[llvm] bd82cec - [X86] Update PR20841 test description to make it clear we SHOULDN'T be folding EFLAGS with XADD

[Simon Pilgrim via llvm-commits]

Author: Simon Pilgrim
Date: 2021-05-04T13:29:19+01:00
New Revision: bd82cec016ad1cb4ef75d710ff6628837c4d5151

URL: https://github.com/llvm/llvm-project/commit/bd82cec016ad1cb4ef75d710ff6628837c4d5151
DIFF: https://github.com/llvm/llvm-project/commit/bd82cec016ad1cb4ef75d710ff6628837c4d5151.diff

LOG: [X86] Update PR20841 test description to make it clear we SHOULDN'T be folding EFLAGS with XADD

Added: 
    

Modified: 
    llvm/test/CodeGen/X86/atomic-flags.ll

Removed: 
    


################################################################################
diff  --git a/llvm/test/CodeGen/X86/atomic-flags.ll b/llvm/test/CodeGen/X86/atomic-flags.ll
index dfd916f81216..e804340148af 100644
--- a/llvm/test/CodeGen/X86/atomic-flags.ll
+++ b/llvm/test/CodeGen/X86/atomic-flags.ll
@@ -124,7 +124,9 @@ L4:
   ret i32 4
 }
 
-; FIXME: PR20841 - ensure we reuse the ZF flag from XADD for compares with zero.
+; PR20841 - ensure we don't reuse the ZF flag from XADD for compares with zero,
+; the flags are set for the result of the add result (the value stored to memory),
+; not the value returned by the atomicrmw add.
 
 define zeroext i1 @xadd_cmp0_i64(i64* %x) nounwind {
 ; X64-LABEL: xadd_cmp0_i64: