On 12/01/2016 03:03 AM, Jaze Lee wrote:
> Hello,
> Dose this issue can be fixed? Or the only method is to use offload
> to workaround.
Can you try the below patch? We simply need to make sure that we always
regrow, when we enter an idle state. The logic was a little warped, the
below is cleaner.
diff --git a/backend.c b/backend.c
index 8616fc2..ac71521 100644
--- a/backend.c
+++ b/backend.c
@@ -441,11 +441,8 @@ static int wait_for_completions(struct thread_data *td, struct timeval *time)
int min_evts = 0;
int ret;
- if (td->flags & TD_F_REGROW_LOGS) {
- ret = io_u_quiesce(td);
- regrow_logs(td);
- return ret;
- }
+ if (td->flags & TD_F_REGROW_LOGS)
+ return io_u_quiesce(td);
/*
* if the queue is full, we MUST reap at least 1 event
diff --git a/io_u.c b/io_u.c
index 428d4b7..7420629 100644
--- a/io_u.c
+++ b/io_u.c
@@ -653,6 +653,9 @@ int io_u_quiesce(struct thread_data *td)
completed += ret;
}
+ if (td->flags & TD_F_REGROW_LOGS)
+ regrow_logs(td);
+
return completed;
}
--
Jens Axboe
--
To unsubscribe from this list: send the line "unsubscribe fio" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
