[llvm] [SPIRV] Add handling for `uinc_wrap` and `udec_wrap` atomics (PR #179114)

[Manuel Carrasco via llvm-commits]

================
@@ -627,3 +634,26 @@ bool SPIRVTargetLowering::insertLogicalCopyOnResult(
       .constrainAllUses(*STI.getInstrInfo(), *STI.getRegisterInfo(),
                         *STI.getRegBankInfo());
 }
+
+TargetLowering::AtomicExpansionKind
+SPIRVTargetLowering::shouldExpandAtomicRMWInIR(const AtomicRMWInst *RMW) const {
+  switch (RMW->getOperation()) {
+  case AtomicRMWInst::FAdd:
+  case AtomicRMWInst::FSub:
+  case AtomicRMWInst::FMin:
+  case AtomicRMWInst::FMax:
+    return AtomicExpansionKind::None;
+  case AtomicRMWInst::UIncWrap:
+  case AtomicRMWInst::UDecWrap:
+    return AtomicExpansionKind::CmpXChg;
+  default:
+    return TargetLowering::shouldExpandAtomicRMWInIR(RMW);
+  }
+}
+
+TargetLowering::AtomicExpansionKind
+SPIRVTargetLowering::shouldCastAtomicRMWIInIR(AtomicRMWInst *RMWI) const {
+  // Do not cast atomic exchange at all since SPIR-V natively supports
----------------
mgcarrasco wrote:

If I got it right, the PR is not changing how pointers are handled because that is not required for uinc_wrap or udec_wrap support. So there are no regressions. Thus, I'd not address that particular problem in this PR. A todo looks fine to me explaining that it may be a good idea to cast pointers because SPIRV doesn't support that (if it makes sense). 

https://github.com/llvm/llvm-project/pull/179114