Olaf Faaland [Thu, 15 Oct 2015 20:08:27 +0000 (13:08 -0700)]
Identify locks flagged by lockdep
When running a kernel with CONFIG_LOCKDEP=y, lockdep reports possible
recursive locking in some cases and possible circular locking dependency
in others, within the SPL and ZFS modules.
This patch uses a mutex type defined in SPL, MUTEX_NOLOCKDEP, to mark
such mutexes when they are initialized. This mutex type causes
attempts to take or release those locks to be wrapped in lockdep_off()
and lockdep_on() calls to silence the dependency checker and allow the
use of lock_stats to examine contention.
For RW locks, it uses an analogous lock type, RW_NOLOCKDEP.
The goal is that these locks are ultimately changed back to type
MUTEX_DEFAULT or RW_DEFAULT, after the locks are annotated to reflect
their relationship (e.g. z_name_lock below) or any real problem with the
lock dependencies are fixed.
Some of the affected locks are:
tc_open_lock:
=============
This is an array of locks, all with same name, which txg_quiesce must
take all of in order to move txg to next state. All default to the same
lockdep class, and so to lockdep appears recursive.
zp->z_name_lock:
================
In zfs_rmdir,
dzp = znode for the directory (input to zfs_dirent_lock)
zp = znode for the entry being removed (output of zfs_dirent_lock)
zfs_rmdir()->zfs_dirent_lock() takes z_name_lock in dzp
zfs_rmdir() takes z_name_lock in zp
Since both dzp and zp are type znode_t, the locks have the same default
class, and lockdep considers it a possible recursive lock attempt.
l->l_rwlock:
============
zap_expand_leaf() sometimes creates two new zap leaf structures, via
these call paths:
Because both zap_leaf_open() and zap_create_leaf() initialize
l->l_rwlock in their (separate) leaf structures, the lockdep class is
the same, and the linux kernel believes these might both be the same
lock, and emits a possible recursive lock warning.
Signed-off-by: Olaf Faaland <faaland1@llnl.gov> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3895
Activate LVM volume groups before looking for zpools.
Original-patch-by: @jgoerzen Signed-off-by: Benjamin Albrecht <git@albrecht.io> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes zfsonlinux/pkg-zfs#102
Closes #4029
Brian Behlendorf [Thu, 17 Dec 2015 17:26:05 +0000 (09:26 -0800)]
Fix zfs_vdev_aggregation_limit bounds checking
Update the bounds checking for zfs_vdev_aggregation_limit so that
it has a floor of zero and a maximum value of the supported block
size for the pool.
Additionally add an early return when zfs_vdev_aggregation_limit
equals zero to disable aggregation. For very fast solid state or
memory devices it may be more expensive to perform the aggregation
than to issue the IO immediately.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Wed, 16 Dec 2015 19:28:15 +0000 (11:28 -0800)]
Fix vdev_queue_aggregate() deadlock
This deadlock may manifest itself in slightly different ways but
at the core it is caused by a memory allocation blocking on file-
system reclaim in the zio pipeline. This is normally impossible
because zio_execute() disables filesystem reclaim by setting
PF_FSTRANS on the thread. However, kmem cache allocations may
still indirectly block on file system reclaim while holding the
critical vq->vq_lock as shown below.
To resolve this issue zio_buf_alloc_flags() is introduced which
allocation flags to be passed. This can then be used in
vdev_queue_aggregate() with KM_NOSLEEP when allocating the
aggregate IO buffer. Since aggregating the IO is purely a
performance optimization we want this to either succeed or fail
quickly. Trying too hard to allocate this memory under the
vq->vq_lock can negatively impact performance and result in
this deadlock.
* z_wr_iss
zio_vdev_io_start
vdev_queue_io -> Takes vq->vq_lock
vdev_queue_io_to_issue
vdev_queue_aggregate
zio_buf_alloc -> Waiting on spl_kmem_cache process
* z_wr_int
zio_vdev_io_done
vdev_queue_io_done
mutex_lock -> Waiting on vq->vq_lock held by z_wr_iss
* txg_sync
spa_sync
dsl_pool_sync
zio_wait -> Waiting on zio being handled by z_wr_int
Brian Behlendorf [Wed, 16 Dec 2015 22:17:49 +0000 (14:17 -0800)]
Fix z_xattr_lock/z_teardown_lock lock inversion
There exists a lock inversion between the z_xattr_lock and the
z_teardown_lock. Detect this case and return EBUSY so zfs_resume_fs()
will mark the inode stale and it can be safely revalidated on next
access.
Chunwei Chen [Tue, 8 Dec 2015 20:26:18 +0000 (12:26 -0800)]
Fix uio_prefaultpages for 0 length iovec
Userspace can freely pass in whatever iovec it feels like, and it's perfectly
legal to pass an iovec which contains a zero length segment. In the current
implementation, uio_prefaultpages would touch an out of bound byte in the
"last byte" logic. While this probably wouldn't cause any critical error, we
would like uio_prefaultpages to be able to continue gracefully.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4078
Brian Behlendorf [Fri, 11 Dec 2015 19:09:41 +0000 (11:09 -0800)]
Handle damaged blk_birth in dsl_deadlist_insert()
If a bit were cleared in `bp->blk_birth` such that the txg birth
was now lower than any other txg_birth in the deadlist, then there
will be no entry before this in the tree.
This should be impossible but regardless error handling code has
been added for this case. By default this is left as a fatal case
and the blk_birth is logged. However, setting `zfs_recover=1` will
cause the bp to be placed at the start of the deadlist even though
it contains an invalid blk_birth.
Commit 5f6d0b6 was originally added to gracefully handle block
pointers with a damaged logical size. However, it incorrectly
assumed that all passed arc_done_func_t could handle a NULL
arc_buf_t.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4069
Closes #4080
Richard Yao [Fri, 11 Dec 2015 23:40:05 +0000 (18:40 -0500)]
Unconditionally build zdb and ztest with -DDEBUG
Illumos unconditionally builds zdb and ztest with -DDEBUG. This helps
catch bugs and eliminates the need for commits like 202619623022722f30c2ee49931a4fa6896421c7, which changed ASSERTs to
VERIFYs. The following files in the illumos tree show this:
Brian Behlendorf [Mon, 14 Dec 2015 18:59:25 +0000 (10:59 -0800)]
Hold the zfs_snapentry_t before dispatch
While exceptionally unlikely to cause a problem the zfs_snapentry_t
hold should be taken before the dispatch to prevent any possibility
of the task being processed before the hold.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Chunwei Chen [Fri, 11 Dec 2015 23:24:34 +0000 (15:24 -0800)]
Fix snapshot automount race cause EREMOTE
When a concorrent mount finishes just before calling to
zfsctl_snapshot_ismounted, if we return EISDIR, the VFS will return
with EREMOTE. We should instead just return 0, so VFS may retry and
would likely notice the dentry is alreadly mounted. This will be
inline with when usermode helper return EBUSY.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Thu, 10 Dec 2015 23:53:37 +0000 (15:53 -0800)]
Change zfs_snapshot_lock from mutex to rw lock
By changing the zfs_snapshot_lock from a mutex to a rw lock the
zfsctl_lookup_objset() function can be allowed to run concurrently.
This should reduce the latency of fh_to_dentry lookups in ZFS
snapshots which are being accessed over NFS.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Brian Behlendorf [Thu, 10 Dec 2015 23:47:18 +0000 (15:47 -0800)]
Fix zfsctl_lookup_objset() deadlock
The zfsctl_snapshot_unmount_delay() function must not be called
from zfsctl_lookup_objset() while it is currently holding the
zfs_snapshot_lock. This will result in a deadlock. It is safe
to call zfsctl_snapshot_unmount_delay_impl() directly because the
function already has a reference on the zfs_snapentry_t.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Closes #3997
Brian Behlendorf [Thu, 10 Dec 2015 23:23:26 +0000 (15:23 -0800)]
Set 'zfs_expire_snapshot=0' to disable auto-unmount
There are cases where it's desirable that auto-mounted snapshots
not expire after a fixed duration. They should be unmounted only
when the filesystem they are a snapshot of is unmounted.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
For some arm, powerpc, and sparc platforms it was possible that
neither _ILP32 of _LP64 would be defined. Update the isa_defs.h
header to explicitly set these macros and generate a compile error
in the case neither are defined.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Chunwei Chen <tuxoko@gmail.com>
Closes #4048
Chunwei Chen [Mon, 7 Dec 2015 23:43:53 +0000 (15:43 -0800)]
Use spa as key besides objsetid for snapentry
objsetid is not unique across pool, so using it solely as key would cause
panic when automounting two snapshot on different pools with the same
objsetid. We fix this by adding spa pointer as additional key.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Richard Yao <ryao@gentoo.org>
Issue #3948
Issue #3786
Issue #3887
While stack size will vary by architecture it has historically defaulted to
8K on x86_64 systems. However, as of Linux 3.15 the default thread stack
size was increased to 16K. These kernels are now the default in most non-
enterprise distributions which means we no longer need to assume 8K stacks.
This patch takes advantage of that fact by appropriately reverting stack
conservation changes which were made to ensure stability. Changes which
may have had a negative impact on performance for certain workloads. This
also has the side effect of bringing the code slightly more in line with
upstream.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Richard Yao <ryao@gentoo.org>
Closes #4059
Matthew Ahrens [Fri, 24 Jul 2015 16:53:55 +0000 (09:53 -0700)]
Illumos 5959 - clean up per-dataset feature count code
5959 clean up per-dataset feature count code
Reviewed by: Toomas Soome <tsoome@me.com>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Alex Reece <alex@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
illumos code doesn't check for feature_get_refcount() returning
ENOTSUP (which means feature is disabled) in zdb. zfsonlinux added
a check in https://github.com/zfsonlinux/zfs/commit/784652c
due to #3468. The check was reintroduced here.
Ported-by: Witaut Bajaryn <vitaut.bayaryn@gmail.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3965
Provide a generic interface to prefetch ZAP entries by name. This
functionality is being added for external consumers such as Lustre.
It is based of the existing zap_prefetch_uint64() version which is
used by the deduplication code.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Richard Yao <ryao@gentoo.org>
Closes #4061
ilovezfs [Sun, 22 Nov 2015 12:06:21 +0000 (04:06 -0800)]
Ext4's typical GPT partition type not recognized
Adding additional entries to the efi conversion array will help prevent
the overwriting of the GPTs of disks with in-use file systems in more
cases. Most notably, this adds partition type 8300 "Linux filesystem"
(0FC63DAF-8483-4772-8E79-3D69D8477DE4), which is often used for ext4 and
btrfs, among others.
This commit itself does nothing to address the underlying problematic
behavior that check_slice() isn't called on partitions of an
unrecognized type, even when they contain a currently mounted file
system.
The additional entries were derived from these two resources:
https://en.wikipedia.org/wiki/GUID_Partition_Table
http://sourceforge.net/p/gptfdisk/code/ci/master/tree/parttypes.cc
Signed-off-by: ilovezfs <ilovezfs@icloud.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #4016
Yuri Pankov [Thu, 23 Feb 2012 03:11:44 +0000 (07:11 +0400)]
Illumos 934 - FreeBSD's GPT not recognized
Reviewed by: Alexander Eremin <alexander.r.eremin@gmail.com>
Reviewed by: Garrett D'Amore <garrett@damore.org>
Reviewed by: Andrew Stormont <Andrew.Stormont@nexenta.com>
Reviewed by: Richard Elling <richard.elling@richardelling.com>
Approved by: Gordon Ross <gwr@nexenta.com>
Richard Yao [Wed, 25 Nov 2015 15:39:57 +0000 (10:39 -0500)]
Only trigger SET_ERROR tracepoint event on error
Currently, the SET_ERROR tracepoint triggers regardless of whether there
is an error or not. On Illumos, SET_ERROR only triggers on an actual
error, which is avoids irrelevant noise. Linux 2.6.38 added support for
conditional tracepoints, so we modify SET_ERROR to use them when they
are avaliable for functionality equivalent to the Illumos functionality.
Signed-off-by: Richard Yao <ryao@gentoo.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4043
Chunwei Chen [Fri, 20 Nov 2015 23:50:06 +0000 (15:50 -0800)]
Fix zdb calling behavior in ztest
The current zdb calling behaviour is really fragile, and is guaranteed to
segfault if ztest is not installed in either /sbin or /usr/sbin. With this
patch, the ztest will try to call zdb in the following order.
1. Use environmental variable ZDB_PATH if provided.
2. If ztest resides in build tree, guess the in tree zdb path.
3. Just pass zdb to popen and let it search it in PATH.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3126
Adding VPATH support, commit 47a4a6f, required that a `src`
and `obj` line be added to the top of the Makefiles. They
must be removed from the Makefiles when builtin.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue zfsonlinux/spl#481
Issue zfsonlinux/spl#498
Chunwei Chen [Mon, 23 Nov 2015 23:06:46 +0000 (15:06 -0800)]
Linux 4.4 compat: xattr operations takes xattr_handler
The xattr_hander->{list,get,set} were changed to take a xattr_handler,
and handler_flags argument was removed and should be accessed by
handler->flags.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #4021
Chunwei Chen [Mon, 23 Nov 2015 22:47:29 +0000 (14:47 -0800)]
Linux 4.4 compat: make_request_fn returns blk_qc_t
As part of block polling support in Linux 4.4, make_request_fn should
return a cookie value of type blk_qc_t. For now, we make zvol_request
always return BLK_QC_T_NONE until we assess whether and how we want
to support block polling.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #4021
tuxoko [Fri, 30 Oct 2015 23:10:01 +0000 (16:10 -0700)]
Fix zfs_dirty_data_max overflow on 32-bit
On 32 bit, the calculation of zfs_dirty_data_max from phymem will overflow,
causing it to be smaller than zfs_dirty_data_sync, and will cause txg being
delayed while no one write to disk. The end result is horrendous write speed.
On 4G ram 32-bit VM, before this patch, simple dd results in ~7MB/s. Now it
can reach speed on par with 64-bit VM.
Signed-off-by: Chunwei Chen <tuxoko@gmail.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3973
Chunwei Chen [Tue, 17 Nov 2015 00:39:52 +0000 (16:39 -0800)]
Fix snapshot automount behavior when concurrent or fail
When concurrent threads accessing the snapdir, one will succeed the user
helper mount while others will get EBUSY. However, the original code treats
those EBUSY threads as success and goes on to do zfsctl_snapshot_add, which
causes repeated avl_add and thus panic.
Also, if the snapshot is already mounted somewhere else, a thread accessing
the snapdir will also get EBUSY from user helper mount. And it will cause
strange things as doing follow_down_one will fail and then follow_up will jump
up to the mountpoint of the filesystem and confuse the hell out of VFS.
The patch fix both behavior by returning 0 immediately for the EBUSY threads.
Note, this will have a side effect for the second case where the VFS will
retry several times before returning ELOOP.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4018
Jason Zaman [Sat, 24 Oct 2015 06:01:08 +0000 (14:01 +0800)]
sysmacros: Make P2ROUNDUP not trigger int overflow
The original P2ROUNDUP and P2ROUNDUP_TYPED macros contain -x which
triggers PaX's integer overflow detection for unsigned integers.
Replace the macros with an equivalent version that does not trigger
the overflow.
Axioms:
A. (-(x)) === (~((x) - 1)) === (~(x) + 1) under two's complement.
B. ~(x & y) === ((~(x)) | (~(y))) under De Morgan's law.
C. ~(~x) === x under the law of excluded middle.
Proof:
0. (-(-(x) & -(align))) original
1. (~(-(x) & -(align)) + 1) by A
2. (((~(-(x))) | (~(-(align)))) + 1) by B
3. (((~(~((x) - 1))) | (~(~((align) - 1)))) + 1) by A
4. (((((x) - 1)) | (((align) - 1))) + 1) by C
Q.E.D.
Signed-off-by: Jason Zaman <jason@perfinion.com> Reviewed-by: Chris Dunlop <chris@onthe.net.au> Reviewed-by: Richard Yao <ryao@gentoo.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3949
Brian Behlendorf [Mon, 16 Nov 2015 17:47:43 +0000 (09:47 -0800)]
zimport.sh: Add configure/make option support
Allow the following environment variables to control the build
behavior of the zimport.sh script. This can be useful when you
want a debug build or require specific build options. The
default values are:
CONFIG_OPTIONS=""
MAKE_OPTIONS="-s -j$(nproc)"
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Mon, 16 Nov 2015 23:00:38 +0000 (15:00 -0800)]
Follow 0/-E convention for module load errors
Because errors during module load are so rare it went unnoticed that
it was possible that a positive errno was returned. This would result
in the module being loaded, nothing being initialized, and a system
panic shortly thereafter. This is what was causing the hard failures
in the automated testing.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
AndCycle [Tue, 10 Nov 2015 14:01:26 +0000 (22:01 +0800)]
Obey arc_meta_limit default size when changing arc_max
When decreasing the maximum ARC size preserve the 3/4 default
ratio for the arc_meta_limit. Otherwise, the arc_meta_limit
may be set the same as arc_max.
Signed-off-by: AndCycle <andcycle@andcycle.idv.tw> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4001
The TEST file is provided as a hint to the automated test infra-
structure. It controls which regression tests are run and how they
are run. This file along with any lines in the commit messages
which start with TEST_* are sourced by the test scripts and can
be used to override the default values. For complete details see:
https://github.com/zfsonlinux/zfs-buildbot/
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
As of gcc 5.1.1 20150618 (Red Hat 5.1.1-4) the -Werror=maybe-uninitialized
check detects that 'snapname' in recv_incremental_replication() may not be
initialized. Explicitly initialize the variable to resolved the warning.
libzfs_sendrecv.c: In function ‘recv_incremental_replication’:
libzfs_sendrecv.c:2019:2: error: ‘snapname’ may be used uninitialized in
(void) snprintf(buf, sizeof (buf), "%s@%s", fsname, snapname);
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Chunwei Chen [Fri, 9 Oct 2015 19:27:01 +0000 (12:27 -0700)]
Fix fail path in zfs_znode_alloc
When sa_bulk_lookup() fails, unlock_new_inode() will spit out a WARNING. It
will also recursive deadlock on ZFS_OBJ_HOLD_ENTER in zfs_zinactive().
Since we never call insert_inode_locked in fail path, I_NEW is never set, the
inode is never hashed. So unlock_new_inode() can be safely remove it.
We set z_sa_hdl to NULL in fail path so that iput path will stop at
zfs_inactive() without entering zfs_zinactive(). This way we can avoid the
deadlock and prevent double sa_handle_destroy().
Signed-off-by: Chunwei Chen <david.chen@osnexus.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3899
Chunwei Chen [Tue, 13 Oct 2015 21:13:52 +0000 (14:13 -0700)]
Fix use-after-free in vdev_disk_physio_completion
Currently, vdev_disk_physio_completion will try to wake up an waiter without
first checking the existence. This creates a race window in which complete is
called after dr is freed.
We add dr_wait in dio_request to indicate the existence of waiter. Also,
remove dr_rw since no one is using it, and reorder dr_ref to make the struct
more compact in 64bit.
Justin T. Gibbs [Tue, 13 Oct 2015 21:09:45 +0000 (14:09 -0700)]
Illumos 6267 - dn_bonus evicted too early
6267 dn_bonus evicted too early
Reviewed by: Richard Yao <ryao@gentoo.org>
Reviewed by: Xin LI <delphij@freebsd.org>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
James Lee [Sun, 20 Sep 2015 02:00:36 +0000 (22:00 -0400)]
zfs-import: Perform verbatim import using cache file
This change modifies the import service to use the default cache file
to perform a verbatim import of pools at boot. This fixes code that
searches all devices and imported all visible pools.
Using the cache file is in keeping with the way ZFS has always worked,
how Solaris, Illumos, FreeBSD, and systemd performs imports, and is how
it is written in the man page (zpool(1M,8)):
All pools in this cache are automatically imported when the
system boots.
Importantly, the cache contains important information for importing
multipath devices, and helps control which pools get imported in more
dynamic environments like SANs, which may have thousands of visible
and constantly changing pools, which the ZFS_POOL_EXCEPTIONS variable
is not equipped to handle. Verbatim imports prevent rogue pools from
being automatically imported and mounted where they shouldn't be.
The change also stops the service from exporting pools at shutdown.
Exporting pools is only meant to be performed explicitly by the
administrator of the system.
The old behavior of searching and importing all visible pools is
preserved and can be switched on by heeding the warning and toggling
the ZPOOL_IMPORT_ALL_VISIBLE variable in /etc/default/zfs.
Signed-off-by: James Lee <jlee@thestaticvoid.com> Signed-off-by: Richard Yao <ryao@gentoo.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3777
Closes #3526
Brian Behlendorf [Tue, 13 Oct 2015 16:17:01 +0000 (09:17 -0700)]
Fix 'arc_c < arc_c_min' panic
Strictly enforce keeping 'arc_c >= arc_c_min'. The ASSERTs are
left in place to catch this in a debug build but logic has been
added to gracefully handle in a production build.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3904
For consistency all systemd unit files and init scripts now share
the same names. This prevents an issue where the zed is started
twice on systems where both the systemd and sysv infrastructure is
installed concurrently.
For backward compatibility a 'zed' alias has been added. This
allows the user to interact with the service using either the
name 'zed' or 'zfs-zed'.
Signed-off-by: Turbo Fredriksson <turbo@bayour.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3837
Modern versions of dkms cleanup the build directory after installing.
This resulted in 'dkms uninstall' never running because the check
added by commit 866c162 which verifies the existance of the
zfs.release build product would never be true.
This patch resolves the issue by updating the conditional to check
in the explicitly installed zfs_config.h file for the version.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3862
* Fix regression - "OVERLAY_MOUNTS" should have been "DO_OVERLAY_MOUNTS".
* Fix update-rc.d commands in postinst. Thanx to subzero79@GitHub.
* Fix make sure a filesystem exists before trying to mount in mount_fs()
* Fix local variable usage.
* Fix to read_mtab():
* Strip control characters (space - \040) from /proc/mounts GLOBALY,
not just first occurrence.
* Don't replace unprintable characters ([/-. ]) for use in the variable
name with underscore. No need, just remove them all together.
* Add check_boolean() to check if a user configure option is
set ('yes', 'Yes', 'YES' or any combination there of) OR '1'.
Anything else is considered 'unset'.
* Add a ZFS_POOL_IMPORT to the default config.
* This is a semi colon separated list of pools to import ONLY.
* This is intended for systems which have _a lot_ of pools (from
a SAN for example) and it would be to many to put in the
ZFS_POOL_EXCEPTIONS variable..
* Add a config option "ZPOOL_IMPORT_OPTS" for adding additional options
to "zpool import".
* Add documentation and the chance of overriding the ZPOOL_CACHE
variable in the config file.
* Remove "sort" from find_pools() and setup_snapshot_booting().
Sometimes not available, and not really necessary.
Signed-off-by: Turbo Fredriksson <turbo@bayour.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov>
Issue #3816
When doing uioskip to skip an iovec to the very end, the current loop
condition will falsely check pass the end of iovec. We fix this checking
uio_iovcnt first.
The solution is to delete the assertion. This could potentially
occur in uiomove as well, which contains analogous assertions
that appear similarly unnecessary, so we remove those as well.
Reported-by: Jonathan Vasquez <jvasquez1011@gmail.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Richard Yao <ryao@gentoo.org>
Issue #3792
Brian Behlendorf [Thu, 24 Sep 2015 23:32:25 +0000 (16:32 -0700)]
Fix synchronous behavior in __vdev_disk_physio()
Commit b39c22b set the READ_SYNC and WRITE_SYNC flags for a bio
based on the ZIO_PRIORITY_* flag passed in. This had the unnoticed
side-effect of making the vdev_disk_io_start() synchronous for
certain I/Os.
This in turn resulted in vdev_disk_io_start() being able to
re-dispatch zio's which would result in a RCU stalls when a disk
was removed from the system. Additionally, this could negatively
impact performance and explains the performance regressions reported
in both #3829 and #3780.
This patch resolves the issue by making the blocking behavior
dependent on a 'wait' flag being passed rather than overloading
the passed bio flags.
Finally, the WRITE_SYNC and READ_SYNC behavior is restricted to
non-rotational devices where there is no benefit to queuing to
aggregate the I/O.
Brian Behlendorf [Wed, 23 Sep 2015 22:59:04 +0000 (15:59 -0700)]
Avoid blocking in arc_reclaim_thread()
As described in the comment above arc_reclaim_thread() it's critical
that the reclaim thread be careful about blocking. Just like it must
never wait on a hash lock, it must never wait on a task which can in
turn wait on the CV in arc_get_data_buf(). This will deadlock, see
issue #3822 for full backtraces showing the problem.
To resolve this issue arc_kmem_reap_now() has been updated to use the
asynchronous arc prune function. This means that arc_prune_async()
may now be called while there are still outstanding arc_prune_tasks.
However, this isn't a problem because arc_prune_async() already
keeps a reference count preventing multiple outstanding tasks per
registered consumer. Functionally, this behavior is the same as
the counterpart illumos function dnlc_reduce_cache().
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Tim Chase <tim@chase2k.com>
Issue #3808
Issue #3834
Issue #3822
Brian Behlendorf [Wed, 23 Sep 2015 21:28:43 +0000 (14:28 -0700)]
Disable zpl_nr_cached_objects() callback
The zpl_nr_cached_objects() function has been disabled because in the
current code it doesn't provide any critical functionality and it may
result in a deadlock under certain circumstances. However, because
we expect to need these hooks in the future this code has not been
entirely removed.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3719
Brian Behlendorf [Wed, 23 Sep 2015 20:00:28 +0000 (13:00 -0700)]
Allow NFS activity to defer snapshot unmounts
Accessing a snapshot via NFS should cause an auto-unmount of that
snapshot to be deferred until such as time as the snapshot is idle.
This is analogous to the zpl_revalidate logic employed by locally
mounted snapshots.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3794
Commit torvalds/linux@4246a0b63bd8f56a1469b12eafeb875b1041a451
("block: add a bi_error field to struct bio") dropped the error
argument from bio_endio in favor of newly introduced bio->bi_error.
This also replaces bio->bi_flags value BIO_UPTODATE.
bio_endio was a 3 argument function until Linux 2.6.24, which made it
a 2 argument function, and now the prototype has changed yet again to
a 1 argument function. Support for pre 2.6.24 kernels was already
dropped with 37f9dac592bf ("zvol processing should use struct bio")
which assumed the 2 argument version in zvol_request(). Remaining code
to support the 3 argument version is hereby removed.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Lukas Wunner <lukas@wunner.de>
Issue #3799
Don Brady [Thu, 17 Sep 2015 23:55:22 +0000 (17:55 -0600)]
Add large block support to zpios(1) benchmark
As part of the large block support effort, it makes sense to add
support for large blocks to **zpios(1)**. The specifying of a zfs
block size for zpios is optional and will default to 128K if the
block size is not specified.
`zpios ... -S size | --blocksize size ...`
This will use *size* ZFS blocks for each test, specified as a comma
delimited list with an optional unit suffix. The supported range is
powers of two from 128K through 16M. A range of block sizes can be
tested as follows: `-S 128K,256K,512K,1M`
Example run below
(non realistic results from a VM and output abbreviated for space)
Tab-indent continuation lines in the "scan:" section of "zpool status".
All other sections use a tab, which makes them easy to parse. Only the
"scan:" section had its continuation lines indented with four spaces.
This makes them consistent with the others.
Signed-off-by: Remy Blank <remy.blank@pobox.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Richard Yao <ryao@gentoo.org>
Closes #3769
Ned Bass [Wed, 16 Sep 2015 09:49:09 +0000 (02:49 -0700)]
Honor xattr=sa dataset property
ZFS incorrectly uses directory-based extended attributes even when
xattr=sa is specified as a dataset property or mount option. Support to
honor temporary mount options including "xattr" was added in commit 0282c4137e7409e6d85289f4955adf07fac834f5. There are two issues with the
mount option handling:
* Libzfs has historically included "xattr" in its list of default mount
options. This overrides the dataset property, so the dataset is always
configured to use directory-based xattrs even when the xattr dataset
property is set to off or sa. Address this by removing "xattr" from
the set of default mount options in libzfs.
* There was no way to enable system attribute-based extended attributes
using temporary mount options. Add the mount options "saxattr" and
"dirxattr" which enable the xattr behavior their names suggest. This
approach has the advantages of mirroring the valid xattr dataset
property values and following existing conventions for mount option
names.
Signed-off-by: Ned Bass <bass6@llnl.gov> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3787
Richard Yao [Fri, 18 Sep 2015 12:32:52 +0000 (08:32 -0400)]
Discard on zvols should not exceed the length of a block
37f9dac592bf5889c3efb305c48ac39b4c7dd140 replaced the end-start
calculation with a cached value, but neglected to update it on discard
operations. This can cause us to discard data not requested, causing
data loss on zvols.
Reported-by: Richard Connon <richard.connon@zynstra.com> Signed-off-by: Richard Yao <ryao@gentoo.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3798
Arne Jansen [Fri, 11 Sep 2015 16:18:56 +0000 (09:18 -0700)]
Illumos 6214 - zpools going south
6214 zpools going south
Reviewed by: Igor Kozhukhov <ikozhukhov@gmail.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Dan McDonald <danmcd@omniti.com>
Reviewed by: Saso Kiselkov <skiselkov.ml@gmail.com>
Reintroduce b_compress to the l2arc_buf_hdr_t. In commit b9541d6
the compression flags were moved to the generic b_flags in the
arc_buf_hdr_t. This is a problem because l2arc_compress_buf()
may manipulate the compression flags and this can only be done
safely under the hash lock which is not held. See Illumos 6214
for a detailed analysis of the race.
HDR_GET_COMPRESS() macro was removed from arc_buf_info().
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3757
Brian Behlendorf [Tue, 18 Aug 2015 20:51:20 +0000 (13:51 -0700)]
Prefetch start and end of volumes
When adding a zvol to the system prefetch zvol_prefetch_bytes from the
start and end of the volume. Prefetching these regions of the volume is
desirable because they are likely to be accessed immediately by blkid(8),
the kernel scanning for a partition table, or another task which probes
the devices.
Signed-off-by: Richard Yao <ryao@gentoo.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3659
Richard Yao [Mon, 7 Sep 2015 16:03:19 +0000 (12:03 -0400)]
Reintroduce IO accounting on zvols on Linux 3.19+
zfsonlinux/zfs@e20cd6f7a8922709b1aa2ecefd783390102d79e0 caused us to
lose IO accounting on zvols. When I originally wrote that last year, the
symbols we needed to maintain IO accounting were GPL exported, but
torvalds/linux@394ffa503bc40e32d7f54a9b817264e81ce131b4 provided
suitable symbols for restoring this functionality 4 months later. We
can call them to restore the IO accounting on Linux 3.19 and later as
well as any older kernels where that patch is backported.
Signed-off-by: Richard Yao <ryao@gentoo.org> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3741
Internally ZFS keeps a small log to facilitate debugging. By default
the log is disabled, to enable it set zfs_dbgmsg_enable=1. The contents
of the log can be accessed by reading the /proc/spl/kstat/zfs/dbgmsg file.
Writing 0 to this proc file clears the log.
$ echo 1 >/sys/module/zfs/parameters/zfs_dbgmsg_enable
$ echo 0 >/proc/spl/kstat/zfs/dbgmsg
$ zpool import tank
$ cat /proc/spl/kstat/zfs/dbgmsg
1 0 0x01 -1 0 24923575255422525836565501
timestamp message 1441141408 spa=tank async request task=1 1441141408 txg 70 open pool version 5000; software version 5000/5; ... 1441141409 spa=tank async request task=32 1441141409 txg 72 import pool version 5000; software version 5000/5; ... 1441141414 command: lt-zpool import tank
Note the zfs_dbgmsg() and dprintf() functions are both now mapped to
the same log. As mentioned above the kernel debug log can be accessed
though the /proc/spl/kstat/zfs/dbgmsg kstat. For user space consumers
log messages are immediately written to stdout after applying the
ZFS_DEBUG environment variable.
$ ZFS_DEBUG=on ./cmd/ztest/ztest -V
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov>
Closes #3728
Brian Behlendorf [Fri, 28 Aug 2015 21:54:32 +0000 (14:54 -0700)]
Support accessing .zfs/snapshot via NFS
This patch is based on the previous work done by @andrey-ve and
@yshui. It triggers the automount by using kern_path() to traverse
to the known snapshout mount point. Once the snapshot is mounted
NFS can access the contents of the snapshot.
Allowing NFS clients to access to the .zfs/snapshot directory would
normally mean that a root user on a client mounting an export with
'no_root_squash' would be able to use mkdir/rmdir/mv to manipulate
snapshots on the server. To prevent configuration mistakes a
zfs_admin_snapshot module option was added which disables the
mkdir/rmdir/mv functionally. System administators desiring this
functionally must explicitly enable it.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #2797
Closes #1655
Closes #616
Richard Yao [Fri, 10 Oct 2014 15:23:23 +0000 (11:23 -0400)]
Support secure discard on zvols
Linux 2.6.36 introduced REQ_SECURE to indicate when discards *must* be
processed, such that we cannot do optimizations like block alignment.
Consequently, the discard semantics prior to 2.6.36 require us to always
process unaligned discards. Previously, we would do this optimization
regardless. This patch changes things to correctly restrict this
optimization to situations where REQ_SECURE exists, but is not included
in the flags.
Richard Yao [Fri, 4 Jul 2014 22:43:47 +0000 (18:43 -0400)]
zvol processing should use struct bio
Internally, zvols are files exposed through the block device API. This
is intended to reduce overhead when things require block devices.
However, the ZoL zvol code emulates a traditional block device in that
it has a top half and a bottom half. This is an unnecessary source of
overhead that does not exist on any other OpenZFS platform does this.
This patch removes it. Early users of this patch reported double digit
performance gains in IOPS on zvols in the range of 50% to 80%.
Comments in the code suggest that the current implementation was done to
obtain IO merging from Linux's IO elevator. However, the DMU already
does write merging while arc_read() should implicitly merge read IOs
because only 1 thread is permitted to fetch the buffer into ARC. In
addition, commercial ZFSOnLinux distributions report that regular files
are more performant than zvols under the current implementation, and the
main consumers of zvols are VMs and iSCSI targets, which have their own
elevators to merge IOs.
Some minor refactoring allows us to register zfs_request() as our
->make_request() handler in place of the generic_make_request()
function. This eliminates the layer of code that broke IO requests on
zvols into a top half and a bottom half. This has several benefits:
1. No per zvol spinlocks.
2. No redundant IO elevator processing.
3. Interrupts are disabled only when actually necessary.
4. No redispatching of IOs when all taskq threads are busy.
5. Linux's page out routines will properly block.
6. Many autotools checks become obsolete.
An unfortunate consequence of eliminating the layer that
generic_make_request() is that we no longer calls the instrumentation
hooks for block IO accounting. Those hooks are GPL-exported, so we
cannot call them ourselves and consequently, we lose the ability to do
IO monitoring via iostat. Since zvols are internally files mapped as
block devices, this should be okay. Anyone who is willing to accept the
performance penalty for the block IO layer's accounting could use the
loop device in between the zvol and its consumer. Alternatively, perf
and ftrace likely could be used. Also, tools like latencytop will still
work. Tools such as latencytop sometimes provide a better view of
performance bottlenecks than the traditional block IO accounting tools
do.
Lastly, if direct reclaim occurs during spacemap loading and swap is on
a zvol, this code will deadlock. That deadlock could already occur with
sync=always on zvols. Given that swap on zvols is not yet production
ready, this is not a blocker.
Brian Behlendorf [Mon, 31 Aug 2015 23:46:01 +0000 (16:46 -0700)]
Add temporary mount options
Add the required kernel side infrastructure to parse arbitrary
mount options. This enables us to support temporary mount
options in largely the same way it is handled on other platforms.
See the 'Temporary Mount Point Properties' section of zfs(8)
for complete details.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #985
Closes #3351
Richard Yao [Mon, 31 Aug 2015 15:36:32 +0000 (11:36 -0400)]
VDEV_REQ_FUA should be mapped to REQ_FUA
Pre-2.6.37 kernels support REQ_FUA in request flags, but not in BIO
flags. zvols are the only consumer of VDEV_REQ_FUA and since they are
passed requests, they should be obey the REQ_FUA flag like later
kernels. This optimization will only matter on 2.6.36 and 2.6.37 because
the zvol rework changes things to use bio, where we no longer are able
to distinguish on earlier kernels
Tim Chase [Mon, 31 Aug 2015 01:59:23 +0000 (20:59 -0500)]
Dbuf hash table should be sized as is the arc hash table
Commit 49ddb315066e372f31bda29a5c546a9eccc8b418 added the
zfs_arc_average_blocksize parameter to allow control over the size of
the arc hash table. The dbuf hash table's size should be determined
similarly.
Signed-off-by: Tim Chase <tim@chase2k.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3721
Allow for easy turning of a pools reserved free space. Previous
versions of ZFS (v0.6.4 and earlier) held 1/64 of the pools capacity
in reserve. Commits 3d45fdd and 0c60cc3 increased this to 1/32.
Setting spa_slop_shift=6 will restore the previous default setting.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3724
projects / mirror_zfs.git / log