamedBufferStorageMemEXT = (FUNC_glNamedBufferStorageMemEXT)self->eglGetProcAddress("glNamedBufferStorageMemEXT"); + self->glMemoryObjectParameterivEXT = (FUNC_glMemoryObjectParameterivEXT)self->eglGetProcAddress("glMemoryObjectParameterivEXT"); + + if(!self->eglExportDMABUFImageQueryMESA) { + fprintf(stderr, "gsr error: gsr_egl_load failed: could not find eglExportDMABUFImageQueryMESA\n"); + return false; + } + + if(!self->eglExportDMABUFImageMESA) { + fprintf(stderr, "gsr error: gsr_egl_load failed: could not find eglExportDMABUFImageMESA\n"); + return false; + } if(!self->glEGLImageTargetTexture2DOES) { fprintf(stderr, "gsr error: gsr_egl_load failed: could not find glEGLImageTargetTexture2DOES\n"); @@ -359,23 +257,61 @@ static bool gsr_egl_proc_load_egl(gsr_egl *self) { return true; } +static bool gsr_egl_load_glx(gsr_egl *self, void *library) { + const dlsym_assign required_dlsym[] = { + { (void**)&self->glXGetProcAddress, "glXGetProcAddress" }, + { (void**)&self->glXChooseFBConfig, "glXChooseFBConfig" }, + { (void**)&self->glXMakeContextCurrent, "glXMakeContextCurrent" }, + { (void**)&self->glXCreateNewContext, "glXCreateNewContext" }, + { (void**)&self->glXDestroyContext, "glXDestroyContext" }, + { (void**)&self->glXSwapBuffers, "glXSwapBuffers" }, + + { NULL, NULL } + }; + + if(!dlsym_load_list(library, required_dlsym)) { + fprintf(stderr, "gsr error: gsr_egl_load failed: missing required symbols in libGLX.so.0\n"); + return false; + } + + self->glXCreateContextAttribsARB = (FUNC_glXCreateContextAttribsARB)self->glXGetProcAddress((const unsigned char*)"glXCreateContextAttribsARB"); + if(!self->glXCreateContextAttribsARB) { + fprintf(stderr, "gsr error: gsr_egl_load_glx failed: could not find glXCreateContextAttribsARB\n"); + return false; + } + + self->glXSwapIntervalEXT = (FUNC_glXSwapIntervalEXT)self->glXGetProcAddress((const unsigned char*)"glXSwapIntervalEXT"); + self->glXSwapIntervalMESA = (FUNC_glXSwapIntervalMESA)self->glXGetProcAddress((const unsigned char*)"glXSwapIntervalMESA"); + self->glXSwapIntervalSGI = (FUNC_glXSwapIntervalSGI)self->glXGetProcAddress((const unsigned char*)"glXSwapIntervalSGI"); + + return true; +} + static bool gsr_egl_load_gl(gsr_egl *self, void *library) { - dlsym_assign required_dlsym[] = { + const dlsym_assign required_dlsym[] = { { (void**)&self->glGetError, "glGetError" }, { (void**)&self->glGetString, "glGetString" }, + { (void**)&self->glFlush, "glFlush" }, + { (void**)&self->glFinish, "glFinish" }, { (void**)&self->glClear, "glClear" }, { (void**)&self->glClearColor, "glClearColor" }, { (void**)&self->glGenTextures, "glGenTextures" }, { (void**)&self->glDeleteTextures, "glDeleteTextures" }, + { (void**)&self->glActiveTexture, "glActiveTexture" }, { (void**)&self->glBindTexture, "glBindTexture" }, + { (void**)&self->glBindImageTexture, "glBindImageTexture" }, { (void**)&self->glTexParameteri, "glTexParameteri" }, + { (void**)&self->glTexParameteriv, "glTexParameteriv" }, + { (void**)&self->glTexParameterfv, "glTexParameterfv" }, { (void**)&self->glGetTexLevelParameteriv, "glGetTexLevelParameteriv" }, { (void**)&self->glTexImage2D, "glTexImage2D" }, - { (void**)&self->glCopyImageSubData, "glCopyImageSubData" }, - { (void**)&self->glClearTexImage, "glClearTexImage" }, + { (void**)&self->glTexSubImage2D, "glTexSubImage2D" }, + { (void**)&self->glGetTexImage, "glGetTexImage" }, { (void**)&self->glGenFramebuffers, "glGenFramebuffers" }, { (void**)&self->glBindFramebuffer, "glBindFramebuffer" }, { (void**)&self->glDeleteFramebuffers, "glDeleteFramebuffers" }, + { (void**)&self->glDispatchCompute, "glDispatchCompute" }, + { (void**)&self->glMemoryBarrier, "glMemoryBarrier" }, { (void**)&self->glViewport, "glViewport" }, { (void**)&self->glFramebufferTexture2D, "glFramebufferTexture2D" }, { (void**)&self->glDrawBuffers, "glDrawBuffers" }, @@ -406,9 +342,21 @@ static bool gsr_egl_load_gl(gsr_egl *self, void *library) { { (void**)&self->glEnableVertexAttribArray, "glEnableVertexAttribArray" }, { (void**)&self->glDrawArrays, "glDrawArrays" }, { (void**)&self->glEnable, "glEnable" }, + { (void**)&self->glDisable, "glDisable" }, { (void**)&self->glBlendFunc, "glBlendFunc" }, + { (void**)&self->glPixelStorei, "glPixelStorei" }, { (void**)&self->glGetUniformLocation, "glGetUniformLocation" }, { (void**)&self->glUniform1f, "glUniform1f" }, + { (void**)&self->glUniform2f, "glUniform2f" }, + { (void**)&self->glUniform1i, "glUniform1i" }, + { (void**)&self->glUniform2i, "glUniform2i" }, + { (void**)&self->glUniformMatrix2fv, "glUniformMatrix2fv" }, + { (void**)&self->glDebugMessageCallback, "glDebugMessageCallback" }, + { (void**)&self->glScissor, "glScissor" }, + { (void**)&self->glReadPixels, "glReadPixels" }, + { (void**)&self->glMapBuffer, "glMapBuffer" }, + { (void**)&self->glUnmapBuffer, "glUnmapBuffer" }, + { (void**)&self->glGetIntegerv, "glGetIntegerv" }, { NULL, NULL } }; @@ -421,56 +369,139 @@ static bool gsr_egl_load_gl(gsr_egl *self, void *library) { return true; } -bool gsr_egl_load(gsr_egl *self, Display *dpy, bool wayland) { - memset(self, 0, sizeof(gsr_egl)); - self->x11.dpy = dpy; +#define GL_DEBUG_TYPE_ERROR 0x824C +#define GL_DEBUG_SEVERITY_NOTIFICATION 0x826B +static void debug_callback(unsigned int source, unsigned int type, unsigned int id, unsigned int severity, int length, const char* message, const void* userParam) { + (void)source; + (void)id; + (void)length; + (void)userParam; + if(severity != GL_DEBUG_SEVERITY_NOTIFICATION) + fprintf(stderr, "gsr info: gl callback: %s type = 0x%x, severity = 0x%x, message = %s\n", type == GL_DEBUG_TYPE_ERROR ? "** GL ERROR **" : "", type, severity, message); +} + +/* TODO: check for glx swap control extension string (GLX_EXT_swap_control, etc) */ +static void set_vertical_sync_enabled(gsr_egl *egl, int enabled) { + int result = 0; + + if(egl->glXSwapIntervalEXT) { + assert(gsr_window_get_display_server(egl->window) == GSR_DISPLAY_SERVER_X11); + Display *display = gsr_window_get_display(egl->window); + const Window window = (Window)gsr_window_get_window(egl->window); + egl->glXSwapIntervalEXT(display, window, enabled ? 1 : 0); + } else if(egl->glXSwapIntervalMESA) { + result = egl->glXSwapIntervalMESA(enabled ? 1 : 0); + } else if(egl->glXSwapIntervalSGI) { + result = egl->glXSwapIntervalSGI(enabled ? 1 : 0); + } else { + static int warned = 0; + if (!warned) { + warned = 1; + fprintf(stderr, "gsr warning: setting vertical sync not supported\n"); + } + } - void *egl_lib = NULL; - void *gl_lib = NULL; + if(result != 0) + fprintf(stderr, "gsr warning: setting vertical sync failed\n"); +} + +static void gsr_egl_disable_vsync(gsr_egl *self) { + switch(self->context_type) { + case GSR_GL_CONTEXT_TYPE_EGL: { + self->eglSwapInterval(self->egl_display, 0); + break; + } + case GSR_GL_CONTEXT_TYPE_GLX: { + set_vertical_sync_enabled(self, 0); + break; + } + } +} + +bool gsr_egl_load(gsr_egl *self, gsr_window *window, bool is_monitor_capture, bool enable_debug) { + memset(self, 0, sizeof(gsr_egl)); + self->context_type = GSR_GL_CONTEXT_TYPE_EGL; + self->window = window; dlerror(); /* clear */ - egl_lib = dlopen("libEGL.so.1", RTLD_LAZY); - if(!egl_lib) { + self->egl_library = dlopen("libEGL.so.1", RTLD_LAZY); + if(!self->egl_library) { fprintf(stderr, "gsr error: gsr_egl_load: failed to load libEGL.so.1, error: %s\n", dlerror()); goto fail; } - gl_lib = dlopen("libGL.so.1", RTLD_LAZY); - if(!egl_lib) { + self->glx_library = dlopen("libGLX.so.0", RTLD_LAZY); + + self->gl_library = dlopen("libGL.so.1", RTLD_LAZY); + if(!self->gl_library) { fprintf(stderr, "gsr error: gsr_egl_load: failed to load libGL.so.1, error: %s\n", dlerror()); goto fail; } - if(!gsr_egl_load_egl(self, egl_lib)) + if(!gsr_egl_load_egl(self, self->egl_library)) + goto fail; + + /* In some distros (alpine for example libGLX doesn't exist, but libGL can be used instead) */ + if(!gsr_egl_load_glx(self, self->glx_library ? self->glx_library : self->gl_library)) goto fail; - if(!gsr_egl_load_gl(self, gl_lib)) + if(!gsr_egl_load_gl(self, self->gl_library)) goto fail; if(!gsr_egl_proc_load_egl(self)) goto fail; - if(!gsr_egl_create_window(self, wayland)) + if(!gsr_egl_create_window(self)) goto fail; + if(!gl_get_gpu_info(self, &self->gpu_info)) + goto fail; + + if(self->eglQueryDisplayAttribEXT && self->eglQueryDeviceStringEXT) { + intptr_t device = 0; + if(self->eglQueryDisplayAttribEXT(self->egl_display, EGL_DEVICE_EXT, &device) && device) + self->dri_card_path = self->eglQueryDeviceStringEXT((void*)device, EGL_DRM_DEVICE_FILE_EXT); + } + + /* Nvfbc requires glx */ + if(gsr_window_get_display_server(self->window) == GSR_DISPLAY_SERVER_X11 && is_monitor_capture && self->gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA) { + self->context_type = GSR_GL_CONTEXT_TYPE_GLX; + self->dri_card_path = NULL; + if(!gsr_egl_switch_to_glx_context(self)) + goto fail; + } + self->glEnable(GL_BLEND); self->glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA); + self->glPixelStorei(GL_PACK_ALIGNMENT, 1); + self->glPixelStorei(GL_UNPACK_ALIGNMENT, 1); + + if(enable_debug) { + self->glEnable(GL_DEBUG_OUTPUT); + self->glDebugMessageCallback(debug_callback, NULL); + } + + gsr_egl_disable_vsync(self); + + if(self->gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA) { + /* This fixes nvenc codecs unable to load on openSUSE tumbleweed because of a cuda error. Don't ask me why */ + const bool inside_flatpak = getenv("FLATPAK_ID") != NULL; + if(inside_flatpak) + system("flatpak-spawn --host -- nvidia-smi -f /dev/null"); + else + system("nvidia-smi -f /dev/null"); + } - self->egl_library = egl_lib; - self->gl_library = gl_lib; return true; fail: - if(egl_lib) - dlclose(egl_lib); - if(gl_lib) - dlclose(gl_lib); - memset(self, 0, sizeof(gsr_egl)); + gsr_egl_unload(self); return false; } void gsr_egl_unload(gsr_egl *self) { if(self->egl_context) { + self->eglMakeCurrent(self->egl_display, NULL, NULL, NULL); self->eglDestroyContext(self->egl_display, self->egl_context); self->egl_context = NULL; } @@ -485,59 +516,13 @@ void gsr_egl_unload(gsr_egl *self) { self->egl_display = NULL; } - if(self->x11.window) { - XDestroyWindow(self->x11.dpy, self->x11.window); - self->x11.window = None; - } - - gsr_egl_cleanup_frame(self); - - if(self->wayland.frame_callback) { - zwlr_export_dmabuf_frame_v1_destroy(self->wayland.frame_callback); - self->wayland.frame_callback = NULL; - } - - if(self->wayland.export_manager) { - zwlr_export_dmabuf_manager_v1_destroy(self->wayland.export_manager); - self->wayland.export_manager = NULL; - } - - if(self->wayland.window) { - wl_egl_window_destroy(self->wayland.window); - self->wayland.window = NULL; - } - - if(self->wayland.surface) { - wl_surface_destroy(self->wayland.surface); - self->wayland.surface = NULL; - } - - for(int i = 0; i < self->wayland.num_outputs; ++i) { - if(self->wayland.outputs[i].output) { - wl_output_destroy(self->wayland.outputs[i].output); - self->wayland.outputs[i].output = NULL; - } - - if(self->wayland.outputs[i].name) { - free(self->wayland.outputs[i].name); - self->wayland.outputs[i].name = NULL; - } - } - self->wayland.num_outputs = 0; - - if(self->wayland.compositor) { - wl_compositor_destroy(self->wayland.compositor); - self->wayland.compositor = NULL; - } - - if(self->wayland.registry) { - wl_registry_destroy(self->wayland.registry); - self->wayland.registry = NULL; - } - - if(self->wayland.dpy) { - wl_display_disconnect(self->wayland.dpy); - self->wayland.dpy = NULL; + if(self->glx_context) { + assert(gsr_window_get_display_server(self->window) == GSR_DISPLAY_SERVER_X11); + Display *display = gsr_window_get_display(self->window); + self->glXMakeContextCurrent(display, None, None, NULL); + self->glXDestroyContext(display, self->glx_context); + self->glx_context = NULL; + self->glx_fb_config = NULL; } if(self->egl_library) { @@ -545,6 +530,11 @@ void gsr_egl_unload(gsr_egl *self) { self->egl_library = NULL; } + if(self->glx_library) { + dlclose(self->glx_library); + self->glx_library = NULL; + } + if(self->gl_library) { dlclose(self->gl_library); self->gl_library = NULL; @@ -553,55 +543,16 @@ void gsr_egl_unload(gsr_egl *self) { memset(self, 0, sizeof(gsr_egl)); } -bool gsr_egl_supports_wayland_capture(gsr_egl *self) { - // TODO: wlroots capture is broken right now (black screen) on amd and multiple monitors - // so it has to be disabled right now. Find out why it happens and fix it. - (void)self; - return false; - //return !!self->wayland.export_manager && self->wayland.num_outputs > 0; -} - -bool gsr_egl_start_capture(gsr_egl *self, const char *monitor_to_capture) { - assert(monitor_to_capture); - if(!monitor_to_capture) - return false; - - if(!self->wayland.dpy) - return false; - - if(!gsr_egl_supports_wayland_capture(self)) - return false; - - if(self->wayland.frame_callback) - return false; - - self->wayland.output_to_capture = get_wayland_output_by_name(self, monitor_to_capture); - if(!self->wayland.output_to_capture) - return false; - - frame_capture_output(self); - return true; -} - -void gsr_egl_update(gsr_egl *self) { - if(!self->wayland.dpy) - return; - - // TODO: pselect on wl_display_get_fd before doing dispatch - wl_display_dispatch(self->wayland.dpy); -} - -void gsr_egl_cleanup_frame(gsr_egl *self) { - if(!self->wayland.dpy) - return; - - if(self->fd > 0) { - close(self->fd); - self->fd = 0; - } - - if(self->wayland.current_frame) { - //zwlr_export_dmabuf_frame_v1_destroy(self->wayland.current_frame); - self->wayland.current_frame = NULL; +void gsr_egl_swap_buffers(gsr_egl *self) { + /* This uses less cpu than swap buffer on nvidia */ + // TODO: Do these and remove swap + //self->glFlush(); + //self->glFinish(); + if(self->egl_display) { + self->eglSwapBuffers(self->egl_display, self->egl_surface); + } else if(gsr_window_get_display_server(self->window) == GSR_DISPLAY_SERVER_X11) { + Display *display = gsr_window_get_display(self->window); + const Window window = (Window)gsr_window_get_window(self->window); + self->glXSwapBuffers(display, window); } } diff --git a/src/encoder/encoder.c b/src/encoder/encoder.c new file mode 100644 index 0000000..0f8eda5 --- /dev/null +++ b/src/encoder/encoder.c @@ -0,0 +1,155 @@ +#include "../../include/encoder/encoder.h" +#include "../../include/utils.h" + +#include <string.h> +#include <stdio.h> + +#include <libavcodec/avcodec.h> +#include <libavformat/avformat.h> + +bool gsr_encoder_init(gsr_encoder *self, gsr_replay_storage replay_storage, size_t replay_buffer_num_packets, double replay_buffer_time, const char *replay_directory) { + memset(self, 0, sizeof(*self)); + self->num_recording_destinations = 0; + self->recording_destination_id_counter = 0; + + if(pthread_mutex_init(&self->file_write_mutex, NULL) != 0) { + fprintf(stderr, "gsr error: gsr_encoder_init: failed to create mutex\n"); + return false; + } + self->mutex_created = true; + + if(replay_buffer_num_packets > 0) { + self->replay_buffer = gsr_replay_buffer_create(replay_storage, replay_directory, replay_buffer_time, replay_buffer_num_packets); + if(!self->replay_buffer) { + fprintf(stderr, "gsr error: gsr_encoder_init: failed to create replay buffer\n"); + gsr_encoder_deinit(self); + return false; + } + } + + return true; +} + +void gsr_encoder_deinit(gsr_encoder *self) { + if(self->mutex_created) { + self->mutex_created = false; + pthread_mutex_destroy(&self->file_write_mutex); + } + + if(self->replay_buffer) { + gsr_replay_buffer_destroy(self->replay_buffer); + self->replay_buffer = NULL; + } + + self->num_recording_destinations = 0; + self->recording_destination_id_counter = 0; +} + +void gsr_encoder_receive_packets(gsr_encoder *self, AVCodecContext *codec_context, int64_t pts, int stream_index) { + for(;;) { + AVPacket *av_packet = av_packet_alloc(); + if(!av_packet) + break; + + av_packet->data = NULL; + av_packet->size = 0; + int res = avcodec_receive_packet(codec_context, av_packet); + if(res == 0) { // we have a packet, send the packet to the muxer + av_packet->stream_index = stream_index; + av_packet->pts = pts; + av_packet->dts = pts; + + if(self->replay_buffer) { + const double time_now = clock_get_monotonic_seconds(); + if(!gsr_replay_buffer_append(self->replay_buffer, av_packet, time_now)) + fprintf(stderr, "gsr error: gsr_encoder_receive_packets: failed to add replay buffer data\n"); + } + + pthread_mutex_lock(&self->file_write_mutex); + const bool is_keyframe = av_packet->flags & AV_PKT_FLAG_KEY; + for(size_t i = 0; i < self->num_recording_destinations; ++i) { + gsr_encoder_recording_destination *recording_destination = &self->recording_destinations[i]; + if(recording_destination->codec_context != codec_context) + continue; + + if(is_keyframe) + recording_destination->has_received_keyframe = true; + else if(!recording_destination->has_received_keyframe) + continue; + + av_packet->pts = pts - recording_destination->start_pts; + av_packet->dts = pts - recording_destination->start_pts; + + av_packet_rescale_ts(av_packet, codec_context->time_base, recording_destination->stream->time_base); + // TODO: Is av_interleaved_write_frame needed?. Answer: might be needed for mkv but dont use it! it causes frames to be inconsistent, skipping frames and duplicating frames. + // TODO: av_interleaved_write_frame might be needed for cfr, or always for flv + const int ret = av_write_frame(recording_destination->format_context, av_packet); + if(ret < 0) { + char error_buffer[AV_ERROR_MAX_STRING_SIZE]; + if(av_strerror(ret, error_buffer, sizeof(error_buffer)) < 0) + snprintf(error_buffer, sizeof(error_buffer), "Unknown error"); + fprintf(stderr, "gsr error: gsr_encoder_receive_packets: failed to write frame index %d to muxer, reason: %s (%d)\n", av_packet->stream_index, error_buffer, ret); + } + } + pthread_mutex_unlock(&self->file_write_mutex); + + av_packet_free(&av_packet); + } else if (res == AVERROR(EAGAIN)) { // we have no packet + // fprintf(stderr, "No packet!\n"); + av_packet_free(&av_packet); + break; + } else if (res == AVERROR_EOF) { // this is the end of the stream + av_packet_free(&av_packet); + fprintf(stderr, "End of stream!\n"); + break; + } else { + av_packet_free(&av_packet); + fprintf(stderr, "Unexpected error: %d\n", res); + break; + } + } +} + +size_t gsr_encoder_add_recording_destination(gsr_encoder *self, AVCodecContext *codec_context, AVFormatContext *format_context, AVStream *stream, int64_t start_pts) { + if(self->num_recording_destinations >= GSR_MAX_RECORDING_DESTINATIONS) { + fprintf(stderr, "gsr error: gsr_encoder_add_recording_destination: failed to add destination, reached the max amount of recording destinations (%d)\n", GSR_MAX_RECORDING_DESTINATIONS); + return (size_t)-1; + } + + for(size_t i = 0; i < self->num_recording_destinations; ++i) { + if(self->recording_destinations[i].stream == stream) { + fprintf(stderr, "gsr error: gsr_encoder_add_recording_destination: failed to add destination, the stream %p already exists as an output\n", (void*)stream); + return (size_t)-1; + } + } + + pthread_mutex_lock(&self->file_write_mutex); + gsr_encoder_recording_destination *recording_destination = &self->recording_destinations[self->num_recording_destinations]; + recording_destination->id = self->recording_destination_id_counter; + recording_destination->codec_context = codec_context; + recording_destination->format_context = format_context; + recording_destination->stream = stream; + recording_destination->start_pts = start_pts; + recording_destination->has_received_keyframe = false; + + ++self->recording_destination_id_counter; + ++self->num_recording_destinations; + pthread_mutex_unlock(&self->file_write_mutex); + + return recording_destination->id; +} + +bool gsr_encoder_remove_recording_destination(gsr_encoder *self, size_t id) { + bool found = false; + pthread_mutex_lock(&self->file_write_mutex); + for(size_t i = 0; i < self->num_recording_destinations; ++i) { + if(self->recording_destinations[i].id == id) { + self->recording_destinations[i] = self->recording_destinations[self->num_recording_destinations - 1]; + --self->num_recording_destinations; + found = true; + break; + } + } + pthread_mutex_unlock(&self->file_write_mutex); + return found; +} diff --git a/src/encoder/video/nvenc.c b/src/encoder/video/nvenc.c new file mode 100644 index 0000000..5f578c2 --- /dev/null +++ b/src/encoder/video/nvenc.c @@ -0,0 +1,237 @@ +#include "../../../include/encoder/video/nvenc.h" +#include "../../../include/egl.h" +#include "../../../include/cuda.h" +#include "../../../include/window/window.h" + +#include <libavcodec/avcodec.h> +#include <libavutil/hwcontext_cuda.h> + +#include <stdlib.h> + +typedef struct { + gsr_video_encoder_nvenc_params params; + + unsigned int target_textures[2]; + + AVBufferRef *device_ctx; + + gsr_cuda cuda; + CUgraphicsResource cuda_graphics_resources[2]; + CUarray mapped_arrays[2]; + CUstream cuda_stream; +} gsr_video_encoder_nvenc; + +static bool gsr_video_encoder_nvenc_setup_context(gsr_video_encoder_nvenc *self, AVCodecContext *video_codec_context) { + self->device_ctx = av_hwdevice_ctx_alloc(AV_HWDEVICE_TYPE_CUDA); + if(!self->device_ctx) { + fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_context failed: failed to create hardware device context\n"); + return false; + } + + AVHWDeviceContext *hw_device_context = (AVHWDeviceContext*)self->device_ctx->data; + AVCUDADeviceContext *cuda_device_context = (AVCUDADeviceContext*)hw_device_context->hwctx; + cuda_device_context->cuda_ctx = self->cuda.cu_ctx; + if(av_hwdevice_ctx_init(self->device_ctx) < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_context failed: failed to create hardware device context\n"); + av_buffer_unref(&self->device_ctx); + return false; + } + + AVBufferRef *frame_context = av_hwframe_ctx_alloc(self->device_ctx); + if(!frame_context) { + fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_context failed: failed to create hwframe context\n"); + av_buffer_unref(&self->device_ctx); + return false; + } + + AVHWFramesContext *hw_frame_context = (AVHWFramesContext*)frame_context->data; + hw_frame_context->width = video_codec_context->width; + hw_frame_context->height = video_codec_context->height; + hw_frame_context->sw_format = self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? AV_PIX_FMT_P010LE : AV_PIX_FMT_NV12; + hw_frame_context->format = video_codec_context->pix_fmt; + hw_frame_context->device_ctx = (AVHWDeviceContext*)self->device_ctx->data; + + if (av_hwframe_ctx_init(frame_context) < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_context failed: failed to initialize hardware frame context " + "(note: ffmpeg version needs to be > 4.0)\n"); + av_buffer_unref(&self->device_ctx); + //av_buffer_unref(&frame_context); + return false; + } + + self->cuda_stream = cuda_device_context->stream; + video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); + av_buffer_unref(&frame_context); + return true; +} + +static bool cuda_register_opengl_texture(gsr_cuda *cuda, CUgraphicsResource *cuda_graphics_resource, CUarray *mapped_array, unsigned int texture_id) { + CUresult res; + res = cuda->cuGraphicsGLRegisterImage(cuda_graphics_resource, texture_id, GL_TEXTURE_2D, CU_GRAPHICS_REGISTER_FLAGS_NONE); + if (res != CUDA_SUCCESS) { + const char *err_str = "unknown"; + cuda->cuGetErrorString(res, &err_str); + fprintf(stderr, "gsr error: cuda_register_opengl_texture: cuGraphicsGLRegisterImage failed, error: %s, texture " "id: %u\n", err_str, texture_id); + return false; + } + + res = cuda->cuGraphicsResourceSetMapFlags(*cuda_graphics_resource, CU_GRAPHICS_MAP_RESOURCE_FLAGS_NONE); + res = cuda->cuGraphicsMapResources(1, cuda_graphics_resource, 0); + + res = cuda->cuGraphicsSubResourceGetMappedArray(mapped_array, *cuda_graphics_resource, 0, 0); + return true; +} + +static bool gsr_video_encoder_nvenc_setup_textures(gsr_video_encoder_nvenc *self, AVCodecContext *video_codec_context, AVFrame *frame) { + const int res = av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, frame, 0); + if(res < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_textures: av_hwframe_get_buffer failed: %d\n", res); + return false; + } + + const unsigned int internal_formats_nv12[2] = { GL_R8, GL_RG8 }; + const unsigned int internal_formats_p010[2] = { GL_R16, GL_RG16 }; + const unsigned int formats[2] = { GL_RED, GL_RG }; + const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size + + for(int i = 0; i < 2; ++i) { + self->target_textures[i] = gl_create_texture(self->params.egl, video_codec_context->width / div[i], video_codec_context->height / div[i], self->params.color_depth == GSR_COLOR_DEPTH_8_BITS ? internal_formats_nv12[i] : internal_formats_p010[i], formats[i], GL_NEAREST); + if(self->target_textures[i] == 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_textures: failed to create opengl texture\n"); + return false; + } + + if(!cuda_register_opengl_texture(&self->cuda, &self->cuda_graphics_resources[i], &self->mapped_arrays[i], self->target_textures[i])) { + return false; + } + } + + return true; +} + +static void gsr_video_encoder_nvenc_stop(gsr_video_encoder_nvenc *self, AVCodecContext *video_codec_context); + +static bool gsr_video_encoder_nvenc_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) { + gsr_video_encoder_nvenc *self = encoder->priv; + + const bool is_x11 = gsr_window_get_display_server(self->params.egl->window) == GSR_DISPLAY_SERVER_X11; + const bool overclock = is_x11 ? self->params.overclock : false; + Display *display = is_x11 ? gsr_window_get_display(self->params.egl->window) : NULL; + if(!gsr_cuda_load(&self->cuda, display, overclock)) { + fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_start: failed to load cuda\n"); + gsr_video_encoder_nvenc_stop(self, video_codec_context); + return false; + } + + video_codec_context->width = FFALIGN(video_codec_context->width, 2); + video_codec_context->height = FFALIGN(video_codec_context->height, 2); + + if(video_codec_context->width < 128) + video_codec_context->width = 128; + + if(video_codec_context->height < 128) + video_codec_context->height = 128; + + frame->width = video_codec_context->width; + frame->height = video_codec_context->height; + + if(!gsr_video_encoder_nvenc_setup_context(self, video_codec_context)) { + gsr_video_encoder_nvenc_stop(self, video_codec_context); + return false; + } + + if(!gsr_video_encoder_nvenc_setup_textures(self, video_codec_context, frame)) { + gsr_video_encoder_nvenc_stop(self, video_codec_context); + return false; + } + + return true; +} + +void gsr_video_encoder_nvenc_stop(gsr_video_encoder_nvenc *self, AVCodecContext *video_codec_context) { + self->params.egl->glDeleteTextures(2, self->target_textures); + self->target_textures[0] = 0; + self->target_textures[1] = 0; + + if(video_codec_context->hw_frames_ctx) + av_buffer_unref(&video_codec_context->hw_frames_ctx); + if(self->device_ctx) + av_buffer_unref(&self->device_ctx); + + if(self->cuda.cu_ctx) { + for(int i = 0; i < 2; ++i) { + if(self->cuda_graphics_resources[i]) { + self->cuda.cuGraphicsUnmapResources(1, &self->cuda_graphics_resources[i], 0); + self->cuda.cuGraphicsUnregisterResource(self->cuda_graphics_resources[i]); + self->cuda_graphics_resources[i] = 0; + } + } + } + + gsr_cuda_unload(&self->cuda); +} + +static void gsr_video_encoder_nvenc_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame, gsr_color_conversion *color_conversion) { + gsr_video_encoder_nvenc *self = encoder->priv; + const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size + for(int i = 0; i < 2; ++i) { + CUDA_MEMCPY2D memcpy_struct; + memcpy_struct.srcXInBytes = 0; + memcpy_struct.srcY = 0; + memcpy_struct.srcMemoryType = CU_MEMORYTYPE_ARRAY; + + memcpy_struct.dstXInBytes = 0; + memcpy_struct.dstY = 0; + memcpy_struct.dstMemoryType = CU_MEMORYTYPE_DEVICE; + + memcpy_struct.srcArray = self->mapped_arrays[i]; + memcpy_struct.srcPitch = frame->width / div[i]; + memcpy_struct.dstDevice = (CUdeviceptr)frame->data[i]; + memcpy_struct.dstPitch = frame->linesize[i]; + memcpy_struct.WidthInBytes = frame->width * (self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? 2 : 1); + memcpy_struct.Height = frame->height / div[i]; + // TODO: Remove this copy if possible + self->cuda.cuMemcpy2DAsync_v2(&memcpy_struct, self->cuda_stream); + } + + // TODO: needed? + self->cuda.cuStreamSynchronize(self->cuda_stream); +} + +static void gsr_video_encoder_nvenc_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) { + gsr_video_encoder_nvenc *self = encoder->priv; + textures[0] = self->target_textures[0]; + textures[1] = self->target_textures[1]; + *num_textures = 2; + *destination_color = self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? GSR_DESTINATION_COLOR_P010 : GSR_DESTINATION_COLOR_NV12; +} + +static void gsr_video_encoder_nvenc_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) { + gsr_video_encoder_nvenc_stop(encoder->priv, video_codec_context); + free(encoder->priv); + free(encoder); +} + +gsr_video_encoder* gsr_video_encoder_nvenc_create(const gsr_video_encoder_nvenc_params *params) { + gsr_video_encoder *encoder = calloc(1, sizeof(gsr_video_encoder)); + if(!encoder) + return NULL; + + gsr_video_encoder_nvenc *encoder_cuda = calloc(1, sizeof(gsr_video_encoder_nvenc)); + if(!encoder_cuda) { + free(encoder); + return NULL; + } + + encoder_cuda->params = *params; + + *encoder = (gsr_video_encoder) { + .start = gsr_video_encoder_nvenc_start, + .copy_textures_to_frame = gsr_video_encoder_nvenc_copy_textures_to_frame, + .get_textures = gsr_video_encoder_nvenc_get_textures, + .destroy = gsr_video_encoder_nvenc_destroy, + .priv = encoder_cuda + }; + + return encoder; +} diff --git a/src/encoder/video/software.c b/src/encoder/video/software.c new file mode 100644 index 0000000..627cdea --- /dev/null +++ b/src/encoder/video/software.c @@ -0,0 +1,126 @@ +#include "../../../include/encoder/video/software.h" +#include "../../../include/egl.h" +#include "../../../include/utils.h" + +#include <libavcodec/avcodec.h> +#include <libavutil/frame.h> + +#include <stdlib.h> + +#define LINESIZE_ALIGNMENT 4 + +typedef struct { + gsr_video_encoder_software_params params; + + unsigned int target_textures[2]; +} gsr_video_encoder_software; + +static bool gsr_video_encoder_software_setup_textures(gsr_video_encoder_software *self, AVCodecContext *video_codec_context, AVFrame *frame) { + int res = av_frame_get_buffer(frame, LINESIZE_ALIGNMENT); + if(res < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_software_setup_textures: av_frame_get_buffer failed: %d\n", res); + return false; + } + + res = av_frame_make_writable(frame); + if(res < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_software_setup_textures: av_frame_make_writable failed: %d\n", res); + return false; + } + + const unsigned int internal_formats_nv12[2] = { GL_R8, GL_RG8 }; + const unsigned int internal_formats_p010[2] = { GL_R16, GL_RG16 }; + const unsigned int formats[2] = { GL_RED, GL_RG }; + const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size + + for(int i = 0; i < 2; ++i) { + self->target_textures[i] = gl_create_texture(self->params.egl, video_codec_context->width / div[i], video_codec_context->height / div[i], self->params.color_depth == GSR_COLOR_DEPTH_8_BITS ? internal_formats_nv12[i] : internal_formats_p010[i], formats[i], GL_NEAREST); + if(self->target_textures[i] == 0) { + fprintf(stderr, "gsr error: gsr_capture_kms_setup_cuda_textures: failed to create opengl texture\n"); + return false; + } + } + + return true; +} + +static void gsr_video_encoder_software_stop(gsr_video_encoder_software *self, AVCodecContext *video_codec_context); + +static bool gsr_video_encoder_software_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) { + gsr_video_encoder_software *self = encoder->priv; + + video_codec_context->width = FFALIGN(video_codec_context->width, LINESIZE_ALIGNMENT); + video_codec_context->height = FFALIGN(video_codec_context->height, 2); + + frame->width = video_codec_context->width; + frame->height = video_codec_context->height; + + if(!gsr_video_encoder_software_setup_textures(self, video_codec_context, frame)) { + gsr_video_encoder_software_stop(self, video_codec_context); + return false; + } + + return true; +} + +void gsr_video_encoder_software_stop(gsr_video_encoder_software *self, AVCodecContext *video_codec_context) { + (void)video_codec_context; + self->params.egl->glDeleteTextures(2, self->target_textures); + self->target_textures[0] = 0; + self->target_textures[1] = 0; +} + +static void gsr_video_encoder_software_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame, gsr_color_conversion *color_conversion) { + gsr_video_encoder_software *self = encoder->priv; + // TODO: hdr support + const unsigned int formats[2] = { GL_RED, GL_RG }; + for(int i = 0; i < 2; ++i) { + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->target_textures[i]); + // We could use glGetTexSubImage and then we wouldn't have to use a specific linesize (LINESIZE_ALIGNMENT) that adds padding, + // but glGetTexSubImage is only available starting from opengl 4.5. + self->params.egl->glGetTexImage(GL_TEXTURE_2D, 0, formats[i], GL_UNSIGNED_BYTE, frame->data[i]); + } + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + // cap_kms->kms.base.egl->eglSwapBuffers(cap_kms->kms.base.egl->egl_display, cap_kms->kms.base.egl->egl_surface); + + //self->params.egl->glFlush(); + //self->params.egl->glFinish(); +} + +static void gsr_video_encoder_software_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) { + gsr_video_encoder_software *self = encoder->priv; + textures[0] = self->target_textures[0]; + textures[1] = self->target_textures[1]; + *num_textures = 2; + *destination_color = self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? GSR_DESTINATION_COLOR_P010 : GSR_DESTINATION_COLOR_NV12; +} + +static void gsr_video_encoder_software_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) { + gsr_video_encoder_software_stop(encoder->priv, video_codec_context); + free(encoder->priv); + free(encoder); +} + +gsr_video_encoder* gsr_video_encoder_software_create(const gsr_video_encoder_software_params *params) { + gsr_video_encoder *encoder = calloc(1, sizeof(gsr_video_encoder)); + if(!encoder) + return NULL; + + gsr_video_encoder_software *encoder_software = calloc(1, sizeof(gsr_video_encoder_software)); + if(!encoder_software) { + free(encoder); + return NULL; + } + + encoder_software->params = *params; + + *encoder = (gsr_video_encoder) { + .start = gsr_video_encoder_software_start, + .copy_textures_to_frame = gsr_video_encoder_software_copy_textures_to_frame, + .get_textures = gsr_video_encoder_software_get_textures, + .destroy = gsr_video_encoder_software_destroy, + .priv = encoder_software + }; + + return encoder; +} diff --git a/src/encoder/video/vaapi.c b/src/encoder/video/vaapi.c new file mode 100644 index 0000000..c7ccd26 --- /dev/null +++ b/src/encoder/video/vaapi.c @@ -0,0 +1,259 @@ +#include "../../../include/encoder/video/vaapi.h" +#include "../../../include/utils.h" +#include "../../../include/egl.h" + +#include <libavcodec/avcodec.h> +#include <libavutil/hwcontext_vaapi.h> +#include <libavutil/intreadwrite.h> + +#include <va/va_drmcommon.h> + +#include <stdlib.h> +#include <unistd.h> +#include <fcntl.h> + +typedef struct { + gsr_video_encoder_vaapi_params params; + + unsigned int target_textures[2]; + + AVBufferRef *device_ctx; + VADisplay va_dpy; + VADRMPRIMESurfaceDescriptor prime; +} gsr_video_encoder_vaapi; + +static bool gsr_video_encoder_vaapi_setup_context(gsr_video_encoder_vaapi *self, AVCodecContext *video_codec_context) { + char render_path[128]; + if(!gsr_card_path_get_render_path(self->params.egl->card_path, render_path)) { + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_context: failed to get /dev/dri/renderDXXX file from %s\n", self->params.egl->card_path); + return false; + } + + if(av_hwdevice_ctx_create(&self->device_ctx, AV_HWDEVICE_TYPE_VAAPI, render_path, NULL, 0) < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_context: failed to create hardware device context\n"); + return false; + } + + AVBufferRef *frame_context = av_hwframe_ctx_alloc(self->device_ctx); + if(!frame_context) { + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_context: failed to create hwframe context\n"); + av_buffer_unref(&self->device_ctx); + return false; + } + + AVHWFramesContext *hw_frame_context = (AVHWFramesContext*)frame_context->data; + hw_frame_context->width = video_codec_context->width; + hw_frame_context->height = video_codec_context->height; + hw_frame_context->sw_format = self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? AV_PIX_FMT_P010LE : AV_PIX_FMT_NV12; + hw_frame_context->format = video_codec_context->pix_fmt; + hw_frame_context->device_ctx = (AVHWDeviceContext*)self->device_ctx->data; + + //hw_frame_context->initial_pool_size = 20; + + AVVAAPIDeviceContext *vactx = ((AVHWDeviceContext*)self->device_ctx->data)->hwctx; + self->va_dpy = vactx->display; + + if (av_hwframe_ctx_init(frame_context) < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_context: failed to initialize hardware frame context " + "(note: ffmpeg version needs to be > 4.0)\n"); + av_buffer_unref(&self->device_ctx); + //av_buffer_unref(&frame_context); + return false; + } + + video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); + av_buffer_unref(&frame_context); + return true; +} + +static uint32_t fourcc(uint32_t a, uint32_t b, uint32_t c, uint32_t d) { + return (d << 24) | (c << 16) | (b << 8) | a; +} + +static bool gsr_video_encoder_vaapi_setup_textures(gsr_video_encoder_vaapi *self, AVCodecContext *video_codec_context, AVFrame *frame) { + const int res = av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, frame, 0); + if(res < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_textures: av_hwframe_get_buffer failed: %d\n", res); + return false; + } + + VASurfaceID target_surface_id = (uintptr_t)frame->data[3]; + + VAStatus va_status = vaExportSurfaceHandle(self->va_dpy, target_surface_id, VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME_2, VA_EXPORT_SURFACE_WRITE_ONLY | VA_EXPORT_SURFACE_SEPARATE_LAYERS, &self->prime); + if(va_status != VA_STATUS_SUCCESS) { + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_textures: vaExportSurfaceHandle failed, error: %d\n", va_status); + return false; + } + vaSyncSurface(self->va_dpy, target_surface_id); + + const uint32_t formats_nv12[2] = { fourcc('R', '8', ' ', ' '), fourcc('G', 'R', '8', '8') }; + const uint32_t formats_p010[2] = { fourcc('R', '1', '6', ' '), fourcc('G', 'R', '3', '2') }; + + if(self->prime.fourcc == VA_FOURCC_NV12 || self->prime.fourcc == VA_FOURCC_P010) { + const uint32_t *formats = self->prime.fourcc == VA_FOURCC_NV12 ? formats_nv12 : formats_p010; + const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size + const float border_colors[2][4] = { + {0.0f, 0.0f, 0.0f, 1.0f}, + {0.5f, 0.5f, 0.0f, 1.0f} + }; + + self->params.egl->glGenTextures(2, self->target_textures); + for(int i = 0; i < 2; ++i) { + const int layer = i; + + int fds[4]; + uint32_t offsets[4]; + uint32_t pitches[4]; + uint64_t modifiers[4]; + for(uint32_t j = 0; j < self->prime.layers[layer].num_planes; ++j) { + // TODO: Close these? in _stop, using self->prime + fds[j] = self->prime.objects[self->prime.layers[layer].object_index[j]].fd; + offsets[j] = self->prime.layers[layer].offset[j]; + pitches[j] = self->prime.layers[layer].pitch[j]; + modifiers[j] = self->prime.objects[self->prime.layers[layer].object_index[j]].drm_format_modifier; + } + + intptr_t img_attr[44]; + setup_dma_buf_attrs(img_attr, formats[i], self->prime.width / div[i], self->prime.height / div[i], + fds, offsets, pitches, modifiers, self->prime.layers[layer].num_planes, true); + + while(self->params.egl->eglGetError() != EGL_SUCCESS){} + EGLImage image = self->params.egl->eglCreateImage(self->params.egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr); + if(!image) { + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_textures: failed to create egl image from drm fd for output drm fd, error: %d\n", self->params.egl->eglGetError()); + return false; + } + + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->target_textures[i]); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->params.egl->glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, border_colors[i]); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST); + + while(self->params.egl->glGetError()) {} + while(self->params.egl->eglGetError() != EGL_SUCCESS){} + self->params.egl->glEGLImageTargetTexture2DOES(GL_TEXTURE_2D, image); + if(self->params.egl->glGetError() != 0 || self->params.egl->eglGetError() != EGL_SUCCESS) { + // TODO: Get the error properly + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_textures: failed to bind egl image to gl texture, error: %d\n", self->params.egl->eglGetError()); + self->params.egl->eglDestroyImage(self->params.egl->egl_display, image); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + return false; + } + + self->params.egl->eglDestroyImage(self->params.egl->egl_display, image); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + } + + return true; + } else { + fprintf(stderr, "gsr error: gsr_video_encoder_vaapi_setup_textures: unexpected fourcc %u for output drm fd, expected nv12 or p010\n", self->prime.fourcc); + return false; + } +} + +static void gsr_video_encoder_vaapi_stop(gsr_video_encoder_vaapi *self, AVCodecContext *video_codec_context); + +static bool gsr_video_encoder_vaapi_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) { + gsr_video_encoder_vaapi *self = encoder->priv; + + if(self->params.egl->gpu_info.vendor == GSR_GPU_VENDOR_AMD && video_codec_context->codec_id == AV_CODEC_ID_HEVC) { + // TODO: dont do this if using ffmpeg reports that this is not needed (AMD driver bug that was fixed recently) + video_codec_context->width = FFALIGN(video_codec_context->width, 64); + video_codec_context->height = FFALIGN(video_codec_context->height, 16); + } else if(self->params.egl->gpu_info.vendor == GSR_GPU_VENDOR_AMD && video_codec_context->codec_id == AV_CODEC_ID_AV1) { + // TODO: Dont do this for VCN 5 and forward which should fix this hardware bug + video_codec_context->width = FFALIGN(video_codec_context->width, 64); + // AMD driver has special case handling for 1080 height to set it to 1082 instead of 1088 (1080 aligned to 16). + // TODO: Set height to 1082 in this case, but it wont work because it will be aligned to 1088. + if(video_codec_context->height == 1080) { + video_codec_context->height = 1080; + } else { + video_codec_context->height = FFALIGN(video_codec_context->height, 16); + } + } else { + video_codec_context->width = FFALIGN(video_codec_context->width, 2); + video_codec_context->height = FFALIGN(video_codec_context->height, 2); + } + + if(FFALIGN(video_codec_context->width, 2) != FFALIGN(frame->width, 2) || FFALIGN(video_codec_context->height, 2) != FFALIGN(frame->height, 2)) { + fprintf(stderr, "gsr warning: gsr_video_encoder_vaapi_start: black bars have been added to the video because of a bug in AMD drivers/hardware. Record with h264 codec instead (-k h264) to get around this issue\n"); + } + + if(video_codec_context->width < 128) + video_codec_context->width = 128; + + if(video_codec_context->height < 128) + video_codec_context->height = 128; + + frame->width = video_codec_context->width; + frame->height = video_codec_context->height; + + if(!gsr_video_encoder_vaapi_setup_context(self, video_codec_context)) { + gsr_video_encoder_vaapi_stop(self, video_codec_context); + return false; + } + + if(!gsr_video_encoder_vaapi_setup_textures(self, video_codec_context, frame)) { + gsr_video_encoder_vaapi_stop(self, video_codec_context); + return false; + } + + return true; +} + +void gsr_video_encoder_vaapi_stop(gsr_video_encoder_vaapi *self, AVCodecContext *video_codec_context) { + self->params.egl->glDeleteTextures(2, self->target_textures); + self->target_textures[0] = 0; + self->target_textures[1] = 0; + + if(video_codec_context->hw_frames_ctx) + av_buffer_unref(&video_codec_context->hw_frames_ctx); + if(self->device_ctx) + av_buffer_unref(&self->device_ctx); + + for(uint32_t i = 0; i < self->prime.num_objects; ++i) { + if(self->prime.objects[i].fd > 0) { + close(self->prime.objects[i].fd); + self->prime.objects[i].fd = 0; + } + } +} + +static void gsr_video_encoder_vaapi_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) { + gsr_video_encoder_vaapi *self = encoder->priv; + textures[0] = self->target_textures[0]; + textures[1] = self->target_textures[1]; + *num_textures = 2; + *destination_color = self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? GSR_DESTINATION_COLOR_P010 : GSR_DESTINATION_COLOR_NV12; +} + +static void gsr_video_encoder_vaapi_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) { + gsr_video_encoder_vaapi_stop(encoder->priv, video_codec_context); + free(encoder->priv); + free(encoder); +} + +gsr_video_encoder* gsr_video_encoder_vaapi_create(const gsr_video_encoder_vaapi_params *params) { + gsr_video_encoder *encoder = calloc(1, sizeof(gsr_video_encoder)); + if(!encoder) + return NULL; + + gsr_video_encoder_vaapi *encoder_vaapi = calloc(1, sizeof(gsr_video_encoder_vaapi)); + if(!encoder_vaapi) { + free(encoder); + return NULL; + } + + encoder_vaapi->params = *params; + + *encoder = (gsr_video_encoder) { + .start = gsr_video_encoder_vaapi_start, + .get_textures = gsr_video_encoder_vaapi_get_textures, + .destroy = gsr_video_encoder_vaapi_destroy, + .priv = encoder_vaapi + }; + + return encoder; +} diff --git a/src/encoder/video/video.c b/src/encoder/video/video.c new file mode 100644 index 0000000..ce3b61b --- /dev/null +++ b/src/encoder/video/video.c @@ -0,0 +1,28 @@ +#include "../../../include/encoder/video/video.h" + +#include <assert.h> + +bool gsr_video_encoder_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) { + assert(!encoder->started); + bool res = encoder->start(encoder, video_codec_context, frame); + if(res) + encoder->started = true; + return res; +} + +void gsr_video_encoder_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) { + assert(encoder->started); + encoder->started = false; + encoder->destroy(encoder, video_codec_context); +} + +void gsr_video_encoder_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame, gsr_color_conversion *color_conversion) { + assert(encoder->started); + if(encoder->copy_textures_to_frame) + encoder->copy_textures_to_frame(encoder, frame, color_conversion); +} + +void gsr_video_encoder_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) { + assert(encoder->started); + encoder->get_textures(encoder, textures, num_textures, destination_color); +} diff --git a/src/encoder/video/vulkan.c b/src/encoder/video/vulkan.c new file mode 100644 index 0000000..802934d --- /dev/null +++ b/src/encoder/video/vulkan.c @@ -0,0 +1,309 @@ +#include "../../../include/encoder/video/vulkan.h" +#include "../../../include/utils.h" +#include "../../../include/egl.h" + +#include <libavcodec/avcodec.h> +#define VK_NO_PROTOTYPES +#include <libavutil/hwcontext_vulkan.h> + +//#include <vulkan/vulkan_core.h> + +#define GL_HANDLE_TYPE_OPAQUE_FD_EXT 0x9586 +#define GL_TEXTURE_TILING_EXT 0x9580 +#define GL_OPTIMAL_TILING_EXT 0x9584 +#define GL_LINEAR_TILING_EXT 0x9585 + +typedef struct { + gsr_video_encoder_vulkan_params params; + unsigned int target_textures[2]; + AVBufferRef *device_ctx; +} gsr_video_encoder_vulkan; + +static bool gsr_video_encoder_vulkan_setup_context(gsr_video_encoder_vulkan *self, AVCodecContext *video_codec_context) { + AVDictionary *options = NULL; + //av_dict_set(&options, "linear_images", "1", 0); + //av_dict_set(&options, "disable_multiplane", "1", 0); +#if 0 + // TODO: Use correct device + if(av_hwdevice_ctx_create(&self->device_ctx, AV_HWDEVICE_TYPE_VULKAN, NULL, options, 0) < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_vulkan_setup_context: failed to create hardware device context\n"); + return false; + } + + AVBufferRef *frame_context = av_hwframe_ctx_alloc(self->device_ctx); + if(!frame_context) { + fprintf(stderr, "gsr error: gsr_video_encoder_vulkan_setup_context: failed to create hwframe context\n"); + av_buffer_unref(&self->device_ctx); + return false; + } + + AVHWFramesContext *hw_frame_context = (AVHWFramesContext*)frame_context->data; + hw_frame_context->width = video_codec_context->width; + hw_frame_context->height = video_codec_context->height; + hw_frame_context->sw_format = self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? AV_PIX_FMT_P010LE : AV_PIX_FMT_NV12; + hw_frame_context->format = video_codec_context->pix_fmt; + hw_frame_context->device_ctx = (AVHWDeviceContext*)self->device_ctx->data; + + //AVVulkanFramesContext *vk_frame_ctx = (AVVulkanFramesContext*)hw_frame_context->hwctx; + //hw_frame_context->initial_pool_size = 20; + + if (av_hwframe_ctx_init(frame_context) < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_vulkan_setup_context: failed to initialize hardware frame context " + "(note: ffmpeg version needs to be > 4.0)\n"); + av_buffer_unref(&self->device_ctx); + //av_buffer_unref(&frame_context); + return false; + } + + video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); + av_buffer_unref(&frame_context); +#endif + return true; +} + +static AVVulkanDeviceContext* video_codec_context_get_vulkan_data(AVCodecContext *video_codec_context) { + AVBufferRef *hw_frames_ctx = video_codec_context->hw_frames_ctx; + if(!hw_frames_ctx) + return NULL; + + AVHWFramesContext *hw_frame_context = (AVHWFramesContext*)hw_frames_ctx->data; + AVHWDeviceContext *device_context = (AVHWDeviceContext*)hw_frame_context->device_ctx; + if(device_context->type != AV_HWDEVICE_TYPE_VULKAN) + return NULL; + + return (AVVulkanDeviceContext*)device_context->hwctx; +} + +static uint32_t get_memory_type_idx(VkPhysicalDevice pdev, const VkMemoryRequirements *mem_reqs, VkMemoryPropertyFlagBits prop_flags, PFN_vkGetPhysicalDeviceMemoryProperties vkGetPhysicalDeviceMemoryProperties) { + VkPhysicalDeviceMemoryProperties pdev_mem_props; + uint32_t i; + + vkGetPhysicalDeviceMemoryProperties(pdev, &pdev_mem_props); + + for (i = 0; i < pdev_mem_props.memoryTypeCount; i++) { + const VkMemoryType *type = &pdev_mem_props.memoryTypes[i]; + + if ((mem_reqs->memoryTypeBits & (1 << i)) && + (type->propertyFlags & prop_flags) == prop_flags) { + return i; + break; + } + } + return UINT32_MAX; +} + +static bool gsr_video_encoder_vulkan_setup_textures(gsr_video_encoder_vulkan *self, AVCodecContext *video_codec_context, AVFrame *frame) { + const int res = av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, frame, 0); + if(res < 0) { + fprintf(stderr, "gsr error: gsr_video_encoder_vulkan_setup_textures: av_hwframe_get_buffer failed: %d\n", res); + return false; + } + + while(self->params.egl->glGetError()) {} +#if 0 + AVVkFrame *target_surface_id = (AVVkFrame*)frame->data[0]; + AVVulkanDeviceContext* vv = video_codec_context_get_vulkan_data(video_codec_context); + const size_t luma_size = frame->width * frame->height; + if(vv) { + PFN_vkGetImageMemoryRequirements vkGetImageMemoryRequirements = (PFN_vkGetImageMemoryRequirements)vv->get_proc_addr(vv->inst, "vkGetImageMemoryRequirements"); + PFN_vkAllocateMemory vkAllocateMemory = (PFN_vkAllocateMemory)vv->get_proc_addr(vv->inst, "vkAllocateMemory"); + PFN_vkGetPhysicalDeviceMemoryProperties vkGetPhysicalDeviceMemoryProperties = (PFN_vkGetPhysicalDeviceMemoryProperties)vv->get_proc_addr(vv->inst, "vkGetPhysicalDeviceMemoryProperties"); + PFN_vkGetMemoryFdKHR vkGetMemoryFdKHR = (PFN_vkGetMemoryFdKHR)vv->get_proc_addr(vv->inst, "vkGetMemoryFdKHR"); + + VkMemoryRequirements mem_reqs = {0}; + vkGetImageMemoryRequirements(vv->act_dev, target_surface_id->img[0], &mem_reqs); + + fprintf(stderr, "size: %lu, alignment: %lu, memory bits: 0x%08x\n", mem_reqs.size, mem_reqs.alignment, mem_reqs.memoryTypeBits); + VkDeviceMemory mem; + { + VkExportMemoryAllocateInfo exp_mem_info; + VkMemoryAllocateInfo mem_alloc_info; + VkMemoryDedicatedAllocateInfoKHR ded_info; + + memset(&exp_mem_info, 0, sizeof(exp_mem_info)); + exp_mem_info.sType = VK_STRUCTURE_TYPE_EXPORT_MEMORY_ALLOCATE_INFO; + exp_mem_info.handleTypes = VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT; + + memset(&ded_info, 0, sizeof(ded_info)); + ded_info.sType = VK_STRUCTURE_TYPE_MEMORY_DEDICATED_ALLOCATE_INFO; + ded_info.image = target_surface_id->img[0]; + + exp_mem_info.pNext = &ded_info; + + memset(&mem_alloc_info, 0, sizeof(mem_alloc_info)); + mem_alloc_info.sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO; + mem_alloc_info.pNext = &exp_mem_info; + mem_alloc_info.allocationSize = target_surface_id->size[0]; + mem_alloc_info.memoryTypeIndex = get_memory_type_idx(vv->phys_dev, &mem_reqs, VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT, vkGetPhysicalDeviceMemoryProperties); + + if (mem_alloc_info.memoryTypeIndex == UINT32_MAX) { + fprintf(stderr, "No suitable memory type index found.\n"); + return VK_NULL_HANDLE; + } + + if (vkAllocateMemory(vv->act_dev, &mem_alloc_info, 0, &mem) != + VK_SUCCESS) + return VK_NULL_HANDLE; + + fprintf(stderr, "memory: %p\n", (void*)mem); + + } + + fprintf(stderr, "target surface id: %p, %zu, %zu\n", (void*)target_surface_id->mem[0], target_surface_id->offset[0], target_surface_id->offset[1]); + fprintf(stderr, "vkGetMemoryFdKHR: %p\n", (void*)vkGetMemoryFdKHR); + + int fd = 0; + VkMemoryGetFdInfoKHR fd_info; + memset(&fd_info, 0, sizeof(fd_info)); + fd_info.sType = VK_STRUCTURE_TYPE_MEMORY_GET_FD_INFO_KHR; + fd_info.memory = target_surface_id->mem[0]; + fd_info.handleType = VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT; + if(vkGetMemoryFdKHR(vv->act_dev, &fd_info, &fd) != VK_SUCCESS) { + fprintf(stderr, "failed!\n"); + } else { + fprintf(stderr, "fd: %d\n", fd); + } + + fprintf(stderr, "glImportMemoryFdEXT: %p, size: %zu\n", (void*)self->params.egl->glImportMemoryFdEXT, target_surface_id->size[0]); + const int tiling = target_surface_id->tiling == VK_IMAGE_TILING_LINEAR ? GL_LINEAR_TILING_EXT : GL_OPTIMAL_TILING_EXT; + + if(tiling != GL_OPTIMAL_TILING_EXT) { + fprintf(stderr, "tiling %d is not supported, only GL_OPTIMAL_TILING_EXT (%d) is supported\n", tiling, GL_OPTIMAL_TILING_EXT); + } + + + unsigned int gl_memory_obj = 0; + self->params.egl->glCreateMemoryObjectsEXT(1, &gl_memory_obj); + + //const int dedicated = GL_TRUE; + //self->params.egl->glMemoryObjectParameterivEXT(gl_memory_obj, GL_DEDICATED_MEMORY_OBJECT_EXT, &dedicated); + + self->params.egl->glImportMemoryFdEXT(gl_memory_obj, target_surface_id->size[0], GL_HANDLE_TYPE_OPAQUE_FD_EXT, fd); + if(!self->params.egl->glIsMemoryObjectEXT(gl_memory_obj)) + fprintf(stderr, "failed to create object!\n"); + + fprintf(stderr, "gl memory obj: %u, error: %d\n", gl_memory_obj, self->params.egl->glGetError()); + + // fprintf(stderr, "0 gl error: %d\n", self->params.egl->glGetError()); + // unsigned int vertex_buffer = 0; + // self->params.egl->glGenBuffers(1, &vertex_buffer); + // self->params.egl->glBindBuffer(GL_ARRAY_BUFFER, vertex_buffer); + // self->params.egl->glBufferStorageMemEXT(GL_ARRAY_BUFFER, target_surface_id->size[0], gl_memory_obj, target_surface_id->offset[0]); + // fprintf(stderr, "1 gl error: %d\n", self->params.egl->glGetError()); + + // fprintf(stderr, "0 gl error: %d\n", self->params.egl->glGetError()); + // unsigned int buffer = 0; + // self->params.egl->glCreateBuffers(1, &buffer); + // self->params.egl->glNamedBufferStorageMemEXT(buffer, target_surface_id->size[0], gl_memory_obj, target_surface_id->offset[0]); + // fprintf(stderr, "1 gl error: %d\n", self->params.egl->glGetError()); + + self->params.egl->glGenTextures(1, &self->target_textures[0]); + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->target_textures[0]); + + fprintf(stderr, "1 gl error: %d\n", self->params.egl->glGetError()); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_TILING_EXT, tiling); + + fprintf(stderr, "tiling: %d\n", tiling); + + fprintf(stderr, "2 gl error: %d\n", self->params.egl->glGetError()); + self->params.egl->glTexStorageMem2DEXT(GL_TEXTURE_2D, 1, GL_R8, frame->width, frame->height, gl_memory_obj, target_surface_id->offset[0]); + + fprintf(stderr, "3 gl error: %d\n", self->params.egl->glGetError()); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + + self->params.egl->glGenTextures(1, &self->target_textures[1]); + self->params.egl->glBindTexture(GL_TEXTURE_2D, self->target_textures[1]); + + fprintf(stderr, "1 gl error: %d\n", self->params.egl->glGetError()); + self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_TILING_EXT, tiling); + + fprintf(stderr, "tiling: %d\n", tiling); + + fprintf(stderr, "2 gl error: %d\n", self->params.egl->glGetError()); + self->params.egl->glTexStorageMem2DEXT(GL_TEXTURE_2D, 1, GL_RG8, frame->width/2, frame->height/2, gl_memory_obj, target_surface_id->offset[0] + luma_size); + + fprintf(stderr, "3 gl error: %d\n", self->params.egl->glGetError()); + self->params.egl->glBindTexture(GL_TEXTURE_2D, 0); + } +#endif + return true; +} + +static void gsr_video_encoder_vulkan_stop(gsr_video_encoder_vulkan *self, AVCodecContext *video_codec_context); + +static bool gsr_video_encoder_vulkan_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) { + gsr_video_encoder_vulkan *self = encoder->priv; + + video_codec_context->width = FFALIGN(video_codec_context->width, 2); + video_codec_context->height = FFALIGN(video_codec_context->height, 2); + + if(video_codec_context->width < 128) + video_codec_context->width = 128; + + if(video_codec_context->height < 128) + video_codec_context->height = 128; + + frame->width = video_codec_context->width; + frame->height = video_codec_context->height; + + if(!gsr_video_encoder_vulkan_setup_context(self, video_codec_context)) { + gsr_video_encoder_vulkan_stop(self, video_codec_context); + return false; + } + + if(!gsr_video_encoder_vulkan_setup_textures(self, video_codec_context, frame)) { + gsr_video_encoder_vulkan_stop(self, video_codec_context); + return false; + } + + return true; +} + +void gsr_video_encoder_vulkan_stop(gsr_video_encoder_vulkan *self, AVCodecContext *video_codec_context) { + self->params.egl->glDeleteTextures(2, self->target_textures); + self->target_textures[0] = 0; + self->target_textures[1] = 0; + + if(video_codec_context->hw_frames_ctx) + av_buffer_unref(&video_codec_context->hw_frames_ctx); + if(self->device_ctx) + av_buffer_unref(&self->device_ctx); +} + +static void gsr_video_encoder_vulkan_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) { + gsr_video_encoder_vulkan *self = encoder->priv; + textures[0] = self->target_textures[0]; + textures[1] = self->target_textures[1]; + *num_textures = 2; + *destination_color = self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? GSR_DESTINATION_COLOR_P010 : GSR_DESTINATION_COLOR_NV12; +} + +static void gsr_video_encoder_vulkan_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) { + gsr_video_encoder_vulkan_stop(encoder->priv, video_codec_context); + free(encoder->priv); + free(encoder); +} + +gsr_video_encoder* gsr_video_encoder_vulkan_create(const gsr_video_encoder_vulkan_params *params) { + gsr_video_encoder *encoder = calloc(1, sizeof(gsr_video_encoder)); + if(!encoder) + return NULL; + + gsr_video_encoder_vulkan *encoder_vulkan = calloc(1, sizeof(gsr_video_encoder_vulkan)); + if(!encoder_vulkan) { + free(encoder); + return NULL; + } + + encoder_vulkan->params = *params; + + *encoder = (gsr_video_encoder) { + .start = gsr_video_encoder_vulkan_start, + .copy_textures_to_frame = NULL, + .get_textures = gsr_video_encoder_vulkan_get_textures, + .destroy = gsr_video_encoder_vulkan_destroy, + .priv = encoder_vulkan + }; + + return encoder; +} diff --git a/src/image_writer.c b/src/image_writer.c new file mode 100644 index 0000000..e153a34 --- /dev/null +++ b/src/image_writer.c @@ -0,0 +1,96 @@ +#include "../include/image_writer.h" +#include "../include/egl.h" +#include "../include/utils.h" + +#define STB_IMAGE_WRITE_IMPLEMENTATION +#include "../external/stb_image_write.h" + +#include <stdlib.h> +#include <stdint.h> +#include <stdio.h> +#include <assert.h> + +/* TODO: Support hdr/10-bit */ +bool gsr_image_writer_init_opengl(gsr_image_writer *self, gsr_egl *egl, int width, int height) { + memset(self, 0, sizeof(*self)); + self->source = GSR_IMAGE_WRITER_SOURCE_OPENGL; + self->egl = egl; + self->width = width; + self->height = height; + self->texture = gl_create_texture(self->egl, self->width, self->height, GL_RGBA8, GL_RGBA, GL_NEAREST); /* TODO: use GL_RGB16 instead of GL_RGB8 for hdr/10-bit */ + if(self->texture == 0) { + fprintf(stderr, "gsr error: gsr_image_writer_init: failed to create texture\n"); + return false; + } + return true; +} + +bool gsr_image_writer_init_memory(gsr_image_writer *self, const void *memory, int width, int height) { + memset(self, 0, sizeof(*self)); + self->source = GSR_IMAGE_WRITER_SOURCE_OPENGL; + self->width = width; + self->height = height; + self->memory = memory; + return true; +} + +void gsr_image_writer_deinit(gsr_image_writer *self) { + if(self->texture) { + self->egl->glDeleteTextures(1, &self->texture); + self->texture = 0; + } +} + +static bool gsr_image_writer_write_memory_to_file(gsr_image_writer *self, const char *filepath, gsr_image_format image_format, int quality, const void *data) { + if(quality < 1) + quality = 1; + else if(quality > 100) + quality = 100; + + bool success = false; + switch(image_format) { + case GSR_IMAGE_FORMAT_JPEG: + success = stbi_write_jpg(filepath, self->width, self->height, 4, data, quality); + break; + case GSR_IMAGE_FORMAT_PNG: + success = stbi_write_png(filepath, self->width, self->height, 4, data, 0); + break; + } + + if(!success) + fprintf(stderr, "gsr error: gsr_image_writer_write_to_file: failed to write image data to output file %s\n", filepath); + + return success; +} + +static bool gsr_image_writer_write_opengl_texture_to_file(gsr_image_writer *self, const char *filepath, gsr_image_format image_format, int quality) { + assert(self->source == GSR_IMAGE_WRITER_SOURCE_OPENGL); + uint8_t *frame_data = malloc(self->width * self->height * 4); + if(!frame_data) { + fprintf(stderr, "gsr error: gsr_image_writer_write_to_file: failed to allocate memory for image frame\n"); + return false; + } + + // TODO: hdr support + self->egl->glBindTexture(GL_TEXTURE_2D, self->texture); + // We could use glGetTexSubImage, but it's only available starting from opengl 4.5 + self->egl->glGetTexImage(GL_TEXTURE_2D, 0, GL_RGBA, GL_UNSIGNED_BYTE, frame_data); + self->egl->glBindTexture(GL_TEXTURE_2D, 0); + + self->egl->glFlush(); + self->egl->glFinish(); + + const bool success = gsr_image_writer_write_memory_to_file(self, filepath, image_format, quality, frame_data); + free(frame_data); + return success; +} + +bool gsr_image_writer_write_to_file(gsr_image_writer *self, const char *filepath, gsr_image_format image_format, int quality) { + switch(self->source) { + case GSR_IMAGE_WRITER_SOURCE_OPENGL: + return gsr_image_writer_write_opengl_texture_to_file(self, filepath, image_format, quality); + case GSR_IMAGE_WRITER_SOURCE_MEMORY: + return gsr_image_writer_write_memory_to_file(self, filepath, image_format, quality, self->memory); + } + return false; +} diff --git a/src/main.cpp b/src/main.cpp index 71452e6..d716fcd 100644 --- a/src/main.cpp +++ b/src/main.cpp @@ -1,25 +1,46 @@ extern "C" { #include "../include/capture/nvfbc.h" -#include "../include/capture/xcomposite_cuda.h" -#include "../include/capture/xcomposite_vaapi.h" -#include "../include/capture/kms_vaapi.h" -#include "../include/capture/kms_cuda.h" +#include "../include/capture/xcomposite.h" +#include "../include/capture/ximage.h" +#include "../include/capture/kms.h" +#ifdef GSR_PORTAL +#include "../include/capture/portal.h" +#include "../dbus/client/dbus_client.h" +#endif +#ifdef GSR_APP_AUDIO +#include "../include/pipewire_audio.h" +#endif +#include "../include/encoder/encoder.h" +#include "../include/encoder/video/nvenc.h" +#include "../include/encoder/video/vaapi.h" +#include "../include/encoder/video/vulkan.h" +#include "../include/encoder/video/software.h" +#include "../include/codec_query/nvenc.h" +#include "../include/codec_query/vaapi.h" +#include "../include/codec_query/vulkan.h" +#include "../include/window/x11.h" +#include "../include/window/wayland.h" #include "../include/egl.h" #include "../include/utils.h" +#include "../include/damage.h" +#include "../include/color_conversion.h" +#include "../include/image_writer.h" +#include "../include/args_parser.h" } #include <assert.h> #include <stdio.h> #include <stdlib.h> #include <string> -#include <vector> -#include <unordered_map> #include <thread> #include <mutex> -#include <map> #include <signal.h> #include <sys/stat.h> #include <unistd.h> +#include <sys/wait.h> +#include <inttypes.h> +#include <libgen.h> +#include <malloc.h> #include "../include/sound.hpp" @@ -31,35 +52,82 @@ extern "C" { #include <libswresample/swresample.h> #include <libavutil/avutil.h> #include <libavutil/time.h> +#include <libavutil/mastering_display_metadata.h> #include <libavfilter/avfilter.h> #include <libavfilter/buffersink.h> #include <libavfilter/buffersrc.h> } -#include <deque> #include <future> +#ifndef GSR_VERSION +#define GSR_VERSION "unknown" +#endif + // TODO: If options are not supported then they are returned (allocated) in the options. This should be free'd. // TODO: Remove LIBAVUTIL_VERSION_MAJOR checks in the future when ubuntu, pop os LTS etc update ffmpeg to >= 5.0 +static const int AUDIO_SAMPLE_RATE = 48000; + static const int VIDEO_STREAM_INDEX = 0; static thread_local char av_error_buffer[AV_ERROR_MAX_STRING_SIZE]; +typedef struct { + const gsr_window *window; +} MonitorOutputCallbackUserdata; + static void monitor_output_callback_print(const gsr_monitor *monitor, void *userdata) { - (void)userdata; - fprintf(stderr, " \"%.*s\" (%dx%d+%d+%d)\n", monitor->name_len, monitor->name, monitor->size.x, monitor->size.y, monitor->pos.x, monitor->pos.y); + const MonitorOutputCallbackUserdata *options = (MonitorOutputCallbackUserdata*)userdata; + vec2i monitor_position = monitor->pos; + vec2i monitor_size = monitor->size; + if(gsr_window_get_display_server(options->window) == GSR_DISPLAY_SERVER_WAYLAND) { + gsr_monitor_rotation monitor_rotation = GSR_MONITOR_ROT_0; + drm_monitor_get_display_server_data(options->window, monitor, &monitor_rotation, &monitor_position); + if(monitor_rotation == GSR_MONITOR_ROT_90 || monitor_rotation == GSR_MONITOR_ROT_270) + std::swap(monitor_size.x, monitor_size.y); + } + fprintf(stderr, " \"%.*s\" (%dx%d+%d+%d)\n", monitor->name_len, monitor->name, monitor_size.x, monitor_size.y, monitor_position.x, monitor_position.y); } typedef struct { - const char *output_name; + char *output_name; } FirstOutputCallback; -static void get_first_output(const gsr_monitor *monitor, void *userdata) { - FirstOutputCallback *first_output = (FirstOutputCallback*)userdata; - if(!first_output->output_name) - first_output->output_name = strndup(monitor->name, monitor->name_len + 1); +static void get_first_output_callback(const gsr_monitor *monitor, void *userdata) { + FirstOutputCallback *data = (FirstOutputCallback*)userdata; + if(!data->output_name) + data->output_name = strdup(monitor->name); +} + +typedef struct { + gsr_window *window; + vec2i position; + char *output_name; + vec2i monitor_pos; + vec2i monitor_size; +} MonitorByPositionCallback; + +static void get_monitor_by_position_callback(const gsr_monitor *monitor, void *userdata) { + MonitorByPositionCallback *data = (MonitorByPositionCallback*)userdata; + + vec2i monitor_position = monitor->pos; + vec2i monitor_size = monitor->size; + if(gsr_window_get_display_server(data->window) == GSR_DISPLAY_SERVER_WAYLAND) { + gsr_monitor_rotation monitor_rotation = GSR_MONITOR_ROT_0; + drm_monitor_get_display_server_data(data->window, monitor, &monitor_rotation, &monitor_position); + if(monitor_rotation == GSR_MONITOR_ROT_90 || monitor_rotation == GSR_MONITOR_ROT_270) + std::swap(monitor_size.x, monitor_size.y); + } + + if(!data->output_name && data->position.x >= monitor_position.x && data->position.x <= monitor_position.x + monitor->size.x + && data->position.y >= monitor_position.y && data->position.y <= monitor_position.y + monitor->size.y) + { + data->output_name = strdup(monitor->name); + data->monitor_pos = monitor_position; + data->monitor_size = monitor->size; + } } static char* av_error_to_string(int err) { @@ -68,35 +136,6 @@ static char* av_error_to_string(int err) { return av_error_buffer; } -enum class VideoQuality { - MEDIUM, - HIGH, - VERY_HIGH, - ULTRA -}; - -enum class VideoCodec { - H264, - H265, - AV1 -}; - -enum class AudioCodec { - AAC, - OPUS, - FLAC -}; - -enum class PixelFormat { - YUV420, - YUV444 -}; - -enum class FramerateMode { - CONSTANT, - VARIABLE -}; - static int x11_error_handler(Display*, XErrorEvent*) { return 0; } @@ -105,124 +144,61 @@ static int x11_io_error_handler(Display*) { return 0; } -struct PacketData { - PacketData() {} - PacketData(const PacketData&) = delete; - PacketData& operator=(const PacketData&) = delete; - - ~PacketData() { - av_free(data.data); - } - - AVPacket data; -}; - -// |stream| is only required for non-replay mode -static void receive_frames(AVCodecContext *av_codec_context, int stream_index, AVStream *stream, int64_t pts, - AVFormatContext *av_format_context, - double replay_start_time, - std::deque<std::shared_ptr<PacketData>> &frame_data_queue, - int replay_buffer_size_secs, - bool &frames_erased, - std::mutex &write_output_mutex) { - for (;;) { - // TODO: Use av_packet_alloc instead because sizeof(av_packet) might not be future proof(?) - AVPacket *av_packet = av_packet_alloc(); - if(!av_packet) - break; - - av_packet->data = NULL; - av_packet->size = 0; - int res = avcodec_receive_packet(av_codec_context, av_packet); - if (res == 0) { // we have a packet, send the packet to the muxer - av_packet->stream_index = stream_index; - av_packet->pts = pts; - av_packet->dts = pts; - - std::lock_guard<std::mutex> lock(write_output_mutex); - if(replay_buffer_size_secs != -1) { - // TODO: Preallocate all frames data and use those instead. - // Why are we doing this you ask? there is a new ffmpeg bug that causes cpu usage to increase over time when you have - // packets that are not being free'd until later. So we copy the packet data, free the packet and then reconstruct - // the packet later on when we need it, to keep packets alive only for a short period. - auto new_packet = std::make_shared<PacketData>(); - new_packet->data = *av_packet; - new_packet->data.data = (uint8_t*)av_malloc(av_packet->size); - memcpy(new_packet->data.data, av_packet->data, av_packet->size); - - double time_now = clock_get_monotonic_seconds(); - double replay_time_elapsed = time_now - replay_start_time; - - frame_data_queue.push_back(std::move(new_packet)); - if(replay_time_elapsed >= replay_buffer_size_secs) { - frame_data_queue.pop_front(); - frames_erased = true; - } - } else { - av_packet_rescale_ts(av_packet, av_codec_context->time_base, stream->time_base); - av_packet->stream_index = stream->index; - // TODO: Is av_interleaved_write_frame needed? - int ret = av_write_frame(av_format_context, av_packet); - if(ret < 0) { - fprintf(stderr, "Error: Failed to write frame index %d to muxer, reason: %s (%d)\n", av_packet->stream_index, av_error_to_string(ret), ret); - } - } - av_packet_free(&av_packet); - } else if (res == AVERROR(EAGAIN)) { // we have no packet - // fprintf(stderr, "No packet!\n"); - av_packet_free(&av_packet); - break; - } else if (res == AVERROR_EOF) { // this is the end of the stream - av_packet_free(&av_packet); - fprintf(stderr, "End of stream!\n"); - break; - } else { - av_packet_free(&av_packet); - fprintf(stderr, "Unexpected error: %d\n", res); - break; - } - } -} - -static const char* audio_codec_get_name(AudioCodec audio_codec) { +static AVCodecID audio_codec_get_id(gsr_audio_codec audio_codec) { switch(audio_codec) { - case AudioCodec::AAC: return "aac"; - case AudioCodec::OPUS: return "opus"; - case AudioCodec::FLAC: return "flac"; - } - assert(false); - return ""; -} - -static AVCodecID audio_codec_get_id(AudioCodec audio_codec) { - switch(audio_codec) { - case AudioCodec::AAC: return AV_CODEC_ID_AAC; - case AudioCodec::OPUS: return AV_CODEC_ID_OPUS; - case AudioCodec::FLAC: return AV_CODEC_ID_FLAC; + case GSR_AUDIO_CODEC_AAC: return AV_CODEC_ID_AAC; + case GSR_AUDIO_CODEC_OPUS: return AV_CODEC_ID_OPUS; + case GSR_AUDIO_CODEC_FLAC: return AV_CODEC_ID_FLAC; } assert(false); return AV_CODEC_ID_AAC; } -static AVSampleFormat audio_codec_get_sample_format(AudioCodec audio_codec, const AVCodec *codec) { +static AVSampleFormat audio_codec_get_sample_format(AVCodecContext *audio_codec_context, gsr_audio_codec audio_codec, const AVCodec *codec, bool mix_audio) { + (void)audio_codec_context; switch(audio_codec) { - case AudioCodec::AAC: { + case GSR_AUDIO_CODEC_AAC: { return AV_SAMPLE_FMT_FLTP; } - case AudioCodec::OPUS: { + case GSR_AUDIO_CODEC_OPUS: { bool supports_s16 = false; bool supports_flt = false; - for(size_t i = 0; codec->sample_fmts && codec->sample_fmts[i] != -1; ++i) { + #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(61, 15, 0) + for(size_t i = 0; codec->sample_fmts && codec->sample_fmts[i] != AV_SAMPLE_FMT_NONE; ++i) { if(codec->sample_fmts[i] == AV_SAMPLE_FMT_S16) { supports_s16 = true; } else if(codec->sample_fmts[i] == AV_SAMPLE_FMT_FLT) { supports_flt = true; } } + #else + const enum AVSampleFormat *sample_fmts = NULL; + if(avcodec_get_supported_config(audio_codec_context, codec, AV_CODEC_CONFIG_SAMPLE_FORMAT, 0, (const void**)&sample_fmts, NULL) >= 0) { + if(sample_fmts) { + for(size_t i = 0; sample_fmts[i] != AV_SAMPLE_FMT_NONE; ++i) { + if(sample_fmts[i] == AV_SAMPLE_FMT_S16) { + supports_s16 = true; + } else if(sample_fmts[i] == AV_SAMPLE_FMT_FLT) { + supports_flt = true; + } + } + } else { + // What a dumb API. It returns NULL if all formats are supported + supports_s16 = true; + supports_flt = true; + } + } + #endif + + // Amix only works with float audio + if(mix_audio) + supports_s16 = false; if(!supports_s16 && !supports_flt) { - fprintf(stderr, "Warning: opus audio codec is chosen but your ffmpeg version does not support s16/flt sample format and performance might be slightly worse. You can either rebuild ffmpeg with libopus instead of the built-in opus, use the flatpak version of gpu screen recorder or record with flac audio codec instead (-ac flac). Falling back to fltp audio sample format instead.\n"); + fprintf(stderr, "Warning: opus audio codec is chosen but your ffmpeg version does not support s16/flt sample format and performance might be slightly worse.\n"); + fprintf(stderr, " You can either rebuild ffmpeg with libopus instead of the built-in opus, use the flatpak version of gpu screen recorder or record with aac audio codec instead (-ac aac).\n"); + fprintf(stderr, " Falling back to fltp audio sample format instead.\n"); } if(supports_s16) @@ -232,7 +208,7 @@ static AVSampleFormat audio_codec_get_sample_format(AudioCodec audio_codec, cons else return AV_SAMPLE_FMT_FLTP; } - case AudioCodec::FLAC: { + case GSR_AUDIO_CODEC_FLAC: { return AV_SAMPLE_FMT_S32; } } @@ -240,14 +216,14 @@ static AVSampleFormat audio_codec_get_sample_format(AudioCodec audio_codec, cons return AV_SAMPLE_FMT_FLTP; } -static int64_t audio_codec_get_get_bitrate(AudioCodec audio_codec) { +static int64_t audio_codec_get_get_bitrate(gsr_audio_codec audio_codec) { switch(audio_codec) { - case AudioCodec::AAC: return 128000; - case AudioCodec::OPUS: return 96000; - case AudioCodec::FLAC: return 96000; + case GSR_AUDIO_CODEC_AAC: return 160000; + case GSR_AUDIO_CODEC_OPUS: return 128000; + case GSR_AUDIO_CODEC_FLAC: return 128000; } assert(false); - return 96000; + return 128000; } static AudioFormat audio_codec_context_get_audio_format(const AVCodecContext *audio_codec_context) { @@ -270,7 +246,8 @@ static AVSampleFormat audio_format_to_sample_format(const AudioFormat audio_form return AV_SAMPLE_FMT_S16; } -static AVCodecContext* create_audio_codec_context(int fps, AudioCodec audio_codec) { +static AVCodecContext* create_audio_codec_context(int fps, gsr_audio_codec audio_codec, bool mix_audio, int64_t audio_bitrate) { + (void)fps; const AVCodec *codec = avcodec_find_encoder(audio_codec_get_id(audio_codec)); if (!codec) { fprintf(stderr, "Error: Could not find %s audio encoder\n", audio_codec_get_name(audio_codec)); @@ -281,10 +258,10 @@ static AVCodecContext* create_audio_codec_context(int fps, AudioCodec audio_code assert(codec->type == AVMEDIA_TYPE_AUDIO); codec_context->codec_id = codec->id; - codec_context->sample_fmt = audio_codec_get_sample_format(audio_codec, codec); - codec_context->bit_rate = audio_codec_get_get_bitrate(audio_codec); - codec_context->sample_rate = 48000; - if(audio_codec == AudioCodec::AAC) + codec_context->sample_fmt = audio_codec_get_sample_format(codec_context, audio_codec, codec, mix_audio); + codec_context->bit_rate = audio_bitrate == 0 ? audio_codec_get_get_bitrate(audio_codec) : audio_bitrate; + codec_context->sample_rate = AUDIO_SAMPLE_RATE; + if(audio_codec == GSR_AUDIO_CODEC_AAC) codec_context->profile = FF_PROFILE_AAC_LOW; #if LIBAVCODEC_VERSION_MAJOR < 60 codec_context->channel_layout = AV_CH_LAYOUT_STEREO; @@ -294,18 +271,68 @@ static AVCodecContext* create_audio_codec_context(int fps, AudioCodec audio_code #endif codec_context->time_base.num = 1; - codec_context->time_base.den = AV_TIME_BASE; - codec_context->framerate.num = fps; - codec_context->framerate.den = 1; + codec_context->time_base.den = codec_context->sample_rate; + codec_context->thread_count = 1; codec_context->flags |= AV_CODEC_FLAG_GLOBAL_HEADER; return codec_context; } -static AVCodecContext *create_video_codec_context(AVPixelFormat pix_fmt, - VideoQuality video_quality, - int fps, const AVCodec *codec, bool is_livestream, gsr_gpu_vendor vendor, FramerateMode framerate_mode) { +static int vbr_get_quality_parameter(AVCodecContext *codec_context, gsr_video_quality video_quality, bool hdr) { + // 8 bit / 10 bit = 80% + const float qp_multiply = hdr ? 8.0f/10.0f : 1.0f; + if(codec_context->codec_id == AV_CODEC_ID_AV1) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + return 160 * qp_multiply; + case GSR_VIDEO_QUALITY_HIGH: + return 130 * qp_multiply; + case GSR_VIDEO_QUALITY_VERY_HIGH: + return 110 * qp_multiply; + case GSR_VIDEO_QUALITY_ULTRA: + return 90 * qp_multiply; + } + } else if(codec_context->codec_id == AV_CODEC_ID_H264) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + return 35 * qp_multiply; + case GSR_VIDEO_QUALITY_HIGH: + return 30 * qp_multiply; + case GSR_VIDEO_QUALITY_VERY_HIGH: + return 25 * qp_multiply; + case GSR_VIDEO_QUALITY_ULTRA: + return 22 * qp_multiply; + } + } else if(codec_context->codec_id == AV_CODEC_ID_HEVC) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + return 35 * qp_multiply; + case GSR_VIDEO_QUALITY_HIGH: + return 30 * qp_multiply; + case GSR_VIDEO_QUALITY_VERY_HIGH: + return 25 * qp_multiply; + case GSR_VIDEO_QUALITY_ULTRA: + return 22 * qp_multiply; + } + } else if(codec_context->codec_id == AV_CODEC_ID_VP8 || codec_context->codec_id == AV_CODEC_ID_VP9) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + return 35 * qp_multiply; + case GSR_VIDEO_QUALITY_HIGH: + return 30 * qp_multiply; + case GSR_VIDEO_QUALITY_VERY_HIGH: + return 25 * qp_multiply; + case GSR_VIDEO_QUALITY_ULTRA: + return 22 * qp_multiply; + } + } + assert(false); + return 22 * qp_multiply; +} +static AVCodecContext *create_video_codec_context(AVPixelFormat pix_fmt, const AVCodec *codec, const gsr_egl &egl, const args_parser &arg_parser) { + const bool use_software_video_encoder = arg_parser.video_encoder == GSR_VIDEO_ENCODER_HW_CPU; + const bool hdr = video_codec_is_hdr(arg_parser.video_codec); AVCodecContext *codec_context = avcodec_alloc_context3(codec); //double fps_ratio = (double)fps / 30.0; @@ -317,238 +344,176 @@ static AVCodecContext *create_video_codec_context(AVPixelFormat pix_fmt, // timebase should be 1/framerate and timestamp increments should be // identical to 1 codec_context->time_base.num = 1; - codec_context->time_base.den = framerate_mode == FramerateMode::CONSTANT ? fps : AV_TIME_BASE; - codec_context->framerate.num = fps; + codec_context->time_base.den = arg_parser.framerate_mode == GSR_FRAMERATE_MODE_CONSTANT ? arg_parser.fps : AV_TIME_BASE; + codec_context->framerate.num = arg_parser.fps; codec_context->framerate.den = 1; codec_context->sample_aspect_ratio.num = 0; codec_context->sample_aspect_ratio.den = 0; - // High values reeduce file size but increases time it takes to seek - if(is_livestream) { + if(arg_parser.low_latency_recording) { codec_context->flags |= (AV_CODEC_FLAG_CLOSED_GOP | AV_CODEC_FLAG_LOW_DELAY); codec_context->flags2 |= AV_CODEC_FLAG2_FAST; //codec_context->gop_size = std::numeric_limits<int>::max(); //codec_context->keyint_min = std::numeric_limits<int>::max(); - codec_context->gop_size = fps * 2; + codec_context->gop_size = arg_parser.fps * arg_parser.keyint; } else { - codec_context->gop_size = fps * 2; + // High values reduce file size but increases time it takes to seek + codec_context->gop_size = arg_parser.fps * arg_parser.keyint; } codec_context->max_b_frames = 0; codec_context->pix_fmt = pix_fmt; - //codec_context->color_range = AVCOL_RANGE_JPEG; // TODO: Amd/nvidia? - //codec_context->color_primaries = AVCOL_PRI_BT709; - //codec_context->color_trc = AVCOL_TRC_BT709; - //codec_context->colorspace = AVCOL_SPC_BT709; + codec_context->color_range = arg_parser.color_range == GSR_COLOR_RANGE_LIMITED ? AVCOL_RANGE_MPEG : AVCOL_RANGE_JPEG; + if(hdr) { + codec_context->color_primaries = AVCOL_PRI_BT2020; + codec_context->color_trc = AVCOL_TRC_SMPTE2084; + codec_context->colorspace = AVCOL_SPC_BT2020_NCL; + } else { + codec_context->color_primaries = AVCOL_PRI_BT709; + codec_context->color_trc = AVCOL_TRC_BT709; + codec_context->colorspace = AVCOL_SPC_BT709; + } //codec_context->chroma_sample_location = AVCHROMA_LOC_CENTER; if(codec->id == AV_CODEC_ID_HEVC) - codec_context->codec_tag = MKTAG('h', 'v', 'c', '1'); - switch(video_quality) { - case VideoQuality::MEDIUM: - //codec_context->qmin = 35; - //codec_context->qmax = 35; - codec_context->bit_rate = 100000;//4500000 + (codec_context->width * codec_context->height)*0.75; - break; - case VideoQuality::HIGH: - //codec_context->qmin = 34; - //codec_context->qmax = 34; - codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75; - break; - case VideoQuality::VERY_HIGH: - //codec_context->qmin = 28; - //codec_context->qmax = 28; - codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75; - break; - case VideoQuality::ULTRA: - //codec_context->qmin = 22; - //codec_context->qmax = 22; - codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75; - break; + codec_context->codec_tag = MKTAG('h', 'v', 'c', '1'); // QuickTime on MacOS requires this or the video wont be playable + + if(arg_parser.bitrate_mode == GSR_BITRATE_MODE_CBR) { + codec_context->bit_rate = arg_parser.video_bitrate; + codec_context->rc_max_rate = codec_context->bit_rate; + //codec_context->rc_min_rate = codec_context->bit_rate; + codec_context->rc_buffer_size = codec_context->bit_rate;//codec_context->bit_rate / 10; + codec_context->rc_initial_buffer_occupancy = 0;//codec_context->bit_rate;//codec_context->bit_rate * 1000; + } else if(arg_parser.bitrate_mode == GSR_BITRATE_MODE_VBR) { + const int quality = vbr_get_quality_parameter(codec_context, arg_parser.video_quality, hdr); + switch(arg_parser.video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + codec_context->qmin = quality; + codec_context->qmax = quality; + codec_context->bit_rate = 100000;//4500000 + (codec_context->width * codec_context->height)*0.75; + break; + case GSR_VIDEO_QUALITY_HIGH: + codec_context->qmin = quality; + codec_context->qmax = quality; + codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75; + break; + case GSR_VIDEO_QUALITY_VERY_HIGH: + codec_context->qmin = quality; + codec_context->qmax = quality; + codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75; + break; + case GSR_VIDEO_QUALITY_ULTRA: + codec_context->qmin = quality; + codec_context->qmax = quality; + codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75; + break; + } + + codec_context->rc_max_rate = codec_context->bit_rate; + //codec_context->rc_min_rate = codec_context->bit_rate; + codec_context->rc_buffer_size = codec_context->bit_rate;//codec_context->bit_rate / 10; + codec_context->rc_initial_buffer_occupancy = codec_context->bit_rate;//codec_context->bit_rate * 1000; + } else { + //codec_context->rc_buffer_size = 50000 * 1000; } //codec_context->profile = FF_PROFILE_H264_MAIN; if (codec_context->codec_id == AV_CODEC_ID_MPEG1VIDEO) codec_context->mb_decision = 2; - // stream->time_base = codec_context->time_base; - // codec_context->ticks_per_frame = 30; - //av_opt_set(codec_context->priv_data, "tune", "hq", 0); - // TODO: Do this for better file size? also allows setting qmin, qmax per frame? which can then be used to dynamically set bitrate to reduce quality - // if live streaming is slow or if the users harddrive is cant handle writing megabytes of data per second. - #if 0 - char qmin_str[32]; - snprintf(qmin_str, sizeof(qmin_str), "%d", codec_context->qmin); - - char qmax_str[32]; - snprintf(qmax_str, sizeof(qmax_str), "%d", codec_context->qmax); - - av_opt_set(codec_context->priv_data, "cq", qmax_str, 0); - av_opt_set(codec_context->priv_data, "rc", "vbr", 0); - av_opt_set(codec_context->priv_data, "qmin", qmin_str, 0); - av_opt_set(codec_context->priv_data, "qmax", qmax_str, 0); - codec_context->bit_rate = 0; - #endif - - if(vendor != GSR_GPU_VENDOR_NVIDIA) { - switch(video_quality) { - case VideoQuality::MEDIUM: - codec_context->global_quality = 180; - break; - case VideoQuality::HIGH: - codec_context->global_quality = 120; - break; - case VideoQuality::VERY_HIGH: - codec_context->global_quality = 100; - break; - case VideoQuality::ULTRA: - codec_context->global_quality = 70; - break; + if(!use_software_video_encoder && egl.gpu_info.vendor != GSR_GPU_VENDOR_NVIDIA && arg_parser.bitrate_mode != GSR_BITRATE_MODE_CBR) { + // 8 bit / 10 bit = 80%, and increase it even more + const float quality_multiply = hdr ? (8.0f/10.0f * 0.7f) : 1.0f; + if(codec_context->codec_id == AV_CODEC_ID_AV1 || codec_context->codec_id == AV_CODEC_ID_H264 || codec_context->codec_id == AV_CODEC_ID_HEVC) { + switch(arg_parser.video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + codec_context->global_quality = 130 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_HIGH: + codec_context->global_quality = 110 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_VERY_HIGH: + codec_context->global_quality = 95 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_ULTRA: + codec_context->global_quality = 85 * quality_multiply; + break; + } + } else if(codec_context->codec_id == AV_CODEC_ID_VP8) { + switch(arg_parser.video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + codec_context->global_quality = 35 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_HIGH: + codec_context->global_quality = 30 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_VERY_HIGH: + codec_context->global_quality = 25 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_ULTRA: + codec_context->global_quality = 10 * quality_multiply; + break; + } + } else if(codec_context->codec_id == AV_CODEC_ID_VP9) { + switch(arg_parser.video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + codec_context->global_quality = 35 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_HIGH: + codec_context->global_quality = 30 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_VERY_HIGH: + codec_context->global_quality = 25 * quality_multiply; + break; + case GSR_VIDEO_QUALITY_ULTRA: + codec_context->global_quality = 10 * quality_multiply; + break; + } } } av_opt_set_int(codec_context->priv_data, "b_ref_mode", 0, 0); //av_opt_set_int(codec_context->priv_data, "cbr", true, 0); - if(vendor != GSR_GPU_VENDOR_NVIDIA) { + if(egl.gpu_info.vendor != GSR_GPU_VENDOR_NVIDIA) { // TODO: More options, better options //codec_context->bit_rate = codec_context->width * codec_context->height; - av_opt_set(codec_context->priv_data, "rc_mode", "CQP", 0); + switch(arg_parser.bitrate_mode) { + case GSR_BITRATE_MODE_QP: { + if(video_codec_is_vulkan(arg_parser.video_codec)) + av_opt_set(codec_context->priv_data, "rc_mode", "cqp", 0); + else if(egl.gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA) + av_opt_set(codec_context->priv_data, "rc", "constqp", 0); + else + av_opt_set(codec_context->priv_data, "rc_mode", "CQP", 0); + break; + } + case GSR_BITRATE_MODE_VBR: { + if(video_codec_is_vulkan(arg_parser.video_codec)) + av_opt_set(codec_context->priv_data, "rc_mode", "vbr", 0); + else if(egl.gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA) + av_opt_set(codec_context->priv_data, "rc", "vbr", 0); + else + av_opt_set(codec_context->priv_data, "rc_mode", "VBR", 0); + break; + } + case GSR_BITRATE_MODE_CBR: { + if(video_codec_is_vulkan(arg_parser.video_codec)) + av_opt_set(codec_context->priv_data, "rc_mode", "cbr", 0); + else if(egl.gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA) + av_opt_set(codec_context->priv_data, "rc", "cbr", 0); + else + av_opt_set(codec_context->priv_data, "rc_mode", "CBR", 0); + break; + } + } //codec_context->global_quality = 4; //codec_context->compression_level = 2; } - //codec_context->rc_max_rate = codec_context->bit_rate; - //codec_context->rc_min_rate = codec_context->bit_rate; - //codec_context->rc_buffer_size = codec_context->bit_rate / 10; - // TODO: Do this when not using cqp - //codec_context->rc_initial_buffer_occupancy = codec_context->bit_rate * 1000; + //av_opt_set(codec_context->priv_data, "bsf", "hevc_metadata=colour_primaries=9:transfer_characteristics=16:matrix_coefficients=9", 0); codec_context->flags |= AV_CODEC_FLAG_GLOBAL_HEADER; return codec_context; } -static bool vaapi_create_codec_context(AVCodecContext *video_codec_context, const char *card_path) { - char render_path[128]; - if(!gsr_card_path_get_render_path(card_path, render_path)) { - fprintf(stderr, "gsr error: failed to get /dev/dri/renderDXXX file from %s\n", card_path); - return false; - } - - AVBufferRef *device_ctx; - if(av_hwdevice_ctx_create(&device_ctx, AV_HWDEVICE_TYPE_VAAPI, render_path, NULL, 0) < 0) { - fprintf(stderr, "Error: Failed to create hardware device context\n"); - return false; - } - - AVBufferRef *frame_context = av_hwframe_ctx_alloc(device_ctx); - if(!frame_context) { - fprintf(stderr, "Error: Failed to create hwframe context\n"); - av_buffer_unref(&device_ctx); - return false; - } - - AVHWFramesContext *hw_frame_context = - (AVHWFramesContext *)frame_context->data; - hw_frame_context->width = video_codec_context->width; - hw_frame_context->height = video_codec_context->height; - hw_frame_context->sw_format = AV_PIX_FMT_NV12; - hw_frame_context->format = video_codec_context->pix_fmt; - hw_frame_context->device_ref = device_ctx; - hw_frame_context->device_ctx = (AVHWDeviceContext*)device_ctx->data; - - hw_frame_context->initial_pool_size = 1; - - if (av_hwframe_ctx_init(frame_context) < 0) { - fprintf(stderr, "Error: Failed to initialize hardware frame context " - "(note: ffmpeg version needs to be > 4.0)\n"); - av_buffer_unref(&device_ctx); - //av_buffer_unref(&frame_context); - return false; - } - - video_codec_context->hw_device_ctx = av_buffer_ref(device_ctx); - video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context); - return true; -} - -static bool check_if_codec_valid_for_hardware(const AVCodec *codec, gsr_gpu_vendor vendor, const char *card_path) { - // Do not use AV_PIX_FMT_CUDA because we dont want to do full check with hardware context - AVCodecContext *codec_context = create_video_codec_context(vendor == GSR_GPU_VENDOR_NVIDIA ? AV_PIX_FMT_YUV420P : AV_PIX_FMT_VAAPI, VideoQuality::VERY_HIGH, 60, codec, false, vendor, FramerateMode::CONSTANT); - if(!codec_context) - return false; - - codec_context->width = 512; - codec_context->height = 512; - - if(vendor != GSR_GPU_VENDOR_NVIDIA) { - if(!vaapi_create_codec_context(codec_context, card_path)) { - avcodec_free_context(&codec_context); - return false; - } - } - - bool success = false; - success = avcodec_open2(codec_context, codec_context->codec, NULL) == 0; - if(codec_context->hw_device_ctx) - av_buffer_unref(&codec_context->hw_device_ctx); - if(codec_context->hw_frames_ctx) - av_buffer_unref(&codec_context->hw_frames_ctx); - avcodec_free_context(&codec_context); - return success; -} - -static const AVCodec* find_h264_encoder(gsr_gpu_vendor vendor, const char *card_path) { - const AVCodec *codec = avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "h264_nvenc" : "h264_vaapi"); - if(!codec) - codec = avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "nvenc_h264" : "vaapi_h264"); - - if(!codec) - return nullptr; - - static bool checked = false; - static bool checked_success = true; - if(!checked) { - checked = true; - if(!check_if_codec_valid_for_hardware(codec, vendor, card_path)) - checked_success = false; - } - return checked_success ? codec : nullptr; -} - -static const AVCodec* find_h265_encoder(gsr_gpu_vendor vendor, const char *card_path) { - const AVCodec *codec = avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "hevc_nvenc" : "hevc_vaapi"); - if(!codec) - codec = avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "nvenc_hevc" : "vaapi_hevc"); - - if(!codec) - return nullptr; - - static bool checked = false; - static bool checked_success = true; - if(!checked) { - checked = true; - if(!check_if_codec_valid_for_hardware(codec, vendor, card_path)) - checked_success = false; - } - return checked_success ? codec : nullptr; -} - -static const AVCodec* find_av1_encoder(gsr_gpu_vendor vendor, const char *card_path) { - const AVCodec *codec = avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "av1_nvenc" : "av1_vaapi"); - if(!codec) - codec = avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "nvenc_av1" : "vaapi_av1"); - - if(!codec) - return nullptr; - - static bool checked = false; - static bool checked_success = true; - if(!checked) { - checked = true; - if(!check_if_codec_valid_for_hardware(codec, vendor, card_path)) - checked_success = false; - } - return checked_success ? codec : nullptr; -} - static void open_audio(AVCodecContext *audio_codec_context) { AVDictionary *options = nullptr; av_dict_set(&options, "strict", "experimental", 0); @@ -587,165 +552,360 @@ static AVFrame* create_audio_frame(AVCodecContext *audio_codec_context) { return frame; } -static void open_video(AVCodecContext *codec_context, VideoQuality video_quality, bool very_old_gpu, gsr_gpu_vendor vendor, PixelFormat pixel_format) { +static void dict_set_profile(AVCodecContext *codec_context, gsr_gpu_vendor vendor, gsr_color_depth color_depth, gsr_video_codec video_codec, AVDictionary **options) { + #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(61, 17, 100) + if(codec_context->codec_id == AV_CODEC_ID_H264) { + // TODO: Only for vaapi + //if(color_depth == GSR_COLOR_DEPTH_10_BITS) + // av_dict_set(options, "profile", "high10", 0); + //else + av_dict_set(options, "profile", "high", 0); + } else if(codec_context->codec_id == AV_CODEC_ID_AV1) { + if(vendor == GSR_GPU_VENDOR_NVIDIA) { + if(color_depth == GSR_COLOR_DEPTH_10_BITS) + av_dict_set_int(options, "highbitdepth", 1, 0); + } else { + av_dict_set(options, "profile", "main", 0); // TODO: use professional instead? + } + } else if(codec_context->codec_id == AV_CODEC_ID_HEVC) { + if(color_depth == GSR_COLOR_DEPTH_10_BITS) + av_dict_set(options, "profile", "main10", 0); + else + av_dict_set(options, "profile", "main", 0); + } + #else + const bool use_nvidia_values = vendor == GSR_GPU_VENDOR_NVIDIA && !video_codec_is_vulkan(video_codec); + if(codec_context->codec_id == AV_CODEC_ID_H264) { + // TODO: Only for vaapi + //if(color_depth == GSR_COLOR_DEPTH_10_BITS) + // av_dict_set_int(options, "profile", AV_PROFILE_H264_HIGH_10, 0); + //else + av_dict_set_int(options, "profile", use_nvidia_values ? 2 : AV_PROFILE_H264_HIGH, 0); + } else if(codec_context->codec_id == AV_CODEC_ID_AV1) { + if(use_nvidia_values) { + if(color_depth == GSR_COLOR_DEPTH_10_BITS) + av_dict_set_int(options, "highbitdepth", 1, 0); + } else { + av_dict_set_int(options, "profile", AV_PROFILE_AV1_MAIN, 0); // TODO: use professional instead? + } + } else if(codec_context->codec_id == AV_CODEC_ID_HEVC) { + if(color_depth == GSR_COLOR_DEPTH_10_BITS) + av_dict_set_int(options, "profile", use_nvidia_values ? 1 : AV_PROFILE_HEVC_MAIN_10, 0); + else + av_dict_set_int(options, "profile", use_nvidia_values ? 0 : AV_PROFILE_HEVC_MAIN, 0); + } + #endif +} + +static void video_software_set_qp(AVCodecContext *codec_context, gsr_video_quality video_quality, bool hdr, AVDictionary **options) { + // 8 bit / 10 bit = 80% + const float qp_multiply = hdr ? 8.0f/10.0f : 1.0f; + if(codec_context->codec_id == AV_CODEC_ID_AV1) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); + break; + } + } else if(codec_context->codec_id == AV_CODEC_ID_H264) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 34 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); + break; + } + } else { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); + break; + } + } +} + +static void open_video_software(AVCodecContext *codec_context, const args_parser &arg_parser) { + const gsr_color_depth color_depth = video_codec_to_bit_depth(arg_parser.video_codec); + const bool hdr = video_codec_is_hdr(arg_parser.video_codec); AVDictionary *options = nullptr; - if(vendor == GSR_GPU_VENDOR_NVIDIA) { - bool supports_p4 = false; - bool supports_p5 = false; - - const AVOption *opt = nullptr; - while((opt = av_opt_next(codec_context->priv_data, opt))) { - if(opt->type == AV_OPT_TYPE_CONST) { - if(strcmp(opt->name, "p4") == 0) - supports_p4 = true; - else if(strcmp(opt->name, "p5") == 0) - supports_p5 = true; - } + + if(arg_parser.bitrate_mode == GSR_BITRATE_MODE_QP) + video_software_set_qp(codec_context, arg_parser.video_quality, hdr, &options); + + av_dict_set(&options, "preset", "veryfast", 0); + av_dict_set(&options, "tune", "film", 0); + av_dict_set(&options, "profile", "high", 0); + + if(codec_context->codec_id == AV_CODEC_ID_H264) { + av_dict_set(&options, "coder", "cabac", 0); // TODO: cavlc is faster than cabac but worse compression. Which to use? + } + + av_dict_set(&options, "strict", "experimental", 0); + + int ret = avcodec_open2(codec_context, codec_context->codec, &options); + if (ret < 0) { + fprintf(stderr, "Error: Could not open video codec: %s\n", av_error_to_string(ret)); + _exit(1); + } +} + +static void video_set_rc(gsr_video_codec video_codec, gsr_gpu_vendor vendor, gsr_bitrate_mode bitrate_mode, AVDictionary **options) { + switch(bitrate_mode) { + case GSR_BITRATE_MODE_QP: { + if(video_codec_is_vulkan(video_codec)) + av_dict_set(options, "rc_mode", "cqp", 0); + else if(vendor == GSR_GPU_VENDOR_NVIDIA) + av_dict_set(options, "rc", "constqp", 0); + else + av_dict_set(options, "rc_mode", "CQP", 0); + break; } + case GSR_BITRATE_MODE_VBR: { + if(video_codec_is_vulkan(video_codec)) + av_dict_set(options, "rc_mode", "vbr", 0); + else if(vendor == GSR_GPU_VENDOR_NVIDIA) + av_dict_set(options, "rc", "vbr", 0); + else + av_dict_set(options, "rc_mode", "VBR", 0); + break; + } + case GSR_BITRATE_MODE_CBR: { + if(video_codec_is_vulkan(video_codec)) + av_dict_set(options, "rc_mode", "cbr", 0); + else if(vendor == GSR_GPU_VENDOR_NVIDIA) + av_dict_set(options, "rc", "cbr", 0); + else + av_dict_set(options, "rc_mode", "CBR", 0); + break; + } + } +} +static void video_hardware_set_qp(AVCodecContext *codec_context, gsr_video_quality video_quality, gsr_gpu_vendor vendor, bool hdr, AVDictionary **options) { + // 8 bit / 10 bit = 80% + const float qp_multiply = hdr ? 8.0f/10.0f : 1.0f; + if(vendor == GSR_GPU_VENDOR_NVIDIA) { + // TODO: Test if these should be in the same range as vaapi if(codec_context->codec_id == AV_CODEC_ID_AV1) { switch(video_quality) { - case VideoQuality::MEDIUM: - av_dict_set_int(&options, "qp", 43, 0); + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); break; - case VideoQuality::HIGH: - av_dict_set_int(&options, "qp", 39, 0); + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); break; - case VideoQuality::VERY_HIGH: - av_dict_set_int(&options, "qp", 34, 0); + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); break; - case VideoQuality::ULTRA: - av_dict_set_int(&options, "qp", 28, 0); + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); break; } - } else if(very_old_gpu || codec_context->codec_id == AV_CODEC_ID_H264) { + } else if(codec_context->codec_id == AV_CODEC_ID_H264) { switch(video_quality) { - case VideoQuality::MEDIUM: - av_dict_set_int(&options, "qp", 37, 0); + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); break; - case VideoQuality::HIGH: - av_dict_set_int(&options, "qp", 32, 0); + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); break; - case VideoQuality::VERY_HIGH: - av_dict_set_int(&options, "qp", 27, 0); + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); break; - case VideoQuality::ULTRA: - av_dict_set_int(&options, "qp", 21, 0); + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); break; } - } else { + } else if(codec_context->codec_id == AV_CODEC_ID_HEVC) { switch(video_quality) { - case VideoQuality::MEDIUM: - av_dict_set_int(&options, "qp", 40, 0); + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); break; - case VideoQuality::HIGH: - av_dict_set_int(&options, "qp", 35, 0); + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); break; - case VideoQuality::VERY_HIGH: - av_dict_set_int(&options, "qp", 30, 0); + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); break; - case VideoQuality::ULTRA: - av_dict_set_int(&options, "qp", 24, 0); + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); break; } - } - - if(!supports_p4 && !supports_p5) - fprintf(stderr, "Info: your ffmpeg version is outdated. It's recommended that you use the flatpak version of gpu-screen-recorder version instead, which you can find at https://flathub.org/apps/details/com.dec05eba.gpu_screen_recorder\n"); - - //if(is_livestream) { - // av_dict_set_int(&options, "zerolatency", 1, 0); - // //av_dict_set(&options, "preset", "llhq", 0); - //} - - // I want to use a good preset for the gpu but all gpus prefer different - // presets. Nvidia and ffmpeg used to support "hq" preset that chose the best preset for the gpu - // with pretty good performance but you now have to choose p1-p7, which are gpu agnostic and on - // older gpus p5-p7 slow the gpu down to a crawl... - // "hq" is now just an alias for p7 in ffmpeg :( - // TODO: Temporary disable because of stuttering? - - // TODO: Preset is set to p5 for now but it should ideally be p6 or p7. - // This change is needed because for certain sizes of a window (or monitor?) such as 971x780 causes encoding to freeze - // when using h264 codec. This is a new(?) nvidia driver bug. - if(very_old_gpu) - av_dict_set(&options, "preset", supports_p4 ? "p4" : "medium", 0); - else - av_dict_set(&options, "preset", supports_p5 ? "p5" : "slow", 0); - - av_dict_set(&options, "tune", "hq", 0); - av_dict_set(&options, "rc", "constqp", 0); - - if(codec_context->codec_id == AV_CODEC_ID_H264) { - switch(pixel_format) { - case PixelFormat::YUV420: - av_dict_set(&options, "profile", "high", 0); + } else if(codec_context->codec_id == AV_CODEC_ID_VP8 || codec_context->codec_id == AV_CODEC_ID_VP9) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); break; - case PixelFormat::YUV444: - av_dict_set(&options, "profile", "high444p", 0); + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); break; - } - } else if(codec_context->codec_id == AV_CODEC_ID_AV1) { - switch(pixel_format) { - case PixelFormat::YUV420: - av_dict_set(&options, "rgb_mode", "yuv420", 0); + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); break; - case PixelFormat::YUV444: - av_dict_set(&options, "rgb_mode", "yuv444", 0); + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); break; } - } else { - //av_dict_set(&options, "profile", "main10", 0); - //av_dict_set(&options, "pix_fmt", "yuv420p16le", 0); } } else { if(codec_context->codec_id == AV_CODEC_ID_AV1) { // Using global_quality option } else if(codec_context->codec_id == AV_CODEC_ID_H264) { switch(video_quality) { - case VideoQuality::MEDIUM: - av_dict_set_int(&options, "qp", 32, 0); + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); break; - case VideoQuality::HIGH: - av_dict_set_int(&options, "qp", 28, 0); + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); break; - case VideoQuality::VERY_HIGH: - av_dict_set_int(&options, "qp", 24, 0); + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); break; - case VideoQuality::ULTRA: - av_dict_set_int(&options, "qp", 18, 0); + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); break; } - } else { + } else if(codec_context->codec_id == AV_CODEC_ID_HEVC) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); + break; + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); + break; + } + } else if(codec_context->codec_id == AV_CODEC_ID_VP8 || codec_context->codec_id == AV_CODEC_ID_VP9) { switch(video_quality) { - case VideoQuality::MEDIUM: - av_dict_set_int(&options, "qp", 34, 0); + case GSR_VIDEO_QUALITY_MEDIUM: + av_dict_set_int(options, "qp", 35 * qp_multiply, 0); break; - case VideoQuality::HIGH: - av_dict_set_int(&options, "qp", 30, 0); + case GSR_VIDEO_QUALITY_HIGH: + av_dict_set_int(options, "qp", 30 * qp_multiply, 0); break; - case VideoQuality::VERY_HIGH: - av_dict_set_int(&options, "qp", 26, 0); + case GSR_VIDEO_QUALITY_VERY_HIGH: + av_dict_set_int(options, "qp", 25 * qp_multiply, 0); break; - case VideoQuality::ULTRA: - av_dict_set_int(&options, "qp", 20, 0); + case GSR_VIDEO_QUALITY_ULTRA: + av_dict_set_int(options, "qp", 22 * qp_multiply, 0); break; } } + } +} + +static void open_video_hardware(AVCodecContext *codec_context, bool low_power, const gsr_egl &egl, const args_parser &arg_parser) { + const gsr_color_depth color_depth = video_codec_to_bit_depth(arg_parser.video_codec); + const bool hdr = video_codec_is_hdr(arg_parser.video_codec); + AVDictionary *options = nullptr; + + if(arg_parser.bitrate_mode == GSR_BITRATE_MODE_QP) + video_hardware_set_qp(codec_context, arg_parser.video_quality, egl.gpu_info.vendor, hdr, &options); + video_set_rc(arg_parser.video_codec, egl.gpu_info.vendor, arg_parser.bitrate_mode, &options); + + // TODO: Enable multipass + + dict_set_profile(codec_context, egl.gpu_info.vendor, color_depth, arg_parser.video_codec, &options); + + if(video_codec_is_vulkan(arg_parser.video_codec)) { + av_dict_set_int(&options, "async_depth", 3, 0); + av_dict_set(&options, "tune", "hq", 0); + av_dict_set(&options, "usage", "record", 0); // TODO: Set to stream when streaming + av_dict_set(&options, "content", "rendered", 0); + } else if(egl.gpu_info.vendor == GSR_GPU_VENDOR_NVIDIA) { + // TODO: These dont seem to be necessary + // av_dict_set_int(&options, "zerolatency", 1, 0); + // if(codec_context->codec_id == AV_CODEC_ID_AV1) { + // av_dict_set(&options, "tune", "ll", 0); + // } else if(codec_context->codec_id == AV_CODEC_ID_H264 || codec_context->codec_id == AV_CODEC_ID_HEVC) { + // av_dict_set(&options, "preset", "llhq", 0); + // av_dict_set(&options, "tune", "ll", 0); + // } + av_dict_set(&options, "tune", "hq", 0); + + switch(arg_parser.tune) { + case GSR_TUNE_PERFORMANCE: + //av_dict_set(&options, "multipass", "qres", 0); + break; + case GSR_TUNE_QUALITY: + av_dict_set(&options, "multipass", "fullres", 0); + av_dict_set(&options, "preset", "p6", 0); + av_dict_set_int(&options, "rc-lookahead", 0, 0); + break; + } + + if(codec_context->codec_id == AV_CODEC_ID_H264) { + // TODO: h264 10bit? + // TODO: + // switch(pixel_format) { + // case GSR_PIXEL_FORMAT_YUV420: + // av_dict_set_int(&options, "profile", AV_PROFILE_H264_HIGH, 0); + // break; + // case GSR_PIXEL_FORMAT_YUV444: + // av_dict_set_int(&options, "profile", AV_PROFILE_H264_HIGH_444, 0); + // break; + // } + } else if(codec_context->codec_id == AV_CODEC_ID_AV1) { + switch(arg_parser.pixel_format) { + case GSR_PIXEL_FORMAT_YUV420: + av_dict_set(&options, "rgb_mode", "yuv420", 0); + break; + case GSR_PIXEL_FORMAT_YUV444: + av_dict_set(&options, "rgb_mode", "yuv444", 0); + break; + } + } else if(codec_context->codec_id == AV_CODEC_ID_HEVC) { + //av_dict_set(&options, "pix_fmt", "yuv420p16le", 0); + } + } else { // TODO: More quality options - av_dict_set(&options, "rc_mode", "CQP", 0); - //av_dict_set_int(&options, "low_power", 1, 0); + if(low_power) + av_dict_set_int(&options, "low_power", 1, 0); + // Improves performance but increases vram. + // TODO: Might need a different async_depth for optimal performance on different amd/intel gpus + av_dict_set_int(&options, "async_depth", 3, 0); if(codec_context->codec_id == AV_CODEC_ID_H264) { - av_dict_set(&options, "profile", "high", 0); - av_dict_set_int(&options, "quality", 5, 0); // quality preset + // Removed because it causes stutter in games for some people + //av_dict_set_int(&options, "quality", 5, 0); // quality preset } else if(codec_context->codec_id == AV_CODEC_ID_AV1) { - av_dict_set(&options, "profile", "main", 0); // TODO: use professional instead? av_dict_set(&options, "tier", "main", 0); - } else { - av_dict_set(&options, "profile", "main", 0); + } else if(codec_context->codec_id == AV_CODEC_ID_HEVC) { + if(hdr) + av_dict_set(&options, "sei", "hdr", 0); } + + // TODO: vp8/vp9 10bit } if(codec_context->codec_id == AV_CODEC_ID_H264) { @@ -761,103 +921,52 @@ static void open_video(AVCodecContext *codec_context, VideoQuality video_quality } } -static void usage_header() { - fprintf(stderr, "usage: gpu-screen-recorder -w <window_id|monitor|focused> [-c <container_format>] [-s WxH] -f <fps> [-a <audio_input>] [-q <quality>] [-r <replay_buffer_size_sec>] [-k h264|h265|av1] [-ac aac|opus|flac] [-oc yes|no] [-fm cfr|vfr] [-v yes|no] [-h|--help] [-o <output_file>] [-mf yes|no]\n"); -} - -static void usage_full() { - usage_header(); - fprintf(stderr, "\n"); - fprintf(stderr, "OPTIONS:\n"); - fprintf(stderr, " -w Window to record, a display, \"screen\", \"screen-direct\", \"screen-direct-force\" or \"focused\".\n"); - fprintf(stderr, " The display is the display (monitor) name in xrandr and if \"screen\", \"screen-direct\" or \"screen-direct-force\" is selected then all displays are recorded.\n"); - fprintf(stderr, " If this is \"focused\" then the currently focused window is recorded. When recording the focused window then the -s option has to be used as well.\n"); - fprintf(stderr, " \"screen-direct\"/\"screen-direct-force\" skips one texture copy for fullscreen applications so it may lead to better performance and it works with VRR monitors\n"); - fprintf(stderr, " when recording fullscreen application but may break some applications, such as mpv in fullscreen mode or might cause games to freeze/crash because of nvidia driver issues.\n"); - fprintf(stderr, " Direct mode doesn't capture cursor either.\n"); - fprintf(stderr, " \"screen-direct-force\" is not recommended unless you use a VRR monitor and you are aware that using this option can cause games to freeze/crash or other issues.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -c Container format for output file, for example mp4, or flv. Only required if no output file is specified or if recording in replay buffer mode.\n"); - fprintf(stderr, " If an output file is specified and -c is not used then the container format is determined from the output filename extension.\n"); - fprintf(stderr, " Only containers that support h264, hevc or av1 are supported, which means that only mp4, mkv, flv (and some others) are supported.\n"); - fprintf(stderr, " WebM is not supported yet.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -s The size (area) to record at in the format WxH, for example 1920x1080. This option is only supported (and required) when -w is \"focused\".\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -f Framerate to record at.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -a Audio device to record from (pulse audio device). Can be specified multiple times. Each time this is specified a new audio track is added for the specified audio device.\n"); - fprintf(stderr, " A name can be given to the audio input device by prefixing the audio input with <name>/, for example \"dummy/alsa_output.pci-0000_00_1b.0.analog-stereo.monitor\".\n"); - fprintf(stderr, " Multiple audio devices can be merged into one audio track by using \"|\" as a separator into one -a argument, for example: -a \"alsa_output1|alsa_output2\".\n"); - fprintf(stderr, " Optional, no audio track is added by default.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -q Video quality. Should be either 'medium', 'high', 'very_high' or 'ultra'. 'high' is the recommended option when live streaming or when you have a slower harddrive.\n"); - fprintf(stderr, " Optional, set to 'very_high' be default.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -r Replay buffer size in seconds. If this is set, then only the last seconds as set by this option will be stored\n"); - fprintf(stderr, " and the video will only be saved when the gpu-screen-recorder is closed. This feature is similar to Nvidia's instant replay feature.\n"); - fprintf(stderr, " This option has be between 5 and 1200. Note that the replay buffer size will not always be precise, because of keyframes. Optional, disabled by default.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -k Video codec to use. Should be either 'auto', 'h264', 'h265', 'av1'. Defaults to 'auto' which defaults to 'h265' unless recording at fps higher than 60. Defaults to 'h264' on intel.\n"); - fprintf(stderr, " Forcefully set to 'h264' if -c is 'flv'.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -ac Audio codec to use. Should be either 'aac', 'opus' or 'flac'. Defaults to 'opus' for .mp4/.mkv files, otherwise defaults to 'aac'.\n"); - fprintf(stderr, " 'opus' and 'flac' is only supported by .mp4/.mkv files. 'opus' is recommended for best performance and smallest audio size.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -oc Overclock memory transfer rate to the maximum performance level. This only applies to NVIDIA on X11 and exists to overcome a bug in NVIDIA driver where performance level. The same issue exists on Wayland but overclocking is not possible on Wayland.\n"); - fprintf(stderr, " is dropped when you record a game. Only needed if you are recording a game that is bottlenecked by GPU.\n"); - fprintf(stderr, " Works only if your have \"Coolbits\" set to \"12\" in NVIDIA X settings, see README for more information. Note! use at your own risk! Optional, disabled by default.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -fm Framerate mode. Should be either 'cfr' or 'vfr'. Defaults to 'cfr' on NVIDIA X11 and 'vfr' on AMD/Intel X11/Wayland or NVIDIA Wayland.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -v Prints per second, fps updates. Optional, set to 'yes' by default.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -h Show this help.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, " -mf Organise replays in folders based on the current date.\n"); - fprintf(stderr, "\n"); - //fprintf(stderr, " -pixfmt The pixel format to use for the output video. yuv420 is the most common format and is best supported, but the color is compressed, so colors can look washed out and certain colors of text can look bad. Use yuv444 for no color compression, but the video may not work everywhere and it may not work with hardware video decoding. Optional, defaults to yuv420\n"); - fprintf(stderr, " -o The output file path. If omitted then the encoded data is sent to stdout. Required in replay mode (when using -r).\n"); - fprintf(stderr, " In replay mode this has to be a directory instead of a file.\n"); - fprintf(stderr, "\n"); - fprintf(stderr, "NOTES:\n"); - fprintf(stderr, " Send signal SIGINT to gpu-screen-recorder (Ctrl+C, or killall gpu-screen-recorder) to stop and save the recording (when not using replay mode).\n"); - fprintf(stderr, " Send signal SIGUSR1 to gpu-screen-recorder (killall -SIGUSR1 gpu-screen-recorder) to save a replay (when in replay mode).\n"); - fprintf(stderr, "\n"); - fprintf(stderr, "EXAMPLES\n"); - fprintf(stderr, " gpu-screen-recorder -w screen -f 60 -a \"$(pactl get-default-sink).monitor\" -o video.mp4\n"); - fprintf(stderr, " gpu-screen-recorder -w screen -f 60 -a \"$(pactl get-default-sink).monitor|$(pactl get-default-source)\" -o video.mp4\n"); - //fprintf(stderr, " gpu-screen-recorder -w screen -f 60 -q ultra -pixfmt yuv444 -o video.mp4\n"); - _exit(1); -} - -static void usage() { - usage_header(); - _exit(1); -} +static const int save_replay_seconds_full = -1; static sig_atomic_t running = 1; -static sig_atomic_t save_replay = 0; +static sig_atomic_t toggle_pause = 0; +static sig_atomic_t toggle_replay_recording = 0; +static sig_atomic_t save_replay_seconds = 0; -static void int_handler(int) { +static void stop_handler(int) { running = 0; } +static void toggle_pause_handler(int) { + toggle_pause = 1; +} + +static void toggle_replay_recording_handler(int) { + toggle_replay_recording = 1; +} + static void save_replay_handler(int) { - save_replay = 1; + save_replay_seconds = save_replay_seconds_full; } -struct Arg { - std::vector<const char*> values; - bool optional = false; - bool list = false; +static void save_replay_10_seconds_handler(int) { + save_replay_seconds = 10; +} - const char* value() const { - if(values.empty()) - return nullptr; - return values.front(); - } -}; +static void save_replay_30_seconds_handler(int) { + save_replay_seconds = 30; +} + +static void save_replay_1_minute_handler(int) { + save_replay_seconds = 60; +} + +static void save_replay_5_minutes_handler(int) { + save_replay_seconds = 60*5; +} + +static void save_replay_10_minutes_handler(int) { + save_replay_seconds = 60*10; +} + +static void save_replay_30_minutes_handler(int) { + save_replay_seconds = 60*30; +} static bool is_hex_num(char c) { return (c >= 'A' && c <= 'F') || (c >= 'a' && c <= 'f') || (c >= '0' && c <= '9'); @@ -891,7 +1000,7 @@ static std::string get_date_str() { time_t now = time(NULL); struct tm *t = localtime(&now); strftime(str, sizeof(str)-1, "%Y-%m-%d_%H-%M-%S", t); - return str; + return str; } static std::string get_date_only_str() { @@ -922,7 +1031,70 @@ static AVStream* create_stream(AVFormatContext *av_format_context, AVCodecContex return stream; } -struct AudioDevice { +static void run_recording_saved_script_async(const char *script_file, const char *video_file, const char *type) { + char script_file_full[PATH_MAX]; + script_file_full[0] = '\0'; + if(!realpath(script_file, script_file_full)) { + fprintf(stderr, "Error: script file not found: %s\n", script_file); + return; + } + + const char *args[7]; + const bool inside_flatpak = getenv("FLATPAK_ID") != NULL; + + if(inside_flatpak) { + args[0] = "flatpak-spawn"; + args[1] = "--host"; + args[2] = "--"; + args[3] = script_file_full; + args[4] = video_file; + args[5] = type; + args[6] = NULL; + } else { + args[0] = script_file_full; + args[1] = video_file; + args[2] = type; + args[3] = NULL; + } + + pid_t pid = fork(); + if(pid == -1) { + perror(script_file_full); + return; + } else if(pid == 0) { // child + setsid(); + signal(SIGHUP, SIG_IGN); + + pid_t second_child = fork(); + if(second_child == 0) { // child + execvp(args[0], (char* const*)args); + perror(script_file_full); + _exit(127); + } else if(second_child != -1) { // parent + _exit(0); + } + } else { // parent + waitpid(pid, NULL, 0); + } +} + +static double audio_codec_get_desired_delay(gsr_audio_codec audio_codec, int fps) { + const double fps_inv = 1.0 / (double)fps; + const double base = 0.01 + 1.0/165.0; + switch(audio_codec) { + case GSR_AUDIO_CODEC_OPUS: + return std::max(0.0, base - fps_inv); + case GSR_AUDIO_CODEC_AAC: + return std::max(0.0, (base + 0.008) * 2.0 - fps_inv); + case GSR_AUDIO_CODEC_FLAC: + // TODO: Test + return std::max(0.0, base - fps_inv); + } + assert(false); + return std::max(0.0, base - fps_inv); +} + +struct AudioDeviceData { SoundDevice sound_device; AudioInput audio_input; AVFilterContext *src_filter_ctx = nullptr; @@ -932,180 +1104,262 @@ struct AudioDevice { // TODO: Cleanup struct AudioTrack { + std::string name; AVCodecContext *codec_context = nullptr; - AVStream *stream = nullptr; - std::vector<AudioDevice> audio_devices; + std::vector<AudioDeviceData> audio_devices; AVFilterGraph *graph = nullptr; AVFilterContext *sink = nullptr; int stream_index = 0; + int64_t pts = 0; }; -static std::future<void> save_replay_thread; -static std::vector<std::shared_ptr<PacketData>> save_replay_packets; -static std::string save_replay_output_filepath; +static bool add_hdr_metadata_to_video_stream(gsr_capture *cap, AVStream *video_stream) { + size_t light_metadata_size = 0; + size_t mastering_display_metadata_size = 0; + AVContentLightMetadata *light_metadata = av_content_light_metadata_alloc(&light_metadata_size); + #if LIBAVUTIL_VERSION_INT < AV_VERSION_INT(59, 37, 100) + AVMasteringDisplayMetadata *mastering_display_metadata = av_mastering_display_metadata_alloc(); + mastering_display_metadata_size = sizeof(*mastering_display_metadata); + #else + AVMasteringDisplayMetadata *mastering_display_metadata = av_mastering_display_metadata_alloc_size(&mastering_display_metadata_size); + #endif -static int create_directory_recursive(char *path) { - int path_len = strlen(path); - char *p = path; - char *end = path + path_len; - for(;;) { - char *slash_p = strchr(p, '/'); + if(!light_metadata || !mastering_display_metadata) { + if(light_metadata) + av_freep(&light_metadata); - // Skips first '/', we don't want to try and create the root directory - if(slash_p == path) { - ++p; - continue; - } + if(mastering_display_metadata) + av_freep(&mastering_display_metadata); - if(!slash_p) - slash_p = end; + return false; + } - char prev_char = *slash_p; - *slash_p = '\0'; - int err = mkdir(path, S_IRWXU); - *slash_p = prev_char; + if(!gsr_capture_set_hdr_metadata(cap, mastering_display_metadata, light_metadata)) { + av_freep(&light_metadata); + av_freep(&mastering_display_metadata); + return false; + } - if(err == -1 && errno != EEXIST) - return err; + // TODO: More error checking - if(slash_p == end) - break; - else - p = slash_p + 1; - } - return 0; + #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(60, 31, 102) + const bool content_light_level_added = av_stream_add_side_data(video_stream, AV_PKT_DATA_CONTENT_LIGHT_LEVEL, (uint8_t*)light_metadata, light_metadata_size) == 0; + #else + const bool content_light_level_added = av_packet_side_data_add(&video_stream->codecpar->coded_side_data, &video_stream->codecpar->nb_coded_side_data, AV_PKT_DATA_CONTENT_LIGHT_LEVEL, light_metadata, light_metadata_size, 0) != NULL; + #endif + + #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(60, 31, 102) + const bool mastering_display_metadata_added = av_stream_add_side_data(video_stream, AV_PKT_DATA_MASTERING_DISPLAY_METADATA, (uint8_t*)mastering_display_metadata, mastering_display_metadata_size) == 0; + #else + const bool mastering_display_metadata_added = av_packet_side_data_add(&video_stream->codecpar->coded_side_data, &video_stream->codecpar->nb_coded_side_data, AV_PKT_DATA_MASTERING_DISPLAY_METADATA, mastering_display_metadata, mastering_display_metadata_size, 0) != NULL; + #endif + + if(!content_light_level_added) + av_freep(&light_metadata); + + if(!mastering_display_metadata_added) + av_freep(&mastering_display_metadata); + + // Return true even on failure because we dont want to retry adding hdr metadata on failure + return true; } -static void save_replay_async(AVCodecContext *video_codec_context, int video_stream_index, std::vector<AudioTrack> &audio_tracks, std::deque<std::shared_ptr<PacketData>> &frame_data_queue, bool frames_erased, std::string output_dir, const char *container_format, const std::string &file_extension, std::mutex &write_output_mutex, bool make_folders) { - if(save_replay_thread.valid()) - return; - - size_t start_index = (size_t)-1; - int64_t video_pts_offset = 0; - int64_t audio_pts_offset = 0; +struct RecordingStartAudio { + const AudioTrack *audio_track; + AVStream *stream; +}; - { - std::lock_guard<std::mutex> lock(write_output_mutex); - start_index = (size_t)-1; - for(size_t i = 0; i < frame_data_queue.size(); ++i) { - const AVPacket &av_packet = frame_data_queue[i]->data; - if((av_packet.flags & AV_PKT_FLAG_KEY) && av_packet.stream_index == video_stream_index) { - start_index = i; - break; - } - } +struct RecordingStartResult { + AVFormatContext *av_format_context = nullptr; + AVStream *video_stream = nullptr; + std::vector<RecordingStartAudio> audio_inputs; +}; - if(start_index == (size_t)-1) - return; +static RecordingStartResult start_recording_create_streams(const char *filename, const char *container_format, AVCodecContext *video_codec_context, const std::vector<AudioTrack> &audio_tracks, bool hdr, gsr_capture *capture) { + AVFormatContext *av_format_context; + avformat_alloc_output_context2(&av_format_context, nullptr, container_format, filename); - if(frames_erased) { - video_pts_offset = frame_data_queue[start_index]->data.pts; - - // Find the next audio packet to use as audio pts offset - for(size_t i = start_index; i < frame_data_queue.size(); ++i) { - const AVPacket &av_packet = frame_data_queue[i]->data; - if(av_packet.stream_index != video_stream_index) { - audio_pts_offset = av_packet.pts; - break; - } - } - } else { - start_index = 0; - } + AVStream *video_stream = create_stream(av_format_context, video_codec_context); + avcodec_parameters_from_context(video_stream->codecpar, video_codec_context); - save_replay_packets.resize(frame_data_queue.size()); - for(size_t i = 0; i < frame_data_queue.size(); ++i) { - save_replay_packets[i] = frame_data_queue[i]; - } + RecordingStartResult result; + result.audio_inputs.reserve(audio_tracks.size()); + + for(const AudioTrack &audio_track : audio_tracks) { + AVStream *audio_stream = create_stream(av_format_context, audio_track.codec_context); + if(!audio_track.name.empty()) + av_dict_set(&audio_stream->metadata, "title", audio_track.name.c_str(), 0); + avcodec_parameters_from_context(audio_stream->codecpar, audio_track.codec_context); + result.audio_inputs.push_back({&audio_track, audio_stream}); + } + + const int open_ret = avio_open(&av_format_context->pb, filename, AVIO_FLAG_WRITE); + if(open_ret < 0) { + fprintf(stderr, "gsr error: start: could not open '%s': %s\n", filename, av_error_to_string(open_ret)); + return result; + } + + AVDictionary *options = nullptr; + av_dict_set(&options, "strict", "experimental", 0); + + const int header_write_ret = avformat_write_header(av_format_context, &options); + av_dict_free(&options); + if(header_write_ret < 0) { + fprintf(stderr, "gsr error: start: error occurred when writing header to output file: %s\n", av_error_to_string(header_write_ret)); + avio_close(av_format_context->pb); + avformat_free_context(av_format_context); + return result; + } + + if(hdr) + add_hdr_metadata_to_video_stream(capture, video_stream); + + result.av_format_context = av_format_context; + result.video_stream = video_stream; + return result; +} + +static bool stop_recording_close_streams(AVFormatContext *av_format_context) { + bool trailer_written = true; + if(av_write_trailer(av_format_context) != 0) { + fprintf(stderr, "gsr error: end: failed to write trailer\n"); + trailer_written = false; } - if (make_folders) { - std::string output_folder = output_dir + '/' + get_date_only_str(); - create_directory_recursive(&output_folder[0]); - save_replay_output_filepath = output_folder + "/Replay_" + get_time_only_str() + "." + file_extension; + const bool closed = avio_close(av_format_context->pb) == 0; + avformat_free_context(av_format_context); + return trailer_written && closed; +} + +static std::future<void> save_replay_thread; +static std::string save_replay_output_filepath; + +static std::string create_new_recording_filepath_from_timestamp(std::string directory, const char *filename_prefix, const std::string &file_extension, bool date_folders) { + std::string output_filepath; + if(date_folders) { + std::string output_folder = directory + '/' + get_date_only_str(); + if(create_directory_recursive(&output_folder[0]) != 0) + fprintf(stderr, "Error: failed to create directory: %s\n", output_folder.c_str()); + output_filepath = output_folder + "/" + filename_prefix + "_" + get_time_only_str() + "." + file_extension; } else { - create_directory_recursive(&output_dir[0]); - save_replay_output_filepath = output_dir + "/Replay_" + get_date_str() + "." + file_extension; + if(create_directory_recursive(&directory[0]) != 0) + fprintf(stderr, "Error: failed to create directory: %s\n", directory.c_str()); + output_filepath = directory + "/" + filename_prefix + "_" + get_date_str() + "." + file_extension; } + return output_filepath; +} - save_replay_thread = std::async(std::launch::async, [video_stream_index, container_format, start_index, video_pts_offset, audio_pts_offset, video_codec_context, &audio_tracks]() mutable { - AVFormatContext *av_format_context; - avformat_alloc_output_context2(&av_format_context, nullptr, container_format, nullptr); +static RecordingStartAudio* get_recording_start_item_by_stream_index(RecordingStartResult &result, int stream_index) { + for(auto &audio_input : result.audio_inputs) { + if(audio_input.stream->index == stream_index) + return &audio_input; + } + return nullptr; +} - AVStream *video_stream = create_stream(av_format_context, video_codec_context); - avcodec_parameters_from_context(video_stream->codecpar, video_codec_context); +static void save_replay_async(AVCodecContext *video_codec_context, int video_stream_index, const std::vector<AudioTrack> &audio_tracks, gsr_replay_buffer *replay_buffer, std::string output_dir, const char *container_format, const std::string &file_extension, bool date_folders, bool hdr, gsr_capture *capture, int current_save_replay_seconds) { + if(save_replay_thread.valid()) + return; - std::unordered_map<int, AudioTrack*> stream_index_to_audio_track_map; - for(AudioTrack &audio_track : audio_tracks) { - stream_index_to_audio_track_map[audio_track.stream_index] = &audio_track; - AVStream *audio_stream = create_stream(av_format_context, audio_track.codec_context); - avcodec_parameters_from_context(audio_stream->codecpar, audio_track.codec_context); - audio_track.stream = audio_stream; - } + const gsr_replay_buffer_iterator search_start_iterator = current_save_replay_seconds == save_replay_seconds_full ? gsr_replay_buffer_iterator{0, 0} : gsr_replay_buffer_find_packet_index_by_time_passed(replay_buffer, current_save_replay_seconds); + const gsr_replay_buffer_iterator video_start_iterator = gsr_replay_buffer_find_keyframe(replay_buffer, search_start_iterator, video_stream_index, false); + if(video_start_iterator.packet_index == (size_t)-1) { + fprintf(stderr, "gsr error: failed to save replay: failed to find a video keyframe. perhaps replay was saved too fast, before anything has been recorded\n"); + return; + } - int ret = avio_open(&av_format_context->pb, save_replay_output_filepath.c_str(), AVIO_FLAG_WRITE); - if (ret < 0) { - fprintf(stderr, "Error: Could not open '%s': %s. Make sure %s is an existing directory with write access\n", save_replay_output_filepath.c_str(), av_error_to_string(ret), save_replay_output_filepath.c_str()); - return; - } + const gsr_replay_buffer_iterator audio_start_iterator = gsr_replay_buffer_find_keyframe(replay_buffer, video_start_iterator, video_stream_index, true); + // if(audio_start_index == (size_t)-1) { + // fprintf(stderr, "gsr error: failed to save replay: failed to find an audio keyframe. perhaps replay was saved too fast, before anything has been recorded\n"); + // return; + // } - AVDictionary *options = nullptr; - av_dict_set(&options, "strict", "experimental", 0); + const int64_t video_pts_offset = gsr_replay_buffer_iterator_get_packet(replay_buffer, video_start_iterator)->pts; + const int64_t audio_pts_offset = audio_start_iterator.packet_index == (size_t)-1 ? 0 : gsr_replay_buffer_iterator_get_packet(replay_buffer, audio_start_iterator)->pts; - ret = avformat_write_header(av_format_context, &options); - if (ret < 0) { - fprintf(stderr, "Error occurred when writing header to output file: %s\n", av_error_to_string(ret)); - return; - } + gsr_replay_buffer *cloned_replay_buffer = gsr_replay_buffer_clone(replay_buffer); + if(!cloned_replay_buffer) { + // TODO: Return this error to mark the replay as failed + fprintf(stderr, "gsr error: failed to save replay: failed to clone replay buffer\n"); + return; + } + + std::string output_filepath = create_new_recording_filepath_from_timestamp(output_dir, "Replay", file_extension, date_folders); + RecordingStartResult recording_start_result = start_recording_create_streams(output_filepath.c_str(), container_format, video_codec_context, audio_tracks, hdr, capture); + if(!recording_start_result.av_format_context) + return; + + save_replay_output_filepath = std::move(output_filepath); + + save_replay_thread = std::async(std::launch::async, [video_stream_index, recording_start_result, video_start_iterator, video_pts_offset, audio_pts_offset, video_codec_context, cloned_replay_buffer]() mutable { + gsr_replay_buffer_iterator replay_iterator = video_start_iterator; + for(;;) { + AVPacket *replay_packet = gsr_replay_buffer_iterator_get_packet(cloned_replay_buffer, replay_iterator); + uint8_t *replay_packet_data = NULL; + if(replay_packet) + replay_packet_data = gsr_replay_buffer_iterator_get_packet_data(cloned_replay_buffer, replay_iterator); + + if(!replay_packet) { + fprintf(stderr, "gsr error: save_replay_async: no replay packet\n"); + break; + } + + if(!replay_packet->data && !replay_packet_data) { + fprintf(stderr, "gsr error: save_replay_async: no replay packet data\n"); + break; + } - for(size_t i = start_index; i < save_replay_packets.size(); ++i) { // TODO: Check if successful AVPacket av_packet; memset(&av_packet, 0, sizeof(av_packet)); - //av_packet_from_data(av_packet, save_replay_packets[i]->data.data, save_replay_packets[i]->data.size); - av_packet.data = save_replay_packets[i]->data.data; - av_packet.size = save_replay_packets[i]->data.size; - av_packet.stream_index = save_replay_packets[i]->data.stream_index; - av_packet.pts = save_replay_packets[i]->data.pts; - av_packet.dts = save_replay_packets[i]->data.pts; - av_packet.flags = save_replay_packets[i]->data.flags; - - AVStream *stream = video_stream; + //av_packet_from_data(av_packet, replay_packet->data, replay_packet->size); + av_packet.data = replay_packet->data ? replay_packet->data : replay_packet_data; + av_packet.size = replay_packet->size; + av_packet.stream_index = replay_packet->stream_index; + av_packet.pts = replay_packet->pts; + av_packet.dts = replay_packet->pts; + av_packet.flags = replay_packet->flags; + //av_packet.duration = replay_packet->duration; + + AVStream *stream = recording_start_result.video_stream; AVCodecContext *codec_context = video_codec_context; if(av_packet.stream_index == video_stream_index) { av_packet.pts -= video_pts_offset; av_packet.dts -= video_pts_offset; } else { - AudioTrack *audio_track = stream_index_to_audio_track_map[av_packet.stream_index]; - stream = audio_track->stream; + RecordingStartAudio *recording_start_audio = get_recording_start_item_by_stream_index(recording_start_result, av_packet.stream_index); + if(!recording_start_audio) { + fprintf(stderr, "gsr error: save_replay_async: failed to find audio stream by index: %d\n", av_packet.stream_index); + free(replay_packet_data); + continue; + } + + const AudioTrack *audio_track = recording_start_audio->audio_track; + stream = recording_start_audio->stream; codec_context = audio_track->codec_context; av_packet.pts -= audio_pts_offset; av_packet.dts -= audio_pts_offset; } - av_packet.stream_index = stream->index; + //av_packet.stream_index = stream->index; av_packet_rescale_ts(&av_packet, codec_context->time_base, stream->time_base); - ret = av_write_frame(av_format_context, &av_packet); + const int ret = av_write_frame(recording_start_result.av_format_context, &av_packet); if(ret < 0) - fprintf(stderr, "Error: Failed to write frame index %d to muxer, reason: %s (%d)\n", stream->index, av_error_to_string(ret), ret); + fprintf(stderr, "Error: Failed to write frame index %d to muxer, reason: %s (%d)\n", av_packet.stream_index, av_error_to_string(ret), ret); + + free(replay_packet_data); //av_packet_free(&av_packet); + if(!gsr_replay_buffer_iterator_next(cloned_replay_buffer, &replay_iterator)) + break; } - if (av_write_trailer(av_format_context) != 0) - fprintf(stderr, "Failed to write trailer\n"); - - avio_close(av_format_context->pb); - avformat_free_context(av_format_context); - av_dict_free(&options); - - for(AudioTrack &audio_track : audio_tracks) { - audio_track.stream = nullptr; - } + stop_recording_close_streams(recording_start_result.av_format_context); + gsr_replay_buffer_destroy(cloned_replay_buffer); }); } @@ -1123,57 +1377,99 @@ static void split_string(const std::string &str, char delimiter, std::function<b } } -static std::vector<AudioInput> parse_audio_input_arg(const char *str) { - std::vector<AudioInput> audio_inputs; - split_string(str, '|', [&audio_inputs](const char *sub, size_t size) { +static bool string_starts_with(const std::string &str, const char *substr) { + int len = strlen(substr); + return (int)str.size() >= len && memcmp(str.data(), substr, len) == 0; +} + +static bool string_ends_with(const char *str, const char *substr) { + int str_len = strlen(str); + int substr_len = strlen(substr); + return str_len >= substr_len && memcmp(str + str_len - substr_len, substr, substr_len) == 0; +} + +static const AudioDevice* get_audio_device_by_name(const std::vector<AudioDevice> &audio_devices, const char *name) { + for(const auto &audio_device : audio_devices) { + if(strcmp(audio_device.name.c_str(), name) == 0) + return &audio_device; + } + return nullptr; +} + +static MergedAudioInputs parse_audio_input_arg(const char *str) { + MergedAudioInputs result; + + split_string(str, '|', [&](const char *sub, size_t size) { AudioInput audio_input; audio_input.name.assign(sub, size); - const size_t index = audio_input.name.find('/'); - if(index != std::string::npos) { - audio_input.description = audio_input.name.substr(0, index); - audio_input.name.erase(audio_input.name.begin(), audio_input.name.begin() + index + 1); + + if(string_starts_with(audio_input.name.c_str(), "app:")) { + audio_input.name.erase(audio_input.name.begin(), audio_input.name.begin() + 4); + audio_input.type = AudioInputType::APPLICATION; + audio_input.inverted = false; + result.audio_inputs.push_back(std::move(audio_input)); + return true; + } else if(string_starts_with(audio_input.name.c_str(), "app-inverse:")) { + audio_input.name.erase(audio_input.name.begin(), audio_input.name.begin() + 12); + audio_input.type = AudioInputType::APPLICATION; + audio_input.inverted = true; + result.audio_inputs.push_back(std::move(audio_input)); + return true; + } else if(string_starts_with(audio_input.name.c_str(), "device:")) { + audio_input.name.erase(audio_input.name.begin(), audio_input.name.begin() + 7); + audio_input.type = AudioInputType::DEVICE; + result.audio_inputs.push_back(std::move(audio_input)); + return true; + } else { + audio_input.type = AudioInputType::DEVICE; + result.audio_inputs.push_back(std::move(audio_input)); + return true; } - audio_inputs.push_back(std::move(audio_input)); - return true; }); - return audio_inputs; -} -// TODO: Does this match all livestreaming cases? -static bool is_livestream_path(const char *str) { - const int len = strlen(str); - if((len >= 7 && memcmp(str, "http://", 7) == 0) || (len >= 8 && memcmp(str, "https://", 8) == 0)) - return true; - else if((len >= 7 && memcmp(str, "rtmp://", 7) == 0) || (len >= 8 && memcmp(str, "rtmps://", 8) == 0)) - return true; - else - return false; + return result; } -// TODO: Proper cleanup -static int init_filter_graph(AVCodecContext *audio_codec_context, AVFilterGraph **graph, AVFilterContext **sink, std::vector<AVFilterContext*> &src_filter_ctx, size_t num_sources) { +static int init_filter_graph(AVCodecContext* audio_codec_context, AVFilterGraph** graph, AVFilterContext** sink, std::vector<AVFilterContext*>& src_filter_ctx, size_t num_sources) { char ch_layout[64]; int err = 0; - - AVFilterGraph *filter_graph = avfilter_graph_alloc(); + ch_layout[0] = '\0'; + + // C89-style variable declaration to + // avoid problems because of goto + AVFilterGraph* filter_graph = nullptr; + AVFilterContext* mix_ctx = nullptr; + + const AVFilter* mix_filter = nullptr; + const AVFilter* abuffersink = nullptr; + AVFilterContext* abuffersink_ctx = nullptr; + char args[512] = { 0 }; +#if LIBAVFILTER_VERSION_INT >= AV_VERSION_INT(7, 107, 100) + bool normalize = false; +#endif + + filter_graph = avfilter_graph_alloc(); if (!filter_graph) { fprintf(stderr, "Unable to create filter graph.\n"); - return AVERROR(ENOMEM); + err = AVERROR(ENOMEM); + goto fail; } - + for(size_t i = 0; i < num_sources; ++i) { const AVFilter *abuffer = avfilter_get_by_name("abuffer"); if (!abuffer) { fprintf(stderr, "Could not find the abuffer filter.\n"); - return AVERROR_FILTER_NOT_FOUND; + err = AVERROR_FILTER_NOT_FOUND; + goto fail; } - + AVFilterContext *abuffer_ctx = avfilter_graph_alloc_filter(filter_graph, abuffer, NULL); if (!abuffer_ctx) { fprintf(stderr, "Could not allocate the abuffer instance.\n"); - return AVERROR(ENOMEM); + err = AVERROR(ENOMEM); + goto fail; } - + #if LIBAVCODEC_VERSION_MAJOR < 60 av_get_channel_layout_string(ch_layout, sizeof(ch_layout), 0, AV_CH_LAYOUT_STEREO); #else @@ -1184,50 +1480,56 @@ static int init_filter_graph(AVCodecContext *audio_codec_context, AVFilterGraph av_opt_set_q (abuffer_ctx, "time_base", audio_codec_context->time_base, AV_OPT_SEARCH_CHILDREN); av_opt_set_int(abuffer_ctx, "sample_rate", audio_codec_context->sample_rate, AV_OPT_SEARCH_CHILDREN); av_opt_set_int(abuffer_ctx, "bit_rate", audio_codec_context->bit_rate, AV_OPT_SEARCH_CHILDREN); - + err = avfilter_init_str(abuffer_ctx, NULL); if (err < 0) { fprintf(stderr, "Could not initialize the abuffer filter.\n"); - return err; + goto fail; } src_filter_ctx.push_back(abuffer_ctx); } - const AVFilter *mix_filter = avfilter_get_by_name("amix"); + mix_filter = avfilter_get_by_name("amix"); if (!mix_filter) { av_log(NULL, AV_LOG_ERROR, "Could not find the mix filter.\n"); - return AVERROR_FILTER_NOT_FOUND; + err = AVERROR_FILTER_NOT_FOUND; + goto fail; } - - char args[512]; + +#if LIBAVFILTER_VERSION_INT >= AV_VERSION_INT(7, 107, 100) + snprintf(args, sizeof(args), "inputs=%d:normalize=%s", (int)num_sources, normalize ? "true" : "false"); +#else snprintf(args, sizeof(args), "inputs=%d", (int)num_sources); - - AVFilterContext *mix_ctx; + fprintf(stderr, "Warning: your ffmpeg version doesn't support disabling normalizing of mixed audio. Volume might be lower than expected\n"); +#endif + err = avfilter_graph_create_filter(&mix_ctx, mix_filter, "amix", args, NULL, filter_graph); if (err < 0) { av_log(NULL, AV_LOG_ERROR, "Cannot create audio amix filter\n"); - return err; + goto fail; } - - const AVFilter *abuffersink = avfilter_get_by_name("abuffersink"); + + abuffersink = avfilter_get_by_name("abuffersink"); if (!abuffersink) { fprintf(stderr, "Could not find the abuffersink filter.\n"); - return AVERROR_FILTER_NOT_FOUND; + err = AVERROR_FILTER_NOT_FOUND; + goto fail; } - - AVFilterContext *abuffersink_ctx = avfilter_graph_alloc_filter(filter_graph, abuffersink, "sink"); + + abuffersink_ctx = avfilter_graph_alloc_filter(filter_graph, abuffersink, "sink"); if (!abuffersink_ctx) { fprintf(stderr, "Could not allocate the abuffersink instance.\n"); - return AVERROR(ENOMEM); + err = AVERROR(ENOMEM); + goto fail; } - + err = avfilter_init_str(abuffersink_ctx, NULL); if (err < 0) { fprintf(stderr, "Could not initialize the abuffersink instance.\n"); - return err; + goto fail; } - + err = 0; for(size_t i = 0; i < src_filter_ctx.size(); ++i) { AVFilterContext *src_ctx = src_filter_ctx[i]; @@ -1238,19 +1540,90 @@ static int init_filter_graph(AVCodecContext *audio_codec_context, AVFilterGraph err = avfilter_link(mix_ctx, 0, abuffersink_ctx, 0); if (err < 0) { av_log(NULL, AV_LOG_ERROR, "Error connecting filters\n"); - return err; + goto fail; } - + err = avfilter_graph_config(filter_graph, NULL); if (err < 0) { av_log(NULL, AV_LOG_ERROR, "Error configuring the filter graph\n"); - return err; + goto fail; } - + *graph = filter_graph; - *sink = abuffersink_ctx; - + *sink = abuffersink_ctx; + return 0; + +fail: + avfilter_graph_free(&filter_graph); + src_filter_ctx.clear(); // possibly unnecessary? + return err; +} + +static gsr_video_encoder* create_video_encoder(gsr_egl *egl, const args_parser &arg_parser) { + const gsr_color_depth color_depth = video_codec_to_bit_depth(arg_parser.video_codec); + gsr_video_encoder *video_encoder = nullptr; + + if(arg_parser.video_encoder == GSR_VIDEO_ENCODER_HW_CPU) { + gsr_video_encoder_software_params params; + params.egl = egl; + params.color_depth = color_depth; + video_encoder = gsr_video_encoder_software_create(¶ms); + return video_encoder; + } + + if(video_codec_is_vulkan(arg_parser.video_codec)) { + gsr_video_encoder_vulkan_params params; + params.egl = egl; + params.color_depth = color_depth; + video_encoder = gsr_video_encoder_vulkan_create(¶ms); + return video_encoder; + } + + switch(egl->gpu_info.vendor) { + case GSR_GPU_VENDOR_AMD: + case GSR_GPU_VENDOR_INTEL: + case GSR_GPU_VENDOR_BROADCOM: { + gsr_video_encoder_vaapi_params params; + params.egl = egl; + params.color_depth = color_depth; + video_encoder = gsr_video_encoder_vaapi_create(¶ms); + break; + } + case GSR_GPU_VENDOR_NVIDIA: { + gsr_video_encoder_nvenc_params params; + params.egl = egl; + params.overclock = arg_parser.overclock; + params.color_depth = color_depth; + video_encoder = gsr_video_encoder_nvenc_create(¶ms); + break; + } + } + + return video_encoder; +} + +static bool get_supported_video_codecs(gsr_egl *egl, gsr_video_codec video_codec, bool use_software_video_encoder, bool cleanup, gsr_supported_video_codecs *video_codecs) { + memset(video_codecs, 0, sizeof(*video_codecs)); + + if(use_software_video_encoder) { + video_codecs->h264.supported = true; + return true; + } + + if(video_codec_is_vulkan(video_codec)) + return gsr_get_supported_video_codecs_vulkan(video_codecs, egl->card_path, cleanup); + + switch(egl->gpu_info.vendor) { + case GSR_GPU_VENDOR_AMD: + case GSR_GPU_VENDOR_INTEL: + case GSR_GPU_VENDOR_BROADCOM: + return gsr_get_supported_video_codecs_vaapi(video_codecs, egl->card_path, cleanup); + case GSR_GPU_VENDOR_NVIDIA: + return gsr_get_supported_video_codecs_nvenc(video_codecs, cleanup); + } + + return false; } static void xwayland_check_callback(const gsr_monitor *monitor, void *userdata) { @@ -1267,245 +1640,345 @@ static bool is_xwayland(Display *display) { return true; bool xwayland_found = false; - for_each_active_monitor_output(display, GSR_CONNECTION_X11, xwayland_check_callback, &xwayland_found); + for_each_active_monitor_output_x11_not_cached(display, xwayland_check_callback, &xwayland_found); return xwayland_found; } -struct ReceivePacketData { - AVCodecContext *codec_context; - int stream_index; - AVStream *stream; - int64_t pts; -}; +static bool is_using_prime_run() { + const char *prime_render_offload = getenv("__NV_PRIME_RENDER_OFFLOAD"); + return (prime_render_offload && strcmp(prime_render_offload, "1") == 0) || getenv("DRI_PRIME"); +} -int main(int argc, char **argv) { - signal(SIGINT, int_handler); - signal(SIGUSR1, save_replay_handler); +static void disable_prime_run() { + unsetenv("__NV_PRIME_RENDER_OFFLOAD"); + unsetenv("__NV_PRIME_RENDER_OFFLOAD_PROVIDER"); + unsetenv("__GLX_VENDOR_LIBRARY_NAME"); + unsetenv("__VK_LAYER_NV_optimus"); + unsetenv("DRI_PRIME"); +} - if(argc <= 1) - usage_full(); +static gsr_window* gsr_window_create(Display *display, bool wayland) { + if(wayland) + return gsr_window_wayland_create(); + else + return gsr_window_x11_create(display); +} + +static void list_system_info(bool wayland) { + printf("display_server|%s\n", wayland ? "wayland" : "x11"); + bool supports_app_audio = false; +#ifdef GSR_APP_AUDIO + supports_app_audio = pulseaudio_server_is_pipewire(); + if(supports_app_audio) { + gsr_pipewire_audio audio; + if(gsr_pipewire_audio_init(&audio)) + gsr_pipewire_audio_deinit(&audio); + else + supports_app_audio = false; + } +#endif + printf("supports_app_audio|%s\n", supports_app_audio ? "yes" : "no"); +} - if(argc == 2 && (strcmp(argv[1], "-h") == 0 || strcmp(argv[1], "--help") == 0)) - usage_full(); +static void list_gpu_info(gsr_egl *egl) { + switch(egl->gpu_info.vendor) { + case GSR_GPU_VENDOR_AMD: + printf("vendor|amd\n"); + break; + case GSR_GPU_VENDOR_INTEL: + printf("vendor|intel\n"); + break; + case GSR_GPU_VENDOR_NVIDIA: + printf("vendor|nvidia\n"); + break; + case GSR_GPU_VENDOR_BROADCOM: + printf("vendor|broadcom\n"); + break; + } + printf("card_path|%s\n", egl->card_path); +} - //av_log_set_level(AV_LOG_TRACE); +static const AVCodec* get_ffmpeg_video_codec(gsr_video_codec video_codec, gsr_gpu_vendor vendor) { + switch(video_codec) { + case GSR_VIDEO_CODEC_H264: + return avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "h264_nvenc" : "h264_vaapi"); + case GSR_VIDEO_CODEC_HEVC: + case GSR_VIDEO_CODEC_HEVC_HDR: + case GSR_VIDEO_CODEC_HEVC_10BIT: + return avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "hevc_nvenc" : "hevc_vaapi"); + case GSR_VIDEO_CODEC_AV1: + case GSR_VIDEO_CODEC_AV1_HDR: + case GSR_VIDEO_CODEC_AV1_10BIT: + return avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "av1_nvenc" : "av1_vaapi"); + case GSR_VIDEO_CODEC_VP8: + return avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "vp8_nvenc" : "vp8_vaapi"); + case GSR_VIDEO_CODEC_VP9: + return avcodec_find_encoder_by_name(vendor == GSR_GPU_VENDOR_NVIDIA ? "vp9_nvenc" : "vp9_vaapi"); + case GSR_VIDEO_CODEC_H264_VULKAN: + return avcodec_find_encoder_by_name("h264_vulkan"); + case GSR_VIDEO_CODEC_HEVC_VULKAN: + return avcodec_find_encoder_by_name("hevc_vulkan"); + } + return nullptr; +} - std::map<std::string, Arg> args = { - { "-w", Arg { {}, false, false } }, - { "-c", Arg { {}, true, false } }, - { "-f", Arg { {}, false, false } }, - { "-s", Arg { {}, true, false } }, - { "-a", Arg { {}, true, true } }, - { "-q", Arg { {}, true, false } }, - { "-o", Arg { {}, true, false } }, - { "-r", Arg { {}, true, false } }, - { "-k", Arg { {}, true, false } }, - { "-ac", Arg { {}, true, false } }, - { "-oc", Arg { {}, true, false } }, - { "-fm", Arg { {}, true, false } }, - { "-pixfmt", Arg { {}, true, false } }, - { "-v", Arg { {}, true, false } }, - { "-mf", Arg { {}, true, false } }, - }; - - for(int i = 1; i < argc; i += 2) { - auto it = args.find(argv[i]); - if(it == args.end()) { - fprintf(stderr, "Invalid argument '%s'\n", argv[i]); - usage(); - } - - if(!it->second.values.empty() && !it->second.list) { - fprintf(stderr, "Expected argument '%s' to only be specified once\n", argv[i]); - usage(); - } - - if(i + 1 >= argc) { - fprintf(stderr, "Missing value for argument '%s'\n", argv[i]); - usage(); - } - - it->second.values.push_back(argv[i + 1]); - } - - for(auto &it : args) { - if(!it.second.optional && !it.second.value()) { - fprintf(stderr, "Missing argument '%s'\n", it.first.c_str()); - usage(); - } - } - - VideoCodec video_codec = VideoCodec::H265; - const char *video_codec_to_use = args["-k"].value(); - if(!video_codec_to_use) - video_codec_to_use = "auto"; - - if(strcmp(video_codec_to_use, "h264") == 0) { - video_codec = VideoCodec::H264; - } else if(strcmp(video_codec_to_use, "h265") == 0) { - video_codec = VideoCodec::H265; - } else if(strcmp(video_codec_to_use, "av1") == 0) { - video_codec = VideoCodec::AV1; - } else if(strcmp(video_codec_to_use, "auto") != 0) { - fprintf(stderr, "Error: -k should either be either 'auto', 'h264', 'h265' or 'av1', got: '%s'\n", video_codec_to_use); - usage(); - } - - AudioCodec audio_codec = AudioCodec::OPUS; - const char *audio_codec_to_use = args["-ac"].value(); - if(!audio_codec_to_use) - audio_codec_to_use = "aac"; - - if(strcmp(audio_codec_to_use, "aac") == 0) { - audio_codec = AudioCodec::AAC; - } else if(strcmp(audio_codec_to_use, "opus") == 0) { - audio_codec = AudioCodec::OPUS; - } else if(strcmp(audio_codec_to_use, "flac") == 0) { - audio_codec = AudioCodec::FLAC; - } else { - fprintf(stderr, "Error: -ac should either be either 'aac', 'opus' or 'flac', got: '%s'\n", audio_codec_to_use); - usage(); +static void set_supported_video_codecs_ffmpeg(gsr_supported_video_codecs *supported_video_codecs, gsr_supported_video_codecs *supported_video_codecs_vulkan, gsr_gpu_vendor vendor) { + if(!get_ffmpeg_video_codec(GSR_VIDEO_CODEC_H264, vendor)) { + supported_video_codecs->h264.supported = false; } - if(audio_codec != AudioCodec::AAC) { - audio_codec_to_use = "aac"; - audio_codec = AudioCodec::AAC; - fprintf(stderr, "Info: audio codec is forcefully set to aac at the moment because of issues with opus/flac. This is a temporary issue\n"); + if(!get_ffmpeg_video_codec(GSR_VIDEO_CODEC_HEVC, vendor)) { + supported_video_codecs->hevc.supported = false; + supported_video_codecs->hevc_hdr.supported = false; + supported_video_codecs->hevc_10bit.supported = false; } - bool overclock = false; - const char *overclock_str = args["-oc"].value(); - if(!overclock_str) - overclock_str = "no"; + if(!get_ffmpeg_video_codec(GSR_VIDEO_CODEC_AV1, vendor)) { + supported_video_codecs->av1.supported = false; + supported_video_codecs->av1_hdr.supported = false; + supported_video_codecs->av1_10bit.supported = false; + } - if(strcmp(overclock_str, "yes") == 0) { - overclock = true; - } else if(strcmp(overclock_str, "no") == 0) { - overclock = false; - } else { - fprintf(stderr, "Error: -oc should either be either 'yes' or 'no', got: '%s'\n", overclock_str); - usage(); + if(!get_ffmpeg_video_codec(GSR_VIDEO_CODEC_VP8, vendor)) { + supported_video_codecs->vp8.supported = false; } - bool verbose = true; - const char *verbose_str = args["-v"].value(); - if(!verbose_str) - verbose_str = "yes"; + if(!get_ffmpeg_video_codec(GSR_VIDEO_CODEC_VP9, vendor)) { + supported_video_codecs->vp9.supported = false; + } - if(strcmp(verbose_str, "yes") == 0) { - verbose = true; - } else if(strcmp(verbose_str, "no") == 0) { - verbose = false; - } else { - fprintf(stderr, "Error: -v should either be either 'yes' or 'no', got: '%s'\n", verbose_str); - usage(); + if(!get_ffmpeg_video_codec(GSR_VIDEO_CODEC_H264_VULKAN, vendor)) { + supported_video_codecs_vulkan->h264.supported = false; } - bool make_folders = false; - const char *make_folders_str = args["-mf"].value(); - if(!make_folders_str) - make_folders_str = "no"; + if(!get_ffmpeg_video_codec(GSR_VIDEO_CODEC_HEVC_VULKAN, vendor)) { + supported_video_codecs_vulkan->hevc.supported = false; + supported_video_codecs_vulkan->hevc_hdr.supported = false; + supported_video_codecs_vulkan->hevc_10bit.supported = false; + } +} - if(strcmp(make_folders_str, "yes") == 0) { - make_folders = true; - } else if(strcmp(make_folders_str, "no") == 0) { - make_folders = false; +static void list_supported_video_codecs(gsr_egl *egl, bool wayland) { + // Dont clean it up on purpose to increase shutdown speed + gsr_supported_video_codecs supported_video_codecs; + get_supported_video_codecs(egl, GSR_VIDEO_CODEC_H264, false, false, &supported_video_codecs); + + gsr_supported_video_codecs supported_video_codecs_vulkan; + get_supported_video_codecs(egl, GSR_VIDEO_CODEC_H264_VULKAN, false, false, &supported_video_codecs_vulkan); + + set_supported_video_codecs_ffmpeg(&supported_video_codecs, &supported_video_codecs_vulkan, egl->gpu_info.vendor); + + if(supported_video_codecs.h264.supported) + puts("h264"); + if(avcodec_find_encoder_by_name("libx264")) + puts("h264_software"); + if(supported_video_codecs.hevc.supported) + puts("hevc"); + if(supported_video_codecs.hevc_hdr.supported && wayland) + puts("hevc_hdr"); + if(supported_video_codecs.hevc_10bit.supported) + puts("hevc_10bit"); + if(supported_video_codecs.av1.supported) + puts("av1"); + if(supported_video_codecs.av1_hdr.supported && wayland) + puts("av1_hdr"); + if(supported_video_codecs.av1_10bit.supported) + puts("av1_10bit"); + if(supported_video_codecs.vp8.supported) + puts("vp8"); + if(supported_video_codecs.vp9.supported) + puts("vp9"); + //if(supported_video_codecs_vulkan.h264.supported) + // puts("h264_vulkan"); + //if(supported_video_codecs_vulkan.hevc.supported) + // puts("hevc_vulkan"); // TODO: hdr, 10 bit +} + +static bool monitor_capture_use_drm(const gsr_window *window, gsr_gpu_vendor vendor) { + return gsr_window_get_display_server(window) == GSR_DISPLAY_SERVER_WAYLAND || vendor != GSR_GPU_VENDOR_NVIDIA; +} + +typedef struct { + const gsr_window *window; + int num_monitors; +} capture_options_callback; + +static void output_monitor_info(const gsr_monitor *monitor, void *userdata) { + capture_options_callback *options = (capture_options_callback*)userdata; + if(gsr_window_get_display_server(options->window) == GSR_DISPLAY_SERVER_WAYLAND) { + vec2i monitor_size = monitor->size; + gsr_monitor_rotation monitor_rotation = GSR_MONITOR_ROT_0; + vec2i monitor_position = {0, 0}; + drm_monitor_get_display_server_data(options->window, monitor, &monitor_rotation, &monitor_position); + if(monitor_rotation == GSR_MONITOR_ROT_90 || monitor_rotation == GSR_MONITOR_ROT_270) + std::swap(monitor_size.x, monitor_size.y); + printf("%.*s|%dx%d\n", monitor->name_len, monitor->name, monitor_size.x, monitor_size.y); } else { - fprintf(stderr, "Error: -mf should either be either 'yes' or 'no', got: '%s'\n", make_folders_str); - usage(); + printf("%.*s|%dx%d\n", monitor->name_len, monitor->name, monitor->size.x, monitor->size.y); } + ++options->num_monitors; +} - PixelFormat pixel_format = PixelFormat::YUV420; - const char *pixfmt = args["-pixfmt"].value(); - if(!pixfmt) - pixfmt = "yuv420"; +static void list_supported_capture_options(const gsr_window *window, const char *card_path, bool list_monitors) { + const bool wayland = gsr_window_get_display_server(window) == GSR_DISPLAY_SERVER_WAYLAND; + if(!wayland) { + puts("window"); + puts("focused"); + } - if(strcmp(pixfmt, "yuv420") == 0) { - pixel_format = PixelFormat::YUV420; - } else if(strcmp(pixfmt, "yuv444") == 0) { - pixel_format = PixelFormat::YUV444; - } else { - fprintf(stderr, "Error: -pixfmt should either be either 'yuv420', or 'yuv444', got: '%s'\n", pixfmt); - usage(); + capture_options_callback options; + options.window = window; + options.num_monitors = 0; + if(list_monitors) { + const bool is_x11 = gsr_window_get_display_server(window) == GSR_DISPLAY_SERVER_X11; + const gsr_connection_type connection_type = is_x11 ? GSR_CONNECTION_X11 : GSR_CONNECTION_DRM; + for_each_active_monitor_output(window, card_path, connection_type, output_monitor_info, &options); } - const Arg &audio_input_arg = args["-a"]; - std::vector<AudioInput> audio_inputs; - if(!audio_input_arg.values.empty()) - audio_inputs = get_pulseaudio_inputs(); - std::vector<MergedAudioInputs> requested_audio_inputs; + if(options.num_monitors > 0) + puts("region"); - // Manually check if the audio inputs we give exist. This is only needed for pipewire, not pulseaudio. - // Pipewire instead DEFAULTS TO THE DEFAULT AUDIO INPUT. THAT'S RETARDED. - // OH, YOU MISSPELLED THE AUDIO INPUT? FUCK YOU - for(const char *audio_input : audio_input_arg.values) { - if(!audio_input || audio_input[0] == '\0') - continue; +#ifdef GSR_PORTAL + // Desktop portal capture on x11 doesn't seem to be hardware accelerated + if(!wayland) + return; - requested_audio_inputs.push_back({parse_audio_input_arg(audio_input)}); - for(AudioInput &request_audio_input : requested_audio_inputs.back().audio_inputs) { - bool match = false; - for(const auto &existing_audio_input : audio_inputs) { - if(strcmp(request_audio_input.name.c_str(), existing_audio_input.name.c_str()) == 0) { - if(request_audio_input.description.empty()) - request_audio_input.description = "gsr-" + existing_audio_input.description; + gsr_dbus_client dbus_client; + if(!gsr_dbus_client_init(&dbus_client, NULL)) + return; - match = true; - break; - } - } + char session_handle[128]; + if(gsr_dbus_client_screencast_create_session(&dbus_client, session_handle, sizeof(session_handle)) == 0) + puts("portal"); - if(!match) { - fprintf(stderr, "Error: Audio input device '%s' is not a valid audio device, expected one of:\n", request_audio_input.name.c_str()); - for(const auto &existing_audio_input : audio_inputs) { - fprintf(stderr, " %s\n", existing_audio_input.name.c_str()); - } - _exit(2); - } - } + gsr_dbus_client_deinit(&dbus_client); +#endif +} + +static void version_command(void *userdata) { + (void)userdata; + puts(GSR_VERSION); + fflush(stdout); + _exit(0); +} + +static void info_command(void *userdata) { + (void)userdata; + bool wayland = false; + Display *dpy = XOpenDisplay(nullptr); + if (!dpy) { + wayland = true; + fprintf(stderr, "Warning: failed to connect to the X server. Assuming wayland is running without Xwayland\n"); } - const char *container_format = args["-c"].value(); - if(container_format && strcmp(container_format, "mkv") == 0) - container_format = "matroska"; + XSetErrorHandler(x11_error_handler); + XSetIOErrorHandler(x11_io_error_handler); - int fps = atoi(args["-f"].value()); - if(fps == 0) { - fprintf(stderr, "Invalid fps argument: %s\n", args["-f"].value()); + if(!wayland) + wayland = is_xwayland(dpy); + + if(!wayland && is_using_prime_run()) { + // Disable prime-run and similar options as it doesn't work, the monitor to capture has to be run on the same device. + // This is fine on wayland since nvidia uses drm interface there and the monitor query checks the monitors connected + // to the drm device. + fprintf(stderr, "Warning: use of prime-run on X11 is not supported. Disabling prime-run\n"); + disable_prime_run(); + } + + gsr_window *window = gsr_window_create(dpy, wayland); + if(!window) { + fprintf(stderr, "Error: failed to create window\n"); _exit(1); } - if(fps < 1) - fps = 1; - - const char *quality_str = args["-q"].value(); - if(!quality_str) - quality_str = "very_high"; - - VideoQuality quality; - if(strcmp(quality_str, "medium") == 0) { - quality = VideoQuality::MEDIUM; - } else if(strcmp(quality_str, "high") == 0) { - quality = VideoQuality::HIGH; - } else if(strcmp(quality_str, "very_high") == 0) { - quality = VideoQuality::VERY_HIGH; - } else if(strcmp(quality_str, "ultra") == 0) { - quality = VideoQuality::ULTRA; - } else { - fprintf(stderr, "Error: -q should either be either 'medium', 'high', 'very_high' or 'ultra', got: '%s'\n", quality_str); - usage(); + + gsr_egl egl; + if(!gsr_egl_load(&egl, window, false, false)) { + fprintf(stderr, "gsr error: failed to load opengl\n"); + _exit(22); } - int replay_buffer_size_secs = -1; - const char *replay_buffer_size_secs_str = args["-r"].value(); - if(replay_buffer_size_secs_str) { - replay_buffer_size_secs = atoi(replay_buffer_size_secs_str); - if(replay_buffer_size_secs < 5 || replay_buffer_size_secs > 1200) { - fprintf(stderr, "Error: option -r has to be between 5 and 1200, was: %s\n", replay_buffer_size_secs_str); - _exit(1); + bool list_monitors = true; + egl.card_path[0] = '\0'; + if(monitor_capture_use_drm(window, egl.gpu_info.vendor)) { + // TODO: Allow specifying another card, and in other places + if(!gsr_get_valid_card_path(&egl, egl.card_path, true)) { + fprintf(stderr, "Error: no /dev/dri/cardX device found. Make sure that you have at least one monitor connected\n"); + list_monitors = false; + } + } + + av_log_set_level(AV_LOG_FATAL); + + puts("section=system_info"); + list_system_info(wayland); + if(egl.gpu_info.is_steam_deck) + puts("is_steam_deck|yes"); + else + puts("is_steam_deck|no"); + printf("gsr_version|%s\n", GSR_VERSION); + puts("section=gpu_info"); + list_gpu_info(&egl); + puts("section=video_codecs"); + list_supported_video_codecs(&egl, wayland); + puts("section=image_formats"); + puts("jpeg"); + puts("png"); + puts("section=capture_options"); + list_supported_capture_options(window, egl.card_path, list_monitors); + + fflush(stdout); + + // Not needed as this will just slow down shutdown + //gsr_egl_unload(&egl); + //gsr_window_destroy(&window); + //if(dpy) + // XCloseDisplay(dpy); + + _exit(0); +} + +static void list_audio_devices_command(void *userdata) { + (void)userdata; + const AudioDevices audio_devices = get_pulseaudio_inputs(); + + if(!audio_devices.default_output.empty()) + puts("default_output|Default output"); + + if(!audio_devices.default_input.empty()) + puts("default_input|Default input"); + + for(const auto &audio_input : audio_devices.audio_inputs) { + printf("%s|%s\n", audio_input.name.c_str(), audio_input.description.c_str()); + } + + fflush(stdout); + _exit(0); +} + +static bool app_audio_query_callback(const char *app_name, void*) { + puts(app_name); + return true; +} + +static void list_application_audio_command(void *userdata) { + (void)userdata; +#ifdef GSR_APP_AUDIO + if(pulseaudio_server_is_pipewire()) { + gsr_pipewire_audio audio; + if(gsr_pipewire_audio_init(&audio)) { + gsr_pipewire_audio_for_each_app(&audio, app_audio_query_callback, NULL); + gsr_pipewire_audio_deinit(&audio); } - replay_buffer_size_secs += 3; // Add a few seconds to account of lost packets because of non-keyframe packets skipped } +#endif + fflush(stdout); + _exit(0); +} + +// |card_path| can be NULL. If not NULL then |vendor| has to be valid +static void list_capture_options_command(const char *card_path, void *userdata) { + (void)userdata; bool wayland = false; Display *dpy = XOpenDisplay(nullptr); if (!dpy) { @@ -1519,468 +1992,1253 @@ int main(int argc, char **argv) { if(!wayland) wayland = is_xwayland(dpy); - gsr_egl egl; - if(!gsr_egl_load(&egl, dpy, wayland)) { - fprintf(stderr, "gsr error: failed to load opengl\n"); + if(!wayland && is_using_prime_run()) { + // Disable prime-run and similar options as it doesn't work, the monitor to capture has to be run on the same device. + // This is fine on wayland since nvidia uses drm interface there and the monitor query checks the monitors connected + // to the drm device. + fprintf(stderr, "Warning: use of prime-run on X11 is not supported. Disabling prime-run\n"); + disable_prime_run(); + } + + gsr_window *window = gsr_window_create(dpy, wayland); + if(!window) { + fprintf(stderr, "Error: failed to create window\n"); _exit(1); } - gsr_gpu_info gpu_inf; - bool very_old_gpu = false; - if(!gl_get_gpu_info(&egl, &gpu_inf)) - _exit(2); + if(card_path) { + list_supported_capture_options(window, card_path, true); + } else { + gsr_egl egl; + if(!gsr_egl_load(&egl, window, false, false)) { + fprintf(stderr, "gsr error: failed to load opengl\n"); + _exit(1); + } - if(gpu_inf.vendor == GSR_GPU_VENDOR_NVIDIA && gpu_inf.gpu_version != 0 && gpu_inf.gpu_version < 900) { - fprintf(stderr, "Info: your gpu appears to be very old (older than maxwell architecture). Switching to lower preset\n"); - very_old_gpu = true; + bool list_monitors = true; + egl.card_path[0] = '\0'; + if(monitor_capture_use_drm(window, egl.gpu_info.vendor)) { + // TODO: Allow specifying another card, and in other places + if(!gsr_get_valid_card_path(&egl, egl.card_path, true)) { + fprintf(stderr, "Error: no /dev/dri/cardX device found. Make sure that you have at least one monitor connected\n"); + list_monitors = false; + } + } + list_supported_capture_options(window, egl.card_path, list_monitors); } - if(gpu_inf.vendor != GSR_GPU_VENDOR_NVIDIA && overclock) { - fprintf(stderr, "Info: overclock option has no effect on amd/intel, ignoring option\n"); - } + fflush(stdout); - if(gpu_inf.vendor == GSR_GPU_VENDOR_NVIDIA && overclock && wayland) { - fprintf(stderr, "Info: overclocking is not possible on nvidia on wayland, ignoring option\n"); - } + // Not needed as this will just slow down shutdown + //gsr_egl_unload(&egl); + //gsr_window_destroy(&window); + //if(dpy) + // XCloseDisplay(dpy); - char card_path[128]; - card_path[0] = '\0'; - if(wayland || gpu_inf.vendor != GSR_GPU_VENDOR_NVIDIA) { - // TODO: Allow specifying another card, and in other places - if(!gsr_get_valid_card_path(card_path)) { - fprintf(stderr, "Error: no /dev/dri/cardX device found\n"); - _exit(2); + _exit(0); +} + +static std::string validate_monitor_get_valid(const gsr_egl *egl, const char* window) { + const bool is_x11 = gsr_window_get_display_server(egl->window) == GSR_DISPLAY_SERVER_X11; + const gsr_connection_type connection_type = is_x11 ? GSR_CONNECTION_X11 : GSR_CONNECTION_DRM; + const bool capture_use_drm = monitor_capture_use_drm(egl->window, egl->gpu_info.vendor); + + std::string window_result = window; + if(strcmp(window_result.c_str(), "screen") == 0) { + FirstOutputCallback data; + data.output_name = NULL; + for_each_active_monitor_output(egl->window, egl->card_path, connection_type, get_first_output_callback, &data); + + if(data.output_name) { + window_result = data.output_name; + free(data.output_name); + } else { + fprintf(stderr, "Error: no usable output found\n"); + _exit(51); + } + } else if(capture_use_drm || (strcmp(window_result.c_str(), "screen-direct") != 0 && strcmp(window_result.c_str(), "screen-direct-force") != 0)) { + gsr_monitor gmon; + if(!get_monitor_by_name(egl, connection_type, window_result.c_str(), &gmon)) { + fprintf(stderr, "gsr error: display \"%s\" not found, expected one of:\n", window_result.c_str()); + fprintf(stderr, " \"screen\"\n"); + if(!capture_use_drm) + fprintf(stderr, " \"screen-direct\"\n"); + + MonitorOutputCallbackUserdata userdata; + userdata.window = egl->window; + for_each_active_monitor_output(egl->window, egl->card_path, connection_type, monitor_output_callback_print, &userdata); + _exit(51); } } + return window_result; +} - // TODO: Fix constant framerate not working properly on amd/intel because capture framerate gets locked to the same framerate as - // game framerate, which doesn't work well when you need to encode multiple duplicate frames (AMD/Intel is slow at encoding!). - // It also appears to skip audio frames on nvidia wayland? why? that should be fine, but it causes video stuttering because of audio/video sync. - FramerateMode framerate_mode; - const char *framerate_mode_str = args["-fm"].value(); - if(!framerate_mode_str) - framerate_mode_str = (gpu_inf.vendor == GSR_GPU_VENDOR_NVIDIA && !wayland) ? "cfr" : "vfr"; +static std::string get_monitor_by_region_center(const gsr_egl *egl, vec2i region_position, vec2i region_size, vec2i *monitor_pos, vec2i *monitor_size) { + const bool is_x11 = gsr_window_get_display_server(egl->window) == GSR_DISPLAY_SERVER_X11; + const gsr_connection_type connection_type = is_x11 ? GSR_CONNECTION_X11 : GSR_CONNECTION_DRM; + + MonitorByPositionCallback data; + data.window = egl->window; + data.position = { region_position.x + region_size.x / 2, region_position.y + region_size.y / 2 }; + data.output_name = NULL; + data.monitor_pos = {0, 0}; + data.monitor_size = {0, 0}; + for_each_active_monitor_output(egl->window, egl->card_path, connection_type, get_monitor_by_position_callback, &data); + + std::string result; + if(data.output_name) { + result = data.output_name; + free(data.output_name); + } + *monitor_pos = data.monitor_pos; + *monitor_size = data.monitor_size; + return result; +} - if(strcmp(framerate_mode_str, "cfr") == 0) { - framerate_mode = FramerateMode::CONSTANT; - } else if(strcmp(framerate_mode_str, "vfr") == 0) { - framerate_mode = FramerateMode::VARIABLE; +static gsr_capture* create_monitor_capture(const args_parser &arg_parser, gsr_egl *egl, bool prefer_ximage) { + if(gsr_window_get_display_server(egl->window) == GSR_DISPLAY_SERVER_X11 && prefer_ximage) { + gsr_capture_ximage_params ximage_params; + ximage_params.egl = egl; + ximage_params.display_to_capture = arg_parser.window; + ximage_params.record_cursor = arg_parser.record_cursor; + ximage_params.output_resolution = arg_parser.output_resolution; + ximage_params.region_size = arg_parser.region_size; + ximage_params.region_position = arg_parser.region_position; + return gsr_capture_ximage_create(&ximage_params); + } + + if(monitor_capture_use_drm(egl->window, egl->gpu_info.vendor)) { + gsr_capture_kms_params kms_params; + kms_params.egl = egl; + kms_params.display_to_capture = arg_parser.window; + kms_params.record_cursor = arg_parser.record_cursor; + kms_params.hdr = video_codec_is_hdr(arg_parser.video_codec); + kms_params.fps = arg_parser.fps; + kms_params.output_resolution = arg_parser.output_resolution; + kms_params.region_size = arg_parser.region_size; + kms_params.region_position = arg_parser.region_position; + return gsr_capture_kms_create(&kms_params); } else { - fprintf(stderr, "Error: -fm should either be either 'cfr' or 'vfr', got: '%s'\n", framerate_mode_str); - usage(); - } + const char *capture_target = arg_parser.window; + const bool direct_capture = strcmp(arg_parser.window, "screen-direct") == 0 || strcmp(arg_parser.window, "screen-direct-force") == 0; + if(direct_capture) { + capture_target = "screen"; + fprintf(stderr, "Warning: %s capture option is not recommended unless you use G-SYNC as Nvidia has driver issues that can cause your system or games to freeze/crash.\n", arg_parser.window); + } - const char *screen_region = args["-s"].value(); - const char *window_str = strdup(args["-w"].value()); + gsr_capture_nvfbc_params nvfbc_params; + nvfbc_params.egl = egl; + nvfbc_params.display_to_capture = capture_target; + nvfbc_params.fps = arg_parser.fps; + nvfbc_params.direct_capture = direct_capture; + nvfbc_params.record_cursor = arg_parser.record_cursor; + nvfbc_params.output_resolution = arg_parser.output_resolution; + nvfbc_params.region_size = arg_parser.region_size; + nvfbc_params.region_position = arg_parser.region_position; + return gsr_capture_nvfbc_create(&nvfbc_params); + } +} - if(screen_region && strcmp(window_str, "focused") != 0) { - fprintf(stderr, "Error: option -s is only available when using -w focused\n"); - usage(); +static std::string region_get_data(gsr_egl *egl, vec2i *region_size, vec2i *region_position) { + vec2i monitor_pos = {0, 0}; + vec2i monitor_size = {0, 0}; + std::string window = get_monitor_by_region_center(egl, *region_position, *region_size, &monitor_pos, &monitor_size); + if(window.empty()) { + const bool is_x11 = gsr_window_get_display_server(egl->window) == GSR_DISPLAY_SERVER_X11; + const gsr_connection_type connection_type = is_x11 ? GSR_CONNECTION_X11 : GSR_CONNECTION_DRM; + fprintf(stderr, "Error: the region %dx%d+%d+%d doesn't match any monitor. Available monitors and their regions:\n", region_size->x, region_size->y, region_position->x, region_position->y); + + MonitorOutputCallbackUserdata userdata; + userdata.window = egl->window; + for_each_active_monitor_output(egl->window, egl->card_path, connection_type, monitor_output_callback_print, &userdata); + _exit(51); + } + + // Capture whole monitor when region size is set to 0x0 + if(region_size->x == 0 && region_size->y == 0) { + region_position->x = 0; + region_position->y = 0; + } else { + region_position->x -= monitor_pos.x; + region_position->y -= monitor_pos.y; } + return window; +} - vec2i region_size = { 0, 0 }; +static gsr_capture* create_capture_impl(args_parser &arg_parser, gsr_egl *egl, bool prefer_ximage) { Window src_window_id = None; bool follow_focused = false; + const bool wayland = gsr_window_get_display_server(egl->window) == GSR_DISPLAY_SERVER_WAYLAND; gsr_capture *capture = nullptr; - if(strcmp(window_str, "focused") == 0) { + if(strcmp(arg_parser.window, "focused") == 0) { if(wayland) { fprintf(stderr, "Error: GPU Screen Recorder window capture only works in a pure X11 session. Xwayland is not supported. You can record a monitor instead on wayland\n"); _exit(2); } - if(!screen_region) { - fprintf(stderr, "Error: option -s is required when using -w focused\n"); - usage(); - } - - if(sscanf(screen_region, "%dx%d", ®ion_size.x, ®ion_size.y) != 2) { - fprintf(stderr, "Error: invalid value for option -s '%s', expected a value in format WxH\n", screen_region); - usage(); - } - - if(region_size.x <= 0 || region_size.y <= 0) { - fprintf(stderr, "Error: invalud value for option -s '%s', expected width and height to be greater than 0\n", screen_region); - usage(); + if(arg_parser.output_resolution.x <= 0 || arg_parser.output_resolution.y <= 0) { + fprintf(stderr, "Error: invalid value for option -s '%dx%d' when using -w focused option. expected width and height to be greater than 0\n", arg_parser.output_resolution.x, arg_parser.output_resolution.y); + args_parser_print_usage(); + _exit(1); } follow_focused = true; - } else if(contains_non_hex_number(window_str)) { - if(wayland || gpu_inf.vendor != GSR_GPU_VENDOR_NVIDIA) { - if(strcmp(window_str, "screen") == 0) { - FirstOutputCallback first_output; - first_output.output_name = NULL; - if(gsr_egl_supports_wayland_capture(&egl)) { - for_each_active_monitor_output(&egl, GSR_CONNECTION_WAYLAND, get_first_output, &first_output); - } else { - for_each_active_monitor_output(card_path, GSR_CONNECTION_DRM, get_first_output, &first_output); - } - - if(first_output.output_name) { - window_str = first_output.output_name; - } else { - fprintf(stderr, "Error: no available output found\n"); - } - } - - if(gsr_egl_supports_wayland_capture(&egl)) { - gsr_monitor gmon; - if(!get_monitor_by_name(&egl, GSR_CONNECTION_WAYLAND, window_str, &gmon)) { - fprintf(stderr, "gsr error: display \"%s\" not found, expected one of:\n", window_str); - fprintf(stderr, " \"screen\"\n"); - for_each_active_monitor_output(&egl, GSR_CONNECTION_WAYLAND, monitor_output_callback_print, NULL); - _exit(1); - } - } else { - gsr_monitor gmon; - if(!get_monitor_by_name(card_path, GSR_CONNECTION_DRM, window_str, &gmon)) { - fprintf(stderr, "gsr error: display \"%s\" not found, expected one of:\n", window_str); - fprintf(stderr, " \"screen\"\n"); - for_each_active_monitor_output(card_path, GSR_CONNECTION_DRM, monitor_output_callback_print, NULL); - _exit(1); - } - } - } else { - if(strcmp(window_str, "screen") != 0 && strcmp(window_str, "screen-direct") != 0 && strcmp(window_str, "screen-direct-force") != 0) { - gsr_monitor gmon; - if(!get_monitor_by_name(dpy, GSR_CONNECTION_X11, window_str, &gmon)) { - fprintf(stderr, "gsr error: display \"%s\" not found, expected one of:\n", window_str); - fprintf(stderr, " \"screen\" (%dx%d+%d+%d)\n", XWidthOfScreen(DefaultScreenOfDisplay(dpy)), XHeightOfScreen(DefaultScreenOfDisplay(dpy)), 0, 0); - fprintf(stderr, " \"screen-direct\" (%dx%d+%d+%d)\n", XWidthOfScreen(DefaultScreenOfDisplay(dpy)), XHeightOfScreen(DefaultScreenOfDisplay(dpy)), 0, 0); - fprintf(stderr, " \"screen-direct-force\" (%dx%d+%d+%d)\n", XWidthOfScreen(DefaultScreenOfDisplay(dpy)), XHeightOfScreen(DefaultScreenOfDisplay(dpy)), 0, 0); - for_each_active_monitor_output(dpy, GSR_CONNECTION_X11, monitor_output_callback_print, NULL); - _exit(1); - } - } + } else if(strcmp(arg_parser.window, "portal") == 0) { +#ifdef GSR_PORTAL + // Desktop portal capture on x11 doesn't seem to be hardware accelerated + if(!wayland) { + fprintf(stderr, "Error: desktop portal capture is not supported on X11\n"); + _exit(1); } - if(gpu_inf.vendor == GSR_GPU_VENDOR_NVIDIA) { - if(wayland) { - gsr_capture_kms_cuda_params kms_params; - kms_params.egl = &egl; - kms_params.display_to_capture = window_str; - kms_params.gpu_inf = gpu_inf; - kms_params.card_path = card_path; - capture = gsr_capture_kms_cuda_create(&kms_params); - if(!capture) - _exit(1); - } else { - const char *capture_target = window_str; - bool direct_capture = strcmp(window_str, "screen-direct") == 0; - if(direct_capture) { - capture_target = "screen"; - // TODO: Temporary disable direct capture because push model causes stuttering when it's direct capturing. This might be a nvfbc bug. This does not happen when using a compositor. - direct_capture = false; - fprintf(stderr, "Warning: screen-direct has temporary been disabled as it causes stuttering. This is likely a NvFBC bug. Falling back to \"screen\".\n"); - } - - if(strcmp(window_str, "screen-direct-force") == 0) { - direct_capture = true; - capture_target = "screen"; - } - - gsr_egl_unload(&egl); - - gsr_capture_nvfbc_params nvfbc_params; - nvfbc_params.dpy = dpy; - nvfbc_params.display_to_capture = capture_target; - nvfbc_params.fps = fps; - nvfbc_params.pos = { 0, 0 }; - nvfbc_params.size = { 0, 0 }; - nvfbc_params.direct_capture = direct_capture; - nvfbc_params.overclock = overclock; - capture = gsr_capture_nvfbc_create(&nvfbc_params); - if(!capture) - _exit(1); - } - } else { - gsr_capture_kms_vaapi_params kms_params; - kms_params.egl = &egl; - kms_params.display_to_capture = window_str; - kms_params.gpu_inf = gpu_inf; - kms_params.card_path = card_path; - kms_params.wayland = wayland; - capture = gsr_capture_kms_vaapi_create(&kms_params); - if(!capture) - _exit(1); - } + gsr_capture_portal_params portal_params; + portal_params.egl = egl; + portal_params.record_cursor = arg_parser.record_cursor; + portal_params.restore_portal_session = arg_parser.restore_portal_session; + portal_params.portal_session_token_filepath = arg_parser.portal_session_token_filepath; + portal_params.output_resolution = arg_parser.output_resolution; + capture = gsr_capture_portal_create(&portal_params); + if(!capture) + _exit(1); +#else + fprintf(stderr, "Error: option '-w portal' used but GPU Screen Recorder was compiled without desktop portal support. Please recompile GPU Screen recorder with the -Dportal=true option\n"); + _exit(2); +#endif + } else if(strcmp(arg_parser.window, "region") == 0) { + const std::string window = region_get_data(egl, &arg_parser.region_size, &arg_parser.region_position); + snprintf(arg_parser.window, sizeof(arg_parser.window), "%s", window.c_str()); + capture = create_monitor_capture(arg_parser, egl, prefer_ximage); + if(!capture) + _exit(1); + } else if(contains_non_hex_number(arg_parser.window)) { + const std::string window = validate_monitor_get_valid(egl, arg_parser.window); + snprintf(arg_parser.window, sizeof(arg_parser.window), "%s", window.c_str()); + capture = create_monitor_capture(arg_parser, egl, prefer_ximage); + if(!capture) + _exit(1); } else { if(wayland) { - fprintf(stderr, "Error: GPU Screen Recorder window capture only works in a pure X11 session. Xwayland is not supported. You can record a monitor instead on wayland\n"); + fprintf(stderr, "Error: GPU Screen Recorder window capture only works in a pure X11 session. Xwayland is not supported. You can record a monitor instead on wayland or use -w portal option which supports window capture if your wayland compositor supports window capture\n"); _exit(2); } errno = 0; - src_window_id = strtol(window_str, nullptr, 0); + src_window_id = strtol(arg_parser.window, nullptr, 0); if(src_window_id == None || errno == EINVAL) { - fprintf(stderr, "Invalid window number %s\n", window_str); - usage(); + fprintf(stderr, "Error: invalid window number %s\n", arg_parser.window); + args_parser_print_usage(); + _exit(1); } } if(!capture) { - switch(gpu_inf.vendor) { - case GSR_GPU_VENDOR_AMD: { - gsr_capture_xcomposite_vaapi_params xcomposite_params; - xcomposite_params.egl = &egl; - xcomposite_params.dpy = dpy; - xcomposite_params.window = src_window_id; - xcomposite_params.follow_focused = follow_focused; - xcomposite_params.region_size = region_size; - xcomposite_params.card_path = card_path; - capture = gsr_capture_xcomposite_vaapi_create(&xcomposite_params); - if(!capture) - _exit(1); - break; - } - case GSR_GPU_VENDOR_INTEL: { - gsr_capture_xcomposite_vaapi_params xcomposite_params; - xcomposite_params.egl = &egl; - xcomposite_params.dpy = dpy; - xcomposite_params.window = src_window_id; - xcomposite_params.follow_focused = follow_focused; - xcomposite_params.region_size = region_size; - xcomposite_params.card_path = card_path; - capture = gsr_capture_xcomposite_vaapi_create(&xcomposite_params); - if(!capture) - _exit(1); + gsr_capture_xcomposite_params xcomposite_params; + xcomposite_params.egl = egl; + xcomposite_params.window = src_window_id; + xcomposite_params.follow_focused = follow_focused; + xcomposite_params.record_cursor = arg_parser.record_cursor; + xcomposite_params.output_resolution = arg_parser.output_resolution; + capture = gsr_capture_xcomposite_create(&xcomposite_params); + if(!capture) + _exit(1); + } + + return capture; +} + +static gsr_color_range image_format_to_color_range(gsr_image_format image_format) { + switch(image_format) { + case GSR_IMAGE_FORMAT_JPEG: return GSR_COLOR_RANGE_LIMITED; + case GSR_IMAGE_FORMAT_PNG: return GSR_COLOR_RANGE_FULL; + } + assert(false); + return GSR_COLOR_RANGE_FULL; +} + +static int video_quality_to_image_quality_value(gsr_video_quality video_quality) { + switch(video_quality) { + case GSR_VIDEO_QUALITY_MEDIUM: + return 75; + case GSR_VIDEO_QUALITY_HIGH: + return 85; + case GSR_VIDEO_QUALITY_VERY_HIGH: + return 90; + case GSR_VIDEO_QUALITY_ULTRA: + return 97; + } + assert(false); + return 90; +} + +// TODO: 10-bit and hdr. +static void capture_image_to_file(args_parser &arg_parser, gsr_egl *egl, gsr_image_format image_format) { + const gsr_color_range color_range = image_format_to_color_range(image_format); + const int fps = 60; + const bool prefer_ximage = true; + gsr_capture *capture = create_capture_impl(arg_parser, egl, prefer_ximage); + + gsr_capture_metadata capture_metadata; + capture_metadata.width = 0; + capture_metadata.height = 0; + capture_metadata.fps = fps; + capture_metadata.video_codec_context = nullptr; + capture_metadata.frame = nullptr; + + int capture_result = gsr_capture_start(capture, &capture_metadata); + if(capture_result != 0) { + fprintf(stderr, "gsr error: capture_image_to_file_wayland: gsr_capture_start failed\n"); + _exit(capture_result); + } + + gsr_image_writer image_writer; + if(!gsr_image_writer_init_opengl(&image_writer, egl, capture_metadata.width, capture_metadata.height)) { + fprintf(stderr, "gsr error: capture_image_to_file_wayland: gsr_image_write_gl_init failed\n"); + _exit(1); + } + + gsr_color_conversion_params color_conversion_params; + memset(&color_conversion_params, 0, sizeof(color_conversion_params)); + color_conversion_params.color_range = color_range; + color_conversion_params.egl = egl; + color_conversion_params.load_external_image_shader = gsr_capture_uses_external_image(capture); + + color_conversion_params.destination_textures[0] = image_writer.texture; + color_conversion_params.num_destination_textures = 1; + color_conversion_params.destination_color = GSR_DESTINATION_COLOR_RGB8; + + gsr_color_conversion color_conversion; + if(gsr_color_conversion_init(&color_conversion, &color_conversion_params) != 0) { + fprintf(stderr, "gsr error: capture_image_to_file_wayland: failed to create color conversion\n"); + _exit(1); + } + + gsr_color_conversion_clear(&color_conversion); + + bool should_stop_error = false; + egl->glClear(0); + + while(running) { + should_stop_error = false; + if(gsr_capture_should_stop(capture, &should_stop_error)) { + running = 0; + break; + } + + // It can fail, for example when capturing portal and the target is a monitor that hasn't been updated. + // Desktop portal wont refresh the image until there is an update. + // TODO: Find out if there is a way to force update desktop portal image. + // This can also happen for example if the system suspends and the monitor to capture's framebuffer is gone, or if the target window disappeared. + if(gsr_capture_capture(capture, &capture_metadata, &color_conversion) == 0) + break; + + usleep(30 * 1000); // 30 ms + } + + gsr_egl_swap_buffers(egl); + + const int image_quality = video_quality_to_image_quality_value(arg_parser.video_quality); + if(!gsr_image_writer_write_to_file(&image_writer, arg_parser.filename, image_format, image_quality)) { + fprintf(stderr, "gsr error: capture_image_to_file_wayland: failed to write opengl texture to image output file %s\n", arg_parser.filename); + _exit(1); + } + + gsr_image_writer_deinit(&image_writer); + gsr_capture_destroy(capture); + _exit(should_stop_error ? 3 : 0); +} + +static AVPixelFormat get_pixel_format(gsr_video_codec video_codec, gsr_gpu_vendor vendor, bool use_software_video_encoder) { + if(use_software_video_encoder) { + return AV_PIX_FMT_NV12; + } else { + if(video_codec_is_vulkan(video_codec)) + return AV_PIX_FMT_VULKAN; + else + return vendor == GSR_GPU_VENDOR_NVIDIA ? AV_PIX_FMT_CUDA : AV_PIX_FMT_VAAPI; + } +} + +static void match_app_audio_input_to_available_apps(const std::vector<AudioInput> &requested_audio_inputs, const std::vector<std::string> &app_audio_names) { + for(const AudioInput &request_audio_input : requested_audio_inputs) { + if(request_audio_input.type != AudioInputType::APPLICATION || request_audio_input.inverted) + continue; + + bool match = false; + for(const std::string &app_name : app_audio_names) { + if(strcasecmp(app_name.c_str(), request_audio_input.name.c_str()) == 0) { + match = true; break; } - case GSR_GPU_VENDOR_NVIDIA: { - gsr_capture_xcomposite_cuda_params xcomposite_params; - xcomposite_params.egl = &egl; - xcomposite_params.dpy = dpy; - xcomposite_params.window = src_window_id; - xcomposite_params.follow_focused = follow_focused; - xcomposite_params.region_size = region_size; - xcomposite_params.overclock = overclock; - capture = gsr_capture_xcomposite_cuda_create(&xcomposite_params); - if(!capture) - _exit(1); - break; + } + + if(!match) { + fprintf(stderr, "gsr warning: no audio application with the name \"%s\" was found, expected one of the following:\n", request_audio_input.name.c_str()); + for(const std::string &app_name : app_audio_names) { + fprintf(stderr, " * %s\n", app_name.c_str()); } + fprintf(stderr, " assuming this is intentional (if you are trying to record audio for applications that haven't started yet).\n"); } } +} + +// Manually check if the audio inputs we give exist. This is only needed for pipewire, not pulseaudio. +// Pipewire instead DEFAULTS TO THE DEFAULT AUDIO INPUT. THAT'S RETARDED. +// OH, YOU MISSPELLED THE AUDIO INPUT? FUCK YOU +static std::vector<MergedAudioInputs> parse_audio_inputs(const AudioDevices &audio_devices, const Arg *audio_input_arg) { + std::vector<MergedAudioInputs> requested_audio_inputs; + + for(int i = 0; i < audio_input_arg->num_values; ++i) { + const char *audio_input = audio_input_arg->values[i]; + if(!audio_input || audio_input[0] == '\0') + continue; + + requested_audio_inputs.push_back(parse_audio_input_arg(audio_input)); + for(AudioInput &request_audio_input : requested_audio_inputs.back().audio_inputs) { + if(request_audio_input.type != AudioInputType::DEVICE) + continue; + + bool match = false; - const char *filename = args["-o"].value(); - if(filename) { - if(replay_buffer_size_secs != -1) { - if(!container_format) { - fprintf(stderr, "Error: option -c is required when using option -r\n"); - usage(); + if(request_audio_input.name == "default_output") { + if(audio_devices.default_output.empty()) { + fprintf(stderr, "Error: -a default_output was specified but no default audio output is specified in the audio server\n"); + _exit(2); + } + match = true; + } else if(request_audio_input.name == "default_input") { + if(audio_devices.default_input.empty()) { + fprintf(stderr, "Error: -a default_input was specified but no default audio input is specified in the audio server\n"); + _exit(2); + } + match = true; + } else { + const bool name_is_existing_audio_device = get_audio_device_by_name(audio_devices.audio_inputs, request_audio_input.name.c_str()) != nullptr; + if(name_is_existing_audio_device) + match = true; } - struct stat buf; - if(stat(filename, &buf) != -1 && !S_ISDIR(buf.st_mode)) { - fprintf(stderr, "Error: File \"%s\" exists but it's not a directory\n", filename); - usage(); + if(!match) { + fprintf(stderr, "Error: Audio device '%s' is not a valid audio device, expected one of:\n", request_audio_input.name.c_str()); + if(!audio_devices.default_output.empty()) + fprintf(stderr, " default_output (Default output)\n"); + if(!audio_devices.default_input.empty()) + fprintf(stderr, " default_input (Default input)\n"); + for(const auto &audio_device_input : audio_devices.audio_inputs) { + fprintf(stderr, " %s (%s)\n", audio_device_input.name.c_str(), audio_device_input.description.c_str()); + } + _exit(50); } } - } else { - if(replay_buffer_size_secs == -1) { - filename = "/dev/stdout"; - } else { - fprintf(stderr, "Error: Option -o is required when using option -r\n"); - usage(); - } + } - if(!container_format) { - fprintf(stderr, "Error: option -c is required when not using option -o\n"); - usage(); - } + return requested_audio_inputs; +} + +static bool audio_inputs_has_app_audio(const std::vector<AudioInput> &audio_inputs) { + for(const auto &audio_input : audio_inputs) { + if(audio_input.type == AudioInputType::APPLICATION) + return true; } + return false; +} - AVFormatContext *av_format_context; - // The output format is automatically guessed by the file extension - avformat_alloc_output_context2(&av_format_context, nullptr, container_format, filename); - if (!av_format_context) { - if(container_format) - fprintf(stderr, "Error: Container format '%s' (argument -c) is not valid\n", container_format); - else - fprintf(stderr, "Error: Failed to deduce container format from file extension\n"); - _exit(1); +static bool merged_audio_inputs_has_app_audio(const std::vector<MergedAudioInputs> &merged_audio_inputs) { + for(const auto &merged_audio_input : merged_audio_inputs) { + if(audio_inputs_has_app_audio(merged_audio_input.audio_inputs)) + return true; } + return false; +} - const AVOutputFormat *output_format = av_format_context->oformat; +// Should use amix if more than 1 audio device and 0 application audio, merged +static bool audio_inputs_should_use_amix(const std::vector<AudioInput> &audio_inputs) { + int num_audio_devices = 0; + int num_app_audio = 0; - std::string file_extension = output_format->extensions; - { - size_t comma_index = file_extension.find(','); - if(comma_index != std::string::npos) - file_extension = file_extension.substr(0, comma_index); + for(const auto &audio_input : audio_inputs) { + if(audio_input.type == AudioInputType::DEVICE) + ++num_audio_devices; + else if(audio_input.type == AudioInputType::APPLICATION) + ++num_app_audio; + } + + return num_audio_devices > 1 && num_app_audio == 0; +} + +static bool merged_audio_inputs_should_use_amix(const std::vector<MergedAudioInputs> &merged_audio_inputs) { + for(const auto &merged_audio_input : merged_audio_inputs) { + if(audio_inputs_should_use_amix(merged_audio_input.audio_inputs)) + return true; } + return false; +} +static void validate_merged_audio_inputs_app_audio(const std::vector<MergedAudioInputs> &merged_audio_inputs, const std::vector<std::string> &app_audio_names) { + for(const auto &merged_audio_input : merged_audio_inputs) { + int num_app_audio = 0; + int num_app_inverted_audio = 0; + + for(const auto &audio_input : merged_audio_input.audio_inputs) { + if(audio_input.type == AudioInputType::APPLICATION) { + if(audio_input.inverted) + ++num_app_inverted_audio; + else + ++num_app_audio; + } + } + + match_app_audio_input_to_available_apps(merged_audio_input.audio_inputs, app_audio_names); + + if(num_app_audio > 0 && num_app_inverted_audio > 0) { + fprintf(stderr, "gsr error: argument -a was provided with both app: and app-inverse:, only one of them can be used for one audio track\n"); + _exit(2); + } + } +} + +static gsr_audio_codec select_audio_codec_with_fallback(gsr_audio_codec audio_codec, const std::string &file_extension, bool uses_amix) { switch(audio_codec) { - case AudioCodec::AAC: { + case GSR_AUDIO_CODEC_AAC: { + if(file_extension == "webm") { + //audio_codec_to_use = "opus"; + audio_codec = GSR_AUDIO_CODEC_OPUS; + fprintf(stderr, "Warning: .webm files only support opus audio codec, changing audio codec from aac to opus\n"); + } break; } - case AudioCodec::OPUS: { - if(file_extension != "mp4" && file_extension != "mkv") { - audio_codec_to_use = "aac"; - audio_codec = AudioCodec::AAC; - fprintf(stderr, "Warning: opus audio codec is only supported by .mp4 and .mkv files, falling back to aac instead\n"); + case GSR_AUDIO_CODEC_OPUS: { + // TODO: Also check mpegts? + if(file_extension != "mp4" && file_extension != "mkv" && file_extension != "webm") { + //audio_codec_to_use = "aac"; + audio_codec = GSR_AUDIO_CODEC_AAC; + fprintf(stderr, "Warning: opus audio codec is only supported by .mp4, .mkv and .webm files, falling back to aac instead\n"); } break; } - case AudioCodec::FLAC: { - if(file_extension != "mp4" && file_extension != "mkv") { - audio_codec_to_use = "aac"; - audio_codec = AudioCodec::AAC; + case GSR_AUDIO_CODEC_FLAC: { + // TODO: Also check mpegts? + if(file_extension == "webm") { + //audio_codec_to_use = "opus"; + audio_codec = GSR_AUDIO_CODEC_OPUS; + fprintf(stderr, "Warning: .webm files only support opus audio codec, changing audio codec from flac to opus\n"); + } else if(file_extension != "mp4" && file_extension != "mkv") { + //audio_codec_to_use = "aac"; + audio_codec = GSR_AUDIO_CODEC_AAC; fprintf(stderr, "Warning: flac audio codec is only supported by .mp4 and .mkv files, falling back to aac instead\n"); + } else if(uses_amix) { + // TODO: remove this? is it true anymore? + //audio_codec_to_use = "opus"; + audio_codec = GSR_AUDIO_CODEC_OPUS; + fprintf(stderr, "Warning: flac audio codec is not supported when mixing audio sources, falling back to opus instead\n"); } break; } } + return audio_codec; +} - const double target_fps = 1.0 / (double)fps; +static bool video_codec_only_supports_low_power_mode(const gsr_supported_video_codecs &supported_video_codecs, gsr_video_codec video_codec) { + switch(video_codec) { + case GSR_VIDEO_CODEC_H264: return supported_video_codecs.h264.low_power; + case GSR_VIDEO_CODEC_HEVC: return supported_video_codecs.hevc.low_power; + case GSR_VIDEO_CODEC_HEVC_HDR: return supported_video_codecs.hevc_hdr.low_power; + case GSR_VIDEO_CODEC_HEVC_10BIT: return supported_video_codecs.hevc_10bit.low_power; + case GSR_VIDEO_CODEC_AV1: return supported_video_codecs.av1.low_power; + case GSR_VIDEO_CODEC_AV1_HDR: return supported_video_codecs.av1_hdr.low_power; + case GSR_VIDEO_CODEC_AV1_10BIT: return supported_video_codecs.av1_10bit.low_power; + case GSR_VIDEO_CODEC_VP8: return supported_video_codecs.vp8.low_power; + case GSR_VIDEO_CODEC_VP9: return supported_video_codecs.vp9.low_power; + case GSR_VIDEO_CODEC_H264_VULKAN: return supported_video_codecs.h264.low_power; + case GSR_VIDEO_CODEC_HEVC_VULKAN: return supported_video_codecs.hevc.low_power; // TODO: hdr, 10 bit + } + return false; +} - const bool video_codec_auto = strcmp(video_codec_to_use, "auto") == 0; - if(video_codec_auto) { - if(gpu_inf.vendor == GSR_GPU_VENDOR_INTEL) { - const AVCodec *h264_codec = find_h264_encoder(gpu_inf.vendor, card_path); - if(!h264_codec) { - fprintf(stderr, "Info: using h265 encoder because a codec was not specified and your gpu does not support h264\n"); - video_codec_to_use = "h265"; - video_codec = VideoCodec::H265; - } else { - fprintf(stderr, "Info: using h264 encoder because a codec was not specified\n"); - video_codec_to_use = "h264"; - video_codec = VideoCodec::H264; - } - } else { - const AVCodec *h265_codec = find_h265_encoder(gpu_inf.vendor, card_path); - - // h265 generally allows recording at a higher resolution than h264 on nvidia cards. On a gtx 1080 4k is the max resolution for h264 but for h265 it's 8k. - // Another important info is that when recording at a higher fps than.. 60? h265 has very bad performance. For example when recording at 144 fps the fps drops to 1 - // while with h264 the fps doesn't drop. - if(!h265_codec) { - fprintf(stderr, "Info: using h264 encoder because a codec was not specified and your gpu does not support h265\n"); - video_codec_to_use = "h264"; - video_codec = VideoCodec::H264; - } else if(fps > 60) { - fprintf(stderr, "Info: using h264 encoder because a codec was not specified and fps is more than 60\n"); - video_codec_to_use = "h264"; - video_codec = VideoCodec::H264; - } else { - fprintf(stderr, "Info: using h265 encoder because a codec was not specified\n"); - video_codec_to_use = "h265"; - video_codec = VideoCodec::H265; - } - } - } +static const AVCodec* pick_video_codec(gsr_video_codec *video_codec, gsr_egl *egl, bool use_software_video_encoder, bool video_codec_auto, bool is_flv, bool *low_power) { + // TODO: software encoder for hevc, av1, vp8 and vp9 + *low_power = false; - // TODO: Allow hevc, vp9 and av1 in (enhanced) flv (supported since ffmpeg 6.1) - const bool is_flv = strcmp(file_extension.c_str(), "flv") == 0; - if(video_codec != VideoCodec::H264 && is_flv) { - video_codec_to_use = "h264"; - video_codec = VideoCodec::H264; - fprintf(stderr, "Warning: h265 is not compatible with flv, falling back to h264 instead.\n"); + gsr_supported_video_codecs supported_video_codecs; + if(!get_supported_video_codecs(egl, *video_codec, use_software_video_encoder, true, &supported_video_codecs)) { + fprintf(stderr, "Error: failed to query for supported video codecs\n"); + _exit(11); } const AVCodec *video_codec_f = nullptr; - switch(video_codec) { - case VideoCodec::H264: - video_codec_f = find_h264_encoder(gpu_inf.vendor, card_path); + + switch(*video_codec) { + case GSR_VIDEO_CODEC_H264: { + if(use_software_video_encoder) + video_codec_f = avcodec_find_encoder_by_name("libx264"); + else if(supported_video_codecs.h264.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); + break; + } + case GSR_VIDEO_CODEC_HEVC: { + if(supported_video_codecs.hevc.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); + break; + } + case GSR_VIDEO_CODEC_HEVC_HDR: { + if(supported_video_codecs.hevc_hdr.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); + break; + } + case GSR_VIDEO_CODEC_HEVC_10BIT: { + if(supported_video_codecs.hevc_10bit.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); + break; + } + case GSR_VIDEO_CODEC_AV1: { + if(supported_video_codecs.av1.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); + break; + } + case GSR_VIDEO_CODEC_AV1_HDR: { + if(supported_video_codecs.av1_hdr.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); break; - case VideoCodec::H265: - video_codec_f = find_h265_encoder(gpu_inf.vendor, card_path); + } + case GSR_VIDEO_CODEC_AV1_10BIT: { + if(supported_video_codecs.av1_10bit.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); break; - case VideoCodec::AV1: - video_codec_f = find_av1_encoder(gpu_inf.vendor, card_path); + } + case GSR_VIDEO_CODEC_VP8: { + if(supported_video_codecs.vp8.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); break; + } + case GSR_VIDEO_CODEC_VP9: { + if(supported_video_codecs.vp9.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); + break; + } + case GSR_VIDEO_CODEC_H264_VULKAN: { + if(supported_video_codecs.h264.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); + break; + } + case GSR_VIDEO_CODEC_HEVC_VULKAN: { + // TODO: hdr, 10 bit + if(supported_video_codecs.hevc.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); + break; + } } if(!video_codec_auto && !video_codec_f && !is_flv) { - switch(video_codec) { - case VideoCodec::H264: { - fprintf(stderr, "Warning: selected video codec h264 is not supported, trying h265 instead\n"); - video_codec_to_use = "h265"; - video_codec = VideoCodec::H265; - video_codec_f = find_h265_encoder(gpu_inf.vendor, card_path); + switch(*video_codec) { + case GSR_VIDEO_CODEC_H264: { + fprintf(stderr, "Warning: selected video codec h264 is not supported, trying hevc instead\n"); + *video_codec = GSR_VIDEO_CODEC_HEVC; + if(supported_video_codecs.hevc.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); break; } - case VideoCodec::H265: { - fprintf(stderr, "Warning: selected video codec h265 is not supported, trying h264 instead\n"); - video_codec_to_use = "h264"; - video_codec = VideoCodec::H264; - video_codec_f = find_h264_encoder(gpu_inf.vendor, card_path); + case GSR_VIDEO_CODEC_HEVC: + case GSR_VIDEO_CODEC_HEVC_HDR: + case GSR_VIDEO_CODEC_HEVC_10BIT: { + fprintf(stderr, "Warning: selected video codec hevc is not supported, trying h264 instead\n"); + *video_codec = GSR_VIDEO_CODEC_H264; + if(supported_video_codecs.h264.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); break; } - case VideoCodec::AV1: { + case GSR_VIDEO_CODEC_AV1: + case GSR_VIDEO_CODEC_AV1_HDR: + case GSR_VIDEO_CODEC_AV1_10BIT: { fprintf(stderr, "Warning: selected video codec av1 is not supported, trying h264 instead\n"); - video_codec_to_use = "h264"; - video_codec = VideoCodec::H264; - video_codec_f = find_h264_encoder(gpu_inf.vendor, card_path); + *video_codec = GSR_VIDEO_CODEC_H264; + if(supported_video_codecs.h264.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); break; } - } - } - - if(!video_codec_f) { - const char *video_codec_name = ""; - switch(video_codec) { - case VideoCodec::H264: { - video_codec_name = "h265"; + case GSR_VIDEO_CODEC_VP8: + case GSR_VIDEO_CODEC_VP9: + // TODO: Cant fallback to other codec because webm only supports vp8/vp9 break; - } - case VideoCodec::H265: { - video_codec_name = "h265"; + case GSR_VIDEO_CODEC_H264_VULKAN: { + fprintf(stderr, "Warning: selected video codec h264_vulkan is not supported, trying h264 instead\n"); + *video_codec = GSR_VIDEO_CODEC_H264; + // Need to do a query again because this time it's without vulkan + if(!get_supported_video_codecs(egl, *video_codec, use_software_video_encoder, true, &supported_video_codecs)) { + fprintf(stderr, "Error: failed to query for supported video codecs\n"); + _exit(11); + } + if(supported_video_codecs.h264.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); break; } - case VideoCodec::AV1: { - video_codec_name = "av1"; + case GSR_VIDEO_CODEC_HEVC_VULKAN: { + fprintf(stderr, "Warning: selected video codec hevc_vulkan is not supported, trying hevc instead\n"); + *video_codec = GSR_VIDEO_CODEC_HEVC; + // Need to do a query again because this time it's without vulkan + if(!get_supported_video_codecs(egl, *video_codec, use_software_video_encoder, true, &supported_video_codecs)) { + fprintf(stderr, "Error: failed to query for supported video codecs\n"); + _exit(11); + } + if(supported_video_codecs.hevc.supported) + video_codec_f = get_ffmpeg_video_codec(*video_codec, egl->gpu_info.vendor); break; } } + } + if(!video_codec_f) { + const char *video_codec_name = video_codec_to_string(*video_codec); fprintf(stderr, "Error: your gpu does not support '%s' video codec. If you are sure that your gpu does support '%s' video encoding and you are using an AMD/Intel GPU,\n" - " then it's possible that your distro has disabled hardware accelerated video encoding for '%s' video codec.\n" - " This may be the case on corporate distros such as Manjaro.\n" - " You can test this by running 'vainfo | grep VAEntrypointEncSlice' to see if it matches any H264/HEVC/AV1 profile. vainfo is part of libva-utils.\n" - " On such distros, you need to manually install mesa from source to enable H264/HEVC/AV1 hardware acceleration, or use a more user friendly distro.\n", video_codec_name, video_codec_name, video_codec_name); + " then make sure you have installed the GPU specific vaapi packages (intel-media-driver, libva-intel-driver, libva-mesa-driver and linux-firmware).\n" + " It's also possible that your distro has disabled hardware accelerated video encoding for '%s' video codec.\n" + " This may be the case on corporate distros such as Manjaro, Fedora or OpenSUSE.\n" + " You can test this by running 'vainfo | grep VAEntrypointEncSlice' to see if it matches any H264/HEVC/AV1/VP8/VP9 profile.\n" + " On such distros, you need to manually install mesa from source to enable H264/HEVC hardware acceleration, or use a more user friendly distro. Alternatively record with AV1 if supported by your GPU.\n" + " You can alternatively use the flatpak version of GPU Screen Recorder (https://flathub.org/apps/com.dec05eba.gpu_screen_recorder) which bypasses system issues with patented H264/HEVC codecs.\n" + " Make sure you have mesa-extra freedesktop runtime installed when using the flatpak (this should be the default), which can be installed with this command:\n" + " flatpak install --system org.freedesktop.Platform.GL.default//23.08-extra\n" + " If your GPU doesn't support hardware accelerated video encoding then you can use '-encoder cpu' option to encode with your cpu instead.\n", video_codec_name, video_codec_name, video_codec_name); _exit(2); } - const bool is_livestream = is_livestream_path(filename); + *low_power = video_codec_only_supports_low_power_mode(supported_video_codecs, *video_codec); + + return video_codec_f; +} + +static const AVCodec* select_video_codec_with_fallback(gsr_video_codec *video_codec, const char *file_extension, bool use_software_video_encoder, gsr_egl *egl, bool *low_power) { + const bool video_codec_auto = *video_codec == (gsr_video_codec)GSR_VIDEO_CODEC_AUTO; + if(video_codec_auto) { + if(strcmp(file_extension, "webm") == 0) { + fprintf(stderr, "Info: using vp8 encoder because a codec was not specified and the file extension is .webm\n"); + *video_codec = GSR_VIDEO_CODEC_VP8; + } else { + fprintf(stderr, "Info: using h264 encoder because a codec was not specified\n"); + *video_codec = GSR_VIDEO_CODEC_H264; + } + } + + // TODO: Allow hevc, vp9 and av1 in (enhanced) flv (supported since ffmpeg 6.1) + const bool is_flv = strcmp(file_extension, "flv") == 0; + if(is_flv) { + if(*video_codec != GSR_VIDEO_CODEC_H264) { + *video_codec = GSR_VIDEO_CODEC_H264; + fprintf(stderr, "Warning: hevc/av1 is not compatible with flv, falling back to h264 instead.\n"); + } + + // if(audio_codec != GSR_AUDIO_CODEC_AAC) { + // audio_codec_to_use = "aac"; + // audio_codec = GSR_AUDIO_CODEC_AAC; + // fprintf(stderr, "Warning: flv only supports aac, falling back to aac instead.\n"); + // } + } + + const bool is_hls = strcmp(file_extension, "m3u8") == 0; + if(is_hls) { + if(video_codec_is_av1(*video_codec)) { + *video_codec = GSR_VIDEO_CODEC_HEVC; + fprintf(stderr, "Warning: av1 is not compatible with hls (m3u8), falling back to hevc instead.\n"); + } + + // if(audio_codec != GSR_AUDIO_CODEC_AAC) { + // audio_codec_to_use = "aac"; + // audio_codec = GSR_AUDIO_CODEC_AAC; + // fprintf(stderr, "Warning: hls (m3u8) only supports aac, falling back to aac instead.\n"); + // } + } + + if(use_software_video_encoder && *video_codec != GSR_VIDEO_CODEC_H264) { + fprintf(stderr, "Error: \"-encoder cpu\" option is currently only available when using h264 codec option (-k)\n"); + args_parser_print_usage(); + _exit(1); + } + + return pick_video_codec(video_codec, egl, use_software_video_encoder, video_codec_auto, is_flv, low_power); +} + +static std::vector<AudioDeviceData> create_device_audio_inputs(const std::vector<AudioInput> &audio_inputs, AVCodecContext *audio_codec_context, int num_channels, double num_audio_frames_shift, std::vector<AVFilterContext*> &src_filter_ctx, bool use_amix) { + std::vector<AudioDeviceData> audio_track_audio_devices; + for(size_t i = 0; i < audio_inputs.size(); ++i) { + const auto &audio_input = audio_inputs[i]; + AVFilterContext *src_ctx = nullptr; + if(use_amix) + src_ctx = src_filter_ctx[i]; + + AudioDeviceData audio_device; + audio_device.audio_input = audio_input; + audio_device.src_filter_ctx = src_ctx; + + if(audio_input.name.empty()) { + audio_device.sound_device.handle = NULL; + audio_device.sound_device.frames = 0; + } else { + const std::string description = "gsr-" + audio_input.name; + if(sound_device_get_by_name(&audio_device.sound_device, audio_input.name.c_str(), description.c_str(), num_channels, audio_codec_context->frame_size, audio_codec_context_get_audio_format(audio_codec_context)) != 0) { + fprintf(stderr, "Error: failed to get \"%s\" audio device\n", audio_input.name.c_str()); + _exit(1); + } + } + + audio_device.frame = create_audio_frame(audio_codec_context); + audio_device.frame->pts = -audio_codec_context->frame_size * num_audio_frames_shift; + + audio_track_audio_devices.push_back(std::move(audio_device)); + } + return audio_track_audio_devices; +} + +#ifdef GSR_APP_AUDIO +static AudioDeviceData create_application_audio_audio_input(const MergedAudioInputs &merged_audio_inputs, AVCodecContext *audio_codec_context, int num_channels, double num_audio_frames_shift, gsr_pipewire_audio *pipewire_audio) { + AudioDeviceData audio_device; + audio_device.frame = create_audio_frame(audio_codec_context); + audio_device.frame->pts = -audio_codec_context->frame_size * num_audio_frames_shift; + + char random_str[8]; + if(!generate_random_characters_standard_alphabet(random_str, sizeof(random_str))) { + fprintf(stderr, "gsr error: failed to generate random string\n"); + _exit(1); + } + std::string combined_sink_name = "gsr-combined-"; + combined_sink_name.append(random_str, sizeof(random_str)); + + if(!gsr_pipewire_audio_create_virtual_sink(pipewire_audio, combined_sink_name.c_str())) { + fprintf(stderr, "gsr error: failed to create virtual sink for application audio\n"); + _exit(1); + } + + combined_sink_name += ".monitor"; + + if(sound_device_get_by_name(&audio_device.sound_device, combined_sink_name.c_str(), "gpu-screen-recorder", num_channels, audio_codec_context->frame_size, audio_codec_context_get_audio_format(audio_codec_context)) != 0) { + fprintf(stderr, "Error: failed to setup audio recording to combined sink\n"); + _exit(1); + } + + std::vector<const char*> audio_devices_sources; + for(const auto &audio_input : merged_audio_inputs.audio_inputs) { + if(audio_input.type == AudioInputType::DEVICE) + audio_devices_sources.push_back(audio_input.name.c_str()); + } + + bool app_audio_inverted = false; + std::vector<const char*> app_names; + for(const auto &audio_input : merged_audio_inputs.audio_inputs) { + if(audio_input.type == AudioInputType::APPLICATION) { + app_names.push_back(audio_input.name.c_str()); + app_audio_inverted = audio_input.inverted; + } + } + + if(!audio_devices_sources.empty()) { + if(!gsr_pipewire_audio_add_link_from_sources_to_sink(pipewire_audio, audio_devices_sources.data(), audio_devices_sources.size(), combined_sink_name.c_str())) { + fprintf(stderr, "gsr error: failed to add application audio link\n"); + _exit(1); + } + } + + if(app_audio_inverted) { + if(!gsr_pipewire_audio_add_link_from_apps_to_sink_inverted(pipewire_audio, app_names.data(), app_names.size(), combined_sink_name.c_str())) { + fprintf(stderr, "gsr error: failed to add application audio link\n"); + _exit(1); + } + } else { + if(!gsr_pipewire_audio_add_link_from_apps_to_sink(pipewire_audio, app_names.data(), app_names.size(), combined_sink_name.c_str())) { + fprintf(stderr, "gsr error: failed to add application audio link\n"); + _exit(1); + } + } + + return audio_device; +} +#endif + +static bool get_image_format_from_filename(const char *filename, gsr_image_format *image_format) { + if(string_ends_with(filename, ".jpg") || string_ends_with(filename, ".jpeg")) { + *image_format = GSR_IMAGE_FORMAT_JPEG; + return true; + } else if(string_ends_with(filename, ".png")) { + *image_format = GSR_IMAGE_FORMAT_PNG; + return true; + } else { + return false; + } +} + +// TODO: replace this with start_recording_create_steams +static bool av_open_file_write_header(AVFormatContext *av_format_context, const char *filename) { + int ret = avio_open(&av_format_context->pb, filename, AVIO_FLAG_WRITE); + if(ret < 0) { + fprintf(stderr, "Error: Could not open '%s': %s\n", filename, av_error_to_string(ret)); + return false; + } + + AVDictionary *options = nullptr; + av_dict_set(&options, "strict", "experimental", 0); + //av_dict_set_int(&av_format_context->metadata, "video_full_range_flag", 1, 0); + + ret = avformat_write_header(av_format_context, &options); + if(ret < 0) + fprintf(stderr, "Error occurred when writing header to output file: %s\n", av_error_to_string(ret)); + + const bool success = ret >= 0; + if(!success) + avio_close(av_format_context->pb); + + av_dict_free(&options); + return success; +} + +static int audio_codec_get_frame_size(gsr_audio_codec audio_codec) { + switch(audio_codec) { + case GSR_AUDIO_CODEC_AAC: return 1024; + case GSR_AUDIO_CODEC_OPUS: return 960; + case GSR_AUDIO_CODEC_FLAC: + assert(false); + return 1024; + } + assert(false); + return 1024; +} + +static size_t calculate_estimated_replay_buffer_packets(int64_t replay_buffer_size_secs, int fps, gsr_audio_codec audio_codec, const std::vector<MergedAudioInputs> &audio_inputs) { + if(replay_buffer_size_secs == -1) + return 0; + + int audio_fps = 0; + if(!audio_inputs.empty()) + audio_fps = AUDIO_SAMPLE_RATE / audio_codec_get_frame_size(audio_codec); + + return replay_buffer_size_secs * (fps + audio_fps * audio_inputs.size()); +} + +static void set_display_server_environment_variables() { + // Some users dont have properly setup environments (no display manager that does systemctl --user import-environment DISPLAY WAYLAND_DISPLAY) + const char *display = getenv("DISPLAY"); + if(!display) { + display = ":0"; + setenv("DISPLAY", display, true); + } + + const char *wayland_display = getenv("WAYLAND_DISPLAY"); + if(!wayland_display) { + wayland_display = "wayland-1"; + setenv("WAYLAND_DISPLAY", wayland_display, true); + } +} + +int main(int argc, char **argv) { + setlocale(LC_ALL, "C"); // Sigh... stupid C + mallopt(M_MMAP_THRESHOLD, 65536); + + signal(SIGINT, stop_handler); + signal(SIGTERM, stop_handler); + signal(SIGUSR1, save_replay_handler); + signal(SIGUSR2, toggle_pause_handler); + signal(SIGRTMIN, toggle_replay_recording_handler); + signal(SIGRTMIN+1, save_replay_10_seconds_handler); + signal(SIGRTMIN+2, save_replay_30_seconds_handler); + signal(SIGRTMIN+3, save_replay_1_minute_handler); + signal(SIGRTMIN+4, save_replay_5_minutes_handler); + signal(SIGRTMIN+5, save_replay_10_minutes_handler); + signal(SIGRTMIN+6, save_replay_30_minutes_handler); + + set_display_server_environment_variables(); + + // Stop nvidia driver from buffering frames + setenv("__GL_MaxFramesAllowed", "1", true); + // If this is set to 1 then cuGraphicsGLRegisterImage will fail for egl context with error: invalid OpenGL or DirectX context, + // so we overwrite it + setenv("__GL_THREADED_OPTIMIZATIONS", "0", true); + // Forces low latency encoding mode. Use this environment variable until vaapi supports setting this as a parameter. + // The downside of this is that it always uses maximum power, which is not ideal for replay mode that runs on system startup. + // This option was added in mesa 24.1.4, released in july 17, 2024. + // TODO: Add an option to enable/disable this? + // Seems like the performance issue is not in encoding, but rendering the frame. + // Some frames end up taking 10 times longer. Seems to be an issue with amd gpu power management when letting the application sleep on the cpu side? + setenv("AMD_DEBUG", "lowlatencyenc", true); + // Some people set this to nvidia (for nvdec) or vdpau (for nvidia vdpau), which breaks gpu screen recorder since + // nvidia doesn't support vaapi and nvidia-vaapi-driver doesn't support encoding yet. + // Let vaapi find the match vaapi driver instead of forcing a specific one. + unsetenv("LIBVA_DRIVER_NAME"); + // Some people set this to force all applications to vsync on nvidia, but this makes eglSwapBuffers never return. + unsetenv("__GL_SYNC_TO_VBLANK"); + // Same as above, but for amd/intel + unsetenv("vblank_mode"); + + if(geteuid() == 0) { + fprintf(stderr, "Error: don't run gpu-screen-recorder as the root user\n"); + _exit(1); + } + + args_handlers arg_handlers; + arg_handlers.version = version_command; + arg_handlers.info = info_command; + arg_handlers.list_audio_devices = list_audio_devices_command; + arg_handlers.list_application_audio = list_application_audio_command; + arg_handlers.list_capture_options = list_capture_options_command; + + args_parser arg_parser; + if(!args_parser_parse(&arg_parser, argc, argv, &arg_handlers, NULL)) + _exit(1); + + //av_log_set_level(AV_LOG_TRACE); + + const Arg *audio_input_arg = args_parser_get_arg(&arg_parser, "-a"); + assert(audio_input_arg); + + AudioDevices audio_devices; + if(audio_input_arg->num_values > 0) + audio_devices = get_pulseaudio_inputs(); + + std::vector<MergedAudioInputs> requested_audio_inputs = parse_audio_inputs(audio_devices, audio_input_arg); + + const bool uses_app_audio = merged_audio_inputs_has_app_audio(requested_audio_inputs); + std::vector<std::string> app_audio_names; +#ifdef GSR_APP_AUDIO + gsr_pipewire_audio pipewire_audio; + memset(&pipewire_audio, 0, sizeof(pipewire_audio)); + if(uses_app_audio) { + if(!pulseaudio_server_is_pipewire()) { + fprintf(stderr, "gsr error: your sound server is not PipeWire. Application audio is only available when running PipeWire audio server\n"); + _exit(2); + } + + if(!gsr_pipewire_audio_init(&pipewire_audio)) { + fprintf(stderr, "gsr error: failed to setup PipeWire audio for application audio capture\n"); + _exit(2); + } + + gsr_pipewire_audio_for_each_app(&pipewire_audio, [](const char *app_name, void *userdata) { + std::vector<std::string> *app_audio_names = (std::vector<std::string>*)userdata; + app_audio_names->push_back(app_name); + return true; + }, &app_audio_names); + } +#endif + + validate_merged_audio_inputs_app_audio(requested_audio_inputs, app_audio_names); + + const bool is_replaying = arg_parser.replay_buffer_size_secs != -1; + const bool is_portal_capture = strcmp(arg_parser.window, "portal") == 0; + + bool wayland = false; + Display *dpy = XOpenDisplay(nullptr); + if (!dpy) { + wayland = true; + fprintf(stderr, "Warning: failed to connect to the X server. Assuming wayland is running without Xwayland\n"); + } + + XSetErrorHandler(x11_error_handler); + XSetIOErrorHandler(x11_io_error_handler); + + if(!wayland) + wayland = is_xwayland(dpy); + + if(!wayland && is_using_prime_run()) { + // Disable prime-run and similar options as it doesn't work, the monitor to capture has to be run on the same device. + // This is fine on wayland since nvidia uses drm interface there and the monitor query checks the monitors connected + // to the drm device. + fprintf(stderr, "Warning: use of prime-run on X11 is not supported. Disabling prime-run\n"); + disable_prime_run(); + } + + gsr_window *window = gsr_window_create(dpy, wayland); + if(!window) { + fprintf(stderr, "Error: failed to create window\n"); + _exit(1); + } + + if(is_portal_capture && is_using_prime_run()) { + fprintf(stderr, "Warning: use of prime-run with -w portal option is currently not supported. Disabling prime-run\n"); + disable_prime_run(); + } + + const bool is_monitor_capture = strcmp(arg_parser.window, "focused") != 0 && strcmp(arg_parser.window, "region") != 0 && !is_portal_capture && contains_non_hex_number(arg_parser.window); + gsr_egl egl; + if(!gsr_egl_load(&egl, window, is_monitor_capture, arg_parser.gl_debug)) { + fprintf(stderr, "gsr error: failed to load opengl\n"); + _exit(1); + } + + gsr_shader_enable_debug_output(arg_parser.gl_debug); +#ifndef NDEBUG + gsr_shader_enable_debug_output(true); +#endif + + if(!args_parser_validate_with_gl_info(&arg_parser, &egl)) + _exit(1); + + egl.card_path[0] = '\0'; + if(monitor_capture_use_drm(window, egl.gpu_info.vendor)) { + // TODO: Allow specifying another card, and in other places + if(!gsr_get_valid_card_path(&egl, egl.card_path, is_monitor_capture)) { + fprintf(stderr, "Error: no /dev/dri/cardX device found. Make sure that you have at least one monitor connected or record a single window instead on X11 or record with the -w portal option\n"); + _exit(2); + } + } + + // if(wayland && is_monitor_capture) { + // fprintf(stderr, "gsr warning: it's not possible to sync video to recorded monitor exactly on wayland when recording a monitor." + // " If you experience stutter in the video then record with portal capture option instead (-w portal) or use X11 instead\n"); + // } + + gsr_image_format image_format; + if(get_image_format_from_filename(arg_parser.filename, &image_format)) { + if(audio_input_arg->num_values > 0) { + fprintf(stderr, "Error: can't record audio (-a) when taking a screenshot\n"); + _exit(1); + } + + capture_image_to_file(arg_parser, &egl, image_format); + _exit(0); + } + + AVFormatContext *av_format_context; + // The output format is automatically guessed by the file extension + avformat_alloc_output_context2(&av_format_context, nullptr, arg_parser.container_format, arg_parser.filename); + if (!av_format_context) { + if(arg_parser.container_format) { + fprintf(stderr, "Error: Container format '%s' (argument -c) is not valid\n", arg_parser.container_format); + } else { + fprintf(stderr, "Error: Failed to deduce container format from file extension. Use the '-c' option to specify container format\n"); + args_parser_print_usage(); + _exit(1); + } + _exit(1); + } + + const AVOutputFormat *output_format = av_format_context->oformat; + + std::string file_extension = output_format->extensions ? output_format->extensions : ""; + { + size_t comma_index = file_extension.find(','); + if(comma_index != std::string::npos) + file_extension = file_extension.substr(0, comma_index); + } + + const bool force_no_audio_offset = arg_parser.is_livestream || arg_parser.is_output_piped || (file_extension != "mp4" && file_extension != "mkv" && file_extension != "webm"); + const double target_fps = 1.0 / (double)arg_parser.fps; + + const bool uses_amix = merged_audio_inputs_should_use_amix(requested_audio_inputs); + arg_parser.audio_codec = select_audio_codec_with_fallback(arg_parser.audio_codec, file_extension, uses_amix); + bool low_power = false; + const AVCodec *video_codec_f = select_video_codec_with_fallback(&arg_parser.video_codec, file_extension.c_str(), arg_parser.video_encoder == GSR_VIDEO_ENCODER_HW_CPU, &egl, &low_power); + + gsr_capture *capture = create_capture_impl(arg_parser, &egl, false); + // (Some?) livestreaming services require at least one audio track to work. // If not audio is provided then create one silent audio track. - if(is_livestream && requested_audio_inputs.empty()) { + if(arg_parser.is_livestream && requested_audio_inputs.empty()) { fprintf(stderr, "Info: live streaming but no audio track was added. Adding a silent audio track\n"); MergedAudioInputs mai; - mai.audio_inputs.push_back({ "", "gsr-silent" }); + mai.audio_inputs.push_back({""}); requested_audio_inputs.push_back(std::move(mai)); } - if(is_livestream && framerate_mode != FramerateMode::CONSTANT) { - fprintf(stderr, "Info: framerate mode was forcefully set to \"cfr\" because live streaming was detected\n"); - framerate_mode = FramerateMode::CONSTANT; - framerate_mode_str = "cfr"; - } - AVStream *video_stream = nullptr; std::vector<AudioTrack> audio_tracks; - AVCodecContext *video_codec_context = create_video_codec_context(gpu_inf.vendor == GSR_GPU_VENDOR_NVIDIA ? AV_PIX_FMT_CUDA : AV_PIX_FMT_VAAPI, quality, fps, video_codec_f, is_livestream, gpu_inf.vendor, framerate_mode); - if(replay_buffer_size_secs == -1) + const enum AVPixelFormat video_pix_fmt = get_pixel_format(arg_parser.video_codec, egl.gpu_info.vendor, arg_parser.video_encoder == GSR_VIDEO_ENCODER_HW_CPU); + AVCodecContext *video_codec_context = create_video_codec_context(video_pix_fmt, video_codec_f, egl, arg_parser); + if(!is_replaying) video_stream = create_stream(av_format_context, video_codec_context); - int capture_result = gsr_capture_start(capture, video_codec_context); + if(arg_parser.tune == GSR_TUNE_QUALITY) + video_codec_context->max_b_frames = 2; + + AVFrame *video_frame = av_frame_alloc(); + if(!video_frame) { + fprintf(stderr, "Error: Failed to allocate video frame\n"); + _exit(1); + } + video_frame->format = video_codec_context->pix_fmt; + video_frame->width = 0; + video_frame->height = 0; + video_frame->color_range = video_codec_context->color_range; + video_frame->color_primaries = video_codec_context->color_primaries; + video_frame->color_trc = video_codec_context->color_trc; + video_frame->colorspace = video_codec_context->colorspace; + video_frame->chroma_location = video_codec_context->chroma_sample_location; + + gsr_capture_metadata capture_metadata; + capture_metadata.width = 0; + capture_metadata.height = 0; + capture_metadata.fps = arg_parser.fps; + capture_metadata.video_codec_context = video_codec_context; + capture_metadata.frame = video_frame; + + int capture_result = gsr_capture_start(capture, &capture_metadata); if(capture_result != 0) { fprintf(stderr, "gsr error: gsr_capture_start failed\n"); _exit(capture_result); } - open_video(video_codec_context, quality, very_old_gpu, gpu_inf.vendor, pixel_format); - if(video_stream) + video_codec_context->width = capture_metadata.width; + video_codec_context->height = capture_metadata.height; + video_frame->width = capture_metadata.width; + video_frame->height = capture_metadata.height; + + const size_t estimated_replay_buffer_packets = calculate_estimated_replay_buffer_packets(arg_parser.replay_buffer_size_secs, arg_parser.fps, arg_parser.audio_codec, requested_audio_inputs); + gsr_encoder encoder; + if(!gsr_encoder_init(&encoder, arg_parser.replay_storage, estimated_replay_buffer_packets, arg_parser.replay_buffer_size_secs, arg_parser.filename)) { + fprintf(stderr, "Error: failed to create encoder\n"); + _exit(1); + } + + gsr_video_encoder *video_encoder = create_video_encoder(&egl, arg_parser); + if(!video_encoder) { + fprintf(stderr, "Error: failed to create video encoder\n"); + _exit(1); + } + + if(!gsr_video_encoder_start(video_encoder, video_codec_context, video_frame)) { + fprintf(stderr, "Error: failed to start video encoder\n"); + _exit(1); + } + + capture_metadata.width = video_codec_context->width; + capture_metadata.height = video_codec_context->height; + + gsr_color_conversion_params color_conversion_params; + memset(&color_conversion_params, 0, sizeof(color_conversion_params)); + color_conversion_params.color_range = arg_parser.color_range; + color_conversion_params.egl = &egl; + color_conversion_params.load_external_image_shader = gsr_capture_uses_external_image(capture); + gsr_video_encoder_get_textures(video_encoder, color_conversion_params.destination_textures, &color_conversion_params.num_destination_textures, &color_conversion_params.destination_color); + + gsr_color_conversion color_conversion; + if(gsr_color_conversion_init(&color_conversion, &color_conversion_params) != 0) { + fprintf(stderr, "gsr error: main: failed to create color conversion\n"); + _exit(1); + } + + gsr_color_conversion_clear(&color_conversion); + + if(arg_parser.video_encoder == GSR_VIDEO_ENCODER_HW_CPU) { + open_video_software(video_codec_context, arg_parser); + } else { + open_video_hardware(video_codec_context, low_power, egl, arg_parser); + } + + if(video_stream) { avcodec_parameters_from_context(video_stream->codecpar, video_codec_context); + gsr_encoder_add_recording_destination(&encoder, video_codec_context, av_format_context, video_stream, 0); + } + int audio_max_frame_size = 1024; int audio_stream_index = VIDEO_STREAM_INDEX + 1; for(const MergedAudioInputs &merged_audio_inputs : requested_audio_inputs) { - AVCodecContext *audio_codec_context = create_audio_codec_context(fps, audio_codec); + const bool use_amix = audio_inputs_should_use_amix(merged_audio_inputs.audio_inputs); + AVCodecContext *audio_codec_context = create_audio_codec_context(arg_parser.fps, arg_parser.audio_codec, use_amix, arg_parser.audio_bitrate); AVStream *audio_stream = nullptr; - if(replay_buffer_size_secs == -1) + if(!is_replaying) { audio_stream = create_stream(av_format_context, audio_codec_context); + if(gsr_encoder_add_recording_destination(&encoder, audio_codec_context, av_format_context, audio_stream, 0) == (size_t)-1) + fprintf(stderr, "gsr error: added too many audio sources\n"); + } + + if(audio_stream && !merged_audio_inputs.track_name.empty()) + av_dict_set(&audio_stream->metadata, "title", merged_audio_inputs.track_name.c_str(), 0); open_audio(audio_codec_context); if(audio_stream) @@ -1997,7 +3255,6 @@ int main(int argc, char **argv) { std::vector<AVFilterContext*> src_filter_ctx; AVFilterGraph *graph = nullptr; AVFilterContext *sink = nullptr; - bool use_amix = merged_audio_inputs.audio_inputs.size() > 1; if(use_amix) { int err = init_filter_graph(audio_codec_context, &graph, &sink, src_filter_ctx, merged_audio_inputs.audio_inputs.size()); if(err < 0) { @@ -2008,95 +3265,62 @@ int main(int argc, char **argv) { // TODO: Cleanup above - std::vector<AudioDevice> audio_devices; - for(size_t i = 0; i < merged_audio_inputs.audio_inputs.size(); ++i) { - auto &audio_input = merged_audio_inputs.audio_inputs[i]; - AVFilterContext *src_ctx = nullptr; - if(use_amix) - src_ctx = src_filter_ctx[i]; - - AudioDevice audio_device; - audio_device.audio_input = audio_input; - audio_device.src_filter_ctx = src_ctx; - - if(audio_input.name.empty()) { - audio_device.sound_device.handle = NULL; - audio_device.sound_device.frames = 0; - audio_device.frame = NULL; - } else { - if(sound_device_get_by_name(&audio_device.sound_device, audio_input.name.c_str(), audio_input.description.c_str(), num_channels, audio_codec_context->frame_size, audio_codec_context_get_audio_format(audio_codec_context)) != 0) { - fprintf(stderr, "Error: failed to get \"%s\" sound device\n", audio_input.name.c_str()); - _exit(1); - } - audio_device.frame = create_audio_frame(audio_codec_context); - } + const double audio_fps = (double)audio_codec_context->sample_rate / (double)audio_codec_context->frame_size; + const double timeout_sec = 1000.0 / audio_fps / 1000.0; + + const double audio_startup_time_seconds = force_no_audio_offset ? 0 : audio_codec_get_desired_delay(arg_parser.audio_codec, arg_parser.fps);// * ((double)audio_codec_context->frame_size / 1024.0); + const double num_audio_frames_shift = audio_startup_time_seconds / timeout_sec; - audio_devices.push_back(std::move(audio_device)); + std::vector<AudioDeviceData> audio_track_audio_devices; + if(audio_inputs_has_app_audio(merged_audio_inputs.audio_inputs)) { + assert(!use_amix); +#ifdef GSR_APP_AUDIO + audio_track_audio_devices.push_back(create_application_audio_audio_input(merged_audio_inputs, audio_codec_context, num_channels, num_audio_frames_shift, &pipewire_audio)); +#endif + } else { + audio_track_audio_devices = create_device_audio_inputs(merged_audio_inputs.audio_inputs, audio_codec_context, num_channels, num_audio_frames_shift, src_filter_ctx, use_amix); } AudioTrack audio_track; + audio_track.name = merged_audio_inputs.track_name; audio_track.codec_context = audio_codec_context; - audio_track.stream = audio_stream; - audio_track.audio_devices = std::move(audio_devices); + audio_track.audio_devices = std::move(audio_track_audio_devices); audio_track.graph = graph; audio_track.sink = sink; audio_track.stream_index = audio_stream_index; + audio_track.pts = -audio_codec_context->frame_size * num_audio_frames_shift; audio_tracks.push_back(std::move(audio_track)); ++audio_stream_index; - } - - //av_dump_format(av_format_context, 0, filename, 1); - if (replay_buffer_size_secs == -1 && !(output_format->flags & AVFMT_NOFILE)) { - int ret = avio_open(&av_format_context->pb, filename, AVIO_FLAG_WRITE); - if (ret < 0) { - fprintf(stderr, "Error: Could not open '%s': %s\n", filename, av_error_to_string(ret)); - _exit(1); - } + audio_max_frame_size = std::max(audio_max_frame_size, audio_codec_context->frame_size); } - if(replay_buffer_size_secs == -1) { - AVDictionary *options = nullptr; - av_dict_set(&options, "strict", "experimental", 0); - //av_dict_set_int(&av_format_context->metadata, "video_full_range_flag", 1, 0); + //av_dump_format(av_format_context, 0, filename, 1); - int ret = avformat_write_header(av_format_context, &options); - if (ret < 0) { - fprintf(stderr, "Error occurred when writing header to output file: %s\n", av_error_to_string(ret)); + if(!is_replaying) { + if(!av_open_file_write_header(av_format_context, arg_parser.filename)) _exit(1); - } - - av_dict_free(&options); } - const double start_time_pts = clock_get_monotonic_seconds(); - - double start_time = clock_get_monotonic_seconds(); // todo - target_fps to make first frame start immediately? - double frame_timer_start = start_time; + double fps_start_time = clock_get_monotonic_seconds(); + //double frame_timer_start = fps_start_time; int fps_counter = 0; + int damage_fps_counter = 0; - AVFrame *frame = av_frame_alloc(); - if (!frame) { - fprintf(stderr, "Error: Failed to allocate frame\n"); - _exit(1); - } - frame->format = video_codec_context->pix_fmt; - frame->width = video_codec_context->width; - frame->height = video_codec_context->height; - frame->color_range = video_codec_context->color_range; - frame->color_primaries = video_codec_context->color_primaries; - frame->color_trc = video_codec_context->color_trc; - frame->colorspace = video_codec_context->colorspace; - frame->chroma_location = video_codec_context->chroma_sample_location; + bool paused = false; + double paused_time_offset = 0.0; + double paused_time_start = 0.0; + bool replay_recording = false; + RecordingStartResult replay_recording_start_result; + std::vector<size_t> replay_recording_items; + std::string replay_recording_filepath; + bool force_iframe_frame = false; // Only needed for video since audio frames are always iframes - std::mutex write_output_mutex; std::mutex audio_filter_mutex; const double record_start_time = clock_get_monotonic_seconds(); - std::deque<std::shared_ptr<PacketData>> frame_data_queue; - bool frames_erased = false; - const size_t audio_buffer_size = 1024 * 4 * 2; // max 4 bytes/sample, 2 channels + const size_t audio_buffer_size = audio_max_frame_size * 4 * 2; // max 4 bytes/sample, 2 channels uint8_t *empty_audio = (uint8_t*)malloc(audio_buffer_size); if(!empty_audio) { fprintf(stderr, "Error: failed to create empty audio\n"); @@ -2105,10 +3329,11 @@ int main(int argc, char **argv) { memset(empty_audio, 0, audio_buffer_size); for(AudioTrack &audio_track : audio_tracks) { - for(AudioDevice &audio_device : audio_track.audio_devices) { + for(AudioDeviceData &audio_device : audio_track.audio_devices) { audio_device.thread = std::thread([&]() mutable { const AVSampleFormat sound_device_sample_format = audio_format_to_sample_format(audio_codec_context_get_audio_format(audio_track.codec_context)); - const bool needs_audio_conversion = audio_track.codec_context->sample_fmt != sound_device_sample_format; + // TODO: Always do conversion for now. This fixes issue with stuttering audio on pulseaudio with opus + multiple audio sources merged + const bool needs_audio_conversion = true;//audio_track.codec_context->sample_fmt != sound_device_sample_format; SwrContext *swr = nullptr; if(needs_audio_conversion) { swr = swr_alloc(); @@ -2116,8 +3341,16 @@ int main(int argc, char **argv) { fprintf(stderr, "Failed to create SwrContext\n"); _exit(1); } - av_opt_set_int(swr, "in_channel_layout", AV_CH_LAYOUT_STEREO, 0); - av_opt_set_int(swr, "out_channel_layout", AV_CH_LAYOUT_STEREO, 0); + #if LIBAVUTIL_VERSION_MAJOR <= 56 + av_opt_set_channel_layout(swr, "in_channel_layout", AV_CH_LAYOUT_STEREO, 0); + av_opt_set_channel_layout(swr, "out_channel_layout", AV_CH_LAYOUT_STEREO, 0); + #elif LIBAVUTIL_VERSION_MAJOR >= 59 + av_opt_set_chlayout(swr, "in_chlayout", &audio_track.codec_context->ch_layout, 0); + av_opt_set_chlayout(swr, "out_chlayout", &audio_track.codec_context->ch_layout, 0); + #else + av_opt_set_chlayout(swr, "in_channel_layout", &audio_track.codec_context->ch_layout, 0); + av_opt_set_chlayout(swr, "out_channel_layout", &audio_track.codec_context->ch_layout, 0); + #endif av_opt_set_int(swr, "in_sample_rate", audio_track.codec_context->sample_rate, 0); av_opt_set_int(swr, "out_sample_rate", audio_track.codec_context->sample_rate, 0); av_opt_set_sample_fmt(swr, "in_sample_fmt", sound_device_sample_format, 0); @@ -2125,21 +3358,36 @@ int main(int argc, char **argv) { swr_init(swr); } - const double target_audio_hz = 1.0 / (double)audio_track.codec_context->sample_rate; - double received_audio_time = clock_get_monotonic_seconds(); - const int64_t timeout_ms = std::round((1000.0 / (double)audio_track.codec_context->sample_rate) * 1000.0); - int64_t prev_pts = 0; + const double audio_fps = (double)audio_track.codec_context->sample_rate / (double)audio_track.codec_context->frame_size; + const int64_t timeout_ms = std::round(1000.0 / audio_fps); + const double timeout_sec = 1000.0 / audio_fps / 1000.0; + bool first_frame = true; + int64_t num_received_frames = 0; while(running) { void *sound_buffer; int sound_buffer_size = -1; - if(audio_device.sound_device.handle) - sound_buffer_size = sound_device_read_next_chunk(&audio_device.sound_device, &sound_buffer); + //const double time_before_read_seconds = clock_get_monotonic_seconds(); + if(audio_device.sound_device.handle) { + // TODO: use this instead of calculating time to read. But this can fluctuate and we dont want to go back in time, + // also it's 0.0 for some users??? + double latency_seconds = 0.0; + sound_buffer_size = sound_device_read_next_chunk(&audio_device.sound_device, &sound_buffer, timeout_sec * 2.0, &latency_seconds); + } + const bool got_audio_data = sound_buffer_size >= 0; + //fprintf(stderr, "got audio data: %s\n", got_audio_data ? "yes" : "no"); + //const double time_after_read_seconds = clock_get_monotonic_seconds(); + //const double time_to_read_seconds = time_after_read_seconds - time_before_read_seconds; + //fprintf(stderr, "time to read: %f, %s, %f\n", time_to_read_seconds, got_audio_data ? "yes" : "no", timeout_sec); + const double this_audio_frame_time = clock_get_monotonic_seconds() - paused_time_offset; - const double this_audio_frame_time = clock_get_monotonic_seconds(); - if(got_audio_data) - received_audio_time = this_audio_frame_time; + if(paused) { + if(!audio_device.sound_device.handle) + av_usleep(timeout_ms * 1000); + + continue; + } int ret = av_frame_make_writable(audio_device.frame); if (ret < 0) { @@ -2148,28 +3396,32 @@ int main(int argc, char **argv) { } // TODO: Is this |received_audio_time| really correct? - int64_t num_missing_frames = std::round((this_audio_frame_time - received_audio_time) / target_audio_hz / (int64_t)audio_device.frame->nb_samples); + const int64_t num_expected_frames = std::round((this_audio_frame_time - record_start_time) / timeout_sec); + int64_t num_missing_frames = std::max((int64_t)0LL, num_expected_frames - num_received_frames); + if(got_audio_data) - num_missing_frames = std::max((int64_t)0, num_missing_frames - 1); + num_missing_frames = std::max((int64_t)0LL, num_missing_frames - 1); if(!audio_device.sound_device.handle) num_missing_frames = std::max((int64_t)1, num_missing_frames); - // Jesus is there a better way to do this? I JUST WANT TO KEEP VIDEO AND AUDIO SYNCED HOLY FUCK I WANT TO KILL MYSELF NOW. + // Fucking hell is there a better way to do this? I JUST WANT TO KEEP VIDEO AND AUDIO SYNCED HOLY FUCK I WANT TO KILL MYSELF NOW. // THIS PIECE OF SHIT WANTS EMPTY FRAMES OTHERWISE VIDEO PLAYS TOO FAST TO KEEP UP WITH AUDIO OR THE AUDIO PLAYS TOO EARLY. // BUT WE CANT USE DELAYS TO GIVE DUMMY DATA BECAUSE PULSEAUDIO MIGHT GIVE AUDIO A BIG DELAYED!!! // This garbage is needed because we want to produce constant frame rate videos instead of variable frame rate // videos because bad software such as video editing software and VLC do not support variable frame rate software, // despite nvidia shadowplay and xbox game bar producing variable frame rate videos. // So we have to make sure we produce frames at the same relative rate as the video. - if(num_missing_frames >= 5 || !audio_device.sound_device.handle) { + if((num_missing_frames >= 1 && got_audio_data) || num_missing_frames >= 5 || !audio_device.sound_device.handle) { // TODO: //audio_track.frame->data[0] = empty_audio; - received_audio_time = this_audio_frame_time; - if(needs_audio_conversion) - swr_convert(swr, &audio_device.frame->data[0], audio_device.frame->nb_samples, (const uint8_t**)&empty_audio, audio_track.codec_context->frame_size); - else - audio_device.frame->data[0] = empty_audio; + if(first_frame || num_missing_frames >= 5) { + if(needs_audio_conversion) + swr_convert(swr, &audio_device.frame->data[0], audio_track.codec_context->frame_size, (const uint8_t**)&empty_audio, audio_track.codec_context->frame_size); + else + audio_device.frame->data[0] = empty_audio; + } + first_frame = false; // TODO: Check if duplicate frame can be saved just by writing it with a different pts instead of sending it again std::lock_guard<std::mutex> lock(audio_filter_mutex); @@ -2180,41 +3432,35 @@ int main(int argc, char **argv) { fprintf(stderr, "Error: failed to add audio frame to filter\n"); } } else { - audio_device.frame->pts = (this_audio_frame_time - record_start_time) * (double)AV_TIME_BASE; - const bool same_pts = audio_device.frame->pts == prev_pts; - prev_pts = audio_device.frame->pts; - if(same_pts) - continue; - ret = avcodec_send_frame(audio_track.codec_context, audio_device.frame); if(ret >= 0) { // TODO: Move to separate thread because this could write to network (for example when livestreaming) - receive_frames(audio_track.codec_context, audio_track.stream_index, audio_track.stream, audio_device.frame->pts, av_format_context, record_start_time, frame_data_queue, replay_buffer_size_secs, frames_erased, write_output_mutex); + gsr_encoder_receive_packets(&encoder, audio_track.codec_context, audio_device.frame->pts, audio_track.stream_index); } else { fprintf(stderr, "Failed to encode audio!\n"); } + audio_track.pts += audio_track.codec_context->frame_size; } + + audio_device.frame->pts += audio_track.codec_context->frame_size; + num_received_frames++; } } if(!audio_device.sound_device.handle) - usleep(timeout_ms * 1000); + av_usleep(timeout_ms * 1000); if(got_audio_data) { // TODO: Instead of converting audio, get float audio from alsa. Or does alsa do conversion internally to get this format? if(needs_audio_conversion) - swr_convert(swr, &audio_device.frame->data[0], audio_device.frame->nb_samples, (const uint8_t**)&sound_buffer, audio_track.codec_context->frame_size); + swr_convert(swr, &audio_device.frame->data[0], audio_track.codec_context->frame_size, (const uint8_t**)&sound_buffer, audio_track.codec_context->frame_size); else audio_device.frame->data[0] = (uint8_t*)sound_buffer; + first_frame = false; - audio_device.frame->pts = (this_audio_frame_time - record_start_time) * (double)AV_TIME_BASE; - const bool same_pts = audio_device.frame->pts == prev_pts; - prev_pts = audio_device.frame->pts; - if(same_pts) - continue; + std::lock_guard<std::mutex> lock(audio_filter_mutex); if(audio_track.graph) { - std::lock_guard<std::mutex> lock(audio_filter_mutex); // TODO: av_buffersrc_add_frame if(av_buffersrc_write_frame(audio_device.src_filter_ctx, audio_device.frame) < 0) { fprintf(stderr, "Error: failed to add audio frame to filter\n"); @@ -2223,11 +3469,15 @@ int main(int argc, char **argv) { ret = avcodec_send_frame(audio_track.codec_context, audio_device.frame); if(ret >= 0) { // TODO: Move to separate thread because this could write to network (for example when livestreaming) - receive_frames(audio_track.codec_context, audio_track.stream_index, audio_track.stream, audio_device.frame->pts, av_format_context, record_start_time, frame_data_queue, replay_buffer_size_secs, frames_erased, write_output_mutex); + gsr_encoder_receive_packets(&encoder, audio_track.codec_context, audio_device.frame->pts, audio_track.stream_index); } else { fprintf(stderr, "Failed to encode audio!\n"); } + audio_track.pts += audio_track.codec_context->frame_size; } + + audio_device.frame->pts += audio_track.codec_context->frame_size; + num_received_frames++; } } @@ -2237,161 +3487,383 @@ int main(int argc, char **argv) { } } + std::thread amix_thread; + if(uses_amix) { + amix_thread = std::thread([&]() { + AVFrame *aframe = av_frame_alloc(); + while(running) { + { + std::lock_guard<std::mutex> lock(audio_filter_mutex); + for(AudioTrack &audio_track : audio_tracks) { + if(!audio_track.sink) + continue; + + int err = 0; + while ((err = av_buffersink_get_frame(audio_track.sink, aframe)) >= 0) { + aframe->pts = audio_track.pts; + err = avcodec_send_frame(audio_track.codec_context, aframe); + if(err >= 0){ + // TODO: Move to separate thread because this could write to network (for example when livestreaming) + gsr_encoder_receive_packets(&encoder, audio_track.codec_context, aframe->pts, audio_track.stream_index); + } else { + fprintf(stderr, "Failed to encode audio!\n"); + } + av_frame_unref(aframe); + audio_track.pts += audio_track.codec_context->frame_size; + } + } + } + av_usleep(5 * 1000); // 5 milliseconds + } + av_frame_free(&aframe); + }); + } + // Set update_fps to 24 to test if duplicate/delayed frames cause video/audio desync or too fast/slow video. - const double update_fps = fps + 190; + //const double update_fps = fps + 190; bool should_stop_error = false; - AVFrame *aframe = av_frame_alloc(); - int64_t video_pts_counter = 0; int64_t video_prev_pts = 0; - int64_t audio_prev_pts = 0; + + bool hdr_metadata_set = false; + const bool hdr = video_codec_is_hdr(arg_parser.video_codec); + + double damage_timeout_seconds = arg_parser.framerate_mode == GSR_FRAMERATE_MODE_CONTENT ? 0.5 : 0.1; + damage_timeout_seconds = std::max(damage_timeout_seconds, target_fps); + + bool use_damage_tracking = false; + gsr_damage damage; + memset(&damage, 0, sizeof(damage)); + if(gsr_window_get_display_server(window) == GSR_DISPLAY_SERVER_X11) { + gsr_damage_init(&damage, &egl, arg_parser.record_cursor); + use_damage_tracking = true; + } + + if(is_monitor_capture) + gsr_damage_set_target_monitor(&damage, arg_parser.window); + + double last_capture_seconds = record_start_time; + bool wait_until_frame_time_elapsed = false; while(running) { - double frame_start = clock_get_monotonic_seconds(); + const double frame_start = clock_get_monotonic_seconds(); + + while(gsr_window_process_event(window)) { + gsr_damage_on_event(&damage, gsr_window_get_event_data(window)); + gsr_capture_on_event(capture, &egl); + } + gsr_damage_tick(&damage); + gsr_capture_tick(capture); + + if(!is_monitor_capture) { + Window damage_target_window = 0; + if(capture->get_window_id) + damage_target_window = capture->get_window_id(capture); + + if(damage_target_window != 0) + gsr_damage_set_target_window(&damage, damage_target_window); + } - gsr_capture_tick(capture, video_codec_context, &frame); should_stop_error = false; if(gsr_capture_should_stop(capture, &should_stop_error)) { running = 0; break; } + + bool damaged = false; + if(use_damage_tracking) + damaged = gsr_damage_is_damaged(&damage); + else if(capture->is_damaged) + damaged = capture->is_damaged(capture); + else + damaged = true; + + // TODO: Readd wayland sync warning when removing this + if(arg_parser.framerate_mode != GSR_FRAMERATE_MODE_CONTENT) + damaged = true; + + if(damaged) + ++damage_fps_counter; + ++fps_counter; + const double time_now = clock_get_monotonic_seconds(); + //const double frame_timer_elapsed = time_now - frame_timer_start; + const double elapsed = time_now - fps_start_time; + if (elapsed >= 1.0) { + if(arg_parser.verbose) { + fprintf(stderr, "update fps: %d, damage fps: %d\n", fps_counter, damage_fps_counter); + } + fps_start_time = time_now; + fps_counter = 0; + damage_fps_counter = 0; + } - { - std::lock_guard<std::mutex> lock(audio_filter_mutex); - for(AudioTrack &audio_track : audio_tracks) { - if(!audio_track.sink) - continue; + const double this_video_frame_time = clock_get_monotonic_seconds() - paused_time_offset; + const double time_since_last_frame_captured_seconds = this_video_frame_time - last_capture_seconds; + double frame_time_overflow = time_since_last_frame_captured_seconds - target_fps; + const bool frame_timeout = frame_time_overflow >= 0.0; + + bool force_frame_capture = wait_until_frame_time_elapsed && frame_timeout; + bool allow_capture = !wait_until_frame_time_elapsed || force_frame_capture; + if(arg_parser.framerate_mode == GSR_FRAMERATE_MODE_CONTENT) { + force_frame_capture = false; + allow_capture = frame_timeout; + } - int err = 0; - while ((err = av_buffersink_get_frame(audio_track.sink, aframe)) >= 0) { - const double this_audio_frame_time = clock_get_monotonic_seconds(); - aframe->pts = (this_audio_frame_time - record_start_time) * (double)AV_TIME_BASE; - const bool same_pts = aframe->pts == audio_prev_pts; - audio_prev_pts = aframe->pts; - if(same_pts) { - av_frame_unref(aframe); + bool frame_captured = false; + if((damaged || force_frame_capture) && allow_capture && !paused) { + frame_captured = true; + frame_time_overflow = std::min(std::max(0.0, frame_time_overflow), target_fps); + last_capture_seconds = this_video_frame_time - frame_time_overflow; + wait_until_frame_time_elapsed = false; + + gsr_damage_clear(&damage); + if(capture->clear_damage) + capture->clear_damage(capture); + + // TODO: Dont do this if no damage? + egl.glClear(0); + gsr_capture_capture(capture, &capture_metadata, &color_conversion); + gsr_egl_swap_buffers(&egl); + gsr_video_encoder_copy_textures_to_frame(video_encoder, video_frame, &color_conversion); + + if(hdr && !hdr_metadata_set && !is_replaying && add_hdr_metadata_to_video_stream(capture, video_stream)) + hdr_metadata_set = true; + + const int64_t expected_frames = std::round((this_video_frame_time - record_start_time) / target_fps); + const int num_missed_frames = std::max((int64_t)1LL, expected_frames - video_pts_counter); + + // TODO: Check if duplicate frame can be saved just by writing it with a different pts instead of sending it again + const int num_frames_to_encode = arg_parser.framerate_mode == GSR_FRAMERATE_MODE_CONSTANT ? num_missed_frames : 1; + for(int i = 0; i < num_frames_to_encode; ++i) { + if(arg_parser.framerate_mode == GSR_FRAMERATE_MODE_CONSTANT) { + video_frame->pts = video_pts_counter + i; + } else { + video_frame->pts = (this_video_frame_time - record_start_time) * (double)AV_TIME_BASE; + const bool same_pts = video_frame->pts == video_prev_pts; + video_prev_pts = video_frame->pts; + if(same_pts) continue; - } + } - err = avcodec_send_frame(audio_track.codec_context, aframe); - if(err >= 0){ - // TODO: Move to separate thread because this could write to network (for example when livestreaming) - receive_frames(audio_track.codec_context, audio_track.stream_index, audio_track.stream, aframe->pts, av_format_context, record_start_time, frame_data_queue, replay_buffer_size_secs, frames_erased, write_output_mutex); - } else { - fprintf(stderr, "Failed to encode audio!\n"); - } - av_frame_unref(aframe); + if(force_iframe_frame) { + video_frame->pict_type = AV_PICTURE_TYPE_I; } - } - } - double time_now = clock_get_monotonic_seconds(); - double frame_timer_elapsed = time_now - frame_timer_start; - double elapsed = time_now - start_time; - if (elapsed >= 1.0) { - if(verbose) { - fprintf(stderr, "update fps: %d\n", fps_counter); + int ret = avcodec_send_frame(video_codec_context, video_frame); + if(ret == 0) { + // TODO: Move to separate thread because this could write to network (for example when livestreaming) + gsr_encoder_receive_packets(&encoder, video_codec_context, video_frame->pts, VIDEO_STREAM_INDEX); + } else { + fprintf(stderr, "Error: avcodec_send_frame failed, error: %s\n", av_error_to_string(ret)); + } + + if(force_iframe_frame) { + force_iframe_frame = false; + video_frame->pict_type = AV_PICTURE_TYPE_NONE; + } } - start_time = time_now; - fps_counter = 0; + + video_pts_counter += num_frames_to_encode; } - double frame_time_overflow = frame_timer_elapsed - target_fps; - if (frame_time_overflow >= 0.0) { - frame_time_overflow = std::min(frame_time_overflow, target_fps); - frame_timer_start = time_now - frame_time_overflow; + if(toggle_pause == 1 && !is_replaying) { + const bool new_paused_state = !paused; + if(new_paused_state) { + paused_time_start = clock_get_monotonic_seconds(); + fprintf(stderr, "Paused\n"); + } else { + paused_time_offset += (clock_get_monotonic_seconds() - paused_time_start); + fprintf(stderr, "Unpaused\n"); + } - const double this_video_frame_time = clock_get_monotonic_seconds(); - const int64_t expected_frames = std::round((this_video_frame_time - start_time_pts) / target_fps); - const int num_frames = framerate_mode == FramerateMode::CONSTANT ? std::max((int64_t)0LL, expected_frames - video_pts_counter) : 1; + toggle_pause = 0; + paused = !paused; + } - if(num_frames > 0) { - gsr_capture_capture(capture, frame); + if(toggle_replay_recording && !arg_parser.replay_recording_directory) { + toggle_replay_recording = 0; + printf("Error: Unable to start recording since the -ro option was not specified\n"); + fflush(stdout); + } - // TODO: Check if duplicate frame can be saved just by writing it with a different pts instead of sending it again - for(int i = 0; i < num_frames; ++i) { - if(framerate_mode == FramerateMode::CONSTANT) { - frame->pts = video_pts_counter + i; - } else { - frame->pts = (this_video_frame_time - record_start_time) * (double)AV_TIME_BASE; - const bool same_pts = frame->pts == video_prev_pts; - video_prev_pts = frame->pts; - if(same_pts) - continue; + if(toggle_replay_recording && arg_parser.replay_recording_directory) { + toggle_replay_recording = 0; + const bool new_replay_recording_state = !replay_recording; + if(new_replay_recording_state) { + std::lock_guard<std::mutex> lock(audio_filter_mutex); + replay_recording_items.clear(); + replay_recording_filepath = create_new_recording_filepath_from_timestamp(arg_parser.replay_recording_directory, "Video", file_extension, arg_parser.date_folders); + replay_recording_start_result = start_recording_create_streams(replay_recording_filepath.c_str(), arg_parser.container_format, video_codec_context, audio_tracks, hdr, capture); + if(replay_recording_start_result.av_format_context) { + const size_t video_recording_destination_id = gsr_encoder_add_recording_destination(&encoder, video_codec_context, replay_recording_start_result.av_format_context, replay_recording_start_result.video_stream, video_frame->pts); + if(video_recording_destination_id != (size_t)-1) + replay_recording_items.push_back(video_recording_destination_id); + + for(const auto &audio_input : replay_recording_start_result.audio_inputs) { + const size_t audio_recording_destination_id = gsr_encoder_add_recording_destination(&encoder, audio_input.audio_track->codec_context, replay_recording_start_result.av_format_context, audio_input.stream, audio_input.audio_track->pts); + if(audio_recording_destination_id != (size_t)-1) + replay_recording_items.push_back(audio_recording_destination_id); } - int ret = avcodec_send_frame(video_codec_context, frame); - if(ret == 0) { - // TODO: Move to separate thread because this could write to network (for example when livestreaming) - receive_frames(video_codec_context, VIDEO_STREAM_INDEX, video_stream, frame->pts, av_format_context, - record_start_time, frame_data_queue, replay_buffer_size_secs, frames_erased, write_output_mutex); - } else { - fprintf(stderr, "Error: avcodec_send_frame failed, error: %s\n", av_error_to_string(ret)); - } + replay_recording = true; + force_iframe_frame = true; + fprintf(stderr, "Started recording\n"); + } else { + printf("Error: Failed to start recording\n"); + fflush(stdout); + } + } else if(replay_recording_start_result.av_format_context) { + for(size_t id : replay_recording_items) { + gsr_encoder_remove_recording_destination(&encoder, id); + } + replay_recording_items.clear(); + + if(stop_recording_close_streams(replay_recording_start_result.av_format_context)) { + fprintf(stderr, "Stopped recording\n"); + puts(replay_recording_filepath.c_str()); + fflush(stdout); + if(arg_parser.recording_saved_script) + run_recording_saved_script_async(arg_parser.recording_saved_script, replay_recording_filepath.c_str(), "regular"); + } else { + printf("Error: Failed to save recording\n"); + fflush(stdout); } - gsr_capture_end(capture, frame); - video_pts_counter += num_frames; + replay_recording_start_result = RecordingStartResult{}; + replay_recording = false; + replay_recording_filepath.clear(); } } if(save_replay_thread.valid() && save_replay_thread.wait_for(std::chrono::seconds(0)) == std::future_status::ready) { save_replay_thread.get(); - puts(save_replay_output_filepath.c_str()); - fflush(stdout); - std::lock_guard<std::mutex> lock(write_output_mutex); - save_replay_packets.clear(); + if(save_replay_output_filepath.empty()) { + printf("Error: Failed to save replay\n"); + fflush(stdout); + } else { + puts(save_replay_output_filepath.c_str()); + fflush(stdout); + if(arg_parser.recording_saved_script) + run_recording_saved_script_async(arg_parser.recording_saved_script, save_replay_output_filepath.c_str(), "replay"); + } } - if(save_replay == 1 && !save_replay_thread.valid() && replay_buffer_size_secs != -1) { - save_replay = 0; - save_replay_async(video_codec_context, VIDEO_STREAM_INDEX, audio_tracks, frame_data_queue, frames_erased, filename, container_format, file_extension, write_output_mutex, make_folders); + if(save_replay_seconds != 0 && !save_replay_thread.valid() && is_replaying) { + int current_save_replay_seconds = save_replay_seconds; + if(current_save_replay_seconds > 0) + current_save_replay_seconds += arg_parser.keyint; + + save_replay_seconds = 0; + save_replay_output_filepath.clear(); + save_replay_async(video_codec_context, VIDEO_STREAM_INDEX, audio_tracks, encoder.replay_buffer, arg_parser.filename, arg_parser.container_format, file_extension, arg_parser.date_folders, hdr, capture, current_save_replay_seconds); + + if(arg_parser.restart_replay_on_save && current_save_replay_seconds == save_replay_seconds_full) + gsr_replay_buffer_clear(encoder.replay_buffer); } - double frame_end = clock_get_monotonic_seconds(); - double frame_sleep_fps = 1.0 / update_fps; - double sleep_time = frame_sleep_fps - (frame_end - frame_start); - if(sleep_time > 0.0) - usleep(sleep_time * 1000.0 * 1000.0); + const double frame_end = clock_get_monotonic_seconds(); + const double time_at_frame_end = frame_end - paused_time_offset; + const double time_elapsed_total = time_at_frame_end - record_start_time; + const int64_t frames_elapsed = (int64_t)(time_elapsed_total / target_fps); + const double time_at_next_frame = (frames_elapsed + 1) * target_fps; + double time_to_next_frame = time_at_next_frame - time_elapsed_total; + if(time_to_next_frame > target_fps*1.1) + time_to_next_frame = target_fps; + + const double frame_time = frame_end - frame_start; + const bool frame_deadline_missed = frame_time > target_fps; + if(time_to_next_frame >= 0.0 && !frame_deadline_missed && frame_captured) + av_usleep(time_to_next_frame * 1000.0 * 1000.0); + else { + if(paused) + av_usleep(20.0 * 1000.0); // 20 milliseconds + else if(frame_deadline_missed) + {} + else if(arg_parser.framerate_mode == GSR_FRAMERATE_MODE_CONTENT || !frame_captured) + av_usleep(2.8 * 1000.0); // 2.8 milliseconds + else if(!frame_captured) + av_usleep(1.0 * 1000.0); // 1 milliseconds + wait_until_frame_time_elapsed = true; + } } running = 0; if(save_replay_thread.valid()) { save_replay_thread.get(); - puts(save_replay_output_filepath.c_str()); - fflush(stdout); - std::lock_guard<std::mutex> lock(write_output_mutex); - save_replay_packets.clear(); + if(save_replay_output_filepath.empty()) { + // TODO: Output failed to save + } else { + puts(save_replay_output_filepath.c_str()); + fflush(stdout); + if(arg_parser.recording_saved_script) + run_recording_saved_script_async(arg_parser.recording_saved_script, save_replay_output_filepath.c_str(), "replay"); + } + } + + if(replay_recording_start_result.av_format_context) { + for(size_t id : replay_recording_items) { + gsr_encoder_remove_recording_destination(&encoder, id); + } + replay_recording_items.clear(); + + if(stop_recording_close_streams(replay_recording_start_result.av_format_context)) { + fprintf(stderr, "Stopped recording\n"); + puts(replay_recording_filepath.c_str()); + fflush(stdout); + if(arg_parser.recording_saved_script) + run_recording_saved_script_async(arg_parser.recording_saved_script, replay_recording_filepath.c_str(), "regular"); + } else { + printf("Error: Failed to save recording\n"); + fflush(stdout); + } } for(AudioTrack &audio_track : audio_tracks) { - for(AudioDevice &audio_device : audio_track.audio_devices) { + for(auto &audio_device : audio_track.audio_devices) { audio_device.thread.join(); sound_device_close(&audio_device.sound_device); } } - av_frame_free(&aframe); + if(amix_thread.joinable()) + amix_thread.join(); - if (replay_buffer_size_secs == -1 && av_write_trailer(av_format_context) != 0) { + // TODO: Replace this with start_recording_create_steams + if(!is_replaying && av_write_trailer(av_format_context) != 0) { fprintf(stderr, "Failed to write trailer\n"); } - if(replay_buffer_size_secs == -1 && !(output_format->flags & AVFMT_NOFILE)) + if(!is_replaying) { avio_close(av_format_context->pb); + avformat_free_context(av_format_context); + } + + gsr_damage_deinit(&damage); + gsr_color_conversion_deinit(&color_conversion); + gsr_video_encoder_destroy(video_encoder, video_codec_context); + gsr_encoder_deinit(&encoder); + gsr_capture_destroy(capture); +#ifdef GSR_APP_AUDIO + gsr_pipewire_audio_deinit(&pipewire_audio); +#endif - gsr_capture_destroy(capture, video_codec_context); + if(!is_replaying && arg_parser.recording_saved_script) + run_recording_saved_script_async(arg_parser.recording_saved_script, arg_parser.filename, "regular"); if(dpy) { // TODO: This causes a crash, why? maybe some other library dlclose xlib and that also happened to unload this??? //XCloseDisplay(dpy); } - free((void*)window_str); + //gsr_egl_unload(&egl); + //gsr_window_destroy(&window); + + //av_frame_free(&video_frame); free(empty_audio); + args_parser_deinit(&arg_parser); // We do an _exit here because cuda uses at_exit to do _something_ that causes the program to freeze, // but only on some nvidia driver versions on some gpus (RTX?), and _exit exits the program without calling // the at_exit registered functions. diff --git a/src/overclock.c b/src/overclock.c index 9160a04..df2ae66 100644 --- a/src/overclock.c +++ b/src/overclock.c @@ -2,13 +2,12 @@ #include <X11/Xlib.h> #include <stdio.h> #include <string.h> +#include <stdlib.h> -// HACK!!!: When a program uses cuda (including nvenc) then the nvidia driver drops to performance level 2 (memory transfer rate is dropped and possibly graphics clock). +// HACK!!!: When a program uses cuda (including nvenc) then the nvidia driver drops to max performance level - 1 (memory transfer rate is dropped and possibly graphics clock). // Nvidia does this because in some very extreme cases of cuda there can be memory corruption when running at max memory transfer rate. // So to get around this we overclock memory transfer rate (maybe this should also be done for graphics clock?) to the best performance level while GPU Screen Recorder is running. -// TODO: Does it always drop to performance level 2? - static int min_int(int a, int b) { return a < b ? a : b; } @@ -76,13 +75,13 @@ static unsigned int attribute_type_to_attribute_param_all_levels(NvCTRLAttribute } // Returns 0 on error -static int xnvctrl_get_attribute_max_value(gsr_xnvctrl *xnvctrl, const NVCTRLPerformanceLevelQuery *query, NvCTRLAttributeType attribute_type) { +static int xnvctrl_get_attribute_max_value(gsr_xnvctrl *xnvctrl, int num_performance_levels, NvCTRLAttributeType attribute_type) { NVCTRLAttributeValidValuesRec valid; if(xnvctrl->XNVCTRLQueryValidTargetAttributeValues(xnvctrl->display, NV_CTRL_TARGET_TYPE_GPU, 0, 0, attribute_type_to_attribute_param_all_levels(attribute_type), &valid)) { return valid.u.range.max; } - if(query->num_performance_levels > 0 && xnvctrl->XNVCTRLQueryValidTargetAttributeValues(xnvctrl->display, NV_CTRL_TARGET_TYPE_GPU, 0, query->num_performance_levels - 1, attribute_type_to_attribute_param(attribute_type), &valid)) { + if(num_performance_levels > 0 && xnvctrl->XNVCTRLQueryValidTargetAttributeValues(xnvctrl->display, NV_CTRL_TARGET_TYPE_GPU, 0, num_performance_levels - 1, attribute_type_to_attribute_param(attribute_type), &valid)) { return valid.u.range.max; } @@ -208,6 +207,12 @@ static bool xnvctrl_get_performance_levels(gsr_xnvctrl *xnvctrl, NVCTRLPerforman return success; } +static int compare_mem_transfer_rate_max_asc(const void *a, const void *b) { + const NVCTRLPerformanceLevel *perf_a = a; + const NVCTRLPerformanceLevel *perf_b = b; + return perf_a->mem_transfer_rate_max - perf_b->mem_transfer_rate_max; +} + bool gsr_overclock_load(gsr_overclock *self, Display *display) { memset(self, 0, sizeof(gsr_overclock)); self->num_performance_levels = 0; @@ -234,31 +239,32 @@ bool gsr_overclock_start(gsr_overclock *self) { } self->num_performance_levels = query.num_performance_levels; - int target_transfer_rate_offset = xnvctrl_get_attribute_max_value(&self->xnvctrl, &query, NVCTRL_ATTRIB_GPU_MEM_TRANSFER_RATE) / 2; // Divide by 2 just to be safe that we dont set it too high - if(query.num_performance_levels > 2) { - const int transfer_rate_max_diff = query.performance_level[query.num_performance_levels - 1].mem_transfer_rate_max - query.performance_level[2].mem_transfer_rate_max; - target_transfer_rate_offset = min_int(target_transfer_rate_offset, transfer_rate_max_diff); - } + qsort(query.performance_level, query.num_performance_levels, sizeof(NVCTRLPerformanceLevel), compare_mem_transfer_rate_max_asc); - if(xnvctrl_set_attribute_offset(&self->xnvctrl, self->num_performance_levels, target_transfer_rate_offset, NVCTRL_ATTRIB_GPU_MEM_TRANSFER_RATE)) { - fprintf(stderr, "gsr info: gsr_overclock_start: sucessfully set memory transfer rate offset to %d\n", target_transfer_rate_offset); - } else { - fprintf(stderr, "gsr info: gsr_overclock_start: failed to overclock memory transfer rate offset to %d\n", target_transfer_rate_offset); + int target_transfer_rate_offset = xnvctrl_get_attribute_max_value(&self->xnvctrl, query.num_performance_levels, NVCTRL_ATTRIB_GPU_MEM_TRANSFER_RATE); + if(query.num_performance_levels > 1) { + const int transfer_rate_max_diff = query.performance_level[query.num_performance_levels - 1].mem_transfer_rate_max - query.performance_level[query.num_performance_levels - 2].mem_transfer_rate_max; + target_transfer_rate_offset = min_int(target_transfer_rate_offset, transfer_rate_max_diff); + if(target_transfer_rate_offset >= 0 && xnvctrl_set_attribute_offset(&self->xnvctrl, self->num_performance_levels, target_transfer_rate_offset, NVCTRL_ATTRIB_GPU_MEM_TRANSFER_RATE)) { + fprintf(stderr, "gsr info: gsr_overclock_start: sucessfully set memory transfer rate offset to %d\n", target_transfer_rate_offset); + } else { + fprintf(stderr, "gsr info: gsr_overclock_start: failed to overclock memory transfer rate offset to %d\n", target_transfer_rate_offset); + } } + // TODO: Sort by nv_clock_max // TODO: Enable. Crashes on my system (gtx 1080) so it's disabled for now. Seems to crash even if graphics clock is increasd by 1, let alone 1200 /* - int target_nv_clock_offset = xnvctrl_get_attribute_max_value(&self->xnvctrl, &query, NVCTRL_GPU_NVCLOCK) / 2; // Divide by 2 just to be safe that we dont set it too high - if(query.num_performance_levels > 2) { - const int nv_clock_max_diff = query.performance_level[query.num_performance_levels - 1].nv_clock_max - query.performance_level[2].nv_clock_max; + int target_nv_clock_offset = xnvctrl_get_attribute_max_value(&self->xnvctrl, query.num_performance_levels, NVCTRL_GPU_NVCLOCK); + if(query.num_performance_levels > 1) { + const int nv_clock_max_diff = query.performance_level[query.num_performance_levels - 1].nv_clock_max - query.performance_level[query.num_performance_levels - 2].nv_clock_max; target_nv_clock_offset = min_int(target_nv_clock_offset, nv_clock_max_diff); - } - - if(xnvctrl_set_attribute_offset(&self->xnvctrl, self->num_performance_levels, target_nv_clock_offset, NVCTRL_GPU_NVCLOCK)) { - fprintf(stderr, "gsr info: gsr_overclock_start: sucessfully set nv clock offset to %d\n", target_nv_clock_offset); - } else { - fprintf(stderr, "gsr info: gsr_overclock_start: failed to overclock nv clock offset to %d\n", target_nv_clock_offset); + if(target_nv_clock_offset >= 0 && xnvctrl_set_attribute_offset(&self->xnvctrl, self->num_performance_levels, target_nv_clock_offset, NVCTRL_GPU_NVCLOCK)) { + fprintf(stderr, "gsr info: gsr_overclock_start: sucessfully set nv clock offset to %d\n", target_nv_clock_offset); + } else { + fprintf(stderr, "gsr info: gsr_overclock_start: failed to overclock nv clock offset to %d\n", target_nv_clock_offset); + } } */ diff --git a/src/pipewire_audio.c b/src/pipewire_audio.c new file mode 100644 index 0000000..5b920b4 --- /dev/null +++ b/src/pipewire_audio.c @@ -0,0 +1,808 @@ +#include "../include/pipewire_audio.h" + +#include <pipewire/pipewire.h> +#include <pipewire/extensions/metadata.h> +#include <pipewire/impl-module.h> + +static void on_core_info_cb(void *user_data, const struct pw_core_info *info) { + gsr_pipewire_audio *self = user_data; + //fprintf(stderr, "server name: %s\n", info->name); +} + +static void on_core_error_cb(void *user_data, uint32_t id, int seq, int res, const char *message) { + gsr_pipewire_audio *self = user_data; + //fprintf(stderr, "gsr error: pipewire: error id:%u seq:%d res:%d: %s\n", id, seq, res, message); + pw_thread_loop_signal(self->thread_loop, false); +} + +static void on_core_done_cb(void *user_data, uint32_t id, int seq) { + gsr_pipewire_audio *self = user_data; + if(id == PW_ID_CORE && self->server_version_sync == seq) + pw_thread_loop_signal(self->thread_loop, false); +} + +static const struct pw_core_events core_events = { + PW_VERSION_CORE_EVENTS, + .info = on_core_info_cb, + .done = on_core_done_cb, + .error = on_core_error_cb, +}; + +static gsr_pipewire_audio_node* gsr_pipewire_audio_get_node_by_name_case_insensitive(gsr_pipewire_audio *self, const char *node_name, gsr_pipewire_audio_node_type node_type) { + for(int i = 0; i < self->num_stream_nodes; ++i) { + const gsr_pipewire_audio_node *node = &self->stream_nodes[i]; + if(node->type == node_type && strcasecmp(node->name, node_name) == 0) + return &self->stream_nodes[i]; + } + return NULL; +} + +static gsr_pipewire_audio_port* gsr_pipewire_audio_get_node_port_by_name(gsr_pipewire_audio *self, uint32_t node_id, const char *port_name) { + for(int i = 0; i < self->num_ports; ++i) { + if(self->ports[i].node_id == node_id && strcmp(self->ports[i].name, port_name) == 0) + return &self->ports[i]; + } + return NULL; +} + +static bool requested_link_matches_name_case_insensitive(const gsr_pipewire_audio_requested_link *requested_link, const char *name) { + for(int i = 0; i < requested_link->num_outputs; ++i) { + if(requested_link->outputs[i].type == GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_STANDARD && strcasecmp(requested_link->outputs[i].name, name) == 0) + return true; + } + return false; +} + +static bool requested_link_has_type(const gsr_pipewire_audio_requested_link *requested_link, gsr_pipewire_audio_requested_type type) { + for(int i = 0; i < requested_link->num_outputs; ++i) { + if(requested_link->outputs[i].type == type) + return true; + } + return false; +} + +static void gsr_pipewire_get_node_input_port_by_type(gsr_pipewire_audio *self, const gsr_pipewire_audio_node *input_node, gsr_pipewire_audio_link_input_type input_type, + const gsr_pipewire_audio_port **input_fl_port, const gsr_pipewire_audio_port **input_fr_port) +{ + *input_fl_port = NULL; + *input_fr_port = NULL; + + switch(input_type) { + case GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_STREAM: { + *input_fl_port = gsr_pipewire_audio_get_node_port_by_name(self, input_node->id, "input_FL"); + *input_fr_port = gsr_pipewire_audio_get_node_port_by_name(self, input_node->id, "input_FR"); + break; + } + case GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_SINK: { + *input_fl_port = gsr_pipewire_audio_get_node_port_by_name(self, input_node->id, "playback_FL"); + *input_fr_port = gsr_pipewire_audio_get_node_port_by_name(self, input_node->id, "playback_FR"); + break; + } + } +} + +static void gsr_pipewire_get_node_output_port_by_type(gsr_pipewire_audio *self, const gsr_pipewire_audio_node *output_node, gsr_pipewire_audio_node_type output_type, + const gsr_pipewire_audio_port **output_fl_port, const gsr_pipewire_audio_port **output_fr_port) +{ + *output_fl_port = NULL; + *output_fr_port = NULL; + + switch(output_type) { + case GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_OUTPUT: + *output_fl_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "output_FL"); + *output_fr_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "output_FR"); + break; + case GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_INPUT: + *output_fl_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "monitor_FL"); + *output_fr_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "monitor_FR"); + break; + case GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE: { + *output_fl_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "monitor_FL"); + *output_fr_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "monitor_FR"); + if(!*output_fl_port || !*output_fr_port) { + *output_fl_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "capture_FL"); + *output_fr_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "capture_FR"); + } + if(!*output_fl_port || !*output_fr_port) { + const gsr_pipewire_audio_port *output_mono_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "monitor_MONO"); + if(!output_mono_port) + output_mono_port = gsr_pipewire_audio_get_node_port_by_name(self, output_node->id, "capture_MONO"); + + if(output_mono_port) { + *output_fl_port = output_mono_port; + *output_fr_port = output_mono_port; + } + } + break; + } + } +} + +static void gsr_pipewire_audio_establish_link(gsr_pipewire_audio *self, const gsr_pipewire_audio_port *input_fl_port, const gsr_pipewire_audio_port *input_fr_port, + const gsr_pipewire_audio_port *output_fl_port, const gsr_pipewire_audio_port *output_fr_port) +{ + // TODO: Detect if link already exists before so we dont create these proxies when not needed + + //fprintf(stderr, "linking!\n"); + // TODO: error check and cleanup + { + struct pw_properties *props = pw_properties_new(NULL, NULL); + pw_properties_setf(props, PW_KEY_LINK_OUTPUT_PORT, "%u", output_fl_port->id); + pw_properties_setf(props, PW_KEY_LINK_INPUT_PORT, "%u", input_fl_port->id); + // TODO: Clean this up when removing node + struct pw_proxy *proxy = pw_core_create_object(self->core, "link-factory", PW_TYPE_INTERFACE_Link, PW_VERSION_LINK, &props->dict, 0); + //self->server_version_sync = pw_core_sync(self->core, PW_ID_CORE, self->server_version_sync); + pw_properties_free(props); + } + + { + struct pw_properties *props = pw_properties_new(NULL, NULL); + pw_properties_setf(props, PW_KEY_LINK_OUTPUT_PORT, "%u", output_fr_port->id); + pw_properties_setf(props, PW_KEY_LINK_INPUT_PORT, "%u", input_fr_port->id); + // TODO: Clean this up when removing node + struct pw_proxy *proxy = pw_core_create_object(self->core, "link-factory", PW_TYPE_INTERFACE_Link, PW_VERSION_LINK, &props->dict, 0); + //self->server_version_sync = pw_core_sync(self->core, PW_ID_CORE, self->server_version_sync); + pw_properties_free(props); + } +} + +static void gsr_pipewire_audio_create_link(gsr_pipewire_audio *self, const gsr_pipewire_audio_requested_link *requested_link) { + const gsr_pipewire_audio_node_type requested_link_node_type = requested_link->input_type == GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_STREAM ? GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_INPUT : GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE; + const gsr_pipewire_audio_node *stream_input_node = gsr_pipewire_audio_get_node_by_name_case_insensitive(self, requested_link->input_name, requested_link_node_type); + if(!stream_input_node) + return; + + const gsr_pipewire_audio_port *input_fl_port = NULL; + const gsr_pipewire_audio_port *input_fr_port = NULL; + gsr_pipewire_get_node_input_port_by_type(self, stream_input_node, requested_link->input_type, &input_fl_port, &input_fr_port); + if(!input_fl_port || !input_fr_port) + return; + + for(int i = 0; i < self->num_stream_nodes; ++i) { + const gsr_pipewire_audio_node *output_node = &self->stream_nodes[i]; + if(output_node->type != requested_link->output_type) + continue; + + const bool requested_link_matches_app = requested_link_matches_name_case_insensitive(requested_link, output_node->name); + if(requested_link->inverted) { + if(requested_link_matches_app) + continue; + } else { + if(!requested_link_matches_app) + continue; + } + + const gsr_pipewire_audio_port *output_fl_port = NULL; + const gsr_pipewire_audio_port *output_fr_port = NULL; + gsr_pipewire_get_node_output_port_by_type(self, output_node, requested_link->output_type, &output_fl_port, &output_fr_port); + if(!output_fl_port || !output_fr_port) + continue; + + gsr_pipewire_audio_establish_link(self, input_fl_port, input_fr_port, output_fl_port, output_fr_port); + } +} + +static void gsr_pipewire_audio_create_links(gsr_pipewire_audio *self) { + for(int i = 0; i < self->num_requested_links; ++i) { + gsr_pipewire_audio_create_link(self, &self->requested_links[i]); + } +} + +static void gsr_pipewire_audio_create_link_for_default_devices(gsr_pipewire_audio *self, const gsr_pipewire_audio_requested_link *requested_link, gsr_pipewire_audio_requested_type default_device_type) { + if(default_device_type == GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_STANDARD) + return; + + const char *device_name = default_device_type == GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT ? self->default_output_device_name : self->default_input_device_name; + if(device_name[0] == '\0') + return; + + if(!requested_link_has_type(requested_link, default_device_type)) + return; + + const gsr_pipewire_audio_node_type requested_link_node_type = requested_link->input_type == GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_STREAM ? GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_INPUT : GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE; + const gsr_pipewire_audio_node *stream_input_node = gsr_pipewire_audio_get_node_by_name_case_insensitive(self, requested_link->input_name, requested_link_node_type); + if(!stream_input_node) + return; + + const gsr_pipewire_audio_port *input_fl_port = NULL; + const gsr_pipewire_audio_port *input_fr_port = NULL; + gsr_pipewire_get_node_input_port_by_type(self, stream_input_node, requested_link->input_type, &input_fl_port, &input_fr_port); + if(!input_fl_port || !input_fr_port) + return; + + const gsr_pipewire_audio_node *stream_output_node = gsr_pipewire_audio_get_node_by_name_case_insensitive(self, device_name, GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE); + if(!stream_output_node) + return; + + const gsr_pipewire_audio_port *output_fl_port = NULL; + const gsr_pipewire_audio_port *output_fr_port = NULL; + gsr_pipewire_get_node_output_port_by_type(self, stream_output_node, requested_link->output_type, &output_fl_port, &output_fr_port); + if(!output_fl_port || !output_fr_port) + return; + + gsr_pipewire_audio_establish_link(self, input_fl_port, input_fr_port, output_fl_port, output_fr_port); + //fprintf(stderr, "establishing a link from %u to %u\n", stream_output_node->id, stream_input_node->id); +} + +static void gsr_pipewire_audio_create_links_for_default_devices(gsr_pipewire_audio *self, gsr_pipewire_audio_requested_type default_device_type) { + for(int i = 0; i < self->num_requested_links; ++i) { + gsr_pipewire_audio_create_link_for_default_devices(self, &self->requested_links[i], default_device_type); + } +} + +static void gsr_pipewire_audio_destroy_links_by_output_to_input(gsr_pipewire_audio *self, uint32_t output_node_id, uint32_t input_node_id) { + for(int i = 0; i < self->num_links; ++i) { + if(self->links[i].output_node_id == output_node_id && self->links[i].input_node_id == input_node_id) + pw_registry_destroy(self->registry, self->links[i].id); + } +} + +static void gsr_pipewire_destroy_default_device_link(gsr_pipewire_audio *self, const gsr_pipewire_audio_requested_link *requested_link, gsr_pipewire_audio_requested_type default_device_type) { + if(default_device_type == GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_STANDARD) + return; + + const char *device_name = default_device_type == GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT ? self->default_output_device_name : self->default_input_device_name; + if(device_name[0] == '\0') + return; + + if(!requested_link_has_type(requested_link, default_device_type)) + return; + + /* default_output and default_input can be the same device. In that case both are the same link and we dont want to remove the link */ + const gsr_pipewire_audio_requested_type opposite_device_type = default_device_type == GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT ? GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_INPUT : GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT; + const char *opposite_device_name = opposite_device_type == GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT ? self->default_output_device_name : self->default_input_device_name; + if(requested_link_has_type(requested_link, opposite_device_type) && strcmp(device_name, opposite_device_name) == 0) + return; + + const gsr_pipewire_audio_node_type requested_link_node_type = requested_link->input_type == GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_STREAM ? GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_INPUT : GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE; + const gsr_pipewire_audio_node *stream_input_node = gsr_pipewire_audio_get_node_by_name_case_insensitive(self, requested_link->input_name, requested_link_node_type); + if(!stream_input_node) + return; + + const gsr_pipewire_audio_node *stream_output_node = gsr_pipewire_audio_get_node_by_name_case_insensitive(self, device_name, GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE); + if(!stream_output_node) + return; + + if(requested_link_matches_name_case_insensitive(requested_link, stream_output_node->name)) + return; + + gsr_pipewire_audio_destroy_links_by_output_to_input(self, stream_output_node->id, stream_input_node->id); + //fprintf(stderr, "destroying a link from %u to %u\n", stream_output_node->id, stream_input_node->id); +} + +static void gsr_pipewire_destroy_default_device_links(gsr_pipewire_audio *self, gsr_pipewire_audio_requested_type default_device_type) { + for(int i = 0; i < self->num_requested_links; ++i) { + gsr_pipewire_destroy_default_device_link(self, &self->requested_links[i], default_device_type); + } +} + +static bool json_get_value(const char *json_str, const char *key, char *value, size_t value_size) { + char key_full[32]; + const int key_full_size = snprintf(key_full, sizeof(key_full), "\"%s\":", key); + const char *start = strstr(json_str, key_full); + if(!start) + return false; + + start += key_full_size; + const char *value_start = strchr(start, '"'); + if(!value_start) + return false; + + value_start += 1; + const char *value_end = strchr(value_start, '"'); + if(!value_end) + return false; + + snprintf(value, value_size, "%.*s", (int)(value_end - value_start), value_start); + return true; +} + +static int on_metadata_property_cb(void *data, uint32_t id, const char *key, const char *type, const char *value) { + (void)type; + gsr_pipewire_audio *self = data; + + if(id == PW_ID_CORE && key && value) { + char value_decoded[128]; + if(strcmp(key, "default.audio.sink") == 0) { + if(json_get_value(value, "name", value_decoded, sizeof(value_decoded)) && strcmp(value_decoded, self->default_output_device_name) != 0) { + gsr_pipewire_destroy_default_device_links(self, GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT); + snprintf(self->default_output_device_name, sizeof(self->default_output_device_name), "%s", value_decoded); + gsr_pipewire_audio_create_links_for_default_devices(self, GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT); + } + } else if(strcmp(key, "default.audio.source") == 0) { + if(json_get_value(value, "name", value_decoded, sizeof(value_decoded)) && strcmp(value_decoded, self->default_input_device_name) != 0) { + gsr_pipewire_destroy_default_device_links(self, GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_INPUT); + snprintf(self->default_input_device_name, sizeof(self->default_input_device_name), "%s", value_decoded); + gsr_pipewire_audio_create_links_for_default_devices(self, GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_INPUT); + } + } + } + + return 0; +} + +static const struct pw_metadata_events metadata_events = { + PW_VERSION_METADATA_EVENTS, + .property = on_metadata_property_cb, +}; + +static void on_metadata_proxy_removed_cb(void *data) { + gsr_pipewire_audio *self = data; + if(self->metadata_proxy) { + pw_proxy_destroy(self->metadata_proxy); + self->metadata_proxy = NULL; + } +} + +static void on_metadata_proxy_destroy_cb(void *data) { + gsr_pipewire_audio *self = data; + + spa_hook_remove(&self->metadata_listener); + spa_hook_remove(&self->metadata_proxy_listener); + spa_zero(self->metadata_listener); + spa_zero(self->metadata_proxy_listener); + + self->metadata_proxy = NULL; +} + +static const struct pw_proxy_events metadata_proxy_events = { + PW_VERSION_PROXY_EVENTS, + .removed = on_metadata_proxy_removed_cb, + .destroy = on_metadata_proxy_destroy_cb, +}; + +static bool gsr_pipewire_audio_listen_on_metadata(gsr_pipewire_audio *self, uint32_t id) { + if(self->metadata_proxy) { + pw_proxy_destroy(self->metadata_proxy); + self->metadata_proxy = NULL; + } + + self->metadata_proxy = pw_registry_bind(self->registry, id, PW_TYPE_INTERFACE_Metadata, PW_VERSION_METADATA, 0); + if(!self->metadata_proxy) { + fprintf(stderr, "gsr error: gsr_pipewire_audio_listen_on_metadata: failed to bind to registry\n"); + return false; + } + + pw_proxy_add_object_listener(self->metadata_proxy, &self->metadata_listener, &metadata_events, self); + pw_proxy_add_listener(self->metadata_proxy, &self->metadata_proxy_listener, &metadata_proxy_events, self); + + self->server_version_sync = pw_core_sync(self->core, PW_ID_CORE, self->server_version_sync); + return true; +} + +static void registry_event_global(void *data, uint32_t id, uint32_t permissions, + const char *type, uint32_t version, + const struct spa_dict *props) +{ + //fprintf(stderr, "add: id: %d, type: %s\n", (int)id, type); + if(!props || !type) + return; + + //pw_properties_new_dict(props); + + gsr_pipewire_audio *self = (gsr_pipewire_audio*)data; + if(strcmp(type, PW_TYPE_INTERFACE_Node) == 0) { + const char *node_name = spa_dict_lookup(props, PW_KEY_NODE_NAME); + const char *media_class = spa_dict_lookup(props, PW_KEY_MEDIA_CLASS); + //fprintf(stderr, " node id: %u, node name: %s, media class: %s\n", id, node_name, media_class); + const bool is_stream_output = media_class && strcmp(media_class, "Stream/Output/Audio") == 0; + const bool is_stream_input = media_class && strcmp(media_class, "Stream/Input/Audio") == 0; + const bool is_sink = media_class && strcmp(media_class, "Audio/Sink") == 0; + const bool is_source = media_class && strcmp(media_class, "Audio/Source") == 0; + if(self->num_stream_nodes < GSR_PIPEWIRE_AUDIO_MAX_STREAM_NODES && node_name && (is_stream_output || is_stream_input || is_sink || is_source)) { + //const char *application_binary = spa_dict_lookup(props, PW_KEY_APP_PROCESS_BINARY); + //const char *application_name = spa_dict_lookup(props, PW_KEY_APP_NAME); + //fprintf(stderr, " node name: %s, app binary: %s, app name: %s\n", node_name, application_binary, application_name); + + char *node_name_copy = strdup(node_name); + if(node_name_copy) { + self->stream_nodes[self->num_stream_nodes].id = id; + self->stream_nodes[self->num_stream_nodes].name = node_name_copy; + if(is_stream_output) + self->stream_nodes[self->num_stream_nodes].type = GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_OUTPUT; + else if(is_stream_input) + self->stream_nodes[self->num_stream_nodes].type = GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_INPUT; + else if(is_sink || is_source) + self->stream_nodes[self->num_stream_nodes].type = GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE; + ++self->num_stream_nodes; + + gsr_pipewire_audio_create_links(self); + } + } else if(self->num_stream_nodes >= GSR_PIPEWIRE_AUDIO_MAX_STREAM_NODES) { + fprintf(stderr, "gsr error: reached the maximum amount of audio stream nodes\n"); + } + } else if(strcmp(type, PW_TYPE_INTERFACE_Port) == 0) { + const char *port_name = spa_dict_lookup(props, PW_KEY_PORT_NAME); + + const char *port_direction = spa_dict_lookup(props, PW_KEY_PORT_DIRECTION); + gsr_pipewire_audio_port_direction direction = -1; + if(port_direction && strcmp(port_direction, "in") == 0) + direction = GSR_PIPEWIRE_AUDIO_PORT_DIRECTION_INPUT; + else if(port_direction && strcmp(port_direction, "out") == 0) + direction = GSR_PIPEWIRE_AUDIO_PORT_DIRECTION_OUTPUT; + + const char *node_id = spa_dict_lookup(props, PW_KEY_NODE_ID); + const int node_id_num = node_id ? atoi(node_id) : 0; + + if(self->num_ports < GSR_PIPEWIRE_AUDIO_MAX_PORTS && port_name && direction >= 0 && node_id_num > 0) { + //fprintf(stderr, " port name: %s, node id: %d, direction: %s\n", port_name, node_id_num, port_direction); + char *port_name_copy = strdup(port_name); + if(port_name_copy) { + //fprintf(stderr, " port id: %u, node id: %u, name: %s\n", id, node_id_num, port_name_copy); + self->ports[self->num_ports].id = id; + self->ports[self->num_ports].node_id = node_id_num; + self->ports[self->num_ports].direction = direction; + self->ports[self->num_ports].name = port_name_copy; + ++self->num_ports; + + gsr_pipewire_audio_create_links(self); + } + } else if(self->num_ports >= GSR_PIPEWIRE_AUDIO_MAX_PORTS) { + fprintf(stderr, "gsr error: reached the maximum amount of audio ports\n"); + } + } else if(strcmp(type, PW_TYPE_INTERFACE_Link) == 0) { + const char *output_node = spa_dict_lookup(props, PW_KEY_LINK_OUTPUT_NODE); + const char *input_node = spa_dict_lookup(props, PW_KEY_LINK_INPUT_NODE); + + const uint32_t output_node_id_num = output_node ? atoi(output_node) : 0; + const uint32_t input_node_id_num = input_node ? atoi(input_node) : 0; + if(self->num_links < GSR_PIPEWIRE_AUDIO_MAX_LINKS && output_node_id_num > 0 && input_node_id_num > 0) { + //fprintf(stderr, " new link (%u): %u -> %u\n", id, output_node_id_num, input_node_id_num); + self->links[self->num_links].id = id; + self->links[self->num_links].output_node_id = output_node_id_num; + self->links[self->num_links].input_node_id = input_node_id_num; + ++self->num_links; + } else if(self->num_ports >= GSR_PIPEWIRE_AUDIO_MAX_LINKS) { + fprintf(stderr, "gsr error: reached the maximum amount of audio links\n"); + } + } else if(strcmp(type, PW_TYPE_INTERFACE_Metadata) == 0) { + const char *name = spa_dict_lookup(props, PW_KEY_METADATA_NAME); + if(name && strcmp(name, "default") == 0) + gsr_pipewire_audio_listen_on_metadata(self, id); + } +} + +static bool gsr_pipewire_audio_remove_node_by_id(gsr_pipewire_audio *self, uint32_t node_id) { + for(int i = 0; i < self->num_stream_nodes; ++i) { + if(self->stream_nodes[i].id != node_id) + continue; + + free(self->stream_nodes[i].name); + self->stream_nodes[i] = self->stream_nodes[self->num_stream_nodes - 1]; + --self->num_stream_nodes; + return true; + } + return false; +} + +static bool gsr_pipewire_audio_remove_port_by_id(gsr_pipewire_audio *self, uint32_t port_id) { + for(int i = 0; i < self->num_ports; ++i) { + if(self->ports[i].id != port_id) + continue; + + free(self->ports[i].name); + self->ports[i] = self->ports[self->num_ports - 1]; + --self->num_ports; + return true; + } + return false; +} + +static bool gsr_pipewire_audio_remove_link_by_id(gsr_pipewire_audio *self, uint32_t link_id) { + for(int i = 0; i < self->num_links; ++i) { + if(self->links[i].id != link_id) + continue; + + self->links[i] = self->links[self->num_links - 1]; + --self->num_links; + return true; + } + return false; +} + +static void registry_event_global_remove(void *data, uint32_t id) { + //fprintf(stderr, "remove: %d\n", (int)id); + gsr_pipewire_audio *self = (gsr_pipewire_audio*)data; + if(gsr_pipewire_audio_remove_node_by_id(self, id)) { + //fprintf(stderr, "removed node\n"); + return; + } + + if(gsr_pipewire_audio_remove_port_by_id(self, id)) { + //fprintf(stderr, "removed port\n"); + return; + } + + if(gsr_pipewire_audio_remove_link_by_id(self, id)) { + //fprintf(stderr, "removed link\n"); + return; + } +} + +static const struct pw_registry_events registry_events = { + PW_VERSION_REGISTRY_EVENTS, + .global = registry_event_global, + .global_remove = registry_event_global_remove, +}; + +bool gsr_pipewire_audio_init(gsr_pipewire_audio *self) { + memset(self, 0, sizeof(*self)); + + pw_init(NULL, NULL); + + self->thread_loop = pw_thread_loop_new("gsr screen capture", NULL); + if(!self->thread_loop) { + fprintf(stderr, "gsr error: gsr_pipewire_audio_init: failed to create pipewire thread\n"); + gsr_pipewire_audio_deinit(self); + return false; + } + + self->context = pw_context_new(pw_thread_loop_get_loop(self->thread_loop), NULL, 0); + if(!self->context) { + fprintf(stderr, "gsr error: gsr_pipewire_audio_init: failed to create pipewire context\n"); + gsr_pipewire_audio_deinit(self); + return false; + } + + pw_context_load_module(self->context, "libpipewire-module-link-factory", NULL, NULL); + + if(pw_thread_loop_start(self->thread_loop) < 0) { + fprintf(stderr, "gsr error: gsr_pipewire_audio_init: failed to start thread\n"); + gsr_pipewire_audio_deinit(self); + return false; + } + + pw_thread_loop_lock(self->thread_loop); + + self->core = pw_context_connect(self->context, pw_properties_new(PW_KEY_REMOTE_NAME, NULL, NULL), 0); + if(!self->core) { + pw_thread_loop_unlock(self->thread_loop); + gsr_pipewire_audio_deinit(self); + return false; + } + + // TODO: Error check + pw_core_add_listener(self->core, &self->core_listener, &core_events, self); + + self->registry = pw_core_get_registry(self->core, PW_VERSION_REGISTRY, 0); + pw_registry_add_listener(self->registry, &self->registry_listener, ®istry_events, self); + + self->server_version_sync = pw_core_sync(self->core, PW_ID_CORE, self->server_version_sync); + pw_thread_loop_wait(self->thread_loop); + + pw_thread_loop_unlock(self->thread_loop); + return true; +} + +void gsr_pipewire_audio_deinit(gsr_pipewire_audio *self) { + if(self->thread_loop) { + //pw_thread_loop_wait(self->thread_loop); + pw_thread_loop_stop(self->thread_loop); + } + + for(int i = 0; i < self->num_virtual_sink_proxies; ++i) { + if(self->virtual_sink_proxies[i]) { + pw_proxy_destroy(self->virtual_sink_proxies[i]); + self->virtual_sink_proxies[i] = NULL; + } + } + self->num_virtual_sink_proxies = 0; + + if(self->metadata_proxy) { + spa_hook_remove(&self->metadata_listener); + spa_hook_remove(&self->metadata_proxy_listener); + pw_proxy_destroy(self->metadata_proxy); + spa_zero(self->metadata_listener); + spa_zero(self->metadata_proxy_listener); + self->metadata_proxy = NULL; + } + + spa_hook_remove(&self->registry_listener); + spa_hook_remove(&self->core_listener); + + if(self->core) { + pw_core_disconnect(self->core); + self->core = NULL; + } + + if(self->context) { + pw_context_destroy(self->context); + self->context = NULL; + } + + if(self->thread_loop) { + pw_thread_loop_destroy(self->thread_loop); + self->thread_loop = NULL; + } + + for(int i = 0; i < self->num_stream_nodes; ++i) { + free(self->stream_nodes[i].name); + } + self->num_stream_nodes = 0; + + for(int i = 0; i < self->num_ports; ++i) { + free(self->ports[i].name); + } + self->num_ports = 0; + + self->num_links = 0; + + for(int i = 0; i < self->num_requested_links; ++i) { + for(int j = 0; j < self->requested_links[i].num_outputs; ++j) { + free(self->requested_links[i].outputs[j].name); + } + free(self->requested_links[i].outputs); + free(self->requested_links[i].input_name); + } + self->num_requested_links = 0; + +#if PW_CHECK_VERSION(0, 3, 49) + pw_deinit(); +#endif +} + +static struct pw_properties* gsr_pipewire_create_null_audio_sink(const char *name) { + char props_str[512]; + snprintf(props_str, sizeof(props_str), "{ factory.name=support.null-audio-sink node.name=\"%s\" media.class=Audio/Sink object.linger=false audio.position=[FL FR] monitor.channel-volumes=true monitor.passthrough=true adjust_time=0 node.description=gsr-app-sink slaves=\"\" }", name); + struct pw_properties *props = pw_properties_new_string(props_str); + if(!props) { + fprintf(stderr, "gsr error: gsr_pipewire_create_null_audio_sink: failed to create virtual sink properties\n"); + return NULL; + } + return props; +} + +bool gsr_pipewire_audio_create_virtual_sink(gsr_pipewire_audio *self, const char *name) { + if(self->num_virtual_sink_proxies == GSR_PIPEWIRE_AUDIO_MAX_VIRTUAL_SINKS) { + fprintf(stderr, "gsr error: gsr_pipewire_audio_create_virtual_sink: reached max number of virtual sinks\n"); + return false; + } + + pw_thread_loop_lock(self->thread_loop); + + struct pw_properties *virtual_sink_props = gsr_pipewire_create_null_audio_sink(name); + if(!virtual_sink_props) { + pw_thread_loop_unlock(self->thread_loop); + return false; + } + + struct pw_proxy *virtual_sink_proxy = pw_core_create_object(self->core, "adapter", PW_TYPE_INTERFACE_Node, PW_VERSION_NODE, &virtual_sink_props->dict, 0); + // TODO: + // If these are done then the above needs sizeof(*self) as the last argument + //pw_proxy_add_object_listener(virtual_sink_proxy, &pd->object_listener, &node_events, self); + //pw_proxy_add_listener(virtual_sink_proxy, &pd->proxy_listener, &proxy_events, self); + // TODO: proxy + pw_properties_free(virtual_sink_props); + if(!virtual_sink_proxy) { + fprintf(stderr, "gsr error: gsr_pipewire_audio_create_virtual_sink: failed to create virtual sink\n"); + pw_thread_loop_unlock(self->thread_loop); + return false; + } + + self->server_version_sync = pw_core_sync(self->core, PW_ID_CORE, self->server_version_sync); + pw_thread_loop_wait(self->thread_loop); + pw_thread_loop_unlock(self->thread_loop); + + self->virtual_sink_proxies[self->num_virtual_sink_proxies] = virtual_sink_proxy; + ++self->num_virtual_sink_proxies; + + return true; +} + +static bool string_remove_suffix(char *str, const char *suffix) { + int str_len = strlen(str); + int suffix_len = strlen(suffix); + if(str_len >= suffix_len && memcmp(str + str_len - suffix_len, suffix, suffix_len) == 0) { + str[str_len - suffix_len] = '\0'; + return true; + } else { + return false; + } +} + +static bool gsr_pipewire_audio_add_links_to_output(gsr_pipewire_audio *self, const char **output_names, int num_output_names, const char *input_name, gsr_pipewire_audio_node_type output_type, gsr_pipewire_audio_link_input_type input_type, bool inverted) { + if(self->num_requested_links >= GSR_PIPEWIRE_AUDIO_MAX_REQUESTED_LINKS) { + fprintf(stderr, "gsr error: reached the maximum amount of audio links\n"); + return false; + } + + gsr_pipewire_audio_requested_output *outputs = calloc(num_output_names, sizeof(gsr_pipewire_audio_requested_output)); + if(!outputs) + return false; + + char *input_name_copy = strdup(input_name); + if(!input_name_copy) + goto error; + + if(input_type == GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_SINK) + string_remove_suffix(input_name_copy, ".monitor"); + + for(int i = 0; i < num_output_names; ++i) { + outputs[i].name = strdup(output_names[i]); + if(!outputs[i].name) + goto error; + + outputs[i].type = GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_STANDARD; + if(output_type == GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE) { + string_remove_suffix(outputs[i].name, ".monitor"); + + if(strcmp(outputs[i].name, "default_output") == 0) + outputs[i].type = GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT; + else if(strcmp(outputs[i].name, "default_input") == 0) + outputs[i].type = GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_INPUT; + else + outputs[i].type = GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_STANDARD; + } + } + + pw_thread_loop_lock(self->thread_loop); + self->requested_links[self->num_requested_links].outputs = outputs; + self->requested_links[self->num_requested_links].num_outputs = num_output_names; + self->requested_links[self->num_requested_links].input_name = input_name_copy; + self->requested_links[self->num_requested_links].output_type = output_type; + self->requested_links[self->num_requested_links].input_type = input_type; + self->requested_links[self->num_requested_links].inverted = inverted; + ++self->num_requested_links; + gsr_pipewire_audio_create_link(self, &self->requested_links[self->num_requested_links - 1]); + gsr_pipewire_audio_create_link_for_default_devices(self, &self->requested_links[self->num_requested_links - 1], GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_OUTPUT); + gsr_pipewire_audio_create_link_for_default_devices(self, &self->requested_links[self->num_requested_links - 1], GSR_PIPEWIRE_AUDIO_REQUESTED_TYPE_DEFAULT_INPUT); + pw_thread_loop_unlock(self->thread_loop); + + return true; + + error: + free(input_name_copy); + for(int i = 0; i < num_output_names; ++i) { + free(outputs[i].name); + } + free(outputs); + return false; +} + +bool gsr_pipewire_audio_add_link_from_apps_to_stream(gsr_pipewire_audio *self, const char **app_names, int num_app_names, const char *stream_name_input) { + return gsr_pipewire_audio_add_links_to_output(self, app_names, num_app_names, stream_name_input, GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_OUTPUT, GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_STREAM, false); +} + +bool gsr_pipewire_audio_add_link_from_apps_to_stream_inverted(gsr_pipewire_audio *self, const char **app_names, int num_app_names, const char *stream_name_input) { + return gsr_pipewire_audio_add_links_to_output(self, app_names, num_app_names, stream_name_input, GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_OUTPUT, GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_STREAM, true); +} + +bool gsr_pipewire_audio_add_link_from_apps_to_sink(gsr_pipewire_audio *self, const char **app_names, int num_app_names, const char *sink_name_input) { + return gsr_pipewire_audio_add_links_to_output(self, app_names, num_app_names, sink_name_input, GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_OUTPUT, GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_SINK, false); +} + +bool gsr_pipewire_audio_add_link_from_apps_to_sink_inverted(gsr_pipewire_audio *self, const char **app_names, int num_app_names, const char *sink_name_input) { + return gsr_pipewire_audio_add_links_to_output(self, app_names, num_app_names, sink_name_input, GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_OUTPUT, GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_SINK, true); +} + +bool gsr_pipewire_audio_add_link_from_sources_to_sink(gsr_pipewire_audio *self, const char **source_names, int num_source_names, const char *sink_name_input) { + return gsr_pipewire_audio_add_links_to_output(self, source_names, num_source_names, sink_name_input, GSR_PIPEWIRE_AUDIO_NODE_TYPE_SINK_OR_SOURCE, GSR_PIPEWIRE_AUDIO_LINK_INPUT_TYPE_SINK, false); +} + +void gsr_pipewire_audio_for_each_app(gsr_pipewire_audio *self, gsr_pipewire_audio_app_query_callback callback, void *userdata) { + pw_thread_loop_lock(self->thread_loop); + for(int i = 0; i < self->num_stream_nodes; ++i) { + const gsr_pipewire_audio_node *node = &self->stream_nodes[i]; + if(node->type != GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_OUTPUT) + continue; + + bool duplicate_app = false; + for(int j = i - 1; j >= 0; --j) { + const gsr_pipewire_audio_node *prev_node = &self->stream_nodes[j]; + if(prev_node->type != GSR_PIPEWIRE_AUDIO_NODE_TYPE_STREAM_OUTPUT) + continue; + + if(strcasecmp(node->name, prev_node->name) == 0) { + duplicate_app = true; + break; + } + } + + if(duplicate_app) + continue; + + if(!callback(node->name, userdata)) + break; + } + pw_thread_loop_unlock(self->thread_loop); +} diff --git a/src/pipewire_video.c b/src/pipewire_video.c new file mode 100644 index 0000000..bbc3f5d --- /dev/null +++ b/src/pipewire_video.c @@ -0,0 +1,824 @@ +#include "../include/pipewire_video.h" +#include "../include/egl.h" +#include "../include/utils.h" + +#include <pipewire/pipewire.h> +#include <spa/param/video/format-utils.h> +#include <spa/debug/types.h> + +#include <libdrm/drm_fourcc.h> + +#include <fcntl.h> +#include <unistd.h> + +/* This code is partially based on xr-video-player pipewire implementation which is based on obs-studio's pipewire implementation */ + +/* TODO: Make gsr_pipewire_video_init asynchronous */ +/* TODO: Support hdr when pipewire supports it */ +/* TODO: Test all of the image formats */ + +#ifndef SPA_POD_PROP_FLAG_DONT_FIXATE +#define SPA_POD_PROP_FLAG_DONT_FIXATE (1 << 4) +#endif + +#define CURSOR_META_SIZE(width, height) \ + (sizeof(struct spa_meta_cursor) + sizeof(struct spa_meta_bitmap) + \ + width * height * 4) + +static bool parse_pw_version(gsr_pipewire_video_data_version *dst, const char *version) { + const int n_matches = sscanf(version, "%d.%d.%d", &dst->major, &dst->minor, &dst->micro); + return n_matches == 3; +} + +static bool check_pw_version(const gsr_pipewire_video_data_version *pw_version, int major, int minor, int micro) { + if (pw_version->major != major) + return pw_version->major > major; + if (pw_version->minor != minor) + return pw_version->minor > minor; + return pw_version->micro >= micro; +} + +static void update_pw_versions(gsr_pipewire_video *self, const char *version) { + fprintf(stderr, "gsr info: pipewire: server version: %s\n", version); + fprintf(stderr, "gsr info: pipewire: library version: %s\n", pw_get_library_version()); + fprintf(stderr, "gsr info: pipewire: header version: %s\n", pw_get_headers_version()); + if(!parse_pw_version(&self->server_version, version)) + fprintf(stderr, "gsr error: pipewire: failed to parse server version\n"); +} + +static void on_core_info_cb(void *user_data, const struct pw_core_info *info) { + gsr_pipewire_video *self = user_data; + update_pw_versions(self, info->version); +} + +static void on_core_error_cb(void *user_data, uint32_t id, int seq, int res, const char *message) { + gsr_pipewire_video *self = user_data; + fprintf(stderr, "gsr error: pipewire: error id:%u seq:%d res:%d: %s\n", id, seq, res, message); + pw_thread_loop_signal(self->thread_loop, false); +} + +static void on_core_done_cb(void *user_data, uint32_t id, int seq) { + gsr_pipewire_video *self = user_data; + if (id == PW_ID_CORE && self->server_version_sync == seq) + pw_thread_loop_signal(self->thread_loop, false); +} + +static bool is_cursor_format_supported(const enum spa_video_format format) { + switch(format) { + case SPA_VIDEO_FORMAT_RGBx: return true; + case SPA_VIDEO_FORMAT_BGRx: return true; + case SPA_VIDEO_FORMAT_RGBA: return true; + case SPA_VIDEO_FORMAT_BGRA: return true; + case SPA_VIDEO_FORMAT_RGB: return true; + case SPA_VIDEO_FORMAT_BGR: return true; + case SPA_VIDEO_FORMAT_ARGB: return true; + case SPA_VIDEO_FORMAT_ABGR: return true; +#if PW_CHECK_VERSION(0, 3, 41) + case SPA_VIDEO_FORMAT_xRGB_210LE: return true; + case SPA_VIDEO_FORMAT_xBGR_210LE: return true; + case SPA_VIDEO_FORMAT_ARGB_210LE: return true; + case SPA_VIDEO_FORMAT_ABGR_210LE: return true; +#endif + default: break; + } + return false; +} + +static const struct pw_core_events core_events = { + PW_VERSION_CORE_EVENTS, + .info = on_core_info_cb, + .done = on_core_done_cb, + .error = on_core_error_cb, +}; + +static void on_process_cb(void *user_data) { + gsr_pipewire_video *self = user_data; + struct spa_meta_cursor *cursor = NULL; + //struct spa_meta *video_damage = NULL; + + /* Find the most recent buffer */ + struct pw_buffer *pw_buf = NULL; + for(;;) { + struct pw_buffer *aux = pw_stream_dequeue_buffer(self->stream); + if(!aux) + break; + if(pw_buf) + pw_stream_queue_buffer(self->stream, pw_buf); + pw_buf = aux; + } + + if(!pw_buf) { + fprintf(stderr, "gsr info: pipewire: out of buffers!\n"); + return; + } + + struct spa_buffer *buffer = pw_buf->buffer; + const bool has_buffer = buffer->datas[0].chunk->size != 0; + if(!has_buffer) + goto read_metadata; + + pthread_mutex_lock(&self->mutex); + + if(buffer->datas[0].type == SPA_DATA_DmaBuf) { + for(size_t i = 0; i < self->dmabuf_num_planes; ++i) { + if(self->dmabuf_data[i].fd > 0) { + close(self->dmabuf_data[i].fd); + self->dmabuf_data[i].fd = -1; + } + } + + self->dmabuf_num_planes = buffer->n_datas; + if(self->dmabuf_num_planes > GSR_PIPEWIRE_VIDEO_DMABUF_MAX_PLANES) + self->dmabuf_num_planes = GSR_PIPEWIRE_VIDEO_DMABUF_MAX_PLANES; + + for(size_t i = 0; i < self->dmabuf_num_planes; ++i) { + self->dmabuf_data[i].fd = dup(buffer->datas[i].fd); + self->dmabuf_data[i].offset = buffer->datas[i].chunk->offset; + self->dmabuf_data[i].stride = buffer->datas[i].chunk->stride; + } + + self->damaged = true; + } else { + // TODO: + } + + // TODO: Move down to read_metadata + struct spa_meta_region *region = spa_buffer_find_meta_data(buffer, SPA_META_VideoCrop, sizeof(*region)); + if(region && spa_meta_region_is_valid(region)) { + // fprintf(stderr, "gsr info: pipewire: crop Region available (%dx%d+%d+%d)\n", + // region->region.position.x, region->region.position.y, + // region->region.size.width, region->region.size.height); + self->crop.x = region->region.position.x; + self->crop.y = region->region.position.y; + self->crop.width = region->region.size.width; + self->crop.height = region->region.size.height; + self->crop.valid = true; + } else { + self->crop.valid = false; + } + + pthread_mutex_unlock(&self->mutex); + +read_metadata: + + // video_damage = spa_buffer_find_meta(buffer, SPA_META_VideoDamage); + // if(video_damage) { + // struct spa_meta_region *r = spa_meta_first(video_damage); + // if(spa_meta_check(r, video_damage)) { + // //fprintf(stderr, "damage: %d,%d %ux%u\n", r->region.position.x, r->region.position.y, r->region.size.width, r->region.size.height); + // pthread_mutex_lock(&self->mutex); + // self->damaged = true; + // pthread_mutex_unlock(&self->mutex); + // } + // } + + cursor = spa_buffer_find_meta_data(buffer, SPA_META_Cursor, sizeof(*cursor)); + self->cursor.valid = cursor && spa_meta_cursor_is_valid(cursor); + + if (self->cursor.visible && self->cursor.valid) { + pthread_mutex_lock(&self->mutex); + + struct spa_meta_bitmap *bitmap = NULL; + if (cursor->bitmap_offset) + bitmap = SPA_MEMBER(cursor, cursor->bitmap_offset, struct spa_meta_bitmap); + + if (bitmap && bitmap->size.width > 0 && bitmap->size.height && is_cursor_format_supported(bitmap->format)) { + const uint8_t *bitmap_data = SPA_MEMBER(bitmap, bitmap->offset, uint8_t); + fprintf(stderr, "gsr info: pipewire: cursor bitmap update, size: %dx%d, format: %s\n", + (int)bitmap->size.width, (int)bitmap->size.height, spa_debug_type_find_name(spa_type_video_format, bitmap->format)); + + const size_t bitmap_size = bitmap->size.width * bitmap->size.height * 4; + uint8_t *new_bitmap_data = realloc(self->cursor.data, bitmap_size); + if(new_bitmap_data) { + self->cursor.data = new_bitmap_data; + /* TODO: Convert bgr and other image formats to rgb here */ + memcpy(self->cursor.data, bitmap_data, bitmap_size); + } + + self->cursor.hotspot_x = cursor->hotspot.x; + self->cursor.hotspot_y = cursor->hotspot.y; + self->cursor.width = bitmap->size.width; + self->cursor.height = bitmap->size.height; + } + + self->cursor.x = cursor->position.x; + self->cursor.y = cursor->position.y; + pthread_mutex_unlock(&self->mutex); + + //fprintf(stderr, "gsr info: pipewire: cursor: %d %d %d %d\n", cursor->hotspot.x, cursor->hotspot.y, cursor->position.x, cursor->position.y); + } + + pw_stream_queue_buffer(self->stream, pw_buf); +} + +static void on_param_changed_cb(void *user_data, uint32_t id, const struct spa_pod *param) { + gsr_pipewire_video *self = user_data; + + if (!param || id != SPA_PARAM_Format) + return; + + int result = spa_format_parse(param, &self->format.media_type, &self->format.media_subtype); + if (result < 0) + return; + + if (self->format.media_type != SPA_MEDIA_TYPE_video || self->format.media_subtype != SPA_MEDIA_SUBTYPE_raw) + return; + + pthread_mutex_lock(&self->mutex); + spa_format_video_raw_parse(param, &self->format.info.raw); + pthread_mutex_unlock(&self->mutex); + + uint32_t buffer_types = 0; + const bool has_modifier = spa_pod_find_prop(param, NULL, SPA_FORMAT_VIDEO_modifier) != NULL; + if(has_modifier || check_pw_version(&self->server_version, 0, 3, 24)) + buffer_types |= 1 << SPA_DATA_DmaBuf; + + fprintf(stderr, "gsr info: pipewire: negotiated format:\n"); + + fprintf(stderr, "gsr info: pipewire: Format: %d (%s)\n", + self->format.info.raw.format, + spa_debug_type_find_name(spa_type_video_format, self->format.info.raw.format)); + + if(has_modifier) { + fprintf(stderr, "gsr info: pipewire: Modifier: 0x%" PRIx64 "\n", self->format.info.raw.modifier); + } + + fprintf(stderr, "gsr info: pipewire: Size: %dx%d\n", self->format.info.raw.size.width, self->format.info.raw.size.height); + fprintf(stderr, "gsr info: pipewire: Framerate: %d/%d\n", self->format.info.raw.framerate.num, self->format.info.raw.framerate.denom); + + uint8_t params_buffer[1024]; + struct spa_pod_builder pod_builder = SPA_POD_BUILDER_INIT(params_buffer, sizeof(params_buffer)); + const struct spa_pod *params[4]; + + params[0] = spa_pod_builder_add_object( + &pod_builder, SPA_TYPE_OBJECT_ParamMeta, SPA_PARAM_Meta, + SPA_PARAM_META_type, SPA_POD_Id(SPA_META_VideoCrop), + SPA_PARAM_META_size, + SPA_POD_Int(sizeof(struct spa_meta_region))); + + params[1] = spa_pod_builder_add_object( + &pod_builder, SPA_TYPE_OBJECT_ParamMeta, SPA_PARAM_Meta, + SPA_PARAM_META_type, SPA_POD_Id(SPA_META_VideoDamage), + SPA_PARAM_META_size, SPA_POD_CHOICE_RANGE_Int( + sizeof(struct spa_meta_region) * 16, + sizeof(struct spa_meta_region) * 1, + sizeof(struct spa_meta_region) * 16)); + + params[2] = spa_pod_builder_add_object( + &pod_builder, SPA_TYPE_OBJECT_ParamMeta, SPA_PARAM_Meta, + SPA_PARAM_META_type, SPA_POD_Id(SPA_META_Cursor), + SPA_PARAM_META_size, + SPA_POD_CHOICE_RANGE_Int(CURSOR_META_SIZE(64, 64), + CURSOR_META_SIZE(1, 1), + CURSOR_META_SIZE(1024, 1024))); + + params[3] = spa_pod_builder_add_object( + &pod_builder, SPA_TYPE_OBJECT_ParamBuffers, SPA_PARAM_Buffers, + SPA_PARAM_BUFFERS_dataType, SPA_POD_Int(buffer_types)); + + pw_stream_update_params(self->stream, params, 4); + self->negotiated = true; +} + +static void on_state_changed_cb(void *user_data, enum pw_stream_state old, enum pw_stream_state state, const char *error) { + (void)old; + gsr_pipewire_video *self = user_data; + + fprintf(stderr, "gsr info: pipewire: stream %p state: \"%s\" (error: %s)\n", + (void*)self->stream, pw_stream_state_as_string(state), + error ? error : "none"); +} + +static const struct pw_stream_events stream_events = { + PW_VERSION_STREAM_EVENTS, + .state_changed = on_state_changed_cb, + .param_changed = on_param_changed_cb, + .process = on_process_cb, +}; + +static inline struct spa_pod *build_format(struct spa_pod_builder *b, + const gsr_pipewire_video_video_info *ovi, + uint32_t format, const uint64_t *modifiers, + size_t modifier_count) +{ + struct spa_pod_frame format_frame; + + spa_pod_builder_push_object(b, &format_frame, SPA_TYPE_OBJECT_Format, SPA_PARAM_EnumFormat); + spa_pod_builder_add(b, SPA_FORMAT_mediaType, SPA_POD_Id(SPA_MEDIA_TYPE_video), 0); + spa_pod_builder_add(b, SPA_FORMAT_mediaSubtype, SPA_POD_Id(SPA_MEDIA_SUBTYPE_raw), 0); + + spa_pod_builder_add(b, SPA_FORMAT_VIDEO_format, SPA_POD_Id(format), 0); + + if (modifier_count > 0) { + struct spa_pod_frame modifier_frame; + + spa_pod_builder_prop(b, SPA_FORMAT_VIDEO_modifier, SPA_POD_PROP_FLAG_MANDATORY | SPA_POD_PROP_FLAG_DONT_FIXATE); + spa_pod_builder_push_choice(b, &modifier_frame, SPA_CHOICE_Enum, 0); + + /* The first element of choice pods is the preferred value. Here + * we arbitrarily pick the first modifier as the preferred one. + */ + // TODO: + spa_pod_builder_long(b, modifiers[0]); + + for(uint32_t i = 0; i < modifier_count; i++) + spa_pod_builder_long(b, modifiers[i]); + + spa_pod_builder_pop(b, &modifier_frame); + } + + spa_pod_builder_add(b, SPA_FORMAT_VIDEO_size, + SPA_POD_CHOICE_RANGE_Rectangle( + &SPA_RECTANGLE(32, 32), + &SPA_RECTANGLE(1, 1), + &SPA_RECTANGLE(16384, 16384)), + SPA_FORMAT_VIDEO_framerate, + SPA_POD_CHOICE_RANGE_Fraction( + &SPA_FRACTION(ovi->fps_num, ovi->fps_den), + &SPA_FRACTION(0, 1), &SPA_FRACTION(500, 1)), + 0); + return spa_pod_builder_pop(b, &format_frame); +} + +/* https://gstreamer.freedesktop.org/documentation/additional/design/mediatype-video-raw.html?gi-language=c#formats */ +/* For some reason gstreamer formats are in opposite order to drm formats */ +static int64_t spa_video_format_to_drm_format(const enum spa_video_format format) { + switch(format) { + case SPA_VIDEO_FORMAT_RGBx: return DRM_FORMAT_XBGR8888; + case SPA_VIDEO_FORMAT_BGRx: return DRM_FORMAT_XRGB8888; + case SPA_VIDEO_FORMAT_RGBA: return DRM_FORMAT_ABGR8888; + case SPA_VIDEO_FORMAT_BGRA: return DRM_FORMAT_ARGB8888; + case SPA_VIDEO_FORMAT_RGB: return DRM_FORMAT_XBGR8888; + case SPA_VIDEO_FORMAT_BGR: return DRM_FORMAT_XRGB8888; + case SPA_VIDEO_FORMAT_ARGB: return DRM_FORMAT_XRGB8888; + case SPA_VIDEO_FORMAT_ABGR: return DRM_FORMAT_XRGB8888; +#if PW_CHECK_VERSION(0, 3, 41) + case SPA_VIDEO_FORMAT_xRGB_210LE: return DRM_FORMAT_XRGB2101010; + case SPA_VIDEO_FORMAT_xBGR_210LE: return DRM_FORMAT_XBGR2101010; + case SPA_VIDEO_FORMAT_ARGB_210LE: return DRM_FORMAT_ARGB2101010; + case SPA_VIDEO_FORMAT_ABGR_210LE: return DRM_FORMAT_ABGR2101010; +#endif + default: break; + } + return DRM_FORMAT_INVALID; +} + +#if PW_CHECK_VERSION(0, 3, 41) +#define GSR_PIPEWIRE_VIDEO_NUM_VIDEO_FORMATS GSR_PIPEWIRE_VIDEO_MAX_VIDEO_FORMATS +#else +#define GSR_PIPEWIRE_VIDEO_NUM_VIDEO_FORMATS 8 +#endif + +static const enum spa_video_format video_formats[GSR_PIPEWIRE_VIDEO_MAX_VIDEO_FORMATS] = { + SPA_VIDEO_FORMAT_BGRA, + SPA_VIDEO_FORMAT_BGRx, + SPA_VIDEO_FORMAT_BGR, + SPA_VIDEO_FORMAT_RGBx, + SPA_VIDEO_FORMAT_RGBA, + SPA_VIDEO_FORMAT_RGB, + SPA_VIDEO_FORMAT_ARGB, + SPA_VIDEO_FORMAT_ABGR, +#if PW_CHECK_VERSION(0, 3, 41) + SPA_VIDEO_FORMAT_xRGB_210LE, + SPA_VIDEO_FORMAT_xBGR_210LE, + SPA_VIDEO_FORMAT_ARGB_210LE, + SPA_VIDEO_FORMAT_ABGR_210LE +#endif +}; + +static bool gsr_pipewire_video_build_format_params(gsr_pipewire_video *self, struct spa_pod_builder *pod_builder, struct spa_pod **params, uint32_t *num_params) { + *num_params = 0; + + if(!check_pw_version(&self->server_version, 0, 3, 33)) + return false; + + for(size_t i = 0; i < GSR_PIPEWIRE_VIDEO_NUM_VIDEO_FORMATS; i++) { + if(self->supported_video_formats[i].modifiers_size == 0) + continue; + params[*num_params] = build_format(pod_builder, &self->video_info, self->supported_video_formats[i].format, self->modifiers + self->supported_video_formats[i].modifiers_index, self->supported_video_formats[i].modifiers_size); + ++(*num_params); + } + + return true; +} + +static void renegotiate_format(void *data, uint64_t expirations) { + (void)expirations; + gsr_pipewire_video *self = (gsr_pipewire_video*)data; + + pw_thread_loop_lock(self->thread_loop); + + struct spa_pod *params[GSR_PIPEWIRE_VIDEO_NUM_VIDEO_FORMATS]; + uint32_t num_video_formats = 0; + uint8_t params_buffer[4096]; + struct spa_pod_builder pod_builder = SPA_POD_BUILDER_INIT(params_buffer, sizeof(params_buffer)); + if (!gsr_pipewire_video_build_format_params(self, &pod_builder, params, &num_video_formats)) { + fprintf(stderr, "gsr error: renegotiate_format: failed to build formats\n"); + pw_thread_loop_unlock(self->thread_loop); + return; + } + + pw_stream_update_params(self->stream, (const struct spa_pod**)params, num_video_formats); + pw_thread_loop_unlock(self->thread_loop); +} + +static bool spa_video_format_get_modifiers(gsr_pipewire_video *self, const enum spa_video_format format, uint64_t *modifiers, int32_t max_modifiers, int32_t *num_modifiers) { + *num_modifiers = 0; + + if(max_modifiers == 0) { + fprintf(stderr, "gsr error: spa_video_format_get_modifiers: no space for modifiers left\n"); + //modifiers[0] = DRM_FORMAT_MOD_LINEAR; + //modifiers[1] = DRM_FORMAT_MOD_INVALID; + //*num_modifiers = 2; + return false; + } + + if(!self->egl->eglQueryDmaBufModifiersEXT) { + fprintf(stderr, "gsr error: spa_video_format_get_modifiers: failed to initialize modifiers because eglQueryDmaBufModifiersEXT is not available\n"); + //modifiers[0] = DRM_FORMAT_MOD_LINEAR; + //modifiers[1] = DRM_FORMAT_MOD_INVALID; + //*num_modifiers = 2; + return false; + } + + const int64_t drm_format = spa_video_format_to_drm_format(format); + if(drm_format == DRM_FORMAT_INVALID) { + fprintf(stderr, "gsr error: spa_video_format_get_modifiers: unsupported format: %d\n", (int)format); + return false; + } + + if(!self->egl->eglQueryDmaBufModifiersEXT(self->egl->egl_display, drm_format, max_modifiers, modifiers, NULL, num_modifiers)) { + fprintf(stderr, "gsr error: spa_video_format_get_modifiers: eglQueryDmaBufModifiersEXT failed with drm format %d, %" PRIi64 "\n", (int)format, drm_format); + //modifiers[0] = DRM_FORMAT_MOD_LINEAR; + //modifiers[1] = DRM_FORMAT_MOD_INVALID; + //*num_modifiers = 2; + *num_modifiers = 0; + return false; + } + + // if(*num_modifiers + 2 <= max_modifiers) { + // modifiers[*num_modifiers + 0] = DRM_FORMAT_MOD_LINEAR; + // modifiers[*num_modifiers + 1] = DRM_FORMAT_MOD_INVALID; + // *num_modifiers += 2; + // } + return true; +} + +static void gsr_pipewire_video_init_modifiers(gsr_pipewire_video *self) { + for(size_t i = 0; i < GSR_PIPEWIRE_VIDEO_NUM_VIDEO_FORMATS; i++) { + self->supported_video_formats[i].format = video_formats[i]; + int32_t num_modifiers = 0; + spa_video_format_get_modifiers(self, self->supported_video_formats[i].format, self->modifiers + self->num_modifiers, GSR_PIPEWIRE_VIDEO_MAX_MODIFIERS - self->num_modifiers, &num_modifiers); + self->supported_video_formats[i].modifiers_index = self->num_modifiers; + self->supported_video_formats[i].modifiers_size = num_modifiers; + } +} + +static bool gsr_pipewire_video_setup_stream(gsr_pipewire_video *self) { + struct spa_pod *params[GSR_PIPEWIRE_VIDEO_NUM_VIDEO_FORMATS]; + uint32_t num_video_formats = 0; + uint8_t params_buffer[4096]; + struct spa_pod_builder pod_builder = SPA_POD_BUILDER_INIT(params_buffer, sizeof(params_buffer)); + + self->thread_loop = pw_thread_loop_new("gsr screen capture", NULL); + if(!self->thread_loop) { + fprintf(stderr, "gsr error: gsr_pipewire_video_setup_stream: failed to create pipewire thread\n"); + goto error; + } + + self->context = pw_context_new(pw_thread_loop_get_loop(self->thread_loop), NULL, 0); + if(!self->context) { + fprintf(stderr, "gsr error: gsr_pipewire_video_setup_stream: failed to create pipewire context\n"); + goto error; + } + + if(pw_thread_loop_start(self->thread_loop) < 0) { + fprintf(stderr, "gsr error: gsr_pipewire_video_setup_stream: failed to start thread\n"); + goto error; + } + + pw_thread_loop_lock(self->thread_loop); + + // TODO: Why pass 5 to fcntl? + self->core = pw_context_connect_fd(self->context, fcntl(self->fd, F_DUPFD_CLOEXEC, 5), NULL, 0); + if(!self->core) { + pw_thread_loop_unlock(self->thread_loop); + fprintf(stderr, "gsr error: gsr_pipewire_video_setup_stream: failed to connect to fd %d\n", self->fd); + goto error; + } + + // TODO: Error check + pw_core_add_listener(self->core, &self->core_listener, &core_events, self); + + self->server_version_sync = pw_core_sync(self->core, PW_ID_CORE, 0); + pw_thread_loop_wait(self->thread_loop); + + gsr_pipewire_video_init_modifiers(self); + + // TODO: Cleanup? + self->reneg = pw_loop_add_event(pw_thread_loop_get_loop(self->thread_loop), renegotiate_format, self); + if(!self->reneg) { + pw_thread_loop_unlock(self->thread_loop); + fprintf(stderr, "gsr error: gsr_pipewire_video_setup_stream: pw_loop_add_event failed\n"); + goto error; + } + + self->stream = pw_stream_new(self->core, "com.dec05eba.gpu_screen_recorder", + pw_properties_new(PW_KEY_MEDIA_TYPE, "Video", + PW_KEY_MEDIA_CATEGORY, "Capture", + PW_KEY_MEDIA_ROLE, "Screen", NULL)); + if(!self->stream) { + pw_thread_loop_unlock(self->thread_loop); + fprintf(stderr, "gsr error: gsr_pipewire_video_setup_stream: failed to create stream\n"); + goto error; + } + pw_stream_add_listener(self->stream, &self->stream_listener, &stream_events, self); + + if(!gsr_pipewire_video_build_format_params(self, &pod_builder, params, &num_video_formats)) { + pw_thread_loop_unlock(self->thread_loop); + fprintf(stderr, "gsr error: gsr_pipewire_video_setup_stream: failed to build format params\n"); + goto error; + } + + if(pw_stream_connect( + self->stream, PW_DIRECTION_INPUT, self->node, + PW_STREAM_FLAG_AUTOCONNECT | PW_STREAM_FLAG_MAP_BUFFERS, (const struct spa_pod**)params, + num_video_formats) < 0) + { + pw_thread_loop_unlock(self->thread_loop); + fprintf(stderr, "gsr error: gsr_pipewire_video_setup_stream: failed to connect stream\n"); + goto error; + } + + pw_thread_loop_unlock(self->thread_loop); + return true; + + error: + if(self->thread_loop) { + //pw_thread_loop_wait(self->thread_loop); + pw_thread_loop_stop(self->thread_loop); + } + + if(self->stream) { + pw_stream_disconnect(self->stream); + pw_stream_destroy(self->stream); + self->stream = NULL; + } + + if(self->core) { + pw_core_disconnect(self->core); + self->core = NULL; + } + + if(self->context) { + pw_context_destroy(self->context); + self->context = NULL; + } + + if(self->thread_loop) { + pw_thread_loop_destroy(self->thread_loop); + self->thread_loop = NULL; + } + return false; +} + +static int pw_init_counter = 0; +bool gsr_pipewire_video_init(gsr_pipewire_video *self, int pipewire_fd, uint32_t pipewire_node, int fps, bool capture_cursor, gsr_egl *egl) { + if(pw_init_counter == 0) + pw_init(NULL, NULL); + ++pw_init_counter; + + memset(self, 0, sizeof(*self)); + self->egl = egl; + self->fd = pipewire_fd; + self->node = pipewire_node; + if(pthread_mutex_init(&self->mutex, NULL) != 0) { + fprintf(stderr, "gsr error: gsr_pipewire_video_init: failed to initialize mutex\n"); + gsr_pipewire_video_deinit(self); + return false; + } + self->mutex_initialized = true; + self->video_info.fps_num = fps; + self->video_info.fps_den = 1; + self->cursor.visible = capture_cursor; + + if(!gsr_pipewire_video_setup_stream(self)) { + gsr_pipewire_video_deinit(self); + return false; + } + + return true; +} + +void gsr_pipewire_video_deinit(gsr_pipewire_video *self) { + if(self->thread_loop) { + //pw_thread_loop_wait(self->thread_loop); + pw_thread_loop_stop(self->thread_loop); + } + + if(self->stream) { + pw_stream_disconnect(self->stream); + pw_stream_destroy(self->stream); + self->stream = NULL; + } + + if(self->core) { + pw_core_disconnect(self->core); + self->core = NULL; + } + + if(self->context) { + pw_context_destroy(self->context); + self->context = NULL; + } + + if(self->thread_loop) { + pw_thread_loop_destroy(self->thread_loop); + self->thread_loop = NULL; + } + + if(self->fd > 0) { + close(self->fd); + self->fd = -1; + } + + for(size_t i = 0; i < self->dmabuf_num_planes; ++i) { + if(self->dmabuf_data[i].fd > 0) { + close(self->dmabuf_data[i].fd); + self->dmabuf_data[i].fd = -1; + } + } + self->dmabuf_num_planes = 0; + + self->negotiated = false; + + if(self->mutex_initialized) { + pthread_mutex_destroy(&self->mutex); + self->mutex_initialized = false; + } + + if(self->cursor.data) { + free(self->cursor.data); + self->cursor.data = NULL; + } + + --pw_init_counter; + if(pw_init_counter == 0) { +#if PW_CHECK_VERSION(0, 3, 49) + pw_deinit(); +#endif + } +} + +static EGLImage gsr_pipewire_video_create_egl_image(gsr_pipewire_video *self, const int *fds, const uint32_t *offsets, const uint32_t *pitches, const uint64_t *modifiers, bool use_modifiers) { + intptr_t img_attr[44]; + setup_dma_buf_attrs(img_attr, spa_video_format_to_drm_format(self->format.info.raw.format), self->format.info.raw.size.width, self->format.info.raw.size.height, + fds, offsets, pitches, modifiers, self->dmabuf_num_planes, use_modifiers); + while(self->egl->eglGetError() != EGL_SUCCESS){} + EGLImage image = self->egl->eglCreateImage(self->egl->egl_display, 0, EGL_LINUX_DMA_BUF_EXT, NULL, img_attr); + if(!image || self->egl->eglGetError() != EGL_SUCCESS) { + if(image) + self->egl->eglDestroyImage(self->egl->egl_display, image); + return NULL; + } + return image; +} + +static EGLImage gsr_pipewire_video_create_egl_image_with_fallback(gsr_pipewire_video *self) { + int fds[GSR_PIPEWIRE_VIDEO_DMABUF_MAX_PLANES]; + uint32_t offsets[GSR_PIPEWIRE_VIDEO_DMABUF_MAX_PLANES]; + uint32_t pitches[GSR_PIPEWIRE_VIDEO_DMABUF_MAX_PLANES]; + uint64_t modifiers[GSR_PIPEWIRE_VIDEO_DMABUF_MAX_PLANES]; + for(size_t i = 0; i < self->dmabuf_num_planes; ++i) { + fds[i] = self->dmabuf_data[i].fd; + offsets[i] = self->dmabuf_data[i].offset; + pitches[i] = self->dmabuf_data[i].stride; + modifiers[i] = self->format.info.raw.modifier; + } + + EGLImage image = NULL; + if(self->no_modifiers_fallback) { + image = gsr_pipewire_video_create_egl_image(self, fds, offsets, pitches, modifiers, false); + } else { + image = gsr_pipewire_video_create_egl_image(self, fds, offsets, pitches, modifiers, true); + if(!image) { + fprintf(stderr, "gsr error: gsr_pipewire_video_create_egl_image_with_fallback: failed to create egl image with modifiers, trying without modifiers\n"); + self->no_modifiers_fallback = true; + image = gsr_pipewire_video_create_egl_image(self, fds, offsets, pitches, modifiers, false); + } + } + return image; +} + +static bool gsr_pipewire_video_bind_image_to_texture(gsr_pipewire_video *self, EGLImage image, unsigned int texture_id, bool external_texture) { + const int texture_target = external_texture ? GL_TEXTURE_EXTERNAL_OES : GL_TEXTURE_2D; + while(self->egl->glGetError() != 0){} + self->egl->glBindTexture(texture_target, texture_id); + self->egl->glEGLImageTargetTexture2DOES(texture_target, image); + const bool success = self->egl->glGetError() == 0; + self->egl->glBindTexture(texture_target, 0); + return success; +} + +static void gsr_pipewire_video_bind_image_to_texture_with_fallback(gsr_pipewire_video *self, gsr_texture_map texture_map, EGLImage image) { + if(self->external_texture_fallback) { + gsr_pipewire_video_bind_image_to_texture(self, image, texture_map.external_texture_id, true); + } else { + if(!gsr_pipewire_video_bind_image_to_texture(self, image, texture_map.texture_id, false)) { + fprintf(stderr, "gsr error: gsr_pipewire_video_map_texture: failed to bind image to texture, trying with external texture\n"); + self->external_texture_fallback = true; + gsr_pipewire_video_bind_image_to_texture(self, image, texture_map.external_texture_id, true); + } + } +} + +static void gsr_pipewire_video_update_cursor_texture(gsr_pipewire_video *self, gsr_texture_map texture_map) { + if(!self->cursor.data) + return; + + const float border_color[4] = { 0.0f, 0.0f, 0.0f, 0.0f }; + self->egl->glBindTexture(GL_TEXTURE_2D, texture_map.cursor_texture_id); + // TODO: glTextureSubImage2D if same size + self->egl->glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, self->cursor.width, self->cursor.height, 0, GL_RGBA, GL_UNSIGNED_BYTE, self->cursor.data); + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->egl->glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, border_color); + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); + self->egl->glBindTexture(GL_TEXTURE_2D, 0); + + free(self->cursor.data); + self->cursor.data = NULL; +} + +bool gsr_pipewire_video_map_texture(gsr_pipewire_video *self, gsr_texture_map texture_map, gsr_pipewire_video_region *region, gsr_pipewire_video_region *cursor_region, gsr_pipewire_video_dmabuf_data *dmabuf_data, int *num_dmabuf_data, uint32_t *fourcc, uint64_t *modifiers, bool *using_external_image) { + for(int i = 0; i < GSR_PIPEWIRE_VIDEO_DMABUF_MAX_PLANES; ++i) { + memset(&dmabuf_data[i], 0, sizeof(gsr_pipewire_video_dmabuf_data)); + } + *num_dmabuf_data = 0; + *using_external_image = self->external_texture_fallback; + *fourcc = 0; + *modifiers = 0; + pthread_mutex_lock(&self->mutex); + + if(!self->negotiated || self->dmabuf_data[0].fd <= 0) { + pthread_mutex_unlock(&self->mutex); + return false; + } + + EGLImage image = gsr_pipewire_video_create_egl_image_with_fallback(self); + if(image) { + gsr_pipewire_video_bind_image_to_texture_with_fallback(self, texture_map, image); + *using_external_image = self->external_texture_fallback; + self->egl->eglDestroyImage(self->egl->egl_display, image); + } + + gsr_pipewire_video_update_cursor_texture(self, texture_map); + + region->x = 0; + region->y = 0; + + region->width = self->format.info.raw.size.width; + region->height = self->format.info.raw.size.height; + + if(self->crop.valid) { + region->x = self->crop.x; + region->y = self->crop.y; + + region->width = self->crop.width; + region->height = self->crop.height; + } + + /* TODO: Test if cursor hotspot is correct */ + cursor_region->x = self->cursor.x - self->cursor.hotspot_x; + cursor_region->y = self->cursor.y - self->cursor.hotspot_y; + + cursor_region->width = self->cursor.width; + cursor_region->height = self->cursor.height; + + for(size_t i = 0; i < self->dmabuf_num_planes; ++i) { + dmabuf_data[i] = self->dmabuf_data[i]; + self->dmabuf_data[i].fd = -1; + } + *num_dmabuf_data = self->dmabuf_num_planes; + *fourcc = spa_video_format_to_drm_format(self->format.info.raw.format); + *modifiers = self->format.info.raw.modifier; + self->dmabuf_num_planes = 0; + + pthread_mutex_unlock(&self->mutex); + return true; +} + +bool gsr_pipewire_video_is_damaged(gsr_pipewire_video *self) { + bool damaged = false; + pthread_mutex_lock(&self->mutex); + damaged = self->damaged; + pthread_mutex_unlock(&self->mutex); + return damaged; +} + +void gsr_pipewire_video_clear_damage(gsr_pipewire_video *self) { + pthread_mutex_lock(&self->mutex); + self->damaged = false; + pthread_mutex_unlock(&self->mutex); +} diff --git a/src/replay_buffer/replay_buffer.c b/src/replay_buffer/replay_buffer.c new file mode 100644 index 0000000..92aa645 --- /dev/null +++ b/src/replay_buffer/replay_buffer.c @@ -0,0 +1,91 @@ +#include "../../include/replay_buffer/replay_buffer.h" +#include "../../include/replay_buffer/replay_buffer_ram.h" +#include "../../include/replay_buffer/replay_buffer_disk.h" + +#include <stdlib.h> +#include <string.h> +#include <assert.h> + +gsr_replay_buffer* gsr_replay_buffer_create(gsr_replay_storage replay_storage, const char *replay_directory, double replay_buffer_time, size_t replay_buffer_num_packets) { + gsr_replay_buffer *replay_buffer = NULL; + switch(replay_storage) { + case GSR_REPLAY_STORAGE_RAM: + replay_buffer = gsr_replay_buffer_ram_create(replay_buffer_num_packets); + break; + case GSR_REPLAY_STORAGE_DISK: + replay_buffer = gsr_replay_buffer_disk_create(replay_directory, replay_buffer_time); + break; + } + + replay_buffer->mutex_initialized = false; + replay_buffer->original_replay_buffer = NULL; + if(pthread_mutex_init(&replay_buffer->mutex, NULL) != 0) { + gsr_replay_buffer_destroy(replay_buffer); + return NULL; + } + + replay_buffer->mutex_initialized = true; + return replay_buffer; +} + +void gsr_replay_buffer_destroy(gsr_replay_buffer *self) { + self->destroy(self); + if(self->mutex_initialized && !self->original_replay_buffer) { + pthread_mutex_destroy(&self->mutex); + self->mutex_initialized = false; + } + self->original_replay_buffer = NULL; + free(self); +} + +void gsr_replay_buffer_lock(gsr_replay_buffer *self) { + if(self->original_replay_buffer) { + gsr_replay_buffer_lock(self->original_replay_buffer); + return; + } + + if(self->mutex_initialized) + pthread_mutex_lock(&self->mutex); +} + +void gsr_replay_buffer_unlock(gsr_replay_buffer *self) { + if(self->original_replay_buffer) { + gsr_replay_buffer_unlock(self->original_replay_buffer); + return; + } + + if(self->mutex_initialized) + pthread_mutex_unlock(&self->mutex); +} + +bool gsr_replay_buffer_append(gsr_replay_buffer *self, const AVPacket *av_packet, double timestamp) { + return self->append(self, av_packet, timestamp); +} + +void gsr_replay_buffer_clear(gsr_replay_buffer *self) { + self->clear(self); +} + +AVPacket* gsr_replay_buffer_iterator_get_packet(gsr_replay_buffer *self, gsr_replay_buffer_iterator iterator) { + return self->iterator_get_packet(self, iterator); +} + +uint8_t* gsr_replay_buffer_iterator_get_packet_data(gsr_replay_buffer *self, gsr_replay_buffer_iterator iterator) { + return self->iterator_get_packet_data(self, iterator); +} + +gsr_replay_buffer* gsr_replay_buffer_clone(gsr_replay_buffer *self) { + return self->clone(self); +} + +gsr_replay_buffer_iterator gsr_replay_buffer_find_packet_index_by_time_passed(gsr_replay_buffer *self, int seconds) { + return self->find_packet_index_by_time_passed(self, seconds); +} + +gsr_replay_buffer_iterator gsr_replay_buffer_find_keyframe(gsr_replay_buffer *self, gsr_replay_buffer_iterator start_iterator, int stream_index, bool invert_stream_index) { + return self->find_keyframe(self, start_iterator, stream_index, invert_stream_index); +} + +bool gsr_replay_buffer_iterator_next(gsr_replay_buffer *self, gsr_replay_buffer_iterator *iterator) { + return self->iterator_next(self, iterator); +} diff --git a/src/replay_buffer/replay_buffer_disk.c b/src/replay_buffer/replay_buffer_disk.c new file mode 100644 index 0000000..3fff9f3 --- /dev/null +++ b/src/replay_buffer/replay_buffer_disk.c @@ -0,0 +1,437 @@ +#include "../../include/replay_buffer/replay_buffer_disk.h" +#include "../../include/utils.h" + +#include <stdlib.h> +#include <string.h> +#include <stdio.h> +#include <fcntl.h> +#include <unistd.h> +#include <time.h> +#include <errno.h> +#include <assert.h> + +#define REPLAY_BUFFER_FILE_SIZE_BYTES 1024 * 1024 * 256 /* 256MB */ +#define FILE_PREFIX "Replay" + +static void gsr_replay_buffer_disk_set_impl_funcs(gsr_replay_buffer_disk *self); + +static void gsr_av_packet_disk_init(gsr_av_packet_disk *self, const AVPacket *av_packet, size_t data_index, double timestamp) { + self->packet = *av_packet; + self->packet.data = NULL; + self->data_index = data_index; + self->timestamp = timestamp; +} + +static gsr_replay_buffer_file* gsr_replay_buffer_file_create(char *replay_directory, size_t replay_storage_counter, double timestamp, int *replay_storage_fd) { + gsr_replay_buffer_file *self = calloc(1, sizeof(gsr_replay_buffer_file)); + if(!self) { + fprintf(stderr, "gsr error: gsr_av_packet_file_init: failed to create buffer file\n"); + return NULL; + } + + if(create_directory_recursive(replay_directory) != 0) { + fprintf(stderr, "gsr error: gsr_av_packet_file_init: failed to create replay directory: %s\n", replay_directory); + free(self); + return NULL; + } + + char filename[PATH_MAX]; + snprintf(filename, sizeof(filename), "%s/%s_%d.gsr", replay_directory, FILE_PREFIX, (int)replay_storage_counter); + *replay_storage_fd = creat(filename, 0700); + if(*replay_storage_fd <= 0) { + fprintf(stderr, "gsr error: gsr_av_packet_file_init: failed to create replay file: %s\n", filename); + free(self); + return NULL; + } + + self->id = replay_storage_counter; + self->start_timestamp = timestamp; + self->end_timestamp = timestamp; + self->ref_counter = 1; + self->fd = -1; + + self->packets = NULL; + self->capacity_num_packets = 0; + self->num_packets = 0; + return self; +} + +static gsr_replay_buffer_file* gsr_replay_buffer_file_ref(gsr_replay_buffer_file *self) { + if(self->ref_counter >= 1) + ++self->ref_counter; + return self; +} + +static void gsr_replay_buffer_file_free(gsr_replay_buffer_file *self, const char *replay_directory) { + self->ref_counter = 0; + + if(self->fd > 0) { + close(self->fd); + self->fd = -1; + } + + char filename[PATH_MAX]; + snprintf(filename, sizeof(filename), "%s/%s_%d.gsr", replay_directory, FILE_PREFIX, (int)self->id); + remove(filename); + + if(self->packets) { + free(self->packets); + self->packets = NULL; + } + self->num_packets = 0; + self->capacity_num_packets = 0; + + free(self); +} + +static void gsr_replay_buffer_file_unref(gsr_replay_buffer_file *self, const char *replay_directory) { + if(self->ref_counter > 0) + --self->ref_counter; + + if(self->ref_counter <= 0) + gsr_replay_buffer_file_free(self, replay_directory); +} + +static void gsr_replay_buffer_disk_clear(gsr_replay_buffer *replay_buffer) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + gsr_replay_buffer_lock(&self->replay_buffer); + + for(size_t i = 0; i < self->num_files; ++i) { + gsr_replay_buffer_file_unref(self->files[i], self->replay_directory); + } + self->num_files = 0; + + if(self->storage_fd > 0) { + close(self->storage_fd); + self->storage_fd = 0; + } + + self->storage_num_bytes_written = 0; + gsr_replay_buffer_unlock(&self->replay_buffer); +} + +static void gsr_replay_buffer_disk_destroy(gsr_replay_buffer *replay_buffer) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + gsr_replay_buffer_disk_clear(replay_buffer); + + if(self->owns_directory) { + remove(self->replay_directory); + self->owns_directory = false; + } +} + +static bool file_write_all(int fd, const uint8_t *data, size_t size, size_t *bytes_written_total) { + *bytes_written_total = 0; + while(*bytes_written_total < size) { + const ssize_t bytes_written = write(fd, data + *bytes_written_total, size - *bytes_written_total); + if(bytes_written == -1) { + if(errno == EAGAIN) + continue; + else + return false; + } + *bytes_written_total += bytes_written; + } + return true; +} + +static bool gsr_replay_buffer_disk_create_next_file(gsr_replay_buffer_disk *self, double timestamp) { + if(self->num_files + 1 >= GSR_REPLAY_BUFFER_CAPACITY_NUM_FILES) { + fprintf(stderr, "gsr error: gsr_replay_buffer_disk_create_next_file: too many replay buffer files created! (> %d), either reduce the replay buffer time or report this as a bug\n", (int)GSR_REPLAY_BUFFER_CAPACITY_NUM_FILES); + return false; + } + + gsr_replay_buffer_file *replay_buffer_file = gsr_replay_buffer_file_create(self->replay_directory, self->storage_counter, timestamp, &self->storage_fd); + if(!replay_buffer_file) + return false; + + self->files[self->num_files] = replay_buffer_file; + ++self->num_files; + ++self->storage_counter; + return true; +} + +static bool gsr_replay_buffer_disk_append_to_current_file(gsr_replay_buffer_disk *self, const AVPacket *av_packet, double timestamp) { + gsr_replay_buffer_file *replay_buffer_file = self->files[self->num_files - 1]; + replay_buffer_file->end_timestamp = timestamp; + + if(replay_buffer_file->num_packets + 1 >= replay_buffer_file->capacity_num_packets) { + size_t new_capacity_num_packets = replay_buffer_file->capacity_num_packets * 2; + if(new_capacity_num_packets == 0) + new_capacity_num_packets = 256; + + void *new_packets = realloc(replay_buffer_file->packets, new_capacity_num_packets * sizeof(gsr_av_packet_disk)); + if(!new_packets) { + fprintf(stderr, "gsr error: gsr_replay_buffer_disk_append_to_current_file: failed to reallocate replay buffer file packets\n"); + return false; + } + + replay_buffer_file->capacity_num_packets = new_capacity_num_packets; + replay_buffer_file->packets = new_packets; + } + + gsr_av_packet_disk *packet = &replay_buffer_file->packets[replay_buffer_file->num_packets]; + gsr_av_packet_disk_init(packet, av_packet, self->storage_num_bytes_written, timestamp); + ++replay_buffer_file->num_packets; + + size_t bytes_written = 0; + const bool file_written = file_write_all(self->storage_fd, av_packet->data, av_packet->size, &bytes_written); + self->storage_num_bytes_written += bytes_written; + if(self->storage_num_bytes_written >= REPLAY_BUFFER_FILE_SIZE_BYTES) { + self->storage_num_bytes_written = 0; + close(self->storage_fd); + self->storage_fd = 0; + } + + return file_written; +} + +static void gsr_replay_buffer_disk_remove_first_file(gsr_replay_buffer_disk *self) { + gsr_replay_buffer_file_unref(self->files[0], self->replay_directory); + for(size_t i = 1; i < self->num_files; ++i) { + self->files[i - 1] = self->files[i]; + } + --self->num_files; +} + +static bool gsr_replay_buffer_disk_append(gsr_replay_buffer *replay_buffer, const AVPacket *av_packet, double timestamp) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + bool success = false; + gsr_replay_buffer_lock(&self->replay_buffer); + + if(self->storage_fd <= 0) { + if(!gsr_replay_buffer_disk_create_next_file(self, timestamp)) + goto done; + } + + const bool data_written = gsr_replay_buffer_disk_append_to_current_file(self, av_packet, timestamp); + + if(self->num_files > 1) { + const double buffer_time_accumulated = timestamp - self->files[1]->start_timestamp; + if(buffer_time_accumulated >= self->replay_buffer_time) + gsr_replay_buffer_disk_remove_first_file(self); + } + + success = data_written; + + done: + gsr_replay_buffer_unlock(&self->replay_buffer); + return success; +} + +static AVPacket* gsr_replay_buffer_disk_iterator_get_packet(gsr_replay_buffer *replay_buffer, gsr_replay_buffer_iterator iterator) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + assert(iterator.file_index < self->num_files); + assert(iterator.packet_index < self->files[iterator.file_index]->num_packets); + return &self->files[iterator.file_index]->packets[iterator.packet_index].packet; +} + +static uint8_t* gsr_replay_buffer_disk_iterator_get_packet_data(gsr_replay_buffer *replay_buffer, gsr_replay_buffer_iterator iterator) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + assert(iterator.file_index < self->num_files); + gsr_replay_buffer_file *file = self->files[iterator.file_index]; + assert(iterator.packet_index < file->num_packets); + + if(file->fd <= 0) { + char filename[PATH_MAX]; + snprintf(filename, sizeof(filename), "%s/%s_%d.gsr", self->replay_directory, FILE_PREFIX, (int)file->id); + file->fd = open(filename, O_RDONLY); + if(file->fd <= 0) { + fprintf(stderr, "gsr error: gsr_replay_buffer_disk_iterator_get_packet_data: failed to open file\n"); + return NULL; + } + } + + const gsr_av_packet_disk *packet = &self->files[iterator.file_index]->packets[iterator.packet_index]; + if(lseek(file->fd, packet->data_index, SEEK_SET) == -1) { + fprintf(stderr, "gsr error: gsr_replay_buffer_disk_iterator_get_packet_data: failed to seek\n"); + return NULL; + } + + uint8_t *packet_data = malloc(packet->packet.size); + if(read(file->fd, packet_data, packet->packet.size) != packet->packet.size) { + fprintf(stderr, "gsr error: gsr_replay_buffer_disk_iterator_get_packet_data: failed to read data from file\n"); + free(packet_data); + return NULL; + } + + return packet_data; +} + +static gsr_replay_buffer* gsr_replay_buffer_disk_clone(gsr_replay_buffer *replay_buffer) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + gsr_replay_buffer_disk *destination = calloc(1, sizeof(gsr_replay_buffer_disk)); + if(!destination) + return NULL; + + gsr_replay_buffer_disk_set_impl_funcs(destination); + gsr_replay_buffer_lock(&self->replay_buffer); + + destination->replay_buffer.original_replay_buffer = replay_buffer; + destination->replay_buffer.mutex = self->replay_buffer.mutex; + destination->replay_buffer.mutex_initialized = self->replay_buffer.mutex_initialized; + destination->replay_buffer_time = self->replay_buffer_time; + destination->storage_counter = self->storage_counter; + destination->storage_num_bytes_written = self->storage_num_bytes_written; + destination->storage_fd = 0; // We only want to read from the clone. If there is a need to write to it in the future then TODO change this + + for(size_t i = 0; i < self->num_files; ++i) { + destination->files[i] = gsr_replay_buffer_file_ref(self->files[i]); + } + destination->num_files = self->num_files; + + snprintf(destination->replay_directory, sizeof(destination->replay_directory), "%s", self->replay_directory); + destination->owns_directory = false; + + gsr_replay_buffer_unlock(&self->replay_buffer); + return (gsr_replay_buffer*)destination; +} + +/* Binary search */ +static size_t gsr_replay_buffer_file_find_packet_index_by_time_passed(const gsr_replay_buffer_file *self, int seconds) { + const double now = clock_get_monotonic_seconds(); + if(self->num_packets == 0) { + return 0; + } + + size_t lower_bound = 0; + size_t upper_bound = self->num_packets; + size_t index = 0; + + for(;;) { + index = lower_bound + (upper_bound - lower_bound) / 2; + const gsr_av_packet_disk *packet = &self->packets[index]; + const double time_passed_since_packet = now - packet->timestamp; + if(time_passed_since_packet >= seconds) { + if(lower_bound == index) + break; + lower_bound = index; + } else { + if(upper_bound == index) + break; + upper_bound = index; + } + } + + return index; +} + +/* Binary search */ +static gsr_replay_buffer_iterator gsr_replay_buffer_disk_find_file_index_by_time_passed(gsr_replay_buffer *replay_buffer, int seconds) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + gsr_replay_buffer_lock(&self->replay_buffer); + + const double now = clock_get_monotonic_seconds(); + if(self->num_files == 0) { + gsr_replay_buffer_unlock(&self->replay_buffer); + return (gsr_replay_buffer_iterator){0, 0}; + } + + size_t lower_bound = 0; + size_t upper_bound = self->num_files; + size_t file_index = 0; + + for(;;) { + file_index = lower_bound + (upper_bound - lower_bound) / 2; + const gsr_replay_buffer_file *file = self->files[file_index]; + const double time_passed_since_file_start = now - file->start_timestamp; + const double time_passed_since_file_end = now - file->end_timestamp; + if(time_passed_since_file_start >= seconds && time_passed_since_file_end <= seconds) { + break; + } else if(time_passed_since_file_start >= seconds) { + if(lower_bound == file_index) + break; + lower_bound = file_index; + } else { + if(upper_bound == file_index) + break; + upper_bound = file_index; + } + } + + const gsr_replay_buffer_file *file = self->files[file_index]; + const size_t packet_index = gsr_replay_buffer_file_find_packet_index_by_time_passed(file, seconds); + + gsr_replay_buffer_unlock(&self->replay_buffer); + return (gsr_replay_buffer_iterator){packet_index, file_index}; +} + +static gsr_replay_buffer_iterator gsr_replay_buffer_disk_find_keyframe(gsr_replay_buffer *replay_buffer, gsr_replay_buffer_iterator start_iterator, int stream_index, bool invert_stream_index) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + gsr_replay_buffer_iterator keyframe_iterator = {(size_t)-1, 0}; + gsr_replay_buffer_lock(&self->replay_buffer); + size_t packet_index = start_iterator.packet_index; + for(size_t file_index = start_iterator.file_index; file_index < self->num_files; ++file_index) { + const gsr_replay_buffer_file *file = self->files[file_index]; + for(; packet_index < file->num_packets; ++packet_index) { + const gsr_av_packet_disk *packet = &file->packets[packet_index]; + if((packet->packet.flags & AV_PKT_FLAG_KEY) && (invert_stream_index ? packet->packet.stream_index != stream_index : packet->packet.stream_index == stream_index)) { + keyframe_iterator.packet_index = packet_index; + keyframe_iterator.file_index = file_index; + goto done; + } + } + packet_index = 0; + } + done: + gsr_replay_buffer_unlock(&self->replay_buffer); + return keyframe_iterator; +} + +static bool gsr_replay_buffer_disk_iterator_next(gsr_replay_buffer *replay_buffer, gsr_replay_buffer_iterator *iterator) { + gsr_replay_buffer_disk *self = (gsr_replay_buffer_disk*)replay_buffer; + if(iterator->file_index >= self->num_files) + return false; + + if(iterator->packet_index + 1 >= self->files[iterator->file_index]->num_packets) { + if(iterator->file_index + 1 >= self->num_files) + return false; + + if(self->files[iterator->file_index + 1]->num_packets == 0) + return false; + + ++iterator->file_index; + iterator->packet_index = 0; + return true; + } else { + ++iterator->packet_index; + return true; + } +} + +static void get_current_time(char *time_str, size_t time_str_size) { + time_t now = time(NULL); + struct tm *t = localtime(&now); + strftime(time_str, time_str_size - 1, "%Y-%m-%d_%H-%M-%S", t); +} + +static void gsr_replay_buffer_disk_set_impl_funcs(gsr_replay_buffer_disk *self) { + self->replay_buffer.destroy = gsr_replay_buffer_disk_destroy; + self->replay_buffer.append = gsr_replay_buffer_disk_append; + self->replay_buffer.clear = gsr_replay_buffer_disk_clear; + self->replay_buffer.iterator_get_packet = gsr_replay_buffer_disk_iterator_get_packet; + self->replay_buffer.iterator_get_packet_data = gsr_replay_buffer_disk_iterator_get_packet_data; + self->replay_buffer.clone = gsr_replay_buffer_disk_clone; + self->replay_buffer.find_packet_index_by_time_passed = gsr_replay_buffer_disk_find_file_index_by_time_passed; + self->replay_buffer.find_keyframe = gsr_replay_buffer_disk_find_keyframe; + self->replay_buffer.iterator_next = gsr_replay_buffer_disk_iterator_next; +} + +gsr_replay_buffer* gsr_replay_buffer_disk_create(const char *replay_directory, double replay_buffer_time) { + assert(replay_buffer_time > 0); + gsr_replay_buffer_disk *replay_buffer = calloc(1, sizeof(gsr_replay_buffer_disk)); + if(!replay_buffer) + return NULL; + + char time_str[128]; + get_current_time(time_str, sizeof(time_str)); + + replay_buffer->num_files = 0; + replay_buffer->storage_counter = 0; + replay_buffer->replay_buffer_time = replay_buffer_time; + snprintf(replay_buffer->replay_directory, sizeof(replay_buffer->replay_directory), "%s/gsr-replay-%s.gsr", replay_directory, time_str); + replay_buffer->owns_directory = true; + + gsr_replay_buffer_disk_set_impl_funcs(replay_buffer); + return (gsr_replay_buffer*)replay_buffer; +} diff --git a/src/replay_buffer/replay_buffer_ram.c b/src/replay_buffer/replay_buffer_ram.c new file mode 100644 index 0000000..890588f --- /dev/null +++ b/src/replay_buffer/replay_buffer_ram.c @@ -0,0 +1,256 @@ +#include "../../include/replay_buffer/replay_buffer_ram.h" +#include "../../include/utils.h" + +#include <stdlib.h> +#include <string.h> +#include <assert.h> + +#include <libavutil/mem.h> + +static void gsr_replay_buffer_ram_set_impl_funcs(gsr_replay_buffer_ram *self); + +static gsr_av_packet_ram* gsr_av_packet_ram_create(const AVPacket *av_packet, double timestamp) { + gsr_av_packet_ram *self = malloc(sizeof(gsr_av_packet_ram)); + if(!self) + return NULL; + + self->ref_counter = 1; + self->packet = *av_packet; + self->timestamp = timestamp; + // Why are we doing this you ask? there is a ffmpeg bug that causes cpu usage to increase over time when you have + // packets that are not being free'd until later. So we copy the packet data, free the packet and then reconstruct + // the packet later on when we need it, to keep packets alive only for a short period. + self->packet.data = av_memdup(av_packet->data, av_packet->size); + if(!self->packet.data) { + free(self); + return NULL; + } + + return self; +} + +static gsr_av_packet_ram* gsr_av_packet_ram_ref(gsr_av_packet_ram *self) { + if(self->ref_counter >= 1) + ++self->ref_counter; + return self; +} + +static void gsr_av_packet_ram_free(gsr_av_packet_ram *self) { + self->ref_counter = 0; + if(self->packet.data) { + av_free(self->packet.data); + self->packet.data = NULL; + } + free(self); +} + +static void gsr_av_packet_ram_unref(gsr_av_packet_ram *self) { + if(self->ref_counter >= 1) + --self->ref_counter; + + if(self->ref_counter <= 0) + gsr_av_packet_ram_free(self); +} + +static void gsr_replay_buffer_ram_destroy(gsr_replay_buffer *replay_buffer) { + gsr_replay_buffer_ram *self = (gsr_replay_buffer_ram*)replay_buffer; + gsr_replay_buffer_lock(&self->replay_buffer); + for(size_t i = 0; i < self->num_packets; ++i) { + if(self->packets[i]) { + gsr_av_packet_ram_unref(self->packets[i]); + self->packets[i] = NULL; + } + } + self->num_packets = 0; + gsr_replay_buffer_unlock(&self->replay_buffer); + + if(self->packets) { + free(self->packets); + self->packets = NULL; + } + + self->capacity_num_packets = 0; + self->index = 0; +} + +static bool gsr_replay_buffer_ram_append(gsr_replay_buffer *replay_buffer, const AVPacket *av_packet, double timestamp) { + gsr_replay_buffer_ram *self = (gsr_replay_buffer_ram*)replay_buffer; + gsr_replay_buffer_lock(&self->replay_buffer); + gsr_av_packet_ram *packet = gsr_av_packet_ram_create(av_packet, timestamp); + if(!packet) { + gsr_replay_buffer_unlock(&self->replay_buffer); + return false; + } + + if(self->packets[self->index]) { + gsr_av_packet_ram_unref(self->packets[self->index]); + self->packets[self->index] = NULL; + } + self->packets[self->index] = packet; + + self->index = (self->index + 1) % self->capacity_num_packets; + ++self->num_packets; + if(self->num_packets > self->capacity_num_packets) + self->num_packets = self->capacity_num_packets; + + gsr_replay_buffer_unlock(&self->replay_buffer); + return true; +} + +static void gsr_replay_buffer_ram_clear(gsr_replay_buffer *replay_buffer) { + gsr_replay_buffer_ram *self = (gsr_replay_buffer_ram*)replay_buffer; + gsr_replay_buffer_lock(&self->replay_buffer); + for(size_t i = 0; i < self->num_packets; ++i) { + if(self->packets[i]) { + gsr_av_packet_ram_unref(self->packets[i]); + self->packets[i] = NULL; + } + } + self->num_packets = 0; + self->index = 0; + gsr_replay_buffer_unlock(&self->replay_buffer); +} + +static gsr_av_packet_ram* gsr_replay_buffer_ram_get_packet_at_index(gsr_replay_buffer *replay_buffer, size_t index) { + gsr_replay_buffer_ram *self = (gsr_replay_buffer_ram*)replay_buffer; + assert(index < self->num_packets); + size_t start_index = 0; + if(self->num_packets < self->capacity_num_packets) + start_index = self->num_packets - self->index; + else + start_index = self->index; + + const size_t offset = (start_index + index) % self->capacity_num_packets; + return self->packets[offset]; +} + +static AVPacket* gsr_replay_buffer_ram_iterator_get_packet(gsr_replay_buffer *replay_buffer, gsr_replay_buffer_iterator iterator) { + return &gsr_replay_buffer_ram_get_packet_at_index(replay_buffer, iterator.packet_index)->packet; +} + +static uint8_t* gsr_replay_buffer_ram_iterator_get_packet_data(gsr_replay_buffer *replay_buffer, gsr_replay_buffer_iterator iterator) { + (void)replay_buffer; + (void)iterator; + return NULL; +} + +static gsr_replay_buffer* gsr_replay_buffer_ram_clone(gsr_replay_buffer *replay_buffer) { + gsr_replay_buffer_ram *self = (gsr_replay_buffer_ram*)replay_buffer; + gsr_replay_buffer_ram *destination = calloc(1, sizeof(gsr_replay_buffer_ram)); + if(!destination) + return NULL; + + gsr_replay_buffer_ram_set_impl_funcs(destination); + gsr_replay_buffer_lock(&self->replay_buffer); + + destination->replay_buffer.original_replay_buffer = replay_buffer; + destination->replay_buffer.mutex = self->replay_buffer.mutex; + destination->replay_buffer.mutex_initialized = self->replay_buffer.mutex_initialized; + destination->capacity_num_packets = self->capacity_num_packets; + destination->index = self->index; + destination->packets = calloc(destination->capacity_num_packets, sizeof(gsr_av_packet_ram*)); + if(!destination->packets) { + free(destination); + gsr_replay_buffer_unlock(&self->replay_buffer); + return NULL; + } + + destination->num_packets = self->num_packets; + for(size_t i = 0; i < destination->num_packets; ++i) { + destination->packets[i] = gsr_av_packet_ram_ref(self->packets[i]); + } + + gsr_replay_buffer_unlock(&self->replay_buffer); + return (gsr_replay_buffer*)destination; +} + +/* Binary search */ +static gsr_replay_buffer_iterator gsr_replay_buffer_ram_find_packet_index_by_time_passed(gsr_replay_buffer *replay_buffer, int seconds) { + gsr_replay_buffer_ram *self = (gsr_replay_buffer_ram*)replay_buffer; + gsr_replay_buffer_lock(&self->replay_buffer); + + const double now = clock_get_monotonic_seconds(); + if(self->num_packets == 0) { + gsr_replay_buffer_unlock(&self->replay_buffer); + return (gsr_replay_buffer_iterator){0, 0}; + } + + size_t lower_bound = 0; + size_t upper_bound = self->num_packets; + size_t index = 0; + + for(;;) { + index = lower_bound + (upper_bound - lower_bound) / 2; + const gsr_av_packet_ram *packet = gsr_replay_buffer_ram_get_packet_at_index(replay_buffer, index); + const double time_passed_since_packet = now - packet->timestamp; + if(time_passed_since_packet >= seconds) { + if(lower_bound == index) + break; + lower_bound = index; + } else { + if(upper_bound == index) + break; + upper_bound = index; + } + } + + gsr_replay_buffer_unlock(&self->replay_buffer); + return (gsr_replay_buffer_iterator){index, 0}; +} + +static gsr_replay_buffer_iterator gsr_replay_buffer_ram_find_keyframe(gsr_replay_buffer *replay_buffer, gsr_replay_buffer_iterator start_iterator, int stream_index, bool invert_stream_index) { + gsr_replay_buffer_ram *self = (gsr_replay_buffer_ram*)replay_buffer; + size_t keyframe_index = (size_t)-1; + gsr_replay_buffer_lock(&self->replay_buffer); + for(size_t i = start_iterator.packet_index; i < self->num_packets; ++i) { + const gsr_av_packet_ram *packet = gsr_replay_buffer_ram_get_packet_at_index(replay_buffer, i); + if((packet->packet.flags & AV_PKT_FLAG_KEY) && (invert_stream_index ? packet->packet.stream_index != stream_index : packet->packet.stream_index == stream_index)) { + keyframe_index = i; + break; + } + } + gsr_replay_buffer_unlock(&self->replay_buffer); + return (gsr_replay_buffer_iterator){keyframe_index, 0}; +} + +static bool gsr_replay_buffer_ram_iterator_next(gsr_replay_buffer *replay_buffer, gsr_replay_buffer_iterator *iterator) { + gsr_replay_buffer_ram *self = (gsr_replay_buffer_ram*)replay_buffer; + if(iterator->packet_index + 1 < self->num_packets) { + ++iterator->packet_index; + return true; + } else { + return false; + } +} + +static void gsr_replay_buffer_ram_set_impl_funcs(gsr_replay_buffer_ram *self) { + self->replay_buffer.destroy = gsr_replay_buffer_ram_destroy; + self->replay_buffer.append = gsr_replay_buffer_ram_append; + self->replay_buffer.clear = gsr_replay_buffer_ram_clear; + self->replay_buffer.iterator_get_packet = gsr_replay_buffer_ram_iterator_get_packet; + self->replay_buffer.iterator_get_packet_data = gsr_replay_buffer_ram_iterator_get_packet_data; + self->replay_buffer.clone = gsr_replay_buffer_ram_clone; + self->replay_buffer.find_packet_index_by_time_passed = gsr_replay_buffer_ram_find_packet_index_by_time_passed; + self->replay_buffer.find_keyframe = gsr_replay_buffer_ram_find_keyframe; + self->replay_buffer.iterator_next = gsr_replay_buffer_ram_iterator_next; +} + +gsr_replay_buffer* gsr_replay_buffer_ram_create(size_t replay_buffer_num_packets) { + assert(replay_buffer_num_packets > 0); + gsr_replay_buffer_ram *replay_buffer = calloc(1, sizeof(gsr_replay_buffer_ram)); + if(!replay_buffer) + return NULL; + + replay_buffer->capacity_num_packets = replay_buffer_num_packets; + replay_buffer->num_packets = 0; + replay_buffer->index = 0; + replay_buffer->packets = calloc(replay_buffer->capacity_num_packets, sizeof(gsr_av_packet_ram*)); + if(!replay_buffer->packets) { + gsr_replay_buffer_ram_destroy(&replay_buffer->replay_buffer); + free(replay_buffer); + return NULL; + } + + gsr_replay_buffer_ram_set_impl_funcs(replay_buffer); + return (gsr_replay_buffer*)replay_buffer; +} diff --git a/src/shader.c b/src/shader.c index f8d7eb2..ba4db80 100644 --- a/src/shader.c +++ b/src/shader.c @@ -1,15 +1,18 @@ #include "../include/shader.h" +#include "../include/egl.h" #include <stdio.h> #include <assert.h> +static bool print_compile_errors = false; + static int min_int(int a, int b) { return a < b ? a : b; } -static unsigned int loader_shader(gsr_egl *egl, unsigned int type, const char *source) { +static unsigned int load_shader(gsr_egl *egl, unsigned int type, const char *source) { unsigned int shader_id = egl->glCreateShader(type); if(shader_id == 0) { - fprintf(stderr, "gsr error: loader_shader: failed to create shader, error: %d\n", egl->glGetError()); + fprintf(stderr, "gsr error: load_shader: failed to create shader, error: %d\n", egl->glGetError()); return 0; } @@ -22,10 +25,10 @@ static unsigned int loader_shader(gsr_egl *egl, unsigned int type, const char *s int info_length = 0; egl->glGetShaderiv(shader_id, GL_INFO_LOG_LENGTH, &info_length); - if(info_length > 1) { + if(info_length > 1 && print_compile_errors) { char info_log[4096]; egl->glGetShaderInfoLog(shader_id, min_int(4096, info_length), NULL, info_log); - fprintf(stderr, "gsr error: loader shader: failed to compile shader, error:\n%s\n", info_log); + fprintf(stderr, "gsr error: load_shader: failed to compile shader, error:\n%s\nshader source:\n%s\n", info_log, source); } egl->glDeleteShader(shader_id); @@ -35,28 +38,36 @@ static unsigned int loader_shader(gsr_egl *egl, unsigned int type, const char *s return shader_id; } -static unsigned int load_program(gsr_egl *egl, const char *vertex_shader, const char *fragment_shader) { +static unsigned int load_program(gsr_egl *egl, const char *vertex_shader, const char *fragment_shader, const char *compute_shader) { unsigned int vertex_shader_id = 0; unsigned int fragment_shader_id = 0; + unsigned int compute_shader_id = 0; unsigned int program_id = 0; int linked = 0; + bool success = false; if(vertex_shader) { - vertex_shader_id = loader_shader(egl, GL_VERTEX_SHADER, vertex_shader); + vertex_shader_id = load_shader(egl, GL_VERTEX_SHADER, vertex_shader); if(vertex_shader_id == 0) - goto err; + goto done; } if(fragment_shader) { - fragment_shader_id = loader_shader(egl, GL_FRAGMENT_SHADER, fragment_shader); + fragment_shader_id = load_shader(egl, GL_FRAGMENT_SHADER, fragment_shader); if(fragment_shader_id == 0) - goto err; + goto done; + } + + if(compute_shader) { + compute_shader_id = load_shader(egl, GL_COMPUTE_SHADER, compute_shader); + if(compute_shader_id == 0) + goto done; } program_id = egl->glCreateProgram(); if(program_id == 0) { fprintf(stderr, "gsr error: load_program: failed to create shader program, error: %d\n", egl->glGetError()); - goto err; + goto done; } if(vertex_shader_id) @@ -65,6 +76,9 @@ static unsigned int load_program(gsr_egl *egl, const char *vertex_shader, const if(fragment_shader_id) egl->glAttachShader(program_id, fragment_shader_id); + if(compute_shader_id) + egl->glAttachShader(program_id, compute_shader_id); + egl->glLinkProgram(program_id); egl->glGetProgramiv(program_id, GL_LINK_STATUS, &linked); @@ -78,37 +92,36 @@ static unsigned int load_program(gsr_egl *egl, const char *vertex_shader, const fprintf(stderr, "gsr error: load program: linking shader program failed, error:\n%s\n", info_log); } - goto err; + goto done; } - if(fragment_shader_id) - egl->glDeleteShader(fragment_shader_id); - if(vertex_shader_id) - egl->glDeleteShader(vertex_shader_id); - - return program_id; + success = true; + done: - err: - if(program_id) - egl->glDeleteProgram(program_id); + if(!success) { + if(program_id) + egl->glDeleteProgram(program_id); + } + if(compute_shader_id) + egl->glDeleteShader(compute_shader_id); if(fragment_shader_id) egl->glDeleteShader(fragment_shader_id); if(vertex_shader_id) egl->glDeleteShader(vertex_shader_id); - return 0; + return program_id; } -int gsr_shader_init(gsr_shader *self, gsr_egl *egl, const char *vertex_shader, const char *fragment_shader) { +int gsr_shader_init(gsr_shader *self, gsr_egl *egl, const char *vertex_shader, const char *fragment_shader, const char *compute_shader) { assert(egl); self->egl = egl; self->program_id = 0; - if(!vertex_shader && !fragment_shader) { - fprintf(stderr, "gsr error: gsr_shader_init: vertex shader and fragment shader can't be NULL at the same time\n"); + if(!vertex_shader && !fragment_shader && !compute_shader) { + fprintf(stderr, "gsr error: gsr_shader_init: vertex, fragment shader and compute shaders can't be NULL at the same time\n"); return -1; } - self->program_id = load_program(self->egl, vertex_shader, fragment_shader); + self->program_id = load_program(self->egl, vertex_shader, fragment_shader, compute_shader); if(self->program_id == 0) return -1; @@ -140,3 +153,7 @@ void gsr_shader_use(gsr_shader *self) { void gsr_shader_use_none(gsr_shader *self) { self->egl->glUseProgram(0); } + +void gsr_shader_enable_debug_output(bool enable) { + print_compile_errors = enable; +} diff --git a/src/sound.cpp b/src/sound.cpp index c3aa4d4..5a0ce77 100644 --- a/src/sound.cpp +++ b/src/sound.cpp @@ -8,12 +8,16 @@ extern "C" { #include <string.h> #include <cmath> #include <time.h> +#include <mutex> #include <pulse/pulseaudio.h> #include <pulse/mainloop.h> #include <pulse/xmalloc.h> #include <pulse/error.h> +#define RECONNECT_TRY_TIMEOUT_SECONDS 0.5 +#define DEVICE_NAME_MAX_SIZE 128 + #define CHECK_DEAD_GOTO(p, rerror, label) \ do { \ if (!(p)->context || !PA_CONTEXT_IS_GOOD(pa_context_get_state((p)->context)) || \ @@ -29,6 +33,12 @@ extern "C" { } \ } while(false); +enum class DeviceType { + STANDARD, + DEFAULT_OUTPUT, + DEFAULT_INPUT +}; + struct pa_handle { pa_context *context; pa_stream *stream; @@ -41,49 +51,153 @@ struct pa_handle { size_t output_index, output_length; int operation_success; + double latency_seconds; + + pa_buffer_attr attr; + pa_sample_spec ss; + + std::mutex reconnect_mutex; + DeviceType device_type; + char stream_name[256]; + bool reconnect; + double reconnect_last_tried_seconds; + + char device_name[DEVICE_NAME_MAX_SIZE]; + char default_output_device_name[DEVICE_NAME_MAX_SIZE]; + char default_input_device_name[DEVICE_NAME_MAX_SIZE]; }; -static void pa_sound_device_free(pa_handle *s) { - assert(s); +static void pa_sound_device_free(pa_handle *p) { + assert(p); + + if (p->stream) { + pa_stream_unref(p->stream); + p->stream = NULL; + } + + if (p->context) { + pa_context_disconnect(p->context); + pa_context_unref(p->context); + p->context = NULL; + } + + if (p->mainloop) { + pa_mainloop_free(p->mainloop); + p->mainloop = NULL; + } + + if (p->output_data) { + free(p->output_data); + p->output_data = NULL; + } + + pa_xfree(p); +} + +static void subscribe_update_default_devices(pa_context*, const pa_server_info *server_info, void *userdata) { + pa_handle *handle = (pa_handle*)userdata; + std::lock_guard<std::mutex> lock(handle->reconnect_mutex); + + if(server_info->default_sink_name) { + // TODO: Size check + snprintf(handle->default_output_device_name, sizeof(handle->default_output_device_name), "%s.monitor", server_info->default_sink_name); + if(handle->device_type == DeviceType::DEFAULT_OUTPUT && strcmp(handle->device_name, handle->default_output_device_name) != 0) { + handle->reconnect = true; + handle->reconnect_last_tried_seconds = clock_get_monotonic_seconds(); + // TODO: Size check + snprintf(handle->device_name, sizeof(handle->device_name), "%s", handle->default_output_device_name); + } + } + + if(server_info->default_source_name) { + // TODO: Size check + snprintf(handle->default_input_device_name, sizeof(handle->default_input_device_name), "%s", server_info->default_source_name); + if(handle->device_type == DeviceType::DEFAULT_INPUT && strcmp(handle->device_name, handle->default_input_device_name) != 0) { + handle->reconnect = true; + handle->reconnect_last_tried_seconds = clock_get_monotonic_seconds(); + // TODO: Size check + snprintf(handle->device_name, sizeof(handle->device_name), "%s", handle->default_input_device_name); + } + } +} + +static void subscribe_cb(pa_context *c, pa_subscription_event_type_t t, uint32_t idx, void *userdata) { + (void)idx; + pa_handle *handle = (pa_handle*)userdata; + if((t & PA_SUBSCRIPTION_EVENT_FACILITY_MASK) == PA_SUBSCRIPTION_EVENT_SERVER) { + pa_operation *pa = pa_context_get_server_info(c, subscribe_update_default_devices, handle); + if(pa) + pa_operation_unref(pa); + } +} - if (s->stream) - pa_stream_unref(s->stream); +static void store_default_devices(pa_context*, const pa_server_info *server_info, void *userdata) { + pa_handle *handle = (pa_handle*)userdata; + if(server_info->default_sink_name) + snprintf(handle->default_output_device_name, sizeof(handle->default_output_device_name), "%s.monitor", server_info->default_sink_name); + if(server_info->default_source_name) + snprintf(handle->default_input_device_name, sizeof(handle->default_input_device_name), "%s", server_info->default_source_name); +} - if (s->context) { - pa_context_disconnect(s->context); - pa_context_unref(s->context); +static bool startup_get_default_devices(pa_handle *p, const char *device_name) { + pa_operation *pa = pa_context_get_server_info(p->context, store_default_devices, p); + while(pa) { + pa_operation_state state = pa_operation_get_state(pa); + if(state == PA_OPERATION_DONE) { + pa_operation_unref(pa); + break; + } else if(state == PA_OPERATION_CANCELLED) { + pa_operation_unref(pa); + return false; + } + pa_mainloop_iterate(p->mainloop, 1, NULL); } - if (s->mainloop) - pa_mainloop_free(s->mainloop); + if(p->default_output_device_name[0] == '\0') { + fprintf(stderr, "Error: failed to find default audio output device\n"); + return false; + } - if (s->output_data) { - free(s->output_data); - s->output_data = NULL; + if(strcmp(device_name, "default_output") == 0) { + snprintf(p->device_name, sizeof(p->device_name), "%s", p->default_output_device_name); + p->device_type = DeviceType::DEFAULT_OUTPUT; + } else if(strcmp(device_name, "default_input") == 0) { + snprintf(p->device_name, sizeof(p->device_name), "%s", p->default_input_device_name); + p->device_type = DeviceType::DEFAULT_INPUT; + } else { + snprintf(p->device_name, sizeof(p->device_name), "%s", device_name); + p->device_type = DeviceType::STANDARD; } - pa_xfree(s); + return true; } static pa_handle* pa_sound_device_new(const char *server, const char *name, - const char *dev, + const char *device_name, const char *stream_name, const pa_sample_spec *ss, const pa_buffer_attr *attr, int *rerror) { pa_handle *p; - int error = PA_ERR_INTERNAL, r; + int error = PA_ERR_INTERNAL; + pa_operation *pa = NULL; p = pa_xnew0(pa_handle, 1); - p->read_data = NULL; - p->read_length = 0; - p->read_index = 0; + p->attr = *attr; + p->ss = *ss; + snprintf(p->stream_name, sizeof(p->stream_name), "%s", stream_name); + + p->reconnect = true; + p->reconnect_last_tried_seconds = clock_get_monotonic_seconds() - 1000.0; + p->default_output_device_name[0] = '\0'; + p->default_input_device_name[0] = '\0'; + p->device_type = DeviceType::STANDARD; - const int buffer_size = attr->maxlength; + const int buffer_size = attr->fragsize; void *buffer = malloc(buffer_size); if(!buffer) { - fprintf(stderr, "failed to allocate buffer for audio\n"); + fprintf(stderr, "Error: failed to allocate buffer for audio\n"); *rerror = -1; return NULL; } @@ -117,56 +231,96 @@ static pa_handle* pa_sound_device_new(const char *server, pa_mainloop_iterate(p->mainloop, 1, NULL); } - if (!(p->stream = pa_stream_new(p->context, stream_name, ss, NULL))) { - error = pa_context_errno(p->context); + if(!startup_get_default_devices(p, device_name)) goto fail; + + pa_context_set_subscribe_callback(p->context, subscribe_cb, p); + pa = pa_context_subscribe(p->context, PA_SUBSCRIPTION_MASK_SERVER, NULL, NULL); + if(pa) + pa_operation_unref(pa); + + return p; + +fail: + if (rerror) + *rerror = error; + pa_sound_device_free(p); + return NULL; +} + +static bool pa_sound_device_should_reconnect(pa_handle *p, double now, char *device_name, size_t device_name_size) { + std::lock_guard<std::mutex> lock(p->reconnect_mutex); + if(p->reconnect && now - p->reconnect_last_tried_seconds >= RECONNECT_TRY_TIMEOUT_SECONDS) { + p->reconnect_last_tried_seconds = now; + // TODO: Size check + snprintf(device_name, device_name_size, "%s", p->device_name); + return true; + } + return false; +} + +static bool pa_sound_device_handle_reconnect(pa_handle *p, char *device_name, size_t device_name_size, double now) { + int r; + if(!pa_sound_device_should_reconnect(p, now, device_name, device_name_size)) + return true; + + if(p->stream) { + pa_stream_disconnect(p->stream); + pa_stream_unref(p->stream); + p->stream = NULL; } - r = pa_stream_connect_record(p->stream, dev, attr, + if(!(p->stream = pa_stream_new(p->context, p->stream_name, &p->ss, NULL))) { + //pa_context_errno(p->context); + return false; + } + + r = pa_stream_connect_record(p->stream, device_name, &p->attr, (pa_stream_flags_t)(PA_STREAM_INTERPOLATE_TIMING|PA_STREAM_ADJUST_LATENCY|PA_STREAM_AUTO_TIMING_UPDATE)); - if (r < 0) { - error = pa_context_errno(p->context); - goto fail; + if(r < 0) { + //pa_context_errno(p->context); + return false; } - for (;;) { + for(;;) { pa_stream_state_t state = pa_stream_get_state(p->stream); - if (state == PA_STREAM_READY) + if(state == PA_STREAM_READY) break; - if (!PA_STREAM_IS_GOOD(state)) { - error = pa_context_errno(p->context); - goto fail; + if(!PA_STREAM_IS_GOOD(state)) { + //pa_context_errno(p->context); + return false; } pa_mainloop_iterate(p->mainloop, 1, NULL); } - return p; - -fail: - if (rerror) - *rerror = error; - pa_sound_device_free(p); - return NULL; + std::lock_guard<std::mutex> lock(p->reconnect_mutex); + p->reconnect = false; + return true; } -// Returns a negative value on failure or if |p->output_length| data is not available within the time frame specified by the sample rate -static int pa_sound_device_read(pa_handle *p) { +static int pa_sound_device_read(pa_handle *p, double timeout_seconds) { assert(p); - const int64_t timeout_ms = std::round((1000.0 / (double)pa_stream_get_sample_spec(p->stream)->rate) * 1000.0); const double start_time = clock_get_monotonic_seconds(); + char device_name[DEVICE_NAME_MAX_SIZE]; bool success = false; int r = 0; int *rerror = &r; + pa_usec_t latency = 0; + int negative = 0; + + if(!pa_sound_device_handle_reconnect(p, device_name, sizeof(device_name), start_time)) + goto fail; + CHECK_DEAD_GOTO(p, rerror, fail); while (p->output_index < p->output_length) { - if((clock_get_monotonic_seconds() - start_time) * 1000 >= timeout_ms) + if(clock_get_monotonic_seconds() - start_time >= timeout_seconds) return -1; if(!p->read_data) { @@ -195,6 +349,15 @@ static int pa_sound_device_read(pa_handle *p) { CHECK_DEAD_GOTO(p, rerror, fail); continue; } + + pa_operation_unref(pa_stream_update_timing_info(p->stream, NULL, NULL)); + // TODO: Deal with one pa_stream_peek not being enough. In that case we need to add multiple of these together(?) + if(pa_stream_get_latency(p->stream, &latency, &negative) >= 0) { + p->latency_seconds = negative ? -(double)latency : latency; + if(p->latency_seconds < 0.0) + p->latency_seconds = 0.0; + p->latency_seconds *= 0.0000001; + } } const size_t space_free_in_output_buffer = p->output_length - p->output_index; @@ -254,16 +417,16 @@ int sound_device_get_by_name(SoundDevice *device, const char *device_name, const ss.channels = num_channels; pa_buffer_attr buffer_attr; + buffer_attr.fragsize = period_frame_size * audio_format_to_get_bytes_per_sample(audio_format) * num_channels; // 2/4 bytes/sample, @num_channels channels buffer_attr.tlength = -1; buffer_attr.prebuf = -1; buffer_attr.minreq = -1; - buffer_attr.maxlength = period_frame_size * audio_format_to_get_bytes_per_sample(audio_format) * num_channels; // 2/4 bytes/sample, @num_channels channels - buffer_attr.fragsize = buffer_attr.maxlength; + buffer_attr.maxlength = buffer_attr.fragsize; int error = 0; pa_handle *handle = pa_sound_device_new(nullptr, description, device_name, description, &ss, &buffer_attr, &error); if(!handle) { - fprintf(stderr, "pa_sound_device_new() failed: %s. Audio input device %s might not be valid\n", pa_strerror(error), description); + fprintf(stderr, "Error: pa_sound_device_new() failed: %s. Audio input device %s might not be valid\n", pa_strerror(error), device_name); return -1; } @@ -278,13 +441,15 @@ void sound_device_close(SoundDevice *device) { device->handle = NULL; } -int sound_device_read_next_chunk(SoundDevice *device, void **buffer) { +int sound_device_read_next_chunk(SoundDevice *device, void **buffer, double timeout_sec, double *latency_seconds) { pa_handle *pa = (pa_handle*)device->handle; - if(pa_sound_device_read(pa) < 0) { + if(pa_sound_device_read(pa, timeout_sec) < 0) { //fprintf(stderr, "pa_simple_read() failed: %s\n", pa_strerror(error)); + *latency_seconds = 0.0; return -1; } *buffer = pa->output_data; + *latency_seconds = pa->latency_seconds; return device->frames; } @@ -308,26 +473,138 @@ static void pa_state_cb(pa_context *c, void *userdata) { } } -static void pa_sourcelist_cb(pa_context *ctx, const pa_source_info *source_info, int eol, void *userdata) { - (void)ctx; +static void pa_sourcelist_cb(pa_context*, const pa_source_info *source_info, int eol, void *userdata) { if(eol > 0) return; - std::vector<AudioInput> *inputs = (std::vector<AudioInput>*)userdata; - inputs->push_back({ source_info->name, source_info->description }); + AudioDevices *audio_devices = (AudioDevices*)userdata; + audio_devices->audio_inputs.push_back({ source_info->name, source_info->description }); +} + +static void pa_server_info_cb(pa_context*, const pa_server_info *server_info, void *userdata) { + AudioDevices *audio_devices = (AudioDevices*)userdata; + if(server_info->default_sink_name) + audio_devices->default_output = std::string(server_info->default_sink_name) + ".monitor"; + if(server_info->default_source_name) + audio_devices->default_input = server_info->default_source_name; } -std::vector<AudioInput> get_pulseaudio_inputs() { - std::vector<AudioInput> inputs; +static void server_info_callback(pa_context*, const pa_server_info *server_info, void *userdata) { + bool *is_server_pipewire = (bool*)userdata; + if(server_info->server_name && strstr(server_info->server_name, "PipeWire")) + *is_server_pipewire = true; +} + +static void get_pulseaudio_default_inputs(AudioDevices &audio_devices) { + int state = 0; + int pa_ready = 0; + pa_operation *pa_op = NULL; + pa_mainloop *main_loop = pa_mainloop_new(); + if(!main_loop) + return; pa_context *ctx = pa_context_new(pa_mainloop_get_api(main_loop), "gpu-screen-recorder"); - pa_context_connect(ctx, NULL, PA_CONTEXT_NOFLAGS, NULL); + if(pa_context_connect(ctx, NULL, PA_CONTEXT_NOFLAGS, NULL) < 0) + goto done; + + pa_context_set_state_callback(ctx, pa_state_cb, &pa_ready); + + for(;;) { + // Not ready + if(pa_ready == 0) { + pa_mainloop_iterate(main_loop, 1, NULL); + continue; + } + + switch(state) { + case 0: { + pa_op = pa_context_get_server_info(ctx, pa_server_info_cb, &audio_devices); + ++state; + break; + } + } + + // Couldn't get connection to the server + if(pa_ready == 2 || (state == 1 && pa_op && pa_operation_get_state(pa_op) == PA_OPERATION_DONE)) + break; + + pa_mainloop_iterate(main_loop, 1, NULL); + } + + done: + if(pa_op) + pa_operation_unref(pa_op); + pa_context_disconnect(ctx); + pa_context_unref(ctx); + pa_mainloop_free(main_loop); +} + +AudioDevices get_pulseaudio_inputs() { + AudioDevices audio_devices; int state = 0; int pa_ready = 0; + pa_operation *pa_op = NULL; + + // TODO: Do this in the same connection below instead of two separate connections + get_pulseaudio_default_inputs(audio_devices); + + pa_mainloop *main_loop = pa_mainloop_new(); + if(!main_loop) + return audio_devices; + + pa_context *ctx = pa_context_new(pa_mainloop_get_api(main_loop), "gpu-screen-recorder"); + if(pa_context_connect(ctx, NULL, PA_CONTEXT_NOFLAGS, NULL) < 0) + goto done; + pa_context_set_state_callback(ctx, pa_state_cb, &pa_ready); + for(;;) { + // Not ready + if(pa_ready == 0) { + pa_mainloop_iterate(main_loop, 1, NULL); + continue; + } + + switch(state) { + case 0: { + pa_op = pa_context_get_source_info_list(ctx, pa_sourcelist_cb, &audio_devices); + ++state; + break; + } + } + + // Couldn't get connection to the server + if(pa_ready == 2 || (state == 1 && pa_op && pa_operation_get_state(pa_op) == PA_OPERATION_DONE)) + break; + + pa_mainloop_iterate(main_loop, 1, NULL); + } + + done: + if(pa_op) + pa_operation_unref(pa_op); + pa_context_disconnect(ctx); + pa_context_unref(ctx); + pa_mainloop_free(main_loop); + return audio_devices; +} + +bool pulseaudio_server_is_pipewire() { + int state = 0; + int pa_ready = 0; pa_operation *pa_op = NULL; + bool is_server_pipewire = false; + + pa_mainloop *main_loop = pa_mainloop_new(); + if(!main_loop) + return is_server_pipewire; + + pa_context *ctx = pa_context_new(pa_mainloop_get_api(main_loop), "gpu-screen-recorder"); + if(pa_context_connect(ctx, NULL, PA_CONTEXT_NOFLAGS, NULL) < 0) + goto done; + + pa_context_set_state_callback(ctx, pa_state_cb, &pa_ready); for(;;) { // Not ready @@ -338,24 +615,24 @@ std::vector<AudioInput> get_pulseaudio_inputs() { switch(state) { case 0: { - pa_op = pa_context_get_source_info_list(ctx, pa_sourcelist_cb, &inputs); + pa_op = pa_context_get_server_info(ctx, server_info_callback, &is_server_pipewire); ++state; break; } } // Couldn't get connection to the server - if(pa_ready == 2 || (state == 1 && pa_op && pa_operation_get_state(pa_op) == PA_OPERATION_DONE)) { - if(pa_op) - pa_operation_unref(pa_op); - pa_context_disconnect(ctx); - pa_context_unref(ctx); + if(pa_ready == 2 || (state == 1 && pa_op && pa_operation_get_state(pa_op) == PA_OPERATION_DONE)) break; - } pa_mainloop_iterate(main_loop, 1, NULL); } + done: + if(pa_op) + pa_operation_unref(pa_op); + pa_context_disconnect(ctx); + pa_context_unref(ctx); pa_mainloop_free(main_loop); - return inputs; + return is_server_pipewire; } diff --git a/src/utils.c b/src/utils.c index 94b0037..9b4a4b6 100644 --- a/src/utils.c +++ b/src/utils.c @@ -1,12 +1,23 @@ #include "../include/utils.h" +#include "../include/window/window.h" + #include <time.h> #include <string.h> #include <stdio.h> #include <unistd.h> #include <fcntl.h> +#include <stdlib.h> +#include <sys/stat.h> +#include <sys/random.h> +#include <errno.h> +#include <assert.h> + #include <xf86drmMode.h> #include <xf86drm.h> -#include <stdlib.h> +#include <X11/Xatom.h> +#include <X11/extensions/Xrandr.h> +#include <libavcodec/avcodec.h> +#include <libavutil/hwcontext_vaapi.h> double clock_get_monotonic_seconds(void) { struct timespec ts; @@ -16,6 +27,25 @@ double clock_get_monotonic_seconds(void) { return (double)ts.tv_sec + (double)ts.tv_nsec * 0.000000001; } +bool generate_random_characters(char *buffer, int buffer_size, const char *alphabet, size_t alphabet_size) { + /* TODO: Use other functions on other platforms than linux */ + if(getrandom(buffer, buffer_size, 0) < buffer_size) { + fprintf(stderr, "Failed to get random bytes, error: %s\n", strerror(errno)); + return false; + } + + for(int i = 0; i < buffer_size; ++i) { + unsigned char c = *(unsigned char*)&buffer[i]; + buffer[i] = alphabet[c % alphabet_size]; + } + + return true; +} + +bool generate_random_characters_standard_alphabet(char *buffer, int buffer_size) { + return generate_random_characters(buffer, buffer_size, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789", 62); +} + static const XRRModeInfo* get_mode_info(const XRRScreenResources *sr, RRMode id) { for(int i = 0; i < sr->nmode; ++i) { if(sr->modes[i].id == id) @@ -24,29 +54,73 @@ static const XRRModeInfo* get_mode_info(const XRRScreenResources *sr, RRMode id) return NULL; } -static void for_each_active_monitor_output_x11(Display *display, active_monitor_callback callback, void *userdata) { +static gsr_monitor_rotation x11_rotation_to_gsr_rotation(int rot) { + switch(rot) { + case RR_Rotate_0: return GSR_MONITOR_ROT_0; + case RR_Rotate_90: return GSR_MONITOR_ROT_90; + case RR_Rotate_180: return GSR_MONITOR_ROT_180; + case RR_Rotate_270: return GSR_MONITOR_ROT_270; + } + return GSR_MONITOR_ROT_0; +} + +static uint32_t x11_output_get_connector_id(Display *dpy, RROutput output, Atom randr_connector_id_atom) { + Atom type = 0; + int format = 0; + unsigned long bytes_after = 0; + unsigned long nitems = 0; + unsigned char *prop = NULL; + XRRGetOutputProperty(dpy, output, randr_connector_id_atom, 0, 128, false, false, AnyPropertyType, &type, &format, &nitems, &bytes_after, &prop); + + long result = 0; + if(type == XA_INTEGER && format == 32) + result = *(long*)prop; + + free(prop); + return result; +} + +static vec2i get_monitor_size_rotated(int width, int height, gsr_monitor_rotation rotation) { + vec2i size = { .x = width, .y = height }; + if(rotation == GSR_MONITOR_ROT_90 || rotation == GSR_MONITOR_ROT_270) { + int tmp_x = size.x; + size.x = size.y; + size.y = tmp_x; + } + return size; +} + +void for_each_active_monitor_output_x11_not_cached(Display *display, active_monitor_callback callback, void *userdata) { XRRScreenResources *screen_res = XRRGetScreenResources(display, DefaultRootWindow(display)); if(!screen_res) return; + const Atom randr_connector_id_atom = XInternAtom(display, "CONNECTOR_ID", False); + char display_name[256]; for(int i = 0; i < screen_res->noutput; ++i) { XRROutputInfo *out_info = XRRGetOutputInfo(display, screen_res, screen_res->outputs[i]); if(out_info && out_info->crtc && out_info->connection == RR_Connected) { XRRCrtcInfo *crt_info = XRRGetCrtcInfo(display, screen_res, out_info->crtc); if(crt_info && crt_info->mode) { + // We want to use the current mode info width/height (mode_info->width/height) instead of crtc info width/height (crt_info->width/height) because crtc info + // is scaled if the monitor is scaled (xrandr --output DP-1 --scale 1.5). Normally this is not an issue for x11 applications, + // but gpu screen recorder captures the drm framebuffer instead of x11 api. This drm framebuffer which doesn't increase in size when using xrandr scaling. + // Maybe a better option would be to get the drm crtc size instead. const XRRModeInfo *mode_info = get_mode_info(screen_res, crt_info->mode); if(mode_info && out_info->nameLen < (int)sizeof(display_name)) { - memcpy(display_name, out_info->name, out_info->nameLen); - display_name[out_info->nameLen] = '\0'; + snprintf(display_name, sizeof(display_name), "%.*s", (int)out_info->nameLen, out_info->name); + const gsr_monitor_rotation rotation = x11_rotation_to_gsr_rotation(crt_info->rotation); + const vec2i monitor_size = get_monitor_size_rotated(mode_info->width, mode_info->height, rotation); - gsr_monitor monitor = { + const gsr_monitor monitor = { .name = display_name, .name_len = out_info->nameLen, .pos = { .x = crt_info->x, .y = crt_info->y }, - .size = { .x = (int)crt_info->width, .y = (int)crt_info->height }, - .crt_info = crt_info, - .connector_id = 0 // TODO: Get connector id + .size = monitor_size, + .connector_id = x11_output_get_connector_id(display, screen_res->outputs[i], randr_connector_id_atom), + .rotation = rotation, + .monitor_identifier = out_info->crtc }; callback(&monitor, userdata); } @@ -61,14 +135,22 @@ static void for_each_active_monitor_output_x11(Display *display, active_monitor_ XRRFreeScreenResources(screen_res); } -typedef struct { - int type; - int count; -} drm_connector_type_count; - -#define CONNECTOR_TYPE_COUNTS 32 +/* TODO: Support more connector types */ +int get_connector_type_by_name(const char *name) { + int len = strlen(name); + if(len >= 5 && strncmp(name, "HDMI-", 5) == 0) + return 1; + else if(len >= 3 && strncmp(name, "DP-", 3) == 0) + return 2; + else if(len >= 12 && strncmp(name, "DisplayPort-", 12) == 0) + return 3; + else if(len >= 4 && strncmp(name, "eDP-", 4) == 0) + return 4; + else + return -1; +} -static drm_connector_type_count* drm_connector_types_get_index(drm_connector_type_count *type_counts, int *num_type_counts, int connector_type) { +drm_connector_type_count* drm_connector_types_get_index(drm_connector_type_count *type_counts, int *num_type_counts, int connector_type) { for(int i = 0; i < *num_type_counts; ++i) { if(type_counts[i].type == connector_type) return &type_counts[i]; @@ -80,10 +162,15 @@ static drm_connector_type_count* drm_connector_types_get_index(drm_connector_typ const int index = *num_type_counts; type_counts[index].type = connector_type; type_counts[index].count = 0; + type_counts[index].count_active = 0; ++*num_type_counts; return &type_counts[index]; } +uint32_t monitor_identifier_from_type_and_count(int monitor_type_index, int monitor_type_count) { + return ((uint32_t)monitor_type_index << 16) | ((uint32_t)monitor_type_count); +} + static bool connector_get_property_by_name(int drmfd, drmModeConnectorPtr props, const char *name, uint64_t *result) { for(int i = 0; i < props->count_props; ++i) { drmModePropertyPtr prop = drmModeGetProperty(drmfd, props->props[i]); @@ -99,30 +186,12 @@ static bool connector_get_property_by_name(int drmfd, drmModeConnectorPtr props, return false; } -static void for_each_active_monitor_output_wayland(gsr_egl *egl, active_monitor_callback callback, void *userdata) { - if(!gsr_egl_supports_wayland_capture(egl)) +static void for_each_active_monitor_output_drm(const char *card_path, active_monitor_callback callback, void *userdata) { + int fd = open(card_path, O_RDONLY); + if(fd == -1) { + fprintf(stderr, "gsr error: for_each_active_monitor_output_drm failed, failed to open \"%s\", error: %s\n", card_path, strerror(errno)); return; - - for(int i = 0; i < egl->wayland.num_outputs; ++i) { - if(!egl->wayland.outputs[i].name) - continue; - - gsr_monitor monitor = { - .name = egl->wayland.outputs[i].name, - .name_len = strlen(egl->wayland.outputs[i].name), - .pos = { .x = egl->wayland.outputs[i].pos.x, .y = egl->wayland.outputs[i].pos.y }, - .size = { .x = egl->wayland.outputs[i].size.x, .y = egl->wayland.outputs[i].size.y }, - .crt_info = NULL, - .connector_id = 0 - }; - callback(&monitor, userdata); } -} - -static void for_each_active_monitor_output_drm(const char *drm_card_path, active_monitor_callback callback, void *userdata) { - int fd = open(drm_card_path, O_RDONLY); - if(fd == -1) - return; drmSetClientCap(fd, DRM_CLIENT_CAP_ATOMIC, 1); @@ -148,19 +217,24 @@ static void for_each_active_monitor_output_drm(const char *drm_card_path, active continue; } + if(connector_type) + ++connector_type->count_active; + uint64_t crtc_id = 0; connector_get_property_by_name(fd, connector, "CRTC_ID", &crtc_id); drmModeCrtcPtr crtc = drmModeGetCrtc(fd, crtc_id); if(connector_type && crtc_id > 0 && crtc && connection_name_len + 5 < (int)sizeof(display_name)) { const int display_name_len = snprintf(display_name, sizeof(display_name), "%s-%d", connection_name, connector_type->count); + const int connector_type_index_name = get_connector_type_by_name(display_name); gsr_monitor monitor = { .name = display_name, .name_len = display_name_len, .pos = { .x = crtc->x, .y = crtc->y }, .size = { .x = (int)crtc->width, .y = (int)crtc->height }, - .crt_info = NULL, - .connector_id = connector->connector_id + .connector_id = connector->connector_id, + .rotation = GSR_MONITOR_ROT_0, + .monitor_identifier = connector_type_index_name != -1 ? monitor_identifier_from_type_and_count(connector_type_index_name, connector_type->count_active) : 0 }; callback(&monitor, userdata); } @@ -176,16 +250,14 @@ static void for_each_active_monitor_output_drm(const char *drm_card_path, active close(fd); } -void for_each_active_monitor_output(void *connection, gsr_connection_type connection_type, active_monitor_callback callback, void *userdata) { +void for_each_active_monitor_output(const gsr_window *window, const char *card_path, gsr_connection_type connection_type, active_monitor_callback callback, void *userdata) { switch(connection_type) { case GSR_CONNECTION_X11: - for_each_active_monitor_output_x11(connection, callback, userdata); - break; case GSR_CONNECTION_WAYLAND: - for_each_active_monitor_output_wayland(connection, callback, userdata); + gsr_window_for_each_active_monitor_output_cached(window, callback, userdata); break; case GSR_CONNECTION_DRM: - for_each_active_monitor_output_drm(connection, callback, userdata); + for_each_active_monitor_output_drm(card_path, callback, userdata); break; } } @@ -195,20 +267,96 @@ static void get_monitor_by_name_callback(const gsr_monitor *monitor, void *userd if(!data->found_monitor && strcmp(data->name, monitor->name) == 0) { data->monitor->pos = monitor->pos; data->monitor->size = monitor->size; + data->monitor->connector_id = monitor->connector_id; + data->monitor->rotation = monitor->rotation; + data->monitor->monitor_identifier = monitor->monitor_identifier; data->found_monitor = true; } } -bool get_monitor_by_name(void *connection, gsr_connection_type connection_type, const char *name, gsr_monitor *monitor) { +bool get_monitor_by_name(const gsr_egl *egl, gsr_connection_type connection_type, const char *name, gsr_monitor *monitor) { get_monitor_by_name_userdata userdata; userdata.name = name; userdata.name_len = strlen(name); userdata.monitor = monitor; userdata.found_monitor = false; - for_each_active_monitor_output(connection, connection_type, get_monitor_by_name_callback, &userdata); + for_each_active_monitor_output(egl->window, egl->card_path, connection_type, get_monitor_by_name_callback, &userdata); return userdata.found_monitor; } +typedef struct { + const gsr_monitor *monitor; + gsr_monitor_rotation rotation; + vec2i position; + bool match_found; +} get_monitor_by_connector_id_userdata; + +static bool vec2i_eql(vec2i a, vec2i b) { + return a.x == b.x && a.y == b.y; +} + +static void get_monitor_by_name_and_size_callback(const gsr_monitor *monitor, void *userdata) { + get_monitor_by_connector_id_userdata *data = (get_monitor_by_connector_id_userdata*)userdata; + if(monitor->name && data->monitor->name && strcmp(monitor->name, data->monitor->name) == 0 && vec2i_eql(monitor->size, data->monitor->size)) { + data->rotation = monitor->rotation; + data->position = monitor->pos; + data->match_found = true; + } +} + +static void get_monitor_by_connector_id_callback(const gsr_monitor *monitor, void *userdata) { + get_monitor_by_connector_id_userdata *data = (get_monitor_by_connector_id_userdata*)userdata; + if(monitor->connector_id == data->monitor->connector_id || + (!monitor->connector_id && monitor->monitor_identifier == data->monitor->monitor_identifier)) + { + data->rotation = monitor->rotation; + data->position = monitor->pos; + data->match_found = true; + } +} + +bool drm_monitor_get_display_server_data(const gsr_window *window, const gsr_monitor *monitor, gsr_monitor_rotation *monitor_rotation, vec2i *monitor_position) { + *monitor_rotation = GSR_MONITOR_ROT_0; + *monitor_position = (vec2i){0, 0}; + + if(gsr_window_get_display_server(window) == GSR_DISPLAY_SERVER_WAYLAND) { + { + get_monitor_by_connector_id_userdata userdata; + userdata.monitor = monitor; + userdata.rotation = GSR_MONITOR_ROT_0; + userdata.position = (vec2i){0, 0}; + userdata.match_found = false; + gsr_window_for_each_active_monitor_output_cached(window, get_monitor_by_name_and_size_callback, &userdata); + if(userdata.match_found) { + *monitor_rotation = userdata.rotation; + *monitor_position = userdata.position; + return true; + } + } + { + get_monitor_by_connector_id_userdata userdata; + userdata.monitor = monitor; + userdata.rotation = GSR_MONITOR_ROT_0; + userdata.position = (vec2i){0, 0}; + userdata.match_found = false; + gsr_window_for_each_active_monitor_output_cached(window, get_monitor_by_connector_id_callback, &userdata); + *monitor_rotation = userdata.rotation; + *monitor_position = userdata.position; + return userdata.match_found; + } + } else { + get_monitor_by_connector_id_userdata userdata; + userdata.monitor = monitor; + userdata.rotation = GSR_MONITOR_ROT_0; + userdata.position = (vec2i){0, 0}; + userdata.match_found = false; + gsr_window_for_each_active_monitor_output_cached(window, get_monitor_by_connector_id_callback, &userdata); + *monitor_rotation = userdata.rotation; + *monitor_position = userdata.position; + return userdata.match_found; + } +} + bool gl_get_gpu_info(gsr_egl *egl, gsr_gpu_info *info) { const char *software_renderers[] = { "llvmpipe", "SWR", "softpipe", NULL }; bool supported = true; @@ -216,6 +364,7 @@ bool gl_get_gpu_info(gsr_egl *egl, gsr_gpu_info *info) { const unsigned char *gl_renderer = egl->glGetString(GL_RENDERER); info->gpu_version = 0; + info->is_steam_deck = false; if(!gl_vendor) { fprintf(stderr, "gsr error: failed to get gpu vendor\n"); @@ -235,10 +384,14 @@ bool gl_get_gpu_info(gsr_egl *egl, gsr_gpu_info *info) { if(strstr((const char*)gl_vendor, "AMD")) info->vendor = GSR_GPU_VENDOR_AMD; + else if(strstr((const char*)gl_vendor, "Mesa") && gl_renderer && strstr((const char*)gl_renderer, "AMD")) + info->vendor = GSR_GPU_VENDOR_AMD; else if(strstr((const char*)gl_vendor, "Intel")) info->vendor = GSR_GPU_VENDOR_INTEL; else if(strstr((const char*)gl_vendor, "NVIDIA")) info->vendor = GSR_GPU_VENDOR_NVIDIA; + else if(strstr((const char*)gl_vendor, "Broadcom")) + info->vendor = GSR_GPU_VENDOR_BROADCOM; else { fprintf(stderr, "gsr error: unknown gpu vendor: %s\n", gl_vendor); supported = false; @@ -248,53 +401,66 @@ bool gl_get_gpu_info(gsr_egl *egl, gsr_gpu_info *info) { if(gl_renderer) { if(info->vendor == GSR_GPU_VENDOR_NVIDIA) sscanf((const char*)gl_renderer, "%*s %*s %*s %d", &info->gpu_version); + info->is_steam_deck = strstr((const char*)gl_renderer, "vangogh") != NULL; } end: return supported; } -bool gsr_get_valid_card_path(char *output) { - for(int i = 0; i < 10; ++i) { - drmVersion *ver = NULL; - drmModePlaneResPtr planes = NULL; - bool found_screen_card = false; +bool try_card_has_valid_plane(const char *card_path) { + drmVersion *ver = NULL; + drmModePlaneResPtr planes = NULL; + bool found_screen_card = false; - sprintf(output, DRM_DEV_NAME, DRM_DIR_NAME, i); - int fd = open(output, O_RDONLY); - if(fd == -1) - continue; - - ver = drmGetVersion(fd); - if(!ver || strstr(ver->name, "nouveau")) - goto next; + int fd = open(card_path, O_RDONLY); + if(fd == -1) + return false; - drmSetClientCap(fd, DRM_CLIENT_CAP_UNIVERSAL_PLANES, 1); + ver = drmGetVersion(fd); + if(!ver || strstr(ver->name, "nouveau")) + goto next; - planes = drmModeGetPlaneResources(fd); - if(!planes) - goto next; + drmSetClientCap(fd, DRM_CLIENT_CAP_UNIVERSAL_PLANES, 1); - for(uint32_t j = 0; j < planes->count_planes; ++j) { - drmModePlanePtr plane = drmModeGetPlane(fd, planes->planes[j]); - if(!plane) - continue; + planes = drmModeGetPlaneResources(fd); + if(!planes) + goto next; - if(plane->fb_id) - found_screen_card = true; + for(uint32_t j = 0; j < planes->count_planes; ++j) { + drmModePlanePtr plane = drmModeGetPlane(fd, planes->planes[j]); + if(!plane) + continue; - drmModeFreePlane(plane); - if(found_screen_card) - break; - } + if(plane->fb_id) + found_screen_card = true; - next: - if(planes) - drmModeFreePlaneResources(planes); - if(ver) - drmFreeVersion(ver); - close(fd); + drmModeFreePlane(plane); if(found_screen_card) + break; + } + + next: + if(planes) + drmModeFreePlaneResources(planes); + if(ver) + drmFreeVersion(ver); + close(fd); + if(found_screen_card) + return true; + + return false; +} + +bool gsr_get_valid_card_path(gsr_egl *egl, char *output, bool is_monitor_capture) { + if(egl->dri_card_path) { + snprintf(output, 128, "%s", egl->dri_card_path); + return is_monitor_capture ? try_card_has_valid_plane(output) : true; + } + + for(int i = 0; i < 10; ++i) { + snprintf(output, 128, DRM_DEV_NAME, DRM_DIR_NAME, i); + if(try_card_has_valid_plane(output)) return true; } return false; @@ -307,7 +473,7 @@ bool gsr_card_path_get_render_path(const char *card_path, char *render_path) { char *render_path_tmp = drmGetRenderDeviceNameFromFd(fd); if(render_path_tmp) { - strncpy(render_path, render_path_tmp, 128); + snprintf(render_path, 128, "%s", render_path_tmp); free(render_path_tmp); close(fd); return true; @@ -317,6 +483,194 @@ bool gsr_card_path_get_render_path(const char *card_path, char *render_path) { return false; } -int even_number_ceil(int value) { - return value + (value & 1); +int create_directory_recursive(char *path) { + int path_len = strlen(path); + char *p = path; + char *end = path + path_len; + for(;;) { + char *slash_p = strchr(p, '/'); + + // Skips first '/', we don't want to try and create the root directory + if(slash_p == path) { + ++p; + continue; + } + + if(!slash_p) + slash_p = end; + + char prev_char = *slash_p; + *slash_p = '\0'; + int err = mkdir(path, S_IRWXU); + *slash_p = prev_char; + + if(err == -1 && errno != EEXIST) + return err; + + if(slash_p == end) + break; + else + p = slash_p + 1; + } + return 0; +} + +void setup_dma_buf_attrs(intptr_t *img_attr, uint32_t format, uint32_t width, uint32_t height, const int *fds, const uint32_t *offsets, const uint32_t *pitches, const uint64_t *modifiers, int num_planes, bool use_modifier) { + size_t img_attr_index = 0; + + img_attr[img_attr_index++] = EGL_LINUX_DRM_FOURCC_EXT; + img_attr[img_attr_index++] = format; + + img_attr[img_attr_index++] = EGL_WIDTH; + img_attr[img_attr_index++] = width; + + img_attr[img_attr_index++] = EGL_HEIGHT; + img_attr[img_attr_index++] = height; + + if(num_planes >= 1) { + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE0_FD_EXT; + img_attr[img_attr_index++] = fds[0]; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE0_OFFSET_EXT; + img_attr[img_attr_index++] = offsets[0]; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE0_PITCH_EXT; + img_attr[img_attr_index++] = pitches[0]; + + if(use_modifier) { + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE0_MODIFIER_LO_EXT; + img_attr[img_attr_index++] = modifiers[0] & 0xFFFFFFFFULL; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE0_MODIFIER_HI_EXT; + img_attr[img_attr_index++] = modifiers[0] >> 32ULL; + } + } + + if(num_planes >= 2) { + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE1_FD_EXT; + img_attr[img_attr_index++] = fds[1]; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE1_OFFSET_EXT; + img_attr[img_attr_index++] = offsets[1]; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE1_PITCH_EXT; + img_attr[img_attr_index++] = pitches[1]; + + if(use_modifier) { + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE1_MODIFIER_LO_EXT; + img_attr[img_attr_index++] = modifiers[1] & 0xFFFFFFFFULL; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE1_MODIFIER_HI_EXT; + img_attr[img_attr_index++] = modifiers[1] >> 32ULL; + } + } + + if(num_planes >= 3) { + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE2_FD_EXT; + img_attr[img_attr_index++] = fds[2]; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE2_OFFSET_EXT; + img_attr[img_attr_index++] = offsets[2]; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE2_PITCH_EXT; + img_attr[img_attr_index++] = pitches[2]; + + if(use_modifier) { + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE2_MODIFIER_LO_EXT; + img_attr[img_attr_index++] = modifiers[2] & 0xFFFFFFFFULL; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE2_MODIFIER_HI_EXT; + img_attr[img_attr_index++] = modifiers[2] >> 32ULL; + } + } + + if(num_planes >= 4) { + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE3_FD_EXT; + img_attr[img_attr_index++] = fds[3]; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE3_OFFSET_EXT; + img_attr[img_attr_index++] = offsets[3]; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE3_PITCH_EXT; + img_attr[img_attr_index++] = pitches[3]; + + if(use_modifier) { + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE3_MODIFIER_LO_EXT; + img_attr[img_attr_index++] = modifiers[3] & 0xFFFFFFFFULL; + + img_attr[img_attr_index++] = EGL_DMA_BUF_PLANE3_MODIFIER_HI_EXT; + img_attr[img_attr_index++] = modifiers[3] >> 32ULL; + } + } + + img_attr[img_attr_index++] = EGL_NONE; + assert(img_attr_index <= 44); +} + +static VADisplay video_codec_context_get_vaapi_display(AVCodecContext *video_codec_context) { + AVBufferRef *hw_frames_ctx = video_codec_context->hw_frames_ctx; + if(!hw_frames_ctx) + return NULL; + + AVHWFramesContext *hw_frame_context = (AVHWFramesContext*)hw_frames_ctx->data; + AVHWDeviceContext *device_context = (AVHWDeviceContext*)hw_frame_context->device_ctx; + if(device_context->type != AV_HWDEVICE_TYPE_VAAPI) + return NULL; + + AVVAAPIDeviceContext *vactx = device_context->hwctx; + return vactx->display; +} + +bool video_codec_context_is_vaapi(AVCodecContext *video_codec_context) { + if(!video_codec_context) + return false; + + AVBufferRef *hw_frames_ctx = video_codec_context->hw_frames_ctx; + if(!hw_frames_ctx) + return false; + + AVHWFramesContext *hw_frame_context = (AVHWFramesContext*)hw_frames_ctx->data; + AVHWDeviceContext *device_context = (AVHWDeviceContext*)hw_frame_context->device_ctx; + return device_context->type == AV_HWDEVICE_TYPE_VAAPI; +} + +vec2i scale_keep_aspect_ratio(vec2i from, vec2i to) { + if(from.x == 0 || from.y == 0) + return (vec2i){0, 0}; + + const double height_to_width_ratio = (double)from.y / (double)from.x; + from.x = to.x; + from.y = from.x * height_to_width_ratio; + + if(from.y > to.y) { + const double width_height_ratio = (double)from.x / (double)from.y; + from.y = to.y; + from.x = from.y * width_height_ratio; + } + + return from; +} + +unsigned int gl_create_texture(gsr_egl *egl, int width, int height, int internal_format, unsigned int format, int filter) { + float border_color[4] = { 0.0f, 0.0f, 0.0f, 1.0f }; + if(format == GL_RG) { // UV + border_color[0] = 0.5f; + border_color[1] = 0.5f; + border_color[2] = 0.0f; + border_color[3] = 1.0f; + } + + unsigned int texture_id = 0; + egl->glGenTextures(1, &texture_id); + egl->glBindTexture(GL_TEXTURE_2D, texture_id); + egl->glTexImage2D(GL_TEXTURE_2D, 0, internal_format, width, height, 0, format, GL_UNSIGNED_BYTE, NULL); + + egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + egl->glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, border_color); + egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, filter); + egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, filter); + + egl->glBindTexture(GL_TEXTURE_2D, 0); + return texture_id; } diff --git a/src/window/wayland.c b/src/window/wayland.c new file mode 100644 index 0000000..b7907be --- /dev/null +++ b/src/window/wayland.c @@ -0,0 +1,402 @@ +#include "../../include/window/wayland.h" + +#include "../../include/vec2.h" +#include "../../include/defs.h" + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdint.h> +#include <wayland-client.h> +#include <wayland-egl.h> +#include "xdg-output-unstable-v1-client-protocol.h" + +#define GSR_MAX_OUTPUTS 32 + +typedef struct gsr_window_wayland gsr_window_wayland; + +typedef struct { + uint32_t wl_name; + struct wl_output *output; + struct zxdg_output_v1 *xdg_output; + vec2i pos; + vec2i size; + int32_t transform; + char *name; +} gsr_wayland_output; + +struct gsr_window_wayland { + struct wl_display *display; + struct wl_egl_window *window; + struct wl_registry *registry; + struct wl_surface *surface; + struct wl_compositor *compositor; + gsr_wayland_output outputs[GSR_MAX_OUTPUTS]; + int num_outputs; + struct zxdg_output_manager_v1 *xdg_output_manager; +}; + +static void output_handle_geometry(void *data, struct wl_output *wl_output, + int32_t x, int32_t y, int32_t phys_width, int32_t phys_height, + int32_t subpixel, const char *make, const char *model, + int32_t transform) { + (void)wl_output; + (void)phys_width; + (void)phys_height; + (void)subpixel; + (void)make; + (void)model; + gsr_wayland_output *gsr_output = data; + gsr_output->pos.x = x; + gsr_output->pos.y = y; + gsr_output->transform = transform; +} + +static void output_handle_mode(void *data, struct wl_output *wl_output, uint32_t flags, int32_t width, int32_t height, int32_t refresh) { + (void)wl_output; + (void)flags; + (void)refresh; + gsr_wayland_output *gsr_output = data; + gsr_output->size.x = width; + gsr_output->size.y = height; +} + +static void output_handle_done(void *data, struct wl_output *wl_output) { + (void)data; + (void)wl_output; +} + +static void output_handle_scale(void* data, struct wl_output *wl_output, int32_t factor) { + (void)data; + (void)wl_output; + (void)factor; +} + +static void output_handle_name(void *data, struct wl_output *wl_output, const char *name) { + (void)wl_output; + gsr_wayland_output *gsr_output = data; + if(gsr_output->name) { + free(gsr_output->name); + gsr_output->name = NULL; + } + gsr_output->name = strdup(name); +} + +static void output_handle_description(void *data, struct wl_output *wl_output, const char *description) { + (void)data; + (void)wl_output; + (void)description; +} + +static const struct wl_output_listener output_listener = { + .geometry = output_handle_geometry, + .mode = output_handle_mode, + .done = output_handle_done, + .scale = output_handle_scale, + .name = output_handle_name, + .description = output_handle_description, +}; + +static void registry_add_object(void *data, struct wl_registry *registry, uint32_t name, const char *interface, uint32_t version) { + (void)version; + gsr_window_wayland *window_wayland = data; + if(strcmp(interface, "wl_compositor") == 0) { + if(window_wayland->compositor) + return; + + window_wayland->compositor = wl_registry_bind(registry, name, &wl_compositor_interface, 1); + } else if(strcmp(interface, wl_output_interface.name) == 0) { + if(version < 4) { + fprintf(stderr, "gsr warning: wl output interface version is < 4, expected >= 4 to capture a monitor\n"); + return; + } + + if(window_wayland->num_outputs == GSR_MAX_OUTPUTS) { + fprintf(stderr, "gsr warning: reached maximum outputs (%d), ignoring output %u\n", GSR_MAX_OUTPUTS, name); + return; + } + + gsr_wayland_output *gsr_output = &window_wayland->outputs[window_wayland->num_outputs]; + window_wayland->num_outputs++; + *gsr_output = (gsr_wayland_output) { + .wl_name = name, + .output = wl_registry_bind(registry, name, &wl_output_interface, 4), + .pos = { .x = 0, .y = 0 }, + .size = { .x = 0, .y = 0 }, + .transform = 0, + .name = NULL, + }; + wl_output_add_listener(gsr_output->output, &output_listener, gsr_output); + } else if(strcmp(interface, zxdg_output_manager_v1_interface.name) == 0) { + if(version < 1) { + fprintf(stderr, "gsr warning: xdg output interface version is < 1, expected >= 1 to capture a monitor\n"); + return; + } + + if(window_wayland->xdg_output_manager) + return; + + window_wayland->xdg_output_manager = wl_registry_bind(registry, name, &zxdg_output_manager_v1_interface, 1); + } +} + +static void registry_remove_object(void *data, struct wl_registry *registry, uint32_t name) { + (void)data; + (void)registry; + (void)name; + // TODO: Remove output +} + +static struct wl_registry_listener registry_listener = { + .global = registry_add_object, + .global_remove = registry_remove_object, +}; + +static void xdg_output_logical_position(void *data, struct zxdg_output_v1 *zxdg_output_v1, int32_t x, int32_t y) { + (void)zxdg_output_v1; + gsr_wayland_output *gsr_xdg_output = data; + gsr_xdg_output->pos.x = x; + gsr_xdg_output->pos.y = y; +} + +static void xdg_output_handle_logical_size(void *data, struct zxdg_output_v1 *xdg_output, int32_t width, int32_t height) { + (void)data; + (void)xdg_output; + (void)width; + (void)height; +} + +static void xdg_output_handle_done(void *data, struct zxdg_output_v1 *xdg_output) { + (void)data; + (void)xdg_output; +} + +static void xdg_output_handle_name(void *data, struct zxdg_output_v1 *xdg_output, const char *name) { + (void)data; + (void)xdg_output; + (void)name; +} + +static void xdg_output_handle_description(void *data, struct zxdg_output_v1 *xdg_output, const char *description) { + (void)data; + (void)xdg_output; + (void)description; +} + +static const struct zxdg_output_v1_listener xdg_output_listener = { + .logical_position = xdg_output_logical_position, + .logical_size = xdg_output_handle_logical_size, + .done = xdg_output_handle_done, + .name = xdg_output_handle_name, + .description = xdg_output_handle_description, +}; + +static void gsr_window_wayland_set_monitor_outputs_from_xdg_output(gsr_window_wayland *self) { + if(!self->xdg_output_manager) { + fprintf(stderr, "gsr warning: zxdg_output_manager not found. registered monitor positions might be incorrect\n"); + return; + } + + for(int i = 0; i < self->num_outputs; ++i) { + self->outputs[i].xdg_output = zxdg_output_manager_v1_get_xdg_output(self->xdg_output_manager, self->outputs[i].output); + zxdg_output_v1_add_listener(self->outputs[i].xdg_output, &xdg_output_listener, &self->outputs[i]); + } + + // Fetch xdg_output + wl_display_roundtrip(self->display); +} + +static void gsr_window_wayland_deinit(gsr_window_wayland *self) { + if(self->window) { + wl_egl_window_destroy(self->window); + self->window = NULL; + } + + if(self->surface) { + wl_surface_destroy(self->surface); + self->surface = NULL; + } + + for(int i = 0; i < self->num_outputs; ++i) { + if(self->outputs[i].output) { + wl_output_destroy(self->outputs[i].output); + self->outputs[i].output = NULL; + } + + if(self->outputs[i].name) { + free(self->outputs[i].name); + self->outputs[i].name = NULL; + } + + if(self->outputs[i].xdg_output) { + zxdg_output_v1_destroy(self->outputs[i].xdg_output); + self->outputs[i].output = NULL; + } + } + self->num_outputs = 0; + + if(self->xdg_output_manager) { + zxdg_output_manager_v1_destroy(self->xdg_output_manager); + self->xdg_output_manager = NULL; + } + + if(self->compositor) { + wl_compositor_destroy(self->compositor); + self->compositor = NULL; + } + + if(self->registry) { + wl_registry_destroy(self->registry); + self->registry = NULL; + } + + if(self->display) { + wl_display_disconnect(self->display); + self->display = NULL; + } +} + +static bool gsr_window_wayland_init(gsr_window_wayland *self) { + self->display = wl_display_connect(NULL); + if(!self->display) { + fprintf(stderr, "gsr error: gsr_window_wayland_init failed: failed to connect to the Wayland server\n"); + goto fail; + } + + self->registry = wl_display_get_registry(self->display); // TODO: Error checking + wl_registry_add_listener(self->registry, ®istry_listener, self); // TODO: Error checking + + // Fetch globals + wl_display_roundtrip(self->display); + + // Fetch wl_output + wl_display_roundtrip(self->display); + + gsr_window_wayland_set_monitor_outputs_from_xdg_output(self); + + if(!self->compositor) { + fprintf(stderr, "gsr error: gsr_window_wayland_init failed: failed to find compositor\n"); + goto fail; + } + + self->surface = wl_compositor_create_surface(self->compositor); + if(!self->surface) { + fprintf(stderr, "gsr error: gsr_window_wayland_init failed: failed to create surface\n"); + goto fail; + } + + self->window = wl_egl_window_create(self->surface, 16, 16); + if(!self->window) { + fprintf(stderr, "gsr error: gsr_window_wayland_init failed: failed to create window\n"); + goto fail; + } + + return true; + + fail: + gsr_window_wayland_deinit(self); + return false; +} + +static void gsr_window_wayland_destroy(gsr_window *window) { + gsr_window_wayland *self = window->priv; + gsr_window_wayland_deinit(self); + free(self); + free(window); +} + +static bool gsr_window_wayland_process_event(gsr_window *window) { + gsr_window_wayland *self = window->priv; + // TODO: pselect on wl_display_get_fd before doing dispatch + const bool events_available = wl_display_dispatch_pending(self->display) > 0; + wl_display_flush(self->display); + return events_available; +} + +static gsr_display_server gsr_wayland_get_display_server(void) { + return GSR_DISPLAY_SERVER_WAYLAND; +} + +static void* gsr_window_wayland_get_display(gsr_window *window) { + gsr_window_wayland *self = window->priv; + return self->display; +} + +static void* gsr_window_wayland_get_window(gsr_window *window) { + gsr_window_wayland *self = window->priv; + return self->window; +} + +static gsr_monitor_rotation wayland_transform_to_gsr_rotation(int32_t rot) { + switch(rot) { + case 0: return GSR_MONITOR_ROT_0; + case 1: return GSR_MONITOR_ROT_90; + case 2: return GSR_MONITOR_ROT_180; + case 3: return GSR_MONITOR_ROT_270; + } + return GSR_MONITOR_ROT_0; +} + +static void gsr_window_wayland_for_each_active_monitor_output_cached(const gsr_window *window, active_monitor_callback callback, void *userdata) { + const gsr_window_wayland *self = window->priv; + drm_connector_type_count type_counts[CONNECTOR_TYPE_COUNTS]; + int num_type_counts = 0; + + for(int i = 0; i < self->num_outputs; ++i) { + const gsr_wayland_output *output = &self->outputs[i]; + if(!output->name) + continue; + + const int connector_type_index = get_connector_type_by_name(output->name); + drm_connector_type_count *connector_type = NULL; + if(connector_type_index != -1) + connector_type = drm_connector_types_get_index(type_counts, &num_type_counts, connector_type_index); + + if(connector_type) { + ++connector_type->count; + ++connector_type->count_active; + } + + const gsr_monitor monitor = { + .name = output->name, + .name_len = strlen(output->name), + .pos = { .x = output->pos.x, .y = output->pos.y }, + .size = { .x = output->size.x, .y = output->size.y }, + .connector_id = 0, + .rotation = wayland_transform_to_gsr_rotation(output->transform), + .monitor_identifier = connector_type ? monitor_identifier_from_type_and_count(connector_type_index, connector_type->count_active) : 0 + }; + callback(&monitor, userdata); + } +} + +gsr_window* gsr_window_wayland_create(void) { + gsr_window *window = calloc(1, sizeof(gsr_window)); + if(!window) + return window; + + gsr_window_wayland *window_wayland = calloc(1, sizeof(gsr_window_wayland)); + if(!window_wayland) { + free(window); + return NULL; + } + + if(!gsr_window_wayland_init(window_wayland)) { + free(window_wayland); + free(window); + return NULL; + } + + *window = (gsr_window) { + .destroy = gsr_window_wayland_destroy, + .process_event = gsr_window_wayland_process_event, + .get_event_data = NULL, + .get_display_server = gsr_wayland_get_display_server, + .get_display = gsr_window_wayland_get_display, + .get_window = gsr_window_wayland_get_window, + .for_each_active_monitor_output_cached = gsr_window_wayland_for_each_active_monitor_output_cached, + .priv = window_wayland + }; + + return window; +} diff --git a/src/window/window.c b/src/window/window.c new file mode 100644 index 0000000..1c6a24e --- /dev/null +++ b/src/window/window.c @@ -0,0 +1,30 @@ +#include "../../include/window/window.h" +#include <stddef.h> + +void gsr_window_destroy(gsr_window *self); + +bool gsr_window_process_event(gsr_window *self) { + return self->process_event(self); +} + +XEvent* gsr_window_get_event_data(gsr_window *self) { + if(self->get_event_data) + return self->get_event_data(self); + return NULL; +} + +gsr_display_server gsr_window_get_display_server(const gsr_window *self) { + return self->get_display_server(); +} + +void* gsr_window_get_display(gsr_window *self) { + return self->get_display(self); +} + +void* gsr_window_get_window(gsr_window *self) { + return self->get_window(self); +} + +void gsr_window_for_each_active_monitor_output_cached(const gsr_window *self, active_monitor_callback callback, void *userdata) { + self->for_each_active_monitor_output_cached(self, callback, userdata); +} diff --git a/src/window/x11.c b/src/window/x11.c new file mode 100644 index 0000000..964422d --- /dev/null +++ b/src/window/x11.c @@ -0,0 +1,162 @@ +#include "../../include/window/x11.h" + +#include "../../include/vec2.h" +#include "../../include/defs.h" +#include "../../include/utils.h" + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdint.h> +#include <X11/Xlib.h> + +#define GSR_MAX_OUTPUTS 32 + +typedef struct { + char *name; + vec2i pos; + vec2i size; + uint32_t connector_id; + gsr_monitor_rotation rotation; + uint32_t monitor_identifier; /* crtc id */ +} gsr_x11_output; + +typedef struct { + Display *display; + Window window; + gsr_x11_output outputs[GSR_MAX_OUTPUTS]; + int num_outputs; + XEvent xev; +} gsr_window_x11; + +static void store_x11_monitor(const gsr_monitor *monitor, void *userdata) { + gsr_window_x11 *window_x11 = userdata; + if(window_x11->num_outputs == GSR_MAX_OUTPUTS) { + fprintf(stderr, "gsr warning: reached maximum outputs (%d), ignoring output %s\n", GSR_MAX_OUTPUTS, monitor->name); + return; + } + + char *monitor_name = strdup(monitor->name); + if(!monitor_name) + return; + + const int index = window_x11->num_outputs; + window_x11->outputs[index].name = monitor_name; + window_x11->outputs[index].pos = monitor->pos; + window_x11->outputs[index].size = monitor->size; + window_x11->outputs[index].connector_id = monitor->connector_id; + window_x11->outputs[index].rotation = monitor->rotation; + window_x11->outputs[index].monitor_identifier = monitor->monitor_identifier; + ++window_x11->num_outputs; +} + +static void gsr_window_x11_deinit(gsr_window_x11 *self) { + if(self->window) { + XDestroyWindow(self->display, self->window); + self->window = None; + } + + for(int i = 0; i < self->num_outputs; ++i) { + if(self->outputs[i].name) { + free(self->outputs[i].name); + self->outputs[i].name = NULL; + } + } + self->num_outputs = 0; +} + +static bool gsr_window_x11_init(gsr_window_x11 *self) { + self->window = XCreateWindow(self->display, DefaultRootWindow(self->display), 0, 0, 16, 16, 0, CopyFromParent, InputOutput, CopyFromParent, 0, NULL); + if(!self->window) { + fprintf(stderr, "gsr error: gsr_window_x11_init failed: failed to create gl window\n"); + return false; + } + + self->num_outputs = 0; + for_each_active_monitor_output_x11_not_cached(self->display, store_x11_monitor, self); + return true; +} + +static void gsr_window_x11_destroy(gsr_window *window) { + gsr_window_x11 *self = window->priv; + gsr_window_x11_deinit(self); + free(self); + free(window); +} + +static bool gsr_window_x11_process_event(gsr_window *window) { + gsr_window_x11 *self = window->priv; + if(XPending(self->display)) { + XNextEvent(self->display, &self->xev); + return true; + } + return false; +} + +static XEvent* gsr_window_x11_get_event_data(gsr_window *window) { + gsr_window_x11 *self = window->priv; + return &self->xev; +} + +static gsr_display_server gsr_window_x11_get_display_server(void) { + return GSR_DISPLAY_SERVER_X11; +} + +static void* gsr_window_x11_get_display(gsr_window *window) { + gsr_window_x11 *self = window->priv; + return self->display; +} + +static void* gsr_window_x11_get_window(gsr_window *window) { + gsr_window_x11 *self = window->priv; + return (void*)self->window; +} + +static void gsr_window_x11_for_each_active_monitor_output_cached(const gsr_window *window, active_monitor_callback callback, void *userdata) { + const gsr_window_x11 *self = window->priv; + for(int i = 0; i < self->num_outputs; ++i) { + const gsr_x11_output *output = &self->outputs[i]; + const gsr_monitor monitor = { + .name = output->name, + .name_len = strlen(output->name), + .pos = output->pos, + .size = output->size, + .connector_id = output->connector_id, + .rotation = output->rotation, + .monitor_identifier = output->monitor_identifier + }; + callback(&monitor, userdata); + } +} + +gsr_window* gsr_window_x11_create(Display *display) { + gsr_window *window = calloc(1, sizeof(gsr_window)); + if(!window) + return window; + + gsr_window_x11 *window_x11 = calloc(1, sizeof(gsr_window_x11)); + if(!window_x11) { + free(window); + return NULL; + } + + window_x11->display = display; + if(!gsr_window_x11_init(window_x11)) { + free(window_x11); + free(window); + return NULL; + } + + *window = (gsr_window) { + .destroy = gsr_window_x11_destroy, + .process_event = gsr_window_x11_process_event, + .get_event_data = gsr_window_x11_get_event_data, + .get_display_server = gsr_window_x11_get_display_server, + .get_display = gsr_window_x11_get_display, + .get_window = gsr_window_x11_get_window, + .for_each_active_monitor_output_cached = gsr_window_x11_for_each_active_monitor_output_cached, + .priv = window_x11 + }; + + return window; +} diff --git a/src/window_texture.c b/src/window_texture.c index 7448323..4846bdc 100644 --- a/src/window_texture.c +++ b/src/window_texture.c @@ -16,6 +16,7 @@ int window_texture_init(WindowTexture *window_texture, Display *display, Window window_texture->display = display; window_texture->window = window; window_texture->pixmap = None; + window_texture->image = NULL; window_texture->texture_id = 0; window_texture->redirected = 0; window_texture->egl = egl; @@ -34,6 +35,11 @@ static void window_texture_cleanup(WindowTexture *self, int delete_texture) { self->texture_id = 0; } + if(self->image) { + self->egl->eglDestroyImage(self->egl->egl_display, self->image); + self->image = NULL; + } + if(self->pixmap) { XFreePixmap(self->display, self->pixmap); self->pixmap = None; @@ -79,8 +85,10 @@ int window_texture_on_resize(WindowTexture *self) { texture_id = self->texture_id; } - self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); - self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); + const float border_color[4] = { 0.0f, 0.0f, 0.0f, 0.0f }; + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER); + self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER); + self->egl->glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, border_color); self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); self->egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); @@ -101,14 +109,14 @@ int window_texture_on_resize(WindowTexture *self) { self->pixmap = pixmap; self->texture_id = texture_id; + self->image = image; cleanup: self->egl->glBindTexture(GL_TEXTURE_2D, 0); - if(image) - self->egl->eglDestroyImage(self->egl->egl_display, image); - if(result != 0) { + if(image) + self->egl->eglDestroyImage(self->egl->egl_display, image); if(texture_id != 0) self->egl->glDeleteTextures(1, &texture_id); if(pixmap) @@ -120,4 +128,4 @@ int window_texture_on_resize(WindowTexture *self) { unsigned int window_texture_get_opengl_texture_id(WindowTexture *self) { return self->texture_id; -}
\ No newline at end of file +} diff --git a/src/xnvctrl.c b/src/xnvctrl.c index b738455..af46493 100644 --- a/src/xnvctrl.c +++ b/src/xnvctrl.c @@ -15,7 +15,7 @@ bool gsr_xnvctrl_load(gsr_xnvctrl *self, Display *display) { return false; } - dlsym_assign required_dlsym[] = { + const dlsym_assign required_dlsym[] = { { (void**)&self->XNVCTRLQueryExtension, "XNVCTRLQueryExtension" }, { (void**)&self->XNVCTRLSetTargetAttributeAndGetStatus, "XNVCTRLSetTargetAttributeAndGetStatus" }, { (void**)&self->XNVCTRLQueryValidTargetAttributeValues, "XNVCTRLQueryValidTargetAttributeValues" }, |