X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=plugins%2Fgpu_unai%2Fgpulib_if.cpp;h=453cc7a6f9e65ec148cdc6efd48e54ebc338a67c;hb=4949d4fff530344f0f4a1676bc45d0203749f291;hp=710a7137353f8dd666c19778fa1bb91caed9e288;hpb=c89fac55014151734b4abb0d8a7e3747e5ce629a;p=pcsx_rearmed.git diff --git a/plugins/gpu_unai/gpulib_if.cpp b/plugins/gpu_unai/gpulib_if.cpp index 710a7137..453cc7a6 100644 --- a/plugins/gpu_unai/gpulib_if.cpp +++ b/plugins/gpu_unai/gpulib_if.cpp @@ -70,7 +70,7 @@ #define DOWNSCALE_VRAM_SIZE (1024 * 512 * 2 * 2 + 4096) -INLINE void scale_640_to_320(uint16_t *dest, const uint16_t *src, bool isRGB24) { +INLINE void scale_640_to_320(uint16_t *dest, const le16_t *src, bool isRGB24) { size_t uCount = 320; if(isRGB24) { @@ -84,17 +84,17 @@ INLINE void scale_640_to_320(uint16_t *dest, const uint16_t *src, bool isRGB24) src8 += 4; } while(--uCount); } else { - const uint16_t* src16 = src; + const le16_t *src16 = src; uint16_t* dst16 = dest; do { - *dst16++ = *src16; + *dst16++ = le16_to_u16(*src16); src16 += 2; } while(--uCount); } } -INLINE void scale_512_to_320(uint16_t *dest, const uint16_t *src, bool isRGB24) { +INLINE void scale_512_to_320(uint16_t *dest, const le16_t *src, bool isRGB24) { size_t uCount = 64; if(isRGB24) { @@ -122,17 +122,17 @@ INLINE void scale_512_to_320(uint16_t *dest, const uint16_t *src, bool isRGB24) src8 += 4; } while(--uCount); } else { - const uint16_t* src16 = src; + const le16_t* src16 = src; uint16_t* dst16 = dest; do { - *dst16++ = *src16++; - *dst16++ = *src16; + *dst16++ = le16_to_u16(*src16++); + *dst16++ = le16_to_u16(*src16); src16 += 2; - *dst16++ = *src16++; - *dst16++ = *src16; + *dst16++ = le16_to_u16(*src16++); + *dst16++ = le16_to_u16(*src16); src16 += 2; - *dst16++ = *src16; + *dst16++ = le16_to_u16(*src16); src16 += 2; } while(--uCount); } @@ -141,7 +141,7 @@ INLINE void scale_512_to_320(uint16_t *dest, const uint16_t *src, bool isRGB24) static uint16_t *get_downscale_buffer(int *x, int *y, int *w, int *h, int *vram_h) { uint16_t *dest = gpu_unai.downscale_vram; - const uint16_t *src = gpu_unai.vram; + const le16_t *src = gpu_unai.vram; bool isRGB24 = (gpu_unai.GPU_GP1 & 0x00200000 ? true : false); int stride = 1024, dstride = 1024, lines = *h, orig_w = *w; @@ -184,7 +184,12 @@ static uint16_t *get_downscale_buffer(int *x, int *y, int *w, int *h, int *vram_ size_t size = isRGB24 ? *w * 3 : *w * 2; do { - memcpy(dest + fb_offset_dest, src + fb_offset_src, size); +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + for (unsigned int i; i < size; i += 2) + dest[fb_offset_dest + i] = le16_to_u16(src[fb_offset_src + i]); +#else + memcpy(dest + fb_offset_dest, (u16 *)src + fb_offset_src, size); +#endif fb_offset_src = (fb_offset_src + stride) & fb_mask; fb_offset_dest = (fb_offset_dest + dstride) & fb_mask; } while(--lines); @@ -223,7 +228,7 @@ static void unmap_downscale_buffer(void) int renderer_init(void) { memset((void*)&gpu_unai, 0, sizeof(gpu_unai)); - gpu_unai.vram = (u16*)gpu.vram; + gpu_unai.vram = (le16_t *)gpu.vram; // Original standalone gpu_unai initialized TextureWindow[]. I added the // same behavior here, since it seems unsafe to leave [2],[3] unset when @@ -392,11 +397,12 @@ static void gpuGP0Cmd_0xEx(gpu_unai_t &gpu_unai, u32 cmd_word) extern const unsigned char cmd_lengths[256]; -int do_cmd_list(u32 *list, int list_len, int *last_cmd) +int do_cmd_list(u32 *_list, int list_len, int *last_cmd) { u32 cmd = 0, len, i; - u32 *list_start = list; - u32 *list_end = list + list_len; + le32_t *list = (le32_t *)_list; + le32_t *list_start = list; + le32_t *list_end = list + list_len; //TODO: set ilace_mask when resolution changes instead of every time, // eliminate #ifdef below. @@ -411,7 +417,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) for (; list < list_end; list += 1 + len) { - cmd = *list >> 24; + cmd = le32_to_u32(*list) >> 24; len = cmd_lengths[cmd]; if (list + 1 + len > list_end) { cmd = -1; @@ -447,8 +453,8 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x25: case 0x26: case 0x27: { // Textured 3-pt poly - gpuSetCLUT (gpu_unai.PacketBuffer.U4[2] >> 16); - gpuSetTexture(gpu_unai.PacketBuffer.U4[4] >> 16); + gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); + gpuSetTexture(le32_to_u32(gpu_unai.PacketBuffer.U4[4]) >> 16); u32 driver_idx = (gpu_unai.blit_mask?1024:0) | @@ -483,8 +489,8 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x2D: case 0x2E: case 0x2F: { // Textured 4-pt poly - gpuSetCLUT (gpu_unai.PacketBuffer.U4[2] >> 16); - gpuSetTexture(gpu_unai.PacketBuffer.U4[4] >> 16); + gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); + gpuSetTexture(le32_to_u32(gpu_unai.PacketBuffer.U4[4]) >> 16); u32 driver_idx = (gpu_unai.blit_mask?1024:0) | @@ -524,8 +530,8 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x35: case 0x36: case 0x37: { // Gouraud-shaded, textured 3-pt poly - gpuSetCLUT (gpu_unai.PacketBuffer.U4[2] >> 16); - gpuSetTexture (gpu_unai.PacketBuffer.U4[5] >> 16); + gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); + gpuSetTexture (le32_to_u32(gpu_unai.PacketBuffer.U4[5]) >> 16); PP driver = gpuPolySpanDrivers[ (gpu_unai.blit_mask?1024:0) | Dithering | @@ -553,8 +559,8 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x3D: case 0x3E: case 0x3F: { // Gouraud-shaded, textured 4-pt poly - gpuSetCLUT (gpu_unai.PacketBuffer.U4[2] >> 16); - gpuSetTexture (gpu_unai.PacketBuffer.U4[5] >> 16); + gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); + gpuSetTexture (le32_to_u32(gpu_unai.PacketBuffer.U4[5]) >> 16); PP driver = gpuPolySpanDrivers[ (gpu_unai.blit_mask?1024:0) | Dithering | @@ -576,7 +582,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x48 ... 0x4F: { // Monochrome line strip u32 num_vertexes = 1; - u32 *list_position = &(list[2]); + le32_t *list_position = &list[2]; // Shift index right by one, as untextured prims don't use lighting u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; @@ -594,7 +600,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) cmd = -1; goto breakloop; } - if((*list_position & 0xf000f000) == 0x50005000) + if((le32_raw(*list_position) & HTOLE32(0xf000f000)) == HTOLE32(0x50005000)) break; } @@ -615,7 +621,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x58 ... 0x5F: { // Gouraud-shaded line strip u32 num_vertexes = 1; - u32 *list_position = &(list[2]); + le32_t *list_position = &list[2]; // Shift index right by one, as untextured prims don't use lighting u32 driver_idx = (Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1; @@ -637,7 +643,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) cmd = -1; goto breakloop; } - if((*list_position & 0xf000f000) == 0x50005000) + if((le32_raw(*list_position) & HTOLE32(0xf000f000)) == HTOLE32(0x50005000)) break; } @@ -656,7 +662,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x65: case 0x66: case 0x67: { // Textured rectangle (variable size) - gpuSetCLUT (gpu_unai.PacketBuffer.U4[2] >> 16); + gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); //senquack - Only color 808080h-878787h allows skipping lighting calculation: @@ -673,7 +679,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) // alone, I don't want to slow rendering down too much. (TODO) //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) // Strip lower 3 bits of each color and determine if lighting should be used: - if ((gpu_unai.PacketBuffer.U4[0] & 0xF8F8F8) != 0x808080) + if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) driver_idx |= Lighting; PS driver = gpuSpriteSpanDrivers[driver_idx]; gpuDrawS(packet, driver); @@ -683,7 +689,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x69: case 0x6A: case 0x6B: { // Monochrome rectangle (1x1 dot) - gpu_unai.PacketBuffer.U4[2] = 0x00010001; + gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00010001); PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; gpuDrawT(packet, driver); } break; @@ -692,7 +698,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x71: case 0x72: case 0x73: { // Monochrome rectangle (8x8) - gpu_unai.PacketBuffer.U4[2] = 0x00080008; + gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00080008); PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; gpuDrawT(packet, driver); } break; @@ -701,14 +707,14 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x75: case 0x76: case 0x77: { // Textured rectangle (8x8) - gpu_unai.PacketBuffer.U4[3] = 0x00080008; - gpuSetCLUT (gpu_unai.PacketBuffer.U4[2] >> 16); + gpu_unai.PacketBuffer.U4[3] = u32_to_le32(0x00080008); + gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); //senquack - Only color 808080h-878787h allows skipping lighting calculation: //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) // Strip lower 3 bits of each color and determine if lighting should be used: - if ((gpu_unai.PacketBuffer.U4[0] & 0xF8F8F8) != 0x808080) + if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) driver_idx |= Lighting; PS driver = gpuSpriteSpanDrivers[driver_idx]; gpuDrawS(packet, driver); @@ -718,7 +724,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) case 0x79: case 0x7A: case 0x7B: { // Monochrome rectangle (16x16) - gpu_unai.PacketBuffer.U4[2] = 0x00100010; + gpu_unai.PacketBuffer.U4[2] = u32_to_le32(0x00100010); PT driver = gpuTileSpanDrivers[(Blending_Mode | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>3)) >> 1]; gpuDrawT(packet, driver); } break; @@ -728,7 +734,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) #ifdef __arm__ if ((gpu_unai.GPU_GP1 & 0x180) == 0 && (gpu_unai.Masking | gpu_unai.PixelMSB) == 0) { - gpuSetCLUT (gpu_unai.PacketBuffer.U4[2] >> 16); + gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); gpuDrawS16(packet); break; } @@ -736,13 +742,13 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) #endif case 0x7E: case 0x7F: { // Textured rectangle (16x16) - gpu_unai.PacketBuffer.U4[3] = 0x00100010; - gpuSetCLUT (gpu_unai.PacketBuffer.U4[2] >> 16); + gpu_unai.PacketBuffer.U4[3] = u32_to_le32(0x00100010); + gpuSetCLUT (le32_to_u32(gpu_unai.PacketBuffer.U4[2]) >> 16); u32 driver_idx = Blending_Mode | gpu_unai.TEXT_MODE | gpu_unai.Masking | Blending | (gpu_unai.PixelMSB>>1); //senquack - Only color 808080h-878787h allows skipping lighting calculation: //if ((gpu_unai.PacketBuffer.U1[0]>0x5F) && (gpu_unai.PacketBuffer.U1[1]>0x5F) && (gpu_unai.PacketBuffer.U1[2]>0x5F)) // Strip lower 3 bits of each color and determine if lighting should be used: - if ((gpu_unai.PacketBuffer.U4[0] & 0xF8F8F8) != 0x808080) + if ((le32_raw(gpu_unai.PacketBuffer.U4[0]) & HTOLE32(0xF8F8F8)) != HTOLE32(0x808080)) driver_idx |= Lighting; PS driver = gpuSpriteSpanDrivers[driver_idx]; gpuDrawS(packet, driver); @@ -771,7 +777,7 @@ int do_cmd_list(u32 *list, int list_len, int *last_cmd) goto breakloop; #endif case 0xE1 ... 0xE6: { // Draw settings - gpuGP0Cmd_0xEx(gpu_unai, gpu_unai.PacketBuffer.U4[0]); + gpuGP0Cmd_0xEx(gpu_unai, le32_to_u32(gpu_unai.PacketBuffer.U4[0])); } break; } } @@ -784,7 +790,7 @@ breakloop: return list - list_start; } -void renderer_sync_ecmds(uint32_t *ecmds) +void renderer_sync_ecmds(u32 *ecmds) { int dummy; do_cmd_list(&ecmds[1], 6, &dummy); @@ -806,7 +812,7 @@ void renderer_set_interlace(int enable, int is_odd) // Handle any gpulib settings applicable to gpu_unai: void renderer_set_config(const struct rearmed_cbs *cbs) { - gpu_unai.vram = (u16*)gpu.vram; + gpu_unai.vram = (le16_t *)gpu.vram; gpu_unai.config.ilace_force = cbs->gpu_unai.ilace_force; gpu_unai.config.pixel_skip = cbs->gpu_unai.pixel_skip; gpu_unai.config.lighting = cbs->gpu_unai.lighting;