X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?p=pcsx_rearmed.git;a=blobdiff_plain;f=plugins%2Fgpu_unai%2Fgpu.cpp;h=1552bed90af2c2a92cce7b7bb6e88e1a06649277;hp=991610cf9cf5cc6eb788d906f09b579e39368bf0;hb=ecd502e11f1d17998924f2de5909380b75c67d49;hpb=7c49c8a2a0cd00c453425b0f88c7b117223c92f4 diff --git a/plugins/gpu_unai/gpu.cpp b/plugins/gpu_unai/gpu.cpp index 991610cf..1552bed9 100644 --- a/plugins/gpu_unai/gpu.cpp +++ b/plugins/gpu_unai/gpu.cpp @@ -24,23 +24,29 @@ #include "debug.h" int skipCount = 2; /* frame skip (0,1,2,3...) */ -int skCount=0; /* internal frame skip */ +int skCount = 0; /* internal frame skip */ int linesInterlace = 0; /* internal lines interlace */ int linesInterlace_user = 0; /* Lines interlace */ -bool isSkip=false; /* skip frame (info coming from GPU) */ -bool skipFrame=false; /* skip frame (according to frame skip) */ +bool isSkip = false; /* skip frame (info coming from GPU) */ +bool wasSkip = false; +bool skipFrame = false; /* skip frame (according to frame skip) */ bool alt_fps = false; /* Alternative FPS algorithm */ bool show_fps = false; /* Show FPS statistics */ -bool isPAL=false; /* PAL video timing */ +bool isPAL = false; /* PAL video timing */ bool progressInterlace_flag = false; /* Progressive interlace flag */ bool progressInterlace = false; /* Progressive interlace option*/ bool frameLimit = false; /* frames to wait */ + bool light = true; /* lighting */ bool blend = true; /* blending */ +bool FrameToRead = false; /* load image in progress */ +bool FrameToWrite = false; /* store image in progress */ +bool fb_dirty = false; bool enableAbbeyHack = false; /* Abe's Odyssey hack */ + u8 BLEND_MODE; u8 TEXT_MODE; u8 Masking; @@ -58,9 +64,6 @@ s32 px,py; s32 x_end,y_end; u16* pvram; -u32 FrameToRead; -u32 FrameToWrite; - u32 GP0; s32 PacketCount; s32 PacketIndex; @@ -92,7 +95,8 @@ u32 lInc; u32 tInc, tMsk; GPUPacket PacketBuffer; -u16 GPU_FrameBuffer[FRAME_BUFFER_SIZE/2] __attribute__((aligned(16))); // FRAME_BUFFER_SIZE is defined in bytes +// FRAME_BUFFER_SIZE is defined in bytes; 512K is guard memory for out of range reads +u16 GPU_FrameBuffer[(FRAME_BUFFER_SIZE+512*1024)/2] __attribute__((aligned(2048))); u32 GPU_GP1; /////////////////////////////////////////////////////////////////////////////// @@ -103,12 +107,10 @@ u32 GPU_GP1; // GPU Raster Macros #define GPU_RGB16(rgb) ((((rgb)&0xF80000)>>9)|(((rgb)&0xF800)>>6)|(((rgb)&0xF8)>>3)) -#define GPU_EXPANDSIGN_POLY(x) (((s32)(x)<<20)>>20) -//#define GPU_EXPANDSIGN_POLY(x) (((s32)(x)<<21)>>21) -#define GPU_EXPANDSIGN_SPRT(x) (((s32)(x)<<21)>>21) +#define GPU_EXPANDSIGN(x) (((s32)(x)<<21)>>21) -//#define GPU_TESTRANGE(x) { if((u32)(x+1024) > 2047) return; } -#define GPU_TESTRANGE(x) { if ((x<-1023) || (x>1023)) return; } +#define CHKMAX_X 1024 +#define CHKMAX_Y 512 #define GPU_SWAP(a,b,t) {(t)=(a);(a)=(b);(b)=(t);} @@ -180,6 +182,14 @@ long GPU_freeze(unsigned int bWrite, GPUFreeze_t* p2) if (bWrite) { p2->GPU_gp1 = GPU_GP1; + memset(p2->Control, 0, sizeof(p2->Control)); + // save resolution and registers for P.E.Op.S. compatibility + p2->Control[3] = (3 << 24) | ((GPU_GP1 >> 23) & 1); + p2->Control[4] = (4 << 24) | ((GPU_GP1 >> 29) & 3); + p2->Control[5] = (5 << 24) | (DisplayArea[0] | (DisplayArea[1] << 10)); + p2->Control[6] = (6 << 24) | (2560 << 12); + p2->Control[7] = (7 << 24) | (DisplayArea[4] | (DisplayArea[5] << 10)); + p2->Control[8] = (8 << 24) | ((GPU_GP1 >> 17) & 0x3f) | ((GPU_GP1 >> 10) & 0x40); memcpy(p2->FrameBuffer, (u16*)GPU_FrameBuffer, FRAME_BUFFER_SIZE); return (1); } @@ -187,6 +197,10 @@ long GPU_freeze(unsigned int bWrite, GPUFreeze_t* p2) { GPU_GP1 = p2->GPU_gp1; memcpy((u16*)GPU_FrameBuffer, p2->FrameBuffer, FRAME_BUFFER_SIZE); + GPU_writeStatus((5 << 24) | p2->Control[5]); + GPU_writeStatus((7 << 24) | p2->Control[7]); + GPU_writeStatus((8 << 24) | p2->Control[8]); + gpuSetTexture(GPU_GP1); return (1); } return (0); @@ -258,8 +272,9 @@ void GPU_writeDataMem(u32* dmaAddress, s32 dmaCount) { if (FrameToWrite) { - while (dmaCount--) + while (dmaCount) { + dmaCount--; data = *dmaAddress++; if ((&pvram[px])>(VIDEO_END)) pvram-=512*1024; pvram[px] = data; @@ -269,7 +284,7 @@ void GPU_writeDataMem(u32* dmaAddress, s32 dmaCount) pvram += 1024; if (++py>=y_end) { - FrameToWrite = 0; + FrameToWrite = false; GPU_GP1 &= ~0x08000000; break; } @@ -282,7 +297,7 @@ void GPU_writeDataMem(u32* dmaAddress, s32 dmaCount) pvram += 1024; if (++py>=y_end) { - FrameToWrite = 0; + FrameToWrite = false; GPU_GP1 &= ~0x08000000; break; } @@ -298,6 +313,7 @@ void GPU_writeDataMem(u32* dmaAddress, s32 dmaCount) } GPU_GP1 = (GPU_GP1 | 0x14000000) & ~0x60000000; + fb_dirty = true; pcsx4all_prof_end_with_resume(PCSX4ALL_PROF_GPU,PCSX4ALL_PROF_HW_WRITE); pcsx4all_prof_resume(PCSX4ALL_PROF_CPU); } @@ -315,7 +331,7 @@ INLINE int CheckForEndlessLoop(u32 *laddr) } /////////////////////////////////////////////////////////////////////////////// -void GPU_dmaChain(u32* baseAddr, u32 dmaVAddr) +long GPU_dmaChain(u32* baseAddr, u32 dmaVAddr) { #ifdef DEBUG_ANALYSIS dbg_anacnt_GPU_dmaChain++; @@ -323,6 +339,7 @@ void GPU_dmaChain(u32* baseAddr, u32 dmaVAddr) pcsx4all_prof_start_with_pause(PCSX4ALL_PROF_GPU,PCSX4ALL_PROF_HW_WRITE); u32 data, *address, count, offset; unsigned int DMACommandCounter = 0; + long dma_words = 0; GPU_GP1 &= ~0x14000000; lUsedAddr[0]=lUsedAddr[1]=lUsedAddr[2]=(u32*)0x1fffff; @@ -339,9 +356,12 @@ void GPU_dmaChain(u32* baseAddr, u32 dmaVAddr) else dmaVAddr = 0x1FFFFF; if(count>0) GPU_writeDataMem(address,count); + dma_words += 1 + count; } GPU_GP1 = (GPU_GP1 | 0x14000000) & ~0x60000000; pcsx4all_prof_end_with_resume(PCSX4ALL_PROF_GPU,PCSX4ALL_PROF_HW_WRITE); + + return dma_words; } /////////////////////////////////////////////////////////////////////////////// @@ -365,7 +385,7 @@ void GPU_writeData(u32 data) pvram += 1024; if (++py>=y_end) { - FrameToWrite = 0; + FrameToWrite = false; GPU_GP1 &= ~0x08000000; } } @@ -379,7 +399,7 @@ void GPU_writeData(u32 data) pvram += 1024; if (++py>=y_end) { - FrameToWrite = 0; + FrameToWrite = false; GPU_GP1 &= ~0x08000000; } } @@ -390,6 +410,7 @@ void GPU_writeData(u32 data) gpuCheckPacket(data); } GPU_GP1 |= 0x14000000; + fb_dirty = true; pcsx4all_prof_end_with_resume(PCSX4ALL_PROF_GPU,PCSX4ALL_PROF_HW_WRITE); pcsx4all_prof_resume(PCSX4ALL_PROF_CPU); @@ -411,7 +432,7 @@ void GPU_readDataMem(u32* dmaAddress, s32 dmaCount) { if ((&pvram[px])>(VIDEO_END)) pvram-=512*1024; // lower 16 bit - u32 data = (unsigned long)pvram[px]; + u32 data = pvram[px]; if (++px>=x_end) { @@ -421,7 +442,7 @@ void GPU_readDataMem(u32* dmaAddress, s32 dmaCount) if ((&pvram[px])>(VIDEO_END)) pvram-=512*1024; // higher 16 bit (always, even if it's an odd width) - data |= (unsigned long)(pvram[px])<<16; + data |= (u32)(pvram[px])<<16; *dmaAddress++ = data; @@ -431,7 +452,7 @@ void GPU_readDataMem(u32* dmaAddress, s32 dmaCount) pvram += 1024; if (++py>=y_end) { - FrameToRead = 0; + FrameToRead = false; GPU_GP1 &= ~0x08000000; break; } @@ -464,7 +485,7 @@ u32 GPU_readData(void) pvram += 1024; if (++py>=y_end) { - FrameToRead = 0; + FrameToRead = false; GPU_GP1 &= ~0x08000000; } } @@ -476,7 +497,7 @@ u32 GPU_readData(void) pvram +=1024; if (++py>=y_end) { - FrameToRead = 0; + FrameToRead = false; GPU_GP1 &= ~0x08000000; } } @@ -512,11 +533,11 @@ void GPU_writeStatus(u32 data) break; case 0x01: GPU_GP1 &= ~0x08000000; - PacketCount = FrameToRead = FrameToWrite = 0; + PacketCount = 0; FrameToRead = FrameToWrite = false; break; case 0x02: GPU_GP1 &= ~0x08000000; - PacketCount = FrameToRead = FrameToWrite = 0; + PacketCount = 0; FrameToRead = FrameToWrite = false; break; case 0x03: GPU_GP1 = (GPU_GP1 & ~0x00800000) | ((data & 1) << 23); @@ -529,10 +550,17 @@ void GPU_writeStatus(u32 data) case 0x05: DisplayArea[0] = (data & 0x000003FF); //(short)(data & 0x3ff); DisplayArea[1] = ((data & 0x0007FC00)>>10); //(data & 0x000FFC00) >> 10; //(short)((data>>10)&0x1ff); + fb_dirty = true; + wasSkip = isSkip; + if (isSkip) + isSkip = false; + else + isSkip = skipFrame; break; case 0x07: DisplayArea[4] = data & 0x000003FF; //(short)(data & 0x3ff); DisplayArea[5] = (data & 0x000FFC00) >> 10; //(short)((data>>10) & 0x3ff); + fb_dirty = true; break; case 0x08: { @@ -543,6 +571,7 @@ void GPU_writeStatus(u32 data) DisplayArea[3] = VerticalResolution[(GPU_GP1 >> 19) & 3]; isPAL = (data & 0x08) ? true : false; // if 1 - PAL mode, else NTSC } + fb_dirty = true; break; case 0x10: switch (data & 0xffff) { @@ -614,15 +643,16 @@ INLINE void gpuVideoOutput(void) int incY = (h0==480) ? 2 : 1; h0=(h0==480 ? 2048 : 1024); - if (!progressInterlace) { const int li=linesInterlace; + bool pi=progressInterlace; + bool pif=progressInterlace_flag; switch ( w0 ) { case 256: for(int y1=y0+h1; y0=1000000) @@ -837,6 +811,7 @@ void GPU_updateLace(void) isSkip=true; } } + fb_dirty=false; pcsx4all_prof_end_with_resume(PCSX4ALL_PROF_GPU,PCSX4ALL_PROF_COUNTERS); } @@ -847,22 +822,18 @@ void GPU_updateLace(void) extern "C" { -extern void bgr555_to_rgb565(void *dst, void *src, int bytes); -extern void bgr888_to_rgb888(void *dst, void *src, int bytes); static const struct rearmed_cbs *cbs; -static void *screen_buf; +static s16 old_res_horz, old_res_vert, old_rgb24; static void blit(void) { - static s16 old_res_horz, old_res_vert, old_rgb24; + u16 *base = (u16 *)GPU_FrameBuffer; s16 isRGB24 = (GPU_GP1 & 0x00200000) ? 1 : 0; s16 h0, x0, y0, w0, h1; - u8 *dest = (u8 *)screen_buf; - u16 *srcs; - x0 = DisplayArea[0] & ~3; // alignment needed by blitter + x0 = DisplayArea[0] & ~1; // alignment needed by blitter y0 = DisplayArea[1]; - srcs = &((u16*)GPU_FrameBuffer)[FRAME_OFFSET(x0,y0)]; + base += FRAME_OFFSET(x0, y0); w0 = DisplayArea[2]; h0 = DisplayArea[3]; // video mode @@ -870,30 +841,18 @@ static void blit(void) h1 = DisplayArea[5] - DisplayArea[4]; // display needed if (h0 == 480) h1 = Min2(h1*2,480); + if (h1 <= 0) + return; + if (w0 != old_res_horz || h1 != old_res_vert || isRGB24 != old_rgb24) { old_res_horz = w0; old_res_vert = h1; old_rgb24 = (s16)isRGB24; - cbs->pl_fbdev_set_mode(w0, h1, isRGB24 ? 24 : 16); - } - - if (isRGB24) - { - for (; h1-- > 0; dest += w0 * 3, srcs += 1024) - { - bgr888_to_rgb888(dest, srcs, w0 * 3); - } - } - else - { - for (; h1-- > 0; dest += w0 * 2, srcs += 1024) - { - bgr555_to_rgb565(dest, srcs, w0 * 2); - } + cbs->pl_vout_set_mode(w0, h1, w0, h1, isRGB24 ? 24 : 16); } - screen_buf = cbs->pl_fbdev_flip(); + cbs->pl_vout_flip(base, 1024, isRGB24, w0, h1); } void GPU_updateLace(void) @@ -901,20 +860,32 @@ void GPU_updateLace(void) // Interlace bit toggle GPU_GP1 ^= 0x80000000; - if (!((GPU_GP1&0x08000000) || (GPU_GP1&0x00800000))) + if (!fb_dirty || (GPU_GP1&0x08800000)) + return; + + if (!wasSkip) { blit(); + fb_dirty = false; + skCount = 0; + } + else { + skCount++; + if (skCount >= 8) + wasSkip = isSkip = 0; + } + + skipFrame = cbs->fskip_advice || cbs->frameskip == 1; } long GPUopen(unsigned long *, char *, char *) { - cbs->pl_fbdev_open(); - screen_buf = cbs->pl_fbdev_flip(); + cbs->pl_vout_open(); return 0; } long GPUclose(void) { - cbs->pl_fbdev_close(); + cbs->pl_vout_close(); return 0; } @@ -928,7 +899,15 @@ long GPUfreeze(unsigned int ulGetFreezeData, GPUFreeze_t* p2) void GPUrearmedCallbacks(const struct rearmed_cbs *cbs_) { + enableAbbeyHack = cbs_->gpu_unai.abe_hack; + light = !cbs_->gpu_unai.no_light; + blend = !cbs_->gpu_unai.no_blend; + if (cbs_->pl_vout_set_raw_vram) + cbs_->pl_vout_set_raw_vram((void *)GPU_FrameBuffer); + cbs = cbs_; + if (cbs->pl_set_gpu_caps) + cbs->pl_set_gpu_caps(0); } } /* extern "C" */