[x86] Switch to a formulation of a for loop that is much more obviously
authorChandler Carruth <chandlerc@gmail.com>
Wed, 6 Aug 2014 10:16:33 +0000 (10:16 +0000)
committerChandler Carruth <chandlerc@gmail.com>
Wed, 6 Aug 2014 10:16:33 +0000 (10:16 +0000)
commit8f23ba26d25d03d7deb7c7ccd46824ed713bc04b
tree6bacc3356b7dca57c8d7cfce2b6bac6c82080f42
parent6fd227c9ad44582dee3c096ea6a1e1c429e8d253
[x86] Switch to a formulation of a for loop that is much more obviously
not corrupting the mask by mutating it more times than intended. No
functionality changed (the results were non-overlapping so the old
version "worked" but was non-obvious).

llvm-svn: 214953
llvm/lib/Target/X86/X86ISelLowering.cpp