gl: Finalize BGRA storage format internals

- Performance is terrible but it works properly now

rpcs3/Emu/RSX/GL/GLCompute.cpp

@@ -302,8 +302,8 @@ namespace gl
 		m_program.uniforms["region_offset"] = color2i(region.x, region.y);
 		m_program.uniforms["region_size"] = color2i(region.width, region.height);
 
-		auto depth_view = src->get_view(0xAAE4, rsx::default_remap_vector, gl::image_aspect::depth);
+		auto depth_view = src->get_view(GL_REMAP_IDENTITY, rsx::default_remap_vector, gl::image_aspect::depth);
-		auto stencil_view = src->get_view(0xAAE4, rsx::default_remap_vector, gl::image_aspect::stencil);
+		auto stencil_view = src->get_view(GL_REMAP_IDENTITY, rsx::default_remap_vector, gl::image_aspect::stencil);
 
 		depth_view->bind(cmd, GL_COMPUTE_BUFFER_SLOT(0));
 		stencil_view->bind(cmd, GL_COMPUTE_BUFFER_SLOT(1));
@@ -343,7 +343,7 @@ namespace gl
 		m_program.uniforms["is_bgra"] = (layout.format == static_cast<GLenum>(gl::texture::format::bgra));
 		m_program.uniforms["block_width"] = static_cast<u32>(layout.size);
 
-		auto data_view = src->get_view(0xAAE4, rsx::default_remap_vector, gl::image_aspect::color);
+		auto data_view = src->get_view(GL_REMAP_IDENTITY, rsx::default_remap_vector, gl::image_aspect::color);
 
 		data_view->bind(cmd, GL_COMPUTE_BUFFER_SLOT(0));
 		dst->bind_range(gl::buffer::target::ssbo, GL_COMPUTE_BUFFER_SLOT(1), out_offset, row_pitch * 4 * region.height);

rpcs3/Emu/RSX/GL/GLRenderTargets.cpp

@@ -13,7 +13,7 @@ color_format rsx::internals::surface_color_format_to_gl(rsx::surface_color_forma
 		return{ ::gl::texture::type::ushort_5_6_5, ::gl::texture::format::rgb, ::gl::texture::internal_format::rgb565, true };
 
 	case rsx::surface_color_format::a8r8g8b8:
-		return{ ::gl::texture::type::uint_8_8_8_8, ::gl::texture::format::bgra, ::gl::texture::internal_format::bgra8, false };
+		return{ ::gl::texture::type::uint_8_8_8_8_rev, ::gl::texture::format::bgra, ::gl::texture::internal_format::bgra8, true };
 
 	//These formats discard their alpha component, forced to 0 or 1
 	//All XBGR formats will have remapping before they can be read back in shaders as DRGB8
@@ -27,19 +27,19 @@ color_format rsx::internals::surface_color_format_to_gl(rsx::surface_color_forma
 		{ ::gl::texture::channel::zero, ::gl::texture::channel::r, ::gl::texture::channel::g, ::gl::texture::channel::b } };
 
 	case rsx::surface_color_format::x8r8g8b8_z8r8g8b8:
-		return{ ::gl::texture::type::uint_8_8_8_8, ::gl::texture::format::bgra, ::gl::texture::internal_format::bgra8, false,
+		return{ ::gl::texture::type::uint_8_8_8_8_rev, ::gl::texture::format::bgra, ::gl::texture::internal_format::bgra8, true,
 		{ ::gl::texture::channel::zero, ::gl::texture::channel::r, ::gl::texture::channel::g, ::gl::texture::channel::b } };
 
 	case rsx::surface_color_format::x8b8g8r8_o8b8g8r8:
-		return{ ::gl::texture::type::uint_8_8_8_8, ::gl::texture::format::rgba, ::gl::texture::internal_format::rgba8, false,
+		return{ ::gl::texture::type::uint_8_8_8_8_rev, ::gl::texture::format::rgba, ::gl::texture::internal_format::rgba8, true,
-		{ ::gl::texture::channel::one, ::gl::texture::channel::b, ::gl::texture::channel::g, ::gl::texture::channel::r } };
+		{ ::gl::texture::channel::one, ::gl::texture::channel::r, ::gl::texture::channel::g, ::gl::texture::channel::b } };
 
 	case rsx::surface_color_format::x8b8g8r8_z8b8g8r8:
-		return{ ::gl::texture::type::uint_8_8_8_8, ::gl::texture::format::rgba, ::gl::texture::internal_format::rgba8, false,
+		return{ ::gl::texture::type::uint_8_8_8_8_rev, ::gl::texture::format::rgba, ::gl::texture::internal_format::rgba8, true,
-		{ ::gl::texture::channel::zero, ::gl::texture::channel::b, ::gl::texture::channel::g, ::gl::texture::channel::r } };
+		{ ::gl::texture::channel::zero, ::gl::texture::channel::r, ::gl::texture::channel::g, ::gl::texture::channel::b } };
 
 	case rsx::surface_color_format::x8r8g8b8_o8r8g8b8:
-		return{ ::gl::texture::type::uint_8_8_8_8, ::gl::texture::format::bgra, ::gl::texture::internal_format::bgra8, false,
+		return{ ::gl::texture::type::uint_8_8_8_8_rev, ::gl::texture::format::bgra, ::gl::texture::internal_format::bgra8, true,
 		{ ::gl::texture::channel::one, ::gl::texture::channel::r, ::gl::texture::channel::g, ::gl::texture::channel::b } };
 
 	case rsx::surface_color_format::w16z16y16x16:
@@ -61,8 +61,7 @@ color_format rsx::internals::surface_color_format_to_gl(rsx::surface_color_forma
 		{ ::gl::texture::channel::r, ::gl::texture::channel::r, ::gl::texture::channel::r, ::gl::texture::channel::r } };
 
 	case rsx::surface_color_format::a8b8g8r8:
-		return{ ::gl::texture::type::uint_8_8_8_8, ::gl::texture::format::rgba, ::gl::texture::internal_format::rgba8, false,
+		return{ ::gl::texture::type::uint_8_8_8_8_rev, ::gl::texture::format::rgba, ::gl::texture::internal_format::rgba8, true };
-		{ ::gl::texture::channel::a, ::gl::texture::channel::b, ::gl::texture::channel::g, ::gl::texture::channel::r } };
 
 	default:
 		fmt::throw_exception("Unsupported surface color format 0x%x", static_cast<u32>(color_format));

rpcs3/Emu/RSX/GL/GLTexture.cpp

@@ -169,9 +169,9 @@ namespace gl
 		case texture::internal_format::rgba4:
 			return { GL_BGRA, GL_UNSIGNED_SHORT_4_4_4_4, 2, false };
 		case texture::internal_format::rgba8:
-			return { GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, 4, false };
+			return { GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, 4, true };
 		case texture::internal_format::bgra8:
-			return { GL_BGRA, GL_UNSIGNED_INT_8_8_8_8, 4, false };
+			return { GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV, 4, true };
 		case texture::internal_format::rgba16f:
 			return { GL_RGBA, GL_HALF_FLOAT, 2, true };
 		case texture::internal_format::rgba32f:

rpcs3/Emu/RSX/GL/GLTextureCache.h

@@ -713,7 +713,7 @@ namespace gl
 				{
 				case CELL_GCM_TEXTURE_A8R8G8B8:
 				{
-					cached.set_format(gl::texture::format::bgra, gl::texture::type::uint_8_8_8_8, false);
+					cached.set_format(gl::texture::format::bgra, gl::texture::type::uint_8_8_8_8_rev, true);
 					break;
 				}
 				case CELL_GCM_TEXTURE_R5G6B5:
@@ -814,8 +814,7 @@ namespace gl
 				return (ifmt == gl::texture::internal_format::rgb565);
 			case CELL_GCM_TEXTURE_A8R8G8B8:
 			case CELL_GCM_TEXTURE_D8R8G8B8:
-				return (ifmt == gl::texture::internal_format::rgba8 ||
+				return (ifmt == gl::texture::internal_format::bgra8 ||
-						ifmt == gl::texture::internal_format::bgra8 ||
 						ifmt == gl::texture::internal_format::depth24_stencil8 ||
 						ifmt == gl::texture::internal_format::depth32f_stencil8);
 			case CELL_GCM_TEXTURE_B8:

rpcs3/Emu/RSX/GL/glutils/image.cpp

@@ -24,7 +24,7 @@ namespace gl
 		glGenTextures(1, &m_id);
 
 		// Must bind to initialize the new texture
-		gl::get_command_context()->bind_texture(GL_TEMP_IMAGE_SLOT, target, m_id);
+		gl::get_command_context()->bind_texture(GL_TEMP_IMAGE_SLOT, target, m_id, GL_TRUE);
 
 		const GLenum storage_fmt = sizedfmt_to_ifmt(sized_format);
 		switch (target)
@@ -251,7 +251,7 @@ namespace gl
 			component_swizzle[2] = argb_swizzle[3];
 			component_swizzle[3] = argb_swizzle[0];
 
-			gl::get_command_context()->bind_texture(GL_TEMP_IMAGE_SLOT, m_target, m_id);
+			gl::get_command_context()->bind_texture(GL_TEMP_IMAGE_SLOT, m_target, m_id, GL_TRUE);
 			glTexParameteriv(m_target, GL_TEXTURE_SWIZZLE_RGBA, reinterpret_cast<GLint*>(component_swizzle));
 		}
 		else
@@ -267,7 +267,7 @@ namespace gl
 			constexpr u32 depth_stencil_mask = (image_aspect::depth | image_aspect::stencil);
 			ensure((range.aspect_mask & depth_stencil_mask) != depth_stencil_mask); // "Invalid aspect mask combination"
 
-			gl::get_command_context()->bind_texture(GL_TEMP_IMAGE_SLOT, m_target, m_id);
+			gl::get_command_context()->bind_texture(GL_TEMP_IMAGE_SLOT, m_target, m_id, GL_TRUE);
 			glTexParameteri(m_target, GL_DEPTH_STENCIL_TEXTURE_MODE, GL_STENCIL_INDEX);
 		}
 	}
@@ -287,8 +287,9 @@ namespace gl
 		cmd->bind_texture(layer, m_target, m_id);
 	}
 
-	texture_view* viewable_image::get_view(u32 remap_encoding, const std::pair<std::array<u8, 4>, std::array<u8, 4>>& remap, GLenum aspect_flags)
+	texture_view* viewable_image::get_view(u32 remap_encoding, const std::pair<std::array<u8, 4>, std::array<u8, 4>>& remap_, GLenum aspect_flags)
 	{
+		auto remap = remap_;
 		const u64 view_aspect = static_cast<u64>(aspect_flags) & aspect();
 		ensure(view_aspect);
 

rpcs3/Emu/RSX/GL/glutils/state_tracker.hpp

@@ -287,12 +287,12 @@ namespace gl
 			glUseProgram(program);
 		}
 
-		void bind_texture(GLuint layer, GLenum target, GLuint name)
+		void bind_texture(GLuint layer, GLenum target, GLuint name, GLboolean force = GL_FALSE)
 		{
 			ensure(layer < 48);
 
 			auto& bound = bound_textures[layer][target];
-			if (bound != name)
+			if (bound != name || force)
 			{
 				glActiveTexture(GL_TEXTURE0 + layer);
 				glBindTexture(target, name);

rpcs3/Emu/RSX/Program/GLSLSnippets/CopyRGBA8ToBuffer.glsl

@@ -77,17 +77,7 @@ void main()
 		}
 
 		uvec4 bytes = uvec4(color * 255);
-		uint result;
+		uint result = bytes.x | (bytes.y << 8u) | (bytes.z << 16u) | (bytes.w << 24u); // UINT_8_8_8_8_REV
-		
-		if (block_width > 1)
-		{
-			// Simulate BE packing as in UINT_8_8_8_8
-			result = bytes.w | (bytes.z << 8u) | (bytes.y << 16u) | (bytes.x << 24u);
-		}
-		else
-		{
-			result = bytes.x | (bytes.y << 8u) | (bytes.z << 16u) | (bytes.w << 24u);
-		}
 
 		uint output_id = input_coord_to_output_id(coord);
 		data[output_id] = result;

rpcs3/Emu/RSX/RSXTexture.cpp

@@ -242,26 +242,7 @@ namespace rsx
 			break;
 		}
 
-		//Remapping tables; format is A-R-G-B
+		return decode_remap_encoding(remap_ctl);
-		//Remap input table. Contains channel index to read color from
-		const std::array<u8, 4> remap_inputs =
-		{
-			static_cast<u8>(remap_ctl & 0x3),
-			static_cast<u8>((remap_ctl >> 2) & 0x3),
-			static_cast<u8>((remap_ctl >> 4) & 0x3),
-			static_cast<u8>((remap_ctl >> 6) & 0x3),
-		};
-
-		//Remap control table. Controls whether the remap value is used, or force either 0 or 1
-		const std::array<u8, 4> remap_lookup =
-		{
-			static_cast<u8>((remap_ctl >> 8) & 0x3),
-			static_cast<u8>((remap_ctl >> 10) & 0x3),
-			static_cast<u8>((remap_ctl >> 12) & 0x3),
-			static_cast<u8>((remap_ctl >> 14) & 0x3),
-		};
-
-		return std::make_pair(remap_inputs, remap_lookup);
 	}
 
 	f32 fragment_texture::bias() const

rpcs3/Emu/RSX/rsx_utils.h

@@ -217,6 +217,30 @@ namespace rsx
 		{ CELL_GCM_TEXTURE_REMAP_REMAP, CELL_GCM_TEXTURE_REMAP_REMAP, CELL_GCM_TEXTURE_REMAP_REMAP, CELL_GCM_TEXTURE_REMAP_REMAP }
 	};
 
+	static inline std::pair<std::array<u8, 4>, std::array<u8, 4>> decode_remap_encoding(u32 remap_ctl)
+	{
+		// Remapping tables; format is A-R-G-B
+		// Remap input table. Contains channel index to read color from
+		const std::array<u8, 4> remap_inputs =
+		{
+			static_cast<u8>(remap_ctl & 0x3),
+			static_cast<u8>((remap_ctl >> 2) & 0x3),
+			static_cast<u8>((remap_ctl >> 4) & 0x3),
+			static_cast<u8>((remap_ctl >> 6) & 0x3),
+		};
+
+		// Remap control table. Controls whether the remap value is used, or force either 0 or 1
+		const std::array<u8, 4> remap_lookup =
+		{
+			static_cast<u8>((remap_ctl >> 8) & 0x3),
+			static_cast<u8>((remap_ctl >> 10) & 0x3),
+			static_cast<u8>((remap_ctl >> 12) & 0x3),
+			static_cast<u8>((remap_ctl >> 14) & 0x3),
+		};
+
+		return std::make_pair(remap_inputs, remap_lookup);
+	}
+
 	template <typename T>
 	void pad_texture(void* input_pixels, void* output_pixels, u16 input_width, u16 input_height, u16 output_width, u16 /*output_height*/)
 	{