[clang] [Clang] Remove 't' from __builtin_amdgcn_global_atomic_fadd_f32/f64 (PR #173480)
Juan Manuel Martinez Caamaño via cfe-commits
cfe-commits at lists.llvm.org
Wed Dec 24 04:09:40 PST 2025
https://github.com/jmmartinez created https://github.com/llvm/llvm-project/pull/173480
Allows for type checking depending on the built-in signature.
>From 892659b38b9ed790e5baa89848c738319ea32804 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
<jmartinezcaamao at gmail.com>
Date: Wed, 24 Dec 2025 11:43:15 +0100
Subject: [PATCH 1/2] Pre-commit test: [Clang] Remove 't' from
__builtin_amdgcn_global_atomic_fadd_f32/f64
---
.../CodeGenHIP/amdgpu-global-atomic-fadd.hip | 65 +++++++++++++++++++
.../SemaHIP/amdgpu-global-atomic-fadd-err.hip | 39 +++++++++++
2 files changed, 104 insertions(+)
create mode 100644 clang/test/CodeGenHIP/amdgpu-global-atomic-fadd.hip
create mode 100644 clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip
diff --git a/clang/test/CodeGenHIP/amdgpu-global-atomic-fadd.hip b/clang/test/CodeGenHIP/amdgpu-global-atomic-fadd.hip
new file mode 100644
index 0000000000000..1fa3a20442064
--- /dev/null
+++ b/clang/test/CodeGenHIP/amdgpu-global-atomic-fadd.hip
@@ -0,0 +1,65 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 6
+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -target-cpu gfx942 -emit-llvm -fcuda-is-device %s -o - | FileCheck %s
+
+#define __device__ __attribute__((device))
+
+__device__ float global_float;
+__device__ double global_double;
+
+// CHECK-LABEL: define dso_local void @_Z33test_global_atomic_fadd_f32_validPff(
+// CHECK-SAME: ptr noundef [[PTR:%.*]], float noundef [[VAL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[PTR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5)
+// CHECK-NEXT: [[VAL_ADDR:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[RESULT:%.*]] = alloca float, align 4, addrspace(5)
+// CHECK-NEXT: [[PTR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[PTR_ADDR]] to ptr
+// CHECK-NEXT: [[VAL_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VAL_ADDR]] to ptr
+// CHECK-NEXT: [[RESULT_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RESULT]] to ptr
+// CHECK-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR_ASCAST]], align 8
+// CHECK-NEXT: store float [[VAL]], ptr [[VAL_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
+// CHECK-NEXT: [[TMP1:%.*]] = addrspacecast ptr [[TMP0]] to ptr addrspace(1)
+// CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[VAL_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw fadd ptr addrspace(1) [[TMP1]], float [[TMP2]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4:![0-9]+]], !amdgpu.ignore.denormal.mode [[META4]]
+// CHECK-NEXT: store float [[TMP3]], ptr [[RESULT_ASCAST]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[VAL_ADDR_ASCAST]], align 4
+// CHECK-NEXT: [[TMP5:%.*]] = atomicrmw fadd ptr addrspace(1) @global_float, float [[TMP4]] syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory [[META4]], !amdgpu.ignore.denormal.mode [[META4]]
+// CHECK-NEXT: store float [[TMP5]], ptr [[RESULT_ASCAST]], align 4
+// CHECK-NEXT: ret void
+//
+__device__ void test_global_atomic_fadd_f32_valid(float *ptr, float val) {
+ float result;
+ result = __builtin_amdgcn_global_atomic_fadd_f32(ptr, val);
+ result = __builtin_amdgcn_global_atomic_fadd_f32(&global_float, val);
+}
+
+// CHECK-LABEL: define dso_local void @_Z33test_global_atomic_fadd_f64_validPdd(
+// CHECK-SAME: ptr noundef [[PTR:%.*]], double noundef [[VAL:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[PTR_ADDR:%.*]] = alloca ptr, align 8, addrspace(5)
+// CHECK-NEXT: [[VAL_ADDR:%.*]] = alloca double, align 8, addrspace(5)
+// CHECK-NEXT: [[RESULT:%.*]] = alloca double, align 8, addrspace(5)
+// CHECK-NEXT: [[PTR_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[PTR_ADDR]] to ptr
+// CHECK-NEXT: [[VAL_ADDR_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[VAL_ADDR]] to ptr
+// CHECK-NEXT: [[RESULT_ASCAST:%.*]] = addrspacecast ptr addrspace(5) [[RESULT]] to ptr
+// CHECK-NEXT: store ptr [[PTR]], ptr [[PTR_ADDR_ASCAST]], align 8
+// CHECK-NEXT: store double [[VAL]], ptr [[VAL_ADDR_ASCAST]], align 8
+// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR_ADDR_ASCAST]], align 8
+// CHECK-NEXT: [[TMP1:%.*]] = addrspacecast ptr [[TMP0]] to ptr addrspace(1)
+// CHECK-NEXT: [[TMP2:%.*]] = load double, ptr [[VAL_ADDR_ASCAST]], align 8
+// CHECK-NEXT: [[TMP3:%.*]] = atomicrmw fadd ptr addrspace(1) [[TMP1]], double [[TMP2]] syncscope("agent") monotonic, align 8, !amdgpu.no.fine.grained.memory [[META4]]
+// CHECK-NEXT: store double [[TMP3]], ptr [[RESULT_ASCAST]], align 8
+// CHECK-NEXT: [[TMP4:%.*]] = load double, ptr [[VAL_ADDR_ASCAST]], align 8
+// CHECK-NEXT: [[TMP5:%.*]] = atomicrmw fadd ptr addrspace(1) @global_double, double [[TMP4]] syncscope("agent") monotonic, align 8, !amdgpu.no.fine.grained.memory [[META4]]
+// CHECK-NEXT: store double [[TMP5]], ptr [[RESULT_ASCAST]], align 8
+// CHECK-NEXT: ret void
+//
+__device__ void test_global_atomic_fadd_f64_valid(double *ptr, double val) {
+ double result;
+ result = __builtin_amdgcn_global_atomic_fadd_f64(ptr, val);
+ result = __builtin_amdgcn_global_atomic_fadd_f64(&global_double, val);
+}
+//.
+// CHECK: [[META4]] = !{}
+//.
diff --git a/clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip b/clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip
new file mode 100644
index 0000000000000..9bf4a841e8dfd
--- /dev/null
+++ b/clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip
@@ -0,0 +1,39 @@
+// RUN: %clang_cc1 -fsyntax-only -triple amdgcn -target-cpu gfx942 -verify %s -fcuda-is-device
+// RUN: %clang_cc1 -fsyntax-only -triple x86_64 -aux-triple amdgcn -verify %s
+// expected-no-diagnostics
+
+#define __device__ __attribute__((device))
+#define __shared__ __attribute__((shared))
+
+__device__ float global_float;
+__device__ float global_double;
+
+__device__ void test_global_atomic_fadd_f32_valid(float *ptr, float val) {
+ float result;
+ result = __builtin_amdgcn_global_atomic_fadd_f32(ptr, val);
+ result = __builtin_amdgcn_global_atomic_fadd_f32(&global_float, val);
+}
+
+__device__ void test_global_atomic_fadd_f32_errors(float *ptr, float val,
+ __shared__ float *lds_ptr,
+ double *ptr_d) {
+ float result;
+ result = __builtin_amdgcn_global_atomic_fadd_f32(ptr, val, 0);
+ result = __builtin_amdgcn_global_atomic_fadd_f32(lds_ptr, val);
+ result = __builtin_amdgcn_global_atomic_fadd_f32(ptr_d, val);
+}
+
+__device__ void test_global_atomic_fadd_f64_valid(double *ptr, double val) {
+ double result;
+ result = __builtin_amdgcn_global_atomic_fadd_f64(ptr, val);
+ result = __builtin_amdgcn_global_atomic_fadd_f32(&global_double, val);
+}
+
+__device__ void test_global_atomic_fadd_f64_errors(double *ptr, double val,
+ __shared__ double *lds_ptr,
+ float *ptr_f) {
+ double result;
+ result = __builtin_amdgcn_global_atomic_fadd_f64(ptr, val, 0);
+ result = __builtin_amdgcn_global_atomic_fadd_f64(lds_ptr, val);
+ result = __builtin_amdgcn_global_atomic_fadd_f64(ptr_f, val);
+}
>From 619a5e0741ae15c1bc34ad2b85a6e9e87efa4c1d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
<jmartinezcaamao at gmail.com>
Date: Wed, 24 Dec 2025 11:43:57 +0100
Subject: [PATCH 2/2] [Clang] Remove 't' from
__builtin_amdgcn_global_atomic_fadd_f32/f64
---
clang/include/clang/Basic/BuiltinsAMDGPU.def | 4 ++--
clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip | 9 ++++-----
2 files changed, 6 insertions(+), 7 deletions(-)
diff --git a/clang/include/clang/Basic/BuiltinsAMDGPU.def b/clang/include/clang/Basic/BuiltinsAMDGPU.def
index 2623bd476f08f..224aa2ea30bad 100644
--- a/clang/include/clang/Basic/BuiltinsAMDGPU.def
+++ b/clang/include/clang/Basic/BuiltinsAMDGPU.def
@@ -266,8 +266,8 @@ TARGET_BUILTIN(__builtin_amdgcn_perm, "UiUiUiUi", "nc", "gfx8-insts")
TARGET_BUILTIN(__builtin_amdgcn_fmed3h, "hhhh", "nc", "gfx9-insts")
-TARGET_BUILTIN(__builtin_amdgcn_global_atomic_fadd_f64, "dd*1d", "t", "gfx90a-insts")
-TARGET_BUILTIN(__builtin_amdgcn_global_atomic_fadd_f32, "ff*1f", "t", "atomic-fadd-rtn-insts")
+TARGET_BUILTIN(__builtin_amdgcn_global_atomic_fadd_f64, "dd*1d", "", "gfx90a-insts")
+TARGET_BUILTIN(__builtin_amdgcn_global_atomic_fadd_f32, "ff*1f", "", "atomic-fadd-rtn-insts")
TARGET_BUILTIN(__builtin_amdgcn_global_atomic_fadd_v2f16, "V2hV2h*1V2h", "t", "atomic-buffer-global-pk-add-f16-insts")
TARGET_BUILTIN(__builtin_amdgcn_global_atomic_fmin_f64, "dd*1d", "t", "gfx90a-insts")
TARGET_BUILTIN(__builtin_amdgcn_global_atomic_fmax_f64, "dd*1d", "t", "gfx90a-insts")
diff --git a/clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip b/clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip
index 9bf4a841e8dfd..dc8697debc9bc 100644
--- a/clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip
+++ b/clang/test/SemaHIP/amdgpu-global-atomic-fadd-err.hip
@@ -1,6 +1,5 @@
// RUN: %clang_cc1 -fsyntax-only -triple amdgcn -target-cpu gfx942 -verify %s -fcuda-is-device
// RUN: %clang_cc1 -fsyntax-only -triple x86_64 -aux-triple amdgcn -verify %s
-// expected-no-diagnostics
#define __device__ __attribute__((device))
#define __shared__ __attribute__((shared))
@@ -18,9 +17,9 @@ __device__ void test_global_atomic_fadd_f32_errors(float *ptr, float val,
__shared__ float *lds_ptr,
double *ptr_d) {
float result;
- result = __builtin_amdgcn_global_atomic_fadd_f32(ptr, val, 0);
+ result = __builtin_amdgcn_global_atomic_fadd_f32(ptr, val, 0); // expected-error{{too many arguments to function call, expected 2, have 3}}
result = __builtin_amdgcn_global_atomic_fadd_f32(lds_ptr, val);
- result = __builtin_amdgcn_global_atomic_fadd_f32(ptr_d, val);
+ result = __builtin_amdgcn_global_atomic_fadd_f32(ptr_d, val); // expected-error{{cannot initialize a parameter of type}}
}
__device__ void test_global_atomic_fadd_f64_valid(double *ptr, double val) {
@@ -33,7 +32,7 @@ __device__ void test_global_atomic_fadd_f64_errors(double *ptr, double val,
__shared__ double *lds_ptr,
float *ptr_f) {
double result;
- result = __builtin_amdgcn_global_atomic_fadd_f64(ptr, val, 0);
+ result = __builtin_amdgcn_global_atomic_fadd_f64(ptr, val, 0); // expected-error{{too many arguments to function call, expected 2, have 3}}
result = __builtin_amdgcn_global_atomic_fadd_f64(lds_ptr, val);
- result = __builtin_amdgcn_global_atomic_fadd_f64(ptr_f, val);
+ result = __builtin_amdgcn_global_atomic_fadd_f64(ptr_f, val); // expected-error{{cannot initialize a parameter of type}}
}
More information about the cfe-commits
mailing list