Merge remote-tracking branch 'nccl/master' into develop

[ROCm/rccl commit: 84081064a0]
2024-10-02 09:29:22 -05:00
parent 74aa13afbe 792c33598f
commit 9059445acb
54 changed files with 2158 additions and 964 deletions
@@ -33,6 +33,15 @@ ncclResult_t PtrCheck(void* ptr, const char* opname, const char* ptrname) {
  return ncclSuccess;
 }

+ncclResult_t CommCheck(struct ncclComm* comm, const char* opname, const char* ptrname) {
+  NCCLCHECK(PtrCheck(comm, opname, ptrname));
+  if (comm->startMagic != NCCL_MAGIC || comm->endMagic != NCCL_MAGIC) {
+    WARN("Error: corrupted comm object detected");
+    return ncclInvalidArgument;
+  }
+  return ncclSuccess;
+}
+
 ncclResult_t ArgsCheck(struct ncclInfo* info) {
  // First, the easy ones
  if (info->root < 0 || info->root >= info->comm->nRanks) {