render: wrap U coord on texture fetch overflow

render_line: error calculation now mimic HW

Author: JaCzekanski

src/device/gpu/render/render_line.cpp

@@ -40,7 +40,7 @@ void Render::drawLine(gpu::GPU* gpu, const primitive::Line& line) {
     int dx = x1 - x0;
     int dy = y1 - y0;
     int derror = std::abs(dy) * 2;
-    int error = 0;
+    int error = !steep;
     int _y = y0;
 
     float length = sqrtf(powf(x1 - x0, 2) + powf(y1 - y0, 2));

src/device/gpu/render/texture_utils.h

@@ -9,18 +9,20 @@ enum class ColorDepth { NONE, BIT_4, BIT_8, BIT_16 };
 namespace {
 // Using unsigned vectors allows compiler to generate slightly faster division code
 INLINE uint16_t tex4bit(gpu::GPU* gpu, glm::uvec2 tex, glm::uvec2 texPage, glm::uvec2 clut) {
-    uint16_t index = gpuVRAM[texPage.y + tex.y][texPage.x + tex.x / 4];
+    uint16_t index = gpuVRAM[texPage.y + tex.y][(texPage.x + tex.x / 4) & 1023];
     uint8_t entry = (index >> ((tex.x & 3) * 4)) & 0xf;
     return gpuVRAM[clut.y][clut.x + entry];
 }
 
 INLINE uint16_t tex8bit(gpu::GPU* gpu, glm::uvec2 tex, glm::uvec2 texPage, glm::uvec2 clut) {
-    uint16_t index = gpuVRAM[texPage.y + tex.y][texPage.x + tex.x / 2];
+    uint16_t index = gpuVRAM[texPage.y + tex.y][(texPage.x + tex.x / 2) & 1023];
     uint8_t entry = (index >> ((tex.x & 1) * 8)) & 0xff;
     return gpuVRAM[clut.y][clut.x + entry];
 }
 
-INLINE uint16_t tex16bit(gpu::GPU* gpu, glm::uvec2 tex, glm::uvec2 texPage) { return gpuVRAM[texPage.y + tex.y][texPage.x + tex.x]; }
+INLINE uint16_t tex16bit(gpu::GPU* gpu, glm::uvec2 tex, glm::uvec2 texPage) {
+    return gpuVRAM[texPage.y + tex.y][(texPage.x + tex.x) & 1023];
+}
 
 template <ColorDepth bits>
 INLINE PSXColor fetchTex(gpu::GPU* gpu, glm::uvec2 texel, const glm::ivec2 texPage, const glm::ivec2 clut) {