[llvm-branch-commits] [llvm] [X86] Remove extra MOV after widening atomic load (PR #138635)

[via llvm-branch-commits]

================
@@ -1200,6 +1200,13 @@ def : Pat<(i16 (atomic_load_nonext_16 addr:$src)), (MOV16rm addr:$src)>;
 def : Pat<(i32 (atomic_load_nonext_32 addr:$src)), (MOV32rm addr:$src)>;
 def : Pat<(i64 (atomic_load_nonext_64 addr:$src)), (MOV64rm addr:$src)>;
 
+def : Pat<(v4i32 (scalar_to_vector (i32 (anyext (i16 (atomic_load_16 addr:$src)))))),
+           (MOVDI2PDIrm addr:$src)>;   // load atomic <2 x i8>
----------------
jofrn wrote:

Without loss of generality, do we not need the `v2` in `--check-prefixes=CHECK,CHECKv2-O0` due to divergence of asm? https://github.com/llvm/llvm-project/blob/d27d0c7a5266f89f9d62464e71be98421aae598d/llvm/test/CodeGen/X86/atomic-load-store.ll#L7

https://github.com/llvm/llvm-project/pull/138635