diff options
| author | Youngjun Park <youngjun.park@lge.com> | 2026-07-13 13:50:14 +0900 |
|---|---|---|
| committer | Andrew Morton <akpm@linux-foundation.org> | 2026-08-06 18:56:56 -0700 |
| commit | 4f93f0956f446e270222c58f821de7500a163229 (patch) | |
| tree | 6a9c0a3d718b46335f9ee4f604a267fb8a5cfcaf | |
| parent | 20dab9d72acdd2288d8a966a8e318274b1e26254 (diff) | |
mm/swap: clarify unreachable cond_resched() in swap_reclaim_full_clusters
Commit 66366d291f66 ("mm/swap: add cond_resched() in
swap_reclaim_full_clusters to prevent softlockup") added cond_resched() to
prevent soft lockups during heavy swap reclaim.
Currently, there are two call sites for this function:
1. swap_reclaim_work(): called with force=true in a workqueue context.
2. cluster_alloc_swap_entry(): called with force=false, holding a
local_lock and potentially a global cluster spinlock (atomic context).
In the second case, calling cond_resched() would normally cause a
"scheduling while atomic" bug. However, it is currently safe because when
force=false, 'to_scan' is initialized to 1. The loop decrements it to 0
and breaks before ever reaching cond_resched().
This implicit dependency is hard to notice and recently triggered a false
positive in AI Sashiko review.
Add a comment to explicitly clarify that cond_resched() is unreachable in
atomic contexts. This improves readability and prevents future misuse if
the loop logic or 'to_scan' initialization is modified.
Link: https://sashiko.dev/#/patchset/20260713025644.170839-1-youngjun.park@lge.com?part=4
Link: https://lore.kernel.org/20260713045014.219653-1-youngjun.park@lge.com
Signed-off-by: Youngjun Park <youngjun.park@lge.com>
Cc: Baoquan He <baoquan.he@linux.dev>
Cc: Barry Song <baohua@kernel.org>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <kasong@tencent.com>
Cc: Kemeng Shi <shikemeng@huaweicloud.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
| -rw-r--r-- | mm/swapfile.c | 6 |
1 files changed, 6 insertions, 0 deletions
diff --git a/mm/swapfile.c b/mm/swapfile.c index 775c385490a5..095d43838d4e 100644 --- a/mm/swapfile.c +++ b/mm/swapfile.c @@ -1072,6 +1072,12 @@ static void swap_reclaim_full_clusters(struct swap_info_struct *si, bool force) swap_cluster_unlock(ci); if (to_scan <= 0) break; + + /* + * When 'force' is false, 'to_scan' is initialized to 1. + * The loop breaks above, making this cond_resched() unreachable + * in atomic contexts. + */ cond_resched(); } } |
