Re: [PATCH v2 1/1] block: fix blk_queue_split() resource exhaustion

From: Mikulas Patocka
Date: Fri Jan 06 2017 - 12:34:54 EST




On Fri, 6 Jan 2017, Mikulas Patocka wrote:

>
>
> On Wed, 4 Jan 2017, Mike Snitzer wrote:
>
> > On Wed, Jan 04 2017 at 12:12am -0500,
> > NeilBrown <neilb@xxxxxxxx> wrote:
> >
> > > > Suggested-by: NeilBrown <neilb@xxxxxxxx>
> > > > Signed-off-by: Jack Wang <jinpu.wang@xxxxxxxxxxxxxxxx>
> > > > ---
> > > > block/blk-core.c | 20 ++++++++++++++++++++
> > > > 1 file changed, 20 insertions(+)
> > > >
> > > > diff --git a/block/blk-core.c b/block/blk-core.c
> > > > index 9e3ac56..47ef373 100644
> > > > --- a/block/blk-core.c
> > > > +++ b/block/blk-core.c
> > > > @@ -2138,10 +2138,30 @@ blk_qc_t generic_make_request(struct bio *bio)
> > > > struct request_queue *q = bdev_get_queue(bio->bi_bdev);
> > > >
> > > > if (likely(blk_queue_enter(q, __GFP_DIRECT_RECLAIM) == 0)) {
> > > > + struct bio_list lower, same, hold;
> > > > +
> > > > + /* Create a fresh bio_list for all subordinate requests */
> > > > + bio_list_init(&hold);
> > > > + bio_list_merge(&hold, &bio_list_on_stack);
> > > > + bio_list_init(&bio_list_on_stack);
> > > >
> > > > ret = q->make_request_fn(q, bio);
> > > >
> > > > blk_queue_exit(q);
> > > > + /* sort new bios into those for a lower level
> > > > + * and those for the same level
> > > > + */
> > > > + bio_list_init(&lower);
> > > > + bio_list_init(&same);
> > > > + while ((bio = bio_list_pop(&bio_list_on_stack)) != NULL)
> > > > + if (q == bdev_get_queue(bio->bi_bdev))
> > > > + bio_list_add(&same, bio);
> > > > + else
> > > > + bio_list_add(&lower, bio);
> > > > + /* now assemble so we handle the lowest level first */
> > > > + bio_list_merge(&bio_list_on_stack, &lower);
> > > > + bio_list_merge(&bio_list_on_stack, &same);
> > > > + bio_list_merge(&bio_list_on_stack, &hold);
> > > >
> > > > bio = bio_list_pop(current->bio_list);
> > > > } else {
> > > > --
> > > > 2.7.4
> >
> > Mikulas, would you be willing to try the below patch with the
> > dm-snapshot deadlock scenario and report back on whether it fixes that?
> >
> > Patch below looks to be the same as here:
> > https://marc.info/?l=linux-raid&m=148232453107685&q=p3
> >
> > Neil and/or others if that isn't the patch that should be tested please
> > provide a pointer to the latest.
> >
> > Thanks,
> > Mike
>
> The bad news is that this doesn't fix the snapshot deadlock.
>
> I created a test program for the snapshot deadlock bug (it was originally
> created years ago to test for a different bug, so it contains some cruft).
> You also need to insert "if (ci->sector_count) msleep(100);" to the end of
> __split_and_process_non_flush to make the kernel sleep when splitting the
> bio.
>
> And with the above above patch, the snapshot deadlock bug still happens.
>
> Mikulas
>
>
> #define _XOPEN_SOURCE 500
> #define _GNU_SOURCE
> #include <stdio.h>
> #include <stdlib.h>
> #include <unistd.h>
> #include <fcntl.h>
> #include <string.h>
> #include <errno.h>
> #include <malloc.h>
> #include <pthread.h>
> #include <asm/unistd.h>
>
> /*
> * Change "VG" symbol to a volume group name that you are using.
> *
> * You must apply this patch to the kernel to trigger the bug:
> * Index: linux-4.10-rc2/drivers/md/dm.c
> * ===================================================================
> * --- linux-4.10-rc2.orig/drivers/md/dm.c
> * +++ linux-4.10-rc2/drivers/md/dm.c
> * @@ -1223,6 +1223,9 @@ static int __split_and_process_non_flush
> * ci->sector += len;
> * ci->sector_count -= len;
> *
> * + if (ci->sector_count)
> * + msleep(100);
> * +
> * return 0;
> * }
> *
> */
>
> #define VG "vg1"
> #define LV "test_lv"
> #define LV_SNAP "test_snap"
> #define MEGABYTES "12"
> #define SNAP_MEGABYTES "16"
> #define THREADS 1
> #define BS 4096
> #define SKEW 512
> #define ORIG_PATTERN 'p'
> #define NEW_PATTERN 'n'
>
> enum {
> IOPRIO_CLASS_NONE,
> IOPRIO_CLASS_RT,
> IOPRIO_CLASS_BE,
> IOPRIO_CLASS_IDLE,
> };
>
> enum {
> IOPRIO_WHO_PROCESS = 1,
> IOPRIO_WHO_PGRP,
> IOPRIO_WHO_USER,
> };
>
> #define IOPRIO_CLASS_SHIFT 13
>
> static inline int ioprio_set(int which, int who, int ioprio)
> {
> return syscall(__NR_ioprio_set, which, who, ioprio);
> }
>
> static inline int ioprio_get(int which, int who)
> {
> return syscall(__NR_ioprio_get, which, who);
> }
>
> #define PRIO_READER ((IOPRIO_CLASS_IDLE << IOPRIO_CLASS_SHIFT) | 0xff)
> #define PRIO_WRITER (IOPRIO_CLASS_RT << IOPRIO_CLASS_SHIFT)
>
> static void do_cmd(char *cmd, int ign_err)
> {
> int r;
> fprintf(stderr, "* %s\n", cmd);
> r = system(cmd);
> if (r) {
> if (r == -1) {
> perror("system");
> } else {
> if (ign_err) return;
> fprintf(stderr, "return code %x\n", r);
> }
> exit(1);
> }
> }
>
> static char pattern[BS];
>
> static int h_orig, h_snap;
> static int n;
> static long long test_of;
> static pthread_rwlock_t rw_lock_1;
> static pthread_rwlock_t rw_lock_2;
> static pthread_rwlock_t rw_lock_3;
> static volatile int started = 0;
>
> static void pthread_error(int r)
> {
> fprintf(stderr, "pthread_error: %s\n", strerror(r));
> exit(1);
> }
>
> static void *test_read(long long of)
> {
> int r;
> char *t = memalign(BS, BS);
> if (!t) perror("memalign"), exit(1);
> if ((r = pread(h_snap, t, BS, of)) != BS) {
> fprintf(stderr, "can't read (%d): %s\n", r, strerror(errno));
> exit(1);
> }
> if (memcmp(pattern, t, BS)) {
> int i;
> for (i = 0; i < BS; i++) if (t[i] != pattern[i]) break;
> fprintf(stderr, "!!!! SNAPSHOT VOLUME DAMAGE AT BLOCK OFFSET %llX, BYTE OFFSET %X: %02x != %02x\n", of, i, (unsigned char)t[i], (unsigned char)pattern[i]);
> exit(2);
> }
> free(t);
> return NULL;
> }
>
> static void *test_thread(void *_)
> {
> int r;
> _ = _;
> //fprintf(stderr, "start\n");
> if ((r = ioprio_set(IOPRIO_WHO_PROCESS, 0, PRIO_READER))) perror("ioprio_set"), exit(1);
> if ((r = pthread_rwlock_rdlock(&rw_lock_2))) pthread_error(r);
> started = 1;
> if ((r = ioprio_get(IOPRIO_WHO_PROCESS, 0)) != PRIO_READER) {
> if (r == -1) perror("ioprio_get");
> else fprintf(stderr, "reader priority not set: %x\n", r);
> exit(1);
> }
> again:
> if ((r = pthread_rwlock_rdlock(&rw_lock_1))) pthread_error(r);
> if ((r = pthread_rwlock_unlock(&rw_lock_2))) pthread_error(r);
> if (test_of == -1) {
> if ((r = pthread_rwlock_unlock(&rw_lock_1))) pthread_error(r);
> //fprintf(stderr, "return\n");
> return NULL;
> }
> //fprintf(stderr, "test(%lld)\n", test_of);
> test_read(test_of);
> if ((r = pthread_rwlock_rdlock(&rw_lock_3))) pthread_error(r);
> if ((r = pthread_rwlock_unlock(&rw_lock_1))) pthread_error(r);
> if ((r = pthread_rwlock_rdlock(&rw_lock_2))) pthread_error(r);
> if ((r = pthread_rwlock_unlock(&rw_lock_3))) pthread_error(r);
> goto again;
> }
>
> int main(void)
> {
> int i, j, r;
> char *np;
> pthread_t thr[THREADS];
>
> memset(pattern, ORIG_PATTERN, sizeof pattern);
>
> do_cmd("lvremove -f "VG"/"LV_SNAP"", 1);
> do_cmd("lvremove -f "VG"/"LV"", 1);
> do_cmd("lvcreate -L "MEGABYTES" -n "LV" "VG"", 0);
>
> h_orig = open("/dev/mapper/"VG"-"LV"", O_RDWR);
> if (h_orig < 0) perror("open orig"), exit(1);
> if (lseek(h_orig, SKEW, SEEK_SET) == -1) perror("lseek"), exit(1);
> n = 0;
> while (write(h_orig, pattern, BS) == BS) {
> n++;
> fprintf(stderr, "creating %llx...\r", (long long)n * BS + SKEW);
> }
> if (fsync(h_orig)) perror("fsync"), exit(1);
> fprintf(stderr,"\n");
> lseek(h_orig, 0, SEEK_SET);
> close(h_orig);
>
> do_cmd("lvcreate -L "SNAP_MEGABYTES" -n "LV_SNAP" -s "VG"/"LV"", 0);
>
> h_orig = open("/dev/mapper/"VG"-"LV"", O_RDWR | O_DIRECT);
> if (h_orig < 0) perror("open orig"), exit(1);
>
> h_snap = open("/dev/mapper/"VG"-"LV_SNAP"", O_RDONLY | O_DIRECT);
> if (h_snap < 0) perror("open snap"), exit(1);
>
> if ((r = pthread_rwlock_init(&rw_lock_1, NULL))) pthread_error(r);
> if ((r = pthread_rwlock_init(&rw_lock_2, NULL))) pthread_error(r);
> if ((r = pthread_rwlock_init(&rw_lock_3, NULL))) pthread_error(r);
> if ((r = pthread_rwlock_wrlock(&rw_lock_1))) pthread_error(r);
> if ((r = pthread_rwlock_wrlock(&rw_lock_3))) pthread_error(r);
>
> if ((r = ioprio_set(IOPRIO_WHO_PROCESS, 0, PRIO_WRITER))) perror("ioprio_set"), exit(1);
>
> for (j = 0; j < THREADS; j++) {
> if ((r = pthread_create(&thr[j], NULL, test_thread, NULL))) pthread_error(r);
> }
> while (!started) usleep(1000);
>
> if ((r = ioprio_get(IOPRIO_WHO_PROCESS, 0)) != PRIO_WRITER) {
> if (r == -1) perror("ioprio_get");
> else fprintf(stderr, "writer priority not set: %x\n", r);
> exit(1);
> }
>
> np = memalign(BS, BS);
> if (!np) perror("memalign"), exit(1);
> memset(np, NEW_PATTERN, BS);
> for (i = 0; i < n; i++) {
> test_of = (off_t)i * BS + SKEW;
> fprintf(stderr, "testing %llx...\r", test_of);
> if ((r = pthread_rwlock_unlock(&rw_lock_1))) pthread_error(r);
> sched_yield();
> if (pwrite(h_orig, np, BS, test_of) != BS) {
> fprintf(stderr, "can't write (%d): %s\n", r, strerror(errno));
> exit(1);
> }
> if ((r = pthread_rwlock_wrlock(&rw_lock_2))) pthread_error(r);
> if ((r = pthread_rwlock_unlock(&rw_lock_3))) pthread_error(r);
> if ((r = pthread_rwlock_wrlock(&rw_lock_1))) pthread_error(r);
> if ((r = pthread_rwlock_unlock(&rw_lock_2))) pthread_error(r);
> if ((r = pthread_rwlock_wrlock(&rw_lock_3))) pthread_error(r);
> }
> fprintf(stderr,"\n");
>
> test_of = -1;
> if ((r = pthread_rwlock_unlock(&rw_lock_1))) pthread_error(r);
>
> for (j = 0; j < THREADS; j++) {
> if ((r = pthread_join(thr[j], NULL))) pthread_error(r);
> }
>
> fprintf(stderr, "TEST PASSED OK.\n");
>
> return 0;
> }
>
>

Here I post a patch that fixes the snapshot deadlock. On schedule(), it
redirects bios on current->bio_list to helper workqueues.

Mikulas


>From f126e182a053ef2e44a3e70b86df84d2b003530b Mon Sep 17 00:00:00 2001
From: Mikulas Patocka <mpatocka@xxxxxxxxxx>
Date: Tue, 27 May 2014 11:03:36 -0400
Subject: block: flush queued bios when process blocks to avoid deadlock

The block layer uses per-process bio list to avoid recursion in
generic_make_request. When generic_make_request is called recursively,
the bio is added to current->bio_list and generic_make_request returns
immediately. The top-level instance of generic_make_request takes bios
from current->bio_list and processes them.

Commit df2cb6daa4 ("block: Avoid deadlocks with bio allocation by
stacking drivers") created a workqueue for every bio set and code
in bio_alloc_bioset() that tries to resolve some low-memory deadlocks by
redirecting bios queued on current->bio_list to the workqueue if the
system is low on memory. However another deadlock (see below **) may
happen, without any low memory condition, because generic_make_request
is queuing bios to current->bio_list (rather than submitting them).

Fix this deadlock by redirecting any bios on current->bio_list to the
bio_set's rescue workqueue on every schedule call. Consequently, when
the process blocks on a mutex, the bios queued on current->bio_list are
dispatched to independent workqueus and they can complete without
waiting for the mutex to be available.

Also, now we can remove punt_bios_to_rescuer() and bio_alloc_bioset()'s
calls to it because bio_alloc_bioset() will implicitly punt all bios on
current->bio_list if it performs a blocking allocation.

** Here is the dm-snapshot deadlock that was observed:

1) Process A sends one-page read bio to the dm-snapshot target. The bio
spans snapshot chunk boundary and so it is split to two bios by device
mapper.

2) Device mapper creates the first sub-bio and sends it to the snapshot
driver.

3) The function snapshot_map calls track_chunk (that allocates a structure
dm_snap_tracked_chunk and adds it to tracked_chunk_hash) and then remaps
the bio to the underlying device and exits with DM_MAPIO_REMAPPED.

4) The remapped bio is submitted with generic_make_request, but it isn't
issued - it is added to current->bio_list instead.

5) Meanwhile, process B (dm's kcopyd) executes pending_complete for the
chunk affected be the first remapped bio, it takes down_write(&s->lock)
and then loops in __check_for_conflicting_io, waiting for
dm_snap_tracked_chunk created in step 3) to be released.

6) Process A continues, it creates a second sub-bio for the rest of the
original bio.

7) snapshot_map is called for this new bio, it waits on
down_write(&s->lock) that is held by Process B (in step 5).

Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1267650
Signed-off-by: Mikulas Patocka <mpatocka@xxxxxxxxxx>
Signed-off-by: Mike Snitzer <snitzer@xxxxxxxxxx>
Depends-on: df2cb6daa4 ("block: Avoid deadlocks with bio allocation by stacking drivers")
Cc: stable@xxxxxxxxxxxxxxx

---
block/bio.c | 77 +++++++++++++++++++------------------------------
include/linux/blkdev.h | 24 ++++++++++-----
kernel/sched/core.c | 7 +---
3 files changed, 50 insertions(+), 58 deletions(-)

Index: linux-4.9-rc3/block/bio.c
===================================================================
--- linux-4.9-rc3.orig/block/bio.c 2016-11-02 23:05:03.000000000 +0100
+++ linux-4.9-rc3/block/bio.c 2016-11-02 23:05:21.000000000 +0100
@@ -353,35 +353,37 @@ static void bio_alloc_rescue(struct work
}
}

-static void punt_bios_to_rescuer(struct bio_set *bs)
+/**
+ * blk_flush_bio_list
+ * @tsk: task_struct whose bio_list must be flushed
+ *
+ * Pop bios queued on @tsk->bio_list and submit each of them to
+ * their rescue workqueue.
+ *
+ * If the bio doesn't have a bio_set, we leave it on @tsk->bio_list.
+ * If the bio is allocated from fs_bio_set, we must leave it to avoid
+ * deadlock on loopback block device.
+ * Stacking bio drivers should use bio_set, so this shouldn't be
+ * an issue.
+ */
+void blk_flush_bio_list(struct task_struct *tsk)
{
- struct bio_list punt, nopunt;
struct bio *bio;
+ struct bio_list list = *tsk->bio_list;
+ bio_list_init(tsk->bio_list);

- /*
- * In order to guarantee forward progress we must punt only bios that
- * were allocated from this bio_set; otherwise, if there was a bio on
- * there for a stacking driver higher up in the stack, processing it
- * could require allocating bios from this bio_set, and doing that from
- * our own rescuer would be bad.
- *
- * Since bio lists are singly linked, pop them all instead of trying to
- * remove from the middle of the list:
- */
-
- bio_list_init(&punt);
- bio_list_init(&nopunt);
-
- while ((bio = bio_list_pop(current->bio_list)))
- bio_list_add(bio->bi_pool == bs ? &punt : &nopunt, bio);
-
- *current->bio_list = nopunt;
-
- spin_lock(&bs->rescue_lock);
- bio_list_merge(&bs->rescue_list, &punt);
- spin_unlock(&bs->rescue_lock);
+ while ((bio = bio_list_pop(&list))) {
+ struct bio_set *bs = bio->bi_pool;
+ if (unlikely(!bs) || bs == fs_bio_set) {
+ bio_list_add(tsk->bio_list, bio);
+ continue;
+ }

- queue_work(bs->rescue_workqueue, &bs->rescue_work);
+ spin_lock(&bs->rescue_lock);
+ bio_list_add(&bs->rescue_list, bio);
+ queue_work(bs->rescue_workqueue, &bs->rescue_work);
+ spin_unlock(&bs->rescue_lock);
+ }
}

/**
@@ -421,7 +423,6 @@ static void punt_bios_to_rescuer(struct
*/
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;
struct bio_vec *bvl = NULL;
@@ -455,23 +456,11 @@ struct bio *bio_alloc_bioset(gfp_t gfp_m
* 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_DIRECT_RECLAIM; if that fails, we punt those
- * bios we would be blocking to the rescuer workqueue before
- * we retry with the original gfp_flags.
+ * workqueue per bio_set. If an allocation would block (due to
+ * __GFP_DIRECT_RECLAIM) the scheduler will first punt all bios
+ * on current->bio_list to the rescuer workqueue.
*/
-
- if (current->bio_list && !bio_list_empty(current->bio_list))
- gfp_mask &= ~__GFP_DIRECT_RECLAIM;
-
p = mempool_alloc(bs->bio_pool, gfp_mask);
- if (!p && gfp_mask != saved_gfp) {
- punt_bios_to_rescuer(bs);
- gfp_mask = saved_gfp;
- p = mempool_alloc(bs->bio_pool, gfp_mask);
- }
-
front_pad = bs->front_pad;
inline_vecs = BIO_INLINE_VECS;
}
@@ -486,12 +475,6 @@ struct bio *bio_alloc_bioset(gfp_t gfp_m
unsigned long idx = 0;

bvl = bvec_alloc(gfp_mask, nr_iovecs, &idx, bs->bvec_pool);
- if (!bvl && gfp_mask != saved_gfp) {
- punt_bios_to_rescuer(bs);
- gfp_mask = saved_gfp;
- bvl = bvec_alloc(gfp_mask, nr_iovecs, &idx, bs->bvec_pool);
- }
-
if (unlikely(!bvl))
goto err_free;

Index: linux-4.9-rc3/include/linux/blkdev.h
===================================================================
--- linux-4.9-rc3.orig/include/linux/blkdev.h 2016-11-02 23:05:03.000000000 +0100
+++ linux-4.9-rc3/include/linux/blkdev.h 2016-11-02 23:05:21.000000000 +0100
@@ -1118,6 +1118,22 @@ static inline bool blk_needs_flush_plug(
!list_empty(&plug->cb_list));
}

+extern void blk_flush_bio_list(struct task_struct *tsk);
+
+static inline void blk_flush_queued_io(struct task_struct *tsk)
+{
+ /*
+ * Flush any queued bios to corresponding rescue threads.
+ */
+ if (tsk->bio_list && !bio_list_empty(tsk->bio_list))
+ blk_flush_bio_list(tsk);
+ /*
+ * Flush any plugged IO that is queued.
+ */
+ if (blk_needs_flush_plug(tsk))
+ blk_schedule_flush_plug(tsk);
+}
+
/*
* tag stuff
*/
@@ -1729,16 +1745,10 @@ static inline void blk_flush_plug(struct
{
}

-static inline void blk_schedule_flush_plug(struct task_struct *task)
+static inline void blk_flush_queued_io(struct task_struct *tsk)
{
}

-
-static inline bool blk_needs_flush_plug(struct task_struct *tsk)
-{
- return false;
-}
-
static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask,
sector_t *error_sector)
{
Index: linux-4.9-rc3/kernel/sched/core.c
===================================================================
--- linux-4.9-rc3.orig/kernel/sched/core.c 2016-11-02 23:05:03.000000000 +0100
+++ linux-4.9-rc3/kernel/sched/core.c 2016-11-02 23:05:21.000000000 +0100
@@ -3440,11 +3440,10 @@ static inline void sched_submit_work(str
if (!tsk->state || tsk_is_pi_blocked(tsk))
return;
/*
- * If we are going to sleep and we have plugged IO queued,
+ * If we are going to sleep and we have queued IO,
* make sure to submit it to avoid deadlocks.
*/
- if (blk_needs_flush_plug(tsk))
- blk_schedule_flush_plug(tsk);
+ blk_flush_queued_io(tsk);
}

asmlinkage __visible void __sched schedule(void)
@@ -5067,7 +5066,7 @@ long __sched io_schedule_timeout(long ti
long ret;

current->in_iowait = 1;
- blk_schedule_flush_plug(current);
+ blk_flush_queued_io(current);

delayacct_blkio_start();
rq = raw_rq();