projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 11f7402) | patch
AMDGPU: Remove leftover implicit operands when folding immediates
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Thu, 6 Oct 2016 17:54:30 +0000 (17:54 +0000)
committerMatt Arsenault <Matthew.Arsenault@amd.com>
Thu, 6 Oct 2016 17:54:30 +0000 (17:54 +0000)
commitc2ee42cd16213e896947b9e1ee1c30defa006f99
tree2aff178ca11ab25f3514ccb5d081ec0d7b0ed16ctree | snapshot
parent11f740207561c27fb1301e5eba00ba869696bbd0commit | diff
AMDGPU: Remove leftover implicit operands when folding immediates

When constant folding an operation to a copy or an immediate
mov, the implicit uses/defs of the old instruction were left behind,
e.g. replacing v_or_b32 left the implicit exec use on the new copy.

llvm-svn: 283471
llvm/lib/Target/AMDGPU/SIFoldOperands.cpp diff | blob | history
Domain: System / Toolchain;