[GlobalOpt] Precommit tests for PR84694

[gandhi56]

PR link: #84694

[gandhi56]

ping

[fhahn]

Thanks for splitting off the tests. Would it make sense to have them all in a single file?

Also would be good to add a prefix like [GlobalOpt] to the title.

[fhahn]

Better to avoid mentioning internal functions here and in the name of the file?

[gandhi56]

Sorry, where is this?

[llvmbot]

@llvm/pr-subscribers-llvm-transforms

Author: Anshil Gandhi (gandhi56)

Changes

PR link: #84694

---

Full diff: https://github.com/llvm/llvm-project/pull/87443.diff

2 Files Affected:

• (added) llvm/test/Transforms/GlobalOpt/cleanup-pointer-root-users-negative.ll (+26)
• (added) llvm/test/Transforms/GlobalOpt/cleanup-pointer-root-users-ops.ll (+71)

diff --git a/llvm/test/Transforms/GlobalOpt/cleanup-pointer-root-users-negative.ll b/llvm/test/Transforms/GlobalOpt/cleanup-pointer-root-users-negative.ll
new file mode 100644
index 00000000000000..97504ffb3fd59d
--- /dev/null
+++ b/llvm/test/Transforms/GlobalOpt/cleanup-pointer-root-users-negative.ll
@@ -0,0 +1,26 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
+; RUN: opt -passes=globalopt < %s -S | FileCheck %s
+
+@gv = internal unnamed_addr global [3 x ptr] zeroinitializer, align 16
+@gv2 = internal unnamed_addr global i32 0, align 4
+
+;; This test includes a load from @gv. No stores
+;; or memintrinsics with destination @gv should be removed.
+define i32 @main_with_load_from_b() local_unnamed_addr {
+; CHECK-LABEL: define i32 @main_with_load_from_b() local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[E:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store ptr [[E]], ptr getelementptr inbounds ([3 x ptr], ptr @gv, i64 0, i64 2), align 16
+; CHECK-NEXT:    [[LOAD_B:%.*]] = load ptr, ptr getelementptr inbounds ([3 x ptr], ptr @gv, i64 0, i64 2), align 16
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr @gv, ptr @gv2, i64 8, i1 false)
+; CHECK-NEXT:    ret i32 0
+;
+entry:
+  %e = alloca i32, align 4
+  store ptr %e, ptr getelementptr inbounds ([3 x ptr], ptr @gv, i64 0, i64 2), align 16
+  %load.b = load ptr, ptr getelementptr inbounds ([3 x ptr], ptr @gv, i64 0, i64 2), align 16
+  call void @llvm.memcpy.p0i8.p0i8.i64(ptr getelementptr inbounds ([3 x ptr], ptr @gv, i64 0, i64 0), ptr @gv2, i64 8, i1 false)
+  ret i32 0
+}
+
+declare void @llvm.memcpy.p0i8.p0i8.i64(ptr nocapture, ptr nocapture readonly, i64, i1) local_unnamed_addr
diff --git a/llvm/test/Transforms/GlobalOpt/cleanup-pointer-root-users-ops.ll b/llvm/test/Transforms/GlobalOpt/cleanup-pointer-root-users-ops.ll
new file mode 100644
index 00000000000000..6b3d0885809dfc
--- /dev/null
+++ b/llvm/test/Transforms/GlobalOpt/cleanup-pointer-root-users-ops.ll
@@ -0,0 +1,71 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
+; RUN: opt -passes=globalopt < %s -S | FileCheck %s
+
+@a = internal unnamed_addr global i32 0, align 4
+@b = internal unnamed_addr global [3 x ptr] zeroinitializer, align 16
+
+;; This test is extracted from the issue reported in #64680, with an
+;; additional memcpy and a memset. Ensure all stores and memintrinsics with
+;; destination @b are removed as @b is dead.
+define i32 @main() local_unnamed_addr {
+; CHECK-LABEL: define i32 @main() local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[E:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[DOTPR:%.*]] = load i32, ptr @a, align 4
+; CHECK-NEXT:    [[CMP1:%.*]] = icmp slt i32 [[DOTPR]], 3
+; CHECK-NEXT:    br i1 [[CMP1]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
+; CHECK:       for.body:
+; CHECK-NEXT:    store i32 8, ptr [[E]], align 4
+; CHECK-NEXT:    call void @bar()
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, ptr [[E]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    call void @foo()
+; CHECK-NEXT:    br label [[IF_END]]
+; CHECK:       if.end:
+; CHECK-NEXT:    store ptr [[E]], ptr getelementptr inbounds ([3 x ptr], ptr @b, i64 0, i64 2), align 16
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, ptr @a, align 4
+; CHECK-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
+; CHECK-NEXT:    store i32 [[INC]], ptr @a, align 4
+; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP1]], 2
+; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
+; CHECK:       for.end:
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr @b, ptr null, i64 8, i1 false)
+; CHECK-NEXT:    ret i32 0
+;
+entry:
+  %e = alloca i32, align 4
+  %.pr = load i32, ptr @a, align 4
+  %cmp1 = icmp slt i32 %.pr, 3
+  br i1 %cmp1, label %for.body, label %for.end
+
+for.body:                                         ; preds = %entry, %if.end
+  store i32 8, ptr %e, align 4
+  call void @bar()
+  %0 = load i32, ptr %e, align 4
+  %tobool.not = icmp eq i32 %0, 0
+  br i1 %tobool.not, label %if.then, label %if.end
+
+if.then:                                          ; preds = %for.body
+  call void @foo()
+  br label %if.end
+
+if.end:                                           ; preds = %if.then, %for.body
+  store ptr %e, ptr getelementptr inbounds ([3 x ptr], ptr @b, i64 0, i64 2), align 16
+  %1 = load i32, ptr @a, align 4
+  %inc = add nsw i32 %1, 1
+  store i32 %inc, ptr @a, align 4
+  %cmp = icmp slt i32 %1, 2
+  call void @llvm.memset.p0i8.i64(ptr getelementptr inbounds ([3 x ptr], ptr @b, i64 0, i64 0), i8 0, i64 8, i1 false)
+  br i1 %cmp, label %for.body, label %for.end
+
+for.end:                                          ; preds = %if.end, %entry
+  call void @llvm.memcpy.p0i8.p0i8.i64(ptr getelementptr inbounds ([3 x ptr], ptr @b, i64 0, i64 0), ptr null, i64 8, i1 false)
+  ret i32 0
+}
+
+declare void @bar() local_unnamed_addr
+declare void @foo() local_unnamed_addr
+declare void @llvm.memcpy.p0i8.p0i8.i64(ptr nocapture, ptr nocapture readonly, i64, i1) local_unnamed_addr
+declare void @llvm.memset.p0i8.i64(ptr nocapture, i8, i64, i1) local_unnamed_addr

[gandhi56]

Thanks for splitting off the tests. Would it make sense to have them all in a single file?

Also would be good to add a prefix like [GlobalOpt] to the title.

Done, I figured it would be best to add these tests in the existing test file specific to gep and constexpr operators.

[gandhi56]

ping

[fhahn]

LGTM, thanks (looks like GH won't send emails when a conversation has been resolved)

[gandhi56]

That's a bummer, thanks for the review!