OSDN Git Service

[X86] Don't use constant condition for select instruction when testing masking ops.
authorCraig Topper <craig.topper@intel.com>
Sun, 15 Oct 2017 06:05:50 +0000 (06:05 +0000)
committerCraig Topper <craig.topper@intel.com>
Sun, 15 Oct 2017 06:05:50 +0000 (06:05 +0000)
commit37bf8530346faf7459ce5244ab8ba968cb1030ef
treeeef4eab7af593897db95915334b2137569de3830
parenteec5b16c8825de1564a555dec4618ea48a8de7a6
[X86] Don't use constant condition for select instruction when testing masking ops.

We should be able to fold constant conditions by converting to shuffles, but fixing that would break these tests in their current form. Since they are really trying to test masking ops, add a non-constant mask to the selects.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@315848 91177308-0d34-0410-b5e6-96231b3b80d8
15 files changed:
test/CodeGen/X86/avx512-shuffle-schedule.ll
test/CodeGen/X86/avx512-shuffles/broadcast-scalar-fp.ll
test/CodeGen/X86/avx512-shuffles/broadcast-scalar-int.ll
test/CodeGen/X86/avx512-shuffles/broadcast-vector-fp.ll
test/CodeGen/X86/avx512-shuffles/broadcast-vector-int.ll
test/CodeGen/X86/avx512-shuffles/duplicate-high.ll
test/CodeGen/X86/avx512-shuffles/duplicate-low.ll
test/CodeGen/X86/avx512-shuffles/in_lane_permute.ll
test/CodeGen/X86/avx512-shuffles/partial_permute.ll
test/CodeGen/X86/avx512-shuffles/permute.ll
test/CodeGen/X86/avx512-shuffles/shuffle-interleave.ll
test/CodeGen/X86/avx512-shuffles/shuffle-vec.ll
test/CodeGen/X86/avx512-shuffles/shuffle.ll
test/CodeGen/X86/avx512-shuffles/unpack.ll
test/CodeGen/X86/vector-shuffle-256-v16.ll