diff mbox series

[5/5] cpu: Better output when waiting for a very long job

Message ID 20180815051039.7019-5-benh@kernel.crashing.org
State Accepted
Headers show
Series [1/5] phb4: Workaround PHB errata with CFG write UR/CA errors | expand

Checks

Context Check Description
snowpatch_ozlabs/apply_patch success master/apply_patch Successfully applied
snowpatch_ozlabs/make_check success Test make_check on branch master

Commit Message

Benjamin Herrenschmidt Aug. 15, 2018, 5:10 a.m. UTC 
Instead of printing at the end if the job took more than 1s,
print in the loop every 5s along with a backtrace. This will
give us some output if the job is deadlocked.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---
 core/cpu.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

Comments

Stewart Smith Aug. 16, 2018, 9:21 a.m. UTC | #1 
Benjamin Herrenschmidt <benh@kernel.crashing.org> writes:
> Instead of printing at the end if the job took more than 1s,
> print in the loop every 5s along with a backtrace. This will
> give us some output if the job is deadlocked.
>
> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> ---
>  core/cpu.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)

Cool, as discussed on Slack, I've taken 2-5 now and bumped the time
period in 5 to 30s rather than 5s.

So, 2-5 merged to master as of b78d823faf4f727a113ad5907d10a45c55ca7138
diff mbox series

Patch

diff --git a/core/cpu.c b/core/cpu.c
index a4c4d4ff..7ae2ba11 100644
--- a/core/cpu.c
+++ b/core/cpu.c
@@ -300,13 +300,14 @@  void cpu_wait_job(struct cpu_job *job, bool free_it)
 		time_wait_ms(10);
 		time_waited += 10;
 		lwsync();
+		if ((time_waited % 5000) == 0) {
+			prlog(PR_INFO, "cpu_wait_job(%s) for %lums\n",
+			      job->name, time_waited);
+			backtrace();
+		}
 	}
 	lwsync();
 
-	if (time_waited > 1000)
-		prlog(PR_DEBUG, "cpu_wait_job(%s) for %lums\n",
-		      job->name, time_waited);
-
 	if (free_it)
 		free(job);
 }