[llvm] [LoongArch] Introduce `32s` target feature for LA32S ISA extensions (PR #139695)

[via llvm-commits]

================
@@ -214,8 +214,9 @@ static void doAtomicBinOpExpansion(const LoongArchInstrInfo *TII,
       .addReg(ScratchReg)
       .addReg(AddrReg)
       .addImm(0);
-  BuildMI(LoopMBB, DL, TII->get(LoongArch::BEQZ))
+  BuildMI(LoopMBB, DL, TII->get(LoongArch::BEQ))
----------------
heiher wrote:

I agree. While `BEQZ` does offer a wider branch range than `BEQ`, their latency and throughput are the same. For expanding the pseudo-atomic seqences, I believe BEQ's range is sufficient. I chose not to split it further to avoid unnecessary divergence between the 32-bit and 64-bit code paths.

https://github.com/llvm/llvm-project/pull/139695