Linux-mm Archive on lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] mm/damon/sysfs: make next_update_jiffies per-kdamond
@ 2026-10-17  6:31 Karthikeyan KS
  2026-10-17 16:01 ` SJ Park
  0 siblings, 1 reply; 2+ messages in thread
From: Karthikeyan KS @ 2026-10-17  6:31 UTC (permalink / raw)
  To: sj; +Cc: akpm, damon, linux-mm, linux-kernel, Karthikeyan KS

damon_sysfs_next_update_jiffies was a single static variable shared by
all kdamonds, so with two or more kdamonds using refresh_ms their
threads race on it in damon_sysfs_repeat_call_fn().

Signed-off-by: Karthikeyan KS <karthiproffesional@gmail.com>
---
Testing notes for review:

- Found and confirmed with KCSAN.
- Repro setup: QEMU (x86_64, 2GB RAM, multiple vCPUs), two paddr
  kdamonds, both refresh_ms non-zero. kdamond.0 watching [0, 1GB),
  kdamond.1 watching [1GB, 2GB).
- On the unpatched kernel, KCSAN kept reporting the race on
  damon_sysfs_next_update_jiffies, again and again.
- Ran the same setup after the fix, KCSAN still on, nothing reported.
- Tried the same with vaddr kdamonds too, could not repro there.

 mm/damon/sysfs.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/mm/damon/sysfs.c b/mm/damon/sysfs.c
index 160292f077dd..872b89f58bd1 100644
--- a/mm/damon/sysfs.c
+++ b/mm/damon/sysfs.c
@@ -1789,6 +1789,7 @@ struct damon_sysfs_kdamond {
 	struct damon_sysfs_contexts *contexts;
 	struct damon_ctx *damon_ctx;
 	unsigned int refresh_ms;
+	unsigned long damon_sysfs_next_update_jiffies;
 };
 
 static struct damon_sysfs_kdamond *damon_sysfs_kdamond_alloc(void)
@@ -2215,18 +2216,16 @@ static struct damon_ctx *damon_sysfs_build_ctx(
 	return ctx;
 }
 
-/* Tracks when the next periodic re-commit via repeat_call_fn() is due. */
-static unsigned long damon_sysfs_next_update_jiffies;
-
 static int damon_sysfs_repeat_call_fn(void *data)
 {
 	struct damon_sysfs_kdamond *sysfs_kdamond = data;
 
 	if (!sysfs_kdamond->refresh_ms)
 		return 0;
-	if (time_before(jiffies, damon_sysfs_next_update_jiffies))
+	if (time_before(jiffies,
+			sysfs_kdamond->damon_sysfs_next_update_jiffies))
 		return 0;
-	damon_sysfs_next_update_jiffies = jiffies +
+	sysfs_kdamond->damon_sysfs_next_update_jiffies = jiffies +
 		msecs_to_jiffies(sysfs_kdamond->refresh_ms);
 
 	if (!mutex_trylock(&damon_sysfs_lock))
@@ -2274,7 +2273,7 @@ static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond)
 	}
 	kdamond->damon_ctx = ctx;
 
-	damon_sysfs_next_update_jiffies =
+	kdamond->damon_sysfs_next_update_jiffies =
 		jiffies + msecs_to_jiffies(kdamond->refresh_ms);
 
 	repeat_call_control->fn = damon_sysfs_repeat_call_fn;
-- 
2.43.0



^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] mm/damon/sysfs: make next_update_jiffies per-kdamond
  2026-10-17  6:31 [PATCH] mm/damon/sysfs: make next_update_jiffies per-kdamond Karthikeyan KS
@ 2026-10-17 16:01 ` SJ Park
  0 siblings, 0 replies; 2+ messages in thread
From: SJ Park @ 2026-10-17 16:01 UTC (permalink / raw)
  To: Karthikeyan KS; +Cc: SJ Park, akpm, damon, linux-mm, linux-kernel

Hi Karthikeyan,

On Sat, 17 Oct 2026 06:31:52 +0000 Karthikeyan KS <karthiproffesional@gmail.com> wrote:

> damon_sysfs_next_update_jiffies was a single static variable shared by
> all kdamonds, so with two or more kdamonds using refresh_ms their
> threads race on it in damon_sysfs_repeat_call_fn().

Thank you for this patch.  However, I think the same issue is already fixed by
another patch [1].  Could you please check?

[1] https://lore.kernel.org/20260902002725.108635-5-sj@kernel.org


Thanks,
SJ

[...]


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-10-17 16:01 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-10-17  6:31 [PATCH] mm/damon/sysfs: make next_update_jiffies per-kdamond Karthikeyan KS
2026-10-17 16:01 ` SJ Park

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox