Implement more instructions, implement function parameters and arguments

5 files changed, 397 insertions, 43 deletions

diff --git a/executor/executor.h b/executor/executor.h
index fdf6e67..5be9abc 100644
--- a/executor/executor.h
+++ b/executor/executor.h
@@ -20,6 +20,10 @@ CHECK_RESULT int amal_executor_init(amal_executor **self);
 void amal_executor_deinit(amal_executor *self);
 CHECK_RESULT int amal_executor_run(amal_executor *self);
 
+/* These functions are called for every file in the program. Every file has its own list of strings, intermediates, functions and external functions */
+CHECK_RESULT int amal_executor_instructions_start(amal_executor *self, u16 num_functions);
+CHECK_RESULT int amal_executor_instructions_end(amal_executor *self);
+
 CHECK_RESULT int amal_exec_nop(amal_executor *self);
 CHECK_RESULT int amal_exec_setz(amal_executor *self, u8 dst_reg);
 CHECK_RESULT int amal_exec_mov(amal_executor *self, u8 dst_reg, u8 src_reg);
@@ -34,12 +38,12 @@ CHECK_RESULT int amal_exec_div(amal_executor *self, u8 dst_reg, u8 src_reg1, u8
 CHECK_RESULT int amal_exec_push(amal_executor *self, u8 reg);
 CHECK_RESULT int amal_exec_pushi(amal_executor *self, i64 imm);
 CHECK_RESULT int amal_exec_pushd(amal_executor *self, BufferView data);
-/*CHECK_RESULT int amal_exec_call(u8 dst_reg, BufferView data);
-CHECK_RESULT int amal_exec_callr(u8 dst_reg, BufferView data);*/
+CHECK_RESULT int amal_exec_call(amal_executor *self, u16 func_index, u8 num_args);
+/*CHECK_RESULT int amal_exec_callr(u8 dst_reg, BufferView data);*/
 CHECK_RESULT int amal_exec_cmp(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2);
 CHECK_RESULT int amal_exec_jz(amal_executor *self, u8 dst_reg, i16 offset);
 CHECK_RESULT int amal_exec_jmp(amal_executor *self, i16 offset);
-CHECK_RESULT int amal_exec_ret(amal_executor *self);
+CHECK_RESULT int amal_exec_ret(amal_executor *self, u8 reg);
 CHECK_RESULT int amal_exec_func_start(amal_executor *self, u16 num_regs);
 CHECK_RESULT int amal_exec_func_end(amal_executor *self);
 
diff --git a/executor/interpreter/executor.c b/executor/interpreter/executor.c
new file mode 100644
index 0000000..0180b08
--- /dev/null
+++ b/executor/interpreter/executor.c
@@ -0,0 +1,197 @@
+#include "../executor.h"
+#include "../../include/program.h"
+#include "../../include/std/alloc.h"
+#include <assert.h>
+
+typedef struct {
+    usize *stack;
+    usize stack_size;
+    usize stack_index;
+    isize reg[AMAL_PROGRAM_NUM_REGISTERS];
+} amal_executor_impl;
+
+#define IMPL \
+    amal_executor_impl *impl; \
+    impl = (amal_executor_impl*)self;
+
+#define FOUR_MEGABYTES 1024*1024*4
+
+static int executor_ensure_stack_capacity(amal_executor_impl *self, usize bytes_to_add) {
+    const isize overflow = self->stack_size - self->stack_index * sizeof(isize) + bytes_to_add;
+    if(overflow > 0) {
+        void *new_data;
+        const usize new_stack_size = self->stack_size * 1.5;
+        if(new_stack_size > FOUR_MEGABYTES)
+            return AMAL_PROGRAM_INSTRUCTION_STACK_OVERFLOW;
+        return_if_error(am_realloc(self->stack, new_stack_size, &new_data));
+        self->stack = new_data;
+        self->stack_size = new_stack_size;
+    }
+    return 0;
+}
+
+int amal_executor_init(amal_executor **self) {
+    amal_executor_impl **impl;
+    impl = (amal_executor_impl**)self;
+    *impl = NULL;
+    return_if_error(am_malloc(sizeof(amal_executor_impl), (void**)impl));
+    (*impl)->stack_size = 4096;
+    return_if_error(am_malloc((*impl)->stack_size, (void**)&(*impl)->stack));
+    (*impl)->stack_index = 0;
+    return 0;
+}
+
+void amal_executor_deinit(amal_executor *self) {
+    IMPL
+    am_free(impl->stack);
+    am_free(impl);
+}
+
+int amal_executor_run(amal_executor *self) {
+    (void)self;
+    assert(bool_false && "TODO: Implement!");
+    return 0;
+}
+
+int amal_exec_nop(amal_executor *self) {
+    (void)self;
+    return 0;
+}
+
+int amal_exec_setz(amal_executor *self, u8 dst_reg) {
+    IMPL
+    impl->reg[dst_reg] = 0;
+    return 0;
+}
+
+int amal_exec_mov(amal_executor *self, u8 dst_reg, u8 src_reg) {
+    IMPL
+    impl->reg[dst_reg] = impl->reg[src_reg];
+    return 0;
+}
+
+int amal_exec_movi(amal_executor *self, u8 dst_reg, i64 imm) {
+    IMPL
+    impl->reg[dst_reg] = imm;
+    return 0;
+}
+
+int amal_exec_movd(amal_executor *self, u8 dst_reg, BufferView data) {
+    IMPL
+    impl->reg[dst_reg] = (uintptr_t)data.data;
+    return 0;
+}
+
+int amal_exec_add(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
+    IMPL
+    impl->reg[dst_reg] = impl->reg[src_reg1] + impl->reg[src_reg2];
+    return 0;
+}
+
+int amal_exec_sub(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
+    IMPL
+    impl->reg[dst_reg] = impl->reg[src_reg1] - impl->reg[src_reg2];
+    return 0;
+}
+
+int amal_exec_imul(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
+    IMPL
+    impl->reg[dst_reg] = impl->reg[src_reg1] * impl->reg[src_reg2];
+    return 0;
+}
+
+int amal_exec_mul(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
+    IMPL
+    impl->reg[dst_reg] = impl->reg[src_reg1] * impl->reg[src_reg2];
+    return 0;
+}
+
+int amal_exec_idiv(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
+    IMPL
+    impl->reg[dst_reg] = impl->reg[src_reg1] / impl->reg[src_reg2];
+    return 0;
+}
+
+int amal_exec_div(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
+    IMPL
+    impl->reg[dst_reg] = impl->reg[src_reg1] / impl->reg[src_reg2];
+    return 0;
+}
+
+int amal_exec_push(amal_executor *self, u8 reg) {
+    IMPL
+    return_if_error(executor_ensure_stack_capacity(impl, sizeof(isize)));
+    impl->stack[impl->stack_index++] = impl->reg[reg];
+    return 0;
+}
+
+int amal_exec_pushi(amal_executor *self, i64 imm) {
+    IMPL
+    return_if_error(executor_ensure_stack_capacity(impl, sizeof(isize)));
+    impl->stack[impl->stack_index++] = imm;
+    return 0;
+}
+
+int amal_exec_pushd(amal_executor *self, BufferView data) {
+    IMPL
+    return_if_error(executor_ensure_stack_capacity(impl, sizeof(isize)));
+    impl->stack[impl->stack_index++] = (uintptr_t)data.data;
+    return 0;
+}
+
+/*int amal_exec_call(u8 dst_reg, BufferView data);
+int amal_exec_callr(u8 dst_reg, BufferView data);*/
+int amal_exec_cmp(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
+    IMPL
+    impl->reg[dst_reg] = (impl->reg[src_reg1] == impl->reg[src_reg2]);
+    return 0;
+}
+
+int amal_exec_jz(amal_executor *self, u8 dst_reg, i16 offset) {
+    (void)self;
+    (void)dst_reg;
+    (void)offset;
+    /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement!");
+    return 0;
+}
+
+int amal_exec_jmp(amal_executor *self, i16 offset) {
+    (void)self;
+    (void)offset;
+    /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement!");
+    return 0;
+}
+
+int amal_exec_ret(amal_executor *self) {
+    (void)self;
+    /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement RET. RET needs to restore the stack before returning");
+    return 0;
+}
+
+int amal_exec_func_start(amal_executor *self, u16 num_regs) {
+    /*
+        TODO: Validate stack size, or maybe remove all validation? do we really need validation?
+        If we need security, we could fork the process instead.
+    */
+
+    /*
+        Some registers need to be preserved before entering a function scope and these registers are different on different platforms.
+        32-bit: EBX, ESI, EDI, EBP
+        64-bit Windows: RBX, RSI, RDI, RBP, R12-R15, XMM6-XMM15
+        64-bit Linux,BSD,Mac: RBX, RBP, R12-R15
+    */
+    /* TODO: Preserve registers and stack frame */
+    /*return executor_ensure_stack_capacity(impl, num_regs * sizeof(isize));*/
+    (void)self;
+    (void)num_regs;
+    return 0;
+}
+
+int amal_exec_func_end(amal_executor *self) {
+    (void)self;
+    /* TODO: Restore registers and stack frame and ret */
+    return 0;
+}
diff --git a/executor/x86_64/asm.c b/executor/x86_64/asm.c
index 73ae568..f032538 100644
--- a/executor/x86_64/asm.c
+++ b/executor/x86_64/asm.c
@@ -35,6 +35,10 @@ static void ins_start(Asm *self) {
     asm_debug_str_buffer_index = 0;
 }
 
+static int max(int a, int b) {
+    return a >= b ? a : b;
+}
+
 static void ins_end(Asm *self, const char *fmt, ...) {
     usize ins_end_offset;
     usize i;
@@ -46,7 +50,7 @@ static void ins_end(Asm *self, const char *fmt, ...) {
         fprintf(stderr, "%02x ", ((u8*)self->code)[i]);
     }
     /* Same padding for all instructions, no matter how long they are */
-    for(i = 0; i < 35 - (ins_end_offset - ins_start_offset)*3; ++i) {
+    for(i = 0; i < (usize)max(0, 35 - (ins_end_offset - ins_start_offset)*3); ++i) {
         putc(' ', stderr);
     }
     vfprintf(stderr, fmt, args);
@@ -160,9 +164,9 @@ void asm_ptr_init_index_disp(AsmPtr *self, Reg64 base, Reg64 index, i32 disp) {
 }
 
 int asm_init(Asm *self) {
-    self->size = am_pagesize();
-    amal_log_debug("asm: page size: %u", self->size);
-    self->code = mmap(NULL, self->size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+    self->allocated_size = am_pagesize();
+    amal_log_debug("asm: page size: %u", self->allocated_size);
+    self->code = mmap(NULL, self->allocated_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
     if(self->code == MAP_FAILED)
         return -errno;
     self->code_it = self->code;
@@ -171,12 +175,17 @@ int asm_init(Asm *self) {
 
 void asm_deinit(Asm *self) {
     if(self->code)
-        munmap(self->code, self->size);
+        munmap(self->code, self->allocated_size);
     self->code = NULL;
     self->code_it = NULL;
-    self->size = 0;
+    self->allocated_size = 0;
+}
+
+usize asm_get_size(Asm *self) {
+    return self->code_it - (u8*)self->code;
 }
 
+#if 0
 static void asm_print_code_hex(Asm *self) {
     u8 *ptr;
     int off;
@@ -196,13 +205,14 @@ static void asm_print_code_hex(Asm *self) {
     if(off != 0)
         putc('\n', stdout);
 }
+#endif
 
 int asm_execute(Asm *self) {
     void (*func)();
-    if(mprotect(self->code, self->size, PROT_READ | PROT_EXEC) != 0)
+    if(mprotect(self->code, self->allocated_size, PROT_READ | PROT_EXEC) != 0)
         return -errno;
 
-    asm_print_code_hex(self);
+    /*asm_print_code_hex(self);*/
 
     /* TODO: Verify if this is valid on all platforms. According to ISO C standard it isn't? */
     *(void**)(&func) = self->code;
@@ -214,17 +224,17 @@ int asm_execute(Asm *self) {
 static CHECK_RESULT int asm_ensure_capacity(Asm *self, usize size) {
     usize current_offset;
     current_offset = (u8*)self->code_it - (u8*)self->code;
-    if(current_offset + size > self->size) {
+    if(current_offset + size > self->allocated_size) {
         void *new_mem;
         usize new_size;
-        new_size = self->size + am_pagesize();
+        new_size = self->allocated_size + am_pagesize();
         new_mem = mmap(NULL, new_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
         if(self->code == MAP_FAILED)
             return -errno;
 
-        am_memcpy(new_mem, self->code, self->size);
+        am_memcpy(new_mem, self->code, self->allocated_size);
         self->code = new_mem;
-        self->size = new_size;
+        self->allocated_size = new_size;
         self->code_it = (u8*)self->code + current_offset;
     }
     return 0;
@@ -232,7 +242,7 @@ static CHECK_RESULT int asm_ensure_capacity(Asm *self, usize size) {
 
 #ifdef DEBUG
 static isize asm_get_capacity_left(Asm *self) {
-    return (isize)self->size - (isize)((u8*)self->code_it - (u8*)self->code);
+    return (isize)self->allocated_size - (isize)((u8*)self->code_it - (u8*)self->code);
 }
 #endif
 
@@ -346,21 +356,20 @@ int asm_mov_rm(Asm *self, Reg64 dst, AsmPtr *src) {
     return 0;
 }
 
+/* Note: This shows as instruction movabs in intel assembly format */
 int asm_mov_ri(Asm *self, Reg64 dst, i64 immediate) {
     ins_start(self);
-    /* 10 bytes is the maximum size of the instruction. We don't how how large it will be so we prepare for the largest size */
     return_if_error(asm_ensure_capacity(self, 10));
     *self->code_it++ = REX_W;
     *self->code_it++ = 0xB8 + dst;
     am_memcpy(self->code_it, &immediate, sizeof(immediate));
     self->code_it += sizeof(immediate);
-    ins_end(self, "mov %s, %ld", reg64_to_str(dst), immediate);
+    ins_end(self, "mov %s, 0x%x", reg64_to_str(dst), immediate);
     return 0;
 }
 
 int asm_mov_rr(Asm *self, Reg64 dst, Reg64 src) {
     ins_start(self);
-    /* 3 bytes is the maximum size of the instruction. We don't how how large it will be so we prepare for the largest size */
     return_if_error(asm_ensure_capacity(self, 3));
     *self->code_it++ = REX_W;
     *self->code_it++ = 0x89;
@@ -371,7 +380,6 @@ int asm_mov_rr(Asm *self, Reg64 dst, Reg64 src) {
 
 int asm_add_rr(Asm *self, Reg64 dst, Reg64 src) {
     ins_start(self);
-    /* 3 bytes is the maximum size of the instruction. We don't how how large it will be so we prepare for the largest size */
     return_if_error(asm_ensure_capacity(self, 3));
     *self->code_it++ = REX_W;
     *self->code_it++ = 0x01;
@@ -382,7 +390,6 @@ int asm_add_rr(Asm *self, Reg64 dst, Reg64 src) {
 
 int asm_sub_rr(Asm *self, Reg64 dst, Reg64 src) {
     ins_start(self);
-    /* 3 bytes is the maximum size of the instruction. We don't how how large it will be so we prepare for the largest size */
     return_if_error(asm_ensure_capacity(self, 3));
     *self->code_it++ = REX_W;
     *self->code_it++ = 0x29;
@@ -393,7 +400,6 @@ int asm_sub_rr(Asm *self, Reg64 dst, Reg64 src) {
 
 int asm_imul_rr(Asm *self, Reg64 dst, Reg64 src) {
     ins_start(self);
-    /* 3 bytes is the maximum size of the instruction. We don't how how large it will be so we prepare for the largest size */
     return_if_error(asm_ensure_capacity(self, 4));
     *self->code_it++ = REX_W;
     *self->code_it++ = 0x0F;
@@ -403,6 +409,25 @@ int asm_imul_rr(Asm *self, Reg64 dst, Reg64 src) {
     return 0;
 }
 
+int asm_cqo(Asm *self) {
+    ins_start(self);
+    return_if_error(asm_ensure_capacity(self, 2));
+    *self->code_it++ = REX_W;
+    *self->code_it++ = 0x99;
+    ins_end(self, "cqo");
+    return 0;
+}
+
+int asm_idiv_rr(Asm *self, Reg64 src) {
+    ins_start(self);
+    return_if_error(asm_ensure_capacity(self, 4));
+    *self->code_it++ = REX_W;
+    *self->code_it++ = 0xF7;
+    asm_rr(self, src, 0x7);
+    ins_end(self, "idiv %s", reg64_to_str(src));
+    return 0;
+}
+
 int asm_pushr(Asm *self, Reg64 reg) {
     ins_start(self);
     return_if_error(asm_ensure_capacity(self, 1));
@@ -419,6 +444,26 @@ int asm_popr(Asm *self, Reg64 reg) {
     return 0;
 }
 
+/*
+    Note: This is sometimes called with @relative 0 (will print call -5), in which case it's most likely a dummy call until the relative position
+    is later changed with @asm_override_call_rel32. TODO: Update the ins_end debug print to take that into account somehow
+*/
+int asm_call_rel32(Asm *self, i32 relative) {
+    ins_start(self);
+    relative -= 5; /* In x86, the relative position starts from the next instruction */
+    return_if_error(asm_ensure_capacity(self, 5));
+    *self->code_it++ = 0xE8;
+    am_memcpy(self->code_it, &relative, sizeof(relative));
+    self->code_it += sizeof(relative);
+    ins_end(self, "call 0x%x", relative);
+    return 0;
+}
+
+void asm_override_call_rel32(Asm *self, u32 asm_index, i32 new_relative) {
+    new_relative -= 5; /* In x86, the relative position starts from the next instruction */
+    am_memcpy((u8*)self->code + asm_index + 1, &new_relative, sizeof(new_relative));
+}
+
 /* TODO: Remove these !*/
 
 /* /r */
@@ -559,11 +604,12 @@ int asm_ret(Asm *self, u16 bytes) {
     if(bytes == 0) {
         return_if_error(asm_ensure_capacity(self, 1));
         *self->code_it++ = 0xC3;
+        ins_end(self, "ret");
     } else {
         return_if_error(asm_ensure_capacity(self, 3));
         *self->code_it++ = 0xC2;
         am_memcpy(self->code_it, &bytes, sizeof(bytes));
+        ins_end(self, "ret 0x%x", bytes);
     }
-    ins_end(self, "ret 0x%x", bytes);
     return 0;
 }
diff --git a/executor/x86_64/asm.h b/executor/x86_64/asm.h
index b374e44..7e5ac67 100644
--- a/executor/x86_64/asm.h
+++ b/executor/x86_64/asm.h
@@ -7,7 +7,7 @@
 typedef struct {
     void *code;
     u8 *code_it;
-    usize size;
+    usize allocated_size;
 } Asm;
 
 typedef enum {
@@ -47,6 +47,8 @@ void asm_ptr_init_index_disp(AsmPtr *self, Reg64 base, Reg64 index, i32 disp);
 CHECK_RESULT int asm_init(Asm *self);
 void asm_deinit(Asm *self);
 
+usize asm_get_size(Asm *self);
+
 CHECK_RESULT int asm_execute(Asm *self);
 
 CHECK_RESULT int asm_nop(Asm *self);
@@ -68,9 +70,23 @@ CHECK_RESULT int asm_mov_rr(Asm *self, Reg64 dst, Reg64 src);
 CHECK_RESULT int asm_add_rr(Asm *self, Reg64 dst, Reg64 src);
 CHECK_RESULT int asm_sub_rr(Asm *self, Reg64 dst, Reg64 src);
 CHECK_RESULT int asm_imul_rr(Asm *self, Reg64 dst, Reg64 src);
+/* Sign extend RAX into RDX, this is needed for some operations, such as idiv */
+CHECK_RESULT int asm_cqo(Asm *self);
+/*
+    Divide RDX:RAX by @src. Store the quotient in RAX and the remainder in RDX.
+    @asm_cqo should be called before this, since RAX needs to be sign extended into RDX
+*/
+CHECK_RESULT int asm_idiv_rr(Asm *self, Reg64 src);
 
 CHECK_RESULT int asm_pushr(Asm *self, Reg64 reg);
 CHECK_RESULT int asm_popr(Asm *self, Reg64 reg);
+/*
+    In x86 assembly, the @relative position starts from the next instruction.
+    This offset shouldn't be calculated by the caller and is instead managed
+    by this asm library itself.
+*/
+CHECK_RESULT int asm_call_rel32(Asm *self, i32 relative);
+void asm_override_call_rel32(Asm *self, u32 asm_index, i32 new_relative);
 
 
 
diff --git a/executor/x86_64/executor.c b/executor/x86_64/executor.c
index b53ccea..b7aa91f 100644
--- a/executor/x86_64/executor.c
+++ b/executor/x86_64/executor.c
@@ -1,5 +1,6 @@
 #include "../executor.h"
 #include "../../include/std/alloc.h"
+#include "../../include/std/buffer.h"
 #include "asm.h"
 #include <assert.h>
 
@@ -11,7 +12,16 @@
 */
 
 typedef struct {
+    u32 asm_index;
+    u16 func_index;
+} CallDefer;
+
+typedef struct {
     Asm asm;
+    usize *function_indices;
+    u16 num_functions;
+    u16 func_counter;
+    Buffer/*CallDefer*/ call_defer;
 } amal_executor_impl;
 
 #define IMPL \
@@ -26,13 +36,20 @@ static i64 abs_i64(i64 value) {
 
 int amal_executor_init(amal_executor **self) {
     amal_executor_impl **impl;
-    return_if_error(am_malloc(sizeof(amal_executor_impl), (void**)self));
     impl = (amal_executor_impl**)self;
+    *impl = NULL;
+    return_if_error(am_malloc(sizeof(amal_executor_impl), (void**)impl));
+    (*impl)->function_indices = NULL;
+    (*impl)->num_functions = 0;
+    (*impl)->func_counter = 0;
+    ignore_result_int(buffer_init(&(*impl)->call_defer, NULL));
     return asm_init(&(*impl)->asm);
 }
 
 void amal_executor_deinit(amal_executor *self) {
     IMPL
+    buffer_deinit(&impl->call_defer);
+    am_free(impl->function_indices);
     asm_deinit(&impl->asm);
     am_free(impl);
 }
@@ -42,6 +59,30 @@ int amal_executor_run(amal_executor *self) {
     return asm_execute(&impl->asm);
 }
 
+int amal_executor_instructions_start(amal_executor *self, u16 num_functions) {
+    void *new_data;
+    IMPL
+    return_if_error(am_realloc(impl->function_indices, num_functions * sizeof(*impl->function_indices), &new_data));
+    impl->function_indices = new_data;
+    impl->num_functions = num_functions;
+    impl->func_counter = 0;
+    buffer_clear(&impl->call_defer);
+    return 0;
+}
+
+int amal_executor_instructions_end(amal_executor *self) {
+    CallDefer *call_defer, *call_defer_end;
+    IMPL
+
+    call_defer = buffer_begin(&impl->call_defer);
+    call_defer_end = buffer_end(&impl->call_defer);
+    for(; call_defer != call_defer_end; ++call_defer) {
+        const isize func_offset = (isize)impl->function_indices[call_defer->func_index] - (isize)call_defer->asm_index;
+        asm_override_call_rel32(&impl->asm, call_defer->asm_index, func_offset);
+    }
+    return 0;
+}
+
 int amal_exec_nop(amal_executor *self) {
     IMPL
     return asm_nop(&impl->asm);
@@ -158,16 +199,25 @@ int amal_exec_mul(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
     return_if_error(asm_mul_rr(&self->asm, RAX, RCX));
     return_if_error(asm_mov_mr(&self->asm, &dst, RAX));
     #endif
+    assert(bool_false && "TODO: Implement!");
     return 0;
 }
 
 int amal_exec_idiv(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
-    (void)self;
-    (void)dst_reg;
-    (void)src_reg1;
-    (void)src_reg2;
-    /* TODO: Implement! */
-    return 0;
+    AsmPtr dst;
+    AsmPtr reg1;
+    AsmPtr reg2;
+    IMPL
+
+    asm_ptr_init_disp(&dst, RBP, get_register_stack_offset(dst_reg));
+    asm_ptr_init_disp(&reg1, RBP, get_register_stack_offset(src_reg1));
+    asm_ptr_init_disp(&reg2, RBP, get_register_stack_offset(src_reg2));
+
+    return_if_error(asm_mov_rm(&impl->asm, RAX, &reg1));
+    return_if_error(asm_mov_rm(&impl->asm, RCX, &reg2));
+    return_if_error(asm_cqo(&impl->asm));
+    return_if_error(asm_idiv_rr(&impl->asm, RCX));
+    return asm_mov_mr(&impl->asm, &dst, RAX);
 }
 
 int amal_exec_div(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
@@ -176,20 +226,24 @@ int amal_exec_div(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
     (void)src_reg1;
     (void)src_reg2;
     /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement!");
     return 0;
 }
 
 int amal_exec_push(amal_executor *self, u8 reg) {
-    (void)self;
-    (void)reg;
-    /* TODO: Implement! */
-    return 0;
+    AsmPtr reg_ptr;
+    IMPL
+
+    asm_ptr_init_disp(&reg_ptr, RBP, get_register_stack_offset(reg));
+    return_if_error(asm_mov_rm(&impl->asm, RAX, &reg_ptr));
+    return asm_pushr(&impl->asm, RAX);
 }
 
 int amal_exec_pushi(amal_executor *self, i64 imm) {
     (void)self;
     (void)imm;
     /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement!");
     return 0;
 }
 
@@ -197,17 +251,48 @@ int amal_exec_pushd(amal_executor *self, BufferView data) {
     (void)self;
     (void)data;
     /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement!");
+    return 0;
+}
+
+int amal_exec_call(amal_executor *self, u16 func_index, u8 num_args) {
+    isize asm_size;
+    IMPL
+    /* TODO: Preserve necessary registers before call? */
+    /* TODO: This assumes all arguments are isize */
+    asm_size = asm_get_size(&impl->asm);
+    if(func_index < impl->func_counter) {
+        return_if_error(asm_call_rel32(&impl->asm, (isize)impl->function_indices[func_index] - asm_size));
+    } else {
+        /*
+            The location of the function has not been defined yet. Use call instruction with dummy data and change
+            the location once the location to the function is known
+        */
+        CallDefer call_defer;
+        call_defer.asm_index = asm_size;
+        call_defer.func_index = func_index;
+        return_if_error(buffer_append(&impl->call_defer, &call_defer, sizeof(call_defer)));
+        return_if_error(asm_call_rel32(&impl->asm, 0));
+    }
+
+    if(num_args > 0)
+        return asm_add_rm64_imm(&impl->asm, RSP, num_args * sizeof(isize));
     return 0;
 }
 
-/*int amal_exec_call(u8 dst_reg, BufferView data);
-int amal_exec_callr(u8 dst_reg, BufferView data);*/
+/*
+int amal_exec_callr(u8 dst_reg, BufferView data) {
+
+}
+*/
+
 int amal_exec_cmp(amal_executor *self, u8 dst_reg, u8 src_reg1, u8 src_reg2) {
     (void)self;
     (void)dst_reg;
     (void)src_reg1;
     (void)src_reg2;
     /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement!");
     return 0;
 }
 
@@ -216,6 +301,7 @@ int amal_exec_jz(amal_executor *self, u8 dst_reg, i16 offset) {
     (void)dst_reg;
     (void)offset;
     /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement!");
     return 0;
 }
 
@@ -223,14 +309,18 @@ int amal_exec_jmp(amal_executor *self, i16 offset) {
     (void)self;
     (void)offset;
     /* TODO: Implement! */
+    assert(bool_false && "TODO: Implement!");
     return 0;
 }
 
-int amal_exec_ret(amal_executor *self) {
-    (void)self;
-    /* TODO: Implement! */
-    assert(bool_false && "TODO: Implement RET. RET needs to restore the stack before returning");
-    return 0;
+int amal_exec_ret(amal_executor *self, u8 reg) {
+    AsmPtr ret_reg;
+    IMPL
+
+    asm_ptr_init_disp(&ret_reg, RBP, get_register_stack_offset(reg));
+    /* Result is returned in RAX register. TODO: Make this work for larger data */
+    return_if_error(asm_mov_rm(&impl->asm, RAX, &ret_reg));
+    return amal_exec_func_end(self);
 }
 
 int amal_exec_func_start(amal_executor *self, u16 num_regs) {
@@ -246,10 +336,11 @@ int amal_exec_func_start(amal_executor *self, u16 num_regs) {
         64-bit Linux,BSD,Mac: RBX, RBP, R12-R15
     */
     IMPL
+    impl->function_indices[impl->func_counter++] = asm_get_size(&impl->asm);
     return_if_error(asm_pushr(&impl->asm, RBX));
     return_if_error(asm_pushr(&impl->asm, RBP));
     return_if_error(asm_mov_rr(&impl->asm, RBP, RSP));
-    return asm_sub_rm64_imm(&impl->asm, RSP, num_regs * sizeof(usize));
+    return asm_sub_rm64_imm(&impl->asm, RSP, num_regs * sizeof(isize));
 }
 
 int amal_exec_func_end(amal_executor *self) {