switched threading to SDL; works on windows now

8 files changed, 90 insertions, 157 deletions

diff --git a/bitmap_calls.cpp b/bitmap_calls.cpp
index 2cdb463..132aec5 100644
--- a/bitmap_calls.cpp
+++ b/bitmap_calls.cpp
@@ -177,10 +177,7 @@ BitmapPackRGB(pixel_buffer *Buffer) {
 
 internal void
 OutputToViewport(pixel_buffer *CompBuffer, project_state *State, GLuint textureID) {
-    if (D)
-        Convert4x4Chunk(CompBuffer, 1);
-    else
-        CopyToBuffer(CompBuffer, 0);
+    Convert4x4Chunk(CompBuffer, 1);
     EndRenderState(State);
     glBindTexture(GL_TEXTURE_2D, textureID);
     glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, CompBuffer->Width, CompBuffer->Height, GL_RGBA, GL_UNSIGNED_BYTE,
diff --git a/build.sh b/build.sh
index 8b5a63e..35ac264 100755
--- a/build.sh
+++ b/build.sh
@@ -8,7 +8,7 @@ WARNING_FLAGS="
     -Wno-missing-field-initializers -Wno-sign-compare -Wno-write-strings -Wno-unused-but-set-parameter \
     -Wno-missing-braces -Wno-format-security
     -fno-exceptions -Wno-strict-aliasing \
-    -DDEBUG=1 -DARM=0 -DTHREADED=0 \
+    -DDEBUG=1 -DARM=0 -DTHREADED=1 \
 "
 
 if [[ "$WINDOWS" == 1 ]]; then
diff --git a/createcalls.cpp b/createcalls.cpp
index 0dbf75c..1c20b1a 100644
--- a/createcalls.cpp
+++ b/createcalls.cpp
@@ -311,6 +311,18 @@ CreateDebugLayer(project_data *File, memory *Memory, uint16 Width, uint16 Height
 }
 
 internal void
+LoadTestFootage(project_data *File, project_state *State, memory *Memory)
+{
+    CreateLayerFromSource(File, State, Memory, "../asset/24.mp4");
+    // project_layer *Layer1 = CreateDebugLayer(&File, &Memory, 9, 14);
+    // project_layer *Layer1 = CreateSolidLayer(&File, &Memory, 9, 13, V4(1.0, 1.0, 1.0, 1.0));
+    // Layer1->x.CurrentValue.f = 7;
+    // Layer1->y.CurrentValue.f = 4;
+    // Layer1->StartFrame = 0;
+    // Layer1->EndFrame = File.EndFrame;
+}
+
+internal void
 CreateDemoScene(project_data *File, memory *Memory)
 {
     project_layer *Layer1 = CreateSolidLayer(File, Memory, 720, 1280, V4(0.0, 0.0, 0.0, 1.0));
@@ -339,3 +351,22 @@ CreateDemoScene(project_data *File, memory *Memory)
     Layer3->x.IsToggled = true;
     Layer3->y.IsToggled = true;
 }
+
+internal void
+CreateGrid(project_data *File, memory *Memory) {
+    uint16 Amount = 8;
+    real32 XInc = File->Width / Amount;
+    real32 YInc = File->Height / Amount;
+    for (int16 j = 0; j < 8; j++) {
+        for (int16 i = 0; i < 8; i++) {
+            project_layer *Layer = CreateSolidLayer(File, Memory, 200, 200, V4(0.6, 0.3, 0.4, 1.0));
+            Layer->x.CurrentValue.f = (XInc*i);
+            Layer->y.CurrentValue.f = (XInc*j);
+            Layer->opacity.CurrentValue.f = 0.25;
+            Layer->StartFrame = 0;
+            Layer->EndFrame = File->EndFrame;
+            ManualKeyframeInsertF(&Layer->rotation, Memory, i, 0);
+            ManualKeyframeInsertF(&Layer->rotation, Memory, 40+i, 360);
+        }
+    }
+}
diff --git a/main.cpp b/main.cpp
index 3a517be..4d513c4 100644
--- a/main.cpp
+++ b/main.cpp
@@ -5,15 +5,6 @@
 #include <sys/mman.h>
 #endif
 
-// TODO(fox): Switch all thread code to SDL!
-#if WINDOWS
-#else
-#if THREADED
-#include <pthread.h>
-#else
-#endif
-#endif
-
 #if ARM
 #include <arm_neon.h>
 #else
@@ -101,11 +92,10 @@ typedef double real64;
 #include "main.h"
 #include "debug.h"
 
-global_variable uint32 volatile CompletedJobs;
-global_variable uint32 volatile NextEntryToDo;
-global_variable uint32 volatile EntryCount;
+SDL_atomic_t CurrentEntry;
+SDL_atomic_t QueuedEntries;
+SDL_atomic_t CompletedEntries;
 global_variable bool32 IsRendering = false;
-global_variable bool32 D = true;
 global_variable instruction_mode InstructionMode = scalar_only;
 
 
@@ -314,11 +304,9 @@ int main(int argc, char *argv[]) {
         InstructionMode = avx_enabled;
     }
 
-    InstructionMode = scalar_only;
-
     project_data File = {};
-    File.Width = 1280;
-    File.Height = 720;
+    File.Width = 1283;
+    File.Height = 723;
     File.NumberOfFrames = 65;
     File.FPS = 30;
     File.CurrentFrame = 1;
@@ -337,14 +325,6 @@ int main(int argc, char *argv[]) {
 
     ui UI = {};
 
-    // File.NumberOfSources = 2;
-    // File.Source[0] = (char *)AllocateMemory(&Memory, STRING_SIZE, F_Strings);
-    // File.Source[1] = (char *)AllocateMemory(&Memory, STRING_SIZE, F_Strings);
-    // sprintf(File.Source[0], "../asset/b.jpg");
-    // sprintf(File.Source[1], "../asset/24.mp4");
-    // CreateLayerFromSource(&File, &State, &Memory, File.Source[0]);
-    // CreateLayerFromSource(&File, &State, &Memory, File.Source[1]);
-
     // shm_unlink("/testl");
     // int fd = shm_open("/testl", O_CREAT | O_EXCL | O_RDWR,
     //                   S_IRUSR | S_IWUSR);
@@ -367,68 +347,6 @@ int main(int argc, char *argv[]) {
     // if (sem_init(&shmp->sem2, 1, 0) == -1)
     //     Assert(0);
 
-    // CreateLayerFromSource(&File, &State, &Memory, "../asset/24.mp4");
-    // project_layer *Layer1 = CreateDebugLayer(&File, &Memory, 9, 14);
-    // project_layer *Layer1 = CreateSolidLayer(&File, &Memory, 9, 13, V4(1.0, 1.0, 1.0, 1.0));
-    // Layer1->x.CurrentValue.f = 7;
-    // Layer1->y.CurrentValue.f = 4;
-    // Layer1->StartFrame = 0;
-    // Layer1->EndFrame = File.EndFrame;
-
-    // CreateDebugLayer(&File, &Memory, 12, 8);
-    // File.Layer[0]->Name = "debug";
-    // File.Layer[0]->StartFrame = 0;
-    // File.Layer[0]->EndFrame = 65;
-
-    // CreateLayer(&File, &Memory);
-
-    // CreateRenderInfo(File.Layer[0], &Memory, File, image, "./asset/r.jpg");
-
-    // File.Layer[0]->Name = "Robot";
-    // File.Layer[0]->x.CurrentValue.f = 200;
-    // File.Layer[0]->y.CurrentValue.f = 250;
-    // File.Layer[0]->scale.CurrentValue.f = 1.1;
-    // File.Layer[0]->opacity.CurrentValue.f = 1.0;
-    // File.Layer[0]->StartFrame = 0;
-    // File.Layer[0]->EndFrame = 65;
-
-    // ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, 2, 100);
-    // ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, 60, 500);
-
-
-    //  ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, 3, 300);
-    //  ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, 8, 800);
-    //  ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, 5, 500);
-    //  ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, 6, 600);
-    //  ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, 7, 700);
-
-    //  File.Layer[0]->x.KeyframeBlock[0]->Keyframe[2].IsSelected = true;
-    //  File.Layer[0]->x.KeyframeBlock[0]->Keyframe[3].IsSelected = true;
-    //  File.Layer[0]->x.KeyframeBlock[0]->Keyframe[4].IsSelected = true;
-    //  File.Layer[0]->x.KeyframeBlock[0]->Keyframe[6].IsSelected = true;
-
-    // DeleteSelectedKeyframes(&File, &Memory);
-
-    // for (int16 i = 0; i < 10; i++) {
-    //     ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, i*6 + 3, i*100);
-    // }
-
-    // int16 kef = 2;
-    // for (int16 i = 0; i < kef; i++) {
-    //     int16 p = kef - i;
-    //     ManualKeyframeInsertF(&File.Layer[0]->x, &Memory, p*4, p*100);
-    // }
-    // &File.Layer[1]->x.KeyframeBlock[0]->Keyframe[2];
-    // KeyframeDelete(&File.Layer[1]->x, &Memory, 2);
-
-    // AddEffect(File.Layer[0], &Memory, 0);
-    // AddEffect(File.Layer[0], &Memory, 0);
-
-    // for (int i = 0; i < 3; i++)
-        // CreateLayer(&File, &Memory);
-
-    // DebugPrintMemoryUsage(Memory);
-
     SDL_Init(SDL_INIT_VIDEO);
 
     Semaphore = SDL_CreateSemaphore(0);
@@ -583,18 +501,24 @@ int main(int argc, char *argv[]) {
         //      File.CurrentFrame = shmp->shared_framenumber;
         // }
 
+        // Right now IsRendering does nothing. I have it here if we want to
+        // completely detatch the rendering updater onto its own thread so the
+        // UI never lags.
+
         if (State.UpdateFrame && !IsRendering) {
             MainFunction(0, &Memory, &State, &File, &Cache, &CompBuffer);
             State.UpdateFrame = 0;
-            OutputToViewport(&CompBuffer, &State, textureID);
         }
 
 #if THREADED
+        uint32 C = SDL_AtomicGet(&CompletedEntries);
         if (IsRendering) {
-            while (CompletedJobs != 16) {
+            while (C != 16) {
+                C = SDL_AtomicGet(&CompletedEntries);
                 CheckQueue(RenderInfo, 8);
             }
-            if (CompletedJobs == 16) {
+            C = SDL_AtomicGet(&CompletedEntries);
+            if (C == 16) {
                 Convert4x4Chunk(&CompBuffer, 1);
                 EndRenderState(&State);
                 glBindTexture(GL_TEXTURE_2D, textureID);
@@ -608,6 +532,7 @@ int main(int argc, char *argv[]) {
             }
         }
 #else
+        OutputToViewport(&CompBuffer, &State, textureID);
 #endif
 
         ImGui::Render();
diff --git a/main.h b/main.h
index d5a0126..3c81c43 100644
--- a/main.h
+++ b/main.h
@@ -386,6 +386,7 @@ struct project_state
     bool32 IsRunning = 1;
     bool32 IsPlaying;
     bool32 DemoButton = 1;
+    bool32 GridButton = 1;
 
     uint16 NumberOfSelectedLayers;
     int16 MostRecentlySelectedLayer = -1;        // convenience for the properties panel
diff --git a/my_imgui_widgets.cpp b/my_imgui_widgets.cpp
index 1190430..440694b 100644
--- a/my_imgui_widgets.cpp
+++ b/my_imgui_widgets.cpp
@@ -200,7 +200,7 @@ ImGui_Viewport(project_data File, project_state *State, ui *UI, pixel_buffer Com
     // {
     //     Debug.ToggleRenders = true;
     // }
-    ImGui::OpenPopupOnItemClick("context", ImGuiPopupFlags_MouseButtonRight);
+    ImGui::OpenPopupOnItemClick("context", ImGuiPopupFlags_MouseButtonMiddle);
     if (ImGui::BeginPopup("context")) {
         if (ImGui::MenuItem("Scalar", NULL, false, InstructionMode != scalar_only)) { InstructionMode = scalar_only; }
         if (ImGui::MenuItem("SSE", NULL, false, InstructionMode != sse_enabled)) { InstructionMode = sse_enabled; }
@@ -295,6 +295,15 @@ ImGui_File(project_data *File, project_state *State, memory *Memory, ui *UI, ImG
             State->DemoButton = false;
         }
     }
+    if (State->GridButton) {
+        ImGui::SameLine();
+        if (ImGui::Button("Generate square grid")) {
+            CreateGrid(File, Memory);
+            State->UpdateKeyframes = true;
+            State->UpdateFrame = true;
+            State->GridButton = false;
+        }
+    }
     for (int16 i = 0; i < File->NumberOfSources; i++) {
         ImGui::PushID(i);
         ImGui::InputText("##source", File->Source[i], STRING_SIZE);
@@ -932,14 +941,16 @@ ImGui_Timeline(project_data *File, project_state *State, memory *Memory, ui *UI,
 
     if (IsRectTouching(WindowMinAbs, WindowMaxAbs, io.MousePos, io.MousePos + 1)) {
         if (io.KeyCtrl && io.MouseWheel) {
-            real32 ZoomAmount = io.MouseWheel*8;
+            real32 ZoomAmount = io.MouseWheel*16;
             real32 LocalMousePos = ImGui::GetMousePos().x - TimelineStartingPos.x;
             real32 ZoomRatio = LocalMousePos / UI->TimelineZoom;
             UI->TimelineZoom += ZoomAmount;
             UI->ScrollXOffset -= ZoomAmount*ZoomRatio;
+        } else if (io.KeyShift && io.MouseWheel) {
+            UI->ScrollXOffset += io.MouseWheel*16;
         } else {
-            UI->ScrollXOffset += io.MouseWheelH*8;
-            UI->ScrollYOffset += io.MouseWheel*8;
+            UI->ScrollXOffset += io.MouseWheelH*16;
+            UI->ScrollYOffset += io.MouseWheel*16;
         }
     }
 
@@ -987,6 +998,9 @@ ImGui_ProcessInputs(project_data *File, project_state *State, pixel_buffer *Comp
     if (ImGui::IsKeyPressed(ImGuiKey_A)  && State->NumberOfSelectedLayers)
         TransformsInteract(File, State, UI, sliding_anchorpoint);
 
+    if (ImGui::IsKeyPressed(ImGuiKey_1))
+        LoadTestFootage(File, State, Memory);
+
     if (ImGui::IsKeyPressed(ImGuiKey_Delete))
     {
         switch (State->RecentSelectionType)
@@ -1010,11 +1024,6 @@ ImGui_ProcessInputs(project_data *File, project_state *State, pixel_buffer *Comp
     }
 
 #if DEBUG
-    if (ImGui::IsKeyPressed(ImGuiKey_Z))
-    {
-        // SwitchBool(D);
-        // State->UpdateFrame = true;
-    }
     if (ImGui::IsKeyPressed(ImGuiKey_M))
     {
         Debug.Markers[Debug.MarkerIndex] = File->CurrentFrame;
diff --git a/prenderer.cpp b/prenderer.cpp
index 356ecd7..72c2893 100644
--- a/prenderer.cpp
+++ b/prenderer.cpp
@@ -97,13 +97,6 @@ internal void
 EndRenderState(project_state *State)
 {
       IsRendering = false;
-      DEBUG_CycleCountEnd(3);
-      //TODO(fox): proper pixel accounting
-      // Debug.ExecutionAmount[4] += 1280*720;
-
-      // printf("%lu %lu, avg %lu\n", Debug.EndCycleCount[3], Debug.ExecutionAmount[4],
-      //                              Debug.EndCycleCount[3] / Debug.ExecutionAmount[4]);
-      // Debug = {};
 
       for (int16 i = 0; i < State->NumberOfLayersToRender; i++)
       {
@@ -112,9 +105,9 @@ EndRenderState(project_state *State)
 
       State->NumberOfLayersToRender = 0;
 #if THREADED
-      __atomic_store_n(&EntryCount, 0, __ATOMIC_SEQ_CST);
-      __atomic_store_n(&NextEntryToDo, 0, __ATOMIC_SEQ_CST);
-      __atomic_store_n(&CompletedJobs, 0, __ATOMIC_SEQ_CST);
+      SDL_AtomicSet(&CurrentEntry, 0);
+      SDL_AtomicSet(&QueuedEntries, 0);
+      SDL_AtomicSet(&CompletedEntries, 0);
 #endif
 
 }
@@ -138,9 +131,6 @@ QueueCurrentFrame(project_data *File, pixel_buffer *CompBuffer, project_state *S
     IsRendering = true;
     render_queue RenderInfo = {File, State, CompBuffer};
 
-    uint16 TileWidth = (CompBuffer->Width - (CompBuffer->Width & 3)) / 4;
-    uint16 TileHeight = (CompBuffer->Height - (CompBuffer->Height & 3)) / 4;
-
     for (int16 i = 0; i < File->NumberOfLayers; i++)
     {
         if (File->Layer[i]->StartFrame <= File->CurrentFrame &&
@@ -153,54 +143,30 @@ QueueCurrentFrame(project_data *File, pixel_buffer *CompBuffer, project_state *S
     }
 
 #if THREADED
+
+    uint16 TileWidth = CompBuffer->Width / 4;
+    uint16 TileHeight = CompBuffer->Height / 4;
+
     DEBUG_CycleCountStart(3);
     for (int y = 0; y < 4; y++) {
         for (int x = 0; x < 4; x++) {
             // if (x == y) {
             rectangle RenderRegion = {TileWidth*x, TileHeight*y, TileWidth + TileWidth*x, TileHeight + TileHeight*y};
+            if (RenderRegion.Max.x > CompBuffer->Width)
+                RenderRegion.Max.x = CompBuffer->Width;
+            if (RenderRegion.Max.y > CompBuffer->Height)
+                RenderRegion.Max.y = CompBuffer->Height;
             PushRect(RenderRegion);
             // }
         }
     }
 
-    // while (CompletedJobs != 16) {
-    //     // CheckQueue(RenderInfo, 8);
-    // }
-    // DEBUG_CycleCountEnd(3);
-    // // //TODO(fox): proper pixel accounting
-    // Debug.ExecutionAmount[4] += 1280*720;
-
-    // for (int16 i = 0; i < State->NumberOfLayersToRender; i++)
-    // {
-    //     State->LayersToRender[i] = 0;
-    // }
-
-    // State->NumberOfLayersToRender = 0;
-
 #else
-    // DEBUG_CycleCountStart(3);
 
     rectangle RenderRegion = {0, 0, (int32)CompBuffer->Width, (int32)CompBuffer->Height};
     RenderLayers(&RenderInfo, RenderRegion);
 
-    // DEBUG_CycleCountEnd(3);
-    // Debug.ExecutionAmount[4] += 1280*720;
-
-    for (int16 i = 0; i < State->NumberOfLayersToRender; i++)
-    {
-        State->LayersToRender[i] = 0;
-    }
-
-    State->NumberOfLayersToRender = 0;
-
 #endif
-
-    // printf("Completed jobs: %i\n", CompletedJobs);
-    // printf("Next: %i\n", NextEntryToDo);
-    // Assert(CompletedJobs == 4*4);
-    // __atomic_store_n(&EntryCount, 0, __ATOMIC_SEQ_CST);
-    // __atomic_store_n(&NextEntryToDo, 0, __ATOMIC_SEQ_CST);
-    // __atomic_store_n(&CompletedJobs, 0, __ATOMIC_SEQ_CST);
 }
 
 
diff --git a/threading.cpp b/threading.cpp
index 39e7b75..1b89136 100644
--- a/threading.cpp
+++ b/threading.cpp
@@ -1,26 +1,30 @@
 internal void
 PushRect(rectangle RenderRegion)
 {
-    render_entry *Entry = Entries + EntryCount;
+    uint32 Q = SDL_AtomicGet(&QueuedEntries);
+    render_entry *Entry = Entries + Q;
     Entry->RenderRegion = RenderRegion;
-    __atomic_add_fetch(&EntryCount, 1, __ATOMIC_ACQ_REL);
+    SDL_AtomicAdd(&QueuedEntries, 1);
     SDL_SemPost(Semaphore);
 }
 
+internal void
+RenderLayers(render_queue *RenderInfo, rectangle RenderRegion);
 
 internal bool32
 CheckQueue(render_queue RenderInfo, uint16 Index)
 {
     bool32 Result = 0;
-    uint32 OriginalEntry = NextEntryToDo;
-    if (NextEntryToDo < EntryCount)
+    uint32 Q = SDL_AtomicGet(&QueuedEntries);
+    uint32 C = SDL_AtomicGet(&CurrentEntry);
+    if (Q > C)
         {
-            if (__atomic_compare_exchange_n(&NextEntryToDo, &OriginalEntry, NextEntryToDo + 1, true, __ATOMIC_RELEASE, __ATOMIC_ACQUIRE)) {
-                render_entry *Entry = Entries + OriginalEntry;
+            if (SDL_AtomicCAS(&CurrentEntry, C, C + 1)) {
+                render_entry *Entry = Entries + C;
                 Assert(Entry->RenderRegion.Max.x != 0);
-                RenderLayers(RenderInfo, Entry->RenderRegion);
+                RenderLayers(&RenderInfo, Entry->RenderRegion);
                 // printf("(FINISHED) Thread %i, region X%i Y%i\n", Index, Entry->RenderRegion.Min.x/240, Entry->RenderRegion.Min.y/135);
-                __atomic_add_fetch(&CompletedJobs, 1, __ATOMIC_ACQ_REL);
+                SDL_AtomicAdd(&CompletedEntries, 1);
                 Result = 1;
             }
         }