summaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/video_core/pica.h24
-rw-r--r--src/video_core/rasterizer.cpp23
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp34
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h13
-rw-r--r--src/video_core/renderer_opengl/gl_shader_gen.cpp20
-rw-r--r--src/video_core/renderer_opengl/pica_to_gl.h1
6 files changed, 112 insertions, 3 deletions
diff --git a/src/video_core/pica.h b/src/video_core/pica.h
index 09702d46a..7099c31a0 100644
--- a/src/video_core/pica.h
+++ b/src/video_core/pica.h
@@ -115,7 +115,28 @@ struct Regs {
BitField<24, 5, Semantic> map_w;
} vs_output_attributes[7];
- INSERT_PADDING_WORDS(0x11);
+ INSERT_PADDING_WORDS(0xe);
+
+ enum class ScissorMode : u32 {
+ Disabled = 0,
+ Exclude = 1, // Exclude pixels inside the scissor box
+
+ Include = 3 // Exclude pixels outside the scissor box
+ };
+
+ struct {
+ BitField<0, 2, ScissorMode> mode;
+
+ union {
+ BitField< 0, 16, u32> x1;
+ BitField<16, 16, u32> y1;
+ };
+
+ union {
+ BitField< 0, 16, u32> x2;
+ BitField<16, 16, u32> y2;
+ };
+ } scissor_test;
union {
BitField< 0, 10, s32> x;
@@ -1328,6 +1349,7 @@ ASSERT_REG_POSITION(viewport_depth_range, 0x4d);
ASSERT_REG_POSITION(viewport_depth_near_plane, 0x4e);
ASSERT_REG_POSITION(vs_output_attributes[0], 0x50);
ASSERT_REG_POSITION(vs_output_attributes[1], 0x51);
+ASSERT_REG_POSITION(scissor_test, 0x65);
ASSERT_REG_POSITION(viewport_corner, 0x68);
ASSERT_REG_POSITION(depthmap_enable, 0x6D);
ASSERT_REG_POSITION(texture0_enable, 0x80);
diff --git a/src/video_core/rasterizer.cpp b/src/video_core/rasterizer.cpp
index a84170094..6f369a00e 100644
--- a/src/video_core/rasterizer.cpp
+++ b/src/video_core/rasterizer.cpp
@@ -338,12 +338,26 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0,
return;
}
- // TODO: Proper scissor rect test!
u16 min_x = std::min({vtxpos[0].x, vtxpos[1].x, vtxpos[2].x});
u16 min_y = std::min({vtxpos[0].y, vtxpos[1].y, vtxpos[2].y});
u16 max_x = std::max({vtxpos[0].x, vtxpos[1].x, vtxpos[2].x});
u16 max_y = std::max({vtxpos[0].y, vtxpos[1].y, vtxpos[2].y});
+ // Convert the scissor box coordinates to 12.4 fixed point
+ u16 scissor_x1 = (u16)( regs.scissor_test.x1 << 4);
+ u16 scissor_y1 = (u16)( regs.scissor_test.y1 << 4);
+ // x2,y2 have +1 added to cover the entire sub-pixel area
+ u16 scissor_x2 = (u16)((regs.scissor_test.x2 + 1) << 4);
+ u16 scissor_y2 = (u16)((regs.scissor_test.y2 + 1) << 4);
+
+ if (regs.scissor_test.mode == Regs::ScissorMode::Include) {
+ // Calculate the new bounds
+ min_x = std::max(min_x, scissor_x1);
+ min_y = std::max(min_y, scissor_y1);
+ max_x = std::min(max_x, scissor_x2);
+ max_y = std::min(max_y, scissor_y2);
+ }
+
min_x &= Fix12P4::IntMask();
min_y &= Fix12P4::IntMask();
max_x = ((max_x + Fix12P4::FracMask()) & Fix12P4::IntMask());
@@ -383,6 +397,13 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0,
for (u16 y = min_y + 8; y < max_y; y += 0x10) {
for (u16 x = min_x + 8; x < max_x; x += 0x10) {
+ // Do not process the pixel if it's inside the scissor box and the scissor mode is set to Exclude
+ if (regs.scissor_test.mode == Regs::ScissorMode::Exclude) {
+ if (x >= scissor_x1 && x < scissor_x2 &&
+ y >= scissor_y1 && y < scissor_y2)
+ continue;
+ }
+
// Calculate the barycentric coordinates w0, w1 and w2
int w0 = bias0 + SignedArea(vtxpos[1].xy(), vtxpos[2].xy(), {x, y});
int w1 = bias1 + SignedArea(vtxpos[2].xy(), vtxpos[0].xy(), {x, y});
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 328a4f66b..f8393c618 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -196,6 +196,14 @@ void RasterizerOpenGL::DrawTriangles() {
(GLint)(rect.bottom + regs.viewport_corner.y * color_surface->res_scale_height),
(GLsizei)(viewport_width * color_surface->res_scale_width), (GLsizei)(viewport_height * color_surface->res_scale_height));
+ if (uniform_block_data.data.framebuffer_scale[0] != color_surface->res_scale_width ||
+ uniform_block_data.data.framebuffer_scale[1] != color_surface->res_scale_height) {
+
+ uniform_block_data.data.framebuffer_scale[0] = color_surface->res_scale_width;
+ uniform_block_data.data.framebuffer_scale[1] = color_surface->res_scale_height;
+ uniform_block_data.dirty = true;
+ }
+
// Sync and bind the texture surfaces
const auto pica_textures = regs.GetTextures();
for (unsigned texture_index = 0; texture_index < pica_textures.size(); ++texture_index) {
@@ -353,6 +361,15 @@ void RasterizerOpenGL::NotifyPicaRegisterChanged(u32 id) {
SyncColorWriteMask();
break;
+ // Scissor test
+ case PICA_REG_INDEX(scissor_test.mode):
+ shader_dirty = true;
+ break;
+ case PICA_REG_INDEX(scissor_test.x1): // and y1
+ case PICA_REG_INDEX(scissor_test.x2): // and y2
+ SyncScissorTest();
+ break;
+
// Logic op
case PICA_REG_INDEX(output_merger.logic_op):
SyncLogicOp();
@@ -1002,6 +1019,7 @@ void RasterizerOpenGL::SetShader() {
SyncDepthOffset();
SyncAlphaTest();
SyncCombinerColor();
+ SyncScissorTest();
auto& tev_stages = Pica::g_state.regs.GetTevStages();
for (int index = 0; index < tev_stages.size(); ++index)
SyncTevConstColor(index, tev_stages[index]);
@@ -1166,6 +1184,22 @@ void RasterizerOpenGL::SyncDepthTest() {
PicaToGL::CompareFunc(regs.output_merger.depth_test_func) : GL_ALWAYS;
}
+void RasterizerOpenGL::SyncScissorTest() {
+ const auto& regs = Pica::g_state.regs;
+
+ if (uniform_block_data.data.scissor_x1 != regs.scissor_test.x1 ||
+ uniform_block_data.data.scissor_y1 != regs.scissor_test.y1 ||
+ uniform_block_data.data.scissor_x2 != regs.scissor_test.x2 ||
+ uniform_block_data.data.scissor_y2 != regs.scissor_test.y2) {
+
+ uniform_block_data.data.scissor_x1 = regs.scissor_test.x1;
+ uniform_block_data.data.scissor_y1 = regs.scissor_test.y1;
+ uniform_block_data.data.scissor_x2 = regs.scissor_test.x2;
+ uniform_block_data.data.scissor_y2 = regs.scissor_test.y2;
+ uniform_block_data.dirty = true;
+ }
+}
+
void RasterizerOpenGL::SyncCombinerColor() {
auto combiner_color = PicaToGL::ColorRGBA8(Pica::g_state.regs.tev_combiner_buffer_color.raw);
if (combiner_color != uniform_block_data.data.tev_combiner_buffer_color) {
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index 42482df4b..c5029432b 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -56,6 +56,8 @@ union PicaShaderConfig {
const auto& regs = Pica::g_state.regs;
+ state.scissor_test_mode = regs.scissor_test.mode;
+
state.depthmap_enable = regs.depthmap_enable;
state.alpha_test_func = regs.output_merger.alpha_test.enable ?
@@ -172,6 +174,7 @@ union PicaShaderConfig {
struct State {
Pica::Regs::CompareFunc alpha_test_func;
+ Pica::Regs::ScissorMode scissor_test_mode;
Pica::Regs::TextureConfig::TextureType texture0_type;
std::array<TevStageConfigRaw, 6> tev_stages;
u8 combiner_buffer_input;
@@ -325,9 +328,14 @@ private:
// the end of a uniform block is included in UNIFORM_BLOCK_DATA_SIZE or not.
// Not following that rule will cause problems on some AMD drivers.
struct UniformData {
+ alignas(8) GLvec2 framebuffer_scale;
GLint alphatest_ref;
GLfloat depth_scale;
GLfloat depth_offset;
+ GLint scissor_x1;
+ GLint scissor_y1;
+ GLint scissor_x2;
+ GLint scissor_y2;
alignas(16) GLvec3 fog_color;
alignas(16) GLvec3 lighting_global_ambient;
LightSrc light_src[8];
@@ -335,7 +343,7 @@ private:
alignas(16) GLvec4 tev_combiner_buffer_color;
};
- static_assert(sizeof(UniformData) == 0x3A0, "The size of the UniformData structure has changed, update the structure in the shader");
+ static_assert(sizeof(UniformData) == 0x3C0, "The size of the UniformData structure has changed, update the structure in the shader");
static_assert(sizeof(UniformData) < 16384, "UniformData structure must be less than 16kb as per the OpenGL spec");
/// Sets the OpenGL shader in accordance with the current PICA register state
@@ -384,6 +392,9 @@ private:
/// Syncs the depth test states to match the PICA register
void SyncDepthTest();
+ /// Syncs the scissor test state to match the PICA register
+ void SyncScissorTest();
+
/// Syncs the TEV combiner color buffer to match the PICA register
void SyncCombinerColor();
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp
index 3bace7f01..36513dedc 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp
@@ -539,6 +539,8 @@ in float texcoord0_w;
in vec4 normquat;
in vec3 view;
+in vec4 gl_FragCoord;
+
out vec4 color;
struct LightSrc {
@@ -552,9 +554,14 @@ struct LightSrc {
};
layout (std140) uniform shader_data {
+ vec2 framebuffer_scale;
int alphatest_ref;
float depth_scale;
float depth_offset;
+ int scissor_x1;
+ int scissor_y1;
+ int scissor_x2;
+ int scissor_y2;
vec3 fog_color;
vec3 lighting_global_ambient;
LightSrc light_src[NUM_LIGHTS];
@@ -582,6 +589,19 @@ vec4 secondary_fragment_color = vec4(0.0);
return out;
}
+ // Append the scissor test
+ if (state.scissor_test_mode != Regs::ScissorMode::Disabled) {
+ out += "if (";
+ // Negate the condition if we have to keep only the pixels outside the scissor box
+ if (state.scissor_test_mode == Regs::ScissorMode::Include)
+ out += "!";
+ // x2,y2 have +1 added to cover the entire pixel area
+ out += "(gl_FragCoord.x >= scissor_x1 * framebuffer_scale.x && "
+ "gl_FragCoord.y >= scissor_y1 * framebuffer_scale.y && "
+ "gl_FragCoord.x < (scissor_x2 + 1) * framebuffer_scale.x && "
+ "gl_FragCoord.y < (scissor_y2 + 1) * framebuffer_scale.y)) discard;\n";
+ }
+
out += "float z_over_w = 1.0 - gl_FragCoord.z * 2.0;\n";
out += "float depth = z_over_w * depth_scale + depth_offset;\n";
if (state.depthmap_enable == Pica::Regs::DepthBuffering::WBuffering) {
diff --git a/src/video_core/renderer_opengl/pica_to_gl.h b/src/video_core/renderer_opengl/pica_to_gl.h
index 6dc2758c5..d9b9c9cc2 100644
--- a/src/video_core/renderer_opengl/pica_to_gl.h
+++ b/src/video_core/renderer_opengl/pica_to_gl.h
@@ -17,6 +17,7 @@
#include "video_core/pica.h"
+using GLvec2 = std::array<GLfloat, 2>;
using GLvec3 = std::array<GLfloat, 3>;
using GLvec4 = std::array<GLfloat, 4>;
generated by cgit v1.2.3 (git 2.25.1) at 2024-07-06 16:19:57 +0200