projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b53feca) | patch
AMDGPU: Fix private handling for allowsMisalignedMemoryAccesses
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Mon, 24 Sep 2018 13:18:15 +0000 (13:18 +0000)
committerMatt Arsenault <Matthew.Arsenault@amd.com>
Mon, 24 Sep 2018 13:18:15 +0000 (13:18 +0000)
commitf432011d33e394788d717809ad8955763efe4d9d
tree3ad9a68b7121eb430e0af1c652c178f63bfcbbcetree | snapshot
parentb53feca372725647f7f3eaddee238629616eac4fcommit | diff
AMDGPU: Fix private handling for allowsMisalignedMemoryAccesses

If the alignment is at least 4, this should report true.

Something still seems off with how < 4-byte types are
handled here though.

Fixing this seems to change how some combines get
to where they get, but somehow isn't changing the net
result.

llvm-svn: 342879
llvm/lib/Target/AMDGPU/SIISelLowering.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/byval-frame-setup.ll diff | blob | history
llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores-private.ll diff | blob | history
Domain: System / Toolchain;