mirror of
https://github.com/intel/compute-runtime.git
synced 2025-12-31 12:11:31 +08:00
Fix truncation issues from size_t to uint32_t
Related-To: LOCI-2558 Signed-off-by: Young Jin Yoon <young.jin.yoon@intel.com>
This commit is contained in:
committed by
Compute-Runtime-Automation
parent
72d16c7cf2
commit
702faca7a9
@@ -1449,21 +1449,27 @@ ze_result_t CommandListCoreFamily<gfxCoreFamily>::appendMemoryFill(void *ptr,
|
||||
}
|
||||
|
||||
auto dstAllocation = this->getAlignedAllocation(this->device, ptr, size, false);
|
||||
//get_global_id(0) in FillBufferMiddle only goes upto UINT32_MAX,
|
||||
constexpr size_t getGlobalIdMaxLimit = 4ull * MemoryConstants::gigaByte;
|
||||
//FillBufferMiddle copies data based on uint, so the minimum size is sizeof(uint32_t),
|
||||
//and Max limited to sizeof(uint32_t) * getGlobalIdMaxLimit
|
||||
constexpr size_t fillBufferMiddleMinLimit = sizeof(uint32_t);
|
||||
constexpr size_t fillBufferMiddleMaxLimit = sizeof(uint32_t) * getGlobalIdMaxLimit;
|
||||
if (size >= 4ull * MemoryConstants::gigaByte) {
|
||||
isStateless = true;
|
||||
if (size > fillBufferMiddleMaxLimit) {
|
||||
return ZE_RESULT_ERROR_UNSUPPORTED_SIZE;
|
||||
}
|
||||
}
|
||||
auto lock = device->getBuiltinFunctionsLib()->obtainUniqueOwnership();
|
||||
|
||||
if (patternSize == 1) {
|
||||
if ((patternSize == 1) && (size < fillBufferMiddleMinLimit)) {
|
||||
Kernel *builtinFunction = nullptr;
|
||||
|
||||
if (isStateless) {
|
||||
builtinFunction = device->getBuiltinFunctionsLib()->getStatelessFunction(Builtin::FillBufferImmediate);
|
||||
} else {
|
||||
builtinFunction = device->getBuiltinFunctionsLib()->getFunction(Builtin::FillBufferImmediate);
|
||||
}
|
||||
builtinFunction = device->getBuiltinFunctionsLib()->getFunction(Builtin::FillBufferImmediate);
|
||||
|
||||
uint32_t groupSizeX = builtinFunction->getImmutableData()->getDescriptor().kernelAttributes.simdSize;
|
||||
if (groupSizeX > static_cast<uint32_t>(size)) {
|
||||
if (groupSizeX > size) {
|
||||
groupSizeX = static_cast<uint32_t>(size);
|
||||
}
|
||||
if (builtinFunction->setGroupSize(groupSizeX, 1u, 1u)) {
|
||||
@@ -1478,14 +1484,16 @@ ze_result_t CommandListCoreFamily<gfxCoreFamily>::appendMemoryFill(void *ptr,
|
||||
|
||||
appendEventForProfilingAllWalkers(hSignalEvent, true);
|
||||
|
||||
uint32_t groups = static_cast<uint32_t>(size) / groupSizeX;
|
||||
ze_group_count_t dispatchFuncArgs{groups, 1u, 1u};
|
||||
uint64_t groups = size / groupSizeX;
|
||||
DEBUG_BREAK_IF(groups >= UINT32_MAX);
|
||||
|
||||
ze_group_count_t dispatchFuncArgs{static_cast<uint32_t>(groups), 1u, 1u};
|
||||
res = appendLaunchKernelSplit(builtinFunction->toHandle(), &dispatchFuncArgs, hSignalEvent);
|
||||
if (res) {
|
||||
return res;
|
||||
}
|
||||
|
||||
uint32_t groupRemainderSizeX = static_cast<uint32_t>(size) % groupSizeX;
|
||||
uint32_t groupRemainderSizeX = static_cast<uint32_t>(size % groupSizeX);
|
||||
if (groupRemainderSizeX) {
|
||||
builtinFunction->setGroupSize(groupRemainderSizeX, 1u, 1u);
|
||||
ze_group_count_t dispatchFuncRemainderArgs{1u, 1u, 1u};
|
||||
@@ -1510,13 +1518,14 @@ ze_result_t CommandListCoreFamily<gfxCoreFamily>::appendMemoryFill(void *ptr,
|
||||
|
||||
size_t middleElSize = sizeof(uint32_t);
|
||||
size_t adjustedSize = size / middleElSize;
|
||||
uint32_t groupSizeX = static_cast<uint32_t>(adjustedSize);
|
||||
uint32_t groupSizeX = (adjustedSize >= UINT32_MAX) ? UINT32_MAX - 1 : static_cast<uint32_t>(adjustedSize);
|
||||
uint32_t groupSizeY = 1, groupSizeZ = 1;
|
||||
builtinFunction->suggestGroupSize(groupSizeX, groupSizeY, groupSizeZ, &groupSizeX, &groupSizeY, &groupSizeZ);
|
||||
builtinFunction->setGroupSize(groupSizeX, groupSizeY, groupSizeZ);
|
||||
|
||||
uint32_t groups = static_cast<uint32_t>(adjustedSize) / groupSizeX;
|
||||
uint32_t groupRemainderSizeX = static_cast<uint32_t>(size) % groupSizeX;
|
||||
uint64_t groups = adjustedSize / groupSizeX;
|
||||
DEBUG_BREAK_IF(groups >= UINT32_MAX);
|
||||
uint32_t groupRemainderSizeX = static_cast<uint32_t>(size % groupSizeX);
|
||||
|
||||
size_t patternAllocationSize = alignUp(patternSize, MemoryConstants::cacheLineSize);
|
||||
uint32_t patternSizeInEls = static_cast<uint32_t>(patternAllocationSize / middleElSize);
|
||||
@@ -1527,6 +1536,10 @@ ze_result_t CommandListCoreFamily<gfxCoreFamily>::appendMemoryFill(void *ptr,
|
||||
patternAllocationSize,
|
||||
NEO::GraphicsAllocation::AllocationType::FILL_PATTERN,
|
||||
device->getNEODevice()->getDeviceBitfield()});
|
||||
if (patternGfxAlloc == nullptr) {
|
||||
PRINT_DEBUG_STRING(NEO::DebugManager.flags.PrintDebugMessages.get(), stderr, "out of host memory error, check vm.max_map_count in sysctl for mmap limits, and total memory allocated for the application\n", "");
|
||||
return ZE_RESULT_ERROR_OUT_OF_HOST_MEMORY;
|
||||
}
|
||||
}
|
||||
void *patternGfxAllocPtr = patternGfxAlloc->getUnderlyingBuffer();
|
||||
patternAllocations.push_back(patternGfxAlloc);
|
||||
@@ -1551,14 +1564,14 @@ ze_result_t CommandListCoreFamily<gfxCoreFamily>::appendMemoryFill(void *ptr,
|
||||
|
||||
appendEventForProfilingAllWalkers(hSignalEvent, true);
|
||||
|
||||
ze_group_count_t dispatchFuncArgs{groups, 1u, 1u};
|
||||
ze_group_count_t dispatchFuncArgs{static_cast<uint32_t>(groups), 1u, 1u};
|
||||
res = appendLaunchKernelSplit(builtinFunction->toHandle(), &dispatchFuncArgs, hSignalEvent);
|
||||
if (res) {
|
||||
return res;
|
||||
}
|
||||
|
||||
if (groupRemainderSizeX) {
|
||||
uint32_t dstOffsetRemainder = groups * groupSizeX * static_cast<uint32_t>(middleElSize);
|
||||
uint32_t dstOffsetRemainder = static_cast<uint32_t>(groups) * groupSizeX * static_cast<uint32_t>(middleElSize);
|
||||
uint64_t patternOffsetRemainder = (groupSizeX * groups & (patternSizeInEls - 1)) * middleElSize;
|
||||
|
||||
Kernel *builtinFunctionRemainder;
|
||||
|
||||
Reference in New Issue
Block a user