projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: a1978aa) | patch
AMDGPU/SI: Don't promote to vector if the load/store is volatile.
authorChangpeng Fang <changpeng.fang@gmail.com>
Fri, 12 May 2017 20:31:12 +0000 (20:31 +0000)
committerChangpeng Fang <changpeng.fang@gmail.com>
Fri, 12 May 2017 20:31:12 +0000 (20:31 +0000)
commit161e8c39af4b7ff6ddce76ffe7aa78fc1f35583f
tree83d18d2e3a67ef79e59f203b027b4be2211e8030tree | snapshot
parenta1978aaefd3adc7675e1efbff9f907cd4318e624commit | diff
AMDGPU/SI: Don't promote to vector if the load/store is volatile.

Summary:
  We should not change volatile loads/stores in promoting alloca to vector.

Reviewers:
  arsenm

Differential Revision:
  http://reviews.llvm.org/D33107

llvm-svn: 302943
llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/promote-alloca-volatile.ll diff | blob | history
Domain: System / Toolchain;