From: Mikulas Patocka <mpatocka@redhat.com>
To: Kent Overstreet <koverstreet@google.com>
Cc: linux-bcache@vger.kernel.org, linux-kernel@vger.kernel.org,
dm-devel@redhat.com, tj@kernel.org, vgoyal@redhat.com,
bharrosh@panasas.com, Jens Axboe <axboe@kernel.dk>
Subject: Re: [PATCH v7 9/9] block: Avoid deadlocks with bio allocation by stacking drivers
Date: Wed, 29 Aug 2012 12:24:43 -0400 (EDT) [thread overview]
Message-ID: <Pine.LNX.4.64.1208291210180.774@file.rdu.redhat.com> (raw)
In-Reply-To: <1346175456-1572-10-git-send-email-koverstreet@google.com>
Hi
This fixes the bio allocation problems, but doesn't fix a similar deadlock
in device mapper when allocating from md->io_pool or other mempools in
the target driver.
The problem is that majority of device mapper code assumes that if we
submit a bio, that bio will be finished in a finite time. The commit
d89d87965dcbe6fe4f96a2a7e8421b3a75f634d1 in 2.6.22 broke this assumption.
I suggest - instead of writing workarounds for this current->bio_list
misbehavior, why not remove current->bio_list at all? We could revert
d89d87965dcbe6fe4f96a2a7e8421b3a75f634d1, allocate a per-device workqueue,
test stack usage in generic_make_request, and if it is too high (more than
half of the stack used, or so), put the bio to the target device's
blockqueue.
That could be simpler than allocating per-bioset workqueue and it also
solves more problems (possible deadlocks in dm).
Mikulas
On Tue, 28 Aug 2012, Kent Overstreet wrote:
> Previously, if we ever try to allocate more than once from the same bio
> set while running under generic_make_request() (i.e. a stacking block
> driver), we risk deadlock.
>
> This is because of the code in generic_make_request() that converts
> recursion to iteration; any bios we submit won't actually be submitted
> (so they can complete and eventually be freed) until after we return -
> this means if we allocate a second bio, we're blocking the first one
> from ever being freed.
>
> Thus if enough threads call into a stacking block driver at the same
> time with bios that need multiple splits, and the bio_set's reserve gets
> used up, we deadlock.
>
> This can be worked around in the driver code - we could check if we're
> running under generic_make_request(), then mask out __GFP_WAIT when we
> go to allocate a bio, and if the allocation fails punt to workqueue and
> retry the allocation.
>
> But this is tricky and not a generic solution. This patch solves it for
> all users by inverting the previously described technique. We allocate a
> rescuer workqueue for each bio_set, and then in the allocation code if
> there are bios on current->bio_list we would be blocking, we punt them
> to the rescuer workqueue to be submitted.
>
> Tested it by forcing the rescue codepath to be taken (by disabling the
> first GFP_NOWAIT) attempt, and then ran it with bcache (which does a lot
> of arbitrary bio splitting) and verified that the rescuer was being
> invoked.
>
> Signed-off-by: Kent Overstreet <koverstreet@google.com>
> CC: Jens Axboe <axboe@kernel.dk>
> ---
> fs/bio.c | 73 ++++++++++++++++++++++++++++++++++++++++++++++++++---
> include/linux/bio.h | 9 +++++++
> 2 files changed, 79 insertions(+), 3 deletions(-)
>
> diff --git a/fs/bio.c b/fs/bio.c
> index 31e637a..5d46318 100644
> --- a/fs/bio.c
> +++ b/fs/bio.c
> @@ -285,6 +285,23 @@ void bio_reset(struct bio *bio)
> }
> EXPORT_SYMBOL(bio_reset);
>
> +static void bio_alloc_rescue(struct work_struct *work)
> +{
> + struct bio_set *bs = container_of(work, struct bio_set, rescue_work);
> + struct bio *bio;
> +
> + while (1) {
> + spin_lock(&bs->rescue_lock);
> + bio = bio_list_pop(&bs->rescue_list);
> + spin_unlock(&bs->rescue_lock);
> +
> + if (!bio)
> + break;
> +
> + generic_make_request(bio);
> + }
> +}
> +
> /**
> * bio_alloc_bioset - allocate a bio for I/O
> * @gfp_mask: the GFP_ mask given to the slab allocator
> @@ -307,6 +324,7 @@ EXPORT_SYMBOL(bio_reset);
> */
> struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
> {
> + gfp_t saved_gfp = gfp_mask;
> unsigned front_pad;
> unsigned inline_vecs;
> unsigned long idx = BIO_POOL_NONE;
> @@ -324,13 +342,37 @@ struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
> front_pad = 0;
> inline_vecs = nr_iovecs;
> } else {
> + /*
> + * generic_make_request() converts recursion to iteration; this
> + * means if we're running beneath it, any bios we allocate and
> + * submit will not be submitted (and thus freed) until after we
> + * return.
> + *
> + * This exposes us to a potential deadlock if we allocate
> + * multiple bios from the same bio_set() while running
> + * underneath generic_make_request(). If we were to allocate
> + * multiple bios (say a stacking block driver that was splitting
> + * bios), we would deadlock if we exhausted the mempool's
> + * reserve.
> + *
> + * We solve this, and guarantee forward progress, with a rescuer
> + * workqueue per bio_set. If we go to allocate and there are
> + * bios on current->bio_list, we first try the allocation
> + * without __GFP_WAIT; if that fails, we punt those bios we
> + * would be blocking to the rescuer workqueue before we retry
> + * with the original gfp_flags.
> + */
> +
> + if (current->bio_list && !bio_list_empty(current->bio_list))
> + gfp_mask &= ~__GFP_WAIT;
> +retry:
> p = mempool_alloc(bs->bio_pool, gfp_mask);
> front_pad = bs->front_pad;
> inline_vecs = BIO_INLINE_VECS;
> }
>
> if (unlikely(!p))
> - return NULL;
> + goto err;
>
> bio = p + front_pad;
> bio_init(bio);
> @@ -351,6 +393,19 @@ struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
>
> err_free:
> mempool_free(p, bs->bio_pool);
> +err:
> + if (gfp_mask != saved_gfp) {
> + gfp_mask = saved_gfp;
> +
> + spin_lock(&bs->rescue_lock);
> + bio_list_merge(&bs->rescue_list, current->bio_list);
> + bio_list_init(current->bio_list);
> + spin_unlock(&bs->rescue_lock);
> +
> + queue_work(bs->rescue_workqueue, &bs->rescue_work);
> + goto retry;
> + }
> +
> return NULL;
> }
> EXPORT_SYMBOL(bio_alloc_bioset);
> @@ -1562,6 +1617,9 @@ static void biovec_free_pools(struct bio_set *bs)
>
> void bioset_free(struct bio_set *bs)
> {
> + if (bs->rescue_workqueue)
> + destroy_workqueue(bs->rescue_workqueue);
> +
> if (bs->bio_pool)
> mempool_destroy(bs->bio_pool);
>
> @@ -1597,6 +1655,10 @@ struct bio_set *bioset_create(unsigned int pool_size, unsigned int front_pad)
>
> bs->front_pad = front_pad;
>
> + spin_lock_init(&bs->rescue_lock);
> + bio_list_init(&bs->rescue_list);
> + INIT_WORK(&bs->rescue_work, bio_alloc_rescue);
> +
> bs->bio_slab = bio_find_or_create_slab(front_pad + back_pad);
> if (!bs->bio_slab) {
> kfree(bs);
> @@ -1607,9 +1669,14 @@ struct bio_set *bioset_create(unsigned int pool_size, unsigned int front_pad)
> if (!bs->bio_pool)
> goto bad;
>
> - if (!biovec_create_pools(bs, pool_size))
> - return bs;
> + if (biovec_create_pools(bs, pool_size))
> + goto bad;
> +
> + bs->rescue_workqueue = alloc_workqueue("bioset", WQ_MEM_RECLAIM, 0);
> + if (!bs->rescue_workqueue)
> + goto bad;
>
> + return bs;
> bad:
> bioset_free(bs);
> return NULL;
> diff --git a/include/linux/bio.h b/include/linux/bio.h
> index 3a8345e..84fdaac 100644
> --- a/include/linux/bio.h
> +++ b/include/linux/bio.h
> @@ -492,6 +492,15 @@ struct bio_set {
> mempool_t *bio_integrity_pool;
> #endif
> mempool_t *bvec_pool;
> +
> + /*
> + * Deadlock avoidance for stacking block drivers: see comments in
> + * bio_alloc_bioset() for details
> + */
> + spinlock_t rescue_lock;
> + struct bio_list rescue_list;
> + struct work_struct rescue_work;
> + struct workqueue_struct *rescue_workqueue;
> };
>
> struct biovec_slab {
> --
> 1.7.12
>
next prev parent reply other threads:[~2012-08-29 16:25 UTC|newest]
Thread overview: 75+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-08-28 17:37 [PATCH v7 0/9] Block cleanups, deadlock fix Kent Overstreet
2012-08-28 17:37 ` [PATCH v7 1/9] block: Generalized bio pool freeing Kent Overstreet
2012-08-28 17:37 ` [PATCH v7 2/9] dm: Use bioset's front_pad for dm_rq_clone_bio_info Kent Overstreet
2012-08-28 17:37 ` [PATCH v7 3/9] block: Add bio_reset() Kent Overstreet
2012-08-28 20:31 ` Tejun Heo
2012-08-28 22:17 ` Kent Overstreet
2012-08-28 22:53 ` Kent Overstreet
2012-09-01 2:23 ` Tejun Heo
2012-09-05 20:13 ` Kent Overstreet
2012-08-28 17:37 ` [PATCH v7 4/9] pktcdvd: Switch to bio_kmalloc() Kent Overstreet
2012-08-28 20:32 ` Tejun Heo
2012-08-28 22:24 ` Kent Overstreet
2012-09-04 9:05 ` Jiri Kosina
2012-09-05 19:44 ` Kent Overstreet
2012-08-28 17:37 ` [PATCH v7 5/9] block: Kill bi_destructor Kent Overstreet
2012-08-28 20:36 ` Tejun Heo
2012-08-28 22:07 ` Kent Overstreet
2012-08-28 17:37 ` [PATCH v7 6/9] block: Consolidate bio_alloc_bioset(), bio_kmalloc() Kent Overstreet
2012-08-28 20:41 ` Tejun Heo
2012-08-28 22:03 ` Kent Overstreet
2012-09-01 2:17 ` Tejun Heo
2012-08-28 17:37 ` [PATCH v7 7/9] block: Add bio_clone_bioset(), bio_clone_kmalloc() Kent Overstreet
2012-08-28 20:44 ` Tejun Heo
2012-08-28 22:05 ` Kent Overstreet
2012-09-01 2:19 ` Tejun Heo
2012-08-28 17:37 ` [PATCH v7 8/9] block: Reorder struct bio_set Kent Overstreet
2012-08-28 17:37 ` [PATCH v7 9/9] block: Avoid deadlocks with bio allocation by stacking drivers Kent Overstreet
2012-08-28 20:49 ` Tejun Heo
2012-08-28 22:28 ` Kent Overstreet
2012-08-28 23:01 ` Kent Overstreet
2012-08-29 1:31 ` Vivek Goyal
2012-08-29 3:25 ` Kent Overstreet
2012-08-29 12:57 ` Vivek Goyal
2012-08-29 14:39 ` [dm-devel] " Alasdair G Kergon
2012-08-29 16:26 ` Kent Overstreet
2012-08-29 21:01 ` John Stoffel
2012-08-29 21:08 ` Kent Overstreet
2012-08-28 22:06 ` Vivek Goyal
2012-08-28 22:23 ` Kent Overstreet
2012-08-29 16:24 ` Mikulas Patocka [this message]
2012-08-29 16:50 ` Kent Overstreet
2012-08-29 16:57 ` [dm-devel] " Alasdair G Kergon
2012-08-29 17:07 ` Vivek Goyal
2012-08-29 17:13 ` Kent Overstreet
2012-08-29 17:23 ` [dm-devel] " Alasdair G Kergon
2012-08-29 17:32 ` Kent Overstreet
2012-08-30 22:07 ` Vivek Goyal
2012-08-31 1:43 ` Kent Overstreet
2012-08-31 1:55 ` Kent Overstreet
2012-08-31 15:01 ` Vivek Goyal
2012-09-03 1:26 ` Kent Overstreet
2012-09-03 20:41 ` Mikulas Patocka
2012-09-04 3:41 ` Kent Overstreet
2012-09-04 18:55 ` Tejun Heo
2012-09-04 19:01 ` Tejun Heo
2012-09-04 19:43 ` Kent Overstreet
2012-09-04 19:42 ` Kent Overstreet
2012-09-04 21:03 ` Tejun Heo
2012-09-04 19:26 ` Mikulas Patocka
2012-09-04 19:39 ` Vivek Goyal
2012-09-04 19:51 ` [PATCH] dm: Use bioset's front_pad for dm_target_io Kent Overstreet
2012-09-04 21:20 ` Tejun Heo
2012-09-11 19:28 ` [PATCH 2] " Mikulas Patocka
2012-09-11 19:50 ` Kent Overstreet
2012-09-12 22:31 ` Mikulas Patocka
2012-09-14 23:09 ` [dm-devel] " Alasdair G Kergon
2012-09-01 2:13 ` [PATCH v7 9/9] block: Avoid deadlocks with bio allocation by stacking drivers Tejun Heo
2012-09-03 1:34 ` [PATCH v2] " Kent Overstreet
2012-09-04 15:00 ` [PATCH v7 9/9] " Vivek Goyal
2012-09-03 0:49 ` Dave Chinner
2012-09-03 1:17 ` Kent Overstreet
2012-09-04 13:54 ` Vivek Goyal
2012-09-04 18:26 ` Tejun Heo
2012-09-05 3:57 ` Dave Chinner
2012-09-05 4:37 ` Tejun Heo
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=Pine.LNX.4.64.1208291210180.774@file.rdu.redhat.com \
--to=mpatocka@redhat.com \
--cc=axboe@kernel.dk \
--cc=bharrosh@panasas.com \
--cc=dm-devel@redhat.com \
--cc=koverstreet@google.com \
--cc=linux-bcache@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=tj@kernel.org \
--cc=vgoyal@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).