[PATCH AUTOSEL 6.1 27/49] md: fix soft lockup in status_resync

From: Sasha Levin
Date: Thu May 04 2023 - 15:54:27 EST


From: Yu Kuai <yukuai3@xxxxxxxxxx>

[ Upstream commit 6efddf1e32e2a264694766ca485a4f5e04ee82a7 ]

status_resync() will calculate 'curr_resync - recovery_active' to show
user a progress bar like following:

[============>........] resync = 61.4%

'curr_resync' and 'recovery_active' is updated in md_do_sync(), and
status_resync() can read them concurrently, hence it's possible that
'curr_resync - recovery_active' can overflow to a huge number. In this
case status_resync() will be stuck in the loop to print a large amount
of '=', which will end up soft lockup.

Fix the problem by setting 'resync' to MD_RESYNC_ACTIVE in this case,
this way resync in progress will be reported to user.

Signed-off-by: Yu Kuai <yukuai3@xxxxxxxxxx>
Signed-off-by: Song Liu <song@xxxxxxxxxx>
Link: https://lore.kernel.org/r/20230310073855.1337560-3-yukuai1@xxxxxxxxxxxxxxx
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
drivers/md/md.c | 18 +++++++++---------
1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/md/md.c b/drivers/md/md.c
index d5c362b1602b6..bb73a541bb193 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -8028,16 +8028,16 @@ static int status_resync(struct seq_file *seq, struct mddev *mddev)
} else if (resync > max_sectors) {
resync = max_sectors;
} else {
- resync -= atomic_read(&mddev->recovery_active);
- if (resync < MD_RESYNC_ACTIVE) {
- /*
- * Resync has started, but the subtraction has
- * yielded one of the special values. Force it
- * to active to ensure the status reports an
- * active resync.
- */
+ res = atomic_read(&mddev->recovery_active);
+ /*
+ * Resync has started, but the subtraction has overflowed or
+ * yielded one of the special values. Force it to active to
+ * ensure the status reports an active resync.
+ */
+ if (resync < res || resync - res < MD_RESYNC_ACTIVE)
resync = MD_RESYNC_ACTIVE;
- }
+ else
+ resync -= res;
}

if (resync == MD_RESYNC_NONE) {
--
2.39.2