Improve test to actually check for a folded load.
authorSanjay Patel <spatel@rotateright.com>
Tue, 3 Feb 2015 15:37:18 +0000 (15:37 +0000)
committerSanjay Patel <spatel@rotateright.com>
Tue, 3 Feb 2015 15:37:18 +0000 (15:37 +0000)
commitec60318bf55ecd9beefe13e168b0e3281882b1db
treecd7bb7eeab99e6f59b303ba70ebdea584c9d67d4
parent7df357f5520866a7241828041b32dd20b1f6da60
Improve test to actually check for a folded load.

This test was checking for lack of a "movaps" (an aligned load)
rather than a "movups" (an unaligned load). It also included
a store which complicated the checking.

Add specific CPU runs to prevent subtarget feature flag overrides
from inhibiting this optimization.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@227972 91177308-0d34-0410-b5e6-96231b3b80d8
test/CodeGen/X86/fold-vex.ll