llvm-project.git/clang/test/OpenMP/parallel_master_taskloop_codegen.cpp, branch main Unnamed repository; edit this file 'description' to name the repository. [IRBuilder] Generate nuw GEPs for struct member accesses (#99538) 2024-08-09T12:25:04+00:00 Hari Limaye hari.limaye@arm.com 2024-08-09T12:25:04+00:00 94473f4db6a6f5f12d7c4081455b5b596094eac5 Generate nuw GEPs for struct member accesses, as inbounds + non-negative implies nuw. Regression tests are updated using update scripts where possible, and by find + replace where not. 
Generate nuw GEPs for struct member accesses, as inbounds + non-negative
implies nuw.

Regression tests are updated using update scripts where possible, and by
find + replace where not.
 [OpenMP][NFC] Update clang OpenMP tests 2023-08-23T17:40:31+00:00 Johannes Doerfert johannes@jdoerfert.de 2023-08-20T23:35:42+00:00 5a64ae75b5a05bdfc7ccb12d0b0958478b5d865b Just re-running the script to make future updates easier 
Just re-running the script to make future updates easier
clang: Regenerate OpenMP tests 2023-07-11T22:28:10+00:00 Matt Arsenault Matthew.Arsenault@amd.com 2023-07-08T15:27:58+00:00 a709c49d75c62ecce9e5598994692546dfd1e02b Avoid diffs from no longer hardcoding metadata checks 
Avoid diffs from no longer hardcoding metadata checks
[OpenMP] Prefix outlined and reduction func names with original func's name 2023-04-19T20:00:26+00:00 Itay Bookstein itay.bookstein@nextsilicon.com 2022-12-23T09:35:47+00:00 782c59a4eef0bca8546b0cfbb1e48a9fcd044c93 This patch prefixes omp outlined helpers and reduction funcs with the original function's name. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D140722 
This patch prefixes omp outlined helpers and reduction funcs
with the original function's name.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D140722
Revert "[OpenMP] Prefix outlined and reduction func names with original func's name" 2023-04-19T16:08:49+00:00 Itay Bookstein itay.bookstein@nextsilicon.com 2023-04-19T16:08:11+00:00 6fdd13e0ec61dfa5bd631ec44d2752d965c26436 This reverts commit 029bfc311d4d7d3cd90be81bb08c046848796d02. 
This reverts commit 029bfc311d4d7d3cd90be81bb08c046848796d02.
[OpenMP] Prefix outlined and reduction func names with original func's name 2023-04-19T16:05:21+00:00 Itay Bookstein itay.bookstein@nextsilicon.com 2022-12-23T09:35:47+00:00 029bfc311d4d7d3cd90be81bb08c046848796d02 This patch attempts to prefix omp outlined helpers and reduction funcs with the original function's name. Reviewed By: jdoerfert Differential Revision: https://reviews.llvm.org/D140722 
This patch attempts to prefix omp outlined helpers and reduction funcs
with the original function's name.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D140722
[OpenMP] Convert more tests to opaque pointers (NFC) 2022-10-07T13:36:44+00:00 Nikita Popov npopov@redhat.com 2022-10-07T13:28:44+00:00 40e353d0f9e51938f73a88f48ab8ca7ff31ad918 These were converted using the script at https://gist.github.com/nikic/98357b71fd67756b0f064c9517b62a34 followed by a re-run of update_cc_test_checks.py. 
These were converted using the script at
https://gist.github.com/nikic/98357b71fd67756b0f064c9517b62a34
followed by a re-run of update_cc_test_checks.py.
Revert "[OpenMP] Codegen aggregate for outlined function captures" 2022-09-15T03:08:46+00:00 Dhruva Chakrabarti Dhruva.Chakrabarti@amd.com 2022-09-15T03:08:46+00:00 839ac62c5085d895d3165bc5024db623a7a78813 This reverts commit 7539e9cf811e590d9f12ae39673ca789e26386b4. 
This reverts commit 7539e9cf811e590d9f12ae39673ca789e26386b4.
[OpenMP] Codegen aggregate for outlined function captures 2022-09-15T00:54:05+00:00 Giorgis Georgakoudis georgakoudis1@llnl.gov 2022-09-15T00:09:54+00:00 7539e9cf811e590d9f12ae39673ca789e26386b4 Parallel regions are outlined as functions with capture variables explicitly generated as distinct parameters in the function's argument list. That complicates the fork_call interface in the OpenMP runtime: (1) the fork_call is variadic since there is a variable number of arguments to forward to the outlined function, (2) wrapping/unwrapping arguments happens in the OpenMP runtime, which is sub-optimal, has been a source of ABI bugs, and has a hardcoded limit (16) in the number of arguments, (3) forwarded arguments must cast to pointer types, which complicates debugging. This patch avoids those issues by aggregating captured arguments in a struct to pass to the fork_call. Reviewed By: jdoerfert, jhuber6, ABataev Differential Revision: https://reviews.llvm.org/D102107 
Parallel regions are outlined as functions with capture variables explicitly generated as distinct parameters in the function's argument list. That complicates the fork_call interface in the OpenMP runtime: (1) the fork_call is variadic since there is a variable number of arguments to forward to the outlined function, (2) wrapping/unwrapping arguments happens in the OpenMP runtime, which is sub-optimal, has been a source of ABI bugs, and has a hardcoded limit (16) in the number of arguments, (3)  forwarded arguments must cast to pointer types, which complicates debugging. This patch avoids those issues by aggregating captured arguments in a struct to pass to the fork_call.

Reviewed By: jdoerfert, jhuber6, ABataev

Differential Revision: https://reviews.llvm.org/D102107
Inliner: don't mark call sites as 'nounwind' if that would be redundant 2022-07-20T12:17:23+00:00 Nicolai Hähnle nicolai.haehnle@amd.com 2022-07-15T13:27:26+00:00 1ddc51d89d9d3a3f77d6ca47514a776641dc3e4e When F calls G calls H, G is nounwind, and G is inlined into F, then the inlined call-site to H should be effectively nounwind so as not to lose information during inlining. If H itself is nounwind (which often happens when H is an intrinsic), we no longer mark the callsite explicitly as nounwind. Previously, there were cases where the inlined call-site of H differs from a pre-existing call-site of H in F *only* in the explicitly added nounwind attribute, thus preventing common subexpression elimination. v2: - just check CI->doesNotThrow v3 (resubmit after revert at 344378808778c61d5599f4e0ac783ef7e6f8ed05): - update Clang tests Differential Revision: https://reviews.llvm.org/D129860 
When F calls G calls H, G is nounwind, and G is inlined into F, then the
inlined call-site to H should be effectively nounwind so as not to lose
information during inlining.

If H itself is nounwind (which often happens when H is an intrinsic), we
no longer mark the callsite explicitly as nounwind. Previously, there
were cases where the inlined call-site of H differs from a pre-existing
call-site of H in F *only* in the explicitly added nounwind attribute,
thus preventing common subexpression elimination.

v2:
- just check CI->doesNotThrow

v3 (resubmit after revert at 344378808778c61d5599f4e0ac783ef7e6f8ed05):
- update Clang tests

Differential Revision: https://reviews.llvm.org/D129860