[clang] [CIR][CIRGen] Add support for `__sync_val/bool_compare_and_swap` builtins (PR #186529)

[Andy Kaylor via cfe-commits]

================
@@ -2951,3 +2953,94 @@ int atomic_load_and_store_dynamic_order(int *ptr, int order) {
   // OGCG:      [[CONTINUE_BLK]]:
   // OGCG-NEXT:   %{{.+}} = load i32, ptr %[[RES_SLOT]], align 4
 }
+
+void cmp_bool_int(int* p, int x, int u) {
+  // CIR-LABEL: @cmp_bool_int
+  // CIR: %[[PTR:.*]] = cir.load{{.*}} {{.*}} : !cir.ptr<!cir.ptr<!s32i>>, !cir.ptr<!s32i>
+  // CIR: %[[CMP:.*]] = cir.load{{.*}} {{.*}} : !cir.ptr<!s32i>, !s32i
+  // CIR: %[[UPD:.*]] = cir.load{{.*}} {{.*}} : !cir.ptr<!s32i>, !s32i
+  // CIR: %[[OLD:.*]], %[[RES:.*]] = cir.atomic.cmpxchg success(seq_cst) failure(seq_cst) syncscope(system) %[[PTR]], %[[CMP]], %[[UPD]] align(4) : (!cir.ptr<!s32i>, !s32i, !s32i) -> (!s32i, !cir.bool)
+  // CIR: cir.store{{.*}} %[[RES]], {{.*}} : !cir.bool, !cir.ptr<!cir.bool>
+
+  // LLVM-LABEL: @cmp_bool_int
----------------
andykaylor wrote:

Can you add OGCG checks?

https://github.com/llvm/llvm-project/pull/186529