[clang] [Clang][ARM] Only try to redefine builtins for non-CUDA (PR #128222)

[Nick Sarnie via cfe-commits]

https://github.com/sarnex updated https://github.com/llvm/llvm-project/pull/128222

>From b91e429ae1fc1cf8ea8239aba59789013961533c Mon Sep 17 00:00:00 2001
From: "Sarnie, Nick" <nick.sarnie at intel.com>
Date: Fri, 21 Feb 2025 11:50:48 -0800
Subject: [PATCH 1/2] [Clang][ARM] Only try to redefine builtins for non-CUDA

Signed-off-by: Sarnie, Nick <nick.sarnie at intel.com>
---
 clang/lib/Headers/arm_acle.h | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/clang/lib/Headers/arm_acle.h b/clang/lib/Headers/arm_acle.h
index b1dc90f84ad36..fbd0020258722 100644
--- a/clang/lib/Headers/arm_acle.h
+++ b/clang/lib/Headers/arm_acle.h
@@ -27,6 +27,8 @@
 extern "C" {
 #endif
 
+#if !defined(__CUDA__)
+
 /* 7 SYNCHRONIZATION, BARRIER AND HINT INTRINSICS */
 /* 7.3 Memory barriers */
 #if !__has_builtin(__dmb)
@@ -70,6 +72,7 @@ static __inline__ void __attribute__((__always_inline__, __nodebug__)) __yield(v
   __builtin_arm_yield();
 }
 #endif
+#endif // #if !defined(__CUDA__)
 
 #if defined(__ARM_32BIT_STATE) && __ARM_32BIT_STATE
 #define __dbg(t) __builtin_arm_dbg(t)

>From ee842617b798f9c9f529f96ee6b05afcab1ee95b Mon Sep 17 00:00:00 2001
From: "Sarnie, Nick" <nick.sarnie at intel.com>
Date: Fri, 21 Feb 2025 13:41:04 -0800
Subject: [PATCH 2/2] address feedback

Signed-off-by: Sarnie, Nick <nick.sarnie at intel.com>
---
 clang/lib/Headers/arm_acle.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/clang/lib/Headers/arm_acle.h b/clang/lib/Headers/arm_acle.h
index fbd0020258722..c8857cba83531 100644
--- a/clang/lib/Headers/arm_acle.h
+++ b/clang/lib/Headers/arm_acle.h
@@ -27,7 +27,7 @@
 extern "C" {
 #endif
 
-#if !defined(__CUDA__)
+#if !defined(__CUDA_ARCH__)
 
 /* 7 SYNCHRONIZATION, BARRIER AND HINT INTRINSICS */
 /* 7.3 Memory barriers */
@@ -72,7 +72,7 @@ static __inline__ void __attribute__((__always_inline__, __nodebug__)) __yield(v
   __builtin_arm_yield();
 }
 #endif
-#endif // #if !defined(__CUDA__)
+#endif // #if !defined(__CUDA_ARCH__)
 
 #if defined(__ARM_32BIT_STATE) && __ARM_32BIT_STATE
 #define __dbg(t) __builtin_arm_dbg(t)