fix tag freeing use after free and debugfs name reuse
Subsystems (3)
Hi all,
the first patch fixes a use after free, and the others deal with debugfs name reuse that spews warnings and makes debugfs use impossible for quickly reused gendisk instances. Both of those are rooted in sloppy life time rules for block device tear down.
Compared to the previous separate postings this adds a missing queue quiesce and documents debugfs_mutex better.
Diffstat:
block/blk-core.c | 13 ------------- block/blk-mq-debugfs.c | 29 ++++++++++++++++++----------- block/blk-mq-debugfs.h | 10 ---------- block/blk-mq-sched.c | 11 +++++++++++ block/blk-rq-qos.c | 2 -- block/blk-rq-qos.h | 7 ++++++- block/blk-sysfs.c | 30 ++++++++++++++---------------- block/genhd.c | 42 ++++++++++++------------------------------ include/linux/blkdev.h | 8 ++++---- kernel/trace/blktrace.c | 3 --- 10 files changed, 65 insertions(+), 90 deletions(-)
PATCH 1/4 block: disable the elevator int del_gendisk
The elevator is only used for file system requests, which are stopped in del_gendisk. Move disabling the elevator and freeing the scheduler tags to the end of del_gendisk instead of doing that work in disk_release and blk_cleanup_queue to avoid a use after free on q->tag_set from disk_release as the tag_set might not be alive at that point.
Move the blk_qos_exit call as well, as it just depends on the elevator exit and would be the only reason to keep the not exactly cheap queue freeze in disk_release.
--- block/blk-core.c | 13 ------------- block/genhd.c | 39 +++++++++++---------------------------- 2 files changed, 11 insertions(+), 41 deletions(-)
block/blk-core.c +0 -13
@@ -322,19 +322,6 @@ void blk_cleanup_queue(struct request_queue *q)blk_mq_exit_queue (q); } -/* -* In theory, request pool of sched_tags belongs to request queue. -* However, the current implementation requires tag_set for freeing -* requests, so free the pool now. -* -* Queue has become frozen, there can't be any in-queue requests, so -* it is safe to free requests now. -*/ -mutex_lock (&q->sysfs_lock); -if (q->elevator) -blk_mq_sched_free_rqs (q); -mutex_unlock (&q->sysfs_lock); -/* @q is and will stay empty, shutdown and put */ blk_put_queue (q); }
block/genhd.c +11 -28
@@ -652,6 +652,17 @@ void del_gendisk(struct gendisk *disk)blk_sync_queue (q);blk_flush_integrity (); +blk_mq_cancel_work_sync (q); + +blk_mq_quiesce_queue (q); +if (q->elevator) { +mutex_lock (&q->sysfs_lock); +elevator_exit (q); +mutex_unlock (&q->sysfs_lock); + } +rq_qos_exit (q); +blk_mq_unquiesce_queue (q); +/* * Allow using passthrough request again after the queue is torn down. */
@@ -1120,31 +1131,6 @@ static const struct attribute_group *disk_attr_groups[] = { NULL }; -static void disk_release_mq (struct request_queue *q) -{ -blk_mq_cancel_work_sync (q); - -/* -* There can't be any non non-passthrough bios in flight here, but -* requests stay around longer, including passthrough ones so we -* still need to freeze the queue here. -*/ -blk_mq_freeze_queue (q); - -/* -* Since the I/O scheduler exit code may access cgroup information, -* perform I/O scheduler exit before disassociating from the block -* cgroup controller. -*/ -if (q->elevator) { -mutex_lock (&q->sysfs_lock); -elevator_exit (q); -mutex_unlock (&q->sysfs_lock); - } -rq_qos_exit (q); -__blk_mq_unfreeze_queue (q, true); -} -/** * disk_release - releases all allocated resources of the gendisk * @dev: the device representing this disk
@@ -1166,9 +1152,6 @@ static void disk_release(struct device *dev)might_sleep ();WARN_ON_ONCE (disk_live (disk)); -if (queue_is_mq (disk->queue)) -disk_release_mq (disk->queue); -blkcg_exit_queue (disk->queue);disk_release_events (disk);
Signature · 1 line
2.30.2
Re: [PATCH 1/4] block: disable the elevator int del_gendisk
Quoted 48 lines · Christoph Hellwig
The elevator is only used for file system requests, which are stopped in
del_gendisk. Move disabling the elevator and freeing the scheduler tags to the end ofdel_gendiskinstead of doing that work indisk_releaseandblk_cleanup_queueto avoid a use after free onq->tag_setfromdisk_releaseas thetag_setmight not be alive at that point.Move the
blk_qos_exitcall as well, as it just depends on the elevator exit and would be the only reason to keep the not exactly cheap queue freeze indisk_release.Fixes e155b0c238b2 ("blk-mq: Use shared tags for shared sbitmap support")Reported-by [email protected]Signed-off-by Christoph Hellwig <[email protected]>Tested-by [email protected]--- block/blk-core.c | 13 ------------- block/genhd.c | 39 +++++++++++---------------------------- 2 files changed, 11 insertions(+), 41 deletions(-)block/blk-core.c +0
-13@@ -322,19 +322,6 @@ void blk_cleanup_queue(struct request_queue *q)blk_mq_exit_queue (q); } -/* -* In theory, request pool of sched_tags belongs to request queue. -* However, the current implementation requires tag_set for freeing -* requests, so free the pool now. -* -* Queue has become frozen, there can't be any in-queue requests, so -* it is safe to free requests now. -*/ -mutex_lock (&q->sysfs_lock); -if (q->elevator) -blk_mq_sched_free_rqs (q); -mutex_unlock (&q->sysfs_lock); -/* @q is and will stay empty, shutdown and put */ blk_put_queue (q); }block/genhd.c +0
-0
block/genhd.c void del_gendisk(struct gendisk *disk) · line 657
Quoted 6 lines
block/genhd.c +3
-0@@ -652,6 +652,17 @@ void del_gendisk(struct gendisk *disk)blk_sync_queue (q);blk_flush_integrity (); +blk_mq_cancel_work_sync (q); + +blk_mq_quiesce_queue (q);
quiesce queue adds a bit long delay in del_gendisk, not sure if this way may cause regression in big machines with lots of disks.
block/genhd.c void del_gendisk(struct gendisk *disk) · line 664
Quoted 7 lines
block/genhd.c +7
-0+if (q->elevator) { +mutex_lock (&q->sysfs_lock); +elevator_exit (q); +mutex_unlock (&q->sysfs_lock); + } +rq_qos_exit (q); +blk_mq_unquiesce_queue (q);
Also tearing down elevator here has to be carefully, that means any elevator reference has to hold rcu read lock or .q_usage_counter, meantime it has to be checked, otherwise use-after-free may be caused.
Unfortunately, there are some cases which looks not safe, such as, __blk_mq_update_nr_hw_queues() and blk_mq_has_sqsched().
Another example is bfq_insert_request()<-bfq_insert_requests():
static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
bool at_head)
{
...
spin_unlock_irq(&bfqd->lock);
bfq_update_insert_stats(q, bfqq, idle_timer_disabled,
cmd_flags);
}If last 'rq' is done between unlocking bfqd->lock and calling bfq_update_insert_stats, del_gendisk() may tear down elevator, and UAF is caused.
Thanks,
Ming
Re: [PATCH 1/4] block: disable the elevator int del_gendisk
Quoted 8 lines · Ming Lei
block/genhd.c void del_gendisk(struct gendisk *disk) · line 657
Quoted 5 lines
block/genhd.c +3
-0blk_sync_queue (q);blk_flush_integrity (); +blk_mq_cancel_work_sync (q); + +blk_mq_quiesce_queue (q);quiesce queue adds a bit long delay in
del_gendisk, not sure if this way may cause regression in big machines with lots of disks.
It does. But at least we remove a freeze in the queue teardown path. But either way I'd really like to get things correct first before looking into optimizations.
Quoted 12 lines
block/genhd.c void del_gendisk(struct gendisk *disk) · line 664
Quoted 8 lines
block/genhd.c +7
-0+if (q->elevator) { +mutex_lock (&q->sysfs_lock); +elevator_exit (q); +mutex_unlock (&q->sysfs_lock); + } +rq_qos_exit (q); +blk_mq_unquiesce_queue (q);Also tearing down elevator here has to be carefully, that means any elevator reference has to hold rcu read lock or .q_usage_counter, meantime it has to be checked, otherwise use-after-free may be caused.
This is not a new pattern. We have the same locking here as a sysfs-induced change of the elevator to none which also clears q->elevator under a queue that is frozen and quiesced.
But unlike that path we do fail all requests that could have been queued in the schedule before unfreezing here at least.
Re: [PATCH 1/4] block: disable the elevator int del_gendisk
Quoted 10 lines · Christoph Hellwig
Quoted 8 lines · Ming Lei
block/genhd.c void del_gendisk(struct gendisk *disk) · line 657
Quoted 5 lines
block/genhd.c +3
-0blk_sync_queue (q);blk_flush_integrity (); +blk_mq_cancel_work_sync (q); + +blk_mq_quiesce_queue (q);quiesce queue adds a bit long delay in
del_gendisk, not sure if this way may cause regression in big machines with lots of disks.
Quoted 3 lines
It does. But at least we remove a freeze in the queue teardown path. But either way I'd really like to get things correct first before looking into optimizations.
The removed one works at atomic mode and it is super fast.
Quoted 14 lines
Quoted 13 lines
block/genhd.c void del_gendisk(struct gendisk *disk) · line 664
Quoted 9 lines
block/genhd.c +7
-0+if (q->elevator) { +mutex_lock (&q->sysfs_lock); +elevator_exit (q); +mutex_unlock (&q->sysfs_lock); + } +rq_qos_exit (q); +blk_mq_unquiesce_queue (q);Also tearing down elevator here has to be carefully, that means any elevator reference has to hold rcu read lock or .q_usage_counter, meantime it has to be checked, otherwise use-after-free may be caused.
Quoted 3 lines
This is not a new pattern. We have the same locking here as a sysfs-induced change of the elevator to none which also clears
q->elevatorunder a queue that is frozen and quiesced.
Then looks this pattern has problem in dealing with the examples I mentioned.
And the elevator usage in __blk_mq_update_nr_hw_queues() looks one old problem, but easy to fix by protecting it via sysfs_lock.
And fixing blk_mq_has_sqsched() should be easy too.
I will send patches later.
Thanks,
Ming
Jens Axboe
maintainerRe: [PATCH 1/4] block: disable the elevator int del_gendisk
Quoted 38 lines · Ming Lei
Quoted 10 lines · Christoph Hellwig
Quoted 8 lines · Ming Lei
block/genhd.c void del_gendisk(struct gendisk *disk) · line 657
Quoted 5 lines
block/genhd.c +3
-0blk_sync_queue (q);blk_flush_integrity (); +blk_mq_cancel_work_sync (q); + +blk_mq_quiesce_queue (q);quiesce queue adds a bit long delay in
del_gendisk, not sure if this way may cause regression in big machines with lots of disks.Quoted 3 lines
It does. But at least we remove a freeze in the queue teardown path. But either way I'd really like to get things correct first before looking into optimizations.
The removed one works at atomic mode and it is super fast.
Quoted 14 lines
Quoted 13 lines
block/genhd.c void del_gendisk(struct gendisk *disk) · line 664
Quoted 9 lines
block/genhd.c +7
-0+if (q->elevator) { +mutex_lock (&q->sysfs_lock); +elevator_exit (q); +mutex_unlock (&q->sysfs_lock); + } +rq_qos_exit (q); +blk_mq_unquiesce_queue (q);Also tearing down elevator here has to be carefully, that means any elevator reference has to hold rcu read lock or .q_usage_counter, meantime it has to be checked, otherwise use-after-free may be caused.
Quoted 3 lines
This is not a new pattern. We have the same locking here as a sysfs-induced change of the elevator to none which also clears
q->elevatorunder a queue that is frozen and quiesced.Then looks this pattern has problem in dealing with the examples I mentioned.
Quoted 6 lines
And the elevator usage in
__blk_mq_update_nr_hw_queues()looks one old problem, but easy to fix by protecting it viasysfs_lock.And fixing
blk_mq_has_sqsched()should be easy too.I will send patches later.
Just checking in on this series, Ming did you make any progress?
Signature · 1 line
Jens Axboe
Re: [PATCH 1/4] block: disable the elevator int del_gendisk
Quoted 11 lines · Jens Axboe
Quoted 9 lines
Then looks this pattern has problem in dealing with the examples I mentioned.
And the elevator usage in
__blk_mq_update_nr_hw_queues()looks one old problem, but easy to fix by protecting it viasysfs_lock.And fixing
blk_mq_has_sqsched()should be easy too.I will send patches later.
Just checking in on this series, Ming did you make any progress?
He send the patches in the "blk-mq: three misc patches" series and you already applied them.
Jens Axboe
maintainerRe: [PATCH 1/4] block: disable the elevator int del_gendisk
Quoted 7 lines · Christoph Hellwig
Quoted 6 lines · Jens Axboe
Quoted 6 lines
Then looks this pattern has problem in dealing with the examples I mentioned.
And the elevator usage in
__blk_mq_update_nr_hw_queues()looks one old problem, but easy to fix by protecting it viasysfs_lock.
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
And fixing
blk_mq_has_sqsched()should be easy too.I will send patches later.
Just checking in on this series, Ming did you make any progress?
He send the patches in the "blk-mq: three misc patches" series and you already applied them.
Ah ok, guess they are already queued up for 5.19. Hard to keep track with the backlog.
Signature · 1 line
Jens Axboe
PATCH 2/4 block: serialize all debugfs operations using q->debugfs_mutex
In reply to Christoph HellwigVarious places like I/O schedulers or the QOS infrastructure try to register debugfs files on demans, which can race with creating and removing the main queue debugfs directory. Use the existing debugfs_mutex to serialize all debugfs operations that rely on q->debugfs_dir or the directories hanging off it.
To make the teardown code a little simpler declare all debugfs dentry pointers and not just the main one uncoditionally in blkdev.h.
Move debugfs_mutex next to the dentries that it protects and document what it is used for.
--- block/blk-mq-debugfs.c | 25 ++++++++++++++++++++----- block/blk-mq-debugfs.h | 5 ----- block/blk-mq-sched.c | 11 +++++++++++ block/blk-rq-qos.c | 2 ++ block/blk-rq-qos.h | 7 ++++++- block/blk-sysfs.c | 20 +++++++++----------- include/linux/blkdev.h | 8 ++++---- kernel/trace/blktrace.c | 3 --- 8 files changed, 52 insertions(+), 29 deletions(-)
block/blk-mq-debugfs.c +20 -5
@@ -711,11 +711,6 @@ void blk_mq_debugfs_register(struct request_queue *q) } } -void blk_mq_debugfs_unregister (struct request_queue *q) -{ - q->sched_debugfs_dir =NULL ; -} -static void blk_mq_debugfs_register_ctx (struct blk_mq_hw_ctx *hctx,struct blk_mq_ctx *ctx) {
@@ -746,6 +741,8 @@ void blk_mq_debugfs_register_hctx(struct request_queue *q,void blk_mq_debugfs_unregister_hctx (struct blk_mq_hw_ctx *hctx) { +if (!hctx->queue->debugfs_dir) +return ;debugfs_remove_recursive (hctx->debugfs_dir); hctx->sched_debugfs_dir =NULL ; hctx->debugfs_dir =NULL ;
@@ -773,6 +770,8 @@ void blk_mq_debugfs_register_sched(struct request_queue *q) {struct elevator_type *e = q->elevator->type; +lockdep_assert_held (&q->debugfs_mutex); +/* * If the parent directory has not been created yet, return, we will be * called again later on and the directory/files will be created then.
@@ -790,6 +789,8 @@ void blk_mq_debugfs_register_sched(struct request_queue *q)void blk_mq_debugfs_unregister_sched (struct request_queue *q) { +lockdep_assert_held (&q->debugfs_mutex); +debugfs_remove_recursive (q->sched_debugfs_dir); q->sched_debugfs_dir =NULL ; }
@@ -811,6 +812,10 @@ static const char *rq_qos_id_to_name(enum rq_qos_id id)void blk_mq_debugfs_unregister_rqos (struct rq_qos *rqos) { +lockdep_assert_held (&rqos->q->debugfs_mutex); + +if (!rqos->q->debugfs_dir) +return ;debugfs_remove_recursive (rqos->debugfs_dir); rqos->debugfs_dir =NULL ; }
@@ -820,6 +825,8 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)struct request_queue *q = rqos->q;const char *dir_name =rq_qos_id_to_name (rqos->id); +lockdep_assert_held (&q->debugfs_mutex); +if (rqos->debugfs_dir || !rqos->ops->debugfs_attrs)return ;
@@ -835,6 +842,8 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)void blk_mq_debugfs_unregister_queue_rqos (struct request_queue *q) { +lockdep_assert_held (&q->debugfs_mutex); +debugfs_remove_recursive (q->rqos_debugfs_dir); q->rqos_debugfs_dir =NULL ; }
@@ -844,6 +853,8 @@ void blk_mq_debugfs_register_sched_hctx(struct request_queue *q, {struct elevator_type *e = q->elevator->type; +lockdep_assert_held (&q->debugfs_mutex); +/* * If the parent debugfs directory has not been created yet, return; * We will be called again later on with appropriate parent debugfs
@@ -863,6 +874,10 @@ void blk_mq_debugfs_register_sched_hctx(struct request_queue *q,void blk_mq_debugfs_unregister_sched_hctx (struct blk_mq_hw_ctx *hctx) { +lockdep_assert_held (&hctx->queue->debugfs_mutex); + +if (!hctx->queue->debugfs_dir) +return ;debugfs_remove_recursive (hctx->sched_debugfs_dir); hctx->sched_debugfs_dir =NULL ; }
block/blk-mq-debugfs.h +0 -5
@@ -21,7 +21,6 @@ int __blk_mq_debugfs_rq_show(struct seq_file *m, struct request *rq);int blk_mq_debugfs_rq_show (struct seq_file *m,void *v);void blk_mq_debugfs_register (struct request_queue *q); -void blk_mq_debugfs_unregister (struct request_queue *q);void blk_mq_debugfs_register_hctx (struct request_queue *q,struct blk_mq_hw_ctx *hctx);void blk_mq_debugfs_unregister_hctx (struct blk_mq_hw_ctx *hctx);
@@ -42,10 +41,6 @@ static inline void blk_mq_debugfs_register(struct request_queue *q) { } -static inline void blk_mq_debugfs_unregister (struct request_queue *q) -{ -} -static inline void blk_mq_debugfs_register_hctx (struct request_queue *q,struct blk_mq_hw_ctx *hctx) {
block/blk-mq-sched.c +11 -0
@@ -593,7 +593,9 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)if (ret) goto err_free_map_and_rqs; +mutex_lock (&q->debugfs_mutex);blk_mq_debugfs_register_sched (q); +mutex_unlock (&q->debugfs_mutex);queue_for_each_hw_ctx (q, hctx, i) {if (e->ops.init_hctx) {
@@ -606,7 +608,9 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)return ret; } } +mutex_lock (&q->debugfs_mutex);blk_mq_debugfs_register_sched_hctx (q, hctx); +mutex_unlock (&q->debugfs_mutex); }return 0 ;
@@ -647,14 +651,21 @@ void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)unsigned int flags =0 ;queue_for_each_hw_ctx (q, hctx, i) { +mutex_lock (&q->debugfs_mutex);blk_mq_debugfs_unregister_sched_hctx (hctx); +mutex_unlock (&q->debugfs_mutex); +if (e->type->ops.exit_hctx && hctx->sched_data) { e->type->ops.exit_hctx (hctx, i); hctx->sched_data =NULL ; } flags = hctx->flags; } + +mutex_lock (&q->debugfs_mutex);blk_mq_debugfs_unregister_sched (q); +mutex_unlock (&q->debugfs_mutex); +if (e->type->ops.exit_sched) e->type->ops.exit_sched (e);blk_mq_sched_tags_teardown (q, flags);
block/blk-rq-qos.c +2 -0
@@ -294,7 +294,9 @@ void rq_qos_wait(struct rq_wait *rqw, void *private_data,void rq_qos_exit (struct request_queue *q) { +mutex_lock (&q->debugfs_mutex);blk_mq_debugfs_unregister_queue_rqos (q); +mutex_unlock (&q->debugfs_mutex);while (q->rq_qos) {struct rq_qos *rqos = q->rq_qos;
block/blk-rq-qos.h +6 -1
@@ -104,8 +104,11 @@ static inline void rq_qos_add(struct request_queue *q, struct rq_qos *rqos)blk_mq_unfreeze_queue (q); -if (rqos->ops->debugfs_attrs) +if (rqos->ops->debugfs_attrs) { +mutex_lock (&q->debugfs_mutex);blk_mq_debugfs_register_rqos (rqos); +mutex_unlock (&q->debugfs_mutex); + } }static inline void rq_qos_del (struct request_queue *q,struct rq_qos *rqos)
@@ -129,7 +132,9 @@ static inline void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)blk_mq_unfreeze_queue (q); +mutex_lock (&q->debugfs_mutex);blk_mq_debugfs_unregister_rqos (rqos); +mutex_unlock (&q->debugfs_mutex); }typedef bool (acquire_inflight_cb_t )(struct rq_wait *rqw,void *private_data);
block/blk-sysfs.c +9 -11
@@ -779,14 +779,13 @@ static void blk_release_queue(struct kobject *kobj)if (queue_is_mq (q))blk_mq_release (q); -blk_trace_shutdown (q);mutex_lock (&q->debugfs_mutex); +blk_trace_shutdown (q);debugfs_remove_recursive (q->debugfs_dir); + q->debugfs_dir =NULL ; + q->sched_debugfs_dir =NULL ;mutex_unlock (&q->debugfs_mutex); -if (queue_is_mq (q)) -blk_mq_debugfs_unregister (q); -bioset_exit (&q->bio_split);if (blk_queue_has_srcu (q))
@@ -836,17 +835,16 @@ int blk_register_queue(struct gendisk *disk) goto unlock; } +if (queue_is_mq (q)) +__blk_mq_register_dev (dev, q); +mutex_lock (&q->sysfs_lock); +mutex_lock (&q->debugfs_mutex); q->debugfs_dir =debugfs_create_dir (kobject_name (q->kobj.parent), blk_debugfs_root); -mutex_unlock (&q->debugfs_mutex); - -if (queue_is_mq (q)) { -__blk_mq_register_dev (dev, q); +if (queue_is_mq (q))blk_mq_debugfs_register (q); - } - -mutex_lock (&q->sysfs_lock); +mutex_unlock (&q->debugfs_mutex); ret =disk_register_independent_access_ranges (disk,NULL );if (ret)
include/linux/blkdev.h +4 -4
@@ -482,7 +482,6 @@ struct request_queue {#endif /* CONFIG_BLK_DEV_ZONED */ int node; -struct mutex debugfs_mutex;#ifdef CONFIG_BLK_DEV_IO_TRACE struct blk_trace __rcu *blk_trace;#endif
@@ -526,11 +525,12 @@ struct request_queue {struct bio_set bio_split;struct dentry *debugfs_dir; - -#ifdef CONFIG_BLK_DEBUG_FS struct dentry *sched_debugfs_dir;struct dentry *rqos_debugfs_dir; -#endif +/* +* Serializes all debugfs metadata operations using the above dentries. +*/ +struct mutex debugfs_mutex;bool mq_sysfs_init_done;
kernel/trace/blktrace.c +0 -3
@@ -770,14 +770,11 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)**/ void blk_trace_shutdown (struct request_queue *q) { -mutex_lock (&q->debugfs_mutex);if (rcu_dereference_protected (q->blk_trace,lockdep_is_held (&q->debugfs_mutex))) {__blk_trace_startstop (q,0 );__blk_trace_remove (q); } - -mutex_unlock (&q->debugfs_mutex); }#ifdef CONFIG_BLK_CGROUP
Signature · 1 line
2.30.2
PATCH 3/4 block: remove per-disk debugfs files in blk_unregister_queue
In reply to Christoph HellwigThe block debugfs files are created in blk_register_queue, which is called by add_disk and use a naming scheme based on the disk_name. After del_gendisk returns that name can be reused and thus we must not leave these debugfs files around, otherwise the kernel is unhappy and spews messages like:
Directory XXXXX with parent 'block' already present!
and the newly created devices will not have working debugfs files.
Move the unregistration to blk_unregister_queue instead (which matches the sysfs unregistration) to make sure the debugfs life time rules match those of the disk name.
As part of the move also make sure the whole debugfs unregistration is inside a single debugfs_mutex critical section.
Note that this breaks blktests block/002, which checks that the debugfs directory has not been removed while blktests is running, but that particular check should simply be removed from the test case.
--- block/blk-mq-debugfs.c | 8 -------- block/blk-mq-debugfs.h | 5 ----- block/blk-rq-qos.c | 4 ---- block/blk-sysfs.c | 16 ++++++++-------- 4 files changed, 8 insertions(+), 25 deletions(-)
block/blk-mq-debugfs.c +0 -8
@@ -840,14 +840,6 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)debugfs_create_files (rqos->debugfs_dir, rqos, rqos->ops->debugfs_attrs); } -void blk_mq_debugfs_unregister_queue_rqos (struct request_queue *q) -{ -lockdep_assert_held (&q->debugfs_mutex); - -debugfs_remove_recursive (q->rqos_debugfs_dir); - q->rqos_debugfs_dir =NULL ; -} -void blk_mq_debugfs_register_sched_hctx (struct request_queue *q,struct blk_mq_hw_ctx *hctx) {
block/blk-mq-debugfs.h +0 -5
@@ -35,7 +35,6 @@ void blk_mq_debugfs_unregister_sched_hctx(struct blk_mq_hw_ctx *hctx);void blk_mq_debugfs_register_rqos (struct rq_qos *rqos);void blk_mq_debugfs_unregister_rqos (struct rq_qos *rqos); -void blk_mq_debugfs_unregister_queue_rqos (struct request_queue *q);#else static inline void blk_mq_debugfs_register (struct request_queue *q) {
@@ -82,10 +81,6 @@ static inline void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)static inline void blk_mq_debugfs_unregister_rqos (struct rq_qos *rqos) { } - -static inline void blk_mq_debugfs_unregister_queue_rqos (struct request_queue *q) -{ -}#endif #ifdef CONFIG_BLK_DEBUG_FS_ZONED
block/blk-rq-qos.c +0 -4
@@ -294,10 +294,6 @@ void rq_qos_wait(struct rq_wait *rqw, void *private_data,void rq_qos_exit (struct request_queue *q) { -mutex_lock (&q->debugfs_mutex); -blk_mq_debugfs_unregister_queue_rqos (q); -mutex_unlock (&q->debugfs_mutex); -while (q->rq_qos) {struct rq_qos *rqos = q->rq_qos; q->rq_qos = rqos->next;
block/blk-sysfs.c +8 -8
@@ -779,13 +779,6 @@ static void blk_release_queue(struct kobject *kobj)if (queue_is_mq (q))blk_mq_release (q); -mutex_lock (&q->debugfs_mutex); -blk_trace_shutdown (q); -debugfs_remove_recursive (q->debugfs_dir); - q->debugfs_dir =NULL ; - q->sched_debugfs_dir =NULL ; -mutex_unlock (&q->debugfs_mutex); -bioset_exit (&q->bio_split);if (blk_queue_has_srcu (q))
@@ -946,8 +939,15 @@ void blk_unregister_queue(struct gendisk *disk)/* Now that we've deleted all child objects, we can delete the queue. */ kobject_uevent (&q->kobj,KOBJ_REMOVE );kobject_del (&q->kobj); -mutex_unlock (&q->sysfs_dir_lock); +mutex_lock (&q->debugfs_mutex); +blk_trace_shutdown (q); +debugfs_remove_recursive (q->debugfs_dir); + q->debugfs_dir =NULL ; + q->sched_debugfs_dir =NULL ; + q->rqos_debugfs_dir =NULL ; +mutex_unlock (&q->debugfs_mutex); +kobject_put (&disk_to_dev (disk)->kobj); }
Signature · 1 line
2.30.2
PATCH 4/4 block: freeze the queue earlier in del_gendisk
In reply to Christoph HellwigFreeze the queue earlier in del_gendisk so that the state does not change while we remove debugfs and sysfs files.
Ming mentioned that being able to observer request in debugfs might be useful while the queue is being frozen in del_gendisk, which is made possible by this change.
--- block/genhd.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
block/genhd.c +1 -2
@@ -623,6 +623,7 @@ void del_gendisk(struct gendisk *disk)* Prevent new I/O from crossing bio_queue_enter(). */ blk_queue_start_drain (q); +blk_mq_freeze_queue_wait (q);if (!(disk->flags &GENHD_FL_HIDDEN )) {sysfs_remove_link (&disk_to_dev (disk)->kobj,"bdi" );
@@ -646,8 +647,6 @@ void del_gendisk(struct gendisk *disk)pm_runtime_set_memalloc_noio (disk_to_dev (disk), false);device_del (disk_to_dev (disk)); -blk_mq_freeze_queue_wait (q); -blk_throtl_cancel_bios (disk->queue);blk_sync_queue (q);
Signature · 1 line
2.30.2
REGRESSION: [PATCH 4/4] block: freeze the queue earlier in del_gendisk
Hi,
Quoted 8 lines · Christoph Hellwig
Freeze the queue earlier in
del_gendiskso that the state does not change while we remove debugfs and sysfs files.Ming mentioned that being able to observer request in debugfs might be useful while the queue is being frozen in
del_gendisk, which is made possible by this change.Signed-off-by Christoph Hellwig <[email protected]>
I'm not really sure why this is yet, but this patch in rc4 causes some random failures with mdadm tests.
It seems the 11spare-migration tests starts failing roughly every other run because the block device is not quite cleaned up after mdadm --stop by the time the next mdadm --create commands starts, or rather there appears to be a race now between the newly created device and the one being cleaned up. This results in an infrequent sysfs panic with a duplicate filename error (see the end of this email).
I managed to bisect this and found a09b314005f3a09 to be the problematic commit.
Reverting seems to fix it.
Thanks,
Logan
Quoted 17 lines
--- block/genhd.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)block/genhd.c +1
-0@@ -623,6 +623,7 @@ void del_gendisk(struct gendisk *disk)* Prevent new I/O from crossing bio_queue_enter(). */ blk_queue_start_drain (q); +blk_mq_freeze_queue_wait (q);if (!(disk->flags &GENHD_FL_HIDDEN )) {sysfs_remove_link (&disk_to_dev (disk)->kobj,"bdi" );
block/genhd.c void del_gendisk(struct gendisk *disk) · line 652
Quoted 8 lines
block/genhd.c +0
-2@@ -646,8 +647,6 @@ void del_gendisk(struct gendisk *disk)pm_runtime_set_memalloc_noio (disk_to_dev (disk), false);device_del (disk_to_dev (disk)); -blk_mq_freeze_queue_wait (q); -blk_throtl_cancel_bios (disk->queue);blk_sync_queue (q);
[ 1026.373014] sysfs: cannot create duplicate filename '/devices/virtual/block/md124' [ 1026.374616] CPU: 1 PID: 11046 Comm: mdadm Not tainted 5.19.0-rc4-eid-vmlocalyes-dbg-00065-gff4ec5f79108 #2430 [ 1026.376546] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.14.0-2 04/01/2014 [ 1026.378000] Call Trace: [ 1026.378461] <TASK> [ 1026.378867] dump_stack_lvl+0x5a/0x74 [ 1026.379558] dump_stack+0x10/0x12 [ 1026.380178] sysfs_warn_dup.cold+0x17/0x27 [ 1026.380944] sysfs_create_dir_ns+0x17d/0x190 [ 1026.381765] ? sysfs_create_mount_point+0x80/0x80 [ 1026.382638] ? __kasan_check_read+0x11/0x20 [ 1026.383401] ? class_dir_child_ns_type+0x23/0x30 [ 1026.384267] kobject_add_internal+0x145/0x460 [ 1026.385084] kobject_add+0xf3/0x150 [ 1026.385733] ? kset_create_and_add+0xe0/0xe0 [ 1026.386529] ? __kasan_check_read+0x11/0x20 [ 1026.387306] ? mutex_unlock+0x12/0x20 [ 1026.387986] ? device_add+0x1da/0xf20 [ 1026.388676] device_add+0x224/0xf20 [ 1026.389316] ? kobject_set_name_vargs+0x95/0xb0 [ 1026.390163] ? __fw_devlink_link_to_suppliers+0x180/0x180 [ 1026.391157] ? sprintf+0xae/0xe0 [ 1026.391784] device_add_disk+0x1b8/0x5f0 [ 1026.392520] md_alloc+0x4c9/0x800 [ 1026.393131] ? __kasan_check_read+0x11/0x20 [ 1026.393921] md_probe+0x24/0x30 [ 1026.394506] blk_request_module+0x9a/0x100 [ 1026.395268] blkdev_get_no_open+0x66/0xa0 [ 1026.395993] blkdev_get_by_dev.part.0+0x24/0x570 [ 1026.396854] ? devcgroup_check_permission+0xed/0x240 [ 1026.397770] blkdev_get_by_dev+0x51/0x60 [ 1026.398497] blkdev_open+0xa4/0x140 [ 1026.399146] do_dentry_open+0x2a7/0x6e0 [ 1026.399854] ? blkdev_close+0x50/0x50 [ 1026.400546] vfs_open+0x58/0x60 [ 1026.401125] path_openat+0x77e/0x13f0 [ 1026.401830] ? lookup_open.isra.0+0xaf0/0xaf0 [ 1026.402615] ? kvm_sched_clock_read+0x18/0x40 [ 1026.403441] ? sched_autogroup_detach+0x20/0x20 [ 1026.404267] ? __this_cpu_preempt_check+0x13/0x20 [ 1026.405141] do_filp_open+0x154/0x280 [ 1026.405833] ? may_open_dev+0x60/0x60 [ 1026.406558] ? __kasan_check_read+0x11/0x20 [ 1026.407302] ? do_raw_spin_unlock+0x98/0x100 [ 1026.408067] ? alloc_fd+0x183/0x340 [ 1026.408718] do_sys_openat2+0x119/0x2c0 [ 1026.409437] ? kmem_cache_free+0x156/0x690 [ 1026.410167] ? dput+0x29/0x750 [ 1026.410730] ? build_open_flags+0x280/0x280 [ 1026.411607] ? putname+0x7c/0x90 [ 1026.412164] __x64_sys_openat+0xe7/0x160 [ 1026.412919] ? __ia32_compat_sys_open+0x130/0x130 [ 1026.413630] ? syscall_enter_from_user_mode+0x21/0x60 [ 1026.414271] ? lockdep_hardirqs_on+0x82/0x110 [ 1026.414828] ? trace_hardirqs_on+0x3d/0x100 [ 1026.415376] do_syscall_64+0x3b/0x90 [ 1026.415837] entry_SYSCALL_64_after_hwframe+0x46/0xb0 [ 1026.416473] RIP: 0033:0x7fab048f3be7 [ 1026.416938] Code: 25 00 00 41 00 3d 00 00 41 00 74 47 64 8b 04 25 18 00 00 00 85 c0 75 6b 44 89 e2 48 89 ee bf 9c ff ff ff b8 01 01 00 00 0f 05 <48> 3d 00 f0 ff ff 0f 87 95 00 00 00 48 8b 4c 24 28 64 48 2b 0c 25 [ 1026.419219] RSP: 002b:00007ffc3c3d7ae0 EFLAGS: 00000246 ORIG_RAX: 0000000000000101 [ 1026.420162] RAX: ffffffffffffffda RBX: 00000000000003e8 RCX: 00007fab048f3be7 [ 1026.421045] RDX: 0000000000004082 RSI: 00007ffc3c3d7b70 RDI: 00000000ffffff9c [ 1026.421928] RBP: 00007ffc3c3d7b70 R08: 0000000000000000 R09: 00007ffc3c3d79f0 [ 1026.422809] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000004082 [ 1026.423696] R13: 0000000000000009 R14: 00007ffc3c3d7b68 R15: 0000556054ddd970 [ 1026.424608] </TASK> [ 1026.424982] kobject_add_internal failed for md124 with -EEXIST, don't
try to register things with the same name in the same directory.
Re: REGRESSION: [PATCH 4/4] block: freeze the queue earlier in del_gendisk
Quoted 10 lines · Logan Gunthorpe
I'm not really sure why this is yet, but this patch in rc4 causes some random failures with mdadm tests.
It seems the 11spare-migration tests starts failing roughly every other run because the block device is not quite cleaned up after mdadm --stop by the time the next mdadm --create commands starts, or rather there appears to be a race now between the newly created device and the one being cleaned up. This results in an infrequent sysfs panic with a duplicate filename error (see the end of this email).
Quoted 2 lines
I managed to bisect this and found
a09b314005f3a09to be the problematic commit.
Taking a look at the mddev code this commit just seems to increase the race window of hitting horrible life time problems in md, but I'll also try to reproduce and verify it myself.
Take a look at how md searches for a duplicate name in md_alloc, mddev_alloc_unit and mddev_find_locked based on the all_mddevs list, and how the mddev gets dropped from all_mddevs very early and long before the gendisk is gone in mddev_put. I think what needs to be done is to implement a free_disk method and drop the mddev (and free it) from that. But given how much intricate mess is based on all_mddevs we'll have to be very careful about that.
Re: REGRESSION: [PATCH 4/4] block: freeze the queue earlier in del_gendisk
Quoted 18 lines · Christoph Hellwig
Quoted 10 lines · Logan Gunthorpe
I'm not really sure why this is yet, but this patch in rc4 causes some random failures with mdadm tests.
It seems the 11spare-migration tests starts failing roughly every other run because the block device is not quite cleaned up after mdadm --stop by the time the next mdadm --create commands starts, or rather there appears to be a race now between the newly created device and the one being cleaned up. This results in an infrequent sysfs panic with a duplicate filename error (see the end of this email).
Quoted 2 lines
I managed to bisect this and found
a09b314005f3a09to be the problematic commit.Taking a look at the mddev code this commit just seems to increase the race window of hitting horrible life time problems in md, but I'll also try to reproduce and verify it myself.
Quoted 7 lines
Take a look at how md searches for a duplicate name in
md_alloc,mddev_alloc_unitandmddev_find_lockedbased on theall_mddevslist, and how the mddev gets dropped fromall_mddevsvery early and long before the gendisk is gone inmddev_put. I think what needs to be done is to implement afree_diskmethod and drop the mddev (and free it) from that. But given how much intricate mess is based onall_mddevswe'll have to be very careful about that.
I agree it's a mess, probably buggy and could use a cleanup with a free_disk method. But I'm not sure the all_mdevs lifetime issues are the problem here. If the entry in all_mdevs outlasts the disk, then md_alloc() will just fail earlier. Many test scripts rely on the fact that you can stop an mddev and recreate it immediately after. We need some way of ensuring any deleted disks are fully deleted before trying to make a new mddev, in case the new one has the same name as one being deleted.
The md code deletes the disk in md_delayed_delete(), a work queue item on md_misc_wq. That queue is flushed first in md_misc_wq, but somehow, some of the disk is still not fully deleted by the time flush_workqueue() returns. I'm not sure why that would be.
Logan
Re: REGRESSION: [PATCH 4/4] block: freeze the queue earlier in del_gendisk
Quoted 8 lines · Logan Gunthorpe
I agree it's a mess, probably buggy and could use a cleanup with a
free_diskmethod. But I'm not sure theall_mdevslifetime issues are the problem here. If the entry inall_mdevsoutlasts the disk, thenmd_alloc()will just fail earlier. Many test scripts rely on the fact that you can stop an mddev and recreate it immediately after. We need some way of ensuring any deleted disks are fully deleted before trying to make a new mddev, in case the new one has the same name as one being deleted.
I think those tests are broken. But fortunately that is just an assumption in the tests, while device name reuse is a real problem.
I could not reproduce your problem, but on the for-5.20/block tree I see a hang in 10ddf-geometry when running the tests.
The branch here:
git://git.infradead.org/users/hch/block.git md-lifetime-fixes
http://git.infradead.org/users/hch/block.git/shortlog/refs/heads/md-lifetime-fixes
fixes that for me and does not introduce new regressions. Can you check if that helps your problem?
Re: REGRESSION: [PATCH 4/4] block: freeze the queue earlier in del_gendisk
Quoted 10 lines · Christoph Hellwig
Quoted 8 lines · Logan Gunthorpe
I agree it's a mess, probably buggy and could use a cleanup with a
free_diskmethod. But I'm not sure theall_mdevslifetime issues are the problem here. If the entry inall_mdevsoutlasts the disk, thenmd_alloc()will just fail earlier. Many test scripts rely on the fact that you can stop an mddev and recreate it immediately after. We need some way of ensuring any deleted disks are fully deleted before trying to make a new mddev, in case the new one has the same name as one being deleted.
Quoted 5 lines
I think those tests are broken. But fortunately that is just an assumption in the tests, while device name reuse is a real problem.
I could not reproduce your problem, but on the for-5.20/block tree I see a hang in 10ddf-geometry when running the tests.
Ah, strange. I never saw an issue with that test, though I didn't run that one repeatedly with the latest branch. So maybe it was an intermittent like I saw with 11spare-migration and I just missed it.
Quoted 7 lines
The branch here:
git://git.infradead.org/users/hch/block.gitmd-lifetime-fixes
http://git.infradead.org/users/hch/block.git/shortlog/refs/heads/md-lifetime-fixesfixes that for me and does not introduce new regressions. Can you check if that helps your problem?
Yes, I can confirm those patches fix the bug I was seeing too.
I did a fairly quick review of the patches:
- In the patch that introduces md_free_disk() it looks like md_free()
can still be called from the error path of md_alloc() before mddev->gendisk is set... which seems to make things rather complicated seeing we then can't use free_disk to finish the cleanup if the disk hasn't been created yet. I probably need to take closer look at this but, it might make more sense for the cleanup to remain in md_free() but call kobject_put() in md_free_disk() and del_gendisk() in mdev_delayed_delete(). Then md_alloc() can still use kobject_put() in the error path and it makes a little more sense seeing we'd still be freeing the kobject stuff in it's own release method instead of the disks free method.
- In the patch with md_rdevs_overlap, it looks like we remove thercu_read_lock(), which definitely seems out of place and probably isn't correct. But the comment that was recreated still references the rcu so probably should be changed.
- The last patch has a typo in the title (disk is *a* freed).
Thanks!
Logan
Re: REGRESSION: [PATCH 4/4] block: freeze the queue earlier in del_gendisk
Quoted 13 lines · Logan Gunthorpe
I did a fairly quick review of the patches:
- In the patch that introduces
md_free_disk()it looks likemd_free()
can still be called from the error path ofmd_alloc()beforemddev->gendiskis set... which seems to make things rather complicated seeing we then can't usefree_diskto finish the cleanup if the disk hasn't been created yet. I probably need to take closer look at this but, it might make more sense for the cleanup to remain inmd_free()but callkobject_put()inmd_free_disk()anddel_gendisk()inmdev_delayed_delete(). Thenmd_alloc()can still usekobject_put()in the error path and it makes a little more sense seeing we'd still be freeing the kobject stuff in it's own release method instead of the disks free method.
Uww, yes. I suspect the best fix is to actually stop the kobject from taking part in the md life time directly. Because the kobject contributes a reference to the disk until it is deleted, we might as well stop messing with the refcounts entirely and just call kobject_del on it just before del_gendisk. Let me see what I can do there.
Quoted 5 lines
- In the patch with
md_rdevs_overlap, it looks like we remove thercu_read_lock(), which definitely seems out of place and probably isn't correct. But the comment that was recreated still references the rcu so probably should be changed.
Fixed.
Quoted 1 line
- The last patch has a typo in the title (disk is *a* freed).
Fixed.
Quoted 9 lines · Christoph Hellwig
the first patch fixes a use after free, and the others deal with debugfs name reuse that spews warnings and makes debugfs use impossible for quickly reused gendisk instances. Both of those are rooted in sloppy life time rules for block device tear down.
Compared to the previous separate postings this adds a missing queue quiesce and documents
debugfs_mutexbetter.[...]
Applied, thanks!
[1/4] block: disable the elevator int del_gendisk
commit: 50e34d78815e474d410f342fbe783b18192ca518
[2/4] block: serialize all debugfs operations using q->debugfs_mutex
commit: 5cf9c91ba927119fc6606b938b1895bb2459d3bc
[3/4] block: remove per-disk debugfs files in blk_unregister_queue
commit: 99d055b4fd4bbb309c6cdb51a0d420669f777944
[4/4] block: freeze the queue earlier in del_gendisk
commit: a09b314005f3a0956ebf56e01b3b80339df577ccBest regards,
Signature · 1 line
Jens Axboe