[clang][Sema] Handle target_clones redeclarations that omit the attribute

[henrybw]

This patch adds a case to CheckMultiVersionAdditionalDecl() that detects redeclarations of target_clones functions which omit the attribute, and makes sure they are marked as redeclarations. It also updates the comment at the call site of CheckMultiVersionAdditionalDecl() to reflect this.

Previously, target_clones multiversioned functions that omitted the attribute from subsequent declarations would cause Clang to hit an llvm_unreachable and crash. In the following example, the second declaration (the function definition) should inherit the target_clones attribute from the first declaration (the forward declaration):

__attribute__((target_clones("arch=atom", "default")))
void foo(void);

void foo(void) { /* ... */ }

However, CheckMultiVersionAdditionalDecl() was not recognizing the function definition as a redeclaration of the forward declaration, which prevented Sema::MergeFunctionDecl() from automatically inheriting the attribute.

A side effect of this fix is that Clang now catches redeclarations of target_clones functions that have conflicting types, which previously caused Clang to crash by hitting that same llvm_unreachable. The bad_overload1 case in clang/test/Sema/attr-target-clones.c has been updated to reflect this.

Fixes #165517
Fixes #129483

[github-actions]

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

[llvmbot]

@llvm/pr-subscribers-clang

Author: Henry Baba-Weiss (henrybw)

Changes

This patch adds a case to CheckMultiVersionAdditionalDecl() that detects redeclarations of target_clones functions which omit the attribute, and makes sure they are marked as redeclarations. It also updates the comment at the call site of CheckMultiVersionAdditionalDecl() to reflect this.

Previously, target_clones multiversioned functions that omitted the attribute from subsequent declarations would cause Clang to hit an llvm_unreachable and crash. In the following example, the second declaration (the function definition) should inherit the target_clones attribute from the first declaration (the forward declaration):

__attribute__((target_clones("arch=atom", "default")))
void foo(void);

void foo(void) { /* ... */ }

However, CheckMultiVersionAdditionalDecl() was not recognizing the function definition as a redeclaration of the forward declaration, which prevented Sema::MergeFunctionDecl() from automatically inheriting the attribute.

A side effect of this fix is that Clang now catches redeclarations of target_clones functions that have conflicting types, which previously caused Clang to crash by hitting that same llvm_unreachable. The bad_overload1 case in clang/test/Sema/attr-target-clones.c has been updated to reflect this.

Fixes #165517
Fixes #129483

---

Full diff: https://github.com/llvm/llvm-project/pull/169259.diff

3 Files Affected:

• (modified) clang/lib/Sema/SemaDecl.cpp (+13-2)
• (modified) clang/test/CodeGen/attr-target-clones.c (+29)
• (modified) clang/test/Sema/attr-target-clones.c (+13)

diff --git a/clang/lib/Sema/SemaDecl.cpp b/clang/lib/Sema/SemaDecl.cpp
index b8ca2a376fde8..651437a6f4c30 100644
--- a/clang/lib/Sema/SemaDecl.cpp
+++ b/clang/lib/Sema/SemaDecl.cpp
@@ -11996,6 +11996,16 @@ static bool CheckMultiVersionAdditionalDecl(
     }
   }
 
+  // Redeclarations of a target_clones function may omit the attribute, in which
+  // case it will be inherited during declaration merging.
+  if (NewMVKind == MultiVersionKind::None &&
+      OldMVKind == MultiVersionKind::TargetClones) {
+    NewFD->setIsMultiVersion();
+    Redeclaration = true;
+    OldDecl = OldFD;
+    return false;
+  }
+
   // Else, this is simply a non-redecl case.  Checking the 'value' is only
   // necessary in the Target case, since The CPUSpecific/Dispatch cases are
   // handled in the attribute adding step.
@@ -12119,8 +12129,9 @@ static bool CheckMultiVersionFunction(Sema &S, FunctionDecl *NewFD,
   }
 
   // At this point, we have a multiversion function decl (in OldFD) AND an
-  // appropriate attribute in the current function decl.  Resolve that these are
-  // still compatible with previous declarations.
+  // appropriate attribute in the current function decl (unless it's allowed to
+  // omit the attribute).  Resolve that these are still compatible with previous
+  // declarations.
   return CheckMultiVersionAdditionalDecl(S, OldFD, NewFD, NewCPUDisp,
                                          NewCPUSpec, NewClones, Redeclaration,
                                          OldDecl, Previous);
diff --git a/clang/test/CodeGen/attr-target-clones.c b/clang/test/CodeGen/attr-target-clones.c
index 295b25d6478eb..56db77c2b09a3 100644
--- a/clang/test/CodeGen/attr-target-clones.c
+++ b/clang/test/CodeGen/attr-target-clones.c
@@ -125,6 +125,35 @@ void __attribute__((target_clones("default, arch=ivybridge"))) unused(void) {}
 // WINDOWS: musttail call void @unused.arch_ivybridge.0
 // WINDOWS: musttail call void @unused.default.1
 
+int __attribute__((target_clones("sse4.2, default"))) inherited(void);
+int inherited(void) { return 0; }
+// LINUX: define {{.*}}i32 @inherited.sse4.2.0()
+// LINUX: define {{.*}}i32 @inherited.default.1()
+// LINUX: define weak_odr ptr @inherited.resolver() #[[ATTR_RESOLVER]] comdat
+// LINUX: ret ptr @inherited.sse4.2.0
+// LINUX: ret ptr @inherited.default.1
+
+// DARWIN: define {{.*}}i32 @inherited.sse4.2.0()
+// DARWIN: define {{.*}}i32 @inherited.default.1()
+// DARWIN: define weak_odr ptr @inherited.resolver() #[[ATTR_RESOLVER]] {
+// DARWIN: ret ptr @inherited.sse4.2.0
+// DARWIN: ret ptr @inherited.default.1
+
+// WINDOWS: define dso_local i32 @inherited.sse4.2.0()
+// WINDOWS: define dso_local i32 @inherited.default.1()
+// WINDOWS: define weak_odr dso_local i32 @inherited() #[[ATTR_RESOLVER]] comdat
+// WINDOWS: musttail call i32 @inherited.sse4.2.0
+// WINDOWS: musttail call i32 @inherited.default.1
+
+int test_inherited(void) {
+  // LINUX: define {{.*}}i32 @test_inherited() #[[DEF:[0-9]+]]
+  // DARWIN: define {{.*}}i32 @test_inherited() #[[DEF:[0-9]+]]
+  // WINDOWS: define dso_local i32 @test_inherited() #[[DEF:[0-9]+]]
+  return inherited();
+  // LINUX: call i32 @inherited()
+  // DARWIN: call i32 @inherited()
+  // WINDOWS: call i32 @inherited()
+}
 
 inline int __attribute__((target_clones("arch=sandybridge,default,sse4.2")))
 foo_inline(void) { return 0; }
diff --git a/clang/test/Sema/attr-target-clones.c b/clang/test/Sema/attr-target-clones.c
index 4597ea54d02bf..40688772eeb96 100644
--- a/clang/test/Sema/attr-target-clones.c
+++ b/clang/test/Sema/attr-target-clones.c
@@ -28,6 +28,17 @@ int __attribute__((target_clones("sse4.2", "arch=atom", "default"))) redecl4(voi
 int __attribute__((target_clones("sse4.2", "arch=sandybridge", "default")))
 redecl4(void) { return 1; }
 
+int __attribute__((target_clones("sse4.2", "default"))) redecl5(void);
+int redecl5(void) { return 1; }
+
+int redecl6(void);
+int __attribute__((target_clones("sse4.2", "default"))) redecl6(void) { return 1; }
+
+int __attribute__((target_clones("sse4.2", "default"))) redecl7(void);
+// expected-error@+2 {{multiversioning attributes cannot be combined}}
+// expected-note@-2 {{previous declaration is here}}
+int __attribute__((target("sse4.2"))) redecl7(void) { return 1; }
+
 int __attribute__((target("sse4.2"))) redef2(void) { return 1; }
 // expected-error@+2 {{multiversioning attributes cannot be combined}}
 // expected-note@-2 {{previous declaration is here}}
@@ -87,6 +98,8 @@ int useage(void) {
 int __attribute__((target_clones("sse4.2", "default"))) mv_after_use(void) { return 1; }
 
 void bad_overload1(void) __attribute__((target_clones("mmx", "sse4.2", "default")));
+// expected-error@+2 {{conflicting types for 'bad_overload1'}}
+// expected-note@-2 {{previous declaration is here}}
 void bad_overload1(int p) {}
 
 void bad_overload2(int p) {}

[henrybw]

cc @erichkeane @shafik

[zwuis]

Thank you for your patch! Please add a release note entry to "clang/docs/ReleaseNotes.rst" so that users can know the improvement.

[erichkeane]

I think this is right. I'm happy to let this go and see what happens.

[henrybw]

I think this is right. I'm happy to let this go and see what happens.

@erichkeane Thanks for the review! I don't have commit access, so could you merge this on my behalf?

[github-actions]

@henrybw Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!