fix texconv build

texconv: Add support for modifying txds, add gamepad assets

See merge request skmp/dca3-game!22

dreamcast/Makefile

@@ -79,6 +79,10 @@ OBJS_TEXCONV += \
 	../vendor/librw/src/d3d-x/d3d.texconv.o \
 	../vendor/librw/src/d3d-x/d3d8.texconv.o \
 	../vendor/librw/src/d3d-x/d3d8render.texconv.o \
+	../vendor/librw/src/bmp.texconv.o \
+	../vendor/librw/src/png.texconv.o \
+	../vendor/librw/src/lodepng/lodepng.texconv.o \
+	../vendor/tlsf/tlsf.texconv.o
 
 # Add compilation units to this list to explicity compile them with
 # -O3 optimizations, while the rest get the default (-Os) treatment
@@ -274,6 +278,9 @@ aud2adpcm: aud2adpcm.c
 texconv: $(OBJS_TEXCONV) | pvrtex # You'll have to rebuild pvrtex manually if you change it
 	$(CXX) -o $@ $(OBJS_TEXCONV)
 
+%.texconv.o: %.c
+	$(CXX) -c -O3 -g -MMD -MP -o $@ -I../vendor/koshle $(INCLUDE) -I../vendor/emu -I../vendor/crypto -I../vendor/TriStripper/include $(DEFINES) -DDC_TEXCONV -DDC_SIM -D_INC_WINDOWS $(TEXCONV_FLAGS) $<
+
 %.texconv.o: %.cpp
 	$(CXX) -std=c++2a -c -O3 -g -MMD -MP -o $@ -I../vendor/koshle $(INCLUDE) -I../vendor/emu -I../vendor/crypto -I../vendor/TriStripper/include $(DEFINES) -DDC_TEXCONV -DDC_SIM -D_INC_WINDOWS $(TEXCONV_FLAGS) $<
 
@@ -316,6 +323,22 @@ TXD_OPTS_NEWS = 512 512
 TXD_OPTS_SPLASH1 = 512 512
 TXD_OPTS_SPLASH2 = 512 512
 TXD_OPTS_SPLASH3 = 512 512
+TXD_OPTS_frontend = 512 512 \
+	--delete-tex "fe_arrows4" \
+	--delete-tex "fe_arrows2" \
+	--delete-tex "fe_arrows1" \
+	--delete-tex "fe_controllersh" \
+	--delete-tex "fe_controller" \
+	\
+	--include-tex assets/dc_ps2d.png dc_ps2d \
+	--include-tex assets/dc_ps2f.png dc_ps2f \
+	--include-tex assets/dc_xboxd.png dc_xboxd \
+	--include-tex assets/dc_xboxf.png dc_xboxf \
+	--include-tex assets/ps4_d.png ps4_d \
+	--include-tex assets/ps4_f.png ps4_f \
+	--include-tex assets/xbox_d.png xbox_d \
+	--include-tex assets/xbox_f.png xbox_f
+
 DEFAULT_RES = 512
 
 PVR_ENCODER ?= PVRTEX

dreamcast/common.mk

@@ -269,6 +269,7 @@ RE3_OBJS = \
 	../src/vmu/vmu.o \
 	../vendor/miniLZO/minilzo.o \
 	\
+	../vendor/tlsf/tlsf.o
 
 # Excluded \
 	../src/extras/custompipes.o \
@@ -380,7 +381,9 @@ INCLUDE = \
 \
 -I../vendor/librw \
 \
--I../vendor/miniLZO
+-I../vendor/miniLZO \
+\
+-I../vendor/tlsf
 
 DEFINES = -DRW_DC -DLIBRW $(if $(WITH_LOGGING),-DWITH_LOGGING) $(if $(WITH_DCLOAD),-DDC_CHDIR=/pc) \
 	$(if $(WITH_BEEPS),-DWITH_BEEPS)

dreamcast/texconv.cpp

@@ -3,6 +3,7 @@
 #include <cstring>
 
 #include <set>
+#include <vector>
 
 #include <iostream>
 using namespace std;
@@ -238,6 +239,40 @@ PluginAttach(void)
 
 const char* currentFile;
 
+namespace rw {
+	Image* readBMP(const char *filename);
+	Image* readPNG(const char *filename);
+}
+
+void InsertImage(RwTexDictionary* texDict, const char* file, const char* texName) {
+	RwTexture *tex;
+	RwRaster *raster;
+	RwInt32 width, height, depth, format;
+	RwImage *image = rw::readBMP(file);
+	if (!image) {
+		image = rw::readPNG(file);
+	}
+
+	assert(image);
+
+	RwImageFindRasterFormat(image, rwRASTERTYPETEXTURE, &width, &height, &depth, &format);
+	raster = RwRasterCreate(width, height, depth, format);
+	RwRasterSetFromImage(raster, image);
+
+	tex = RwTextureCreate(raster);
+	RwTextureSetName(tex, texName);
+
+	RwTextureSetFilterMode(tex, rwFILTERLINEAR);
+
+	RwTexDictionaryAddTexture(texDict, tex);
+
+	RwImageDestroy(image);
+}
+
+std::vector<std::pair<const char*, const char*>> ImagesToAdd;
+std::vector<const char*> ImagesToRemove;
+bool listTextures = false;
+
 int main(int argc, const char** argv) {
 	if (argc >= 5) {
 	    int width = atoi(argv[3]);
@@ -247,7 +282,7 @@ int main(int argc, const char** argv) {
         if(height >= 16 && height <= 1024)
 		    rw::dc::maxRasterHeight = height;
 	}
-    for (int i = 0; i < argc; i++) {
+    for (int i = 5; i < argc; i++) {
         if (argv[i] != nullptr) {
             // Downsample Parameter
             if (strcmp(argv[i], "-d") == 0 || strcmp(argv[i], "-D") == 0) {
@@ -271,6 +306,22 @@ int main(int argc, const char** argv) {
                     }
                 }
             }
+
+			if (strcmp(argv[i], "--include-tex") == 0) {
+				assert(i + 2 < argc);
+				ImagesToAdd.emplace_back(argv[i+1], argv[i+2]);
+				i += 2;
+			}
+
+			if (strcmp(argv[i], "--delete-tex") == 0) {
+				assert(i + 1 < argc);
+				ImagesToRemove.emplace_back(argv[i+1]);
+				i += 1;
+			}
+
+			if (strcmp(argv[i], "--list-tex") == 0) {
+				listTextures = true;
+			}
         }
     }
 
@@ -298,6 +349,31 @@ int main(int argc, const char** argv) {
 
 		RwStreamClose(stream, nil);
 
+		if (listTextures) {
+			fprintf(stdout, "Incoming textures:\n");
+			FORLIST(lnk, texDict->textures) {
+				auto tex = rw::Texture::fromDict(lnk);
+				fprintf(stdout, "texture: '%s'\n", tex->name);
+			}
+		}
+
+		for (auto&& removedTextureName: ImagesToRemove) {
+			auto removedTexture = texDict->find(removedTextureName);
+			assert(removedTexture);
+			texDict->remove(removedTexture);
+		}
+		for (auto&& extraTexture: ImagesToAdd) {
+			InsertImage(texDict, extraTexture.first, extraTexture.second);
+		}
+
+		if (listTextures) {
+			fprintf(stdout, "Processing textures:\n");
+			FORLIST(lnk, texDict->textures) {
+				auto tex = rw::Texture::fromDict(lnk);
+				fprintf(stdout, "texture: '%s'\n", tex->name);
+			}
+		}
+
 		auto streamOut = RwStreamOpen(rwSTREAMFILENAME, rwSTREAMWRITE, argv[2]);
 		assert(streamOut && "failed to open output");
 

src/animation/AnimBlendHierarchy.cpp

@@ -52,7 +52,7 @@ CAnimBlendHierarchy::RemoveAnimSequences(void)
 	numSequences = 0;
 }
 
-#ifdef USE_CUSTOM_ALLOCATOR
+#if 0
 void
 CAnimBlendHierarchy::MoveMemory(bool onlyone)
 {
@@ -61,4 +61,4 @@ CAnimBlendHierarchy::MoveMemory(bool onlyone)
 		if(sequences[i].MoveMemory() && onlyone)
 			return;
 }
-#endif
+#endif

src/animation/AnimBlendNode.cpp

@@ -140,7 +140,7 @@ CAnimBlendNode::GetCurrentTranslation(CVector &trans, float weight)
 	float blend = association->GetBlendAmount(weight);
 	if(blend > 0.0f){
 		auto kfAdt = sequence->GetDeltaTime(frameA);
-		float t = (kfAdt - remainingTime)/kfAdt;
+		float t = kfAdt == 0.0f ? 0.0f : (kfAdt - remainingTime)/kfAdt;
 		if(sequence->type & CAnimBlendSequence::KF_TRANS){
 			auto kfAt = sequence->GetTranslation(frameA);
 			auto kfBt = sequence->GetTranslation(frameB);

src/animation/AnimBlendSequence.cpp

@@ -1,3 +1,9 @@
+#include <cstddef>
+
+void* obj_alloc(size_t size, void** storage);
+void obj_free(void* ptr);
+void* obj_move(void* ptr);
+
 #include "common.h"
 
 #include "AnimBlendSequence.h"
@@ -16,7 +22,7 @@ CAnimBlendSequence::CAnimBlendSequence(void)
 CAnimBlendSequence::~CAnimBlendSequence(void)
 {
 	if(keyFrames)
-		RwFree(keyFrames);
+		obj_free(keyFrames);
 }
 
 void
@@ -42,7 +48,8 @@ CAnimBlendSequence::SetNumFrames(int numFrames, bool translation, bool compress)
 		sz = sizeof(KeyFrame);
 		type |= KF_ROT;
 	}
-	keyFrames = RwMalloc(sz * numFrames);
+	keyFrames = obj_alloc(sz * numFrames, &(void*&)keyFrames);
+	assert(keyFrames != nullptr);
 	this->numFrames = numFrames;
 }
 
@@ -64,24 +71,17 @@ CAnimBlendSequence::RemoveQuaternionFlips(void)
 	}
 }
 
-#ifdef USE_CUSTOM_ALLOCATOR
+#if 0
 bool
 CAnimBlendSequence::MoveMemory(void)
 {
-	if(keyFrames){
-		void *newaddr = gMainHeap.MoveMemory(keyFrames);
-		if(newaddr != keyFrames){
+	if (keyFrames) {
+		void* newaddr = obj_move(keyFrames);
+		if (newaddr) {
 			keyFrames = newaddr;
 			return true;
 		}
-	}else if(keyFramesCompressed){
-		void *newaddr = gMainHeap.MoveMemory(keyFramesCompressed);
-		if(newaddr != keyFramesCompressed){
-			keyFramesCompressed = newaddr;
-			return true;
-		}
 	}
 	return false;
 }
-#endif
-
+#endif

src/audio/AudioLogic.cpp

@@ -3789,6 +3789,10 @@ cAudioManager::ProcessPedOneShots(cPedParams &params)
 						if (iSound > 60)
 							iSound = 21;
 					}
+					// In some cases the left and right channels have different loop points
+					// This looks like a data file issue where the left and right channels have different loop points
+					// This is a hot fix to always have the correct loop point for each channel individually
+					SET_LOOP_OFFSETS(m_sQueueSample.m_nSampleIndex)
 					AddSampleToRequestedQueue();
 				}
 			}

src/audio/sampman.h

@@ -174,6 +174,10 @@ public:
 	
 	void UpdateEffectsVolume(void);
 
+#ifdef DC_SH4
+	void UpdateChannelVolume(uint32 nChannel);
+#endif
+
 	void SetEffectsMasterVolume(uint8 nVolume);
 	void SetMusicMasterVolume  (uint8 nVolume);
 	void SetEffectsFadeVolume  (uint8 nVolume);

src/audio/sampman_dc.cpp

@@ -151,7 +151,6 @@ void aica_snd_sfx_freq_vol(int chn, int freq, int vol) {
 
 cSampleManager SampleManager;
 bool8 _bSampmanInitialised = FALSE;
-bool _dcAudioInitialized = false;
 
 uint32 BankStartOffset[MAX_SFX_BANKS];
 char SampleBankDescFilename[] = "sfx/sfx_all.dsc";
@@ -173,6 +172,12 @@ file_t fdPedSfx;
 volatile uint32 nPedSfxReqReadId = 1;
 volatile uint32 nPedSfxReqNextId = 1;
 
+// this is very wasteful and temporary
+#define BANK_STAGE_SIZE 16 * 2048
+static  uint8_t stagingBufferBank[BANK_STAGE_SIZE] __attribute__((aligned(32)));
+static  uint8_t stagingBufferPedSFX[PED_BLOCKSIZE_ADPCM] __attribute__((aligned(32)));
+
+
 struct WavHeader {
     // RIFF Header
     char riff[4];        // RIFF Header Magic header
@@ -462,7 +467,7 @@ cSampleManager::Initialise(void)
 
 	assert(fdPedSfx >= 0);
 
-	_dcAudioInitialized = true;
+	_bSampmanInitialised = true;
 	return TRUE;
 }
 
@@ -485,7 +490,14 @@ char cSampleManager::GetCDAudioDriveLetter(void)
 void
 cSampleManager::UpdateEffectsVolume(void)
 {
-	// TODO
+	if(_bSampmanInitialised) {
+		std::lock_guard<std::mutex> lk(channel_mtx);
+		for (int i = 0; i < MAXCHANNELS+MAX2DCHANNELS; i++) {
+			if (channels[i].ch != -1) {
+				UpdateChannelVolume(i);
+			}
+		}
+	}
 }
 
 
@@ -520,6 +532,7 @@ cSampleManager::SetMonoMode(uint8 nMode)
 {
 }
 
+
 bool8
 cSampleManager::LoadSampleBank(uint8 nBank)
 {
@@ -549,8 +562,8 @@ cSampleManager::LoadSampleBank(uint8 nBank)
 		// TODO: Split per-bank sfx file
 		int fd = fs_open(SampleBankDataFilename, O_RDONLY);
 		assert(fd >= 0);
-		// this is very wasteful and temporary
-		void* stagingBuffer = memalign(32, 32 * 2048);
+
+		auto stagingBuffer = stagingBufferBank;
 		assert(stagingBuffer != 0);
 
 		// Ideally, we'd suspend the CdStream thingy here or read via that instead
@@ -558,7 +571,7 @@ cSampleManager::LoadSampleBank(uint8 nBank)
 		fs_seek(fd, fileStart, SEEK_SET);
 
 		while (fileSize > 0) {
-			size_t readSize = fileSize > 32 * 2048 ? 32 * 2048 : fileSize;
+			size_t readSize = fileSize > BANK_STAGE_SIZE ? BANK_STAGE_SIZE : fileSize;
 			int rs = fs_read(fd, stagingBuffer, readSize);
 			debugf("Read %d bytes, expected %d\n", rs, readSize);
 			assert(rs == readSize);
@@ -568,7 +581,6 @@ cSampleManager::LoadSampleBank(uint8 nBank)
 			debugf("Loaded %d bytes, %d remaining\n", readSize, fileSize);
 		}
 		fs_close(fd);
-		free(stagingBuffer);
 		
 
 		for (int nSfx = BankStartOffset[nBank]; nSfx < BankStartOffset[nBank+1]; nSfx++) {
@@ -717,7 +729,8 @@ cSampleManager::LoadPedComment(uint32 nComment)
 		// TODO: When we can dma directly to AICA, we can use this instead
 		// fs_read(fdPedSfx, SPU_BASE_U8 + (uintptr_t)cmd->dest, cmd->size);
 
-		void* stagingBuffer = memalign(32, cmd->size);
+		// TODO: Merge stagingBufferPedSFX with stagingBuffer
+		void* stagingBuffer = stagingBufferPedSFX;
 		assert(stagingBuffer != 0);
 		debugf("Allocated %d bytes at %p\n", cmd->size, stagingBuffer);
 		int rs = fs_read(fdPedSfx, stagingBuffer, cmd->size);
@@ -725,7 +738,6 @@ cSampleManager::LoadPedComment(uint32 nComment)
 		assert(rs == cmd->size);
 
 		spu_memload((uintptr_t)cmd->dest, stagingBuffer, cmd->size);
-		free(stagingBuffer);
 		nPedSfxReqReadId = nPedSfxReqReadId + 1;
 	});
 
@@ -851,9 +863,10 @@ cSampleManager::InitialiseChannel(uint32 nChannel, uint32 nSfx, uint8 nBank)
 	return TRUE;
 }
 
-void updateVol(uint32 nChannel) {
+void cSampleManager::UpdateChannelVolume(uint32 nChannel) {
 
 	auto newVol = channels[nChannel].emittingVol * channels[nChannel].attenuationVol / 255;
+	newVol = m_nEffectsFadeVolume * newVol * m_nEffectsVolume >> 14;
 	// newVol = 255;
 	// printf("updateVol(nChannel: %d) vol: %d, newVol: %d\n", nChannel, channels[nChannel].vol, newVol);
 	if (channels[nChannel].vol != newVol) {
@@ -876,7 +889,7 @@ cSampleManager::SetChannelVolume(uint32 nChannel, uint32 nVolume)
 	channels[nChannel].emittingVol = linearlize_volume(nVolume);// nVolume * 255 / MAX_VOLUME;
 	channels[nChannel].attenuationVol = 255;
 
-	updateVol(nChannel);
+	UpdateChannelVolume(nChannel);
 	verbosef("SetChannelVolume(nChannel: %d) vol: %d\n", nChannel, nVolume);
 }
 
@@ -1132,7 +1145,8 @@ cSampleManager::StartStreamedFile(uint8 nFile, uint32 nPos, uint8 nStream)
 	debugf("StartStreamedFile(%d, %d, %d)\n", nFile, nPos, nStream);
 	uint32_t seek_aligned = 0;
 	if (nPos) {
-		uint32 seek_bytes = nPos * streams[nStream].rate / (streams[nStream].stereo ? 1000: 2000);
+		uint64_t seek_bytes = (uint64_t)nPos * streams[nStream].rate / (streams[nStream].stereo ? 1000: 2000);
+		assert(seek_bytes <= INT32_MAX);
 		seek_aligned = seek_bytes & ~(streams[nStream].stereo ? (STREAM_STAGING_READ_SIZE_STEREO-1) : (STREAM_STAGING_READ_SIZE_MONO-1));
 	}
 	PreloadStreamedFile(nFile, nStream, seek_aligned);
@@ -1164,7 +1178,9 @@ cSampleManager::GetStreamedFilePosition(uint8 nStream)
 	ASSERT( nStream < MAX_STREAMS );
 	int32 rv = 0;
 
-	return streams[nStream].played_samples * 1000 / streams[nStream].rate;
+	int64_t rv64 = (int64_t)streams[nStream].played_samples * 1000 / streams[nStream].rate;
+	assert(rv64 <= INT32_MAX);
+	rv = (int32)rv64;
 	// if(streams[nStream].fd >= 0) {
 	// 	rv = fs_tell(streams[nStream].fd);
 	// }
@@ -1179,6 +1195,7 @@ cSampleManager::SetStreamedVolumeAndPan(uint8 nVolume, uint8 nPan, uint8 nEffect
 	if (nVolume > MAX_VOLUME)
 		nVolume = MAX_VOLUME;
 	nVolume = linearlize_volume(nVolume); //nVolume * 255 / MAX_VOLUME;
+	nVolume = m_nMusicFadeVolume * nVolume * m_nMusicVolume >> 14;
 	if (streams[nStream].vol != nVolume || streams[nStream].nPan != nPan) {
 		streams[nStream].vol = nVolume;
 		streams[nStream].nPan = nPan;
@@ -1207,7 +1224,10 @@ cSampleManager::GetStreamedFileLength(uint8 nFile)
 	WavHeader hdr;
 	assert(fs_read(fd, &hdr, sizeof(hdr)) == sizeof(hdr));
 
-	rv = hdr.dataSize * 2000 / hdr.numOfChan / hdr.samplesPerSec;
+	uint64_t rv64 = (uint64_t)hdr.dataSize * 2000 / hdr.numOfChan / hdr.samplesPerSec;
+
+	assert(rv64 <= INT32_MAX);
+	rv = (int32)rv64;
 
 	fs_close(fd);
 
@@ -1287,9 +1307,17 @@ void  cSampleManager::SetChannelEmittingVolume(uint32 nChannel, uint32 nVolume)
 	if (nVolume > MAX_VOLUME)
 		nVolume = MAX_VOLUME;
 
+	// reduce channel volume when JB.MP3 or S4_BDBD.MP3 playing
+	if (   MusicManager.GetMusicMode()    == MUSICMODE_CUTSCENE
+		&& MusicManager.GetNextTrack() != STREAMED_SOUND_NEWS_INTRO
+		&& MusicManager.GetNextTrack() != STREAMED_SOUND_CUTSCENE_SAL4_BDBD )
+	{
+		nVolume /= 4;
+	}
+
 	std::lock_guard<std::mutex> lk(channel_mtx);
 	channels[nChannel].emittingVol = linearlize_volume(nVolume); // nVolume * 255 / MAX_VOLUME;
-	updateVol(nChannel);
+	UpdateChannelVolume(nChannel);
 }
 
 float calculatePan(float x, float z) {
@@ -1334,7 +1362,7 @@ void  cSampleManager::SetChannel3DPosition    (uint32 nChannel, float fX, float
 		channels[nChannel].fY = fY;
 		channels[nChannel].fZ = fZ;
 		channels[nChannel].attenuationVol = calculateAttenuation(channels[nChannel].fX, channels[nChannel].fY, channels[nChannel].fZ, channels[nChannel].distMin, channels[nChannel].distMax) * 255;
-		updateVol(nChannel);
+		UpdateChannelVolume(nChannel);
 	}
 
 	SetChannelPan(nChannel, calculatePan(-fX, fZ) * 63 + 64);
@@ -1347,7 +1375,7 @@ void  cSampleManager::SetChannel3DDistances   (uint32 nChannel, float fMax, floa
 	channels[nChannel].distMin = fMin;
 	channels[nChannel].distMax = fMax;
 	channels[nChannel].attenuationVol = calculateAttenuation(channels[nChannel].fX, channels[nChannel].fY, channels[nChannel].fZ, channels[nChannel].distMin, channels[nChannel].distMax) * 255;
-	updateVol(nChannel);
+	UpdateChannelVolume(nChannel);
 }
 #endif
 

src/control/PathFind.cpp

@@ -34,6 +34,9 @@ CPathInfoForObject *InfoForTilePeds;
 CTempDetachedNode *DetachedNodesCars;
 CTempDetachedNode *DetachedNodesPeds;
 
+void* obj_alloc(size_t size, void** storage);
+void obj_free(void* ptr);
+
 bool 
 CPedPath::CalcPedRoute(int8 pathType, CVector position, CVector destination, CVector *pointPoses, int16 *pointsFound, int16 maxPoints)
 {
@@ -245,15 +248,22 @@ CPathFind::Init(void)
 void
 CPathFind::AllocatePathFindInfoMem(int16 numPathGroups)
 {
-	delete[] InfoForTileCars;
-	InfoForTileCars = nil;
-	delete[] InfoForTilePeds;
-	InfoForTilePeds = nil;
+	assert(numPathGroups == 4500);
+	
+	if (InfoForTileCars) {
+		obj_free(InfoForTileCars);
+		InfoForTileCars = nil;
+	}
+	
+	if (InfoForTilePeds) {
+		obj_free(InfoForTilePeds);
+		InfoForTilePeds = nil;
+	}
 
 	// NB: MIAMI doesn't use numPathGroups here but hardcodes 4500
-	InfoForTileCars = new CPathInfoForObject[12*numPathGroups];
+	InfoForTileCars = (CPathInfoForObject*) obj_alloc(sizeof(CPathInfoForObject)*12*numPathGroups, nil);
 	memset(InfoForTileCars, 0, 12*numPathGroups*sizeof(CPathInfoForObject));
-	InfoForTilePeds = new CPathInfoForObject[12*numPathGroups];
+	InfoForTilePeds = (CPathInfoForObject*) obj_alloc(sizeof(CPathInfoForObject)*12*numPathGroups, nil);
 	memset(InfoForTilePeds, 0, 12*numPathGroups*sizeof(CPathInfoForObject));
 
 	// unused
@@ -438,10 +448,15 @@ CPathFind::PreparePathData(void)
 		CountFloodFillGroups(PATH_CAR);
 		CountFloodFillGroups(PATH_PED);
 
-		delete[] InfoForTileCars;
-		InfoForTileCars = nil;
-		delete[] InfoForTilePeds;
-		InfoForTilePeds = nil;
+		if (InfoForTileCars) {
+			obj_free(InfoForTileCars);
+			InfoForTileCars = nil;
+		}
+		
+		if (InfoForTilePeds) {
+			obj_free(InfoForTilePeds);
+			InfoForTilePeds = nil;
+		}
 
 		delete[] DetachedNodesCars;
 		DetachedNodesCars = nil;

src/control/Script4.cpp

@@ -1721,8 +1721,8 @@ int8 CRunningScript::ProcessCommands900To999(int32 command)
 			return 0;
 		int attempts;
 		int model = -1;
-		int index = CGeneral::GetRandomNumberInRange(0, 50);
-		for (attempts = 0; attempts < 50; attempts++) {
+		int index = CGeneral::GetRandomNumberInRange(0, MAXVEHICLESLOADED);
+		for (attempts = 0; attempts < MAXVEHICLESLOADED; attempts++) {
 			if (model != -1)
 				break;
 			model = CStreaming::ms_vehiclesLoaded[index];

src/core/Game.cpp

@@ -1012,8 +1012,12 @@ void CGame::InitialiseWhenRestarting(void)
 	DMAudio.ChangeMusicMode(MUSICMODE_GAME);
 }
 
+bool obj_relocate();
+
 void CGame::Process(void) 
 {
+	obj_relocate();
+	
 	CPad::UpdatePads();
 #ifdef USE_CUSTOM_ALLOCATOR
 	ProcessTidyUpMemory();

src/core/Pad.cpp

@@ -2785,22 +2785,22 @@ bool CPad::WeaponJustDown(void)
 		case 0:	//Xbox Mode
 			if (CPad::GetPad(0)->IsDualAnalog)
 			{
-				if (NewState.RightTrigger > 128)
+				if (NewState.RightTrigger > 128 && !(OldState.RightTrigger > 128))
 					return true;			
 			}
 			else
 			{
-				if (NewState.RightTrigger > 128)
+				if (NewState.RightTrigger > 128 && !(OldState.RightTrigger > 128))
 					return true;			
 			}
 		case 1:	//PS2 Mode
 			if (CPad::GetPad(0)->IsDualAnalog)
 			{
-				return	NewState.B;
+				return	NewState.B && !OldState.B;
 			}
 			else
 			{
-				return	NewState.B;
+				return	NewState.B && !OldState.B;
 			}
 	}
 

src/core/Streaming.cpp

@@ -1164,6 +1164,12 @@ bool re3RemoveLeastUsedModel() {
 	return CStreaming::RemoveLeastUsedModel();
 }
 
+bool re3EmergencyRemoveModel() {
+	auto initial = CStreaming::ms_memoryUsed;
+	CStreaming::DeleteRwObjectsBehindCamera(CStreaming::ms_memoryUsed);
+	return CStreaming::ms_memoryUsed < initial;
+}
+
 bool
 CStreaming::RemoveLeastUsedModel(void)
 {

src/entities/Entity.cpp

@@ -83,6 +83,11 @@ CEntity::~CEntity(void)
 	ResolveReferences();
 }
 
+bool
+CEntity::IsFence(void) {
+	return IsObject() && ::IsFence(static_cast<CObject *>(this)->GetModelIndex());
+}
+
 void
 CEntity::SetModelIndex(uint32 id)
 {

src/entities/Entity.h

@@ -131,6 +131,7 @@ public:
 	bool IsPed(void) { return m_type == ENTITY_TYPE_PED; }
 	bool IsObject(void) { return m_type == ENTITY_TYPE_OBJECT; }
 	bool IsDummy(void) { return m_type == ENTITY_TYPE_DUMMY; }
+	bool IsFence(void);
 
 	RpAtomic *GetAtomic(void) {
 		assert(RwObjectGetType(m_rwObject) == rpATOMIC);

src/entities/Physical.cpp

@@ -59,7 +59,7 @@ CPhysical::CPhysical(void)
 	bInfiniteMass = false;
 	bIsInWater = false;
 	bHitByTrain = false;
-	bSkipLineCol = false;
+	bSkipLineCol = IsFence();
 
 	m_fDistanceTravelled = 0.0f;
 	m_treadable[PATH_CAR] = nil;
@@ -1461,6 +1461,10 @@ CPhysical::ProcessCollisionSectorList(CPtrList *lists)
 			skipCollision = false;
 			altcollision = false;
 
+			if(A->IsFence() && B->IsFence()) {
+				skipCollision = true;
+				A->bSkipLineCol = true;
+			}
 			if(B->IsBuilding())
 				skipCollision = false;
 			else if(IsStreetLight(A->GetModelIndex()) &&
@@ -1922,12 +1926,12 @@ CPhysical::ProcessCollision(void)
 				n = NUMSTEPS(0.09f);
 				step = savedTimeStep / n;
 			}
-		}else if(responsecase == COLLRESPONSE_SMALLBOX || responsecase == COLLRESPONSE_FENCEPART){
+		}else if(responsecase == COLLRESPONSE_SMALLBOX){
 			if(distSq >= sq(0.15f)){
 				n = NUMSTEPS(0.15f);
 				step = savedTimeStep / n;
 			}
-		}else{
+		}else if(responsecase != COLLRESPONSE_FENCEPART){
 			if(distSq >= sq(0.3f)){
 				n = NUMSTEPS(0.3f);
 				step = savedTimeStep / n;

src/vmu/vmu.cpp

@@ -10,7 +10,7 @@
 #       include <dc/sound/sound.h>
 #   endif
 
-extern bool _dcAudioInitialized;
+extern bool _bSampmanInitialised;
 
 // ====== STATIC METHODS =====
 
@@ -86,7 +86,7 @@ void VmuProfiler::run() {
 
 #ifdef DC_SH4
         if(auto *dev = maple_enum_type(0, MAPLE_FUNC_MEMCARD); 
-            dev && _dcAudioInitialized && updated_) 
+            dev && _bSampmanInitialised && updated_) 
         {
             pvr_stats_t pvrStats;   pvr_get_stats(&pvrStats);
             uint32_t    sramStats = snd_mem_available();

vendor/librw/src/dc/rwdc.cpp

@@ -16,6 +16,8 @@ extern const char* currentFile;
 #define texconvf(...) // printf(__VA_ARGS__)
 #endif
 
+#include "tlsf.h"
+
 #include "../../../src/vmu/vmu.h"
 #include "../rwbase.h"
 #include "../rwerror.h"
@@ -39,6 +41,86 @@ extern const char* currentFile;
 
 #define logf(...) // printf(__VA_ARGS__)
 bool re3RemoveLeastUsedModel();
+bool re3EmergencyRemoveModel();
+
+std::map<void*, void**> relocatableAllocs;
+
+uint8_t obj_heap[4 * 1024 * 1024];
+tlsf_t obj_pool;
+
+void obj_init() {
+	obj_pool = tlsf_create_with_pool(obj_heap, sizeof(obj_heap));
+}
+
+void* last_relocation;
+bool obj_relocate();
+
+void* obj_alloc(size_t size, void** storage) {
+	auto rv = tlsf_malloc(obj_pool, size);
+
+	while (rv == nullptr) {
+		if (!re3RemoveLeastUsedModel() && !re3EmergencyRemoveModel()) {
+			fprintf(stderr, "obj_alloc: out of memory, doing full compaction\n");
+			last_relocation = 0;
+			while (obj_relocate())
+				;
+			// last chance
+		}
+		fprintf(stderr, "obj_alloc: soft out of memory\n");
+		rv = tlsf_malloc(obj_pool, size);
+	}
+
+	relocatableAllocs[rv] = storage;
+	
+	return rv;
+}
+
+void obj_free(void* p) {
+	relocatableAllocs.erase(p);
+	tlsf_free(obj_pool, p);
+}
+
+void* obj_move(void* p) {
+	return tlsf_move(obj_pool, p);
+}
+
+bool obj_relocate() {
+	// FILE* f = fopen("/pc/Users/skmp/projects/dca3-game/dreamcast/chunks-sorted-with.txt.native.tmp", "w");
+	// fprintf(f, "ALLOC: %p, %d\n", (uintptr_t)obj_heap & 0xFFFFFF, sizeof(obj_heap));
+	// for(auto allocation: relocatableAllocs) {
+	// 	fprintf(f, "ALLOC: %p, %d\n", (uintptr_t&)allocation.first & 0xFFFFFF, tlsf_block_size(allocation.first));
+	// }
+	// fclose(f);
+
+	// fs_unlink("/pc/Users/skmp/projects/dca3-game/dreamcast/chunks-sorted-with.txt.native");
+	// fs_rename("/pc/Users/skmp/projects/dca3-game/dreamcast/chunks-sorted-with.txt.native.tmp", "/pc/Users/skmp/projects/dca3-game/dreamcast/chunks-sorted-with.txt.native");
+
+	// fprintf(stderr, "obj_relocate: %p\n", last_relocation);
+	int toRelocate = 10 * 1024;
+	auto start = relocatableAllocs.upper_bound(last_relocation);
+	if (start == relocatableAllocs.end())
+		start = relocatableAllocs.begin();
+	while(start != relocatableAllocs.end()) {
+		auto old = start->first;
+		auto storage = start->second;
+		auto oldSize = tlsf_block_size(old);
+		auto newp = obj_move(start->first);
+		if (newp) {
+			toRelocate -= oldSize;
+			*storage = newp;
+			start = relocatableAllocs.erase(start, std::next(start));
+			relocatableAllocs[newp] = storage;
+			last_relocation = newp;
+			// fprintf(stderr, "obj_relocate: %p -> %p, %d\n", old, newp, oldSize);
+			if (toRelocate <= 0)
+				return true;
+		} else {
+			start++;
+		}
+	}
+	last_relocation = 0;
+	return false;
+}
 
 // #include "rwdcimpl.h"
 
@@ -305,6 +387,8 @@ void rw_mat_load_4x4(rw::Matrix* mtx) {
 #define mat_identity(a)
 #define pvr_fog_table_color(a,r,g,b)
 #define pvr_fog_table_linear(s,e)
+#define pvr_fog_table_exp(d)
+#define pvr_fog_table_custom(d)
 #endif
 
 #define mat_trans_single3_nomod(x_, y_, z_, x2, y2, z2) do { \
@@ -663,7 +747,10 @@ void malloc_stats() { }
 #define UNIMPL_LOGV(...)
 #endif
 
+Camera* rwdcCam;
+
 void beginUpdate(Camera* cam)  {
+	rwdcCam = cam;
 	float view[16], proj[16];
 
 	// View Matrix
@@ -1098,6 +1185,7 @@ static bool doAlphaTest;
 
 static uint8_t fogFuncPvr = PVR_FOG_DISABLE;
 static uint32_t fogColor = 0;
+static float fogStart = 0.0f;
 static uint32 cullModePvr;
 
 static inline unsigned pvrCullMode(uint32_t cullMode) {
@@ -1198,8 +1286,9 @@ setRenderState(int32 state, void *pvalue)
 	 	fogFuncPvr = value ? PVR_FOG_TABLE : PVR_FOG_DISABLE;
 	 	break;
 	 case FOGCOLOR:
+#if !defined(DC_TEXCONV)		
         // Set fog color when state changes
-        if(fogColor != value) {
+        if(fogColor != value || fogStart != RwCameraGetFogDistance(rwdcCam)) {
             fogColor = value;
             RGBA c;
             c.red = value;
@@ -1207,9 +1296,22 @@ setRenderState(int32 state, void *pvalue)
             c.blue = value>>16;
             c.alpha = value>>24;
             pvr_fog_table_color(c.alpha / 255.0f, c.red / 255.0f, c.green  / 255.0f, c.blue  / 255.0f);
-            pvr_fog_table_linear(50.0f, 450.0f);
+
+			fogStart = RwCameraGetFogDistance(rwdcCam);
+			float fogEnd = RwCameraGetFarClipPlane(rwdcCam);
+			float fogIntensity[129];
+			uint8_t idx = 0;
+			float startIntensity = (-fogStart) / (fogEnd - fogStart);  //interpolate between start and end to get initial intensity
+			float step = (1.0f - startIntensity) / 129; // we have 129 entries, create a step such that start + (step*129) = 1.0
+			for(int i = 128; i >= 0; i--) {
+				fogIntensity[i] = startIntensity + (idx++ * step);
+			}
+			pvr_fog_far_depth(fogEnd);
+			pvr_fog_table_custom(fogIntensity);
         }
+#endif
 	 	break;
+
 	// case CULLMODE:
 	// 	if(rwStateCache.cullmode != value){
 	// 		rwStateCache.cullmode = value;
@@ -3887,19 +3989,19 @@ imageFindRasterFormat(Image *img, int32 type,
 		;
 
     if(downsampleMode >= HALF) {
-        if(height / 2 >= 64) {
+        if(height / 2 >= 16) {
             height /= 2;
         }
-        if(width / 2 >= 64) {
+        if(width / 2 >= 16) {
             width /= 2;
         }
     }
 
     if(downsampleMode >= QUARTER) {
-        if(height / 2 >= 32) {
+        if(height / 2 >= 16) {
             height /= 2;
         }
-        if(width / 2 >= 32) {
+        if(width / 2 >= 16) {
             width /= 2;
         }
     }
@@ -4362,6 +4464,7 @@ ObjPipeline* makeDefaultPipeline(void)
 static void*
 driverOpen(void *o, int32, int32)
 {
+	obj_init();
     pvr_init(&pvr_params);
 
 	fake_tex = pvr_mem_malloc(sizeof(fake_tex_data));
@@ -4583,7 +4686,7 @@ void*
 destroyNativeData(void *object, int32, int32)
 {
 	auto geo = (Geometry*)object;
-	rwFree(geo->instData);
+	obj_free(geo->instData);
 	geo->instData = nil;
 
 	return object;
@@ -4600,7 +4703,9 @@ readNativeData(Stream *stream, int32 length, void *object, int32, int32)
 		return nil;
 	}
 
-	DCModelDataHeader *header = (DCModelDataHeader *)rwNew(sizeof(DCModelDataHeader) + chunkLen - 8, MEMDUR_EVENT | ID_GEOMETRY);
+	DCModelDataHeader *header = (DCModelDataHeader *)obj_alloc(sizeof(DCModelDataHeader) + chunkLen - 8, &(void*&)geo->instData);
+	assert(header != nullptr);
+
 	geo->instData = header;
 	stream->read32(&header->platform, 4);
 	uint32_t version;

vendor/tlsf/README.md

@@ -0,0 +1,92 @@
+# tlsf
+Two-Level Segregated Fit memory allocator implementation.
+Written by Matthew Conte (matt@baisoku.org).
+Released under the BSD license.
+
+Features
+--------
+  * O(1) cost for malloc, free, realloc, memalign
+  * Extremely low overhead per allocation (4 bytes)
+  * Low overhead per TLSF management of pools (~3kB)
+  * Low fragmentation
+  * Compiles to only a few kB of code and data
+  * Support for adding and removing memory pool regions on the fly
+
+Caveats
+-------
+  * Currently, assumes architecture can make 4-byte aligned accesses
+  * Not designed to be thread safe; the user must provide this
+
+Notes
+-----
+This code was based on the TLSF 1.4 spec and documentation found at:
+
+	http://www.gii.upv.es/tlsf/main/docs
+
+It also leverages the TLSF 2.0 improvement to shrink the per-block overhead from 8 to 4 bytes.
+
+History
+-------
+2016/04/10 - v3.1
+  * Code moved to github
+  * tlsfbits.h rolled into tlsf.c
+  * License changed to BSD
+
+2014/02/08 - v3.0
+  * This version is based on improvements from 3DInteractive GmbH
+  * Interface changed to allow more than one memory pool
+  * Separated pool handling from control structure (adding, removing, debugging)
+  * Control structure and pools can still be constructed in the same memory block
+  * Memory blocks for control structure and pools are checked for alignment
+  * Added functions to retrieve control structure size, alignment size, min and max block size, overhead of pool structure, and overhead of a single allocation
+  * Minimal Pool size is tlsf_block_size_min() + tlsf_pool_overhead()
+  * Pool must be empty when it is removed, in order to allow O(1) removal
+
+2011/10/20 - v2.0
+  * 64-bit support
+  * More compiler intrinsics for ffs/fls
+  * ffs/fls verification during TLSF creation in debug builds
+
+2008/04/04 - v1.9
+  * Add tlsf_heap_check, a heap integrity check
+  * Support a predefined tlsf_assert macro
+  * Fix realloc case where block should shrink; if adjacent block is in use, execution would go down the slow path
+
+2007/02/08 - v1.8
+  * Fix for unnecessary reallocation in tlsf_realloc
+
+2007/02/03 - v1.7
+  * tlsf_heap_walk takes a callback
+  * tlsf_realloc now returns NULL on failure
+  * tlsf_memalign optimization for 4-byte alignment
+  * Usage of size_t where appropriate
+
+2006/11/21 - v1.6
+  * ffs/fls broken out into tlsfbits.h
+  * tlsf_overhead queries per-pool overhead
+
+2006/11/07 - v1.5
+  * Smart realloc implementation
+  * Smart memalign implementation
+
+2006/10/11 - v1.4
+  * Add some ffs/fls implementations
+  * Minor code footprint reduction
+
+2006/09/14 - v1.3
+  * Profiling indicates heavy use of blocks of size 1-128, so implement small block handling
+  * Reduce pool overhead by about 1kb
+  * Reduce minimum block size from 32 to 12 bytes
+  * Realloc bug fix
+
+2006/09/09 - v1.2
+  * Add tlsf_block_size
+  * Static assertion mechanism for invariants
+  * Minor bugfixes 
+
+2006/09/01 - v1.1
+  * Add tlsf_realloc
+  * Add tlsf_walk_heap
+
+2006/08/25 - v1.0
+  * First release

vendor/tlsf/tlsf.h

@@ -0,0 +1,91 @@
+#ifndef INCLUDED_tlsf
+#define INCLUDED_tlsf
+
+/*
+** Two Level Segregated Fit memory allocator, version 3.1.
+** Written by Matthew Conte
+**	http://tlsf.baisoku.org
+**
+** Based on the original documentation by Miguel Masmano:
+**	http://www.gii.upv.es/tlsf/main/docs
+**
+** This implementation was written to the specification
+** of the document, therefore no GPL restrictions apply.
+** 
+** Copyright (c) 2006-2016, Matthew Conte
+** All rights reserved.
+** 
+** Redistribution and use in source and binary forms, with or without
+** modification, are permitted provided that the following conditions are met:
+**     * Redistributions of source code must retain the above copyright
+**       notice, this list of conditions and the following disclaimer.
+**     * Redistributions in binary form must reproduce the above copyright
+**       notice, this list of conditions and the following disclaimer in the
+**       documentation and/or other materials provided with the distribution.
+**     * Neither the name of the copyright holder nor the
+**       names of its contributors may be used to endorse or promote products
+**       derived from this software without specific prior written permission.
+** 
+** THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+** ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+** WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+** DISCLAIMED. IN NO EVENT SHALL MATTHEW CONTE BE LIABLE FOR ANY
+** DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+** (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+** LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+** ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+** (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+** SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <stddef.h>
+
+#if defined(__cplusplus)
+extern "C" {
+#endif
+
+/* tlsf_t: a TLSF structure. Can contain 1 to N pools. */
+/* pool_t: a block of memory that TLSF can manage. */
+typedef void* tlsf_t;
+typedef void* pool_t;
+
+/* Create/destroy a memory pool. */
+tlsf_t tlsf_create(void* mem);
+tlsf_t tlsf_create_with_pool(void* mem, size_t bytes);
+void tlsf_destroy(tlsf_t tlsf);
+pool_t tlsf_get_pool(tlsf_t tlsf);
+
+/* Add/remove memory pools. */
+pool_t tlsf_add_pool(tlsf_t tlsf, void* mem, size_t bytes);
+void tlsf_remove_pool(tlsf_t tlsf, pool_t pool);
+
+/* malloc/memalign/realloc/free replacements. */
+void* tlsf_malloc(tlsf_t tlsf, size_t bytes);
+void* tlsf_move(tlsf_t tlsf, void* ptr);
+void* tlsf_memalign(tlsf_t tlsf, size_t align, size_t bytes);
+void* tlsf_realloc(tlsf_t tlsf, void* ptr, size_t size);
+void tlsf_free(tlsf_t tlsf, void* ptr);
+
+/* Returns internal block size, not original request size */
+size_t tlsf_block_size(void* ptr);
+
+/* Overheads/limits of internal structures. */
+size_t tlsf_size(void);
+size_t tlsf_align_size(void);
+size_t tlsf_block_size_min(void);
+size_t tlsf_block_size_max(void);
+size_t tlsf_pool_overhead(void);
+size_t tlsf_alloc_overhead(void);
+
+/* Debugging. */
+typedef void (*tlsf_walker)(void* ptr, size_t size, int used, void* user);
+void tlsf_walk_pool(pool_t pool, tlsf_walker walker, void* user);
+/* Returns nonzero if any internal consistency check fails. */
+int tlsf_check(tlsf_t tlsf);
+int tlsf_check_pool(pool_t pool);
+
+#if defined(__cplusplus)
+};
+#endif
+
+#endif