caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
0c4d7d14e94de5130d7747bde70c94bce77bb754
clang-p2996/clang/test/OpenMP/target_uses_allocators.c
Shilei Tian 5ba8ecb6cc [Clang][OpenMP] Find the type omp_allocator_handle_t from identifier table 
In Clang, in order to determine the type of `omp_allocator_handle_t`, Clang
checks the type of those predefined allocators. The first one it checks is
`omp_null_allocator`. If the language is C, and the system is 64-bit, what Clang
gets is a `int`, instead of an enum of size 8, given the fact how we define
`omp_allocator_handle_t` in `omp.h`.  If the allocator is captured by a region,
let's say a parallel region, the allocator will be privatized. Because Clang deems
`omp_allocator_handle_t` as an `int`, it will first cast the value returned by
the runtime library (for `libomp` it is a `void *`) to `int`, and then in the
outlined function, it casts back to `omp_allocator_handle_t`. This two casts
completely shaves the first 32-bit of the pointer value returned from `libomp`,
and when the private "new" pointer is fed to another runtime function
`__kmpc_allocate()`, it causes segment fault. That is the root cause of PR54082.
I have no idea why `-fno-pic` could hide this bug.

In this patch, we detect `omp_allocator_handle_t` using roughly the same method
as `omp_event_handle_t`, by looking it up into the identifier table.

Fix #54082.

Reviewed By: ABataev

Differential Revision: https://reviews.llvm.org/D142297
2023-01-24 22:49:05 -05:00

109 lines
6.0 KiB
C
Raw Blame History

// Test host codegen.
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=50 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -verify -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=50 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -include-pch %t %s -emit-llvm -o - | FileCheck %s
#ifndef HEADER
#define HEADER
typedef enum omp_allocator_handle_t {
omp_null_allocator = 0,
omp_default_mem_alloc = 1,
omp_large_cap_mem_alloc = 2,
omp_const_mem_alloc = 3,
omp_high_bw_mem_alloc = 4,
omp_low_lat_mem_alloc = 5,
omp_cgroup_mem_alloc = 6,
omp_pteam_mem_alloc = 7,
omp_thread_mem_alloc = 8,
KMP_ALLOCATOR_MAX_HANDLE = __UINTPTR_MAX__
} omp_allocator_handle_t;
// CHECK: define {{.*}}[[FIE:@.+]]()
void fie(void) {
int x;
#pragma omp target uses_allocators(omp_null_allocator) allocate(omp_null_allocator: x) firstprivate(x)
{}
#pragma omp target uses_allocators(omp_default_mem_alloc) allocate(omp_default_mem_alloc: x) firstprivate(x)
{}
#pragma omp target uses_allocators(omp_large_cap_mem_alloc) allocate(omp_large_cap_mem_alloc: x) firstprivate(x)
{}
#pragma omp target uses_allocators(omp_const_mem_alloc) allocate(omp_const_mem_alloc: x) firstprivate(x)
{}
#pragma omp target uses_allocators(omp_high_bw_mem_alloc) allocate(omp_high_bw_mem_alloc: x) firstprivate(x)
{}
#pragma omp target uses_allocators(omp_low_lat_mem_alloc) allocate(omp_low_lat_mem_alloc: x) firstprivate(x)
{}
#pragma omp target uses_allocators(omp_cgroup_mem_alloc) allocate(omp_cgroup_mem_alloc: x) firstprivate(x)
{}
#pragma omp target uses_allocators(omp_pteam_mem_alloc) allocate(omp_pteam_mem_alloc: x) firstprivate(x)
{}
#pragma omp target uses_allocators(omp_thread_mem_alloc) allocate(omp_thread_mem_alloc: x) firstprivate(x) // expected-warning {{allocator with the 'thread' trait access has unspecified behavior on 'target' directive}}
{}
}
#endif
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr null)
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr null)
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr inttoptr (i64 1 to ptr))
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr inttoptr (i64 2 to ptr))
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr inttoptr (i64 2 to ptr))
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr inttoptr (i64 3 to ptr))
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr inttoptr (i64 3 to ptr))
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr inttoptr (i64 4 to ptr))
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr inttoptr (i64 4 to ptr))
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr inttoptr (i64 5 to ptr))
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr inttoptr (i64 5 to ptr))
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr inttoptr (i64 6 to ptr))
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr inttoptr (i64 6 to ptr))
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr inttoptr (i64 7 to ptr))
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr inttoptr (i64 7 to ptr))
// CHECK: %[[#R0:]] = call i32 @__kmpc_global_thread_num(ptr @1)
// CHECK-NEXT: store i64 %x, ptr %x.addr, align 8
// CHECK-NEXT: %.x..void.addr = call ptr @__kmpc_alloc(i32 %[[#R0]], i64 4, ptr inttoptr (i64 8 to ptr))
// CHECK-NEXT: %[[#R1:]] = load i32, ptr %x.addr, align 4
// CHECK-NEXT: store i32 %[[#R1]], ptr %.x..void.addr, align 4
// CHECK-NEXT: call void @__kmpc_free(i32 %[[#R0]], ptr %.x..void.addr, ptr inttoptr (i64 8 to ptr))
Reference in New Issue View Git Blame Copy Permalink