Update Barrier_All and Sync_All APIs (#72)

* Fix deadlock in `rocshmem_ctx_wg_barrier_all` API in IPC conduit by adding per-context pSync buffers and context IDs
  - Added separate pSync buffers for each device context
  - Resolved deadlock when invoking barrier API (`rocshmem_ctx_wg_barrier_all`) concurrently from multiple contexts

* Update barrier_all functional tests for multi-context support

* Add thread, wavefront, and workgroup-level barrier_all APIs in IPC and RO conduits
  - Implemented barrier_all APIs at thread, wavefront, and workgroup granularity
  - Added support in both IPC and RO conduits
  - Updated functional tests to cover all `barrier_all` APIs

* Add thread, wavefront, and workgroup-level sync_all APIs in IPC and RO conduits
  - Implemented sync_all APIs for thread, wavefront, and workgroup scopes
  - Added support into both IPC and RO conduits
  - Added functional tests to cover all `sync_all` APIs

[ROCm/rocshmem commit: c652f58cef]

projects/rocshmem/include/rocshmem/rocshmem.hpp

@@ -490,6 +490,36 @@ __device__ int rocshmem_team_translate_pe(rocshmem_team_t src_team,
                                            int src_pe,
                                            rocshmem_team_t dest_team);
 
+/**
+ * @brief perform a collective barrier between all PEs in the system.
+ * The caller is blocked until the barrier is resolved.
+ *
+ * This function must be invoked by a single thread within the PE.
+ *
+ * @param[in] handle GPU side handle.
+ *
+ * @return void
+ */
+__device__ ATTR_NO_INLINE void rocshmem_ctx_barrier_all(
+    rocshmem_ctx_t ctx);
+
+__device__ ATTR_NO_INLINE void rocshmem_barrier_all();
+
+/**
+ * @brief perform a collective barrier between all PEs in the system.
+ * The caller is blocked until the barrier is resolved.
+ *
+ * This function must be called as a wave-front collective.
+ *
+ * @param[in] handle GPU side handle.
+ *
+ * @return void
+ */
+__device__ ATTR_NO_INLINE void rocshmem_ctx_wave_barrier_all(
+    rocshmem_ctx_t ctx);
+
+__device__ ATTR_NO_INLINE void rocshmem_wave_barrier_all();
+
 /**
  * @brief perform a collective barrier between all PEs in the system.
  * The caller is blocked until the barrier is resolved.
@@ -515,6 +545,42 @@ __device__ ATTR_NO_INLINE void rocshmem_wg_barrier_all();
  */
 __device__ void rocshmem_barrier(rocshmem_team_t);
 
+/**
+ * @brief registers the arrival of a PE at a barrier.
+ * The caller is blocked until the synchronization is resolved.
+ *
+ * In contrast with the shmem_barrier_all routine, shmem_sync_all only ensures
+ * completion and visibility of previously issued memory stores and does not
+ * ensure completion of remote memory updates issued via OpenSHMEM routines.
+ *
+ * This function must be invoked by a single thread within the PE.
+ *
+ * @param[in] handle GPU side handle.
+ *
+ * @return void
+ */
+__device__ ATTR_NO_INLINE void rocshmem_ctx_sync_all(rocshmem_ctx_t ctx);
+
+__device__ ATTR_NO_INLINE void rocshmem_sync_all();
+
+/**
+ * @brief registers the arrival of a PE at a barrier.
+ * The caller is blocked until the synchronization is resolved.
+ *
+ * In contrast with the shmem_barrier_all routine, shmem_sync_all only ensures
+ * completion and visibility of previously issued memory stores and does not
+ * ensure completion of remote memory updates issued via OpenSHMEM routines.
+ *
+ * This function must be called as a wave-front collective.
+ *
+ * @param[in] handle GPU side handle.
+ *
+ * @return void
+ */
+__device__ ATTR_NO_INLINE void rocshmem_ctx_wave_sync_all(rocshmem_ctx_t ctx);
+
+__device__ ATTR_NO_INLINE void rocshmem_wave_sync_all();
+
 /**
  * @brief registers the arrival of a PE at a barrier.
  * The caller is blocked until the synchronization is resolved.