projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: d150152) | patch
[CUDA][FIX] Make shfl[_sync] for unsigned long long non-recursive
authorJohannes Doerfert <johannes@jdoerfert.de>
Tue, 12 Jul 2022 02:42:16 +0000 (21:42 -0500)
committerJohannes Doerfert <johannes@jdoerfert.de>
Thu, 21 Jul 2022 17:36:54 +0000 (12:36 -0500)
commit48d6f5240187573881f96cc9574ea09592f50723
treef3a782e9031cf73f6cc625c440ff75a1be682091tree | snapshot
parentd150152615074190d20492512da439cd5820b04acommit | diff
[CUDA][FIX] Make shfl[_sync] for unsigned long long non-recursive

A copy-paste error caused UB in the definition of the unsigned long long
versions of the shfl intrinsics. Reported and diagnosed by @trws.

Differential Revision: https://reviews.llvm.org/D129536
clang/lib/Headers/__clang_cuda_intrinsics.h diff | blob | history
clang/test/CodeGenCUDA/shuffle_long_long.cu [new file with mode: 0644] blob
Domain: System / Toolchain;