================
@@ -35,11 +35,11 @@ define void @loop_id(ptr %A) {
 ; CHECK:       [[INNER_HEADER_SPLIT]]:
 ; CHECK-NEXT:    [[TMP0]] = add nuw nsw i64 [[J]], 1
 ; CHECK-NEXT:    [[TMP1:%.*]] = icmp eq i64 [[TMP0]], 256
-; CHECK-NEXT:    br i1 [[TMP1]], label %[[EXIT:.*]], label %[[INNER_HEADER]], 
!llvm.loop [[LOOP1:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP1]], label %[[EXIT:.*]], label %[[INNER_HEADER]]
 ; CHECK:       [[OUTER_LATCH]]:
 ; CHECK-NEXT:    [[I_NEXT]] = add nuw nsw i64 [[I]], 1
 ; CHECK-NEXT:    [[EC_I:%.*]] = icmp eq i64 [[I_NEXT]], 256
-; CHECK-NEXT:    br i1 [[EC_I]], label %[[INNER_HEADER_SPLIT]], label 
%[[OUTER_HEADER]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK-NEXT:    br i1 [[EC_I]], label %[[INNER_HEADER_SPLIT]], label 
%[[OUTER_HEADER]]
----------------
madhur13490 wrote:

The original i loop had llvm.loop.unroll.disable (L10, L319). After this patch, 
the i loop, which is now innermost, can be unrolled even though the user 
disabled unrolling. If we keep attributes that don't depend on nesting 
position, this CHECK should keep the !llvm.loop with unroll.disable.

https://github.com/llvm/llvm-project/pull/235892
_______________________________________________
llvm-branch-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits

Reply via email to