| Message ID | 20250109055810.1402918-7-hch@lst.de |
|---|---|
| State | New |
| Headers | show |
| Series | [01/11] block: fix docs for freezing of queue limits updates | expand |
On Thu, Jan 09, 2025 at 01:07:47PM +0000, John Garry wrote: > Do you mean that the sysfs_lock could be removed in future? I would have > thought that queue limits lock could be used for the same thing, but I am > probably failing to see some lock nesting/ordering issues... More or less. Think about it: what does it even try to protect? Readіng/writing sysfs files vs itself and file removal it serialized by sysfs/kernfs internally. Any information tweaked in sysfs usually also has other places that can modify it. So we'll need a lock independent of sysfs for that anyway. A big part, buy by far all of that is covered by limits_lock. Serializing creating/removing sysfs attribues is supposed to be serialized using sysfs_dir_lock, although that needs a careful audit. It's also used to serialize a few debugfs things, but we'll need to look carefully for what exactly and switch that over to debugfs_mutex or something new. And then there's a bunch of misc cruft that also needs a careful look.
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c index d2aa2177e4ba..e828be777206 100644 --- a/block/blk-sysfs.c +++ b/block/blk-sysfs.c @@ -694,22 +694,24 @@ queue_attr_store(struct kobject *kobj, struct attribute *attr, if (entry->load_module) entry->load_module(disk, page, length); - mutex_lock(&q->sysfs_lock); - blk_mq_freeze_queue(q); if (entry->store_limit) { struct queue_limits lim = queue_limits_start_update(q); res = entry->store_limit(disk, page, length, &lim); if (res < 0) { queue_limits_cancel_update(q); - } else { - res = queue_limits_commit_update(q, &lim); - if (!res) - res = length; + return res; } - } else { - res = entry->store(disk, page, length); + + res = queue_limits_commit_update_frozen(q, &lim); + if (res) + return res; + return length; } + + mutex_lock(&q->sysfs_lock); + blk_mq_freeze_queue(q); + res = entry->store(disk, page, length); blk_mq_unfreeze_queue(q); mutex_unlock(&q->sysfs_lock); return res;
queue_attr_store() always freezes a device queue before calling the attribute store operation. For attributes that control queue limits, the store operation will also lock the queue limits with a call to queue_limits_start_update(). However, some drivers (e.g. SCSI sd) may need to issue commands to a device to obtain limit values from the hardware with the queue limits locked. This creates a potential ABBA deadlock situation if a user attempts to modify a limit (thus freezing the device queue) while the device driver starts a revalidation of the device queue limits. Avoid such deadlock by not freezing the queue before calling the ->store_limit() method in struct queue_sysfs_entry and instead use the queue_limits_commit_update_frozen helper to freeze the queue after taking the limits lock. This also removes taking the sysfs lock for the store_limit method as it doesn't protect anything here, but creates even more nesting. Hopefully it will go away from the actual sysfs methods entirely soon. (commit log adapted from a similar patch from Damien Le Moal) Fixes: ff956a3be95b ("block: use queue_limits_commit_update in queue_discard_max_store") Fixes: 0327ca9d53bf ("block: use queue_limits_commit_update in queue_max_sectors_store") Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Ming Lei <ming.lei@redhat.com> Reviewed-by: Damien Le Moal <dlemoal@kernel.org> Reviewed-by: Nilay Shroff <nilay@linux.ibm.com> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> --- block/blk-sysfs.c | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-)