ocfs2: free replay slots in ocfs2_recovery_exit()
Joseph Qi
maintainerPATCH 1/1 ocfs2: free replay slots in ocfs2_recovery_exit()
Commit ce2fcf1516d6 ("ocfs2: fix memory leak in ocfs2_mount_volume()") added ocfs2_free_replay_slots() calls to the mount error paths out_dismount and out_check_volume to fix a leak of osb->replay_map. However these calls are unlocked while the bail path of the recovery thread, which is woken up by out_dismount right before the call, also frees the replay slots under osb->recovery_lock. Both sides can thus observe a non-NULL osb->replay_map and trigger a double free.
Fix this by moving ocfs2_free_replay_slots() into ocfs2_recovery_exit() after ocfs2_recovery_disable(), which waits for a running recovery thread to exit under osb->recovery_lock, and drop the unlocked call sites. Both ocfs2_dismount_volume() and the out_super path of ocfs2_fill_super() call ocfs2_recovery_exit(), so the replay slots are freed on every path. Since super.c no longer references it, make ocfs2_free_replay_slots() static again.
--- fs/ocfs2/journal.c | 3 ++- fs/ocfs2/journal.h | 1 - fs/ocfs2/super.c | 5 +---- 3 files changed, 3 insertions(+), 6 deletions(-)
fs/ocfs2/journal.c +2 -1
@@ -156,7 +156,7 @@ static void ocfs2_queue_replay_slots(struct ocfs2_super *osb, replay_map->rm_state =REPLAY_DONE ; } -void ocfs2_free_replay_slots (struct ocfs2_super *osb) +static void ocfs2_free_replay_slots (struct ocfs2_super *osb) {struct ocfs2_replay_map *replay_map = osb->replay_map;
@@ -243,6 +243,7 @@ void ocfs2_recovery_exit(struct ocfs2_super *osb)/* XXX: Should we bug if there are dirty entries? */ kfree (rm); +ocfs2_free_replay_slots (osb); } static int __ocfs2_recovery_map_test(struct ocfs2_super *osb,
fs/ocfs2/journal.h +0 -1
@@ -151,7 +151,6 @@ void ocfs2_recovery_exit(struct ocfs2_super *osb);void ocfs2_recovery_disable_quota (struct ocfs2_super *osb);int ocfs2_compute_replay_slots (struct ocfs2_super *osb); -void ocfs2_free_replay_slots (struct ocfs2_super *osb);/* * Journal Control: * Initialize, Load, Shutdown, Wipe a journal.
fs/ocfs2/super.c +1 -4
@@ -1162,7 +1162,6 @@ static int ocfs2_fill_super(struct super_block *sb, struct fs_context *fc) out_dismount:atomic_set (&osb->vol_state,VOLUME_DISABLED );wake_up (&osb->osb_mount_event); -ocfs2_free_replay_slots (osb);ocfs2_dismount_volume (sb,1 ); goto out;
@@ -1776,14 +1775,12 @@ static int ocfs2_mount_volume(struct super_block *sb) status =ocfs2_truncate_log_init (osb);if (status <0 ) {mlog_errno (status); - goto out_check_volume; + goto out_system_inodes; }ocfs2_super_unlock (osb,1 );return 0 ; -out_check_volume: -ocfs2_free_replay_slots (osb); out_system_inodes:if (osb->local_alloc_state ==OCFS2_LA_ENABLED )ocfs2_shutdown_local_alloc (osb);
Signature · 1 line
2.39.3
Joseph Qi
maintainerQuoted 8 lines · Joseph Qi
Commit
ce2fcf1516d6("ocfs2: fix memory leak inocfs2_mount_volume()") addedocfs2_free_replay_slots()calls to the mount error pathsout_dismountandout_check_volumeto fix a leak ofosb->replay_map. However these calls are unlocked while the bail path of the recovery thread, which is woken up byout_dismountright before the call, also frees the replay slots underosb->recovery_lock. Both sides can thus observe a non-NULLosb->replay_mapand trigger a double free.
Quoted 8 lines
Fix this by moving
ocfs2_free_replay_slots()intoocfs2_recovery_exit()afterocfs2_recovery_disable(), which waits for a running recovery thread to exit underosb->recovery_lock, and drop the unlocked call sites. Bothocfs2_dismount_volume()and theout_superpath ofocfs2_fill_super()callocfs2_recovery_exit(), so the replay slots are freed on every path. Sincesuper.cno longer references it, makeocfs2_free_replay_slots()static again.
Sashiko has review comments:
https://sashiko.dev/#/patchset/[email protected]?part=1
This is because it is base on patch:
"ocfs2: exit recovery thread on mount error path"
I'll resend them as a series.
Thanks,
Joseph
Quoted 70 lines
Fixes ce2fcf1516d6 ("ocfs2: fix memory leak in ocfs2_mount_volume()")Cc <[email protected]>Signed-off-by Joseph Qi <[email protected]>--- fs/ocfs2/journal.c | 3 ++- fs/ocfs2/journal.h | 1 - fs/ocfs2/super.c | 5 +---- 3 files changed, 3 insertions(+), 6 deletions(-)fs/ocfs2/journal.c +2
-1@@ -156,7 +156,7 @@ static void ocfs2_queue_replay_slots(struct ocfs2_super *osb, replay_map->rm_state =REPLAY_DONE ; } -void ocfs2_free_replay_slots (struct ocfs2_super *osb) +static void ocfs2_free_replay_slots (struct ocfs2_super *osb) {struct ocfs2_replay_map *replay_map = osb->replay_map;@@ -243,6 +243,7 @@ void ocfs2_recovery_exit(struct ocfs2_super *osb)/* XXX: Should we bug if there are dirty entries? */ kfree (rm); +ocfs2_free_replay_slots (osb); } static int __ocfs2_recovery_map_test(struct ocfs2_super *osb,fs/ocfs2/journal.h +0
-1@@ -151,7 +151,6 @@ void ocfs2_recovery_exit(struct ocfs2_super *osb);void ocfs2_recovery_disable_quota (struct ocfs2_super *osb);int ocfs2_compute_replay_slots (struct ocfs2_super *osb); -void ocfs2_free_replay_slots (struct ocfs2_super *osb);/* * Journal Control: * Initialize, Load, Shutdown, Wipe a journal. fs/ocfs2/super.c +1
-4@@ -1162,7 +1162,6 @@ static int ocfs2_fill_super(struct super_block *sb, struct fs_context *fc) out_dismount:atomic_set (&osb->vol_state,VOLUME_DISABLED );wake_up (&osb->osb_mount_event); -ocfs2_free_replay_slots (osb);ocfs2_dismount_volume (sb,1 ); goto out;@@ -1776,14 +1775,12 @@ static int ocfs2_mount_volume(struct super_block *sb) status =ocfs2_truncate_log_init (osb);if (status <0 ) {mlog_errno (status); - goto out_check_volume; + goto out_system_inodes; }ocfs2_super_unlock (osb,1 );return 0 ; -out_check_volume: -ocfs2_free_replay_slots (osb); out_system_inodes:if (osb->local_alloc_state ==OCFS2_LA_ENABLED )ocfs2_shutdown_local_alloc (osb);