[PATCH][COSMIC-unstable] UBUNTU: SAUCE: (noup) zfs to 0.7.9-3ubuntu4

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

[PATCH][COSMIC-unstable] UBUNTU: SAUCE: (noup) zfs to 0.7.9-3ubuntu4

Colin Ian King-2
From: Colin Ian King <[hidden email]>

BugLink: http://bugs.launchpad.net/bugs/1781364

This is a sync of zfs 0.7.9-3ubuntu4 that fixes a mount/umount deadlock
for upstream ZFS commit ac09630d8b0b ("Fix zpl_mount() deadlock")

ZFS commit 93b43af inadvertently introduced the following scenario which
can result in a deadlock.  This issue was most easily reproduced by
LXD containers using a ZFS storage backend but should be reproducible
under any workload which is frequently mounting and unmounting.

-- THREAD A --
spa_sync()
  spa_sync_upgrades()
    rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); <- Waiting on B

-- THREAD B --
mount_fs()
  zpl_mount()
    zpl_mount_impl()
      dmu_objset_hold()
        dmu_objset_hold_flags()
          dsl_pool_hold()
            dsl_pool_config_enter()
              rrw_enter(&dp->dp_config_rwlock, RW_READER, tag);
    sget()
      sget_userns()
        grab_super()
          down_write(&s->s_umount); <- Waiting on C

-- THREAD C --
cleanup_mnt()
  deactivate_super()
    down_write(&s->s_umount);
    deactivate_locked_super()
      zpl_kill_sb()
        kill_anon_super()
          generic_shutdown_super()
            sync_filesystem()
              zpl_sync_fs()
                zfs_sync()
                  zil_commit()
                    txg_wait_synced() <- Waiting on A

Signed-off-by: Brian Behlendorf <[hidden email]>
(backport from ZFS upstream commit ac09630d8b0bf6c92084a30fdaefd03fd0adbdc1)
Signed-off-by: Colin Ian King <[hidden email]>
---
 zfs/META                     |  2 +-
 zfs/include/sys/zfs_vfsops.h |  1 +
 zfs/module/zfs/zpl_super.c   | 11 ++++++++++-
 3 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/zfs/META b/zfs/META
index 374dda5..0e2d3d2 100644
--- a/zfs/META
+++ b/zfs/META
@@ -2,7 +2,7 @@ Meta:         1
 Name:         zfs
 Branch:       1.0
 Version:      0.7.9
-Release:      3ubuntu3
+Release:      3ubuntu4
 Release-Tags: relext
 License:      CDDL
 Author:       OpenZFS on Linux
diff --git a/zfs/include/sys/zfs_vfsops.h b/zfs/include/sys/zfs_vfsops.h
index 2326da4..927153b 100644
--- a/zfs/include/sys/zfs_vfsops.h
+++ b/zfs/include/sys/zfs_vfsops.h
@@ -32,6 +32,7 @@
 #include <sys/zil.h>
 #include <sys/sa.h>
 #include <sys/rrwlock.h>
+#include <sys/dsl_dataset.h>
 #include <sys/zfs_ioctl.h>
 
 #ifdef __cplusplus
diff --git a/zfs/module/zfs/zpl_super.c b/zfs/module/zfs/zpl_super.c
index fc10271..5c426b0 100644
--- a/zfs/module/zfs/zpl_super.c
+++ b/zfs/module/zfs/zpl_super.c
@@ -271,8 +271,17 @@ zpl_mount_impl(struct file_system_type *fs_type, int flags, zfs_mnt_t *zm)
  if (err)
  return (ERR_PTR(-err));
 
+ /*
+ * The dsl pool lock must be released prior to calling sget().
+ * It is possible sget() may block on the lock in grab_super()
+ * while deactivate_super() holds that same lock and waits for
+ * a txg sync.  If the dsl_pool lock is held over over sget()
+ * this can prevent the pool sync and cause a deadlock.
+ */
+ dsl_pool_rele(dmu_objset_pool(os), FTAG);
  s = zpl_sget(fs_type, zpl_test_super, set_anon_super, flags, os);
- dmu_objset_rele(os, FTAG);
+ dsl_dataset_rele(dmu_objset_ds(os), FTAG);
+
  if (IS_ERR(s))
  return (ERR_CAST(s));
 
--
2.7.4


--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|

APPLIED: [PATCH][COSMIC-unstable] UBUNTU: SAUCE: (noup) zfs to 0.7.9-3ubuntu4

Seth Forshee
On Thu, Jul 12, 2018 at 12:03:14PM +0100, Colin King wrote:

> From: Colin Ian King <[hidden email]>
>
> BugLink: http://bugs.launchpad.net/bugs/1781364
>
> This is a sync of zfs 0.7.9-3ubuntu4 that fixes a mount/umount deadlock
> for upstream ZFS commit ac09630d8b0b ("Fix zpl_mount() deadlock")
>
> ZFS commit 93b43af inadvertently introduced the following scenario which
> can result in a deadlock.  This issue was most easily reproduced by
> LXD containers using a ZFS storage backend but should be reproducible
> under any workload which is frequently mounting and unmounting.
>
> -- THREAD A --
> spa_sync()
>   spa_sync_upgrades()
>     rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); <- Waiting on B
>
> -- THREAD B --
> mount_fs()
>   zpl_mount()
>     zpl_mount_impl()
>       dmu_objset_hold()
>         dmu_objset_hold_flags()
>           dsl_pool_hold()
>             dsl_pool_config_enter()
>               rrw_enter(&dp->dp_config_rwlock, RW_READER, tag);
>     sget()
>       sget_userns()
>         grab_super()
>           down_write(&s->s_umount); <- Waiting on C
>
> -- THREAD C --
> cleanup_mnt()
>   deactivate_super()
>     down_write(&s->s_umount);
>     deactivate_locked_super()
>       zpl_kill_sb()
>         kill_anon_super()
>           generic_shutdown_super()
>             sync_filesystem()
>               zpl_sync_fs()
>                 zfs_sync()
>                   zil_commit()
>                     txg_wait_synced() <- Waiting on A
>
> Signed-off-by: Brian Behlendorf <[hidden email]>
> (backport from ZFS upstream commit ac09630d8b0bf6c92084a30fdaefd03fd0adbdc1)
> Signed-off-by: Colin Ian King <[hidden email]>

Applied to cosmic/master-next and unstable/master, thanks!

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team