| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <20231228125553.2697765-4-yukuai1@huaweicloud.com>
Date: Thu, 28 Dec 2023 20:55:53 +0800
From: Yu Kuai <yukuai1@...weicloud.com>
To: song@...nel.org,
yukuai3@...wei.com,
neilb@...e.de
Cc: linux-raid@...r.kernel.org,
linux-kernel@...r.kernel.org,
yukuai1@...weicloud.com,
yi.zhang@...wei.com,
yangerkun@...wei.com
Subject: [PATCH -next 3/3] md: use interruptible apis in idle/frozen_sync_thread()
From: Yu Kuai <yukuai3@...wei.com>
Before refactoring idle and frozen from action_store, interruptible apis
is used so that hungtask warning won't be triggered if it takes too long
to finish idle/frozen sync_thread. So change to use interruptible apis.
In order not to make stop_sync_thread() more complicated, factor out a
helper prepare_to_stop_sync_thread() to replace stop_sync_thread().
Also return error to user if idle/frozen_sync_thread() failed, otherwise
user will be misleaded.
Fixes: 130443d60b1b ("md: refactor idle/frozen_sync_thread() to fix deadlock")
Fixes: 8e8e2518fcec ("md: Close race when setting 'action' to 'idle'.")
Signed-off-by: Yu Kuai <yukuai3@...wei.com>
---
drivers/md/md.c | 105 ++++++++++++++++++++++++++++++------------------
1 file changed, 67 insertions(+), 38 deletions(-)
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 60f99768a1a9..9ea05de79fe4 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -4846,26 +4846,34 @@ action_show(struct mddev *mddev, char *page)
return sprintf(page, "%s\n", type);
}
+static bool sync_thread_stopped(struct mddev *mddev, int *sync_seq)
+{
+ if (!test_bit(MD_RECOVERY_RUNNING, &mddev->recovery))
+ return true;
+
+ if (sync_seq && *sync_seq != atomic_read(&mddev->sync_seq))
+ return true;
+
+ return false;
+}
+
/**
- * stop_sync_thread() - wait for sync_thread to stop if it's running.
+ * prepare_to_stop_sync_thread() - prepare to stop sync_thread if it's running.
* @mddev: the array.
- * @locked: if set, reconfig_mutex will still be held after this function
- * return; if not set, reconfig_mutex will be released after this
- * function return.
- * @check_seq: if set, only wait for curent running sync_thread to stop, noted
- * that new sync_thread can still start.
+ * @unlock: whether or not caller want to release reconfig_mutex if
+ * sync_thread is not running.
+ *
+ * Return true if sync_thread is running, release reconfig_mutex and do
+ * preparatory work to stop sync_thread, caller should wait for
+ * sync_thread_stopped() to return true. Return false if sync_thread is not
+ * running, reconfig_mutex will be released if @unlock is set.
*/
-static void stop_sync_thread(struct mddev *mddev, bool locked, bool check_seq)
+static bool prepare_to_stop_sync_thread(struct mddev *mddev, bool unlock)
{
- int sync_seq;
-
- if (check_seq)
- sync_seq = atomic_read(&mddev->sync_seq);
-
if (!test_bit(MD_RECOVERY_RUNNING, &mddev->recovery)) {
- if (!locked)
+ if (unlock)
mddev_unlock(mddev);
- return;
+ return false;
}
mddev_unlock(mddev);
@@ -4879,53 +4887,67 @@ static void stop_sync_thread(struct mddev *mddev, bool locked, bool check_seq)
if (work_pending(&mddev->sync_work))
flush_work(&mddev->sync_work);
- wait_event(resync_wait,
- !test_bit(MD_RECOVERY_RUNNING, &mddev->recovery) ||
- (check_seq && sync_seq != atomic_read(&mddev->sync_seq)));
-
- if (locked)
- mddev_lock_nointr(mddev);
+ return true;
}
-static void idle_sync_thread(struct mddev *mddev)
+static int idle_sync_thread(struct mddev *mddev)
{
- mutex_lock(&mddev->sync_mutex);
+ int sync_seq = atomic_read(&mddev->sync_seq);
+ int err = mutex_lock_interruptible(&mddev->sync_mutex);
+
+ if (err)
+ return err;
clear_bit(MD_RECOVERY_FROZEN, &mddev->recovery);
- if (mddev_lock(mddev)) {
+ err = mddev_lock(mddev);
+ if (err) {
mutex_unlock(&mddev->sync_mutex);
- return;
+ return err;
}
- stop_sync_thread(mddev, false, true);
+ if (prepare_to_stop_sync_thread(mddev, true))
+ err = wait_event_interruptible(resync_wait,
+ sync_thread_stopped(mddev, &sync_seq));
+
mutex_unlock(&mddev->sync_mutex);
+
+ return err;
}
-static void frozen_sync_thread(struct mddev *mddev)
+static int frozen_sync_thread(struct mddev *mddev)
{
- mutex_lock(&mddev->sync_mutex);
+ int err = mutex_lock_interruptible(&mddev->sync_mutex);
+
+ if (err)
+ return err;
set_bit(MD_RECOVERY_FROZEN, &mddev->recovery);
- if (mddev_lock(mddev)) {
+ err = mddev_lock(mddev);
+ if (err) {
mutex_unlock(&mddev->sync_mutex);
- return;
+ return err;
}
- stop_sync_thread(mddev, false, false);
+ if (prepare_to_stop_sync_thread(mddev, true))
+ err = wait_event_interruptible(resync_wait,
+ sync_thread_stopped(mddev, NULL));
mutex_unlock(&mddev->sync_mutex);
+
+ return err;
}
static ssize_t
action_store(struct mddev *mddev, const char *page, size_t len)
{
+ int err = 0;
+
if (!mddev->pers || !mddev->pers->sync_request)
return -EINVAL;
-
if (cmd_match(page, "idle"))
- idle_sync_thread(mddev);
+ err = idle_sync_thread(mddev);
else if (cmd_match(page, "frozen"))
- frozen_sync_thread(mddev);
+ err = frozen_sync_thread(mddev);
else if (test_bit(MD_RECOVERY_RUNNING, &mddev->recovery))
return -EBUSY;
else if (cmd_match(page, "resync"))
@@ -4934,7 +4956,6 @@ action_store(struct mddev *mddev, const char *page, size_t len)
clear_bit(MD_RECOVERY_FROZEN, &mddev->recovery);
set_bit(MD_RECOVERY_RECOVER, &mddev->recovery);
} else if (cmd_match(page, "reshape")) {
- int err;
if (mddev->pers->start_reshape == NULL)
return -EINVAL;
err = mddev_lock(mddev);
@@ -4980,7 +5001,7 @@ action_store(struct mddev *mddev, const char *page, size_t len)
set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
md_wakeup_thread(mddev->thread);
sysfs_notify_dirent_safe(mddev->sysfs_action);
- return len;
+ return err ? err : len;
}
static struct md_sysfs_entry md_scan_mode =
@@ -6280,7 +6301,11 @@ static void md_clean(struct mddev *mddev)
static void __md_stop_writes(struct mddev *mddev)
{
- stop_sync_thread(mddev, true, false);
+ if (prepare_to_stop_sync_thread(mddev, false)) {
+ wait_event(resync_wait, sync_thread_stopped(mddev, NULL));
+ mddev_lock_nointr(mddev);
+ }
+
del_timer_sync(&mddev->safemode_timer);
if (mddev->pers && mddev->pers->quiesce) {
@@ -6369,7 +6394,8 @@ static int md_set_readonly(struct mddev *mddev, struct block_device *bdev)
set_bit(MD_RECOVERY_FROZEN, &mddev->recovery);
}
- stop_sync_thread(mddev, false, false);
+ if (prepare_to_stop_sync_thread(mddev, true))
+ wait_event(resync_wait, sync_thread_stopped(mddev, NULL));
wait_event(mddev->sb_wait,
!test_bit(MD_SB_CHANGE_PENDING, &mddev->sb_flags));
mddev_lock_nointr(mddev);
@@ -6421,7 +6447,10 @@ static int do_md_stop(struct mddev *mddev, int mode,
set_bit(MD_RECOVERY_FROZEN, &mddev->recovery);
}
- stop_sync_thread(mddev, true, false);
+ if (prepare_to_stop_sync_thread(mddev, false)) {
+ wait_event(resync_wait, sync_thread_stopped(mddev, NULL));
+ mddev_lock_nointr(mddev);
+ }
mutex_lock(&mddev->open_mutex);
if ((mddev->pers && atomic_read(&mddev->openers) > !!bdev) ||
--
2.39.2
Powered by blists - more mailing lists