I am seeing a lot of failures on nvptx machine (sm_70, cuda11.4) with this patch,

libomptarget :: nvptx64-nvidia-cuda :: offloading/bug49021.cpp
libomptarget :: nvptx64-nvidia-cuda :: offloading/bug49334.cpp
libomptarget :: nvptx64-nvidia-cuda :: offloading/bug49779.cpp
libomptarget :: nvptx64-nvidia-cuda :: offloading/bug51781.c
libomptarget :: nvptx64-nvidia-cuda :: offloading/bug51982.c
libomptarget :: nvptx64-nvidia-cuda :: unified_shared_memory/close_enter_exit.c
libomptarget :: nvptx64-nvidia-cuda :: unified_shared_memory/close_modifier.c
libomptarget :: nvptx64-nvidia-cuda :: unified_shared_memory/shared_update.c
libomptarget :: nvptx64-nvidia-cuda-newRTL :: offloading/bug49021.cpp
libomptarget :: nvptx64-nvidia-cuda-newRTL :: offloading/bug49334.cpp
libomptarget :: nvptx64-nvidia-cuda-newRTL :: offloading/bug51781.c
libomptarget :: nvptx64-nvidia-cuda-newRTL :: unified_shared_memory/close_enter_exit.c
libomptarget :: nvptx64-nvidia-cuda-newRTL :: unified_shared_memory/close_modifier.c
libomptarget :: nvptx64-nvidia-cuda-newRTL :: unified_shared_memory/shared_update.c

On amdgcn, these are the tests failing,

libomptarget :: amdgcn-amd-amdhsa :: offloading/bug49021.cpp
libomptarget :: amdgcn-amd-amdhsa :: offloading/bug51781.c
libomptarget :: amdgcn-amd-amdhsa :: offloading/bug51982.c
libomptarget :: amdgcn-amd-amdhsa-newRTL :: offloading/bug49021.cpp
libomptarget :: amdgcn-amd-amdhsa-newRTL :: offloading/bug51781.c

Hi, this is again causing issues on amdgpu buildbot (https://lab.llvm.org/buildbot/#/builders/193). In the logs, I see that cmake was unable to find libelf dependency which is necessary for openmp plugins to be built and tested. In absence of libelf, the bot, currently, silently ignores and run non-amdgpu tests. I locally reverted this patch and found that tests are again working fine.

LGTM

If its helpful, following is the cmake command I used (cmake version: 3.20)

cmake ../llvm -GNinja -DCMAKE_BUILD_TYPE=Release -DLLVM_TARGETS_TO_BUILD="X86;AMDGPU" -DLLVM_ENABLE_ASSERTIONS=On -DLLVM_ENABLE_RUNTIMES=openmp -DCLANG_DEFAULT_LINKER=lld -DLLVM_ENABLE_PROJECTS="lld;clang"

Hi, this patch is producing cmake errors in our amdgpu buildbot: https://lab.llvm.org/buildbot/#/builders/193. Initially I thought there was something wrong with our configuration, but git bisecting revealed that this commit is causing the issue. If I remove the following line, the build is fine.

project(Runtimes LANGUAGES NONE)

Only removing optnone.

I don't have any concrete evidence but I have some doubt on presence of function pointers causing backend to behave improperly. Also, here removing optnone alone suffices to fix the issue.

I have created a patch (D111218) with fix for amdgcn. This is a temporary fix. I will still keep on looking into it until I find a real root cause.

Abandoning this in favor of D110902 due to dependency issues.

I modified the declare_mapper_target to print the contents of array after target region and found the following output:

2 3 4 5 6 7 8 9 10 11 Sum = 65

Apologies for late reply. Most of the tests now do not try to call malloc, so no page fault errors. But all of them are producing wrong results. For e.g. declare_mapper_target.cpp produces Sum = 132608 with the patch applied. Similarly for other tests as well. So don't know what's happening yet.

• review comments

• rebase after rG9d0eb440ff40
• change return type of hsa_status_t

Rebase

I got this after changing __kmpc_impl_malloc to return 0xdeadbeef. So, this confirms that missing malloc implementation is the root cause.

Memory access fault by GPU node-4 (Agent handle: 0x1bc5000) on address 0xdeadb000. Reason: Page not present or supervisor privilege.

It looks like from IR diff that this patch is adding use of kmpc_alloc_shared method. These methods likely won't work on AMDGPU as device malloc is not available. Not sure what could be done apart from marking those tests as XFAIL on amdgcn. :(

Looks good.

update logic for fine-grained/kernarg pools

This is not dependent on D109875, instead it is the other case. Apologies for not mentioning here. Diff shown here is messed up.

Check if both values are set properly.

run clang-format

Split this patch into two,

1. Current patch now only includes changes for simplifying the host pool lookup
2. D109875 includes plain refactoring.

Even with declare variant separated using ifdef's, the error is still there. So I don't think we have workaround for this.

Hey, Jon, sorry for late reply. I cannot reproduce this issue on nvptx so it seems to occur only on amdgcn. Will it be better if instead the name mangling issue is fixed? Or for the meantime, I could add #ifdef around as a temporary fix. Suggestions?

Looks good! Thanks.

Add amdgcn to the same arch list as nvptx

LGTM

Confirmed locally that this fixes the linking issue. It has fixed the linking issue on amdgcn as well. Thanks for working on this.

Oh sorry, forgot about this. Landing it now.

Remove redundant test

Rebase

@ye-luo and @JonChesterfield can you please test the latest version of this patch? It should work now.

Fixed compilation error for nvptx headers. Tested on both cuda and non-cuda systems.

Addressed review comments.

It required some work to fix the failing lit test case. And many thanks to
@estewart for helping in that.

Missed comment.

Rename method to getCommonDeviceLibNames

Due to the current state of math headers, I was unable to test this patch without ockl. But last time when headers were working, I was actually required to link ockl for a symbol (I forgot the name). I will update once I am able to get the math headers work again.

Extract the options from HIP/OpenMP to a common method in base class.

I need to rebase this.

Move linking logic to a common method.

Looks ok to me. Regression tests and runtime tests went fine. Tested a simple cuda and openmp kernel with sin function on sm_61, didn't see any issue.

Should the name of file be changed as well?

Looks good! Thanks.

• Move constant to openmp_wrappers/cmath
• Using push/pop_macro to avoid redefinition

Looks good! Thanks

Typo

Addressed review comments.

Fix format errors

Addressed review comments.

Fix clang-tidy errors and use scalar for host memory pool.