[llvm] [X86] Always use 128-bit V_SET0/AVX512_128_SET0 patterns, along with SUBREG_TO_REG for extension to 256/512-bit vectors (PR #212950)

[Craig Topper via llvm-commits]

================
@@ -8654,17 +8615,10 @@ X86InstrInfo::foldMemoryOperandImpl(MachineFunction &MF, MachineInstr &MI,
     case X86::AVX512_512_SETALLONES:
       IsAllOnes = true;
       [[fallthrough]];
-    case X86::AVX512_512_SET0:
-      Ty = FixedVectorType::get(Type::getInt32Ty(MF.getFunction().getContext()),
----------------
topperc wrote:

This is bug we can't delete this FixedVectorType::get call.

https://github.com/llvm/llvm-project/pull/212950