private void FreePages(ulong Address, ulong PagesCount)
{
ulong EndAddr = Address + PagesCount * KMemoryManager.PageSize;
int BlockIndex = BlockOrdersCount - 1;
ulong AddressRounded = 0;
ulong EndAddrTruncated = 0;
for (; BlockIndex >= 0; BlockIndex--)
{
KMemoryRegionBlock AllocInfo = Blocks[BlockIndex];
int BlockSize = 1 << AllocInfo.Order;
AddressRounded = BitUtils.AlignUp(Address, BlockSize);
EndAddrTruncated = BitUtils.AlignDown(EndAddr, BlockSize);
if (AddressRounded < EndAddrTruncated)
{
break;
}
}
void FreeRegion(ulong CurrAddress)
{
for (int CurrBlockIndex = BlockIndex;
CurrBlockIndex < BlockOrdersCount && CurrAddress != 0;
CurrBlockIndex++)
{
KMemoryRegionBlock Block = Blocks[CurrBlockIndex];
Block.FreeCount++;
ulong FreedBlocks = (CurrAddress - Block.StartAligned) >> Block.Order;
int Index = (int)FreedBlocks;
for (int Level = Block.MaxLevel - 1; Level >= 0; Level--, Index /= 64)
{
long Mask = Block.Masks[Level][Index / 64];
Block.Masks[Level][Index / 64] = Mask | (1L << (Index & 63));
if (Mask != 0)
{
break;
}
}
int BlockSizeDelta = 1 << (Block.NextOrder - Block.Order);
int FreedBlocksTruncated = BitUtils.AlignDown((int)FreedBlocks, BlockSizeDelta);
if (!Block.TryCoalesce(FreedBlocksTruncated, BlockSizeDelta))
{
break;
}
CurrAddress = Block.StartAligned + ((ulong)FreedBlocksTruncated << Block.Order);
}
}
//Free inside aligned region.
ulong BaseAddress = AddressRounded;
while (BaseAddress < EndAddrTruncated)
{
ulong BlockSize = 1UL << Blocks[BlockIndex].Order;
FreeRegion(BaseAddress);
BaseAddress += BlockSize;
}
int NextBlockIndex = BlockIndex - 1;
//Free region between Address and aligned region start.
BaseAddress = AddressRounded;
for (BlockIndex = NextBlockIndex; BlockIndex >= 0; BlockIndex--)
{
ulong BlockSize = 1UL << Blocks[BlockIndex].Order;
while (BaseAddress - BlockSize >= Address)
{
BaseAddress -= BlockSize;
FreeRegion(BaseAddress);
}
}
//Free region between aligned region end and End Address.
BaseAddress = EndAddrTruncated;
for (BlockIndex = NextBlockIndex; BlockIndex >= 0; BlockIndex--)
{
ulong BlockSize = 1UL << Blocks[BlockIndex].Order;
while (BaseAddress + BlockSize <= EndAddr)
{
FreeRegion(BaseAddress);
BaseAddress += BlockSize;
}
}
}
private void FreePages(ulong address, ulong pagesCount)
{
ulong endAddr = address + pagesCount * KMemoryManager.PageSize;
int blockIndex = _blockOrdersCount - 1;
ulong addressRounded = 0;
ulong endAddrTruncated = 0;
for (; blockIndex >= 0; blockIndex--)
{
KMemoryRegionBlock allocInfo = _blocks[blockIndex];
int blockSize = 1 << allocInfo.Order;
addressRounded = BitUtils.AlignUp(address, blockSize);
endAddrTruncated = BitUtils.AlignDown(endAddr, blockSize);
if (addressRounded < endAddrTruncated)
{
break;
}
}
void FreeRegion(ulong currAddress)
{
for (int currBlockIndex = blockIndex;
currBlockIndex < _blockOrdersCount && currAddress != 0;
currBlockIndex++)
{
KMemoryRegionBlock block = _blocks[currBlockIndex];
block.FreeCount++;
ulong freedBlocks = (currAddress - block.StartAligned) >> block.Order;
int index = (int)freedBlocks;
for (int level = block.MaxLevel - 1; level >= 0; level--, index /= 64)
{
long mask = block.Masks[level][index / 64];
block.Masks[level][index / 64] = mask | (1L << (index & 63));
if (mask != 0)
{
break;
}
}
int blockSizeDelta = 1 << (block.NextOrder - block.Order);
int freedBlocksTruncated = BitUtils.AlignDown((int)freedBlocks, blockSizeDelta);
if (!block.TryCoalesce(freedBlocksTruncated, blockSizeDelta))
{
break;
}
currAddress = block.StartAligned + ((ulong)freedBlocksTruncated << block.Order);
}
}
//Free inside aligned region.
ulong baseAddress = addressRounded;
while (baseAddress < endAddrTruncated)
{
ulong blockSize = 1UL << _blocks[blockIndex].Order;
FreeRegion(baseAddress);
baseAddress += blockSize;
}
int nextBlockIndex = blockIndex - 1;
//Free region between Address and aligned region start.
baseAddress = addressRounded;
for (blockIndex = nextBlockIndex; blockIndex >= 0; blockIndex--)
{
ulong blockSize = 1UL << _blocks[blockIndex].Order;
while (baseAddress - blockSize >= address)
{
baseAddress -= blockSize;
FreeRegion(baseAddress);
}
}
//Free region between aligned region end and End Address.
baseAddress = endAddrTruncated;
for (blockIndex = nextBlockIndex; blockIndex >= 0; blockIndex--)
{
ulong blockSize = 1UL << _blocks[blockIndex].Order;
while (baseAddress + blockSize <= endAddr)
{
FreeRegion(baseAddress);
baseAddress += blockSize;
}
}
}
public ResultCode Initialize(ref AudioRendererConfiguration parameter, uint processHandle, CpuAddress workBuffer, ulong workBufferSize, int sessionId, ulong appletResourceId, IVirtualMemoryManager memoryManager)
{
if (!BehaviourContext.CheckValidRevision(parameter.Revision))
{
return(ResultCode.OperationFailed);
}
if (GetWorkBufferSize(ref parameter) > workBufferSize)
{
return(ResultCode.WorkBufferTooSmall);
}
Debug.Assert(parameter.RenderingDevice == AudioRendererRenderingDevice.Dsp && parameter.ExecutionMode == AudioRendererExecutionMode.Auto);
Logger.Info?.Print(LogClass.AudioRenderer, $"Initializing with REV{BehaviourContext.GetRevisionNumber(parameter.Revision)}");
_behaviourContext.SetUserRevision(parameter.Revision);
_sampleRate = parameter.SampleRate;
_sampleCount = parameter.SampleCount;
_mixBufferCount = parameter.MixBufferCount;
_voiceChannelCountMax = Constants.VoiceChannelCountMax;
_upsamplerCount = parameter.SinkCount + parameter.SubMixBufferCount;
_appletResourceId = appletResourceId;
_memoryPoolCount = parameter.EffectCount + parameter.VoiceCount * Constants.VoiceWaveBufferCount;
_executionMode = parameter.ExecutionMode;
_sessionId = sessionId;
MemoryManager = memoryManager;
if (memoryManager is IRefCounted rc)
{
rc.IncrementReferenceCount();
}
WorkBufferAllocator workBufferAllocator;
_workBufferRegion = MemoryManager.GetWritableRegion(workBuffer, (int)workBufferSize);
_workBufferRegion.Memory.Span.Fill(0);
_workBufferMemoryPin = _workBufferRegion.Memory.Pin();
workBufferAllocator = new WorkBufferAllocator(_workBufferRegion.Memory);
PoolMapper poolMapper = new PoolMapper(processHandle, false);
poolMapper.InitializeSystemPool(ref _dspMemoryPoolState, workBuffer, workBufferSize);
_mixBuffer = workBufferAllocator.Allocate <float>(_sampleCount * (_voiceChannelCountMax + _mixBufferCount), 0x10);
if (_mixBuffer.IsEmpty)
{
return(ResultCode.WorkBufferTooSmall);
}
Memory <float> upSamplerWorkBuffer = workBufferAllocator.Allocate <float>(Constants.TargetSampleCount * (_voiceChannelCountMax + _mixBufferCount) * _upsamplerCount, 0x10);
if (upSamplerWorkBuffer.IsEmpty)
{
return(ResultCode.WorkBufferTooSmall);
}
_depopBuffer = workBufferAllocator.Allocate <float>((ulong)BitUtils.AlignUp(parameter.MixBufferCount, Constants.BufferAlignment), Constants.BufferAlignment);
if (_depopBuffer.IsEmpty)
{
return(ResultCode.WorkBufferTooSmall);
}
// Invalidate DSP cache on what was currently allocated with workBuffer.
AudioProcessorMemoryManager.InvalidateDspCache(_dspMemoryPoolState.Translate(workBuffer, workBufferAllocator.Offset), workBufferAllocator.Offset);
Debug.Assert((workBufferAllocator.Offset % Constants.BufferAlignment) == 0);
Memory <VoiceState> voices = workBufferAllocator.Allocate <VoiceState>(parameter.VoiceCount, VoiceState.Alignment);
if (voices.IsEmpty)
{
return(ResultCode.WorkBufferTooSmall);
}
foreach (ref VoiceState voice in voices.Span)
{
voice.Initialize();
}
// A pain to handle as we can't have VoiceState*, use indices to be a bit more safe
Memory <int> sortedVoices = workBufferAllocator.Allocate <int>(parameter.VoiceCount, 0x10);
if (sortedVoices.IsEmpty)
{
return(ResultCode.WorkBufferTooSmall);
}
// Clear memory (use -1 as it's an invalid index)
sortedVoices.Span.Fill(-1);
Memory <VoiceChannelResource> voiceChannelResources = workBufferAllocator.Allocate <VoiceChannelResource>(parameter.VoiceCount, VoiceChannelResource.Alignment);
if (voiceChannelResources.IsEmpty)
{
return(ResultCode.WorkBufferTooSmall);
}
for (uint id = 0; id < voiceChannelResources.Length; id++)
{
ref VoiceChannelResource voiceChannelResource = ref voiceChannelResources.Span[(int)id];
voiceChannelResource.Id = id;
voiceChannelResource.IsUsed = false;
}
public static bool LoadNsos(KernelContext context, Npdm metaData, byte[] arguments = null, params IExecutable[] executables)
{
ulong argsStart = 0;
uint argsSize = 0;
ulong codeStart = metaData.Is64Bit ? 0x8000000UL : 0x200000UL;
uint codeSize = 0;
ulong[] nsoBase = new ulong[executables.Length];
for (int index = 0; index < executables.Length; index++)
{
IExecutable nso = executables[index];
uint textEnd = nso.TextOffset + (uint)nso.Text.Length;
uint roEnd = nso.RoOffset + (uint)nso.Ro.Length;
uint dataEnd = nso.DataOffset + (uint)nso.Data.Length + nso.BssSize;
uint nsoSize = textEnd;
if (nsoSize < roEnd)
{
nsoSize = roEnd;
}
if (nsoSize < dataEnd)
{
nsoSize = dataEnd;
}
nsoSize = BitUtils.AlignUp(nsoSize, KMemoryManager.PageSize);
nsoBase[index] = codeStart + (ulong)codeSize;
codeSize += nsoSize;
if (arguments != null && argsSize == 0)
{
argsStart = (ulong)codeSize;
argsSize = (uint)BitUtils.AlignDown(arguments.Length * 2 + ArgsTotalSize - 1, KMemoryManager.PageSize);
codeSize += argsSize;
}
}
PtcProfiler.StaticCodeStart = codeStart;
PtcProfiler.StaticCodeSize = (ulong)codeSize;
int codePagesCount = (int)(codeSize / KMemoryManager.PageSize);
int personalMmHeapPagesCount = metaData.PersonalMmHeapSize / KMemoryManager.PageSize;
ProcessCreationInfo creationInfo = new ProcessCreationInfo(
metaData.TitleName,
metaData.Version,
metaData.Aci0.TitleId,
codeStart,
codePagesCount,
(ProcessCreationFlags)metaData.ProcessFlags | ProcessCreationFlags.IsApplication,
0,
personalMmHeapPagesCount);
KernelResult result;
KResourceLimit resourceLimit = new KResourceLimit(context);
long applicationRgSize = (long)context.MemoryRegions[(int)MemoryRegion.Application].Size;
result = resourceLimit.SetLimitValue(LimitableResource.Memory, applicationRgSize);
result |= resourceLimit.SetLimitValue(LimitableResource.Thread, 608);
result |= resourceLimit.SetLimitValue(LimitableResource.Event, 700);
result |= resourceLimit.SetLimitValue(LimitableResource.TransferMemory, 128);
result |= resourceLimit.SetLimitValue(LimitableResource.Session, 894);
if (result != KernelResult.Success)
{
Logger.Error?.Print(LogClass.Loader, $"Process initialization failed setting resource limit values.");
return(false);
}
KProcess process = new KProcess(context);
MemoryRegion memoryRegion = (MemoryRegion)((metaData.Acid.Flags >> 2) & 0xf);
if (memoryRegion > MemoryRegion.NvServices)
{
Logger.Error?.Print(LogClass.Loader, $"Process initialization failed due to invalid ACID flags.");
return(false);
}
var processContextFactory = new ArmProcessContextFactory();
result = process.Initialize(
creationInfo,
metaData.Aci0.KernelAccessControl.Capabilities,
resourceLimit,
memoryRegion,
processContextFactory);
if (result != KernelResult.Success)
{
Logger.Error?.Print(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
for (int index = 0; index < executables.Length; index++)
{
Logger.Info?.Print(LogClass.Loader, $"Loading image {index} at 0x{nsoBase[index]:x16}...");
result = LoadIntoMemory(process, executables[index], nsoBase[index]);
if (result != KernelResult.Success)
{
Logger.Error?.Print(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
}
process.DefaultCpuCore = metaData.DefaultCpuId;
result = process.Start(metaData.MainThreadPriority, (ulong)metaData.MainThreadStackSize);
if (result != KernelResult.Success)
{
Logger.Error?.Print(LogClass.Loader, $"Process start returned error \"{result}\".");
return(false);
}
context.Processes.TryAdd(process.Pid, process);
return(true);
}
public void PrepareForReturn()
{
if (Config.Stage == ShaderStage.Fragment)
{
if (Config.OmapDepth)
{
Operand dest = Attribute(AttributeConsts.FragmentOutputDepth);
Operand src = Register(Config.GetDepthRegister(), RegisterType.Gpr);
this.Copy(dest, src);
}
int regIndex = 0;
for (int rtIndex = 0; rtIndex < 8; rtIndex++)
{
OmapTarget target = Config.OmapTargets[rtIndex];
for (int component = 0; component < 4; component++)
{
if (!target.ComponentEnabled(component))
{
continue;
}
int fragmentOutputColorAttr = AttributeConsts.FragmentOutputColorBase + rtIndex * 16;
Operand src = Register(regIndex, RegisterType.Gpr);
// Perform B <-> R swap if needed, for BGRA formats (not supported on OpenGL).
if (component == 0 || component == 2)
{
Operand isBgra = Attribute(AttributeConsts.FragmentOutputIsBgraBase + rtIndex * 4);
Operand lblIsBgra = Label();
Operand lblEnd = Label();
this.BranchIfTrue(lblIsBgra, isBgra);
this.Copy(Attribute(fragmentOutputColorAttr + component * 4), src);
this.Branch(lblEnd);
MarkLabel(lblIsBgra);
this.Copy(Attribute(fragmentOutputColorAttr + (2 - component) * 4), src);
MarkLabel(lblEnd);
}
else
{
this.Copy(Attribute(fragmentOutputColorAttr + component * 4), src);
}
regIndex++;
}
regIndex = BitUtils.AlignUp(regIndex, 4);
}
}
}
public static SizeInfo GetBlockLinearTextureSize(
int width,
int height,
int depth,
int levels,
int layers,
int blockWidth,
int blockHeight,
int bytesPerPixel,
int gobBlocksInY,
int gobBlocksInZ,
int gobBlocksInTileX,
int gpuLayerSize = 0)
{
bool is3D = depth > 1;
int layerSize = 0;
int[] allOffsets = new int[is3D ? Calculate3DOffsetCount(levels, depth) : levels * layers * depth];
int[] mipOffsets = new int[levels];
int[] sliceSizes = new int[levels];
int[] levelSizes = new int[levels];
int mipGobBlocksInY = gobBlocksInY;
int mipGobBlocksInZ = gobBlocksInZ;
int gobWidth = (GobStride / bytesPerPixel) * gobBlocksInTileX;
int gobHeight = gobBlocksInY * GobHeight;
int depthLevelOffset = 0;
for (int level = 0; level < levels; level++)
{
int w = Math.Max(1, width >> level);
int h = Math.Max(1, height >> level);
int d = Math.Max(1, depth >> level);
w = BitUtils.DivRoundUp(w, blockWidth);
h = BitUtils.DivRoundUp(h, blockHeight);
while (h <= (mipGobBlocksInY >> 1) * GobHeight && mipGobBlocksInY != 1)
{
mipGobBlocksInY >>= 1;
}
while (d <= (mipGobBlocksInZ >> 1) && mipGobBlocksInZ != 1)
{
mipGobBlocksInZ >>= 1;
}
int widthInGobs = BitUtils.DivRoundUp(w * bytesPerPixel, GobStride);
int alignment = gobBlocksInTileX;
if (d < gobBlocksInZ || w <= gobWidth || h <= gobHeight)
{
alignment = 1;
}
widthInGobs = BitUtils.AlignUp(widthInGobs, alignment);
int totalBlocksOfGobsInZ = BitUtils.DivRoundUp(d, mipGobBlocksInZ);
int totalBlocksOfGobsInY = BitUtils.DivRoundUp(BitUtils.DivRoundUp(h, GobHeight), mipGobBlocksInY);
int robSize = widthInGobs * mipGobBlocksInY * mipGobBlocksInZ * GobSize;
if (is3D)
{
int gobSize = mipGobBlocksInY * GobSize;
int sliceSize = totalBlocksOfGobsInY * widthInGobs * gobSize;
int baseOffset = layerSize;
int mask = gobBlocksInZ - 1;
for (int z = 0; z < d; z++)
{
int zLow = z & mask;
int zHigh = z & ~mask;
allOffsets[z + depthLevelOffset] = baseOffset + zLow * gobSize + zHigh * sliceSize;
}
}
mipOffsets[level] = layerSize;
sliceSizes[level] = totalBlocksOfGobsInY * robSize;
levelSizes[level] = totalBlocksOfGobsInZ * sliceSizes[level];
layerSize += levelSizes[level];
depthLevelOffset += d;
}
if (layers > 1)
{
layerSize = AlignLayerSize(
layerSize,
height,
depth,
blockHeight,
gobBlocksInY,
gobBlocksInZ,
gobBlocksInTileX);
}
int totalSize;
if (layerSize < gpuLayerSize)
{
totalSize = (layers - 1) * gpuLayerSize + layerSize;
layerSize = gpuLayerSize;
}
else
{
totalSize = layerSize * layers;
}
if (!is3D)
{
for (int layer = 0; layer < layers; layer++)
{
int baseIndex = layer * levels;
int baseOffset = layer * layerSize;
for (int level = 0; level < levels; level++)
{
allOffsets[baseIndex + level] = baseOffset + mipOffsets[level];
}
}
}
return(new SizeInfo(mipOffsets, allOffsets, sliceSizes, levelSizes, depth, levels, layerSize, totalSize, is3D));
}
public readonly int GetMipStride(int level)
{
return(BitUtils.AlignUp(GetLevelWidth(level) * BytesPerPixel, 4));
}
public static void ConvertLinearToBlockLinear(
Span <byte> dst,
int width,
int height,
int stride,
int bytesPerPixel,
int gobBlocksInY,
ReadOnlySpan <byte> data)
{
int gobHeight = gobBlocksInY * GobHeight;
int strideTrunc = BitUtils.AlignDown(width * bytesPerPixel, 16);
int strideTrunc64 = BitUtils.AlignDown(width * bytesPerPixel, 64);
int xStart = strideTrunc / bytesPerPixel;
int inStrideGap = stride - width * bytesPerPixel;
int alignment = GobStride / bytesPerPixel;
int wAligned = BitUtils.AlignUp(width, alignment);
BlockLinearLayout layoutConverter = new BlockLinearLayout(wAligned, height, gobBlocksInY, 1, bytesPerPixel);
unsafe bool Convert <T>(Span <byte> output, ReadOnlySpan <byte> data) where T : unmanaged
{
fixed(byte *outputPtr = output, dataPtr = data)
{
byte *inPtr = dataPtr;
for (int y = 0; y < height; y++)
{
layoutConverter.SetY(y);
for (int x = 0; x < strideTrunc64; x += 64, inPtr += 64)
{
byte *offset = outputPtr + layoutConverter.GetOffsetWithLineOffset64(x);
byte *offset2 = offset + 0x20;
byte *offset3 = offset + 0x100;
byte *offset4 = offset + 0x120;
Vector128 <byte> value = *(Vector128 <byte> *)inPtr;
Vector128 <byte> value2 = *(Vector128 <byte> *)(inPtr + 16);
Vector128 <byte> value3 = *(Vector128 <byte> *)(inPtr + 32);
Vector128 <byte> value4 = *(Vector128 <byte> *)(inPtr + 48);
*(Vector128 <byte> *)offset = value;
*(Vector128 <byte> *)offset2 = value2;
*(Vector128 <byte> *)offset3 = value3;
*(Vector128 <byte> *)offset4 = value4;
}
for (int x = strideTrunc64; x < strideTrunc; x += 16, inPtr += 16)
{
byte *offset = outputPtr + layoutConverter.GetOffsetWithLineOffset16(x);
*(Vector128 <byte> *)offset = *(Vector128 <byte> *)inPtr;
}
for (int x = xStart; x < width; x++, inPtr += bytesPerPixel)
{
byte *offset = outputPtr + layoutConverter.GetOffset(x);
*(T *)offset = *(T *)inPtr;
}
inPtr += inStrideGap;
}
}
return(true);
}
bool _ = bytesPerPixel switch
{
1 => Convert <byte>(dst, data),
2 => Convert <ushort>(dst, data),
4 => Convert <uint>(dst, data),
8 => Convert <ulong>(dst, data),
12 => Convert <Bpp12Pixel>(dst, data),
16 => Convert <Vector128 <byte> >(dst, data),
_ => throw new NotSupportedException($"Unable to convert ${bytesPerPixel} bpp pixel format.")
};
}
public static Span <byte> ConvertBlockLinearToLinear(
int width,
int height,
int depth,
int sliceDepth,
int levels,
int layers,
int blockWidth,
int blockHeight,
int bytesPerPixel,
int gobBlocksInY,
int gobBlocksInZ,
int gobBlocksInTileX,
SizeInfo sizeInfo,
ReadOnlySpan <byte> data)
{
int outSize = GetTextureSize(
width,
height,
depth,
levels,
layers,
blockWidth,
blockHeight,
bytesPerPixel);
Span <byte> output = new byte[outSize];
int outOffs = 0;
int mipGobBlocksInY = gobBlocksInY;
int mipGobBlocksInZ = gobBlocksInZ;
int gobWidth = (GobStride / bytesPerPixel) * gobBlocksInTileX;
int gobHeight = gobBlocksInY * GobHeight;
for (int level = 0; level < levels; level++)
{
int w = Math.Max(1, width >> level);
int h = Math.Max(1, height >> level);
int d = Math.Max(1, depth >> level);
w = BitUtils.DivRoundUp(w, blockWidth);
h = BitUtils.DivRoundUp(h, blockHeight);
while (h <= (mipGobBlocksInY >> 1) * GobHeight && mipGobBlocksInY != 1)
{
mipGobBlocksInY >>= 1;
}
while (d <= (mipGobBlocksInZ >> 1) && mipGobBlocksInZ != 1)
{
mipGobBlocksInZ >>= 1;
}
int strideTrunc = BitUtils.AlignDown(w * bytesPerPixel, 16);
int strideTrunc64 = BitUtils.AlignDown(w * bytesPerPixel, 64);
int xStart = strideTrunc / bytesPerPixel;
int stride = BitUtils.AlignUp(w * bytesPerPixel, HostStrideAlignment);
int outStrideGap = stride - w * bytesPerPixel;
int alignment = gobWidth;
if (d < gobBlocksInZ || w <= gobWidth || h <= gobHeight)
{
alignment = GobStride / bytesPerPixel;
}
int wAligned = BitUtils.AlignUp(w, alignment);
BlockLinearLayout layoutConverter = new BlockLinearLayout(
wAligned,
h,
mipGobBlocksInY,
mipGobBlocksInZ,
bytesPerPixel);
int sd = Math.Max(1, sliceDepth >> level);
unsafe bool Convert <T>(Span <byte> output, ReadOnlySpan <byte> data) where T : unmanaged
{
fixed(byte *outputPtr = output, dataPtr = data)
{
byte *outPtr = outputPtr + outOffs;
for (int layer = 0; layer < layers; layer++)
{
byte *inBaseOffset = dataPtr + (layer * sizeInfo.LayerSize + sizeInfo.GetMipOffset(level));
for (int z = 0; z < sd; z++)
{
layoutConverter.SetZ(z);
for (int y = 0; y < h; y++)
{
layoutConverter.SetY(y);
for (int x = 0; x < strideTrunc64; x += 64, outPtr += 64)
{
byte *offset = inBaseOffset + layoutConverter.GetOffsetWithLineOffset64(x);
byte *offset2 = offset + 0x20;
byte *offset3 = offset + 0x100;
byte *offset4 = offset + 0x120;
Vector128 <byte> value = *(Vector128 <byte> *)offset;
Vector128 <byte> value2 = *(Vector128 <byte> *)offset2;
Vector128 <byte> value3 = *(Vector128 <byte> *)offset3;
Vector128 <byte> value4 = *(Vector128 <byte> *)offset4;
*(Vector128 <byte> *)outPtr = value;
*(Vector128 <byte> *)(outPtr + 16) = value2;
*(Vector128 <byte> *)(outPtr + 32) = value3;
*(Vector128 <byte> *)(outPtr + 48) = value4;
}
for (int x = strideTrunc64; x < strideTrunc; x += 16, outPtr += 16)
{
byte *offset = inBaseOffset + layoutConverter.GetOffsetWithLineOffset16(x);
*(Vector128 <byte> *)outPtr = *(Vector128 <byte> *)offset;
}
for (int x = xStart; x < w; x++, outPtr += bytesPerPixel)
{
byte *offset = inBaseOffset + layoutConverter.GetOffset(x);
*(T *)outPtr = *(T *)offset;
}
outPtr += outStrideGap;
}
}
}
outOffs += stride * h * d * layers;
}
return(true);
}
bool _ = bytesPerPixel switch
{
1 => Convert <byte>(output, data),
2 => Convert <ushort>(output, data),
4 => Convert <uint>(output, data),
8 => Convert <ulong>(output, data),
12 => Convert <Bpp12Pixel>(output, data),
16 => Convert <Vector128 <byte> >(output, data),
_ => throw new NotSupportedException($"Unable to convert ${bytesPerPixel} bpp pixel format.")
};
}
return(output);
}
/// <summary>
/// Determines if a given texture is "safe" for upscaling from its info.
/// Note that this is different from being compatible - this elilinates targets that would have detrimental effects when scaled.
/// </summary>
/// <param name="info">The texture info to check</param>
/// <returns>True if safe</returns>
private static bool UpscaleSafeMode(TextureInfo info)
{
// While upscaling works for all targets defined by IsUpscaleCompatible, we additionally blacklist targets here that
// may have undesirable results (upscaling blur textures) or simply waste GPU resources (upscaling texture atlas).
if (info.Levels > 3)
{
// Textures with more than 3 levels are likely to be game textures, rather than render textures.
// Small textures with full mips are likely to be removed by the next check.
return(false);
}
if (info.Width < 8 || info.Height < 8)
{
// Discount textures with small dimensions.
return(false);
}
int widthAlignment = (info.IsLinear ? Constants.StrideAlignment : Constants.GobAlignment) / info.FormatInfo.BytesPerPixel;
if (!(info.FormatInfo.Format.IsDepthOrStencil() || info.FormatInfo.Components == 1))
{
// Discount square textures that aren't depth-stencil like. (excludes game textures, cubemap faces, most 3D texture LUT, texture atlas)
// Detect if the texture is possibly square. Widths may be aligned, so to remove the uncertainty we align both the width and height.
bool possiblySquare = BitUtils.AlignUp(info.Width, widthAlignment) == BitUtils.AlignUp(info.Height, widthAlignment);
if (possiblySquare)
{
return(false);
}
}
if (info.Height < 360)
{
int aspectWidth = (int)MathF.Ceiling((info.Height / 9f) * 16f);
int aspectMaxWidth = BitUtils.AlignUp(aspectWidth, widthAlignment);
int aspectMinWidth = BitUtils.AlignDown(aspectWidth, widthAlignment);
if (info.Width >= aspectMinWidth && info.Width <= aspectMaxWidth && info.Height < 360)
{
// Targets that are roughly 16:9 can only be rescaled if they're equal to or above 360p. (excludes blur and bloom textures)
return(false);
}
}
return(true);
}
private static int Align4(int value)
{
return(BitUtils.AlignUp(value, 4));
}
public static bool LoadKernelInitalProcess(Horizon system, KernelInitialProcess kip)
{
int endOffset = kip.DataOffset + kip.Data.Length;
if (kip.BssSize != 0)
{
endOffset = kip.BssOffset + kip.BssSize;
}
int codeSize = BitUtils.AlignUp(kip.TextOffset + endOffset, KMemoryManager.PageSize);
int codePagesCount = codeSize / KMemoryManager.PageSize;
ulong codeBaseAddress = kip.Addr39Bits ? 0x8000000UL : 0x200000UL;
ulong codeAddress = codeBaseAddress + (ulong)kip.TextOffset;
int mmuFlags = 0;
if (AslrEnabled)
{
//TODO: Randomization.
mmuFlags |= 0x20;
}
if (kip.Addr39Bits)
{
mmuFlags |= (int)AddressSpaceType.Addr39Bits << 1;
}
if (kip.Is64Bits)
{
mmuFlags |= 1;
}
ProcessCreationInfo creationInfo = new ProcessCreationInfo(
kip.Name,
kip.ProcessCategory,
kip.TitleId,
codeAddress,
codePagesCount,
mmuFlags,
0,
0);
MemoryRegion memRegion = kip.IsService
? MemoryRegion.Service
: MemoryRegion.Application;
KMemoryRegionManager region = system.MemoryRegions[(int)memRegion];
KernelResult result = region.AllocatePages((ulong)codePagesCount, false, out KPageList pageList);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
KProcess process = new KProcess(system);
result = process.InitializeKip(
creationInfo,
kip.Capabilities,
pageList,
system.ResourceLimit,
memRegion);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
result = LoadIntoMemory(process, kip, codeBaseAddress);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
result = process.Start(kip.MainThreadPriority, (ulong)kip.MainThreadStackSize);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process start returned error \"{result}\".");
return(false);
}
system.Processes.Add(process.Pid, process);
return(true);
}
public static SizeInfo GetBlockLinearTextureSize(
int width,
int height,
int depth,
int levels,
int layers,
int blockWidth,
int blockHeight,
int bytesPerPixel,
int gobBlocksInY,
int gobBlocksInZ,
int gobBlocksInTileX)
{
bool is3D = depth > 1;
int layerSize = 0;
int[] allOffsets = new int[levels * layers * depth];
int[] mipOffsets = new int[levels];
int mipGobBlocksInY = gobBlocksInY;
int mipGobBlocksInZ = gobBlocksInZ;
for (int level = 0; level < levels; level++)
{
int w = Math.Max(1, width >> level);
int h = Math.Max(1, height >> level);
int d = Math.Max(1, depth >> level);
w = BitUtils.DivRoundUp(w, blockWidth);
h = BitUtils.DivRoundUp(h, blockHeight);
while (h <= (mipGobBlocksInY >> 1) * GobHeight && mipGobBlocksInY != 1)
{
mipGobBlocksInY >>= 1;
}
while (d <= (mipGobBlocksInZ >> 1) && mipGobBlocksInZ != 1)
{
mipGobBlocksInZ >>= 1;
}
int widthInGobs = BitUtils.AlignUp(BitUtils.DivRoundUp(w * bytesPerPixel, GobStride), gobBlocksInTileX);
int totalBlocksOfGobsInZ = BitUtils.DivRoundUp(d, mipGobBlocksInZ);
int totalBlocksOfGobsInY = BitUtils.DivRoundUp(BitUtils.DivRoundUp(h, GobHeight), mipGobBlocksInY);
int robSize = widthInGobs * mipGobBlocksInY * mipGobBlocksInZ * GobSize;
if (is3D)
{
int gobSize = mipGobBlocksInY * GobSize;
int sliceSize = totalBlocksOfGobsInY * widthInGobs * gobSize;
int baseOffset = layerSize;
int mask = gobBlocksInZ - 1;
for (int z = 0; z < d; z++)
{
int zLow = z & mask;
int zHigh = z & ~mask;
allOffsets[z * levels + level] = baseOffset + zLow * gobSize + zHigh * sliceSize;
}
}
mipOffsets[level] = layerSize;
layerSize += totalBlocksOfGobsInZ * totalBlocksOfGobsInY * robSize;
}
layerSize = AlignLayerSize(
layerSize,
height,
depth,
blockHeight,
gobBlocksInY,
gobBlocksInZ);
if (!is3D)
{
for (int layer = 0; layer < layers; layer++)
{
int baseIndex = layer * levels;
int baseOffset = layer * layerSize;
for (int level = 0; level < levels; level++)
{
allOffsets[baseIndex + level] = baseOffset + mipOffsets[level];
}
}
}
int totalSize = layerSize * layers;
return(new SizeInfo(mipOffsets, allOffsets, levels, layerSize, totalSize));
}
/// <summary>
/// Performs actual copy of the inline data after the transfer is finished.
/// </summary>
private void FinishTransfer()
{
var memoryManager = _channel.MemoryManager;
var data = MemoryMarshal.Cast <int, byte>(_buffer).Slice(0, _size);
if (_isLinear && _lineCount == 1)
{
memoryManager.Write(_dstGpuVa, data);
}
else
{
var dstCalculator = new OffsetCalculator(
_dstWidth,
_dstHeight,
_dstStride,
_isLinear,
_dstGobBlocksInY,
1);
int srcOffset = 0;
for (int y = _dstY; y < _dstY + _lineCount; y++)
{
int x1 = _dstX;
int x2 = _dstX + _lineLengthIn;
int x1Round = BitUtils.AlignUp(_dstX, 16);
int x2Trunc = BitUtils.AlignDown(x2, 16);
int x = x1;
if (x1Round <= x2)
{
for (; x < x1Round; x++, srcOffset++)
{
int dstOffset = dstCalculator.GetOffset(x, y);
ulong dstAddress = _dstGpuVa + (uint)dstOffset;
memoryManager.Write(dstAddress, data[srcOffset]);
}
}
for (; x < x2Trunc; x += 16, srcOffset += 16)
{
int dstOffset = dstCalculator.GetOffset(x, y);
ulong dstAddress = _dstGpuVa + (uint)dstOffset;
memoryManager.Write(dstAddress, MemoryMarshal.Cast <byte, Vector128 <byte> >(data.Slice(srcOffset, 16))[0]);
}
for (; x < x2; x++, srcOffset++)
{
int dstOffset = dstCalculator.GetOffset(x, y);
ulong dstAddress = _dstGpuVa + (uint)dstOffset;
memoryManager.Write(dstAddress, data[srcOffset]);
}
}
}
_finished = true;
_context.AdvanceSequence();
}
public static bool LoadKip(KernelContext context, KipExecutable kip)
{
int endOffset = kip.DataOffset + kip.Data.Length;
if (kip.BssSize != 0)
{
endOffset = kip.BssOffset + kip.BssSize;
}
int codeSize = BitUtils.AlignUp(kip.TextOffset + endOffset, KMemoryManager.PageSize);
int codePagesCount = codeSize / KMemoryManager.PageSize;
ulong codeBaseAddress = (kip.Header.Flags & 0x10) != 0 ? 0x8000000UL : 0x200000UL;
ulong codeAddress = codeBaseAddress + (ulong)kip.TextOffset;
int mmuFlags = 0;
if (AslrEnabled)
{
// TODO: Randomization.
mmuFlags |= 0x20;
}
if ((kip.Header.Flags & 0x10) != 0)
{
mmuFlags |= (int)AddressSpaceType.Addr39Bits << 1;
}
if ((kip.Header.Flags & 0x08) != 0)
{
mmuFlags |= 1;
}
ProcessCreationInfo creationInfo = new ProcessCreationInfo(
kip.Header.Name,
kip.Header.ProcessCategory,
kip.Header.TitleId,
codeAddress,
codePagesCount,
mmuFlags,
0,
0);
MemoryRegion memoryRegion = (kip.Header.Flags & 0x20) != 0
? MemoryRegion.Service
: MemoryRegion.Application;
KMemoryRegionManager region = context.MemoryRegions[(int)memoryRegion];
KernelResult result = region.AllocatePages((ulong)codePagesCount, false, out KPageList pageList);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
KProcess process = new KProcess(context);
result = process.InitializeKip(
creationInfo,
kip.Capabilities,
pageList,
context.ResourceLimit,
memoryRegion);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
result = LoadIntoMemory(process, kip, codeBaseAddress);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
process.DefaultCpuCore = kip.Header.DefaultCore;
result = process.Start(kip.Header.MainThreadPriority, (ulong)kip.Header.Sections[1].Attribute);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process start returned error \"{result}\".");
return(false);
}
context.Processes.TryAdd(process.Pid, process);
return(true);
}
private KernelResult CopyToClient(KMemoryManager memoryManager, List <KBufferDescriptor> list)
{
foreach (KBufferDescriptor desc in list)
{
MemoryState stateMask;
switch (desc.State)
{
case MemoryState.IpcBuffer0: stateMask = MemoryState.IpcSendAllowedType0; break;
case MemoryState.IpcBuffer1: stateMask = MemoryState.IpcSendAllowedType1; break;
case MemoryState.IpcBuffer3: stateMask = MemoryState.IpcSendAllowedType3; break;
default: return(KernelResult.InvalidCombination);
}
MemoryAttribute attributeMask = MemoryAttribute.Borrowed | MemoryAttribute.Uncached;
if (desc.State == MemoryState.IpcBuffer0)
{
attributeMask |= MemoryAttribute.DeviceMapped;
}
ulong clientAddrTruncated = BitUtils.AlignDown(desc.ClientAddress, KMemoryManager.PageSize);
ulong clientAddrRounded = BitUtils.AlignUp(desc.ClientAddress, KMemoryManager.PageSize);
// Check if address is not aligned, in this case we need to perform 2 copies.
if (clientAddrTruncated != clientAddrRounded)
{
ulong copySize = clientAddrRounded - desc.ClientAddress;
if (copySize > desc.Size)
{
copySize = desc.Size;
}
KernelResult result = memoryManager.CopyDataFromCurrentProcess(
desc.ClientAddress,
copySize,
stateMask,
stateMask,
MemoryPermission.ReadAndWrite,
attributeMask,
MemoryAttribute.None,
desc.ServerAddress);
if (result != KernelResult.Success)
{
return(result);
}
}
ulong clientEndAddr = desc.ClientAddress + desc.Size;
ulong serverEndAddr = desc.ServerAddress + desc.Size;
ulong clientEndAddrTruncated = BitUtils.AlignDown(clientEndAddr, KMemoryManager.PageSize);
ulong clientEndAddrRounded = BitUtils.AlignUp(clientEndAddr, KMemoryManager.PageSize);
ulong serverEndAddrTruncated = BitUtils.AlignDown(serverEndAddr, KMemoryManager.PageSize);
if (clientEndAddrTruncated < clientEndAddrRounded &&
(clientAddrTruncated == clientAddrRounded || clientAddrTruncated < clientEndAddrTruncated))
{
KernelResult result = memoryManager.CopyDataFromCurrentProcess(
clientEndAddrTruncated,
clientEndAddr - clientEndAddrTruncated,
stateMask,
stateMask,
MemoryPermission.ReadAndWrite,
attributeMask,
MemoryAttribute.None,
serverEndAddrTruncated);
if (result != KernelResult.Success)
{
return(result);
}
}
}
return(KernelResult.Success);
}
public KernelResult Start(int mainThreadPriority, ulong stackSize)
{
lock (_processLock)
{
if (_state > ProcessState.CreatedAttached)
{
return(KernelResult.InvalidState);
}
if (ResourceLimit != null && !ResourceLimit.Reserve(LimitableResource.Thread, 1))
{
return(KernelResult.ResLimitExceeded);
}
KResourceLimit threadResourceLimit = ResourceLimit;
KResourceLimit memoryResourceLimit = null;
if (_mainThreadStackSize != 0)
{
throw new InvalidOperationException("Trying to start a process with a invalid state!");
}
ulong stackSizeRounded = BitUtils.AlignUp(stackSize, KMemoryManager.PageSize);
ulong neededSize = stackSizeRounded + _imageSize;
//Check if the needed size for the code and the stack will fit on the
//memory usage capacity of this Process. Also check for possible overflow
//on the above addition.
if (neededSize > _memoryUsageCapacity ||
neededSize < stackSizeRounded)
{
threadResourceLimit?.Release(LimitableResource.Thread, 1);
return(KernelResult.OutOfMemory);
}
if (stackSizeRounded != 0 && ResourceLimit != null)
{
memoryResourceLimit = ResourceLimit;
if (!memoryResourceLimit.Reserve(LimitableResource.Memory, stackSizeRounded))
{
threadResourceLimit?.Release(LimitableResource.Thread, 1);
return(KernelResult.ResLimitExceeded);
}
}
KernelResult result;
KThread mainThread = null;
ulong stackTop = 0;
void CleanUpForError()
{
HandleTable.Destroy();
mainThread?.DecrementReferenceCount();
if (_mainThreadStackSize != 0)
{
ulong stackBottom = stackTop - _mainThreadStackSize;
ulong stackPagesCount = _mainThreadStackSize / KMemoryManager.PageSize;
MemoryManager.UnmapForKernel(stackBottom, stackPagesCount, MemoryState.Stack);
_mainThreadStackSize = 0;
}
memoryResourceLimit?.Release(LimitableResource.Memory, stackSizeRounded);
threadResourceLimit?.Release(LimitableResource.Thread, 1);
}
if (stackSizeRounded != 0)
{
ulong stackPagesCount = stackSizeRounded / KMemoryManager.PageSize;
ulong regionStart = MemoryManager.StackRegionStart;
ulong regionSize = MemoryManager.StackRegionEnd - regionStart;
ulong regionPagesCount = regionSize / KMemoryManager.PageSize;
result = MemoryManager.AllocateOrMapPa(
stackPagesCount,
KMemoryManager.PageSize,
0,
false,
regionStart,
regionPagesCount,
MemoryState.Stack,
MemoryPermission.ReadAndWrite,
out ulong stackBottom);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
_mainThreadStackSize += stackSizeRounded;
stackTop = stackBottom + stackSizeRounded;
}
ulong heapCapacity = _memoryUsageCapacity - _mainThreadStackSize - _imageSize;
result = MemoryManager.SetHeapCapacity(heapCapacity);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
HandleTable = new KHandleTable(System);
result = HandleTable.Initialize(Capabilities.HandleTableSize);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
mainThread = new KThread(System);
result = mainThread.Initialize(
_entrypoint,
0,
stackTop,
mainThreadPriority,
DefaultCpuCore,
this);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
result = HandleTable.GenerateHandle(mainThread, out int mainThreadHandle);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
mainThread.SetEntryArguments(0, mainThreadHandle);
ProcessState oldState = _state;
ProcessState newState = _state != ProcessState.Created
? ProcessState.Attached
: ProcessState.Started;
SetState(newState);
//TODO: We can't call KThread.Start from a non-guest thread.
//We will need to make some changes to allow the creation of
//dummy threads that will be used to initialize the current
//thread on KCoreContext so that GetCurrentThread doesn't fail.
/* Result = MainThread.Start();
*
* if (Result != KernelResult.Success)
* {
* SetState(OldState);
*
* CleanUpForError();
* } */
mainThread.Reschedule(ThreadSchedState.Running);
if (result == KernelResult.Success)
{
mainThread.IncrementReferenceCount();
}
mainThread.DecrementReferenceCount();
return(result);
}
}
public static byte[] DecodeBC4(ReadOnlySpan <byte> data, int width, int height, int depth, int levels, int layers, bool signed)
{
int size = 0;
for (int l = 0; l < levels; l++)
{
size += BitUtils.AlignUp(Math.Max(1, width >> l), 4) * Math.Max(1, height >> l) * Math.Max(1, depth >> l) * layers;
}
// Backends currently expect a stride alignment of 4 bytes, so output width must be aligned.
int alignedWidth = BitUtils.AlignUp(width, 4);
byte[] output = new byte[size];
Span <byte> outputSpan = new Span <byte>(output);
ReadOnlySpan <ulong> data64 = MemoryMarshal.Cast <byte, ulong>(data);
Span <byte> tile = stackalloc byte[BlockWidth * BlockHeight];
Span <byte> rPal = stackalloc byte[8];
Span <uint> tileAsUint = MemoryMarshal.Cast <byte, uint>(tile);
Span <uint> outputLine0 = default;
Span <uint> outputLine1 = default;
Span <uint> outputLine2 = default;
Span <uint> outputLine3 = default;
int imageBaseOOffs = 0;
for (int l = 0; l < levels; l++)
{
int w = BitUtils.DivRoundUp(width, BlockWidth);
int h = BitUtils.DivRoundUp(height, BlockHeight);
for (int l2 = 0; l2 < layers; l2++)
{
for (int z = 0; z < depth; z++)
{
for (int y = 0; y < h; y++)
{
int baseY = y * BlockHeight;
int copyHeight = Math.Min(BlockHeight, height - baseY);
int lineBaseOOffs = imageBaseOOffs + baseY * alignedWidth;
if (copyHeight == 4)
{
outputLine0 = MemoryMarshal.Cast <byte, uint>(outputSpan.Slice(lineBaseOOffs));
outputLine1 = MemoryMarshal.Cast <byte, uint>(outputSpan.Slice(lineBaseOOffs + alignedWidth));
outputLine2 = MemoryMarshal.Cast <byte, uint>(outputSpan.Slice(lineBaseOOffs + alignedWidth * 2));
outputLine3 = MemoryMarshal.Cast <byte, uint>(outputSpan.Slice(lineBaseOOffs + alignedWidth * 3));
}
for (int x = 0; x < w; x++)
{
int baseX = x * BlockWidth;
int copyWidth = Math.Min(BlockWidth, width - baseX);
ulong block = data64[0];
rPal[0] = (byte)block;
rPal[1] = (byte)(block >> 8);
if (signed)
{
BCnLerpAlphaSnorm(rPal);
}
else
{
BCnLerpAlphaUnorm(rPal);
}
BCnDecodeTileAlpha(tile, rPal, block >> 16);
if ((copyWidth | copyHeight) == 4)
{
outputLine0[x] = tileAsUint[0];
outputLine1[x] = tileAsUint[1];
outputLine2[x] = tileAsUint[2];
outputLine3[x] = tileAsUint[3];
}
else
{
int pixelBaseOOffs = lineBaseOOffs + baseX;
for (int tY = 0; tY < copyHeight; tY++)
{
tile.Slice(tY * 4, copyWidth).CopyTo(outputSpan.Slice(pixelBaseOOffs + alignedWidth * tY, copyWidth));
}
}
data64 = data64.Slice(1);
}
}
imageBaseOOffs += alignedWidth * height;
}
}
width = Math.Max(1, width >> 1);
height = Math.Max(1, height >> 1);
depth = Math.Max(1, depth >> 1);
alignedWidth = BitUtils.AlignUp(width, 4);
}
return(output);
}
public static ulong GetTargetSize <T>(ulong currentSize, ulong count, int align) where T : unmanaged
{
return(BitUtils.AlignUp(currentSize, align) + (ulong)Unsafe.SizeOf <T>() * count);
}
public static byte[] DecodeBC5(ReadOnlySpan <byte> data, int width, int height, int depth, int levels, int layers, bool signed)
{
int size = 0;
for (int l = 0; l < levels; l++)
{
size += BitUtils.AlignUp(Math.Max(1, width >> l), 2) * Math.Max(1, height >> l) * Math.Max(1, depth >> l) * layers * 2;
}
// Backends currently expect a stride alignment of 4 bytes, so output width must be aligned.
int alignedWidth = BitUtils.AlignUp(width, 2);
byte[] output = new byte[size];
ReadOnlySpan <ulong> data64 = MemoryMarshal.Cast <byte, ulong>(data);
Span <byte> rTile = stackalloc byte[BlockWidth * BlockHeight * 2];
Span <byte> gTile = stackalloc byte[BlockWidth * BlockHeight * 2];
Span <byte> rPal = stackalloc byte[8];
Span <byte> gPal = stackalloc byte[8];
Span <ushort> outputAsUshort = MemoryMarshal.Cast <byte, ushort>(output);
Span <uint> rTileAsUint = MemoryMarshal.Cast <byte, uint>(rTile);
Span <uint> gTileAsUint = MemoryMarshal.Cast <byte, uint>(gTile);
Span <ulong> outputLine0 = default;
Span <ulong> outputLine1 = default;
Span <ulong> outputLine2 = default;
Span <ulong> outputLine3 = default;
int imageBaseOOffs = 0;
for (int l = 0; l < levels; l++)
{
int w = BitUtils.DivRoundUp(width, BlockWidth);
int h = BitUtils.DivRoundUp(height, BlockHeight);
for (int l2 = 0; l2 < layers; l2++)
{
for (int z = 0; z < depth; z++)
{
for (int y = 0; y < h; y++)
{
int baseY = y * BlockHeight;
int copyHeight = Math.Min(BlockHeight, height - baseY);
int lineBaseOOffs = imageBaseOOffs + baseY * alignedWidth;
if (copyHeight == 4)
{
outputLine0 = MemoryMarshal.Cast <ushort, ulong>(outputAsUshort.Slice(lineBaseOOffs));
outputLine1 = MemoryMarshal.Cast <ushort, ulong>(outputAsUshort.Slice(lineBaseOOffs + alignedWidth));
outputLine2 = MemoryMarshal.Cast <ushort, ulong>(outputAsUshort.Slice(lineBaseOOffs + alignedWidth * 2));
outputLine3 = MemoryMarshal.Cast <ushort, ulong>(outputAsUshort.Slice(lineBaseOOffs + alignedWidth * 3));
}
for (int x = 0; x < w; x++)
{
int baseX = x * BlockWidth;
int copyWidth = Math.Min(BlockWidth, width - baseX);
ulong blockL = data64[0];
ulong blockH = data64[1];
rPal[0] = (byte)blockL;
rPal[1] = (byte)(blockL >> 8);
gPal[0] = (byte)blockH;
gPal[1] = (byte)(blockH >> 8);
if (signed)
{
BCnLerpAlphaSnorm(rPal);
BCnLerpAlphaSnorm(gPal);
}
else
{
BCnLerpAlphaUnorm(rPal);
BCnLerpAlphaUnorm(gPal);
}
BCnDecodeTileAlpha(rTile, rPal, blockL >> 16);
BCnDecodeTileAlpha(gTile, gPal, blockH >> 16);
if ((copyWidth | copyHeight) == 4)
{
outputLine0[x] = InterleaveBytes(rTileAsUint[0], gTileAsUint[0]);
outputLine1[x] = InterleaveBytes(rTileAsUint[1], gTileAsUint[1]);
outputLine2[x] = InterleaveBytes(rTileAsUint[2], gTileAsUint[2]);
outputLine3[x] = InterleaveBytes(rTileAsUint[3], gTileAsUint[3]);
}
else
{
int pixelBaseOOffs = lineBaseOOffs + baseX;
for (int tY = 0; tY < copyHeight; tY++)
{
int line = pixelBaseOOffs + alignedWidth * tY;
for (int tX = 0; tX < copyWidth; tX++)
{
int texel = tY * BlockWidth + tX;
outputAsUshort[line + tX] = (ushort)(rTile[texel] | (gTile[texel] << 8));
}
}
}
data64 = data64.Slice(2);
}
}
imageBaseOOffs += alignedWidth * height;
}
}
width = Math.Max(1, width >> 1);
height = Math.Max(1, height >> 1);
depth = Math.Max(1, depth >> 1);
alignedWidth = BitUtils.AlignUp(width, 2);
}
return(output);
}
public KernelResult Start(int mainThreadPriority, ulong stackSize)
{
lock (_processLock)
{
if (State > ProcessState.CreatedAttached)
{
return(KernelResult.InvalidState);
}
if (ResourceLimit != null && !ResourceLimit.Reserve(LimitableResource.Thread, 1))
{
return(KernelResult.ResLimitExceeded);
}
KResourceLimit threadResourceLimit = ResourceLimit;
KResourceLimit memoryResourceLimit = null;
if (_mainThreadStackSize != 0)
{
throw new InvalidOperationException("Trying to start a process with a invalid state!");
}
ulong stackSizeRounded = BitUtils.AlignUp(stackSize, KPageTableBase.PageSize);
ulong neededSize = stackSizeRounded + _imageSize;
// Check if the needed size for the code and the stack will fit on the
// memory usage capacity of this Process. Also check for possible overflow
// on the above addition.
if (neededSize > _memoryUsageCapacity || neededSize < stackSizeRounded)
{
threadResourceLimit?.Release(LimitableResource.Thread, 1);
return(KernelResult.OutOfMemory);
}
if (stackSizeRounded != 0 && ResourceLimit != null)
{
memoryResourceLimit = ResourceLimit;
if (!memoryResourceLimit.Reserve(LimitableResource.Memory, stackSizeRounded))
{
threadResourceLimit?.Release(LimitableResource.Thread, 1);
return(KernelResult.ResLimitExceeded);
}
}
KernelResult result;
KThread mainThread = null;
ulong stackTop = 0;
void CleanUpForError()
{
HandleTable.Destroy();
mainThread?.DecrementReferenceCount();
if (_mainThreadStackSize != 0)
{
ulong stackBottom = stackTop - _mainThreadStackSize;
ulong stackPagesCount = _mainThreadStackSize / KPageTableBase.PageSize;
MemoryManager.UnmapForKernel(stackBottom, stackPagesCount, MemoryState.Stack);
_mainThreadStackSize = 0;
}
memoryResourceLimit?.Release(LimitableResource.Memory, stackSizeRounded);
threadResourceLimit?.Release(LimitableResource.Thread, 1);
}
if (stackSizeRounded != 0)
{
ulong stackPagesCount = stackSizeRounded / KPageTableBase.PageSize;
ulong regionStart = MemoryManager.StackRegionStart;
ulong regionSize = MemoryManager.StackRegionEnd - regionStart;
ulong regionPagesCount = regionSize / KPageTableBase.PageSize;
result = MemoryManager.MapPages(
stackPagesCount,
KPageTableBase.PageSize,
0,
false,
regionStart,
regionPagesCount,
MemoryState.Stack,
KMemoryPermission.ReadAndWrite,
out ulong stackBottom);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
_mainThreadStackSize += stackSizeRounded;
stackTop = stackBottom + stackSizeRounded;
}
ulong heapCapacity = _memoryUsageCapacity - _mainThreadStackSize - _imageSize;
result = MemoryManager.SetHeapCapacity(heapCapacity);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
HandleTable = new KHandleTable(KernelContext);
result = HandleTable.Initialize(Capabilities.HandleTableSize);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
mainThread = new KThread(KernelContext);
result = mainThread.Initialize(
_entrypoint,
0,
stackTop,
mainThreadPriority,
DefaultCpuCore,
this,
ThreadType.User,
_customThreadStart);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
result = HandleTable.GenerateHandle(mainThread, out int mainThreadHandle);
if (result != KernelResult.Success)
{
CleanUpForError();
return(result);
}
mainThread.SetEntryArguments(0, mainThreadHandle);
ProcessState oldState = State;
ProcessState newState = State != ProcessState.Created
? ProcessState.Attached
: ProcessState.Started;
SetState(newState);
result = mainThread.Start();
if (result != KernelResult.Success)
{
SetState(oldState);
CleanUpForError();
}
if (result == KernelResult.Success)
{
mainThread.IncrementReferenceCount();
}
mainThread.DecrementReferenceCount();
return(result);
}
}
// GetWorkBufferSize(nn::audio::detail::AudioRendererParameterInternal) -> u64
public ResultCode GetAudioRendererWorkBufferSize(ServiceCtx context)
{
AudioRendererParameter parameters = GetAudioRendererParameter(context);
if (AudioRendererCommon.CheckValidRevision(parameters))
{
BehaviorInfo behaviorInfo = new BehaviorInfo();
behaviorInfo.SetUserLibRevision(parameters.Revision);
long size;
int totalMixCount = parameters.SubMixCount + 1;
size = BitUtils.AlignUp(parameters.MixBufferCount * 4, AudioRendererConsts.BufferAlignment) +
parameters.SubMixCount * 0x400 +
totalMixCount * 0x940 +
parameters.VoiceCount * 0x3F0 +
BitUtils.AlignUp(totalMixCount * 8, 16) +
BitUtils.AlignUp(parameters.VoiceCount * 8, 16) +
BitUtils.AlignUp(((parameters.SinkCount + parameters.SubMixCount) * 0x3C0 + parameters.SampleCount * 4) *
(parameters.MixBufferCount + 6), AudioRendererConsts.BufferAlignment) +
(parameters.SinkCount + parameters.SubMixCount) * 0x2C0 +
(parameters.EffectCount + parameters.VoiceCount * 4) * 0x30 +
0x50;
if (behaviorInfo.IsSplitterSupported())
{
size += BitUtils.AlignUp(NodeStates.GetWorkBufferSize(totalMixCount) + EdgeMatrix.GetWorkBufferSize(totalMixCount), 16);
}
size = parameters.SinkCount * 0x170 +
(parameters.SinkCount + parameters.SubMixCount) * 0x280 +
parameters.EffectCount * 0x4C0 +
((size + SplitterContext.CalcWorkBufferSize(behaviorInfo, parameters) + 0x30 * parameters.EffectCount + (4 * parameters.VoiceCount) + 0x8F) & ~0x3FL) +
((parameters.VoiceCount << 8) | 0x40);
if (parameters.PerformanceManagerCount >= 1)
{
size += (PerformanceManager.GetRequiredBufferSizeForPerformanceMetricsPerFrame(behaviorInfo, parameters) *
(parameters.PerformanceManagerCount + 1) + 0xFF) & ~0x3FL;
}
if (behaviorInfo.IsVariadicCommandBufferSizeSupported())
{
size += CommandGenerator.CalculateCommandBufferSize(parameters) + 0x7E;
}
else
{
size += 0x1807E;
}
size = BitUtils.AlignUp(size, 0x1000);
context.ResponseData.Write(size);
Logger.PrintDebug(LogClass.ServiceAudio, $"WorkBufferSize is 0x{size:x16}.");
return(ResultCode.Success);
}
else
{
context.ResponseData.Write(0L);
Logger.PrintWarning(LogClass.ServiceAudio, $"Library Revision REV{AudioRendererCommon.GetRevisionVersion(parameters.Revision)} is not supported!");
return(ResultCode.UnsupportedRevision);
}
}
public static bool LoadKip(KernelContext context, KipExecutable kip)
{
uint endOffset = kip.DataOffset + (uint)kip.Data.Length;
if (kip.BssSize != 0)
{
endOffset = kip.BssOffset + kip.BssSize;
}
uint codeSize = BitUtils.AlignUp(kip.TextOffset + endOffset, KMemoryManager.PageSize);
int codePagesCount = (int)(codeSize / KMemoryManager.PageSize);
ulong codeBaseAddress = kip.Is64BitAddressSpace ? 0x8000000UL : 0x200000UL;
ulong codeAddress = codeBaseAddress + (ulong)kip.TextOffset;
ProcessCreationFlags flags = 0;
if (AslrEnabled)
{
// TODO: Randomization.
flags |= ProcessCreationFlags.EnableAslr;
}
if (kip.Is64BitAddressSpace)
{
flags |= ProcessCreationFlags.AddressSpace64Bit;
}
if (kip.Is64Bit)
{
flags |= ProcessCreationFlags.Is64Bit;
}
ProcessCreationInfo creationInfo = new ProcessCreationInfo(
kip.Name,
kip.Version,
kip.ProgramId,
codeAddress,
codePagesCount,
flags,
0,
0);
MemoryRegion memoryRegion = kip.UsesSecureMemory
? MemoryRegion.Service
: MemoryRegion.Application;
KMemoryRegionManager region = context.MemoryRegions[(int)memoryRegion];
KernelResult result = region.AllocatePages((ulong)codePagesCount, false, out KPageList pageList);
if (result != KernelResult.Success)
{
Logger.Error?.Print(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
KProcess process = new KProcess(context);
var processContextFactory = new ArmProcessContextFactory();
result = process.InitializeKip(
creationInfo,
kip.Capabilities,
pageList,
context.ResourceLimit,
memoryRegion,
processContextFactory);
if (result != KernelResult.Success)
{
Logger.Error?.Print(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
result = LoadIntoMemory(process, kip, codeBaseAddress);
if (result != KernelResult.Success)
{
Logger.Error?.Print(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
process.DefaultCpuCore = kip.IdealCoreId;
result = process.Start(kip.Priority, (ulong)kip.StackSize);
if (result != KernelResult.Success)
{
Logger.Error?.Print(LogClass.Loader, $"Process start returned error \"{result}\".");
return(false);
}
context.Processes.TryAdd(process.Pid, process);
return(true);
}
/// <summary>
/// Get the required work buffer size memory needed for the <see cref="EdgeMatrix"/>.
/// </summary>
/// <param name="nodeCount">The count of nodes.</param>
/// <returns>The size required for the given <paramref name="nodeCount"/>.</returns>
public static int GetWorkBufferSize(int nodeCount)
{
int size = BitUtils.AlignUp(nodeCount * nodeCount, RendererConstants.BufferAlignment);
return(size / Unsafe.SizeOf <byte>());
}
public KMemoryRegionManager(ulong address, ulong size, ulong endAddr)
{
_blocks = new KMemoryRegionBlock[BlockOrders.Length];
Address = address;
Size = size;
EndAddr = endAddr;
_blockOrdersCount = BlockOrders.Length;
for (int blockIndex = 0; blockIndex < _blockOrdersCount; blockIndex++)
{
_blocks[blockIndex] = new KMemoryRegionBlock();
_blocks[blockIndex].Order = BlockOrders[blockIndex];
int nextOrder = blockIndex == _blockOrdersCount - 1 ? 0 : BlockOrders[blockIndex + 1];
_blocks[blockIndex].NextOrder = nextOrder;
int currBlockSize = 1 << BlockOrders[blockIndex];
int nextBlockSize = currBlockSize;
if (nextOrder != 0)
{
nextBlockSize = 1 << nextOrder;
}
ulong startAligned = BitUtils.AlignDown(address, nextBlockSize);
ulong endAddrAligned = BitUtils.AlignDown(endAddr, currBlockSize);
ulong sizeInBlocksTruncated = (endAddrAligned - startAligned) >> BlockOrders[blockIndex];
ulong endAddrRounded = BitUtils.AlignUp(address + size, nextBlockSize);
ulong sizeInBlocksRounded = (endAddrRounded - startAligned) >> BlockOrders[blockIndex];
_blocks[blockIndex].StartAligned = startAligned;
_blocks[blockIndex].SizeInBlocksTruncated = sizeInBlocksTruncated;
_blocks[blockIndex].SizeInBlocksRounded = sizeInBlocksRounded;
ulong currSizeInBlocks = sizeInBlocksRounded;
int maxLevel = 0;
do
{
maxLevel++;
}while ((currSizeInBlocks /= 64) != 0);
_blocks[blockIndex].MaxLevel = maxLevel;
_blocks[blockIndex].Masks = new long[maxLevel][];
currSizeInBlocks = sizeInBlocksRounded;
for (int level = maxLevel - 1; level >= 0; level--)
{
currSizeInBlocks = (currSizeInBlocks + 63) / 64;
_blocks[blockIndex].Masks[level] = new long[currSizeInBlocks];
}
}
if (size != 0)
{
FreePages(address, size / KMemoryManager.PageSize);
}
}
private KernelResult GetReceiveListAddress(
PointerBufferDesc descriptor,
Message message,
uint recvListType,
uint messageSizeInWords,
ulong[] receiveList,
ref uint dstOffset,
out ulong address)
{
ulong recvListBufferAddress = address = 0;
if (recvListType == 0)
{
return(KernelResult.OutOfResource);
}
else if (recvListType == 1 || recvListType == 2)
{
ulong recvListBaseAddr;
ulong recvListEndAddr;
if (recvListType == 1)
{
recvListBaseAddr = message.Address + messageSizeInWords * 4;
recvListEndAddr = message.Address + message.Size;
}
else /* if (recvListType == 2) */
{
ulong packed = receiveList[0];
recvListBaseAddr = packed & 0x7fffffffff;
uint size = (uint)(packed >> 48);
if (size == 0)
{
return(KernelResult.OutOfResource);
}
recvListEndAddr = recvListBaseAddr + size;
}
recvListBufferAddress = BitUtils.AlignUp(recvListBaseAddr + dstOffset, 0x10);
ulong endAddress = recvListBufferAddress + descriptor.BufferSize;
dstOffset = (uint)endAddress - (uint)recvListBaseAddr;
if (recvListBufferAddress + descriptor.BufferSize <= recvListBufferAddress ||
recvListBufferAddress + descriptor.BufferSize > recvListEndAddr)
{
return(KernelResult.OutOfResource);
}
}
else /* if (recvListType > 2) */
{
if (descriptor.ReceiveIndex >= receiveList.Length)
{
return(KernelResult.OutOfResource);
}
ulong packed = receiveList[descriptor.ReceiveIndex];
recvListBufferAddress = packed & 0x7fffffffff;
uint size = (uint)(packed >> 48);
if (recvListBufferAddress == 0 || size == 0 || size < descriptor.BufferSize)
{
return(KernelResult.OutOfResource);
}
}
address = recvListBufferAddress;
return(KernelResult.Success);
}
public static Span <byte> ConvertLinearToBlockLinear(
int width,
int height,
int depth,
int levels,
int layers,
int blockWidth,
int blockHeight,
int bytesPerPixel,
int gobBlocksInY,
int gobBlocksInZ,
int gobBlocksInTileX,
SizeInfo sizeInfo,
ReadOnlySpan <byte> data)
{
Span <byte> output = new byte[sizeInfo.TotalSize];
int inOffs = 0;
int mipGobBlocksInY = gobBlocksInY;
int mipGobBlocksInZ = gobBlocksInZ;
int gobWidth = (GobStride / bytesPerPixel) * gobBlocksInTileX;
int gobHeight = gobBlocksInY * GobHeight;
for (int level = 0; level < levels; level++)
{
int w = Math.Max(1, width >> level);
int h = Math.Max(1, height >> level);
int d = Math.Max(1, depth >> level);
w = BitUtils.DivRoundUp(w, blockWidth);
h = BitUtils.DivRoundUp(h, blockHeight);
while (h <= (mipGobBlocksInY >> 1) * GobHeight && mipGobBlocksInY != 1)
{
mipGobBlocksInY >>= 1;
}
while (d <= (mipGobBlocksInZ >> 1) && mipGobBlocksInZ != 1)
{
mipGobBlocksInZ >>= 1;
}
int stride = BitUtils.AlignUp(w * bytesPerPixel, HostStrideAlignment);
int alignment = gobWidth;
if (d < gobBlocksInZ || w <= gobWidth || h <= gobHeight)
{
alignment = GobStride / bytesPerPixel;
}
int wAligned = BitUtils.AlignUp(w, alignment);
BlockLinearLayout layoutConverter = new BlockLinearLayout(
wAligned,
h,
d,
mipGobBlocksInY,
mipGobBlocksInZ,
bytesPerPixel);
for (int layer = 0; layer < layers; layer++)
{
int outBaseOffset = layer * sizeInfo.LayerSize + sizeInfo.GetMipOffset(level);
for (int z = 0; z < d; z++)
{
for (int y = 0; y < h; y++)
{
for (int x = 0; x < w; x++)
{
int offset = outBaseOffset + layoutConverter.GetOffset(x, y, z);
Span <byte> dest = output.Slice(offset, bytesPerPixel);
data.Slice(inOffs + x * bytesPerPixel, bytesPerPixel).CopyTo(dest);
}
inOffs += stride;
}
}
}
}
return(output);
}
public static bool LoadStaticObjects(
Horizon system,
Npdm metaData,
IExecutable[] staticObjects,
byte[] arguments = null)
{
if (!metaData.Is64Bits)
{
Logger.PrintWarning(LogClass.Loader, "32-bits application detected!");
}
ulong argsStart = 0;
int argsSize = 0;
ulong codeStart = metaData.Is64Bits ? 0x8000000UL : 0x200000UL;
int codeSize = 0;
ulong[] nsoBase = new ulong[staticObjects.Length];
for (int index = 0; index < staticObjects.Length; index++)
{
IExecutable staticObject = staticObjects[index];
int textEnd = staticObject.TextOffset + staticObject.Text.Length;
int roEnd = staticObject.RoOffset + staticObject.Ro.Length;
int dataEnd = staticObject.DataOffset + staticObject.Data.Length + staticObject.BssSize;
int nsoSize = textEnd;
if ((uint)nsoSize < (uint)roEnd)
{
nsoSize = roEnd;
}
if ((uint)nsoSize < (uint)dataEnd)
{
nsoSize = dataEnd;
}
nsoSize = BitUtils.AlignUp(nsoSize, KMemoryManager.PageSize);
nsoBase[index] = codeStart + (ulong)codeSize;
codeSize += nsoSize;
if (arguments != null && argsSize == 0)
{
argsStart = (ulong)codeSize;
argsSize = BitUtils.AlignDown(arguments.Length * 2 + ArgsTotalSize - 1, KMemoryManager.PageSize);
codeSize += argsSize;
}
}
int codePagesCount = codeSize / KMemoryManager.PageSize;
int personalMmHeapPagesCount = metaData.PersonalMmHeapSize / KMemoryManager.PageSize;
ProcessCreationInfo creationInfo = new ProcessCreationInfo(
metaData.TitleName,
metaData.ProcessCategory,
metaData.Aci0.TitleId,
codeStart,
codePagesCount,
metaData.MmuFlags,
0,
personalMmHeapPagesCount);
KernelResult result;
KResourceLimit resourceLimit = new KResourceLimit(system);
long applicationRgSize = (long)system.MemoryRegions[(int)MemoryRegion.Application].Size;
result = resourceLimit.SetLimitValue(LimitableResource.Memory, applicationRgSize);
result |= resourceLimit.SetLimitValue(LimitableResource.Thread, 608);
result |= resourceLimit.SetLimitValue(LimitableResource.Event, 700);
result |= resourceLimit.SetLimitValue(LimitableResource.TransferMemory, 128);
result |= resourceLimit.SetLimitValue(LimitableResource.Session, 894);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization failed setting resource limit values.");
return(false);
}
KProcess process = new KProcess(system);
MemoryRegion memoryRegion = (MemoryRegion)((metaData.Acid.Flags >> 2) & 0xf);
if (memoryRegion > MemoryRegion.NvServices)
{
Logger.PrintError(LogClass.Loader, $"Process initialization failed due to invalid ACID flags.");
return(false);
}
result = process.Initialize(
creationInfo,
metaData.Aci0.KernelAccessControl.Capabilities,
resourceLimit,
memoryRegion);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
for (int index = 0; index < staticObjects.Length; index++)
{
Logger.PrintInfo(LogClass.Loader, $"Loading image {index} at 0x{nsoBase[index]:x16}...");
result = LoadIntoMemory(process, staticObjects[index], nsoBase[index]);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process initialization returned error \"{result}\".");
return(false);
}
}
process.DefaultCpuCore = metaData.DefaultCpuId;
result = process.Start(metaData.MainThreadPriority, (ulong)metaData.MainThreadStackSize);
if (result != KernelResult.Success)
{
Logger.PrintError(LogClass.Loader, $"Process start returned error \"{result}\".");
return(false);
}
system.Processes.Add(process.Pid, process);
return(true);
}
/// <summary>
/// Determines if a given texture is "safe" for upscaling from its info.
/// Note that this is different from being compatible - this elilinates targets that would have detrimental effects when scaled.
/// </summary>
/// <param name="info">The texture info to check</param>
/// <returns>True if safe</returns>
public bool UpscaleSafeMode(TextureInfo info)
{
// While upscaling works for all targets defined by IsUpscaleCompatible, we additionally blacklist targets here that
// may have undesirable results (upscaling blur textures) or simply waste GPU resources (upscaling texture atlas).
if (!(info.FormatInfo.Format.IsDepthOrStencil() || info.FormatInfo.Components == 1))
{
// Discount square textures that aren't depth-stencil like. (excludes game textures, cubemap faces, most 3D texture LUT, texture atlas)
// Detect if the texture is possibly square. Widths may be aligned, so to remove the uncertainty we align both the width and height.
int widthAlignment = (info.IsLinear ? 32 : 64) / info.FormatInfo.BytesPerPixel;
bool possiblySquare = BitUtils.AlignUp(info.Width, widthAlignment) == BitUtils.AlignUp(info.Height, widthAlignment);
if (possiblySquare)
{
return(false);
}
}
int aspect = (int)Math.Round((info.Width / (float)info.Height) * 9);
if (aspect == 16 && info.Height < 360)
{
// Targets that are roughly 16:9 can only be rescaled if they're equal to or above 360p. (excludes blur and bloom textures)
return(false);
}
return(true);
}
public KMemoryRegionManager(ulong Address, ulong Size, ulong EndAddr)
{
Blocks = new KMemoryRegionBlock[BlockOrders.Length];
this.Address = Address;
this.Size = Size;
this.EndAddr = EndAddr;
BlockOrdersCount = BlockOrders.Length;
for (int BlockIndex = 0; BlockIndex < BlockOrdersCount; BlockIndex++)
{
Blocks[BlockIndex] = new KMemoryRegionBlock();
Blocks[BlockIndex].Order = BlockOrders[BlockIndex];
int NextOrder = BlockIndex == BlockOrdersCount - 1 ? 0 : BlockOrders[BlockIndex + 1];
Blocks[BlockIndex].NextOrder = NextOrder;
int CurrBlockSize = 1 << BlockOrders[BlockIndex];
int NextBlockSize = CurrBlockSize;
if (NextOrder != 0)
{
NextBlockSize = 1 << NextOrder;
}
ulong StartAligned = BitUtils.AlignDown(Address, NextBlockSize);
ulong EndAddrAligned = BitUtils.AlignDown(EndAddr, CurrBlockSize);
ulong SizeInBlocksTruncated = (EndAddrAligned - StartAligned) >> BlockOrders[BlockIndex];
ulong EndAddrRounded = BitUtils.AlignUp(Address + Size, NextBlockSize);
ulong SizeInBlocksRounded = (EndAddrRounded - StartAligned) >> BlockOrders[BlockIndex];
Blocks[BlockIndex].StartAligned = StartAligned;
Blocks[BlockIndex].SizeInBlocksTruncated = SizeInBlocksTruncated;
Blocks[BlockIndex].SizeInBlocksRounded = SizeInBlocksRounded;
ulong CurrSizeInBlocks = SizeInBlocksRounded;
int MaxLevel = 0;
do
{
MaxLevel++;
}while ((CurrSizeInBlocks /= 64) != 0);
Blocks[BlockIndex].MaxLevel = MaxLevel;
Blocks[BlockIndex].Masks = new long[MaxLevel][];
CurrSizeInBlocks = SizeInBlocksRounded;
for (int Level = MaxLevel - 1; Level >= 0; Level--)
{
CurrSizeInBlocks = (CurrSizeInBlocks + 63) / 64;
Blocks[BlockIndex].Masks[Level] = new long[CurrSizeInBlocks];
}
}
if (Size != 0)
{
FreePages(Address, Size / KMemoryManager.PageSize);
}
}