intel/llvm
1
0
Fork 0
mirror of https://github.com/intel/llvm.git synced 2026-01-13 11:02:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Offload] Remove check on kernel argument sizes (#162121)

Browse Source 
Summary:
This check is unnecessarily restrictive and currently incorrectly fires
for any size less than eight bytes. Just remove it, we do sanity checks
elsewhere and at some point need to trust the ABI.
This commit is contained in:
Joseph Huber
2025-10-06 12:49:44 -05:00
committed by GitHub
parent 839b91c229
commit 8763812b4c
4 changed files with 6 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
offload/plugins-nextgen/amdgpu/src/rtl.cpp
View File

@@ -3687,11 +3687,6 @@ Error AMDGPUKernelTy::launchImpl(GenericDeviceTy &GenericDevice,
KernelArgsTy &KernelArgs,
KernelLaunchParamsTy LaunchParams,
AsyncInfoWrapperTy &AsyncInfoWrapper) const {
if (ArgsSize != LaunchParams.Size &&
ArgsSize > LaunchParams.Size + getImplicitArgsSize())
return Plugin::error(ErrorCode::INVALID_ARGUMENT,
"invalid kernel arguments size");
AMDGPUPluginTy &AMDGPUPlugin =
static_cast<AMDGPUPluginTy &>(GenericDevice.Plugin);
AMDHostDeviceTy &HostDevice = AMDGPUPlugin.getHostDevice();

2
offload/unittests/OffloadAPI/device_code/CMakeLists.txt
View File

@@ -2,6 +2,7 @@ add_offload_test_device_code(foo.cpp foo)
add_offload_test_device_code(bar.cpp bar)
# Compile with optimizations to eliminate AMDGPU implicit arguments.
add_offload_test_device_code(noargs.cpp noargs -O3)
add_offload_test_device_code(byte.cpp byte)
add_offload_test_device_code(localmem.cpp localmem)
add_offload_test_device_code(localmem_reduction.cpp localmem_reduction)
add_offload_test_device_code(localmem_static.cpp localmem_static)
@@ -14,6 +15,7 @@ add_custom_target(offload_device_binaries DEPENDS
foo.bin
bar.bin
noargs.bin
byte.bin
localmem.bin
localmem_reduction.bin
localmem_static.bin

3
offload/unittests/OffloadAPI/device_code/byte.cpp Normal file
View File

@@ -0,0 +1,3 @@
#include <gpuintrin.h>
extern "C" __gpu_kernel void byte(unsigned char c) { (void)c; }

1
offload/unittests/OffloadAPI/kernel/olLaunchKernel.cpp
View File

@@ -55,6 +55,7 @@ struct LaunchSingleKernelTestBase : LaunchKernelTestBase {
KERNEL_TEST(Foo, foo)
KERNEL_TEST(NoArgs, noargs)
KERNEL_TEST(Byte, byte)
KERNEL_TEST(LocalMem, localmem)
KERNEL_TEST(LocalMemReduction, localmem_reduction)
KERNEL_TEST(LocalMemStatic, localmem_static)