Changed TestBedChild to avoid hang if the call fails (#1875)

Changed `TestBedChild` protocol to send the result code before the return value to avoid hanging if the call fails. Switched `TestBedChild::GetUniqueId` to use this.
This commit is contained in:
corey-derochie-amd
2025-08-22 23:17:34 -06:00
committed by GitHub
parent bf6660ee4e
commit b88c134874
3 changed files with 21 additions and 15 deletions
+1 -1
View File
@@ -142,11 +142,11 @@ namespace RcclUnitTesting
// Tell first rank to get ncclUniqueId
int getIdCmd = TestBedChild::CHILD_GET_UNIQUE_ID;
PIPE_WRITE(0, getIdCmd);
PIPE_CHECK(0);
// Receive back unique ID from first rank
ncclUniqueId id;
PIPE_READ(0, id);
PIPE_CHECK(0);
// Send InitComms command to each active child process
int const cmd = TestBedChild::CHILD_INIT_COMMS;