* [PATCH] selftests/mm: hmm: give migrate_remap_fault a longer timeout
@ 2026-11-19 13:39 Zenghui Yu
2026-11-19 13:54 ` Zenghui Yu
2026-11-19 22:31 ` Alistair Popple
0 siblings, 2 replies; 3+ messages in thread
From: Zenghui Yu @ 2026-11-19 13:39 UTC (permalink / raw)
To: linux-mm
Cc: jgg, leon, akpm, david, liam, vbabka, rppt, mhocko,
Zenghui Yu (Huawei)
From: "Zenghui Yu (Huawei)" <zenghui.yu@linux.dev>
migrate_remap_fault sizes its working set from the PMD page size and runs
24 combinations of migration/mremap parameters. On arm64 with 64K base
pages the PMD size is 512MB, so the test churns through tens of gigabytes
of memory and takes ~32 seconds on my test machine. This just exceeds the
kselftest harness's 30s default per-test timeout (TEST_TIMEOUT_DEFAULT),
and the test gets killed with "Test terminated by timeout" even though it
is making steady progress.
Raise the timeout to 120 seconds via TEST_F_TIMEOUT, so that the test can
run to completion on configurations with large PMD sizes.
Signed-off-by: Zenghui Yu (Huawei) <zenghui.yu@linux.dev>
---
tools/testing/selftests/mm/hmm-tests.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tools/testing/selftests/mm/hmm-tests.c b/tools/testing/selftests/mm/hmm-tests.c
index e5f273ca84c1..840015cdd209 100644
--- a/tools/testing/selftests/mm/hmm-tests.c
+++ b/tools/testing/selftests/mm/hmm-tests.c
@@ -2221,7 +2221,7 @@ TEST_F(hmm, migrate_partial_unmap_fault)
}
}
-TEST_F(hmm, migrate_remap_fault)
+TEST_F_TIMEOUT(hmm, migrate_remap_fault, 120)
{
struct hmm_buffer *buffer;
unsigned long npages;
--
2.56.0
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] selftests/mm: hmm: give migrate_remap_fault a longer timeout
2026-11-19 13:39 [PATCH] selftests/mm: hmm: give migrate_remap_fault a longer timeout Zenghui Yu
@ 2026-11-19 13:54 ` Zenghui Yu
2026-11-19 22:31 ` Alistair Popple
1 sibling, 0 replies; 3+ messages in thread
From: Zenghui Yu @ 2026-11-19 13:54 UTC (permalink / raw)
To: linux-mm; +Cc: jgg, leon, akpm, david, liam, vbabka, rppt, mhocko
On 11/19/26 9:39 PM, Zenghui Yu wrote:
> From: "Zenghui Yu (Huawei)" <zenghui.yu@linux.dev>
>
> migrate_remap_fault sizes its working set from the PMD page size and runs
> 24 combinations of migration/mremap parameters. On arm64 with 64K base
> pages the PMD size is 512MB, so the test churns through tens of gigabytes
> of memory and takes ~32 seconds on my test machine. This just exceeds the
> kselftest harness's 30s default per-test timeout (TEST_TIMEOUT_DEFAULT),
> and the test gets killed with "Test terminated by timeout" even though it
> is making steady progress.
This (and any) failure of the hmm-tests binary is silently reported as PASS
by the kselftest framework because test_hmm.sh always exits with 0, which
needs to be fixed separately.
Thanks,
Zenghui
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] selftests/mm: hmm: give migrate_remap_fault a longer timeout
2026-11-19 13:39 [PATCH] selftests/mm: hmm: give migrate_remap_fault a longer timeout Zenghui Yu
2026-11-19 13:54 ` Zenghui Yu
@ 2026-11-19 22:31 ` Alistair Popple
1 sibling, 0 replies; 3+ messages in thread
From: Alistair Popple @ 2026-11-19 22:31 UTC (permalink / raw)
To: Zenghui Yu
Cc: linux-mm, jgg, leon, akpm, david, liam, vbabka, rppt, mhocko,
balbirs
On 2026-11-20 at 00:39 +1100, Zenghui Yu <zenghui.yu@linux.dev> wrote...
> From: "Zenghui Yu (Huawei)" <zenghui.yu@linux.dev>
>
> migrate_remap_fault sizes its working set from the PMD page size and runs
> 24 combinations of migration/mremap parameters. On arm64 with 64K base
> pages the PMD size is 512MB, so the test churns through tens of gigabytes
> of memory and takes ~32 seconds on my test machine. This just exceeds the
> kselftest harness's 30s default per-test timeout (TEST_TIMEOUT_DEFAULT),
> and the test gets killed with "Test terminated by timeout" even though it
> is making steady progress.
>
> Raise the timeout to 120 seconds via TEST_F_TIMEOUT, so that the test can
> run to completion on configurations with large PMD sizes.
I guess the alternative is to reduce the number of test combinations, but that
seems like it would reduce coverage so this seems reasonable. Adding Balbir to
CC in case he has thoughts though.
Reviewed-by: Alistair Popple <apopple@nvidia.com>
> Signed-off-by: Zenghui Yu (Huawei) <zenghui.yu@linux.dev>
> ---
> tools/testing/selftests/mm/hmm-tests.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/tools/testing/selftests/mm/hmm-tests.c b/tools/testing/selftests/mm/hmm-tests.c
> index e5f273ca84c1..840015cdd209 100644
> --- a/tools/testing/selftests/mm/hmm-tests.c
> +++ b/tools/testing/selftests/mm/hmm-tests.c
> @@ -2221,7 +2221,7 @@ TEST_F(hmm, migrate_partial_unmap_fault)
> }
> }
>
> -TEST_F(hmm, migrate_remap_fault)
> +TEST_F_TIMEOUT(hmm, migrate_remap_fault, 120)
> {
> struct hmm_buffer *buffer;
> unsigned long npages;
> --
> 2.56.0
>
>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2026-11-19 22:31 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-19 13:39 [PATCH] selftests/mm: hmm: give migrate_remap_fault a longer timeout Zenghui Yu
2026-11-19 13:54 ` Zenghui Yu
2026-11-19 22:31 ` Alistair Popple
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox