md: Avoid waking up a thread after it has been freed.
Two related problems:
1/ some error paths call "md_unregister_thread(mddev->thread)"
without subsequently clearing ->thread. A subsequent call
to mddev_unlock will try to wake the thread, and crash.
2/ Most calls to md_wakeup_thread are protected against the thread
disappeared either by:
- holding the ->mutex
- having an active request, so something else must be keeping
the array active.
However mddev_unlock calls md_wakeup_thread after dropping the
mutex and without any certainty of an active request, so the
->thread could theoretically disappear.
So we need a spinlock to provide some protections.
So change md_unregister_thread to take a pointer to the thread
pointer, and ensure that it always does the required locking, and
clears the pointer properly.
Reported-by: "Moshe Melnikov" <moshe@zadarastorage.com>
Signed-off-by: NeilBrown <neilb@suse.de>
cc: stable@kernel.org
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 5404b22..5c95ccb 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -61,6 +61,11 @@
static void autostart_arrays(int part);
#endif
+/* pers_list is a list of registered personalities protected
+ * by pers_lock.
+ * pers_lock does extra service to protect accesses to
+ * mddev->thread when the mutex cannot be held.
+ */
static LIST_HEAD(pers_list);
static DEFINE_SPINLOCK(pers_lock);
@@ -739,7 +744,12 @@
} else
mutex_unlock(&mddev->reconfig_mutex);
+ /* was we've dropped the mutex we need a spinlock to
+ * make sur the thread doesn't disappear
+ */
+ spin_lock(&pers_lock);
md_wakeup_thread(mddev->thread);
+ spin_unlock(&pers_lock);
}
static mdk_rdev_t * find_rdev_nr(mddev_t *mddev, int nr)
@@ -6429,11 +6439,18 @@
return thread;
}
-void md_unregister_thread(mdk_thread_t *thread)
+void md_unregister_thread(mdk_thread_t **threadp)
{
+ mdk_thread_t *thread = *threadp;
if (!thread)
return;
dprintk("interrupting MD-thread pid %d\n", task_pid_nr(thread->tsk));
+ /* Locking ensures that mddev_unlock does not wake_up a
+ * non-existent thread
+ */
+ spin_lock(&pers_lock);
+ *threadp = NULL;
+ spin_unlock(&pers_lock);
kthread_stop(thread->tsk);
kfree(thread);
@@ -7340,8 +7357,7 @@
mdk_rdev_t *rdev;
/* resync has finished, collect result */
- md_unregister_thread(mddev->sync_thread);
- mddev->sync_thread = NULL;
+ md_unregister_thread(&mddev->sync_thread);
if (!test_bit(MD_RECOVERY_INTR, &mddev->recovery) &&
!test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) {
/* success...*/
diff --git a/drivers/md/md.h b/drivers/md/md.h
index 1e586bb..0a309dc 100644
--- a/drivers/md/md.h
+++ b/drivers/md/md.h
@@ -560,7 +560,7 @@
extern int unregister_md_personality(struct mdk_personality *p);
extern mdk_thread_t * md_register_thread(void (*run) (mddev_t *mddev),
mddev_t *mddev, const char *name);
-extern void md_unregister_thread(mdk_thread_t *thread);
+extern void md_unregister_thread(mdk_thread_t **threadp);
extern void md_wakeup_thread(mdk_thread_t *thread);
extern void md_check_recovery(mddev_t *mddev);
extern void md_write_start(mddev_t *mddev, struct bio *bi);
diff --git a/drivers/md/multipath.c b/drivers/md/multipath.c
index 3535c23..d5b5fb3 100644
--- a/drivers/md/multipath.c
+++ b/drivers/md/multipath.c
@@ -514,8 +514,7 @@
{
multipath_conf_t *conf = mddev->private;
- md_unregister_thread(mddev->thread);
- mddev->thread = NULL;
+ md_unregister_thread(&mddev->thread);
blk_sync_queue(mddev->queue); /* the unplug fn references 'conf'*/
mempool_destroy(conf->pool);
kfree(conf->multipaths);
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index f4622dd..d9587df 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -2562,8 +2562,7 @@
raise_barrier(conf);
lower_barrier(conf);
- md_unregister_thread(mddev->thread);
- mddev->thread = NULL;
+ md_unregister_thread(&mddev->thread);
if (conf->r1bio_pool)
mempool_destroy(conf->r1bio_pool);
kfree(conf->mirrors);
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index d7a8468..0cd9672 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -2955,7 +2955,7 @@
return 0;
out_free_conf:
- md_unregister_thread(mddev->thread);
+ md_unregister_thread(&mddev->thread);
if (conf->r10bio_pool)
mempool_destroy(conf->r10bio_pool);
safe_put_page(conf->tmppage);
@@ -2973,8 +2973,7 @@
raise_barrier(conf, 0);
lower_barrier(conf);
- md_unregister_thread(mddev->thread);
- mddev->thread = NULL;
+ md_unregister_thread(&mddev->thread);
blk_sync_queue(mddev->queue); /* the unplug fn references 'conf'*/
if (conf->r10bio_pool)
mempool_destroy(conf->r10bio_pool);
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 43709fa..ac5e8b5 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -4941,8 +4941,7 @@
return 0;
abort:
- md_unregister_thread(mddev->thread);
- mddev->thread = NULL;
+ md_unregister_thread(&mddev->thread);
if (conf) {
print_raid5_conf(conf);
free_conf(conf);
@@ -4956,8 +4955,7 @@
{
raid5_conf_t *conf = mddev->private;
- md_unregister_thread(mddev->thread);
- mddev->thread = NULL;
+ md_unregister_thread(&mddev->thread);
if (mddev->queue)
mddev->queue->backing_dev_info.congested_fn = NULL;
free_conf(conf);