From mboxrd@z Thu Jan 1 00:00:00 1970 From: Coly Li Subject: Re: [PATCH] bcache: add separate workqueue for journal_write to avoid deadlock Date: Thu, 27 Sep 2018 23:23:27 +0800 Message-ID: References: <1538055919-5757-1-git-send-email-fangguoju@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1538055919-5757-1-git-send-email-fangguoju@gmail.com> Content-Language: en-US Sender: linux-kernel-owner@vger.kernel.org To: guoju Cc: kent.overstreet@gmail.com, linux-bcache@vger.kernel.org, linux-kernel@vger.kernel.org List-Id: linux-bcache@vger.kernel.org On 9/27/18 9:45 PM, guoju wrote: > After write SSD completed, bcache schedule journal_write work to > system_wq, that is a public workqueue in system, without WQ_MEM_RECLAIM > flag. system_wq is also a bound wq, and there may be no idle kworker on > current processor. Creating a new kworker may unfortunately need to > reclaim memory first, by shrinking cache and slab used by vfs, which > depends on bcache device. That's a deadlock. > > This patch create a new workqueue for journal_write with WQ_MEM_RECLAIM > flag. It's rescuer thread will work to avoid the deadlock. > > Signed-off-by: guoju Nice catch, this fix is quite important. I will try to submit to Jens ASAP. Thanks. Coly Li > --- > drivers/md/bcache/bcache.h | 1 + > drivers/md/bcache/journal.c | 6 +++--- > drivers/md/bcache/super.c | 8 ++++++++ > 3 files changed, 12 insertions(+), 3 deletions(-) > > diff --git a/drivers/md/bcache/bcache.h b/drivers/md/bcache/bcache.h > index 83504dd..954dad2 100644 > --- a/drivers/md/bcache/bcache.h > +++ b/drivers/md/bcache/bcache.h > @@ -965,6 +965,7 @@ bool bch_alloc_sectors(struct cache_set *c, struct bkey *k, > void bch_write_bdev_super(struct cached_dev *dc, struct closure *parent); > > extern struct workqueue_struct *bcache_wq; > +extern struct workqueue_struct *bch_journal_wq; > extern struct mutex bch_register_lock; > extern struct list_head bch_cache_sets; > > diff --git a/drivers/md/bcache/journal.c b/drivers/md/bcache/journal.c > index 6116bbf..522c742 100644 > --- a/drivers/md/bcache/journal.c > +++ b/drivers/md/bcache/journal.c > @@ -485,7 +485,7 @@ static void do_journal_discard(struct cache *ca) > > closure_get(&ca->set->cl); > INIT_WORK(&ja->discard_work, journal_discard_work); > - schedule_work(&ja->discard_work); > + queue_work(bch_journal_wq, &ja->discard_work); > } > } > > @@ -592,7 +592,7 @@ static void journal_write_done(struct closure *cl) > : &j->w[0]; > > __closure_wake_up(&w->wait); > - continue_at_nobarrier(cl, journal_write, system_wq); > + continue_at_nobarrier(cl, journal_write, bch_journal_wq); > } > > static void journal_write_unlock(struct closure *cl) > @@ -627,7 +627,7 @@ static void journal_write_unlocked(struct closure *cl) > spin_unlock(&c->journal.lock); > > btree_flush_write(c); > - continue_at(cl, journal_write, system_wq); > + continue_at(cl, journal_write, bch_journal_wq); > return; > } > > diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c > index 3ede144..64715a8 100644 > --- a/drivers/md/bcache/super.c > +++ b/drivers/md/bcache/super.c > @@ -47,6 +47,7 @@ > static DEFINE_IDA(bcache_device_idx); > static wait_queue_head_t unregister_wait; > struct workqueue_struct *bcache_wq; > +struct workqueue_struct *bch_journal_wq; > > #define BTREE_MAX_PAGES (256 * 1024 / PAGE_SIZE) > /* limitation of partitions number on single bcache device */ > @@ -2344,6 +2345,9 @@ static void bcache_exit(void) > kobject_put(bcache_kobj); > if (bcache_wq) > destroy_workqueue(bcache_wq); > + if (bch_journal_wq) > + destroy_workqueue(bch_journal_wq); > + > if (bcache_major) > unregister_blkdev(bcache_major, "bcache"); > unregister_reboot_notifier(&reboot); > @@ -2373,6 +2377,10 @@ static int __init bcache_init(void) > if (!bcache_wq) > goto err; > > + bch_journal_wq = alloc_workqueue("bch_journal", WQ_MEM_RECLAIM, 0); > + if (!bch_journal_wq) > + goto err; > + > bcache_kobj = kobject_create_and_add("bcache", fs_kobj); > if (!bcache_kobj) > goto err;