[clang] Do not require GNUInlineAttr for inline builtins

Fix #61691 Differential Revision: https://reviews.llvm.org/D147307
llvm · Apr 13, 2023 · fd8745c · fd8745c
1 parent fd1d608
commit fd8745c
Show file tree

Hide file tree

Showing 3 changed files with 65 additions and 3 deletions.
diff --git a/clang/lib/AST/Decl.cpp b/clang/lib/AST/Decl.cpp
@@ -3302,8 +3302,7 @@ bool FunctionDecl::isInlineBuiltinDeclaration() const {
 
   const FunctionDecl *Definition;
   return hasBody(Definition) && Definition->isInlineSpecified() &&
-         Definition->hasAttr<AlwaysInlineAttr>() &&
-         Definition->hasAttr<GNUInlineAttr>();
+         Definition->hasAttr<AlwaysInlineAttr>();
 }
 
 bool FunctionDecl::isDestroyingOperatorDelete() const {

diff --git a/clang/test/CodeGen/memcpy-inline-builtin.c b/clang/test/CodeGen/memcpy-inline-builtin.c
@@ -3,18 +3,27 @@
 // RUN: %clang_cc1 -triple x86_64 -S -emit-llvm -o - %s | FileCheck %s
 //
 // Verifies that clang detects memcpy inline version and uses it instead of the builtin.
+// Checks alternate version with the `artificial` attribute.
 
 typedef unsigned long size_t;
 
 // Clang requires these attributes for a function to be redefined.
 #define AVAILABLE_EXTERNALLY extern inline __attribute__((always_inline)) __attribute__((gnu_inline))
 
+#define AVAILABLE_EXTERNALLY_ALTERNATE extern inline __attribute__((__always_inline__)) __attribute__((__artificial__))
+
 // Clang recognizes an inline builtin and renames it to prevent conflict with builtins.
 AVAILABLE_EXTERNALLY void *memcpy(void *a, const void *b, size_t c) {
   asm("# memcpy.inline marker");
   return __builtin_memcpy(a, b, c);
 }
 
+// Clang recognizes an inline builtin and renames it to prevent conflict with builtins.
+AVAILABLE_EXTERNALLY_ALTERNATE void *memmove(void *a, const void *b, size_t c) {
+  asm("# memmove.inline marker");
+  return __builtin_memmove(a, b, c);
+}
+
 // CHECK-LABEL: @foo(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[A_ADDR_I:%.*]] = alloca ptr, align 8
@@ -43,6 +52,34 @@ void *foo(void *a, const void *b, size_t c) {
   return memcpy(a, b, c);
 }
 
+// CHECK-LABEL: @foo_alt(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[A_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK-NEXT:    [[B_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK-NEXT:    [[C_ADDR_I:%.*]] = alloca i64, align 8
+// CHECK-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
+// CHECK-NEXT:    [[B_ADDR:%.*]] = alloca ptr, align 8
+// CHECK-NEXT:    [[C_ADDR:%.*]] = alloca i64, align 8
+// CHECK-NEXT:    store ptr [[A:%.*]], ptr [[A_ADDR]], align 8
+// CHECK-NEXT:    store ptr [[B:%.*]], ptr [[B_ADDR]], align 8
+// CHECK-NEXT:    store i64 [[C:%.*]], ptr [[C_ADDR]], align 8
+// CHECK-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[B_ADDR]], align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load i64, ptr [[C_ADDR]], align 8
+// CHECK-NEXT:    store ptr [[TMP0]], ptr [[A_ADDR_I]], align 8
+// CHECK-NEXT:    store ptr [[TMP1]], ptr [[B_ADDR_I]], align 8
+// CHECK-NEXT:    store i64 [[TMP2]], ptr [[C_ADDR_I]], align 8
+// CHECK-NEXT:    call void asm sideeffect "# memmove.inline marker", "~{dirflag},~{fpsr},~{flags}"() #[[ATTR3]], !srcloc !3
+// CHECK-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[A_ADDR_I]], align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[B_ADDR_I]], align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr [[C_ADDR_I]], align 8
+// CHECK-NEXT:    call void @llvm.memmove.p0.p0.i64(ptr align 1 [[TMP3]], ptr align 1 [[TMP4]], i64 [[TMP5]], i1 false)
+// CHECK-NEXT:    ret ptr [[TMP3]]
+//
+void *foo_alt(void *a, const void *b, size_t c) {
+  return memmove(a, b, c);
+}
+
 // CHECK-LABEL: @bar(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
@@ -68,3 +105,29 @@ void bar(void *a, const void *b, size_t c) {
   void *(*cpy)(void *, const void *, size_t) = c > 10 ? memcpy : foo;
   cpy(a, b, c);
 }
+
+// CHECK-LABEL: @bar_alt(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
+// CHECK-NEXT:    [[B_ADDR:%.*]] = alloca ptr, align 8
+// CHECK-NEXT:    [[C_ADDR:%.*]] = alloca i64, align 8
+// CHECK-NEXT:    [[CPY:%.*]] = alloca ptr, align 8
+// CHECK-NEXT:    store ptr [[A:%.*]], ptr [[A_ADDR]], align 8
+// CHECK-NEXT:    store ptr [[B:%.*]], ptr [[B_ADDR]], align 8
+// CHECK-NEXT:    store i64 [[C:%.*]], ptr [[C_ADDR]], align 8
+// CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr [[C_ADDR]], align 8
+// CHECK-NEXT:    [[CMP:%.*]] = icmp ugt i64 [[TMP0]], 10
+// CHECK-NEXT:    [[TMP1:%.*]] = zext i1 [[CMP]] to i64
+// CHECK-NEXT:    [[COND:%.*]] = select i1 [[CMP]], ptr @memmove, ptr @foo_alt
+// CHECK-NEXT:    store ptr [[COND]], ptr [[CPY]], align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[CPY]], align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[B_ADDR]], align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr [[C_ADDR]], align 8
+// CHECK-NEXT:    [[CALL:%.*]] = call ptr [[TMP2]](ptr noundef [[TMP3]], ptr noundef [[TMP4]], i64 noundef [[TMP5]])
+// CHECK-NEXT:    ret void
+//
+void bar_alt(void *a, const void *b, size_t c) {
+  void *(*cpy)(void *, const void *, size_t) = c > 10 ? memmove : foo_alt;
+  cpy(a, b, c);
+}
diff --git a/clang/test/CodeGen/pr9614.c b/clang/test/CodeGen/pr9614.c
@@ -32,7 +32,7 @@ void f(void) {
 
 // CHECK-LABEL: define{{.*}} void @f()
 // CHECK: call void @foo()
-// CHECK: call i32 @abs(i32 noundef 0)
+// CHECK: call i32 @abs(i32 noundef %0)
 // CHECK: call ptr @strrchr(
 // CHECK: call void @llvm.prefetch.p0(
 // CHECK: call ptr @memchr(