[client/host] added new asyncronous memory copy

This changes the method of the memory copy from the host application to
the guest. Instead of performing a full copy from the capture device
into shared memory, and then flagging the new frame, we instead set a
write pointer, flag the client that there is a new frame and then copy
in chunks of 1024 bytes until the entire frame is copied. The client
upon seeing the new frame flag begins to poll at high frequency the
write pointer and upon each update copies as much as it can into the
texture.

This should improve latency but also slightly increase CPU usage on the
client due to the high frequency polling.

c-host/include/interface/capture.h

@@ -21,6 +21,7 @@ Place, Suite 330, Boston, MA 02111-1307 USA
 
 #include <stdbool.h>
 #include <stdint.h>
+#include "common/framebuffer.h"
 
 typedef enum CaptureResult
 {
@@ -55,7 +56,6 @@ typedef struct CaptureFrame
   unsigned int   pitch;
   unsigned int   stride;
   CaptureFormat  format;
-  void         * data;
 }
 CaptureFrame;
 
@@ -84,7 +84,8 @@ typedef struct CaptureInterface
   unsigned int  (*getMaxFrameSize)();
 
   CaptureResult (*capture   )();
-  CaptureResult (*getFrame  )(CaptureFrame   * frame  );
+  CaptureResult (*waitFrame )(CaptureFrame   * frame  );
+  CaptureResult (*getFrame  )(FrameBuffer      frame  );
   CaptureResult (*getPointer)(CapturePointer * pointer);
 }
 CaptureInterface;

c-host/platform/Windows/capture/DXGI/src/dxgi.c

@@ -766,7 +766,7 @@ static CaptureResult dxgi_capture()
   return CAPTURE_RESULT_OK;
 }
 
-static CaptureResult dxgi_getFrame(CaptureFrame * frame)
+static CaptureResult dxgi_waitFrame(CaptureFrame * frame)
 {
   assert(this);
   assert(this->initialized);
@@ -778,7 +778,6 @@ static CaptureResult dxgi_getFrame(CaptureFrame * frame)
   if (this->stop)
     return CAPTURE_RESULT_REINIT;
 
-  // only reset the event if we used the texture
   os_resetEvent(tex->mapped);
 
   frame->width  = this->width;
@@ -787,7 +786,16 @@ static CaptureResult dxgi_getFrame(CaptureFrame * frame)
   frame->stride = this->stride;
   frame->format = this->format;
 
-  memcpy(frame->data, tex->map.pData, this->pitch * this->height);
+  return CAPTURE_RESULT_OK;
+}
+
+static CaptureResult dxgi_getFrame(FrameBuffer frame)
+{
+  assert(this);
+  assert(this->initialized);
+
+  Texture * tex = &this->texture[this->texRIndex];
+  framebuffer_write(frame, tex->map.pData, this->pitch * this->height);
   os_signalEvent(tex->free);
 
   if (++this->texRIndex == this->maxTextures)
@@ -867,6 +875,7 @@ struct CaptureInterface Capture_DXGI =
   .free            = dxgi_free,
   .getMaxFrameSize = dxgi_getMaxFrameSize,
   .capture         = dxgi_capture,
+  .waitFrame       = dxgi_waitFrame,
   .getFrame        = dxgi_getFrame,
   .getPointer      = dxgi_getPointer
 };

c-host/platform/Windows/capture/NVFBC/src/nvfbc.c

@@ -23,6 +23,7 @@ Place, Suite 330, Boston, MA 02111-1307 USA
 #include "windows/debug.h"
 #include "windows/mousehook.h"
 #include "common/option.h"
+#include "common/framebuffer.h"
 #include <assert.h>
 #include <stdlib.h>
 #include <windows.h>
@@ -236,7 +237,7 @@ static CaptureResult nvfbc_capture()
   return CAPTURE_RESULT_OK;
 }
 
-static CaptureResult nvfbc_getFrame(CaptureFrame * frame)
+static CaptureResult nvfbc_waitFrame(CaptureFrame * frame)
 {
   if (!os_waitEvent(this->frameEvent, 1000))
     return CAPTURE_RESULT_TIMEOUT;
@@ -266,7 +267,16 @@ static CaptureResult nvfbc_getFrame(CaptureFrame * frame)
 #endif
 
   frame->format = this->grabInfo.bIsHDR ? CAPTURE_FMT_RGBA10 : CAPTURE_FMT_BGRA;
-  memcpy(frame->data, this->frameBuffer, frame->pitch * frame->height);
+  return CAPTURE_RESULT_OK;
+}
+
+static CaptureResult nvfbc_getFrame(FrameBuffer frame)
+{
+  framebuffer_write(
+    frame,
+    this->frameBuffer,
+    this->grabInfo.dwHeight * this->grabInfo.dwBufferWidth * 4
+  );
   return CAPTURE_RESULT_OK;
 }
 
@@ -310,6 +320,7 @@ struct CaptureInterface Capture_NVFBC =
   .free            = nvfbc_free,
   .getMaxFrameSize = nvfbc_getMaxFrameSize,
   .capture         = nvfbc_capture,
+  .waitFrame       = nvfbc_waitFrame,
   .getFrame        = nvfbc_getFrame,
   .getPointer      = nvfbc_getPointer
 };

c-host/src/app.c

@@ -49,7 +49,7 @@ struct app
 
   uint8_t     * frames;
   unsigned int  frameSize;
-  uint8_t     * frame[MAX_FRAMES];
+  FrameBuffer   frame[MAX_FRAMES];
   unsigned int  frameOffset[MAX_FRAMES];
 
   bool             running;
@@ -168,9 +168,7 @@ static int frameThread(void * opaque)
 
   while(app.running)
   {
-    frame.data = app.frame[frameIndex];
-
-    switch(app.iface->getFrame(&frame))
+    switch(app.iface->waitFrame(&frame))
     {
       case CAPTURE_RESULT_OK:
         break;
@@ -226,7 +224,9 @@ static int frameThread(void * opaque)
     fi->dataPos = app.frameOffset[frameIndex];
     frameValid  = true;
 
+    framebuffer_prepare(app.frame[frameIndex]);
     INTERLOCKED_OR8(&fi->flags, KVMFR_FRAME_FLAG_UPDATE);
+    app.iface->getFrame(app.frame[frameIndex]);
 
     if (++frameIndex == MAX_FRAMES)
       frameIndex = 0;
@@ -369,8 +369,8 @@ int app_main(int argc, char * argv[])
 
   for (int i = 0; i < MAX_FRAMES; ++i)
   {
-    app.frame      [i] = app.frames + i * app.frameSize;
-    app.frameOffset[i] = app.frame[i] - shmemMap;
+    app.frame      [i] = (FrameBuffer)(app.frames + i * app.frameSize);
+    app.frameOffset[i] = (uint8_t *)app.frame[i] - shmemMap;
     DEBUG_INFO("Frame %d          : 0x%" PRIXPTR " (0x%08x)", i, (uintptr_t)app.frame[i], app.frameOffset[i]);
   }