with aclmode=passthrough
Authored by: Albert Lee <trisk@nexenta.com>
Reviewed by: Gordon Ross <gwr@nexenta.com>
Reviewed by: Yuri Pankov <yuri.pankov@nexenta.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Richard Lowe <richlowe@richlowe.net>
Ported-by: Paul B. Henson <henson@acm.org>
OpenZFS-issue: https://www.illumos.org/issues/6764
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/de0f1ddb59Closes#10266
Authored-by: Paul B. Henson <henson@acm.org>
Reviewed by: Albert Lee <trisk@nexenta.com>
Reviewed by: Gordon Ross <gwr@nexenta.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Richard Lowe <richlowe@richlowe.net>
Ported-by: Paul B. Henson <henson@acm.org>
OpenZFS-issue: https://www.illumos.org/issues/3254
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/71dbfc287cCloses#10266
Porting notes:
* Updated zfs_acl_chmod to take 'boolean_t isdir' as first parameter
rather than 'zfsvfs_t *zfsvfs'
* zfs man pages changes mixed between zfs and new zfsprops man pages
Reviewed by: Aram Hvrneanu <aram@nexenta.com>
Reviewed by: Gordon Ross <gwr@nexenta.com>
Reviewed by: Robert Gordon <rbg@openrbg.com>
Reviewed by: Mark.Maybee@oracle.com
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Garrett D'Amore <garrett@nexenta.com>
Ported-by: Paul B. Henson <henson@acm.org>
OpenZFS-issue: https://www.illumos.org/issues/742
OpenZFS-issue: https://www.illumos.org/issues/664
OpenZFS-issue: https://www.illumos.org/issues/279
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/a3c49ce110Closes#10266
When a Thumb-2 kernel is being used, then longjmp must be implemented
using the Thumb-2 instruction set in module/lua/setjmp/setjmp_arm.S.
Original-patch-by: @jsrlabs
Reviewed-by: @awehrfritz
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#7408Closes#9957Closes#9967
zlib_inflateEnd was accidentally a wrapper for inflateInit instead of
inflateEnd, and hilarity ensues.
Fix the typo so we free memory instead of allocating more.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#10225Closes#10252
This patch corrects a bug introduced in 61152d1069. When
resuming a raw base receive, the dmu_recv code always sets
drc->drc_fromsnapobj to the object ID of the previous
snapshot. For incrementals, this is correct, but for base
sends, this should be left at 0. The presence of this ID
eventually allows a check to run which determines whether
or not the incoming stream and the previous snapshot have
matching IVset guids. This check fails becuase it is not
meant to run when there is no previous snapshot. When it
does fail, the user receives an error stating that the
incoming stream has the problem outlined in errata 4.
This patch corrects this issue by simply ensuring
drc->drc_fromsnapobj is left as 0 for base receives.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes#10234Closes#10239
Deduplicated send streams (i.e. `zfs send -D` and `zfs receive` of such
streams) are deprecated. Deduplicated send streams can be received by
first converting them to non-deduplicated with the `zstream redup`
command.
This commit removes the code for sending and receiving deduplicated send
streams. `zfs send -D` will now print a warning, ignore the `-D` flag,
and generate a regular (non-deduplicated) send stream. `zfs receive` of
a deduplicated send stream will print an error message and fail.
The resulting code simplification (especially in the kernel's support
for receiving dedup streams) should help enable future performance
enhancements.
Several new tests are added which leverage `zstream redup`.
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Issue #7887
Issue #10117
Issue #10156Closes#10212
Each metaslab group (of which there is one per top-level vdev) has
several (4, by default) "metaslab group allocators". Each "allocator"
has its own metaslab that it prefers to allocate from (the "primary"
allocator), and each can perform allocations concurrently with the other
allocators. In addition to the primary metaslab, there are several
other fields that need to be tracked separately for each allocator.
These are currently stored as several arrays in the metaslab_group_t,
each array indexed by allocator number.
This change organizes all the metaslab-group-allocator-specific fields
into a new struct, metaslab_group_allocator_t. The metaslab_group_t now
needs only one array indexed by the allocator number - which contains
the metaslab_group_allocator_t's.
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10213
The progress of a send is supposed to be reported by `zfs send -v`, but
it is not. This works by creating a new user thread (with
pthread_create()) which does ZFS_IOC_SEND_PROGRESS ioctls to check how
much progress has been made. This IOCTL finds the specified send (since
there may be multiple concurrent sends in the system). The IOCTL also
checks that the specified send was started by the current process.
On Linux, different threads of the same process are represented as
different `struct task_struct`s (and, confusingly, have different
PID's). To check if if two threads are in the same process, we need to
check if they have the same `struct task_struct:group_leader`.
We used to to this correctly, but it was inadvertently changed by
30af21b025 (Redacted Send) to simply check if the current
`struct task_struct` is the one that started the send.
This commit changes the code back to checking if the send was started by
a `struct task_struct` with the same `group_leader` as the calling
thread.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Chris Wedgwood <cw@f00f.org>
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10215Closes#10216
Propagate changes in HEAD that mostly eliminate object locking.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#10205
Minor fixes on persistent L2ARC improving code readability and fixing
a typo in zdb.c when byte-swapping a log block. It also improves the
pesist_l2arc_007_pos.ksh test by giving it more time to retrieve log
blocks on the cache device.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Adam D. Moss <c@yotes.com>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes#10210
Remove some obsolete legacy compat, rename some misnamed, and add some
missing tunables for FreeBSD.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#10203
Add the FreeBSD platform code to the OpenZFS repository. As of this
commit the source can be compiled and tested on FreeBSD 11 and 12.
Subsequent commits are now required to compile on FreeBSD and Linux.
Additionally, they must pass the ZFS Test Suite on FreeBSD which is
being run by the CI. As of this commit 1230 tests pass on FreeBSD
and there are no unexpected failures.
Reviewed-by: Sean Eric Fagan <sef@ixsystems.com>
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Reviewed-by: Richard Laager <rlaager@wiktel.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#898Closes#8987
The memory and cpu cost of reference count tracking with the current
implementation is significant. For this reason it has always been
disabled by default for the kmods. Apply this same default to user
space so ztest doesn't always incur this performance penalty.
Our intention is to re-enable this by default for ztest once the code
has been optimized. Since we expect to at some point provide a FUSE
implementation we wouldn't want this enabled by default for libzpool.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#10189
We can improve the performance of writes to zvols by using
dmu_tx_hold_write_by_dnode() instead of dmu_tx_hold_write(). This
reduces lock contention on the first block of the dnode object, and also
reduces the amount of CPU needed. The benefit will be highest with
multi-threaded async writes (i.e. writes that don't call zil_commit()).
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10184
This commit makes the L2ARC persistent across reboots. We implement
a light-weight persistent L2ARC metadata structure that allows L2ARC
contents to be recovered after a reboot. This significantly eases the
impact a reboot has on read performance on systems with large caches.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: George Wilson <gwilson@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Saso Kiselkov <skiselkov@gmail.com>
Co-authored-by: Jorgen Lundman <lundman@lundman.net>
Co-authored-by: George Amanakis <gamanakis@gmail.com>
Ported-by: Yuxuan Shui <yshuiv7@gmail.com>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes#925Closes#1823Closes#2672Closes#3744Closes#9582
Set arc_c_min before arc_c_max so that when zfs_arc_min is set lower
than the default allmem/32 zfs_arc_max can also be set lower.
Add warning messages when tunables are being ignored.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#10157Closes#10158
By default it's not possible to open a device already owned by an
active vdev. It's necessary to make an exception to this for vdev
split. The FreeBSD platform code will make an exception if
spa_is splitting is set to to true.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#10178
Commit https://github.com/torvalds/linux/commit/3d745ea5 simplified
the blk_alloc_queue() interface by updating it to take the request
queue as an argument. Add a wrapper function which accepts the new
arguments and internally uses the available interfaces.
Other minor changes include increasing the Linux-Maximum to 5.6 now
that 5.6 has been released. It was not bumped to 5.7 because this
release has not yet been finalized and is still subject to change.
Added local 'struct zvol_state_os *zso' variable to zvol_alloc.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#10181Closes#10187
Added to prevent a possible deadlock, the following comments from
FreeBSD explain the issue. The comment describing vn_io_fault_uiomove:
/*
* Helper function to perform the requested uiomove operation using
* the held pages for io->uio_iov[0].iov_base buffer instead of
* copyin/copyout. Access to the pages with uiomove_fromphys()
* instead of iov_base prevents page faults that could occur due to
* pmap_collect() invalidating the mapping created by
* vm_fault_quick_hold_pages(), or pageout daemon, page laundry or
* object cleanup revoking the write access from page mappings.
*
* Filesystems specified MNTK_NO_IOPF shall use vn_io_fault_uiomove()
* instead of plain uiomove().
*/
This used for vn_io_fault which has the following motivation:
/*
* The vn_io_fault() is a wrapper around vn_read() and vn_write() to
* prevent the following deadlock:
*
* Assume that the thread A reads from the vnode vp1 into userspace
* buffer buf1 backed by the pages of vnode vp2. If a page in buf1 is
* currently not resident, then system ends up with the call chain
* vn_read() -> VOP_READ(vp1) -> uiomove() -> [Page Fault] ->
* vm_fault(buf1) -> vnode_pager_getpages(vp2) -> VOP_GETPAGES(vp2)
* which establishes lock order vp1->vn_lock, then vp2->vn_lock.
* If, at the same time, thread B reads from vnode vp2 into buffer buf2
* backed by the pages of vnode vp1, and some page in buf2 is not
* resident, we get a reversed order vp2->vn_lock, then vp1->vn_lock.
*
* To prevent the lock order reversal and deadlock, vn_io_fault() does
* not allow page faults to happen during VOP_READ() or VOP_WRITE().
* Instead, it first tries to do the whole range i/o with pagefaults
* disabled. If all pages in the i/o buffer are resident and mapped,
* VOP will succeed (ignoring the genuine filesystem errors).
* Otherwise, we get back EFAULT, and vn_io_fault() falls back to do
* i/o in chunks, with all pages in the chunk prefaulted and held
* using vm_fault_quick_hold_pages().
*
* Filesystems using this deadlock avoidance scheme should use the
* array of the held pages from uio, saved in the curthread->td_ma,
* instead of doing uiomove(). A helper function
* vn_io_fault_uiomove() converts uiomove request into
* uiomove_fromphys() over td_ma array.
*
* Since vnode locks do not cover the whole i/o anymore, rangelocks
* make the current i/o request atomic with respect to other i/os and
* truncations.
*/
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#10177
Linux and FreeBSD have different parameters for tunable proc handler.
This has prevented FreeBSD from implementing the ZFS_MODULE_PARAM_CALL
macro.
To complete the sharing of ZFS_MODULE_PARAM_CALL declarations, create
per-platform definitions of the parameter list, ZFS_MODULE_PARAM_ARGS.
With the declarations wired up we discovered an incorrect scope prefix
for spa_slop_shift, so this is now fixed.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#10179
Add a mechanism to wait for delete queue to drain.
When doing redacted send/recv, many workflows involve deleting files
that contain sensitive data. Because of the way zfs handles file
deletions, snapshots taken quickly after a rm operation can sometimes
still contain the file in question, especially if the file is very
large. This can result in issues for redacted send/recv users who
expect the deleted files to be redacted in the send streams, and not
appear in their clones.
This change duplicates much of the zpool wait related logic into a
zfs wait command, which can be used to wait until the internal
deleteq has been drained. Additional wait activities may be added
in the future.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Gallagher <john.gallagher@delphix.com>
Signed-off-by: Paul Dagnelie <pcd@delphix.com>
Closes#9707
In zfs_write(), the loop continues to the next iteration without
accounting for partial copies occurring in uiomove_iov when
copy_from_user/__copy_from_user_inatomic return a non-zero status.
This results in "zfs: accessing past end of object..." in the
kernel log, and the write failing.
Account for partial copies and update uio struct before returning
EFAULT, leave a comment explaining the reason why this is done.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: ilbsmart <wgqimut@gmail.com>
Signed-off-by: Fabio Scaccabarozzi <fsvm88@gmail.com>
Closes#8673Closes#10148
== Summary ==
Prior to this change, sync writes to a zvol are processed serially.
This commit makes zvols process concurrently outstanding sync writes in
parallel, similar to how reads and async writes are already handled.
The result is that the throughput of sync writes is tripled.
== Background ==
When a write comes in for a zvol (e.g. over iscsi), it is processed by
calling `zvol_request()` to initiate the operation. ZFS is expected to
later call `BIO_END_IO()` when the operation completes (possibly from a
different thread). There are a limited number of threads that are
available to call `zvol_request()` - one one per iscsi client (unless
using MC/S). Therefore, to ensure good performance, the latency of
`zvol_request()` is important, so that many i/o operations to the zvol
can be processed concurrently. In other words, if the client has
multiple outstanding requests to the zvol, the zvol should have multiple
outstanding requests to the storage hardware (i.e. issue multiple
concurrent `zio_t`'s).
For reads, and async writes (i.e. writes which can be acknowledged
before the data reaches stable storage), `zvol_request()` achieves low
latency by dispatching the bulk of the work (including waiting for i/o
to disk) to a taskq. The taskq callback (`zvol_read()` or
`zvol_write()`) blocks while waiting for the i/o to disk to complete.
The `zvol_taskq` has 32 threads (by default), so we can have up to 32
concurrent i/os to disk in service of requests to zvols.
However, for sync writes (i.e. writes which must be persisted to stable
storage before they can be acknowledged, by calling `zil_commit()`),
`zvol_request()` does not use `zvol_taskq`. Instead it blocks while
waiting for the ZIL write to disk to complete. This has the effect of
serializing sync writes to each zvol. In other words, each zvol will
only process one sync write at a time, waiting for it to be written to
the ZIL before accepting the next request.
The same issue applies to FLUSH operations, for which `zvol_request()`
calls `zil_commit()` directly.
== Description of change ==
This commit changes `zvol_request()` to use
`taskq_dispatch_ent(zvol_taskq)` for sync writes, and FLUSh operations.
Therefore we can have up to 32 threads (the taskq threads)
simultaneously calling `zil_commit()`, for a theoretical performance
improvement of up to 32x.
To avoid the locking issue described in the comment (which this commit
removes), we acquire the rangelock from the taskq callback (e.g.
`zvol_write()`) rather than from `zvol_request()`. This applies to all
writes (sync and async), reads, and discard operations. This means that
multiple simultaneously-outstanding i/o's which access the same block
can complete in any order. This was previously thought to be incorrect,
but a review of the block device interface requirements revealed that
this is fine - the order is inherently not defined. The shorter hold
time of the rangelock should also have a slight performance improvement.
For an additional slight performance improvement, we use
`taskq_dispatch_ent()` instead of `taskq_dispatch()`, which avoids a
`kmem_alloc()` and eliminates a failure mode. This applies to all
writes (sync and async), reads, and discard operations.
== Performance results ==
We used a zvol as an iscsi target (server) for a Windows initiator
(client), with a single connection (the default - i.e. not MC/S).
We used `diskspd` to generate a workload with 4 threads, doing 1MB
writes to random offsets in the zvol. Without this change we get
231MB/s, and with the change we get 728MB/s, which is 3.15x the original
performance.
We ran a real-world workload, restoring a MSSQL database, and saw
throughput 2.5x the original.
We saw more modest performance wins (typically 1.5x-2x) when using MC/S
with 4 connections, and with different number of client threads (1, 8,
32).
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Reviewed-by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10163
Increasing l2arc_write_size or l2arc_write_boost can result in
l2arc_write_buffers() not having enough space to perform its writes and
panic zio_write_phys().
Instead of resetting l2ad_hand to l2ad_start at the end of
l2arc_write_buffers() and not taking into account a possible
user-mediated increase of l2arc_write_max, we do this in l2arc_evict(),
right after l2arc_write_size() has run. If there is not enough space to
evict (ie we will exceed l2ad_end) we evict to the end of the device,
reset l2ad_hand to l2ad_start, set l2ad_first to 0 and iterate
l2arc_evict(). We avoid infinite iteration of l2arc_evict() by making
sure in l2arc_write_size() that l2ad_start + size does not exceed
l2ad_end.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes#10154
Linux changed the default max ARC size to 1/2 of physical memory to
deal with shortcomings of the Linux SLUB allocator. Other platforms
do not require the same logic.
Implement an arc_default_max() function to determine a default max ARC
size in platform code.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#10155
Make the cityhash code compile into libzfs, in preparation for the new
"zstream" command.
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10152
These paths are never exercised, as the parameters given are always
different cipher and plaintext `crypto_data_t` pointers.
Reviewed-by: Richard Laager <rlaager@wiktel.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Attila Fueloep <attila@fueloep.org>
Signed-off-by: Dirkjan Bussink <d.bussink@gmail.com>
Closes#9661Closes#10015
If a has rollback has occurred while a file is open and unlinked.
Then when the file is closed post rollback it will not exist in the
rolled back version of the unlinked object. Therefore, the call to
zap_remove_int() may correctly return ENOENT and should be allowed.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#6812Closes#9739
This change adds a separate return code to zfs_ioc_recv that is used
for incomplete streams, in addition to the existing return code for
streams that contain corruption.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Paul Dagnelie <pcd@delphix.com>
Closes#10122
There are a couple of x86_64 architectures which support all needed
features to make the accelerated GCM implementation work but the
MOVBE instruction. Those are mainly Intel Sandy- and Ivy-Bridge
and AMD Bulldozer, Piledriver, and Steamroller.
By using MOVBE only if available and replacing it with a MOV
followed by a BSWAP if not, those architectures now benefit from
the new GCM routines and performance is considerably better
compared to the original implementation.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Adam D. Moss <c@yotes.com>
Signed-off-by: Attila Fülöp <attila@fueloep.org>
Followup #9749Closes#10029
For each WRITE record in the stream, `zfs receive` creates a DMU
transaction (`dmu_tx_create()`) and writes this block's data into the
object. If per-block overheads (as opposed to per-byte overheads)
dominate performance (as is often the case with small recordsize), the
per-dmu-transaction overheads can be significant. For example, in some
workloads the `receieve_writer` thread is 100% on CPU, and more than
half of its CPU time is in these per-tx routines (e.g.
dmu_tx_hold_write, dmu_tx_assign, dmu_tx_commit).
To improve performance of `zfs receive`, this commit batches WRITE
records which are to nearby offsets of the same object, and uses one DMU
transaction to write them all. By default the batch size is 1MB, which
for recordsize=8K reduces the number of DMU transactions by 128x for
full send streams (incrementals will depend on how "clumpy" the changed
blocks are).
This commit improves the performance of `dd if=stream | zfs recv`
from 78,800 blocks/sec to 98,100 blocks/sec (25% improvement).
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10099
The normal lock order is that the dp_config_rwlock must be held before
the ds_opening_lock. For example, dmu_objset_hold() does this.
However, dmu_objset_open_impl() is called with the ds_opening_lock held,
and if the dp_config_rwlock is not already held, it will attempt to
acquire it. This may lead to deadlock, since the lock order is
reversed.
Looking at all the callers of dmu_objset_open_impl() (which is
principally the callers of dmu_objset_from_ds()), almost all callers
already have the dp_config_rwlock. However, there are a few places in
the send and receive code paths that do not. For example:
dsl_crypto_populate_key_nvlist, send_cb, dmu_recv_stream,
receive_write_byref, redact_traverse_thread.
This commit resolves the problem by requiring all callers ot
dmu_objset_from_ds() to hold the dp_config_rwlock. In most cases, the
code has been restructured such that we call dmu_objset_from_ds()
earlier on in the send and receive processes, when we already have the
dp_config_rwlock, and save the objset_t until we need it in the middle
of the send or receive (similar to what we already do with the
dsl_dataset_t). Thus we do not need to acquire the dp_config_rwlock in
many new places.
I also cleaned up code in dmu_redact_snap() and send_traverse_thread().
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Paul Zuchowski <pzuchowski@datto.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#9662Closes#10115
Attempt to run scrub or resilver on a new pool containing only special
allocations (special vdev added on creation) caused infinite loop
because of dsl_scan_should_clear() limiting memory usage to 5% of pool
size, which it calculated accounting only normal allocation class.
Addition of special and just in case dedup classes fixes the issue.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#10106Closes#8694
dnode_special_close() waits for the refcount of dn_holds to go to zero
without holding the dn_mtx. dnode_rele_and_unlock() does the final
remove to dn_holds with dn_mtx being held:
refs = zfs_refcount_remove(&dn->dn_holds, tag);
mutex_exit(&dn->dn_mtx);
So, there is a race condition after the remove until dn_mtx is
dropped. During that time, dnode_destroy() can get called, which ends
up in dnode_dest() calling mutex_destroy() and a panic since the lock
is still held.
This change adds a condvar to wait for the final dnode_rele_and_unlock()
to release the dn_mtx before calling dnode_destroy().
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: John Poduska <jpoduska@datto.com>
Closes#7814Closes#10101
Using zfs with Lustre, an arc_read can trigger kernel memory allocation
that in turn leads to a memory reclaim callback and a deadlock within a
single zfs process. This change uses spl_fstrans_mark and
spl_trans_unmark to prevent the reclaim attempt and the deadlock
(https://zfsonlinux.topicbox.com/groups/zfs-devel/T4db2c705ec1804ba).
The stack trace observed is:
__schedule at ffffffff81610f2e
schedule at ffffffff81611558
schedule_preempt_disabled at ffffffff8161184a
__mutex_lock at ffffffff816131e8
arc_buf_destroy at ffffffffa0bf37d7 [zfs]
dbuf_destroy at ffffffffa0bfa6fe [zfs]
dbuf_evict_one at ffffffffa0bfaa96 [zfs]
dbuf_rele_and_unlock at ffffffffa0bfa561 [zfs]
dbuf_rele_and_unlock at ffffffffa0bfa32b [zfs]
osd_object_delete at ffffffffa0b64ecc [osd_zfs]
lu_object_free at ffffffffa06d6a74 [obdclass]
lu_site_purge_objects at ffffffffa06d7fc1 [obdclass]
lu_cache_shrink_scan at ffffffffa06d81b8 [obdclass]
shrink_slab at ffffffff811ca9d8
shrink_node at ffffffff811cfd94
do_try_to_free_pages at ffffffff811cfe63
try_to_free_pages at ffffffff811d01c4
__alloc_pages_slowpath at ffffffff811be7f2
__alloc_pages_nodemask at ffffffff811bf3ed
new_slab at ffffffff81226304
___slab_alloc at ffffffff812272ab
__slab_alloc at ffffffff8122740c
kmem_cache_alloc at ffffffff81227578
spl_kmem_cache_alloc at ffffffffa048a1fd [spl]
arc_buf_alloc_impl at ffffffffa0befba2 [zfs]
arc_read at ffffffffa0bf0924 [zfs]
dbuf_read at ffffffffa0bf9083 [zfs]
dmu_buf_hold_by_dnode at ffffffffa0c04869 [zfs]
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Mark Roper <markroper@gmail.com>
Closes#9987
When doing a zfs send on a dataset with small recordsize (e.g. 8K),
performance is dominated by the per-block overheads. This is especially
true with `zfs send --compressed`, which further reduces the amount of
data sent, for the same number of blocks. Several threads are involved,
but the limiting factor is the `send_prefetch` thread, which is 100% on
CPU.
The main job of the `send_prefetch` thread is to issue zio's for the
data that will be needed by the main thread. It does this by calling
`arc_read(ARC_FLAG_PREFETCH)`. This has an immediate cost of creating
an arc_hdr, which takes around 14% of one CPU. It also induces later
costs by other threads:
* Since the data was only prefetched, dmu_send()->dmu_dump_write() will
need to call arc_read() again to get the data. This will have to
look up the arc_hdr in the hash table and copy the data from the
scatter ABD in the arc_hdr to a linear ABD in arc_buf. This takes
27% of one CPU.
* dmu_dump_write() needs to arc_buf_destroy() This takes 11% of one
CPU.
* arc_adjust() will need to evict this arc_hdr, taking about 50% of one
CPU.
All of these costs can be avoided by bypassing the ARC if the data is
not already cached. This commit changes `zfs send` to check for the
data in the ARC, and if it is not found then we directly call
`zio_read()`, reading the data into a linear ABD which is used by
dmu_dump_write() directly.
The performance improvement is best expressed in terms of how many
blocks can be processed by `zfs send` in one second. This change
increases the metric by 50%, from ~100,000 to ~150,000. When the amount
of data per block is small (e.g. 2KB), there is a corresponding
reduction in the elapsed time of `zfs send >/dev/null` (from 86 minutes
to 58 minutes in this test case).
In addition to improving the performance of `zfs send`, this change
makes `zfs send` not pollute the ARC cache. In most cases the data will
not be reused, so this allows us to keep caching useful data in the MRU
(hit-once) part of the ARC.
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10067
Filesystems allow overlay mounts by default on FreeBSD and Linux.
Respect the native convention by switching the default to overlay=on,
while retaining the option to turn the property off for compatibility
with other operating systems' conventions.
Update documentation and tests accordingly.
Reviewed-by: Richard Laager <rlaager@wiktel.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#10030
Also dprintf_bp() in case BLK_VERIFY_HALT of zfs_blkptr_verify_log()
since dprintf_bp() in zfs_blkptr_verify() will never be executed.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Paul Zuchowski <pzuchowski@datto.com>
Signed-off-by: Justin Keogh <commits@v6y.net>
Closes#10086
Manual trims fall into the category of long-running pool activities
which people might want to wait synchronously for. This change adds
support to 'zpool wait' for waiting for manual trim operations to
complete. It also adds a '-w' flag to 'zpool trim' which can be used to
turn 'zpool trim' into a synchronous operation.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Signed-off-by: John Gallagher <john.gallagher@delphix.com>
Closes#10071
__zio_execute() calls zio_taskq_member() to determine if we are running
in a zio interrupt taskq, in which case we may need to switch to
processing this zio in a zio issue taskq. The call to
zio_taskq_member() can become a performance bottleneck when we are
processing a high rate of zio's.
zio_taskq_member() calls taskq_member() on each of the zio interrupt
taskqs, of which there are 21. This is slow because each call to
taskq_member() does tsd_get(taskq_tsd), which on Linux is relatively
slow.
This commit improves the performance of zio_taskq_member() by having it
cache the value of tsd_get(taskq_tsd), reducing the number of those
calls to 1/21th of the current behavior.
In a test case running `zfs send -c >/dev/null` of a filesystem with
small blocks (average 2.5KB/block), zio_taskq_member() was using 6.7% of
one CPU, and with this change it is reduced to 1.3%. Overall time to
perform the `zfs send` reduced by 10% (~150,000 block/sec to ~165,000
blocks/sec).
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10070
This function should only return "linux" on Linux.
Move the kernel part of the function out of common code.
Fix the tests for FreeBSD.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#10079
FreeBSD has a somewhat more cumbersome locking and refcounting
protocol for the platform counterpart to znode. We need to not call
zrele on the passed zp, but do need to do so on any intermediate zp.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#10075
By adding a zfs_file_private accessor to the common
interfaces and some extensions to FreeBSD platform
code it is now possible to share the implementations
for the aforementioned functions.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#10073
When "zfs destroy" is run, it completes quickly, and in the background
we locate the blocks to free and free them. This background activity
can be observed with `zpool get freeing` and `zpool wait -t free ...`.
This background activity is processed by a single thread (the spa_sync
thread) which calls zio_free() on each of the blocks to free. With even
modest storage performance, the CPU consumption of zio_free() can be the
performance bottleneck.
Performance of zio_free() can be improved by not actually creating a
zio_t in the common case (non-dedup, non-gang), instead calling
metaslab_free() directly. This avoids the CPU cost of allocating the
zio_t, and more importantly the cost of adding and later removing this
zio_t from the parent zio's child list.
The result is that performance of background freeing more than doubles,
from 0.6 million blocks per second to 1.3 million blocks per second.
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Wilson <gwilson@delphix.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#10034
Commit https://github.com/torvalds/linux/commit/9e8d42a0f accidentally
converted the static inline function blkg_tryget() to GPL-only for
kernels built with CONFIG_PREEMPT_RCU=y and CONFIG_BLK_CGROUP=y.
Resolve the build issue by providing our own equivalent functionality
when needed which uses rcu_read_lock_sched() internally as before.
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#9745Closes#10072
The following check currently occurs in three separate locations
in dbuf.c. This change consolidates those checks in to the
dbuf_alloc_arcbuf_from_arcbuf() function.
if (arc_is_encrypted(data)) {
...
} else if (compress_type != ZIO_COMPRESS_OFF) {
...
} else {
...
}
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#10057
As part of the Linux kernel's y2038 changes the time_t type has been
fully retired. Callers are now required to use the time64_t type.
Rather than move to the new type, I've removed the few remaining
places where a time_t is used in the kernel code. They've been
replaced with a uint64_t which is already how ZFS internally
handled these values.
Going forward we should work towards updating the remaining user
space time_t consumers to the 64-bit interfaces.
Reviewed-by: Matthew Macy <mmacy@freebsd.org>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#10052Closes#10064
Paul B. Henson
Brian Behlendorf
George Amanakis
Ryan Moeller
Tom Caputi
Matthew Ahrens
Matthew Macy
Paul Dagnelie
Fabio Scaccabarozzi
Dirkjan Bussink
Attila Fülöp
Alexander Motin
John Poduska
Mark Roper