gsdx-ogl: LINUX-ONLY

* invert the index of fragment output. Seem to work better on Nvidia (strangely no impact for AMD)
* opengl support pitch too, so remove useless copy to an fbo


git-svn-id: http://pcsx2.googlecode.com/svn/branches/gsdx-ogl@5034 96395faa-99c1-11dd-bbfe-3dabce05a288

plugins/GSdx/CMakeLists.txt

@@ -18,6 +18,7 @@ set(CommonFlags
     -Wunused-variable
     -std=c++0x
     -fno-strict-aliasing
+    -DOGL_DEBUG
     )
 
 set(OptimizationFlags
@@ -28,7 +29,7 @@ set(OptimizationFlags
 # Debug - Build
 if(CMAKE_BUILD_TYPE STREQUAL Debug)
     # add defines
-    add_definitions(${CommonFlags} -g -Wall)
+    add_definitions(${CommonFlags} -DOGL_DEBUG -g -Wall)
 endif(CMAKE_BUILD_TYPE STREQUAL Debug)
 
 # Devel - Build

plugins/GSdx/GSDeviceOGL.cpp

@@ -54,8 +54,8 @@
 
 //#define LOUD_DEBUGGING
 #define SHADER_DEBUG
-//#define DUMP_START (500)
-//#define DUMP_LENGTH (40)
+//#define DUMP_START (13000)
+//#define DUMP_LENGTH (200)
 //#define DUMP_ONLY_FRAME  (112)
 
 #ifdef DUMP_START
@@ -219,9 +219,9 @@ bool GSDeviceOGL::Create(GSWnd* wnd)
 
 	glGenFramebuffers(1, &m_fbo);
 	// Setup FBO fragment output
-	OMSetFBO(m_fbo);
-	glDrawBuffer(GL_COLOR_ATTACHMENT0);
-	OMSetFBO(0);
+	// OMSetFBO(m_fbo);
+	// glDrawBuffer(GL_COLOR_ATTACHMENT0);
+	// OMSetFBO(0);
 
 	// ****************************************************************
 	// Vertex buffer state
@@ -706,12 +706,11 @@ void GSDeviceOGL::CopyRect(GSTexture* st, GSTexture* dt, const GSVector4i& r)
 	// GL_NV_copy_image seem like the good extension but not supported on AMD...
 	// Maybe opengl 4.3 !
 	// FIXME check those function work as expected
-	
+
 	// FIXME FBO
 	GLuint fbo_old = m_state.fbo;
 	OMSetFBO(m_fbo);
 
-
 	// Set the input of glCopyTexSubImage2D
 	static_cast<GSTextureOGL*>(st)->Attach(GL_COLOR_ATTACHMENT1);
 	glReadBuffer(GL_COLOR_ATTACHMENT1);
@@ -1012,6 +1011,8 @@ void GSDeviceOGL::IASetVertexBuffer(const void* vertices, size_t count)
 	}
 
 	vb->upload(vertices, map_flags);
+	// FIXME: disable it when code is working
+	CheckDebugLog();
 }
 
 void GSDeviceOGL::IASetPrimitiveTopology(GLenum topology)

plugins/GSdx/GSDeviceOGL.h

@@ -171,6 +171,12 @@ struct GSVertexBufferState {
 	void upload(const void* src, uint32 flags)
 	{
 		uint8* dst = (uint8*) glMapBufferRange(GL_ARRAY_BUFFER, stride*start, stride*count, flags);
+#ifdef OGL_DEBUG
+		if (dst == NULL) {
+			fprintf(stderr, "CRITICAL ERROR map failed for vb !!!\n");
+			return;
+		}
+#endif
 		memcpy(dst, src, stride*count);
 		glUnmapBuffer(GL_ARRAY_BUFFER);
 	}

plugins/GSdx/GSTextureOGL.cpp

@@ -26,8 +26,7 @@ static uint g_state_texture_unit = 0;
 static uint g_state_texture_id = 0;
 
 GSTextureOGL::GSTextureOGL(int type, int w, int h, bool msaa, int format)
-	: m_texture_unit(0),
-	  m_extra_buffer_id(0),
+	: m_extra_buffer_id(0),
 	  m_pbo_size(0)
 {
 	// *************************************************************
@@ -120,7 +119,7 @@ GSTextureOGL::GSTextureOGL(int type, int w, int h, bool msaa, int format)
 	switch (m_type) {
 		case GSTexture::DepthStencil:
 			EnableUnit(1);
-			glTexImage2D(m_texture_target, 0, m_format, w, h, 0, GL_DEPTH_STENCIL, GL_FLOAT_32_UNSIGNED_INT_24_8_REV, NULL);
+			glTexImage2D(m_texture_target, 0, m_format, m_size.x, m_size.y, 0, GL_DEPTH_STENCIL, GL_FLOAT_32_UNSIGNED_INT_24_8_REV, NULL);
 			break;
 		case GSTexture::RenderTarget:
 		case GSTexture::Texture:
@@ -130,7 +129,7 @@ GSTextureOGL::GSTextureOGL(int type, int w, int h, bool msaa, int format)
 			// For the moment SW renderer only use 1 so don't bother
 			EnableUnit(0);
 			if (m_format == GL_RGBA8) {
-				glTexImage2D(m_texture_target, 0, m_format, w, h, 0, GL_RGBA, GL_UNSIGNED_BYTE, NULL);
+				glTexImage2D(m_texture_target, 0, m_format, m_size.x, m_size.y, 0, GL_RGBA, GL_UNSIGNED_BYTE, NULL);
 			}
 			else
 				assert(0); // TODO Later
@@ -161,54 +160,27 @@ void GSTextureOGL::Attach(GLenum attachment)
 
 bool GSTextureOGL::Update(const GSVector4i& r, const void* data, int pitch)
 {
-	// static int update_count = 0;
-	// update_count++;
-	// To update only a part of the texture you can use:
-	// glTexSubImage2D — specify a two-dimensional texture subimage
 	if (m_type == GSTexture::DepthStencil || m_type == GSTexture::Offscreen) assert(0);
 
-	// glTexSubImage2D specifies a two-dimensional subtexture for the current texture unit, specified with glActiveTexture.
-	// If a non-zero named buffer object is bound to the GL_PIXEL_UNPACK_BUFFER target
-	//(see glBindBuffer) while a texture image is
-	// specified, data is treated as a byte offset into the buffer object's data store
 	// FIXME warning order of the y axis
 	// FIXME I'm not confident with GL_UNSIGNED_BYTE type
-	// FIXME add a state check
 
 	EnableUnit(0);
 
-	if (m_format != GL_RGBA8) assert(0);
-
-	// FIXME. That suck but I don't know how to do better for now. I think we need to create a texture buffer and directly upload the copy
-	// The case appears on SW mode. Src pitch is 2x dst pitch.
-	int rowbytes = r.width() << 2;
-	if (pitch != rowbytes) {
-		uint32 map_flags = GL_MAP_WRITE_BIT;
-
-		glBindBuffer(GL_PIXEL_UNPACK_BUFFER, m_extra_buffer_id);
-		if (!m_pbo_size) {
-			m_pbo_size = m_size.x * m_size.y * 4;
-			glBufferData(GL_PIXEL_UNPACK_BUFFER, m_pbo_size, NULL, GL_STREAM_DRAW);
-		} else {
-			GL_MAP_INVALIDATE_BUFFER_BIT;
-		}
-
-		uint8* src = (uint8*) data;
-		uint8* dst = (uint8*) glMapBufferRange(GL_PIXEL_UNPACK_BUFFER, 0, m_pbo_size, map_flags);
-		for(int h = r.height(); h > 0; h--, src += pitch, dst += rowbytes)
-		{
-			memcpy(dst, src, rowbytes);
-		}
-		glUnmapBuffer(GL_PIXEL_UNPACK_BUFFER);
-
-		glTexSubImage2D(m_texture_target, 0, r.x, r.y, r.width(), r.height(), GL_RGBA, GL_UNSIGNED_BYTE, 0 /* offset in UNPACK BUFFER */);
-
-		glBindBuffer(GL_PIXEL_UNPACK_BUFFER, 0);
-
-	} else {
-		glTexSubImage2D(m_texture_target, 0, r.x, r.y, r.width(), r.height(), GL_RGBA, GL_UNSIGNED_BYTE, data);
+	if (m_format != GL_RGBA8) {
+		fprintf(stderr, "wrong pixel format\n");
+		assert(0);
 	}
 
+	// pitch could be different of width*element_size
+	glPixelStorei(GL_UNPACK_ROW_LENGTH, pitch>>2);
+	// FIXME : it crash on colin mcrae rally 3 (others game too) when the size is 16
+	//if (m_size.x != 16)
+	glTexSubImage2D(m_texture_target, 0, r.x, r.y, r.width(), r.height(), GL_RGBA, GL_UNSIGNED_BYTE, data);
+
+	glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); // Restore default behavior
+
+	return true;
 #if 0
 	if(m_dev && m_texture)
 	{

plugins/GSdx/GSTextureOGL.h

@@ -28,7 +28,6 @@ class GSTextureOGL : public GSTexture
 	private:
 		GLenum m_texture_target; // texture target: 2D, rectangle etc...
 		GLuint m_texture_id;	 // the texture id
-		uint m_texture_unit;		 // the texture unit offset
 		uint m_extra_buffer_id;
 		int m_pbo_size;
 

plugins/GSdx/res/tfx.glsl

@@ -257,8 +257,8 @@ void gs_main()
 layout(location = 0) in vertex PSin;
 
 // Same buffer but 2 colors for dual source blending
-layout(location = 0, index = 0) out vec4 SV_Target0;
-layout(location = 0, index = 1) out vec4 SV_Target1;
+layout(location = 0, index = 1) out vec4 SV_Target0;
+layout(location = 0, index = 0) out vec4 SV_Target1;
 
 layout(binding = 0) uniform sampler2D TextureSampler;
 layout(binding = 1) uniform sampler2D PaletteSampler;