Currently mounting an already mounted zfs dataset results in an
error, whereas it is typically allowed with other filesystems.
This causes some bad interactions with mount namespaces. Take
this sequence for example:
- Create a dataset
- Create a snapshot of the dataset
- Create a clone of the snapshot
- Create a new mount namespace
- Rename the original dataset
The rename results in unmounting and remounting the clone in the
original mount namespace, however the remount fails because the
dataset is still mounted in the new mount namespace. (Note that
this means the mount in the new mount namespace is never being
unmounted, so perhaps the unmount/remount of the clone isn't
actually necessary.)
The problem here is a result of the way mounting is implemented
in the kernel module. Since it is not mounting block devices it
uses mount_nodev() instead of the usual mount_bdev(). However,
mount_nodev() is written for filesystems for which each mount is
a new instance (i.e. a new super block), and zfs should be able
to detect when a mount request can be satisfied using an existing
super block.
Change zpl_mount() to call sget() directly with it's own test
callback. Passing the objset_t object as the fs data allows
checking if a superblock already exists for the dataset, and in
that case we just need to return a new reference for the sb's
root dentry.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tom Caputi <tcaputi@datto.com>
Signed-off-by: Alek Pinchuk <apinchuk@datto.com>
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Closes#5796Closes#7207
The HAVE_BLK_QUEUE_WRITE_CACHE_GPL_ONLY case was overlooked in
the original 10f88c5c commit because blk_queue_write_cache()
was available for the in-kernel builds.
Update the blk_queue_flag_{set,clear} wrappers to call the locked
versions to avoid confusion. This is safe for all existing callers.
The blk_queue_set_write_cache() function has been updated to use
these wrappers. This means setting/clearing both QUEUE_FLAG_WC
and QUEUE_FLAG_FUA is no longer atomic but this only done early
in zvol_alloc() prior to any requests so there is no issue.
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Reviewed-by: Kash Pande <kash@tripleback.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#7428Closes#7431
Currently, dnode_check_slots_free() works by checking dn->dn_type
in the dnode to determine if the dnode is reclaimable. However,
there is a small window of time between dnode_free_sync() in the
first call to dsl_dataset_sync() and when the useraccounting code
is run when the type is set DMU_OT_NONE, but the dnode is not yet
evictable, leading to crashes. This patch adds the ability for
dnodes to track which txg they were last dirtied in and adds a
check for this before performing the reclaim.
This patch also corrects several instances when dn_dirty_link was
treated as a list_node_t when it is technically a multilist_node_t.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#7147Closes#7388
This reverts commit cc63068e95.
Under certain circumstances this change can result in an ENOSPC
error when adding new files to a directory. See #7401 for full
details.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Issue #7401
Cloes #7416
mdb doesn't have dmu_ot[], so we need a different mechanism for its
SNPRINTF_BLKPTR() to determine if the BP is encrypted vs authenticated.
Additionally, since it already relies on BP_IS_ENCRYPTED (etc),
SNPRINTF_BLKPTR might as well figure out the "crypt_type" on its own,
rather than making the caller do so.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tom Caputi <tcaputi@datto.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#7390
Currently, the decryption and block authentication code in
the ZIO / ARC layers is a bit inconsistent with regards to
the ereports that are produces and the error codes that are
passed to calling functions. This patch ensures that all of
these errors (which begin as ECKSUM) are converted to EIO
before they leave the ZIO or ARC layer and that ereports
are correctly generated on each decryption / authentication
failure.
In addition, this patch fixes a bug in zio_decrypt() where
ECKSUM never gets written to zio->io_error.
Reviewed by: Matt Ahrens <matt@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#7372
During a receive operation zvol_create_minors_impl() can wait
needlessly for the prefetch thread because both share the same tasks
queue. This results in hung tasks:
<3>INFO: task z_zvol:5541 blocked for more than 120 seconds.
<3> Tainted: P O 3.16.0-4-amd64
<3>"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
The first z_zvol:5541 (zvol_task_cb) is waiting for the long running
traverse_prefetch_thread:260
root@linux:~# cat /proc/spl/taskq
taskq act nthr spwn maxt pri mina
spl_system_taskq/0 1 2 0 64 100 1
active: [260]traverse_prefetch_thread [zfs](0xffff88003347ae40)
wait: 5541
spl_delay_taskq/0 0 1 0 4 100 1
delay: spa_deadman [zfs](0xffff880039924000)
z_zvol/1 1 1 0 1 120 1
active: [5541]zvol_task_cb [zfs](0xffff88001fde6400)
pend: zvol_task_cb [zfs](0xffff88001fde6800)
This change adds a dedicated, per-pool, prefetch taskq to prevent the
traverse code from monopolizing the global (and limited) system_taskq by
inappropriately scheduling long running tasks on it.
Reviewed-by: Albert Lee <trisk@forkgnu.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: loli10K <ezomori.nozomu@gmail.com>
Closes#6330Closes#6890Closes#7343
Authored by: Andriy Gapon <avg@FreeBSD.org>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Don Brady <don.brady@delphix.com>
Reviewed-by: loli10K <ezomori.nozomu@gmail.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Richard Lowe <richlowe@richlowe.net>
Ported-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Porting Notes:
* Re-enabled and tweaked the zpool_upgrade_007_pos test case
to successfully run in under 5 minutes.
OpenZFS-issue: https://www.illumos.org/issues/9164
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/0e776dc06aCloses#6112Closes#7336
zfs_dbgmsg() should record a message by default. As a general
principal, these messages shouldn't be too verbose. Furthermore, the
amount of memory used is limited to 4MB (by default).
dprintf() should only record a message if this is a debug build, and
ZFS_DEBUG_DPRINTF is set in zfs_flags. This flag is not set by default
(even on debug builds). These messages are extremely verbose, and
sometimes nontrivial to compute.
SET_ERROR() should only record a message if ZFS_DEBUG_SET_ERROR is set
in zfs_flags. This flag is not set by default (even on debug builds).
This brings our behavior in line with illumos. Note that the message
format is unchanged (including file, line, and function, even though
these are not recorded on illumos).
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Prakash Surya <prakash.surya@delphix.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#7314
This patch adds some comments describing the purpose of "portable"
dnode and objset flags so that it is clear when new flags should
be added to the repective flag masks. This patch includes no
functional changes.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#7313
The changes piggyback JSON output support on top of channel programs
(#6558). This way the JSON output support is targeted to scripting
use cases and is easily maintainable since it really only touches
one function (zfs_do_channel_program()).
This patch ports Joyent's JSON nvlist library from illumos to enable
easy JSON printing of channel program output nvlist. To keep the
delta small I also took advantage of the fact that printing in
zfs_do_channel_program() was almost always done before exiting
the program.
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Richard Elling <Richard.Elling@RichardElling.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alek Pinchuk <apinchuk@datto.com>
Closes#7281
When the pool is suspended, record whether it was due to an I/O error or
due to MMP writes failing to succeed within the required time.
Change spa_suspended from uint8_t to zio_suspend_reason_t to store the
reason.
When userspace queries pool status via spa_tryimport(), report the
reason the pool was suspended in a new key,
ZPOOL_CONFIG_SUSPENDED_REASON.
In libzfs, when interpreting the returned config nvlist, report
suspension due to MMP with a new pool status enum value,
ZPOOL_STATUS_IO_FAILURE_MMP.
In status_callback(), which generates and emits the message when 'zpool
status' is executed, add a case to print an appropriate message for the
new pool status enum value.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Olaf Faaland <faaland1@llnl.gov>
Closes#7296
We hit an illegal memory access in the zrlock trace point. The problem
is that zrl->zr_owner and zrl->zr_caller are assigned locklessly. And if
zrl->zr_owner got assigned a longer string between when __string()
calculate the strlen, and when __assign_str() does strcpy. The copy will
overflow the buffer.
==
For example:
Initial condition:
zrl->zr_owner = A
zrl->zr_caller = "abc"
Thread A Thread B
-------------------------------------------------
if (zrl->zr_owner == A) {
DTRACE_PROBE2() {
__string() {
strlen(zrl->zr_caller) -> 3
allocate buf[4]
}
zrl->zr_owner = B
zrl->zr_caller = "abcd"
__assign_str() {
strcpy(buf, zrl->zr_caller) <- buffer overflow
==
Dereferencing zrl->zr_owner->pid may also be problematic, in that the
zrl->zr_owner got changed to other task, and that task exits, freeing
the task_struct. This should be very unlikely, as the other task need to
zrl_remove and exit between the dereferencing zr->zr_owner and
zr->zr_owner->pid. Nevertheless, we'll deal with it as well.
To fix the zrl->zr_caller issue, instead of copy the string content, we
just copy the pointer, this is safe because it always points to
__func__, which is static. As for the zrl->zr_owner issue, we pass in
curthread instead of using zrl->zr_owner.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chunwei Chen <david.chen@nutanix.com>
Closes#7291
get_format_prompt_string() and zpool_state_to_name() return
a string literal which is read-only, thus they should return
`const char*`.
zpool_get_prop_string() returns a non-const string after
successful nv-lookup, and returns a string literal otherwise.
Since this function is designed to be used for read-only purpose,
the return type should also be `const char*`.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tomohiro Kusumi <kusumi.tomohiro@osnexus.com>
Closes#7285
Once per pass through the MMP thread's loop, the vdev tree is walked to
find a suitable leaf to write the next MMP block to. If no such leaf is
found, the thread sleeps for a while and resumes at the top of the loop.
Add an entry to multihost_history when no leaf can be found, and record
the reason in the error column. The error code for such entries is a
bitfield, displayed in hex:
0x1 At least one vdev (interior or leaf) was not writeable.
0x2 At least one writeable leaf vdev was found, but it had a pending
MMP write.
timestamp = the time in seconds since the epoch when no leaf could be
found originally.
duration = the time (in ns) during which no MMP block was written for
this reason. This does not include the preceeding inter-write period
nor the following inter-write period.
vdev_guid = the number of sequential cycles of the MMP thread looop when
this occurred.
Sample output, truncated to fit:
For records of skipped MMP writes the right-most column, vdev_path, is
reported as "-".
id txg timestamp error duration mmp_delay vdev_guid ...
936 11 1520036441 0 146264 891422313 1740883117838 ...
937 11 1520036441 0 163956 888356657 7320395061548 ...
938 11 1520036442 0 130690 885314969 7320395061548 ...
939 11 1520036442 0 2001068577 882296582 1740883117838 ...
940 11 1520036443 0 161806 882296582 7320395061548 ...
941 11 1520036443 0x2 0 998020546 1 ...
942 11 1520036444 0 136585 998020546 7320395061548 ...
943 11 1520036444 0x2 0 998020257 1 ...
944 11 1520036445 5 2002662964 994160219 1740883117838 ...
945 11 1520036445 0x2 998073118 994160219 3 ...
946 11 1520036447 0 247136 994160219 7320395061548 ...
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Olaf Faaland <faaland1@llnl.gov>
Closes#7212
As of https://github.com/torvalds/linux/commit/fb6d47a, get_disk()
is now get_disk_and_module(). Add a configure check to determine
if we need to use get_disk_and_module().
Reviewed-by: loli10K <ezomori.nozomu@gmail.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Closes#7264
Change checksum & IO delay ratelimit thresholds from 5/sec to 20/sec.
This allows zed to actually trigger if a bunch of these events arrive in
a short period of time (zed has a threshold of 10 events in 10 sec).
Previously, if you had, say, 100 checksum errors in 1 sec, it would get
ratelimited to 5/sec which wouldn't trigger zed to fault the drive.
Also, convert the checksum and IO delay thresholds to module params for
easy testing.
Reviewed-by: loli10K <ezomori.nozomu@gmail.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes#7252
b1d21733 made it possible for empty metadnode blocks to be
compressed to a hole, fixing a bug that would cause invalid
metadnode MACs when a send stream attempted to free objects
and allowing the blocks to be reclaimed when they were no
longer needed. However, this patch also introduced a race
condition; if a txg sync occurred after a DRR_OBJECT_RANGE
record was received but before any objects were added, the
metadnode block would be compressed to a hole and lose all
of its encryption parameters. This would cause subsequent
DRR_OBJECT records to fail when they attempted to write
their data into an unencrypted block. This patch defers the
DRR_OBJECT_RANGE handling to receive_object() so that the
encryption parameters are set with each object that is
written into that block.
Reviewed-by: Kash Pande <kash@tripleback.net>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#7215Closes#7236
The conditional was reversed which caused garbage values to be used when
calculating dds_ref_dsize.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tom Caputi <tcaputi@datto.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Tim Chase <tim@chase2k.com>
Closes#7234
Provide infrastructure to auto-configure to enum and API changes in the
global page stats used for our free memory calculations.
arc_free_memory has been broken since an API change in Linux v3.14:
2016-07-28 v4.8 599d0c95 mm, vmscan: move LRU lists to node
2016-07-28 v4.8 75ef7184 mm, vmstat: add infrastructure for per-node
vmstats
These commits moved some of global_page_state() into
global_node_page_state(). The API change was particularly egregious as,
instead of breaking the old code, it silently did the wrong thing and we
continued using global_page_state() where we should have been using
global_node_page_state(), thus indexing into the wrong array via
NR_SLAB_RECLAIMABLE et al.
There have been further API changes along the way:
2017-07-06 v4.13 385386cf mm: vmstat: move slab statistics from zone to
node counters
2017-09-06 v4.14 c41f012a mm: rename global_page_state to
global_zone_page_state
...and various (incomplete, as it turns out) attempts to accomodate
these changes in ZoL:
2017-08-24 2209e409 Linux 4.8+ compatibility fix for vm stats
2017-09-16 787acae0 Linux 3.14 compat: IO acct, global_page_state, etc
2017-09-19 661907e6 Linux 4.14 compat: IO acct, global_page_state, etc
The config infrastructure provided here resolves these issues going back
to the original API change in v3.14 and is robust against further Linux
changes in this area.
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Chris Dunlop <chris@onthe.net.au>
Closes#7170
Use refcount_dec_and_test() on 4.16+ kernels, atomic_dec_and_test()
on older kernels. https://lwn.net/Articles/714974/
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes: #7179Closes: #7211
This change implements 'zfs send -b' which can be used to send only
received property values whether or not they are overridden by local
settings.
This can be very useful during "restore" operations from a backup pool
because it allows to send only the property values originally sent
from the backup source, even though they were later modified on the
destination either by a 'zfs set' operation, explicit 'zfs inherit' or
overridden during the receive process via 'zfs receive -o|-x'.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: loli10K <ezomori.nozomu@gmail.com>
Closes#7156
Currently, raw zfs sends transfer the encrypted master keys and
objset_phys_t encryption parameters in the DRR_BEGIN payload of
each send file. Both of these are processed as soon as they are
read in dmu_recv_stream(), meaning that the new keys are set
before the new snapshot is received. In addition to the fact that
this changes the user's keys for the dataset earlier than they
might expect, the keys were never reset to what they originally
were in the event that the receive failed. This patch splits the
processing into objset handling and key handling, the later of
which is moved to dmu_recv_end() so that they key change can be
done atomically.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#7200
Currently, the DMU relies on ZIO layer compression to free LO
dnode blocks that no longer have objects in them. However,
raw receives disable all compression, meaning that these blocks
can never be freed. In addition to the obvious space concerns,
this could also cause incremental raw receives to fail to mount
since the MAC of a hole is different from that of a completely
zeroed block.
This patch corrects this issue by adding a special case in
zio_write_compress() which will attempt to compress these blocks
to a hole even if ZIO_FLAG_RAW_ENCRYPT is set. This patch also
removes the zfs_mdcomp_disable tunable, since tuning it could
cause these same issues.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#7198
Currently, ZIL claiming dirties objsets which causes
dsl_pool_sync() to attempt to perform user accounting on
them. This causes problems for encrypted datasets that were
raw received before the system went offline since they
cannot perform user accounting until they have their keys
loaded. This triggers an ASSERT in zio_encrypt(). Since
encryption was added, the code now depends on the fact that
data should only be written when objsets are owned. This
patch adds a check in dmu_objset_do_userquota_updates()
to ensure that useraccounting is only done when the objsets
are actually owned for write. As part of this work, the
zfsvfs and zvol code was updated so that it no longer lies
about owning objsets readonly.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#6916Closes#7163
CID 173243, 173245: Memory - corruptions (OVERRUN)
Added size argument to lcompat_sprintf() to avoid use of INT_MAX
CID 173244: Integer handling issues (OVERFLOW_BEFORE_WIDEN)
Added cast to uint64_t to avoid a 32 bit overflow warning
CID 173242: Integer handling issues (CONSTANT_EXPRESSION_RESULT)
Conditionally removed unused luai_numisnan() floating point check
CID 173241: Resource leaks (RESOURCE_LEAK)
Added missing close(fd) on error path
CID 173240: (UNINIT)
Fixed uninitialized variable in get_special_prop()
CID 147560: Null pointer dereferences (NULL_RETURNS)
Cleaned up bad code merge in dsl_dataset_promote_check()
CID 28475: Memory - illegal accesses (OVERRUN)
Fixed lcompat_sprintf() to use a size paramater
CID 28418, 28422: Error handling issues (CHECKED_RETURN)
Added function result cast to (void) to avoid warning
CID 23935, 28411, 28412: Memory - corruptions (ARRAY_VS_SINGLETON)
Added casts to avoid exposing result as an array
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Don Brady <don.brady@delphix.com>
Closes#7181
PROBLEM
=======
It's possible for a parent zio to complete even though it has children
which have not completed. This can result in the following panic:
> $C
ffffff01809128c0 vpanic()
ffffff01809128e0 mutex_panic+0x58(fffffffffb94c904, ffffff597dde7f80)
ffffff0180912950 mutex_vector_enter+0x347(ffffff597dde7f80)
ffffff01809129b0 zio_remove_child+0x50(ffffff597dde7c58, ffffff32bd901ac0,
ffffff3373370908)
ffffff0180912a40 zio_done+0x390(ffffff32bd901ac0)
ffffff0180912a70 zio_execute+0x78(ffffff32bd901ac0)
ffffff0180912b30 taskq_thread+0x2d0(ffffff33bae44140)
ffffff0180912b40 thread_start+8()
> ::status
debugging crash dump vmcore.2 (64-bit) from batfs0390
operating system: 5.11 joyent_20170911T171900Z (i86pc)
image uuid: (not set)
panic message: mutex_enter: bad mutex, lp=ffffff597dde7f80
owner=ffffff3c59b39480 thread=ffffff0180912c40
dump content: kernel pages only
The problem is that dbuf_prefetch along with l2arc can create a zio tree
which confuses the parent zio and allows it to complete with while children
still exist. Here's the scenario:
zio tree:
pio
|--- lio
The parent zio, pio, has entered the zio_done stage and begins to check its
children to see there are still some that have not completed. In zio_done(),
the children are checked in the following order:
zio_wait_for_children(zio, ZIO_CHILD_VDEV, ZIO_WAIT_DONE)
zio_wait_for_children(zio, ZIO_CHILD_GANG, ZIO_WAIT_DONE)
zio_wait_for_children(zio, ZIO_CHILD_DDT, ZIO_WAIT_DONE)
zio_wait_for_children(zio, ZIO_CHILD_LOGICAL, ZIO_WAIT_DONE)
If pio, finds any child which has not completed then it stops executing and
goes to sleep. Each call to zio_wait_for_children() will grab the io_lock
while checking the particular child.
In this scenario, the pio has completed the first call to
zio_wait_for_children() to check for any ZIO_CHILD_VDEV children. Since
the only zio in the zio tree right now is the logical zio, lio, then it
completes that call and prepares to check the next child type.
In the meantime, the lio completes and in its callback creates a child vdev
zio, cio. The zio tree looks like this:
zio tree:
pio
|--- lio
|--- cio
The lio then grabs the parent's io_lock and removes itself.
zio tree:
pio
|--- cio
The pio continues to run but has already completed its check for ZIO_CHILD_VDEV
and will erroneously complete. When the child zio, cio, completes it will panic
the system trying to reference the parent zio which has been destroyed.
SOLUTION
========
The fix is to rework the zio_wait_for_children() logic to accept a bitfield
for all the children types that it's interested in checking. The
io_lock will is held the entire time we check all the children types. Since
the function now accepts a bitfield, a simple ZIO_CHILD_BIT() macro is provided
to allow for the conversion between a ZIO_CHILD type and the bitfield used by
the zio_wiat_for_children logic.
Authored by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Andriy Gapon <avg@FreeBSD.org>
Reviewed by: Youzhong Yang <youzhong@gmail.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Dan McDonald <danmcd@omniti.com>
Ported-by: Giuseppe Di Natale <dinatale2@llnl.gov>
OpenZFS-issue: https://www.illumos.org/issues/8857
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/862ff6d99c
Issue #5918Closes#7168
Project quota is a new ZFS system space/object usage accounting
and enforcement mechanism. Similar as user/group quota, project
quota is another dimension of system quota. It bases on the new
object attribute - project ID.
Project ID is a numerical value to indicate to which project an
object belongs. An object only can belong to one project though
you (the object owner or privileged user) can change the object
project ID via 'chattr -p' or 'zfs project [-s] -p' explicitly.
The object also can inherit the project ID from its parent when
created if the parent has the project inherit flag (that can be
set via 'chattr +P' or 'zfs project -s [-p]').
By accounting the spaces/objects belong to the same project, we
can know how many spaces/objects used by the project. And if we
set the upper limit then we can control the spaces/objects that
are consumed by such project. It is useful when multiple groups
and users cooperate for the same project, or a user/group needs
to participate in multiple projects.
Support the following commands and functionalities:
zfs set projectquota@project
zfs set projectobjquota@project
zfs get projectquota@project
zfs get projectobjquota@project
zfs get projectused@project
zfs get projectobjused@project
zfs projectspace
zfs allow projectquota
zfs allow projectobjquota
zfs allow projectused
zfs allow projectobjused
zfs unallow projectquota
zfs unallow projectobjquota
zfs unallow projectused
zfs unallow projectobjused
chattr +/-P
chattr -p project_id
lsattr -p
This patch also supports tree quota based on the project quota via
"zfs project" commands set as following:
zfs project [-d|-r] <file|directory ...>
zfs project -C [-k] [-r] <file|directory ...>
zfs project -c [-0] [-d|-r] [-p id] <file|directory ...>
zfs project [-p id] [-r] [-s] <file|directory ...>
For "df [-i] $DIR" command, if we set INHERIT (project ID) flag on
the $DIR, then the proejct [obj]quota and [obj]used values for the
$DIR's project ID will be shown as the total/free (avail) resource.
Keep the same behavior as EXT4/XFS does.
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by Ned Bass <bass6@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Fan Yong <fan.yong@intel.com>
TEST_ZIMPORT_POOLS="zol-0.6.1 zol-0.6.2 master"
Change-Id: Ib4f0544602e03fb61fd46a849d7ba51a6005693c
Closes#6290
With "casesensitivity=mixed", zap_add() could fail when the number of
files/directories with the same name (varying in case) exceed the
capacity of the leaf node of a Fatzap. This results in a ASSERT()
failure as zfs_link_create() does not expect zap_add() to fail. The fix
is to handle these failures and rollback the transactions.
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed-by: Chunwei Chen <david.chen@nutanix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Sanjeev Bagewadi <sanjeev.bagewadi@gmail.com>
Closes#7011Closes#7054
Related to commit 4859fe796, when directly using the kernel's
refcount functions in kernel compatibility code do not map
refcount_t to zfs_refcount_t. This leads to a type mismatch.
Longer term we should consider renaming refcount_t to
zfs_refcount_t in the zfs code base.
Reviewed-by: Olaf Faaland <faaland1@llnl.gov>
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Reviewed-by: Chunwei Chen <david.chen@nutanix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#7148
A new interface was added to manipulate the version field of an
inode. Add a inode_set_iversion() wrapper for older kernels and
use the new interface when available.
The i_version field was dropped from the trace point due to the
switch to an atomic64_t i_version type.
Reviewed-by: Olaf Faaland <faaland1@llnl.gov>
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Reviewed-by: Chunwei Chen <david.chen@nutanix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#7148
Authored by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Chris Williamson <chris.williamson@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Ported-by: Don Brady <don.brady@delphix.com>
We want to be able to run channel programs outside of synching
context. This would greatly improve performance for channel programs
that just gather information, as they won't have to wait for synching
context anymore.
=== What is implemented?
This feature introduces the following:
- A new command line flag in "zfs program" to specify our intention
to run in open context. (The -n option)
- A new flag/option within the channel program ioctl which selects
the context.
- Appropriate error handling whenever we try a channel program in
open-context that contains zfs.sync* expressions.
- Documentation for the new feature in the manual pages.
=== How do we handle zfs.sync functions in open context?
When such a function is found by the interpreter and we are running
in open context we abort the script and we spit out a descriptive
runtime error. For example, given the script below ...
arg = ...
fs = arg["argv"][1]
err = zfs.sync.destroy(fs)
msg = "destroying " .. fs .. " err=" .. err
return msg
if we run it in open context, we will get back the following error:
Channel program execution failed:
[string "channel program"]:3: running functions from the zfs.sync
submodule requires passing sync=TRUE to lzc_channel_program()
(i.e. do not specify the "-n" command line argument)
stack traceback:
[C]: in function 'destroy'
[string "channel program"]:3: in main chunk
=== What about testing?
We've introduced new wrappers for all channel program tests that
run each channel program as both (startard & open-context) and
expect the appropriate behavior depending on the program using
the zfs.sync module.
OpenZFS-issue: https://www.illumos.org/issues/8677
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/17a49e15Closes#6558
Authored by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Andy Stormont <astormont@racktopsystems.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Ported-by: Don Brady <don.brady@delphix.com>
Every time we want to unmount a snapshot (happens during snapshot
deletion or renaming) we unnecessarily iterate through all the
mountpoints in the VFS layer (see zfs_get_vfs).
The current patch completely gets rid of that code and changes
the approach while keeping the behavior of that code path the
same. Specifically, it puts a hold on the dataset/snapshot and
gets its vfs resource reference directly, instead of linearly
searching for it. If that reference exists we attempt to amount
it.
With the above change, it became obvious that the nvlist
manipulations that we do (add_boolean and add_nvlist) take a
significant amount of time ensuring uniqueness of every new
element even though they don't have too. Thus, we updated the
patch so those nvlists are not trying to enforce the uniqueness
of their elements.
A more complete analysis of the problem solved by this patch
can be found below:
https://sdimitro.github.io/post/snap-unmount-perf/
OpenZFS-issue: https://www.illumos.org/issues/8604
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/126118fb
Authored by: Chris Williamson <chris.williamson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Reviewed by: Brad Lewis <brad.lewis@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Ported-by: Don Brady <don.brady@delphix.com>
ZFS channel programs should be able to create snapshots.
In addition to the base snapshot functionality, this entails extra
logic to handle edge cases which were formerly not possible, such as
creating then destroying a snapshot in the same transaction sync.
OpenZFS-issue: https://www.illumos.org/issues/8600
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/68089b8b
Authored by: Brad Lewis <brad.lewis@delphix.com>
Reviewed by: Chris Williamson <chris.williamson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Ported-by: Don Brady <don.brady@delphix.com>
ZFS channel programs should be able to perform a rollback.
OpenZFS-issue: https://www.illumos.org/issues/8592
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/d46b5ed6
Authored by: Chris Williamson <chris.williamson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Approved by: Garrett D'Amore <garrett@damore.org>
Ported-by: Don Brady <don.brady@delphix.com>
Ported-by: John Kennedy <john.kennedy@delphix.com>
OpenZFS-issue: https://www.illumos.org/issues/7431
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/dfc11533
Porting Notes:
* The CLI long option arguments for '-t' and '-m' don't parse on linux
* Switched from kmem_alloc to vmem_alloc in zcp_lua_alloc
* Lua implementation is built as its own module (zlua.ko)
* Lua headers consumed directly by zfs code moved to 'include/sys/lua/'
* There is no native setjmp/longjump available in stock Linux kernel.
Brought over implementations from illumos and FreeBSD
* The get_temporary_prop() was adapted due to VFS platform differences
* Use of inline functions in lua parser to reduce stack usage per C call
* Skip some ZFS Test Suite ZCP tests on sparc64 to avoid stack overflow
Currently, os_next_write_raw is a single boolean used for determining
whether or not the next call to dmu_objset_sync() should write out
the objset_phys_t as a raw buffer. Since the boolean is not associated
with a txg, the work simply happens during the next txg, which is not
necessarily the correct one. In the current implementation this issue
was misdiagnosed, resulting in a small hack in dmu_objset_sync() which
seemed to resolve the problem.
This patch changes os_next_write_raw to be an array of booleans, one
for each txg in TXG_OFF and removes the hack.
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#6864
Currently, when a raw zfs send file includes a DRR_OBJECT record
that would decrease the number of levels of an existing object,
the object is reallocated with dmu_object_reclaim() which
creates the new dnode using the old object's nlevels. For non-raw
sends this doesn't really matter, but raw sends require that
nlevels on the receive side match that of the send side so that
the checksum-of-MAC tree can be properly maintained. This patch
corrects the issue by freeing the object completely before
allocating it again in this case.
This patch also corrects several issues with dnode_hold_impl()
and related functions that prevented dnodes (particularly
multi-slot dnodes) from being reallocated properly due to
the fact that existing dnodes were not being fully cleaned up
when they were freed.
This patch adds a test to make sure that zfs recv functions
properly with incremental streams containing dnodes of different
sizes.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#6821Closes#6864
The on-disk format for encrypted datasets protects not only
the encrypted and authenticated blocks themselves, but also
the order and interpretation of these blocks. In order to
make this work while maintaining the ability to do raw
sends, the indirect bps maintain a secure checksum of all
the MACs in the block below it along with a few other
fields that determine how the data is interpreted.
Unfortunately, the current on-disk format erroneously
includes some fields which are not portable and thus cannot
support raw sends. It is not possible to easily work around
this issue due to a separate and much smaller bug which
causes indirect blocks for encrypted dnodes to not be
compressed, which conflicts with the previous bug. In
addition, the current code generates incompatible on-disk
formats on big endian and little endian systems due to an
issue with how block pointers are authenticated. Finally,
raw send streams do not currently include dn_maxblkid when
sending both the metadnode and normal dnodes which are
needed in order to ensure that we are correctly maintaining
the portable objset MAC.
This patch zero's out the offending fields when computing
the bp MAC and ensures that these MACs are always
calculated in little endian order (regardless of the host
system's byte order). This patch also registers an errata
for the old on-disk format, which we detect by adding a
"version" field to newly created DSL Crypto Keys. We allow
datasets without a version (version 0) to only be mounted
for read so that they can easily be migrated. We also now
include dn_maxblkid in raw send streams to ensure the MAC
can be maintained correctly.
This patch also contains minor bug fixes and cleanups.
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#6845Closes#6864Closes#7052
This change was missed in 0735ecb334.
Reviewed-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Signed-off-by: András Korn <korn-github.com@elan.rulez.org>
Closes#7096
PROBLEM
=======
When `dmu_tx_assign` is called from `zil_lwb_write_issue`, it's possible
for either `ERESTART` or `EIO` to be returned.
If `ERESTART` is returned, this will cause an assertion to fail directly
in `zil_lwb_write_issue`, where the code assumes the return value is
`EIO` if `dmu_tx_assign` returns a non-zero value. This can occur if the
SPA is suspended when `dmu_tx_assign` is called, and most often occurs
when running `zloop`.
If `EIO` is returned, this can cause assertions to fail elsewhere in the
ZIL code. For example, `zil_commit_waiter_timeout` contains the
following logic:
lwb_t *nlwb = zil_lwb_write_issue(zilog, lwb);
ASSERT3S(lwb->lwb_state, !=, LWB_STATE_OPENED);
In this case, if `dmu_tx_assign` returned `EIO` from within
`zil_lwb_write_issue`, the `lwb` variable passed in will not be issued
to disk. Thus, it's `lwb_state` field will remain `LWB_STATE_OPENED` and
this assertion will fail. `zil_commit_waiter_timeout` assumes that after
it calls `zil_lwb_write_issue`, the `lwb` will be issued to disk, and
doesn't handle the case where this is not true; i.e. it doesn't handle
the case where `dmu_tx_assign` returns `EIO`.
SOLUTION
========
This change modifies the `dmu_tx_assign` function such that `txg_how` is
a bitmask, rather than of the `txg_how_t` enum type. Now, the previous
`TXG_WAITED` semantics can be used via `TXG_NOTHROTTLE`, along with
specifying either `TXG_NOWAIT` or `TXG_WAIT` semantics.
Previously, when `TXG_WAITED` was specified, `TXG_NOWAIT` semantics was
automatically invoked. This was not ideal when using `TXG_WAITED` within
`zil_lwb_write_issued`, leading the problem described above. Rather, we
want to achieve the semantics of `TXG_WAIT`, while also preventing the
`tx` from being penalized via the dirty delay throttling.
With this change, `zil_lwb_write_issued` can acheive the semtantics that
it requires by passing in the value `TXG_WAIT | TXG_NOTHROTTLE` to
`dmu_tx_assign`.
Further, consumers of `dmu_tx_assign` wishing to achieve the old
`TXG_WAITED` semantics can pass in the value `TXG_NOWAIT | TXG_NOTHROTTLE`.
Authored by: Prakash Surya <prakash.surya@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Andriy Gapon <avg@FreeBSD.org>
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
Porting Notes:
- Additionally updated `zfs_tmpfile` to use `TXG_NOTHROTTLE`
OpenZFS-issue: https://www.illumos.org/issues/8997
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/19ea6cb0f9Closes#7084
The intent of this patch is extend the existing deadman code
such that it's flexible enough to be used by both ztest and
on production systems. The proposed changes include:
* Added a new `zfs_deadman_failmode` module option which is
used to dynamically control the behavior of the deadman. It's
loosely modeled after, but independant from, the pool failmode
property. It can be set to wait, continue, or panic.
* wait - Wait for the "hung" I/O (default)
* continue - Attempt to recover from a "hung" I/O
* panic - Panic the system
* Added a new `zfs_deadman_ziotime_ms` module option which is
analogous to `zfs_deadman_synctime_ms` except instead of
applying to a pool TXG sync it applies to zio_wait(). A
default value of 300s is used to define a "hung" zio.
* The ztest deadman thread has been re-enabled by default,
aligned with the upstream OpenZFS code, and then extended
to terminate the process when it takes significantly longer
to complete than expected.
* The -G option was added to ztest to print the internal debug
log when a fatal error is encountered. This same option was
previously added to zdb in commit fa603f82. Update zloop.sh
to unconditionally pass -G to obtain additional debugging.
* The FM_EREPORT_ZFS_DELAY event which was previously posted
when the deadman detect a "hung" pool has been replaced by
a new dedicated FM_EREPORT_ZFS_DEADMAN event.
* The proposed recovery logic attempts to restart a "hung"
zio by calling zio_interrupt() on any outstanding leaf zios.
We may want to further restrict this to zios in either the
ZIO_STAGE_VDEV_IO_START or ZIO_STAGE_VDEV_IO_DONE stages.
Calling zio_interrupt() is expected to only be useful for
cases when an IO has been submitted to the physical device
but for some reasonable the completion callback hasn't been
called by the lower layers. This shouldn't be possible but
has been observed and may be caused by kernel/driver bugs.
* The 'zfs_deadman_synctime_ms' default value was reduced from
1000s to 600s.
* Depending on how ztest fails there may be no cache file to
move. This should not be considered fatal, collect the logs
which are available and carry on.
* Add deadman test cases for spa_deadman() and zio_wait().
* Increase default zfs_deadman_checktime_ms to 60s.
Reviewed-by: Tim Chase <tim@chase2k.com>
Reviewed by: Thomas Caputi <tcaputi@datto.com>
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#6999
usr/src/uts/common/sys/fs/zfs.h
Change ZPROP_INVAL and ZPROP_CONT from macros to enum values. Clang
and GCC both prefer to use unsigned ints to store enums. That was
causing tautological comparison warnings (and likely eliminating
error handling code at compile time) whenever a zfs_prop_t or
zpool_prop_t was compared to ZPROP_INVAL or ZPROP_CONT. Making the
error flags be explicity enum values forces the enum types to be
signed.
ZPROP_INVAL was also compared against two different enum types. I
had to change its name to ZPOOL_PROP_INVAL whenever its compared to
a zpool_prop_t. There are still some places where ZPROP_INVAL or
ZPROP_CONT is compared to a plain int, in code that doesn't know
whether the int is storing a zfs_prop_t or a zpool_prop_t.
usr/src/uts/common/fs/zfs/spa.c
s/ZPROP_INVAL/ZPOOL_PROP_INVAL/
Authored by: Alan Somers <asomers@gmail.com>
Approved by: Gordon Ross <gwr@nexenta.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Igor Kozhukhov <igor@dilos.org>
Reviewed by: George Melikov <mail@gmelikov.ru>
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
OpenZFS-issue: https://www.illumos.org/issues/8652
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/c2de80dc74Closes#7061
Authored by: Sean Eric Fagan <sef@ixsystems.com>
Reviewed by: Alek Pinchuk <pinchuk.alek@gmail.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Gordon Ross <gwr@nexenta.com>
Ported-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Porting Notes:
- Brought #defines in eventdefs.h in line with ZFS on Linux format.
- Updated zfs-events.5 with the new events.
OpenZFS-issue: https://www.illumos.org/issues/8959
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/c862b93eeaCloses#7049
For ztest, which is solely for testing, using a pseudo random
is entirely reasonable. Using /dev/urandom ensures the system
entropy pool doesn't get depleted thus stalling the testing.
This is a particular problem when testing in VMs.
Reviewed-by: Tim Chase <tim@chase2k.com>
Reviewed by: Thomas Caputi <tcaputi@datto.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#7017Closes#7036
As part of the refactoring of ab9f4b0b82,
several uint64_t-s and uint8_t-s were changed to other types. This
caused ZoL github issue #6981, an overflow of a size_t on a 32-bit ARM
machine. In absense of any strong motivation for the type changes, this
simply puts them back, modulo the changes accumulated for ABD.
Compile-tested on amd64 and run-tested on armhf.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Nathaniel Wesley Filardo <nwf@cs.jhu.edu>
Closes#6981Closes#7023
When the compressed ARC feature was added in commit d3c2ae1
the method of reference counting in the ARC was modified. As
part of this accounting change the arc_buf_add_ref() function
was removed entirely.
This would have be fine but the arc_buf_add_ref() function
served a second undocumented purpose of updating the ARC access
information when taking a hold on a dbuf. Without this logic
in place a cached dbuf would not migrate its associated
arc_buf_hdr_t to the MFU list. This would negatively impact
the ARC hit rate, particularly on systems with a small ARC.
This change reinstates the missing call to arc_access() from
dbuf_hold() by implementing a new arc_buf_access() function.
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Tim Chase <tim@chase2k.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#6171Closes#6852Closes#6989
Authored by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: John Kennedy <jwk404@gmail.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Brad Lewis <brad.lewis@delphix.com>
Reviewed by: Igor Kozhukhov <igor@dilos.org>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Robert Mustacchi <rm@joyent.com>
Ported-by: Prakash Surya <prakash.surya@delphix.com>
PROBLEM
=======
There's a race condition that exists if `zil_free_lwb` races with either
`zil_commit_waiter_timeout` and/or `zil_lwb_flush_vdevs_done`.
Here's an example panic due to this bug:
> ::status
debugging crash dump vmcore.0 (64-bit) from ip-10-110-205-40
operating system: 5.11 dlpx-5.2.2.0_2017-12-04-17-28-32b6ba51fb (i86pc)
image uuid: 4af0edfb-e58e-6ed8-cafc-d3e9167c7513
panic message:
BAD TRAP: type=e (#pf Page fault) rp=ffffff0010555970 addr=60 occurred in module "zfs" due to a NULL pointer dereference
dump content: kernel pages only
> $c
zio_shrink+0x12()
zil_lwb_write_issue+0x30d(ffffff03dcd15cc0, ffffff03e0730e20)
zil_commit_waiter_timeout+0xa2(ffffff03dcd15cc0, ffffff03d97ffcf8)
zil_commit_waiter+0xf3(ffffff03dcd15cc0, ffffff03d97ffcf8)
zil_commit+0x80(ffffff03dcd15cc0, 9a9)
zfs_write+0xc34(ffffff03dc38b140, ffffff0010555e60, 40, ffffff03e00fb758, 0)
fop_write+0x5b(ffffff03dc38b140, ffffff0010555e60, 40, ffffff03e00fb758, 0)
write+0x250(42, fffffd7ff4832000, 2000)
sys_syscall+0x177()
If there's an outstanding lwb that's in `zil_commit_waiter_timeout`
waiting to timeout, waiting on it's waiter's CV, we must be sure not to
call `zil_free_lwb`. If we end up calling `zil_free_lwb`, then that LWB
may be freed and can result in a use-after-free situation where the
stale lwb pointer stored in the `zil_commit_waiter_t` structure of the
thread waiting on the waiter's CV is used.
A similar situation can occur if an lwb is issued to disk, and thus in
the `LWB_STATE_ISSUED` state, and `zil_free_lwb` is called while the
disk is servicing that lwb. In this situation, the lwb will be freed by
`zil_free_lwb`, which will result in a use-after-free situation when the
lwb's zio completes, and `zil_lwb_flush_vdevs_done` is called.
This race condition is prevented in `zil_close` by calling `zil_commit`
before `zil_free_lwb` is called, which will ensure all outstanding (i.e.
all lwb's in the `LWB_STATE_OPEN` and/or `LWB_STATE_ISSUED` states)
reach the `LWB_STATE_DONE` state before the lwb's are freed
(`zil_commit` will not return untill all the lwb's are
`LWB_STATE_DONE`).
Further, this race condition is prevented in `zil_sync` by only calling
`zil_free_lwb` for lwb's that do not have their `lwb_buf` pointer set.
All lwb's not in the `LWB_STATE_DONE` state will have a non-null value
for this pointer; the pointer is only cleared in
`zil_lwb_flush_vdevs_done`, at which point the lwb's state will be
changed to `LWB_STATE_DONE`.
This race *is* present in `zil_suspend`, leading to this bug.
At first glance, it would appear as though this would not be true
because `zil_suspend` will call `zil_commit`, just like `zil_close`, but
the problem is that `zil_suspend` will set the zilog's `zl_suspend`
field prior to calling `zil_commit`. Further, in `zil_commit`, if
`zl_suspend` is set, `zil_commit` will take a special branch of logic
and use `txg_wait_synced` instead of performing the normal `zil_commit`
logic.
This call to `txg_wait_synced` might be good enough for the data to
reach disk safely before it returns, but it does not ensure that all
outstanding lwb's reach the `LWB_STATE_DONE` state before it returns.
This is because, if there's an lwb "stuck" in
`zil_commit_waiter_timeout`, waiting for it's lwb to timeout, it will
maintain a non-null value for it's `lwb_buf` field and thus `zil_sync`
will not free that lwb. Thus, even though the lwb's data is already on
disk, the lwb will be left lingering, waiting on the CV, and will
eventually timeout and be issued to disk even though the write is
unnecessary.
So, after `zil_commit` is called from `zil_suspend`, we incorrectly
assume that there are not outstanding lwb's, and proceed to free all
lwb's found on the zilog's lwb list. As a result, we free the lwb that
will later be used `zil_commit_waiter_timeout`.
SOLUTION
========
The solution to this, is to ensure all outstanding lwb's complete before
calling `zil_free_lwb` via `zil_destroy` in `zil_suspend`. This patch
accomplishes this goal by forcing the normal `zil_commit` logic when
called from `zil_sync`.
Now, `zil_suspend` will call `zil_commit_impl` which will always use the
normal logic of waiting/issuing lwb's to disk before it returns. As a
result, any lwb's outstanding when `zil_commit_impl` is called will be
guaranteed to reach the `LWB_STATE_DONE` state by the time it returns.
Further, no new lwb's will be created via `zil_commit` since the zilog's
`zl_suspend` flag will be set. This will force all new callers of
`zil_commit` to use `txg_wait_synced` instead of creating and issuing
new lwb's.
Thus, all lwb's left on the zilog's lwb list when `zil_destroy` is
called will be in the `LWB_STATE_DONE` state, and we'll avoid this race
condition.
OpenZFS-issue: https://www.illumos.org/issues/8909
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/ece62b6f8dCloses#6940
