intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-27 15:53:13 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: properly retrieve sub-device count in allocateRTDispatchGlobals

Browse Source 
Previous method used for retrieiving sub-device count in this path
did not take into account device affinity mask nor device hiearchy
mode, resulting in segmentation faults when attempting to allocate
rtDispatchGlobals structure using improper deviceBitFields.

Related-To: NEO-8422

Signed-off-by: Raiyan Latif <raiyan.latif@intel.com>
This commit is contained in:
Raiyan Latif
2024-03-04 11:35:51 +00:00
committed by Compute-Runtime-Automation
parent 931eaafa11
commit 40471da591
2 changed files with 48 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
shared/source/device/device.cpp
View File

@@ -1019,9 +1019,14 @@ void Device::allocateRTDispatchGlobals(uint32_t maxBvhLevels) {
bool allocFailed = false;
const auto deviceCount = GfxCoreHelper::getSubDevicesCount(executionEnvironment->rootDeviceEnvironments[getRootDeviceIndex()]->getHardwareInfo());
auto dispatchGlobalsSize = deviceCount * dispatchGlobalsStride;
auto rtStackSize = RayTracingHelper::getRTStackSizePerTile(*this, deviceCount, maxBvhLevels, extraBytesLocal, extraBytesGlobal);
uint32_t tileCount = 1;
if (this->getNumSubDevices() > 1) {
// If device encompasses multiple tiles, allocate RTDispatchGlobals for each tile
tileCount = this->getNumSubDevices();
}
auto dispatchGlobalsSize = tileCount * dispatchGlobalsStride;
auto rtStackSize = RayTracingHelper::getRTStackSizePerTile(*this, tileCount, maxBvhLevels, extraBytesLocal, extraBytesGlobal);
std::unique_ptr<RTDispatchGlobalsInfo> dispatchGlobalsInfo = std::make_unique<RTDispatchGlobalsInfo>();
@@ -1039,9 +1044,9 @@ void Device::allocateRTDispatchGlobals(uint32_t maxBvhLevels) {
return;
}
for (unsigned int tile = 0; tile < deviceCount; tile++) {
for (unsigned int tile = 0; tile < tileCount; tile++) {
DeviceBitfield deviceBitfield =
(deviceCount == 1)
(tileCount == 1)
? this->getDeviceBitfield()
: subdevices[tile]->getDeviceBitfield();