AVX-512: Fixed COPY_TO_REGCLASS for mask registers
authorElena Demikhovsky <elena.demikhovsky@intel.com>
Thu, 19 Nov 2015 13:13:00 +0000 (13:13 +0000)
committerElena Demikhovsky <elena.demikhovsky@intel.com>
Thu, 19 Nov 2015 13:13:00 +0000 (13:13 +0000)
commit1c8b4f079651c15f5a54da9f545d87a17026f1fb
tree8fff560dec394cdacab90f5714eecf950d74ef45
parentcc1b3e47c025334300f022921030de782595f6dc
AVX-512: Fixed COPY_TO_REGCLASS for mask registers

Copying one mask register to another under BW should be done with kmovq instruction, otherwise we can loose some bits.
Copying 8 bits under DQ may be done with kmovb.

Differential Revision: http://reviews.llvm.org/D14812

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@253563 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/X86InstrInfo.cpp
test/CodeGen/X86/avx512-gather-scatter-intrin.ll
test/CodeGen/X86/avx512dq-intrinsics.ll
test/CodeGen/X86/avx512dqvl-intrinsics.ll