correct texture address mode, use tlsf instead of sceClibMspace

This commit is contained in:
olebeck 2025-06-29 16:30:20 +02:00
parent 7c15d4f9d8
commit 5005772243
10 changed files with 1639 additions and 171 deletions

View File

@ -61,10 +61,12 @@ if(VITA)
target_sources(miniwin PRIVATE target_sources(miniwin PRIVATE
src/d3drm/backends/gxm/renderer.cpp src/d3drm/backends/gxm/renderer.cpp
src/d3drm/backends/gxm/memory.cpp src/d3drm/backends/gxm/memory.cpp
src/d3drm/backends/gxm/tlsf.c
) )
target_link_libraries(miniwin PRIVATE target_link_libraries(miniwin PRIVATE
SceGxm_stub SceGxm_stub
SceRazorCapture_stub SceRazorCapture_stub
SceRazorHud_stub
) )
list(APPEND GRAPHICS_BACKENDS USE_GXM) list(APPEND GRAPHICS_BACKENDS USE_GXM)
list(REMOVE_ITEM GRAPHICS_BACKENDS USE_SOFTWARE_RENDER USE_SDL_GPU) list(REMOVE_ITEM GRAPHICS_BACKENDS USE_SOFTWARE_RENDER USE_SDL_GPU)

View File

@ -16,5 +16,3 @@ void vita_mem_vertex_usse_free(SceUID uid);
void* vita_mem_fragment_usse_alloc(unsigned int size, SceUID* uid, unsigned int* usse_offset); void* vita_mem_fragment_usse_alloc(unsigned int size, SceUID* uid, unsigned int* usse_offset);
void vita_mem_fragment_usse_free(SceUID uid); void vita_mem_fragment_usse_free(SceUID uid);
bool cdramPool_init();
SceClibMspace cdramPool_get();

View File

@ -1,10 +1,13 @@
#include "memory.h" #include "gxm_memory.h"
#include "utils.h" #include "utils.h"
#include <SDL3/SDL_stdinc.h> #include <SDL3/SDL_stdinc.h>
#include <psp2/gxm.h> #include <psp2/gxm.h>
#include <assert.h>
static SceUID cdramPoolUID = -1; static SceUID cdramPoolUID = -1;
static SceClibMspace cdramPool = NULL; static SceClibMspace cdramPool = NULL;
@ -117,50 +120,3 @@ void vita_mem_fragment_usse_free(SceUID uid)
sceGxmUnmapFragmentUsseMemory(mem); sceGxmUnmapFragmentUsseMemory(mem);
sceKernelFreeMemBlock(uid); sceKernelFreeMemBlock(uid);
} }
bool cdramPool_init()
{
if (cdramPool) {
return true;
}
int poolsize;
int ret;
void* mem;
SceKernelFreeMemorySizeInfo info;
info.size = sizeof(SceKernelFreeMemorySizeInfo);
sceKernelGetFreeMemorySize(&info);
poolsize = ALIGN(info.size_cdram, 256 * 1024);
if (poolsize > info.size_cdram) {
poolsize = ALIGN(info.size_cdram - 256 * 1024, 256 * 1024);
}
poolsize -= 16 * 1024 * 1024;
cdramPoolUID = sceKernelAllocMemBlock("gpu_cdram_pool", SCE_KERNEL_MEMBLOCK_TYPE_USER_CDRAM_RW, poolsize, NULL);
if (cdramPool < 0) {
return false;
}
ret = sceKernelGetMemBlockBase(cdramPoolUID, &mem);
if (ret < 0) {
return false;
}
cdramPool = sceClibMspaceCreate(mem, poolsize);
if (!cdramPool) {
return false;
}
ret = sceGxmMapMemory(
mem,
poolsize,
(SceGxmMemoryAttribFlags) (SCE_GXM_MEMORY_ATTRIB_READ | SCE_GXM_MEMORY_ATTRIB_WRITE)
);
if (ret < 0) {
return false;
}
return true;
}
SceClibMspace cdramPool_get()
{
return cdramPool;
}

View File

@ -0,0 +1,12 @@
#pragma once
#include <psp2/razor_capture.h>
#include <psp2/razor_hud.h>
extern "C" {
extern int sceRazorGpuCaptureSetTrigger(int frames, const char* path);
extern int sceRazorGpuTraceTrigger();
extern int sceRazorGpuTraceSetFilename(const char* filename, int counter);
extern int sceRazorHudSetDisplayEnabled(bool enable);
}

View File

@ -1,7 +1,9 @@
#include "d3drmrenderer_gxm.h" #include "d3drmrenderer_gxm.h"
#include "memory.h" #include "gxm_memory.h"
#include "meshutils.h" #include "meshutils.h"
#include "utils.h" #include "utils.h"
#include "razor.h"
#include "tlsf.h"
#include <SDL3/SDL.h> #include <SDL3/SDL.h>
#include <algorithm> #include <algorithm>
@ -14,6 +16,7 @@
#include "incbin.h" #include "incbin.h"
bool with_razor = false; bool with_razor = false;
bool with_razor_hud = false;
#define VITA_GXM_SCREEN_WIDTH 960 #define VITA_GXM_SCREEN_WIDTH 960
#define VITA_GXM_SCREEN_HEIGHT 544 #define VITA_GXM_SCREEN_HEIGHT 544
@ -23,6 +26,9 @@ bool with_razor = false;
#define VITA_GXM_COLOR_FORMAT SCE_GXM_COLOR_FORMAT_A8B8G8R8 #define VITA_GXM_COLOR_FORMAT SCE_GXM_COLOR_FORMAT_A8B8G8R8
#define VITA_GXM_PIXEL_FORMAT SCE_DISPLAY_PIXELFORMAT_A8B8G8R8 #define VITA_GXM_PIXEL_FORMAT SCE_DISPLAY_PIXELFORMAT_A8B8G8R8
#define CDRAM_POOL_SIZE 64*1024*1024
INCBIN(main_vert_gxp, "shaders/main.vert.gxp"); INCBIN(main_vert_gxp, "shaders/main.vert.gxp");
INCBIN(main_frag_gxp, "shaders/main.frag.gxp"); INCBIN(main_frag_gxp, "shaders/main.frag.gxp");
INCBIN(color_frag_gxp, "shaders/color.frag.gxp"); INCBIN(color_frag_gxp, "shaders/color.frag.gxp");
@ -53,9 +59,6 @@ static const SceGxmBlendInfo blendInfoTransparent = {
.alphaDst = SCE_GXM_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA, .alphaDst = SCE_GXM_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA,
}; };
extern "C" int sceRazorGpuCaptureSetTrigger(int frames, const char* path);
extern "C" int sceRazorGpuCaptureEnableSalvage(const char* path);
extern "C" int sceRazorGpuCaptureSetTriggerNextFrame(const char* path);
static GXMRendererContext gxm_renderer_context; static GXMRendererContext gxm_renderer_context;
@ -77,12 +80,19 @@ static void display_callback(const void* callback_data)
static void load_razor() static void load_razor()
{ {
int mod_id = _sceKernelLoadModule("app0:librazorcapture_es4.suprx", 0, nullptr);
int status; int status;
int mod_id = _sceKernelLoadModule("app0:librazorcapture_es4.suprx", 0, nullptr);
if (!SCE_ERR(sceKernelStartModule, mod_id, 0, nullptr, 0, nullptr, &status)) { if (!SCE_ERR(sceKernelStartModule, mod_id, 0, nullptr, 0, nullptr, &status)) {
with_razor = true; with_razor = true;
} }
/*
int mod_id_hud = _sceKernelLoadModule("app0:librazorhud_es4.suprx", 0, nullptr);
if (!SCE_ERR(sceKernelStartModule, mod_id_hud, 0, nullptr, 0, nullptr, &status)) {
with_razor_hud = true;
}
*/
if (with_razor) { if (with_razor) {
sceRazorGpuCaptureEnableSalvage("ux0:data/gpu_crash.sgx"); sceRazorGpuCaptureEnableSalvage("ux0:data/gpu_crash.sgx");
} }
@ -111,8 +121,64 @@ bool gxm_init()
return err; return err;
} }
gxm_initialized = true; gxm_initialized = true;
return true;
}
return cdramPool_init(); static SceUID cdramAllocatorUID = -1;
static tlsf_t cdramAllocator = nullptr;
bool cdram_allocator_create() {
int ret;
ret = sceKernelAllocMemBlock(
"gpu_cdram_pool",
SCE_KERNEL_MEMBLOCK_TYPE_USER_CDRAM_RW,
CDRAM_POOL_SIZE,
NULL
);
if (ret < 0) {
sceClibPrintf("sceKernelAllocMemBlock failed: %08x\n", ret);
return false;
}
cdramAllocatorUID = ret;
void* mem;
ret = sceKernelGetMemBlockBase(cdramAllocatorUID, &mem);
if (ret < 0) {
sceClibPrintf("sceKernelGetMemBlockBase failed: %08x\n", ret);
return false;
}
ret = sceGxmMapMemory(
mem,
CDRAM_POOL_SIZE,
(SceGxmMemoryAttribFlags) (SCE_GXM_MEMORY_ATTRIB_READ | SCE_GXM_MEMORY_ATTRIB_WRITE)
);
if (ret < 0) {
sceClibPrintf("sceGxmMapMemory failed: %08x\n", ret);
return false;
}
cdramAllocator = SDL_malloc(tlsf_size());
tlsf_create(cdramAllocator);
tlsf_add_pool(cdramAllocator, mem, CDRAM_POOL_SIZE);
return true;
}
int inuse_mem = 0;
inline void* cdram_alloc(size_t size, size_t align) {
sceClibPrintf("cdram_alloc(%d, %d) inuse=%d ", size, align, inuse_mem);
void* ptr = tlsf_memalign(cdramAllocator, align, size);
sceClibPrintf("ptr=%p\n", ptr);
inuse_mem += tlsf_block_size(ptr);
return ptr;
}
inline void cdram_free(void* ptr) {
inuse_mem -= tlsf_block_size(ptr);
sceClibPrintf("cdram_free(%p)\n", ptr);
tlsf_free(cdramAllocator, ptr);
} }
static bool create_gxm_context() static bool create_gxm_context()
@ -130,12 +196,6 @@ static bool create_gxm_context()
const unsigned int patcherVertexUsseSize = 64 * 1024; const unsigned int patcherVertexUsseSize = 64 * 1024;
const unsigned int patcherFragmentUsseSize = 64 * 1024; const unsigned int patcherFragmentUsseSize = 64 * 1024;
data->cdramPool = cdramPool_get();
if (!data->cdramPool) {
SDL_Log("failed to allocate cdramPool");
return false;
}
// allocate buffers // allocate buffers
data->vdmRingBuffer = vita_mem_alloc( data->vdmRingBuffer = vita_mem_alloc(
SCE_KERNEL_MEMBLOCK_TYPE_USER_RW_UNCACHE, SCE_KERNEL_MEMBLOCK_TYPE_USER_RW_UNCACHE,
@ -242,14 +302,13 @@ static void destroy_gxm_context()
SDL_free(gxm_renderer_context.contextHostMem); SDL_free(gxm_renderer_context.contextHostMem);
} }
bool get_gxm_context(SceGxmContext** context, SceGxmShaderPatcher** shaderPatcher, SceClibMspace* cdramPool) bool get_gxm_context(SceGxmContext** context, SceGxmShaderPatcher** shaderPatcher)
{ {
if (!create_gxm_context()) { if (!create_gxm_context()) {
return false; return false;
} }
*context = gxm_renderer_context.context; *context = gxm_renderer_context.context;
*shaderPatcher = gxm_renderer_context.shaderPatcher; *shaderPatcher = gxm_renderer_context.shaderPatcher;
*cdramPool = gxm_renderer_context.cdramPool;
return true; return true;
} }
@ -284,8 +343,6 @@ static void CreateOrthoMatrix(float left, float right, float bottom, float top,
Direct3DRMRenderer* GXMRenderer::Create(DWORD width, DWORD height) Direct3DRMRenderer* GXMRenderer::Create(DWORD width, DWORD height)
{ {
SDL_Log("GXMRenderer::Create width=%d height=%d", width, height);
bool success = gxm_init(); bool success = gxm_init();
if (!success) { if (!success) {
return nullptr; return nullptr;
@ -306,7 +363,12 @@ GXMRenderer::GXMRenderer(DWORD width, DWORD height)
const unsigned int sampleCount = alignedWidth * alignedHeight; const unsigned int sampleCount = alignedWidth * alignedHeight;
const unsigned int depthStrideInSamples = alignedWidth; const unsigned int depthStrideInSamples = alignedWidth;
if (!get_gxm_context(&this->context, &this->shaderPatcher, &this->cdramPool)) { if (!get_gxm_context(&this->context, &this->shaderPatcher)) {
return;
}
if(!cdram_allocator_create()) {
sceClibPrintf("failed to create cdram allocator");
return; return;
} }
@ -533,11 +595,11 @@ GXMRenderer::GXMRenderer(DWORD width, DWORD height)
// clear uniforms // clear uniforms
this->colorShader_uColor = sceGxmProgramFindParameterByName(colorFragmentProgramGxp, "uColor"); // vec4 this->colorShader_uColor = sceGxmProgramFindParameterByName(colorFragmentProgramGxp, "uColor"); // vec4
this->lights = static_cast<decltype(this->lights)>(sceClibMspaceMalloc(this->cdramPool, sizeof(*this->lights))); this->lights = static_cast<decltype(this->lights)>(cdram_alloc(sizeof(*this->lights), 4));
for (int i = 0; i < VITA_GXM_UNIFORM_BUFFER_COUNT; i++) { for (int i = 0; i < VITA_GXM_UNIFORM_BUFFER_COUNT; i++) {
this->quadVertices[i] = (Vertex*) sceClibMspaceMalloc(this->cdramPool, sizeof(Vertex) * 4 * 50); this->quadVertices[i] = static_cast<Vertex*>(cdram_alloc(sizeof(Vertex) * 4 * 50, 4));
} }
this->quadIndices = (uint16_t*) sceClibMspaceMalloc(this->cdramPool, sizeof(uint16_t) * 4); this->quadIndices = static_cast<uint16_t*>(cdram_alloc(sizeof(uint16_t) * 4, 4));
this->quadIndices[0] = 0; this->quadIndices[0] = 0;
this->quadIndices[1] = 1; this->quadIndices[1] = 1;
this->quadIndices[2] = 2; this->quadIndices[2] = 2;
@ -551,6 +613,21 @@ GXMRenderer::GXMRenderer(DWORD width, DWORD height)
this->vertexNotifications[i].value = 0; this->vertexNotifications[i].value = 0;
} }
int count;
auto ids = SDL_GetGamepads(&count);
for(int i = 0; i < count; i++) {
auto id = ids[i];
auto gamepad = SDL_OpenGamepad(id);
if(gamepad != nullptr) {
this->gamepad = gamepad;
break;
}
}
if(with_razor_hud) {
sceRazorGpuTraceSetFilename("ux0:data/gpu_trace", 3);
}
m_initialized = true; m_initialized = true;
} }
@ -600,7 +677,9 @@ void GXMRenderer::AddTextureDestroyCallback(Uint32 id, IDirect3DRMTexture* textu
auto* ctx = static_cast<TextureDestroyContextGXM*>(arg); auto* ctx = static_cast<TextureDestroyContextGXM*>(arg);
auto& cache = ctx->renderer->m_textures[ctx->textureId]; auto& cache = ctx->renderer->m_textures[ctx->textureId];
void* textureData = sceGxmTextureGetData(&cache.gxmTexture); void* textureData = sceGxmTextureGetData(&cache.gxmTexture);
sceClibMspaceFree(ctx->renderer->cdramPool, textureData); cdram_free(textureData);
cache.texture = nullptr;
memset(&cache.gxmTexture, 0, sizeof(cache.gxmTexture));
delete ctx; delete ctx;
}, },
ctx ctx
@ -610,43 +689,92 @@ void GXMRenderer::AddTextureDestroyCallback(Uint32 id, IDirect3DRMTexture* textu
static void convertTextureMetadata( static void convertTextureMetadata(
SDL_Surface* surface, SDL_Surface* surface,
bool* supportedFormat, bool* supportedFormat,
SceGxmTextureFormat* textureFormat, SceGxmTextureFormat* gxmTextureFormat,
size_t* textureSize, size_t* textureSize, // size in bytes
size_t* textureAlignment, size_t* textureAlignment, // alignment in bytes
size_t* textureStride size_t* textureStride, // stride in bytes
size_t* paletteOffset // offset from textureData in bytes
) )
{ {
*supportedFormat = true; int bytesPerPixel;
*textureAlignment = SCE_GXM_TEXTURE_ALIGNMENT; size_t extraDataSize = 0;
switch (surface->format) { switch (surface->format) {
case SDL_PIXELFORMAT_ABGR8888: { case SDL_PIXELFORMAT_INDEX8: {
*textureFormat = SCE_GXM_TEXTURE_FORMAT_U8U8U8U8_ABGR; *supportedFormat = true;
*textureSize = surface->h * surface->pitch; *gxmTextureFormat = SCE_GXM_TEXTURE_FORMAT_P8_ABGR;
*textureStride = surface->pitch; int pixelsSize = surface->w * surface->h;
break; int alignBytes = ALIGNMENT(pixelsSize, SCE_GXM_PALETTE_ALIGNMENT);
} extraDataSize = alignBytes + 256 * 4;
/* *textureAlignment = SCE_GXM_PALETTE_ALIGNMENT;
case SDL_PIXELFORMAT_INDEX8: { *paletteOffset = pixelsSize + alignBytes;
*textureFormat = SCE_GXM_TEXTURE_FORMAT_P8_ABGR; bytesPerPixel = 1;
int pixelsSize = surface->h * surface->pitch; break;
int alignBytes = ALIGNMENT(pixelsSize, SCE_GXM_PALETTE_ALIGNMENT); }
*textureSize = pixelsSize + alignBytes + 0xff; case SDL_PIXELFORMAT_ABGR8888: {
*textureAlignment = SCE_GXM_PALETTE_ALIGNMENT; *supportedFormat = true;
*textureStride = surface->pitch; *gxmTextureFormat = SCE_GXM_TEXTURE_FORMAT_U8U8U8U8_ABGR;
break; *textureAlignment = SCE_GXM_TEXTURE_ALIGNMENT;
} bytesPerPixel = 4;
*/ break;
default: { }
*supportedFormat = false; default: {
} *supportedFormat = false;
*gxmTextureFormat = SCE_GXM_TEXTURE_FORMAT_U8U8U8U8_ABGR;
*textureAlignment = SCE_GXM_TEXTURE_ALIGNMENT;
bytesPerPixel = 4;
break;
}
} }
*textureStride = ALIGN(surface->w, 8)*bytesPerPixel;
*textureSize = (*textureStride)*surface->h+extraDataSize;
} }
void copySurfaceTo(SDL_Surface* src, void* dstData, size_t textureStride) void copySurfaceToGxm(DirectDrawSurfaceImpl* surface, uint8_t* textureData, size_t dstStride, size_t dstSize)
{ {
SDL_Surface* dst = SDL_CreateSurfaceFrom(src->w, src->h, SDL_PIXELFORMAT_ABGR8888, dstData, textureStride); SDL_Surface* src = surface->m_surface;
SDL_BlitSurface(src, nullptr, dst, nullptr); switch(src->format) {
SDL_DestroySurface(dst); case SDL_PIXELFORMAT_ABGR8888: {
for(int y = 0; y < src->h; y++) {
uint8_t* srcRow = (uint8_t*)src->pixels + (y*src->pitch);
uint8_t* dstRow = textureData + (y*dstStride);
size_t rowSize = src->w*4;
if((dstRow - textureData)+rowSize > dstSize) {
sceClibPrintf("buffer overrun!!! size=%d y=%d rowSize=%d\n", dstSize, y, rowSize);
}
memcpy(dstRow, srcRow, rowSize);
}
break;
}
case SDL_PIXELFORMAT_INDEX8: {
LPDIRECTDRAWPALETTE _palette;
surface->GetPalette(&_palette);
auto palette = static_cast<DirectDrawPaletteImpl*>(_palette);
// copy pixels
for(int y = 0; y < src->h; y++) {
void* srcRow = static_cast<uint8_t*>(src->pixels) + (y*src->pitch);
void* dstRow = static_cast<uint8_t*>(textureData) + (y*dstStride);
memcpy(dstRow, srcRow, src->w);
}
int pixelsSize = src->w * src->h;
int alignBytes = ALIGNMENT(pixelsSize, SCE_GXM_PALETTE_ALIGNMENT);
uint8_t* paletteData = textureData + pixelsSize + alignBytes;
memcpy(paletteData, palette->m_palette->colors, 256 * 4);
if((paletteData-textureData) + 256*4 > dstSize) {
sceClibPrintf("buffer overrun!!! textureData=%p paletteData=%p size=%d\n", textureData, paletteData, dstSize);
}
palette->Release();
break;
}
default: {
sceClibPrintf("unsupported format %d\n", SDL_GetPixelFormatName(src->format));
SDL_Surface* dst = SDL_CreateSurfaceFrom(src->w, src->h, SDL_PIXELFORMAT_ABGR8888, textureData, src->w*4);
SDL_BlitSurface(src, nullptr, dst, nullptr);
SDL_DestroySurface(dst);
break;
}
}
} }
Uint32 GXMRenderer::GetTextureId(IDirect3DRMTexture* iTexture) Uint32 GXMRenderer::GetTextureId(IDirect3DRMTexture* iTexture)
@ -655,27 +783,27 @@ Uint32 GXMRenderer::GetTextureId(IDirect3DRMTexture* iTexture)
auto surface = static_cast<DirectDrawSurfaceImpl*>(texture->m_surface); auto surface = static_cast<DirectDrawSurfaceImpl*>(texture->m_surface);
bool supportedFormat; bool supportedFormat;
SceGxmTextureFormat gxmTextureFormat;
size_t textureSize; size_t textureSize;
size_t textureAlignment; size_t textureAlignment;
size_t textureStride; size_t textureStride;
SceGxmTextureFormat textureFormat; size_t paletteOffset;
int textureWidth = surface->m_surface->w; int textureWidth = surface->m_surface->w;
int textureHeight = surface->m_surface->h; int textureHeight = surface->m_surface->h;
convertTextureMetadata( convertTextureMetadata(
surface->m_surface, surface->m_surface,
&supportedFormat, &supportedFormat,
&textureFormat, &gxmTextureFormat,
&textureSize, &textureSize,
&textureAlignment, &textureAlignment,
&textureStride &textureStride,
&paletteOffset
); );
if (!supportedFormat) { if(!supportedFormat) {
textureAlignment = SCE_GXM_TEXTURE_ALIGNMENT; return NO_TEXTURE_ID;
textureStride = textureWidth * 4;
textureSize = textureHeight * textureStride;
textureFormat = SCE_GXM_TEXTURE_FORMAT_U8U8U8U8_ABGR;
} }
for (Uint32 i = 0; i < m_textures.size(); ++i) { for (Uint32 i = 0; i < m_textures.size(); ++i) {
@ -683,69 +811,42 @@ Uint32 GXMRenderer::GetTextureId(IDirect3DRMTexture* iTexture)
if (tex.texture == texture) { if (tex.texture == texture) {
if (tex.version != texture->m_version) { if (tex.version != texture->m_version) {
void* textureData = sceGxmTextureGetData(&tex.gxmTexture); void* textureData = sceGxmTextureGetData(&tex.gxmTexture);
if (!supportedFormat) { copySurfaceToGxm(surface, (uint8_t*)textureData, textureStride, textureSize);
copySurfaceTo(surface->m_surface, textureData, textureStride);
}
else {
memcpy(textureData, surface->m_surface->pixels, textureSize);
}
tex.version = texture->m_version; tex.version = texture->m_version;
} }
return i; return i;
} }
} }
SDL_Log( sceClibPrintf("Create Texture %s w=%d h=%d s=%d size=%d align=%d\n",
"Create Texture %s w=%d h=%d s=%d",
SDL_GetPixelFormatName(surface->m_surface->format), SDL_GetPixelFormatName(surface->m_surface->format),
textureWidth, textureWidth,
textureHeight, textureHeight,
textureStride textureStride,
textureSize,
textureAlignment
); );
// allocate gpu memory // allocate gpu memory
void* textureData = sceClibMspaceMemalign(this->cdramPool, textureAlignment, textureSize); void* textureData = cdram_alloc(textureSize, textureAlignment);
uint8_t* paletteData = nullptr; copySurfaceToGxm(surface, (uint8_t*)textureData, textureStride, textureSize);
if (!supportedFormat) {
SDL_Log(
"unsupported SDL texture format %s, falling back on SDL_PIXELFORMAT_ABGR8888",
SDL_GetPixelFormatName(surface->m_surface->format)
);
copySurfaceTo(surface->m_surface, textureData, textureStride);
}
else if (surface->m_surface->format == SDL_PIXELFORMAT_INDEX8) {
LPDIRECTDRAWPALETTE _palette;
surface->GetPalette(&_palette);
auto palette = static_cast<DirectDrawPaletteImpl*>(_palette);
int pixelsSize = surface->m_surface->w * surface->m_surface->h;
int alignBytes = ALIGNMENT(pixelsSize, SCE_GXM_PALETTE_ALIGNMENT);
SDL_Log("copying indexed texture data from=%p to=%p", surface->m_surface->pixels, textureData);
memcpy(textureData, surface->m_surface->pixels, pixelsSize);
paletteData = (uint8_t*) textureData + pixelsSize + alignBytes;
memcpy(paletteData, palette->m_palette->colors, palette->m_palette->ncolors * sizeof(SDL_Color));
}
else {
SDL_Log("copying texture data from=%p to=%p", surface->m_surface->pixels, textureData);
memcpy(textureData, surface->m_surface->pixels, textureSize);
}
SceGxmTexture gxmTexture; SceGxmTexture gxmTexture;
SCE_ERR( SCE_ERR(
sceGxmTextureInitLinearStrided, sceGxmTextureInitLinear,
&gxmTexture, &gxmTexture,
textureData, textureData,
textureFormat, gxmTextureFormat,
textureWidth, textureWidth,
textureHeight, textureHeight,
textureStride 0
); );
// sceGxmTextureSetMinFilter(&gxmTexture, SCE_GXM_TEXTURE_FILTER_LINEAR); sceGxmTextureSetMinFilter(&gxmTexture, SCE_GXM_TEXTURE_FILTER_LINEAR);
sceGxmTextureSetMagFilter(&gxmTexture, SCE_GXM_TEXTURE_FILTER_LINEAR); sceGxmTextureSetMagFilter(&gxmTexture, SCE_GXM_TEXTURE_FILTER_LINEAR);
if (paletteData) { sceGxmTextureSetUAddrMode(&gxmTexture, SCE_GXM_TEXTURE_ADDR_REPEAT);
sceGxmTextureSetPalette(&gxmTexture, paletteData); sceGxmTextureSetVAddrMode(&gxmTexture, SCE_GXM_TEXTURE_ADDR_REPEAT);
if (gxmTextureFormat == SCE_GXM_TEXTURE_FORMAT_P8_ABGR) {
sceGxmTextureSetPalette(&gxmTexture, (uint8_t*)textureData + paletteOffset);
} }
for (Uint32 i = 0; i < m_textures.size(); ++i) { for (Uint32 i = 0; i < m_textures.size(); ++i) {
@ -754,7 +855,6 @@ Uint32 GXMRenderer::GetTextureId(IDirect3DRMTexture* iTexture)
tex.texture = texture; tex.texture = texture;
tex.version = texture->m_version; tex.version = texture->m_version;
tex.gxmTexture = gxmTexture; tex.gxmTexture = gxmTexture;
tex.textureSize = textureSize;
AddTextureDestroyCallback(i, texture); AddTextureDestroyCallback(i, texture);
return i; return i;
} }
@ -795,7 +895,7 @@ GXMMeshCacheEntry GXMRenderer::GXMUploadMesh(const MeshGroup& meshGroup)
size_t vertexBufferSize = sizeof(Vertex) * vertices.size(); size_t vertexBufferSize = sizeof(Vertex) * vertices.size();
size_t indexBufferSize = sizeof(uint16_t) * indices.size(); size_t indexBufferSize = sizeof(uint16_t) * indices.size();
void* meshData = sceClibMspaceMemalign(this->cdramPool, 4, vertexBufferSize + indexBufferSize); void* meshData = cdram_alloc(vertexBufferSize + indexBufferSize, 4);
Vertex* vertexBuffer = (Vertex*) meshData; Vertex* vertexBuffer = (Vertex*) meshData;
uint16_t* indexBuffer = (uint16_t*) ((uint8_t*) meshData + vertexBufferSize); uint16_t* indexBuffer = (uint16_t*) ((uint8_t*) meshData + vertexBufferSize);
@ -845,7 +945,11 @@ void GXMRenderer::AddMeshDestroyCallback(Uint32 id, IDirect3DRMMesh* mesh)
auto* ctx = static_cast<GXMMeshDestroyContext*>(arg); auto* ctx = static_cast<GXMMeshDestroyContext*>(arg);
auto& cache = ctx->renderer->m_meshes[ctx->id]; auto& cache = ctx->renderer->m_meshes[ctx->id];
cache.meshGroup = nullptr; cache.meshGroup = nullptr;
sceClibMspaceFree(ctx->renderer->cdramPool, cache.meshData); cdram_free(cache.meshData);
cache.meshData = nullptr;
cache.indexBuffer = nullptr;
cache.vertexBuffer = nullptr;
cache.indexCount = 0;
delete ctx; delete ctx;
}, },
ctx ctx
@ -900,12 +1004,59 @@ const char* GXMRenderer::GetName()
} }
bool razor_triggered = false; bool razor_triggered = false;
bool razor_live_started = false;
bool razor_display_enabled = true;
void GXMRenderer::StartScene() void GXMRenderer::StartScene()
{ {
if (sceneStarted) { if (sceneStarted) {
return; return;
} }
bool dpad_up = SDL_GetGamepadButton(this->gamepad, SDL_GAMEPAD_BUTTON_DPAD_UP);
bool dpad_down = SDL_GetGamepadButton(this->gamepad, SDL_GAMEPAD_BUTTON_DPAD_DOWN);
bool dpad_left = SDL_GetGamepadButton(this->gamepad, SDL_GAMEPAD_BUTTON_DPAD_LEFT);
bool dpad_right = SDL_GetGamepadButton(this->gamepad, SDL_GAMEPAD_BUTTON_DPAD_RIGHT);
// hud display
if(with_razor_hud && dpad_up != this->button_dpad_up) {
this->button_dpad_up = dpad_up;
if(dpad_up) {
sceRazorHudSetDisplayEnabled(razor_display_enabled);
}
}
// capture frame
if(with_razor && dpad_down != this->button_dpad_down) {
this->button_dpad_down = dpad_down;
if(dpad_down) {
sceRazorGpuCaptureSetTriggerNextFrame("ux0:/data/capture.sgx");
SDL_Log("trigger razor");
}
}
// toggle live
if(with_razor_hud && dpad_left != this->button_dpad_left) {
this->button_dpad_left = dpad_left;
if(dpad_left) {
if(razor_live_started) {
sceRazorGpuLiveStop();
razor_live_started = false;
} else {
sceRazorGpuLiveStart();
razor_live_started = true;
}
}
}
// trigger trace
if(with_razor_hud && dpad_right != this->button_dpad_right) {
this->button_dpad_right = dpad_right;
if(dpad_right) {
sceRazorGpuTraceTrigger();
}
}
sceGxmBeginScene( sceGxmBeginScene(
this->context, this->context,
0, 0,
@ -922,22 +1073,11 @@ void GXMRenderer::StartScene()
// wait for this uniform buffer to become available // wait for this uniform buffer to become available
this->activeUniformBuffer = (this->activeUniformBuffer + 1) % VITA_GXM_UNIFORM_BUFFER_COUNT; this->activeUniformBuffer = (this->activeUniformBuffer + 1) % VITA_GXM_UNIFORM_BUFFER_COUNT;
sceGxmNotificationWait(&this->fragmentNotifications[this->activeUniformBuffer]); sceGxmNotificationWait(&this->fragmentNotifications[this->activeUniformBuffer]);
// sceClibPrintf("this->activeUniformBuffer: %d notification: %d\n", this->activeUniformBuffer,
// this->fragmentNotifications[this->activeUniformBuffer].value);
} }
int frames = 0;
HRESULT GXMRenderer::BeginFrame() HRESULT GXMRenderer::BeginFrame()
{ {
frames++;
if (with_razor) {
if (!razor_triggered && frames == 10) {
SDL_Log("trigger razor");
sceRazorGpuCaptureSetTriggerNextFrame("ux0:/data/capture.sgx");
razor_triggered = true;
}
}
this->transparencyEnabled = false; this->transparencyEnabled = false;
this->StartScene(); this->StartScene();

File diff suppressed because it is too large Load Diff

View File

@ -0,0 +1,90 @@
#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_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

View File

@ -12,6 +12,8 @@
#include <psp2/types.h> #include <psp2/types.h>
#include <psp2/kernel/clib.h> #include <psp2/kernel/clib.h>
#include "gxm_memory.h"
DEFINE_GUID(GXM_GUID, 0x682656F3, 0x0000, 0x0000, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x58, 0x4D); DEFINE_GUID(GXM_GUID, 0x682656F3, 0x0000, 0x0000, 0x00, 0x00, 0x00, 0x00, 0x00, 0x47, 0x58, 0x4D);
#define VITA_GXM_DISPLAY_BUFFER_COUNT 3 #define VITA_GXM_DISPLAY_BUFFER_COUNT 3
@ -21,7 +23,6 @@ struct GXMTextureCacheEntry {
IDirect3DRMTexture* texture; IDirect3DRMTexture* texture;
Uint32 version; Uint32 version;
SceGxmTexture gxmTexture; SceGxmTexture gxmTexture;
size_t textureSize;
}; };
struct GXMMeshCacheEntry { struct GXMMeshCacheEntry {
@ -73,8 +74,6 @@ typedef struct GXMRendererContext {
void* fragmentRingBuffer; void* fragmentRingBuffer;
void* fragmentUsseRingBuffer; void* fragmentUsseRingBuffer;
SceClibMspace cdramPool;
void* contextHostMem; void* contextHostMem;
SceGxmContext* context; SceGxmContext* context;
@ -159,7 +158,6 @@ class GXMRenderer : public Direct3DRMRenderer {
SceGxmContext* context; SceGxmContext* context;
SceGxmShaderPatcher* shaderPatcher; SceGxmShaderPatcher* shaderPatcher;
SceClibMspace cdramPool;
SceGxmRenderTarget* renderTarget; SceGxmRenderTarget* renderTarget;
void* displayBuffers[VITA_GXM_DISPLAY_BUFFER_COUNT]; void* displayBuffers[VITA_GXM_DISPLAY_BUFFER_COUNT];
@ -212,6 +210,14 @@ class GXMRenderer : public Direct3DRMRenderer {
SceGxmNotification vertexNotifications[VITA_GXM_UNIFORM_BUFFER_COUNT]; SceGxmNotification vertexNotifications[VITA_GXM_UNIFORM_BUFFER_COUNT];
SceGxmNotification fragmentNotifications[VITA_GXM_UNIFORM_BUFFER_COUNT]; SceGxmNotification fragmentNotifications[VITA_GXM_UNIFORM_BUFFER_COUNT];
SDL_Gamepad* gamepad;
bool button_dpad_up;
bool button_dpad_down;
bool button_dpad_left;
bool button_dpad_right;
bool m_initialized = false; bool m_initialized = false;
}; };