[OPENMP][NVPTX]Add __kmpc_syncwarp(int32_t) function.

[ABataev]

Summary

Added function void __kmpc_syncwarp(int32_t) to expose it to the
compiler. It is required to fix the problem with the critical regions in
Cuda9.0+. We cannot use barrier in the critical region, but still need
to reconverge the threads in the warp after. This function allows to do
this.

[grokos]

I guess there is a clang-side patch as well which makes use of the new exposed kmpc function? Can you add it to the description? Also, can you add a basic test with a critical region that doesn't work correctly without this patch?

[ABataev]

In D66672#1643405, @grokos wrote:

I guess there is a clang-side patch as well which makes use of the new exposed kmpc function? Can you add it to the description? Also, can you add a basic test with a critical region that doesn't work correctly without this patch?

The test is libomptarget/deviceRTLs/nvptx/test/parallel/spmd_parallel_regions.cpp. Currently it does not work in Cuda9+.
The clang patch is D66673. It is not completed yet, the tests are not updated. The functional part is finished and tested with Cuda 9.2.

[jdoerfert]

We have uses of __SYNCWARP in the library, right? Can we use the function instead and maybe put it in the language dependent part?

[ABataev]

In D66672#1643428, @jdoerfert wrote:

We have uses of __SYNCWARP in the library, right? Can we use the function instead and maybe put it in the language dependent part?

Move __SYNCWARP macro to target_impl.h? And introduce __kmpc_impl_synwarp()?

[ABataev]

Added target specific __kmpc_impl_syncwarp() function.

[JonChesterfield]

Looks good. Delighted to see extensions to target_impl.

[ABataev]

Is this approved?

[jdoerfert]

LGTM.

openmp/libomptarget/deviceRTLs/nvptx/src/interface.h

 EXTERN void __kmpc_critical(kmp_Ident *loc, int32_t global_tid,
                             kmp_CriticalName *crit);
 EXTERN void __kmpc_end_critical(kmp_Ident *loc, int32_t global_tid,
                                 kmp_CriticalName *crit);
 EXTERN void __kmpc_flush(kmp_Ident *loc);
 
 // vote
 EXTERN int32_t __kmpc_warp_active_thread_mask();
+// syncwarp
+EXTERN void __kmpc_syncwarp(int32_t);
 
 // tasks
 EXTERN kmp_TaskDescr *__kmpc_omp_task_alloc(kmp_Ident *loc,
                                             uint32_t global_tid, int32_t flag,
                                             size_t sizeOfTaskInclPrivate,
                                             size_t sizeOfSharedTable,
                                             kmp_TaskFctPtr sub);
 EXTERN int32_t __kmpc_omp_task(kmp_Ident *loc, uint32_t global_tid,

openmp/libomptarget/deviceRTLs/nvptx/src/sync.cu

 //===------------ sync.h - NVPTX OpenMP synchronizations --------- CUDA -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // Include all synchronization.
 //
 //===----------------------------------------------------------------------===//
 
 #include "omptarget-nvptx.h"
+#include "target_impl.h"
 
 ////////////////////////////////////////////////////////////////////////////////
 // KMP Ordered calls
 ////////////////////////////////////////////////////////////////////////////////
 
 EXTERN void __kmpc_ordered(kmp_Ident *loc, int32_t tid) {
   PRINT0(LD_IO, "call kmpc_ordered\n");
 }

 ////////////////////////////////////////////////////////////////////////////////
 // Vote
 ////////////////////////////////////////////////////////////////////////////////
 
 EXTERN int32_t __kmpc_warp_active_thread_mask() {
   PRINT0(LD_IO, "call __kmpc_warp_active_thread_mask\n");
   return __ACTIVEMASK();
 }
+
+////////////////////////////////////////////////////////////////////////////////
+// Syncwarp
+////////////////////////////////////////////////////////////////////////////////
+
+EXTERN void __kmpc_syncwarp(int32_t Mask) {
+  PRINT0(LD_IO, "call __kmpc_syncwarp\n");
+  __kmpc_impl_syncwarp(Mask);
+}

openmp/libomptarget/deviceRTLs/nvptx/src/target_impl.h

INLINE __kmpc_impl_lanemask_t __kmpc_impl_lanemask_lt() {
   asm("mov.u32 %0, %%lanemask_lt;" : "=r"(res));
   return res;
 }
 
 INLINE int __kmpc_impl_ffs(uint32_t x) { return __ffs(x); }
 
 INLINE int __kmpc_impl_popc(uint32_t x) { return __popc(x); }
 
+INLINE void __kmpc_impl_syncwarp(int32_t Mask) { __SYNCWARP(Mask); }
+
 #endif