From eea69691908d2ff59ef2614fc6f948d021d0ef07 Mon Sep 17 00:00:00 2001
From: Nikita Popov <npopov@redhat.com>
Date: Wed, 28 May 2025 11:57:30 +0200
Subject: [PATCH] [LICM] Only check for provenance captures (#141731)

When performing scalar promotions, only consider provenance captures,
which may lead to non-thread-safe accesses. Address captures can be
ignored.
---
 llvm/lib/Transforms/Scalar/LICM.cpp          |  5 +-
 llvm/test/Transforms/LICM/promote-capture.ll | 52 ++++++++++++++++++++
 2 files changed, 55 insertions(+), 2 deletions(-)

diff --git a/llvm/lib/Transforms/Scalar/LICM.cpp b/llvm/lib/Transforms/Scalar/LICM.cpp
index 006a09b38bc7..7965ed76a81b 100644
--- a/llvm/lib/Transforms/Scalar/LICM.cpp
+++ b/llvm/lib/Transforms/Scalar/LICM.cpp
@@ -1928,8 +1928,9 @@ bool isNotCapturedBeforeOrInLoop(const Value *V, const Loop *L,
   // loop header, as the loop header is reachable from any instruction inside
   // the loop.
   // TODO: ReturnCaptures=true shouldn't be necessary here.
-  return !PointerMayBeCapturedBefore(V, /* ReturnCaptures */ true,
-                                     L->getHeader()->getTerminator(), DT);
+  return capturesNothing(PointerMayBeCapturedBefore(
+      V, /*ReturnCaptures=*/true, L->getHeader()->getTerminator(), DT,
+      /*IncludeI=*/false, CaptureComponents::Provenance));
 }
 
 /// Return true if we can prove that a caller cannot inspect the object if an
diff --git a/llvm/test/Transforms/LICM/promote-capture.ll b/llvm/test/Transforms/LICM/promote-capture.ll
index 5cbe158e47a3..eac670e6f3ed 100644
--- a/llvm/test/Transforms/LICM/promote-capture.ll
+++ b/llvm/test/Transforms/LICM/promote-capture.ll
@@ -158,6 +158,58 @@ exit:
   ret void
 }
 
+define void @test_captured_before_loop_address_only(i32 %len) {
+; CHECK-LABEL: @test_captured_before_loop_address_only(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[COUNT:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 0, ptr [[COUNT]], align 4
+; CHECK-NEXT:    call void @capture(ptr captures(address) [[COUNT]])
+; CHECK-NEXT:    [[COUNT_PROMOTED:%.*]] = load i32, ptr [[COUNT]], align 4
+; CHECK-NEXT:    br label [[LOOP:%.*]]
+; CHECK:       loop:
+; CHECK-NEXT:    [[C_INC2:%.*]] = phi i32 [ [[COUNT_PROMOTED]], [[ENTRY:%.*]] ], [ [[C_INC1:%.*]], [[LATCH:%.*]] ]
+; CHECK-NEXT:    [[I:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[I_NEXT:%.*]], [[LATCH]] ]
+; CHECK-NEXT:    [[COND:%.*]] = call i1 @cond(i32 [[I]])
+; CHECK-NEXT:    br i1 [[COND]], label [[IF:%.*]], label [[LATCH]]
+; CHECK:       if:
+; CHECK-NEXT:    [[C_INC:%.*]] = add i32 [[C_INC2]], 1
+; CHECK-NEXT:    br label [[LATCH]]
+; CHECK:       latch:
+; CHECK-NEXT:    [[C_INC1]] = phi i32 [ [[C_INC]], [[IF]] ], [ [[C_INC2]], [[LOOP]] ]
+; CHECK-NEXT:    [[I_NEXT]] = add nuw i32 [[I]], 1
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[I_NEXT]], [[LEN:%.*]]
+; CHECK-NEXT:    br i1 [[CMP]], label [[EXIT:%.*]], label [[LOOP]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[C_INC1_LCSSA:%.*]] = phi i32 [ [[C_INC1]], [[LATCH]] ]
+; CHECK-NEXT:    store i32 [[C_INC1_LCSSA]], ptr [[COUNT]], align 4
+; CHECK-NEXT:    ret void
+;
+entry:
+  %count = alloca i32
+  store i32 0, ptr %count
+  call void @capture(ptr captures(address) %count)
+  br label %loop
+
+loop:
+  %i = phi i32 [ 0, %entry ], [ %i.next, %latch ]
+  %cond = call i1 @cond(i32 %i)
+  br i1 %cond, label %if, label %latch
+
+if:
+  %c = load i32, ptr %count
+  %c.inc = add i32 %c, 1
+  store i32 %c.inc, ptr %count
+  br label %latch
+
+latch:
+  %i.next = add nuw i32 %i, 1
+  %cmp = icmp eq i32 %i.next, %len
+  br i1 %cmp, label %exit, label %loop
+
+exit:
+  ret void
+}
+
 ; Should not get promoted, because the pointer is captured and may not
 ; be thread-local.
 define void @test_captured_before_loop_byval(ptr byval(i32) align 4 %count, i32 %len) {