R600/SI: Remove VReg_32 register class

Use VGPR_32 register class instead.  These two register classes were
identical and having separate classes was causing
SIInstrInfo::isLegalOperands() to be overly conservative in some cases.

This change is necessary to prevent future paches from missing a folding
opportunity in fneg-fabs.ll.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225382 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Tom Stellard
2015-01-07 20:59:25 +00:00
parent 033a537a84
commit e729f0f935
13 changed files with 151 additions and 153 deletions

View File

@@ -340,7 +340,7 @@ void SILowerControlFlowPass::LoadM0(MachineInstr &MI, MachineInstr *MovRel) {
} else {
assert(AMDGPU::SReg_64RegClass.contains(Save));
assert(AMDGPU::VReg_32RegClass.contains(Idx));
assert(AMDGPU::VGPR_32RegClass.contains(Idx));
// Save the EXEC mask
BuildMI(MBB, &MI, DL, TII->get(AMDGPU::S_MOV_B64), Save)