// Copyright (C) 2003-2008 Dolphin Project. // This program is free software: you can redistribute it and/or modify // it under the terms of the GNU General Public License as published by // the Free Software Foundation, version 2.0. // This program is distributed in the hope that it will be useful, // but WITHOUT ANY WARRANTY; without even the implied warranty of // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the // GNU General Public License 2.0 for more details. // A copy of the GPL 2.0 should have been included with the program. // If not, see http://www.gnu.org/licenses/ // Official SVN repository and contact information can be found at // http://code.google.com/p/dolphin-emu/ // --------------------------------------------------------------------------------------- // includes // ------------- #include "Globals.h" #include "Profiler.h" #include "Config.h" #include "Statistics.h" #include "VertexLoader.h" #include "VertexManager.h" #include "BPStructs.h" #include "Render.h" #include "OpcodeDecoding.h" #include "TextureMngr.h" #include "TextureDecoder.h" #include "TextureConverter.h" #include "VertexShaderManager.h" #include "PixelShaderManager.h" #include "XFB.h" #include "main.h" // --------------------------------------------------------------------------------------- // State translation lookup tables // ------------- static const GLenum glCmpFuncs[8] = { GL_NEVER, GL_LESS, GL_EQUAL, GL_LEQUAL, GL_GREATER, GL_NOTEQUAL, GL_GEQUAL, GL_ALWAYS }; static const GLenum glLogicOpCodes[16] = { GL_CLEAR, GL_AND, GL_AND_REVERSE, GL_COPY, GL_AND_INVERTED, GL_NOOP, GL_XOR, GL_OR, GL_NOR, GL_EQUIV, GL_INVERT, GL_OR_REVERSE, GL_COPY_INVERTED, GL_OR_INVERTED, GL_NAND, GL_SET }; void BPInit() { memset(&bpmem, 0, sizeof(bpmem)); bpmem.bpMask = 0xFFFFFF; } ////////////////////////////////////////////////////////////////////////////////////// // Write to bpmem /* ------------------ Called: At the end of every: OpcodeDecoding.cpp ExecuteDisplayList > Decode() > LoadBPReg TODO: Turn into function table. The (future) DL jit can then call the functions directly, getting rid of dynamic dispatch. Unfortunately, few games use DLs properly - most\ just stuff geometry in them and don't put state changes there. // ------------------ */ void BPWritten(int addr, int changes, int newval) { switch (addr) { case BPMEM_GENMODE: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PRIM_LOG("genmode: texgen=%d, col=%d, ms_en=%d, tev=%d, culmode=%d, ind=%d, zfeeze=%d\n", bpmem.genMode.numtexgens, bpmem.genMode.numcolchans, bpmem.genMode.ms_en, bpmem.genMode.numtevstages+1, bpmem.genMode.cullmode, bpmem.genMode.numindstages, bpmem.genMode.zfreeze); // none, ccw, cw, ccw if (bpmem.genMode.cullmode > 0) { glEnable(GL_CULL_FACE); glFrontFace(bpmem.genMode.cullmode == 2 ? GL_CCW : GL_CW); } else glDisable(GL_CULL_FACE); PixelShaderManager::SetGenModeChanged(); } break; case BPMEM_IND_MTX+0: case BPMEM_IND_MTX+1: case BPMEM_IND_MTX+2: case BPMEM_IND_MTX+3: case BPMEM_IND_MTX+4: case BPMEM_IND_MTX+5: case BPMEM_IND_MTX+6: case BPMEM_IND_MTX+7: case BPMEM_IND_MTX+8: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetIndMatrixChanged((addr-BPMEM_IND_MTX)/3); } break; case BPMEM_RAS1_SS0: case BPMEM_RAS1_SS1: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetIndTexScaleChanged(); } break; case BPMEM_ZMODE: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PRIM_LOG("zmode: test=%d, func=%d, upd=%d\n", bpmem.zmode.testenable, bpmem.zmode.func, bpmem.zmode.updateenable); if (bpmem.zmode.testenable) { glEnable(GL_DEPTH_TEST); glDepthMask(bpmem.zmode.updateenable?GL_TRUE:GL_FALSE); glDepthFunc(glCmpFuncs[bpmem.zmode.func]); } else { // if the test is disabled write is disabled too glDisable(GL_DEPTH_TEST); glDepthMask(GL_FALSE); } if (!bpmem.zmode.updateenable) Renderer::SetRenderMode(Renderer::RM_Normal); } break; case BPMEM_ALPHACOMPARE: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PRIM_LOG("alphacmp: ref0=%d, ref1=%d, comp0=%d, comp1=%d, logic=%d\n", bpmem.alphaFunc.ref0, bpmem.alphaFunc.ref1, bpmem.alphaFunc.comp0, bpmem.alphaFunc.comp1, bpmem.alphaFunc.logic); PixelShaderManager::SetAlpha(bpmem.alphaFunc); } break; case BPMEM_CONSTANTALPHA: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PRIM_LOG("constalpha: alp=%d, en=%d\n", bpmem.dstalpha.alpha, bpmem.dstalpha.enable); SETSTAT(stats.dstAlphaEnable, bpmem.dstalpha.enable); SETSTAT_UINT(stats.dstAlpha, bpmem.dstalpha.alpha); PixelShaderManager::SetDestAlpha(bpmem.dstalpha); } break; case BPMEM_LINEPTWIDTH: { float fratio = xfregs.rawViewport[0] != 0 ? ((float)Renderer::GetTargetWidth() / EFB_WIDTH) : 1.0f; if (bpmem.lineptwidth.linesize > 0) glLineWidth((float)bpmem.lineptwidth.linesize * fratio / 6.0f); // scale by ratio of widths if (bpmem.lineptwidth.pointsize > 0) glPointSize((float)bpmem.lineptwidth.pointsize * fratio / 6.0f); break; } case BPMEM_PE_CONTROL: // GXSetZCompLoc, GXPixModeSync if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; } break; case BPMEM_BLENDMODE: if (changes & 0xFFFF) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PRIM_LOG("blendmode: en=%d, open=%d, colupd=%d, alphaupd=%d, dst=%d, src=%d, sub=%d, mode=%d\n", bpmem.blendmode.blendenable, bpmem.blendmode.logicopenable, bpmem.blendmode.colorupdate, bpmem.blendmode.alphaupdate, bpmem.blendmode.dstfactor, bpmem.blendmode.srcfactor, bpmem.blendmode.subtract, bpmem.blendmode.logicmode); /* Logic Operation Blend Modes -------------------- 0: GL_CLEAR 1: GL_AND 2: GL_AND_REVERSE 3: GL_COPY [Super Smash. Bro. Melee, NES Zelda I, NES Zelda II] 4: GL_AND_INVERTED 5: GL_NOOP 6: GL_XOR 7: GL_OR [Zelda: TP] 8: GL_NOR 9: GL_EQUIV 10: GL_INVERT 11: GL_OR_REVERSE 12: GL_COPY_INVERTED 13: GL_OR_INVERTED 14: GL_NAND 15: GL_SET */ // LogicOp Blending if (changes & 2) { SETSTAT(stats.logicOpMode, bpmem.blendmode.logicopenable != 0 ? bpmem.blendmode.logicmode : stats.logicOpMode); if (bpmem.blendmode.logicopenable) { glEnable(GL_COLOR_LOGIC_OP); // PanicAlert("Logic Op Blend : %i", bpmem.blendmode.logicmode); glLogicOp(glLogicOpCodes[bpmem.blendmode.logicmode]); } else glDisable(GL_COLOR_LOGIC_OP); } // Dithering if (changes & 4) { SETSTAT(stats.dither, bpmem.blendmode.dither); if (bpmem.blendmode.dither) glEnable(GL_DITHER); else glDisable(GL_DITHER); } // Blending if (changes & 0xFE1) { SETSTAT(stats.srcFactor, bpmem.blendmode.srcfactor); SETSTAT(stats.dstFactor, bpmem.blendmode.dstfactor); Renderer::SetBlendMode(false); } // Color Mask if (changes & 0x18) { SETSTAT(stats.alphaUpdate, bpmem.blendmode.alphaupdate); SETSTAT(stats.colorUpdate, bpmem.blendmode.colorupdate); Renderer::SetColorMask(); } } break; case BPMEM_FOGRANGE: case BPMEM_FOGPARAM0: case BPMEM_FOGBEXPONENT: case BPMEM_FOGBMAGNITUDE: case BPMEM_FOGPARAM3: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetFogParamChanged(); } break; case BPMEM_FOGCOLOR: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetFogColorChanged(); } break; case BPMEM_TEXINVALIDATE: break; case BPMEM_SCISSOROFFSET: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; Renderer::SetScissorRect(); } break; case BPMEM_SCISSORTL: case BPMEM_SCISSORBR: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; if (!Renderer::SetScissorRect()) { if (addr == BPMEM_SCISSORBR) { ERROR_LOG(VIDEO, "bad scissor!\n"); } } } break; case BPMEM_ZTEX1: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PRIM_LOG("ztex bias=0x%x\n", bpmem.ztex1.bias); PixelShaderManager::SetZTextureBias(bpmem.ztex1.bias); } break; case BPMEM_ZTEX2: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; if (changes & 3) { PixelShaderManager::SetZTextureTypeChanged(); } #if defined(_DEBUG) || defined(DEBUGFAST) const char* pzop[] = {"DISABLE", "ADD", "REPLACE", "?"}; const char* pztype[] = {"Z8", "Z16", "Z24", "?"}; PRIM_LOG("ztex op=%s, type=%s\n", pzop[bpmem.ztex2.op], pztype[bpmem.ztex2.type]); #endif } break; case 0xf6: // ksel0 case 0xf7: // ksel1 case 0xf8: // ksel2 case 0xf9: // ksel3 case 0xfa: // ksel4 case 0xfb: // ksel5 case 0xfc: // ksel6 case 0xfd: // ksel7 if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetTevKSelChanged(addr-0xf6); } break; case BPMEM_SETDRAWDONE: VertexManager::Flush(); switch (newval & 0xFF) { case 0x02: g_VideoInitialize.pSetPEFinish(); // may generate interrupt DEBUG_LOG(VIDEO, "GXSetDrawDone SetPEFinish (value: 0x%02X)", (newval & 0xFFFF)); break; default: DEBUG_LOG(VIDEO, "GXSetDrawDone ??? (value 0x%02X)", (newval & 0xFFFF)); break; } break; case BPMEM_PE_TOKEN_ID: g_VideoInitialize.pSetPEToken(static_cast(newval & 0xFFFF), FALSE); DEBUG_LOG(VIDEO, "SetPEToken 0x%04x", (newval & 0xFFFF)); break; case BPMEM_PE_TOKEN_INT_ID: g_VideoInitialize.pSetPEToken(static_cast(newval & 0xFFFF), TRUE); DEBUG_LOG(VIDEO, "SetPEToken + INT 0x%04x", (newval & 0xFFFF)); break; case BPMEM_SETGPMETRIC: // Set gp metric? break; // ------------------------ // EFB copy command. This copies a rectangle from the EFB to either RAM in a texture format or to XFB as YUYV. // It can also optionally clear the EFB while copying from it. To emulate this, we of course copy first and clear afterwards. case BPMEM_TRIGGER_EFB_COPY: { DVSTARTSUBPROFILE("LoadBPReg:swap"); VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; // The bottom right is within the rectangle // The values in bpmem.copyTexSrcXY and bpmem.copyTexSrcWH are updated in case 0x49 and 0x4a in this function TRectangle rc = { (int)(bpmem.copyTexSrcXY.x), (int)(bpmem.copyTexSrcXY.y), (int)((bpmem.copyTexSrcXY.x + bpmem.copyTexSrcWH.x + 1)), (int)((bpmem.copyTexSrcXY.y + bpmem.copyTexSrcWH.y + 1)) }; float MValueX = Renderer::GetTargetScaleX(); float MValueY = Renderer::GetTargetScaleY(); // Need another rc here to get it to scale. // Here the bottom right is the out of the rectangle. TRectangle multirc = { (int)(bpmem.copyTexSrcXY.x * MValueX), (int)(bpmem.copyTexSrcXY.y * MValueY), (int)((bpmem.copyTexSrcXY.x * MValueX + (bpmem.copyTexSrcWH.x + 1) * MValueX)), (int)((bpmem.copyTexSrcXY.y * MValueY + (bpmem.copyTexSrcWH.y + 1) * MValueY)) }; UPE_Copy PE_copy; PE_copy.Hex = bpmem.triggerEFBCopy; // -------------------------------------------------------- // Check to where we should copy the image data from the EFB. // -------------------------- if (PE_copy.copy_to_xfb == 0) { if (g_Config.bShowEFBCopyRegions) stats.efb_regions.push_back(rc); // EFB to texture // for some reason it sets bpmem.zcontrol.pixel_format to PIXELFMT_Z24 every time a zbuffer format is given as a dest to GXSetTexCopyDst if (!g_Config.bEFBCopyDisable) { if (g_Config.bCopyEFBToRAM) { TextureConverter::EncodeToRam(bpmem.copyTexDest << 5, bpmem.zcontrol.pixel_format == PIXELFMT_Z24, PE_copy.intensity_fmt > 0, (PE_copy.target_pixel_format / 2) + ((PE_copy.target_pixel_format & 1) * 8), // ?? PE_copy.half_scale > 0, rc); } else { TextureMngr::CopyRenderTargetToTexture(bpmem.copyTexDest << 5, bpmem.zcontrol.pixel_format == PIXELFMT_Z24, PE_copy.intensity_fmt > 0, (PE_copy.target_pixel_format / 2) + ((PE_copy.target_pixel_format & 1) * 8), // ?? PE_copy.half_scale > 0, rc); } } } else { // EFB to XFB if (g_Config.bUseXFB) { // the number of lines copied is determined by the y scale * source efb height float yScale = bpmem.dispcopyyscale / 256.0f; float xfbLines = bpmem.copyTexSrcWH.y + 1.0f * yScale; u8* pXFB = Memory_GetPtr(bpmem.copyTexDest << 5); u32 dstWidth = (bpmem.copyMipMapStrideChannels << 4); u32 dstHeight = (u32)xfbLines; XFB_Write(pXFB, multirc, dstWidth, dstHeight); // FIXME: we draw XFB from here in DC mode. // Bad hack since we can have multiple EFB to XFB copy before a draw. // Plus we should use width and height from VI regs (see VI->Update()). // Dixit donkopunchstania for the info. //DebugLog("(EFB to XFB->XFB_Draw): ptr: %08x | %ix%i", (u32)pXFB, dstWidth, dstHeight); if (g_VideoInitialize.bUseDualCore) XFB_Draw(pXFB, dstWidth, dstHeight, 0); } else { // Hm, we need to compensate for the fact that the copy may be bigger than what is displayed. // Seen in Spartan Warrior. Not sure how to deal with it yet. Renderer::Swap(multirc); } g_VideoInitialize.pCopiedToXFB(); } // -------------------------------------------------------- // Clear the picture after it's done and submitted, to prepare for the next picture // -------------------------- if (PE_copy.clear) { // Clear color Renderer::SetRenderMode(Renderer::RM_Normal); // Clear Z-Buffer target bool bRestoreZBufferTarget = Renderer::UseFakeZTarget(); // Update the view port for clearing the picture glViewport(0, 0, Renderer::GetTargetWidth(), Renderer::GetTargetHeight()); // Always set the scissor in case it was set by the game and has not been reset glScissor(multirc.left, (Renderer::GetTargetHeight() - multirc.bottom), (multirc.right - multirc.left), (multirc.bottom - multirc.top)); // --------------------------- VertexShaderManager::SetViewportChanged(); // Since clear operations use the source rectangle, we have to do // regular renders (glClear clears the entire buffer) if (bpmem.blendmode.colorupdate || bpmem.blendmode.alphaupdate || bpmem.zmode.updateenable) { GLbitfield bits = 0; if (bpmem.blendmode.colorupdate || bpmem.blendmode.alphaupdate) { u32 clearColor = (bpmem.clearcolorAR << 16) | bpmem.clearcolorGB; glClearColor(((clearColor>>16) & 0xff)*(1/255.0f), ((clearColor>>8 ) & 0xff)*(1/255.0f), ((clearColor>>0 ) & 0xff)*(1/255.0f), ((clearColor>>24) & 0xff)*(1/255.0f)); bits |= GL_COLOR_BUFFER_BIT; } if (bpmem.zmode.updateenable) { glClearDepth((float)(bpmem.clearZValue & 0xFFFFFF) / float(0xFFFFFF)); bits |= GL_DEPTH_BUFFER_BIT; } if (bRestoreZBufferTarget) glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT); // don't clear ztarget here glClear(bits); } // Have to clear the target zbuffer if (bpmem.zmode.updateenable && bRestoreZBufferTarget) { glDrawBuffer(GL_COLOR_ATTACHMENT1_EXT); GL_REPORT_ERRORD(); glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE); // red should probably be the LSB glClearColor(((bpmem.clearZValue>>0)&0xff)*(1/255.0f), ((bpmem.clearZValue>>8)&0xff)*(1/255.0f), ((bpmem.clearZValue>>16)&0xff)*(1/255.0f), 0); glClear(GL_COLOR_BUFFER_BIT); Renderer::SetColorMask(); GL_REPORT_ERRORD(); } if (bRestoreZBufferTarget) { // restore target GLenum s_drawbuffers[2] = {GL_COLOR_ATTACHMENT0_EXT, GL_COLOR_ATTACHMENT1_EXT}; glDrawBuffers(2, s_drawbuffers); } } Renderer::RestoreGLState(); } break; // ================================== case BPMEM_LOADTLUT: { DVSTARTSUBPROFILE("LoadBPReg:GXLoadTlut"); VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; u32 tlutTMemAddr = (newval & 0x3FF) << 9; u32 tlutXferCount = (newval & 0x1FFC00) >> 5; u8 *ptr = 0; // TODO - figure out a cleaner way. if (g_VideoInitialize.bWii) ptr = g_VideoInitialize.pGetMemoryPointer(bpmem.tlutXferSrc << 5); else ptr = g_VideoInitialize.pGetMemoryPointer((bpmem.tlutXferSrc & 0xFFFFF) << 5); if (ptr) memcpy_gc(texMem + tlutTMemAddr, ptr, tlutXferCount); else PanicAlert("Invalid palette pointer %08x %08x %08x", bpmem.tlutXferSrc, bpmem.tlutXferSrc << 5, (bpmem.tlutXferSrc & 0xFFFFF)<< 5); // TODO(ector) : kill all textures that use this palette // Not sure if it's a good idea, though. For now, we hash texture palettes } break; default: switch (addr & 0xFC) //texture sampler filter { case 0x28: // tevorder 0-3 case 0x2C: // tevorder 4-7 if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetTevOrderChanged(addr - 0x28); } break; case 0x80: // TEX MODE 0 case 0xA0: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; } break; case 0x84://TEX MODE 1 case 0xA4: break; case 0x88://TEX IMAGE 0 case 0xA8: if (changes) { //textureChanged[((addr&0xE0)==0xA0)*4+(addr&3)] = true; VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; } break; case 0x8C://TEX IMAGE 1 case 0xAC: if (changes) { //textureChanged[((addr&0xE0)==0xA0)*4+(addr&3)] = true; VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; } break; case 0x90://TEX IMAGE 2 case 0xB0: if (changes) { //textureChanged[((addr&0xE0)==0xA0)*4+(addr&3)] = true; VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; } break; case 0x94://TEX IMAGE 3 case 0xB4: if (changes) { //textureChanged[((addr&0xE0)==0xA0)*4+(addr&3)] = true; VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; } break; case 0x98://TEX TLUT case 0xB8: if (changes) { //textureChanged[((addr&0xE0)==0xA0)*4+(addr&3)] = true; VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; } break; case 0x9C://TEX UNKNOWN case 0xBC: //ERROR_LOG("texunknown%x = %x\n", addr, newval); ((u32*)&bpmem)[addr] = newval; break; case 0xE0: case 0xE4: if (addr&1) { // don't compare with changes! VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; int num = (addr>>1)&0x3; PixelShaderManager::SetColorChanged(bpmem.tevregs[num].high.type, num); } else ((u32*)&bpmem)[addr] = newval; break; default: switch (addr&0xF0) { case 0x10: // tevindirect 0-15 if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetTevIndirectChanged(addr - 0x10); } break; case 0x30: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetTexDimsChanged((addr >> 1) & 0x7); } break; case 0xC0: case 0xD0: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; PixelShaderManager::SetTevCombinerChanged((addr & 0x1f) / 2); } break; case 0x20: case 0x80: case 0x90: case 0xA0: case 0xB0: // Just update the bpmem struct, don't do anything else default: if (changes) { VertexManager::Flush(); ((u32*)&bpmem)[addr] = newval; /*switch(addr) { case 0x01: case 0x02: case 0x03: case 0x04: break; // copy filter values case 0x0f: break; // mask case 0x27: break; // tev ind order case 0x44: break; // field mask case 0x45: break; // draw done case 0x46: break; // clock case 0x49: case 0x4a: break; // copy tex src case 0x4b: break; // copy tex dest case 0x4d: break; // copyMipMapStrideChannels case 0x4e: break; // disp copy scale case 0x4f: break; // clear color case 0x50: break; // clear color case 0x51: break; // casez case 0x52: break; // trigger efb copy case 0x53: case 0x54: break; // more copy filters case 0x55: case 0x56: break; // bounding box case 0x64: case 0x65: break; // tlut src dest case 0xe8: break; // fog range case 0xe9: case 0xea: case 0xeb: case 0xec: case 0xed: break; // fog case 0xfe: break; // mask default: // 0x58 = 0xf // 0x69 = 0x49e ERROR_LOG("bp%.2x = %x\n", addr, newval); }*/ } break; } break; } break; } }