1 files changed, 59 insertions, 44 deletions

diff --git a/src/encoder/video/vaapi.c b/src/encoder/video/vaapi.c
index 2df140d..0daf4d8 100644
--- a/src/encoder/video/vaapi.c
+++ b/src/encoder/video/vaapi.c
@@ -4,17 +4,20 @@
 
 #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;
@@ -26,43 +29,40 @@ static bool gsr_video_encoder_vaapi_setup_context(gsr_video_encoder_vaapi *self,
         return false;
     }
 
-    AVBufferRef *device_ctx;
-    if(av_hwdevice_ctx_create(&device_ctx, AV_HWDEVICE_TYPE_VAAPI, render_path, NULL, 0) < 0) {
+    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(device_ctx);
+    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(&device_ctx);
+        av_buffer_unref(&self->device_ctx);
         return false;
     }
 
-    AVHWFramesContext *hw_frame_context =
-        (AVHWFramesContext *)frame_context->data;
+    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.hdr ? AV_PIX_FMT_P010LE : AV_PIX_FMT_NV12;
+    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_ref = device_ctx;
-    hw_frame_context->device_ctx = (AVHWDeviceContext*)device_ctx->data;
+    hw_frame_context->device_ctx = (AVHWDeviceContext*)self->device_ctx->data;
 
     //hw_frame_context->initial_pool_size = 20;
 
-    AVVAAPIDeviceContext *vactx =((AVHWDeviceContext*)device_ctx->data)->hwctx;
+    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(&device_ctx);
+        av_buffer_unref(&self->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);
+    av_buffer_unref(&frame_context);
     return true;
 }
 
@@ -96,20 +96,22 @@ static bool gsr_video_encoder_vaapi_setup_textures(gsr_video_encoder_vaapi *self
         self->params.egl->glGenTextures(2, self->target_textures);
         for(int i = 0; i < 2; ++i) {
             const int layer = i;
-            const int plane = 0;
-
-            const uint64_t modifier = self->prime.objects[self->prime.layers[layer].object_index[plane]].drm_format_modifier;
-            const intptr_t img_attr[] = {
-                EGL_LINUX_DRM_FOURCC_EXT,       formats[i],
-                EGL_WIDTH,                      self->prime.width / div[i],
-                EGL_HEIGHT,                     self->prime.height / div[i],
-                EGL_DMA_BUF_PLANE0_FD_EXT,      self->prime.objects[self->prime.layers[layer].object_index[plane]].fd,
-                EGL_DMA_BUF_PLANE0_OFFSET_EXT,  self->prime.layers[layer].offset[plane],
-                EGL_DMA_BUF_PLANE0_PITCH_EXT,   self->prime.layers[layer].pitch[plane],
-                EGL_DMA_BUF_PLANE0_MODIFIER_LO_EXT, modifier & 0xFFFFFFFFULL,
-                EGL_DMA_BUF_PLANE0_MODIFIER_HI_EXT, modifier >> 32ULL,
-                EGL_NONE
-            };
+
+            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);
@@ -119,10 +121,8 @@ static bool gsr_video_encoder_vaapi_setup_textures(gsr_video_encoder_vaapi *self
             }
 
             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_EDGE);
-            self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
-            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->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){}
@@ -149,13 +149,13 @@ static bool gsr_video_encoder_vaapi_setup_textures(gsr_video_encoder_vaapi *self
 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 *encoder_vaapi = encoder->priv;
+    gsr_video_encoder_vaapi *self = encoder->priv;
 
-    if(encoder_vaapi->params.egl->gpu_info.vendor == GSR_GPU_VENDOR_AMD && video_codec_context->codec_id == AV_CODEC_ID_HEVC) {
+    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(encoder_vaapi->params.egl->gpu_info.vendor == GSR_GPU_VENDOR_AMD && video_codec_context->codec_id == AV_CODEC_ID_AV1) {
+    } 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).
@@ -165,15 +165,31 @@ static bool gsr_video_encoder_vaapi_start(gsr_video_encoder *encoder, AVCodecCon
         } 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(!gsr_video_encoder_vaapi_setup_context(encoder_vaapi, video_codec_context)) {
-        gsr_video_encoder_vaapi_stop(encoder_vaapi, video_codec_context);
+    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(encoder_vaapi, video_codec_context, frame)) {
-        gsr_video_encoder_vaapi_stop(encoder_vaapi, video_codec_context);
+    if(!gsr_video_encoder_vaapi_setup_textures(self, video_codec_context, frame)) {
+        gsr_video_encoder_vaapi_stop(self, video_codec_context);
         return false;
     }
 
@@ -185,10 +201,10 @@ void gsr_video_encoder_vaapi_stop(gsr_video_encoder_vaapi *self, AVCodecContext
     self->target_textures[0] = 0;
     self->target_textures[1] = 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);
+    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) {
@@ -199,11 +215,11 @@ void gsr_video_encoder_vaapi_stop(gsr_video_encoder_vaapi *self, AVCodecContext
 }
 
 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 *encoder_vaapi = encoder->priv;
-    textures[0] = encoder_vaapi->target_textures[0];
-    textures[1] = encoder_vaapi->target_textures[1];
+    gsr_video_encoder_vaapi *self = encoder->priv;
+    textures[0] = self->target_textures[0];
+    textures[1] = self->target_textures[1];
     *num_textures = 2;
-    *destination_color = encoder_vaapi->params.hdr ? GSR_DESTINATION_COLOR_P010 : GSR_DESTINATION_COLOR_NV12;
+    *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) {
@@ -227,7 +243,6 @@ gsr_video_encoder* gsr_video_encoder_vaapi_create(const gsr_video_encoder_vaapi_
 
     *encoder = (gsr_video_encoder) {
         .start = gsr_video_encoder_vaapi_start,
-        .copy_textures_to_frame = NULL,
         .get_textures = gsr_video_encoder_vaapi_get_textures,
         .destroy = gsr_video_encoder_vaapi_destroy,
         .priv = encoder_vaapi