Current rw_tryupgrade does rw_exit and then rw_tryenter(RW_RWITER), and then
does rw_enter(RW_READER) if it fails. This violate the assumption that
rw_tryupgrade should be atomic and could cause extra contention or even lock
inversion.
This patch we implement a proper rw_tryupgrade. For rwsem-spinlock, we take
the spinlock to check rwsem->count and rwsem->wait_list. For normal rwsem, we
use cmpxchg on rwsem->count to change the value from single reader to single
writer.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tim Chase <tim@chase2k.com>
Closeszfsonlinux/zfs#4692Closes#554
GCC for MIPS only defines _LP64 when 64bit,
while no _ILP32 defined when 32bit.
Signed-off-by: YunQiang Su <syq@debian.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#558
Register iterate_shared if it exists so the kernel will used shared
lock and allowing concurrent readdir.
Also, use shared lock when doing llseek with SEEK_DATA or SEEK_HOLE
to allow concurrent seeking.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#4664Closes#4665
Linux 4.7 changes i_mutex to i_rwsem, and we should used inode_lock and
inode_lock_shared to do exclusive and shared lock respectively.
We use spl_inode_lock{,_shared}() to hide the difference. Note that on older
kernel you'll always take an exclusive lock.
We also add all other inode_lock friends. And nested users now should
explicitly call spl_inode_lock_nested with correct subclass.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue zfsonlinux/zfs#4665Closes#549
This reverts commit 8302528617 and
ebecfcd699 which broke the build.
While these patches do apply cleanly and passed previous test
runs they need to be updated to account for the changes made in
commit 241b541574.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3878
The zfs range lock interface no longer tightly depends on a
znode_t and therefore can be used in ztest. This allows the
previous ztest specific implementation to be removed, and for
additional test coverage of the shared version.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Boris Protopopov <boris.protopopov@actifio.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #4023
Issue #4024
struct zvol_state contains a dummy znode, which is around 1KB on x64,
only for zfs_range_lock. But in reality, other than z_range_lock and
z_range_avl, zfs_range_lock only need znode on regular file, which
means we add 1KB on a structure and gain nothing.
In this patch, we remove the dummy znode for zvol_state. In order to
do that, we also need to refactor zfs_range_lock a bit. We move
z_range_lock and z_range_avl pair out of znode_t to form zfs_rlock_t.
This new struct replaces znode_t as the main handle inside the range
lock functions.
We also add pointers to z_size, z_blksz, and z_max_blksz so range lock
code doesn't depend on znode_t. This allows non-ZPL consumers like
Lustre to use the range locks with their equivalent znode_t structure.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Boris Protopopov <boris.protopopov@actifio.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#4510
The was originally using interruptible cv_timedwait_sig, but was changed
to uninterruptible cv_timedwait_hires in ae6d0c6. Use _sig_hires instead
to allow interruptible sleep.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#4633Closes#4634
This reverts commit 4cd77889b6. The
i_generation field in the inode is 32-bit and the SA code expects
64-bit fixed values. Revert this optimization for now until
this is cleanly addressed.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #4538
As described in torvalds/linux@5f3a4a2 the &init_user_ns, and
not the current user_ns, should be passed to posix_acl_from_xattr()
and posix_acl_to_xattr(). Conveniently the init_user_ns is
available through the init credential (kcred).
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Massimo Maggi <me@massimo-maggi.eu>
Closes#4177
6844 dnode_next_offset can detect fictional holes
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
dnode_next_offset is used in a variety of places to iterate over the
holes or allocated blocks in a dnode. It operates under the premise that
it can iterate over the blockpointers of a dnode in open context while
holding only the dn_struct_rwlock as reader. Unfortunately, this premise
does not hold.
When we create the zio for a dbuf, we pass in the actual block pointer
in the indirect block above that dbuf. When we later zero the bp in
zio_write_compress, we are directly modifying the bp. The state of the
bp is now inconsistent from the perspective of dnode_next_offset: the bp
will appear to be a hole until zio_dva_allocate finally finishes filling
it in. In the meantime, dnode_next_offset can detect a hole in the dnode
when none exists.
I was able to experimentally demonstrate this behavior with the
following setup:
1. Create a file with 1 million dbufs.
2. Create a thread that randomly dirties L2 blocks by writing to the
first L0 block under them.
3. Observe dnode_next_offset, waiting for it to skip over a hole in the
middle of a file.
4. Do dnode_next_offset in a loop until we skip over such a non-existent
hole.
The fix is to ensure that it is valid to iterate over the indirect
blocks in a dnode while holding the dn_struct_rwlock by passing the zio
a copy of the BP and updating the actual BP in dbuf_write_ready while
holding the lock.
References:
https://www.illumos.org/issues/6844https://github.com/openzfs/openzfs/pull/82
DLPX-35372
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#4548
This change was lost, somehow, in e5f9a9a. Since the arrays can be
rather large, they need to be allocated with vmem_zalloc() via dfl_alloc()
and freed with vmem_free() via dfl_free().
The new dfl_alloc() function should be used to allocate object of type
dkioc_free_list_t in order that they're allocated from vmem.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Nikolay Borisov <kernel@kyup.com>
Closes#543
To reduce mutex footprint, we detect the existence of owner in kernel mutex,
and rely on it if it exists.
Note that before Linux 3.0, mutex owner is of type thread_info. Also note
that, in Linux 3.18, the condition for owner is changed from
CONFIG_DEBUG_MUTEXES || CONFIG_SMP to
CONFIG_DEBUG_MUTEXES || CONFIG_MUTEX_SPIN_ON_OWNER
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#540
In order to remove the HAVE_PN_UTILS wrappers the pn_alloc() and
pn_free() functions must be implemented. The existing illumos
implementation were used for this purpose.
The `flags` argument which was used in places wrapped by the
HAVE_PN_UTILS condition has beed added back to zfs_remove() and
zfs_link() functions. This removes a small point of divergence
between the ZoL code and upstream.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#4522
The problem for atime:
We have 3 places for atime: inode->i_atime, znode->z_atime and SA. And its
handling is a mess. A huge part of mess regarding atime comes from
zfs_tstamp_update_setup, zfs_inode_update, and zfs_getattr, which behave
inconsistently with those three values.
zfs_tstamp_update_setup clears z_atime_dirty unconditionally as long as you
don't pass ATTR_ATIME. Which means every write(2) operation which only updates
ctime and mtime will cause atime changes to not be written to disk.
Also zfs_inode_update from write(2) will replace inode->i_atime with what's
inside SA(stale). But doesn't touch z_atime. So after read(2) and write(2).
You'll have i_atime(stale), z_atime(new), SA(stale) and z_atime_dirty=0.
Now, if you do stat(2), zfs_getattr will actually replace i_atime with what's
inside, z_atime. So you will have now you'll have i_atime(new), z_atime(new),
SA(stale) and z_atime_dirty=0. These will all gone after umount. And you'll
leave with a stale atime.
The problem for relatime:
We do have a relatime config inside ZFS dataset, but how it should interact
with the mount flag MS_RELATIME is not well defined. It seems it wanted
relatime mount option to override the dataset config by showing it as
temporary in `zfs get`. But at the same time, `zfs set relatime=on|off` would
also seems to want to override the mount option. Not to mention that
MS_RELATIME flag is actually never passed into ZFS, so it never really worked.
How Linux handles atime:
The Linux kernel actually handles atime completely in VFS, except for writing
it to disk. So if we remove the atime handling in ZFS, things would just work,
no matter it's strictatime, relatime, noatime, or even O_NOATIME. And whenever
VFS updates the i_atime, it will notify the underlying filesystem via
sb->dirty_inode().
And also there's one thing to note about atime flags like MS_RELATIME and
other flags like MS_NODEV, etc. They are mount point flags rather than
filesystem(sb) flags. Since native linux filesystem can be mounted at multiple
places at the same time, they can all have different atime settings. So these
flags are never passed down to filesystem drivers.
What this patch tries to do:
We remove znode->z_atime, since we won't gain anything from it. We remove most
of the atime handling and leave it to VFS. The only thing we do with atime is
to write it when dirty_inode() or setattr() is called. We also add
file_accessed() in zpl_read() since it's not provided in vfs_read().
After this patch, only the MS_RELATIME flag will have effect. The setting in
dataset won't do anything. We will make zfstuil to mount ZFS with MS_RELATIME
set according to the setting in dataset in future patch.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #4482
This is foundational work for ZED.
Updates a leaf vdev's persistent device strings on Linux platform
* only applies for a dedicated leaf vdev (aka whole disk)
* updated during pool create|add|attach|import
* used for matching device matching during auto-{online,expand,replace}
* stored in a leaf disk config label (i.e. alongside 'path' NVP)
* can opt-out using env var ZFS_VDEV_DEVID_OPT_OUT=YES
Some examples:
path: '/dev/sdb1'
devid: 'scsi-350000394a8ca4fbc-part1'
phys_path: 'pci-0000:04:00.0-sas-0x50000394a8ca4fbf-lun-0'
path: '/dev/mapper/mpatha'
devid: 'dm-uuid-mpath-35000c5006304de3f'
Signed-off-by: Don Brady <don.brady@intel.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#2856Closes#3978Closes#4416
Effectively provide our own version of assert()/verify() for use
in user space. This minimizes our dependencies and aligns the
user space assertion handling with what's used in the kernel.
Signed-off-by: Carlo Landmeter <clandmeter@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#4449
When compiling with musl libc the return type will be incorrect.
Signed-off-by: Carlo Landmeter <clandmeter@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#4454
This is initial support for x86 vectorized implementations of ZFS parity
and checksum algorithms.
For the compilation phase, configure step checks if toolchain supports relevant
instruction sets. Each implementation must ensure that the code is not passed
to compiler if relevant instruction set is not supported. For this purpose,
following new defines are provided if instruction set is supported:
- HAVE_SSE,
- HAVE_SSE2,
- HAVE_SSE3,
- HAVE_SSSE3,
- HAVE_SSE4_1,
- HAVE_SSE4_2,
- HAVE_AVX,
- HAVE_AVX2.
For detecting if an instruction set can be used in runtime, following functions
are provided in (include/linux/simd_x86.h):
- zfs_sse_available()
- zfs_sse2_available()
- zfs_sse3_available()
- zfs_ssse3_available()
- zfs_sse4_1_available()
- zfs_sse4_2_available()
- zfs_avx_available()
- zfs_avx2_available()
- zfs_bmi1_available()
- zfs_bmi2_available()
These function should be called once, on module load, or initialization.
They are safe to use from user and kernel space.
If an implementation is using more than single instruction set, both compiler
and runtime support for all relevant instruction sets should be checked.
Kernel fpu methods:
- kfpu_begin()
- kfpu_end()
Use __get_cpuid_max and __cpuid_count from <cpuid.h>
Both gcc and clang have support for these. They also handle ebx register
in case it is used for PIC code.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chunwei Chen <tuxoko@gmail.com>
Closes#4381
Add the ZFS Test Suite and test-runner framework from illumos.
This is a continuation of the work done by Turbo Fredriksson to
port the ZFS Test Suite to Linux. While this work was originally
conceived as a stand alone project integrating it directly with
the ZoL source tree has several advantages:
* Allows the ZFS Test Suite to be packaged in zfs-test package.
* Facilitates easy integration with the CI testing.
* Users can locally run the ZFS Test Suite to validate ZFS.
This testing should ONLY be done on a dedicated test system
because the ZFS Test Suite in its current form is destructive.
* Allows the ZFS Test Suite to be run directly in the ZoL source
tree enabled developers to iterate quickly during development.
* Developers can easily add/modify tests in the framework as
features are added or functionality is changed. The tests
will then always be in sync with the implementation.
Full documentation for how to run the ZFS Test Suite is available
in the tests/README.md file.
Warning: This test suite is designed to be run on a dedicated test
system. It will make modifications to the system including, but
not limited to, the following.
* Adding new users
* Adding new groups
* Modifying the following /proc files:
* /proc/sys/kernel/core_pattern
* /proc/sys/kernel/core_uses_pid
* Creating directories under /
Notes:
* Not all of the test cases are expected to pass and by default
these test cases are disabled. The failures are primarily due
to assumption made for illumos which are invalid under Linux.
* When updating these test cases it should be done in as generic
a way as possible so the patch can be submitted back upstream.
Most existing library functions have been updated to be Linux
aware, and the following functions and variables have been added.
* Functions:
* is_linux - Used to wrap a Linux specific section.
* block_device_wait - Waits for block devices to be added to /dev/.
* Variables: Linux Illumos
* ZVOL_DEVDIR "/dev/zvol" "/dev/zvol/dsk"
* ZVOL_RDEVDIR "/dev/zvol" "/dev/zvol/rdsk"
* DEV_DSKDIR "/dev" "/dev/dsk"
* DEV_RDSKDIR "/dev" "/dev/rdsk"
* NEWFS_DEFAULT_FS "ext2" "ufs"
* Many of the disabled test cases fail because 'zfs/zpool destroy'
returns EBUSY. This is largely causes by the asynchronous nature
of device handling on Linux and is expected, the impacted test
cases will need to be updated to handle this.
* There are several test cases which have been disabled because
they can trigger a deadlock. A primary example of this is to
recursively create zpools within zpools. These tests have been
disabled until the root issue can be addressed.
* Illumos specific utilities such as (mkfile) should be added to
the tests/zfs-tests/cmd/ directory. Custom programs required by
the test scripts can also be added here.
* SELinux should be either is permissive mode or disabled when
running the tests. The test cases should be updated to conform
to a standard policy.
* Redundant test functionality has been removed (zfault.sh).
* Existing test scripts (zconfig.sh) should be migrated to use
the framework for consistency and ease of testing.
* The DISKS environment variable currently only supports loopback
devices because of how the ZFS Test Suite expects partitions to
be named (p1, p2, etc). Support must be added to generate the
correct partition name based on the device location and name.
* The ZFS Test Suite is part of the illumos code base at:
https://github.com/illumos/illumos-gate/tree/master/usr/src/test
Original-patch-by: Turbo Fredriksson <turbo@bayour.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Olaf Faaland <faaland1@llnl.gov>
Closes#6Closes#1534
This implementation of rw_tryupgrade() behaves slightly differently
from its counterparts on other platforms. It drops the RW_READER lock
and then acquires the RW_WRITER lock leaving a small window where no
lock is held. On other platforms the lock is never released during
the upgrade process. This is necessary under Linux because the kernel
does not provide an upgrade function.
There are currently no callers in the ZFS code where this change in
behavior is a problem. In fact, in most cases the code is already
written such that if the upgrade fails the RW_READER lock is dropped
and the caller blocks waiting to acquire the lock as RW_WRITER.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Matthew Thode <prometheanfire@gentoo.org>
Closeszfsonlinux/zfs#4388Closes#534
zfsonlinux issue #2217 - zvol minor operations: check snapdev
property before traversing snapshots of a dataset
zfsonlinux issue #3681 - lock order inversion between zvol_open()
and dsl_pool_sync()...zvol_rename_minors()
Create a per-pool zvol taskq for asynchronous zvol tasks.
There are a few key design decisions to be aware of.
* Each taskq must be single threaded to ensure tasks are always
processed in the order in which they were dispatched.
* There is a taskq per-pool in order to keep the pools independent.
This way if one pool is suspended it will not impact another.
* The preferred location to dispatch a zvol minor task is a sync
task. In this context there is easy access to the spa_t and
minimal error handling is required because the sync task must
succeed.
Support for asynchronous zvol minor operations address issue #3681.
Signed-off-by: Boris Protopopov <boris.protopopov@actifio.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#2217Closes#3678Closes#3681
Added by-partlabel and by-partuuid to the default device search
path. Made made device names in by-label more preferable.
Signed-off-by: Thijs Cramer <thijs.cramer@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#3892
Historically libblkid support was detected as part of configure
and optionally enabled. This was done because at the time support
for detecting ZFS pool vdevs had just be added to libblkid and
those updated packages were not yet part of many distributions.
This is no longer the case and any reasonably current distribution
will ship a version of libblkid which can detect ZFS pool vdevs.
This patch makes libblkid mandatory at build time and libblkid
the preferred method of scanning for ZFS pools. For distributions
which include a modern version of libblkid there is no change in
behavior. Explicitly scanning the default search paths is still
supported and can be enabled with the '-s' command line option.
Additionally making libblkid mandatory means that the 'zpool create'
command can reliably detect if a specified device has an existing
non-ZFS filesystem (ext4, xfs) and print a warning.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#2448
The existing algorithm selects a preferred leaf vdev based on offset of the zio
request modulo the number of members in the mirror. It assumes the devices are
of equal performance and that spreading the requests randomly over both drives
will be sufficient to saturate them. In practice this results in the leaf vdevs
being under utilized.
The new algorithm takes into the following additional factors:
* Load of the vdevs (number outstanding I/O requests)
* The locality of last queued I/O vs the new I/O request.
Within the locality calculation additional knowledge about the underlying vdev
is considered such as; is the device backing the vdev a rotating media device.
This results in performance increases across the board as well as significant
increases for predominantly streaming loads and for configurations which don't
have evenly performing devices.
The following are results from a setup with 3 Way Mirror with 2 x HD's and
1 x SSD from a basic test running multiple parrallel dd's.
With pre-fetch disabled (vfs.zfs.prefetch_disable=1):
== Stripe Balanced (default) ==
Read 15360MB using bs: 1048576, readers: 3, took 161 seconds @ 95 MB/s
== Load Balanced (zfslinux) ==
Read 15360MB using bs: 1048576, readers: 3, took 297 seconds @ 51 MB/s
== Load Balanced (locality freebsd) ==
Read 15360MB using bs: 1048576, readers: 3, took 54 seconds @ 284 MB/s
With pre-fetch enabled (vfs.zfs.prefetch_disable=0):
== Stripe Balanced (default) ==
Read 15360MB using bs: 1048576, readers: 3, took 91 seconds @ 168 MB/s
== Load Balanced (zfslinux) ==
Read 15360MB using bs: 1048576, readers: 3, took 108 seconds @ 142 MB/s
== Load Balanced (locality freebsd) ==
Read 15360MB using bs: 1048576, readers: 3, took 48 seconds @ 320 MB/s
In addition to the performance changes the code was also restructured, with
the help of Justin Gibbs, to provide a more logical flow which also ensures
vdevs loads are only calculated from the set of valid candidates.
The following additional sysctls where added to allow the administrator
to tune the behaviour of the load algorithm:
* vfs.zfs.vdev.mirror.rotating_inc
* vfs.zfs.vdev.mirror.rotating_seek_inc
* vfs.zfs.vdev.mirror.rotating_seek_offset
* vfs.zfs.vdev.mirror.non_rotating_inc
* vfs.zfs.vdev.mirror.non_rotating_seek_inc
These changes where based on work started by the zfsonlinux developers:
https://github.com/zfsonlinux/zfs/pull/1487
Reviewed by: gibbs, mav, will
MFC after: 2 weeks
Sponsored by: Multiplay
References:
https://github.com/freebsd/freebsd@5c7a6f5dhttps://github.com/freebsd/freebsd@31b7f68dhttps://github.com/freebsd/freebsd@e186f564
Performance Testing:
https://github.com/zfsonlinux/zfs/pull/4334#issuecomment-189057141
Porting notes:
- The tunables were adjusted to have ZoL-style names.
- The code was modified to use ZoL's vd_nonrot.
- Fixes were done to make cstyle.pl happy
- Merge conflicts were handled manually
- freebsd/freebsd@e186f564bc by my
collegue Andriy Gapon has been included. It applied perfectly, but
added a cstyle regression.
- This replaces 556011dbec entirely.
- A typo "IO'a" has been corrected to say "IO's"
- Descriptions of new tunables were added to man/man5/zfs-module-parameters.5.
Ported-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#4334
The following options have been added to the zpool add, iostat,
list, status, and split subcommands. The default behavior was
not modified, from zfs(8).
-g Display vdev GUIDs instead of the normal short
device names. These GUIDs can be used in-place of
device names for the zpool detach/off‐
line/remove/replace commands.
-L Display real paths for vdevs resolving all symbolic
links. This can be used to lookup the current block
device name regardless of the /dev/disk/ path used
to open it.
-p Display full paths for vdevs instead of only the
last component of the path. This can be used in
conjunction with the -L flag.
This behavior may also be enabled using the following environment
variables.
ZPOOL_VDEV_NAME_GUID
ZPOOL_VDEV_NAME_FOLLOW_LINKS
ZPOOL_VDEV_NAME_PATH
This change is based on worked originally started by Richard Yao
to add a -g option. Then extended by @ilovezfs to add a -L option
for openzfsonosx. Those changes have been merged, re-factored,
a -p option added and extended to all relevant zpool subcommands.
Original-patch-by: Richard Yao <ryao@gentoo.org>
Extended-by: ilovezfs <ilovezfs@icloud.com>
Extended-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: ilovezfs <ilovezfs@icloud.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#2011Closes#4341
Perf profiling of dd on a zvol revealed that my system spent 3.16% of
its time in random_get_pseudo_bytes(). No SPL consumers need
cryptographic strength entropy, so we can reduce our overhead by
changing the implementation to utilize a fast PRNG.
The Linux kernel did not export a suitable PRNG function until it
exported get_random_int() in Linux 3.10. While we could implement an
autotools check so that we use it when it is available or even try to
access the symbol on older kernels where it is not exported using the
fact that it is exported on newer ones as justification, we can instead
implement our own pseudo-random data generator. For this purpose, I have
written one based on a 128-bit pseudo-random number generator proposed
in a paper by Sebastiano Vigna that itself was based on work by the late
George Marsaglia.
http://vigna.di.unimi.it/ftp/papers/xorshiftplus.pdf
Profiling the same benchmark with an earlier variant of this patch that
used a slightly different generator (roughly same number of
instructions) by the same author showed that time spent in
random_get_pseudo_bytes() dropped to 0.06%. That is a factor of 50
improvement. This particular generator algorithm is also well known to
be fast:
http://xorshift.di.unimi.it/#speed
The benchmark numbers there state that it runs at 1.12ns/64-bits or 7.14
GBps of throughput on an Intel Core i7-4770 in what is presumably a
single-threaded context. Using it in `random_get_pseudo_bytes()` in the
manner I have will probably not reach that level of performance, but it
should be fairly high and many times higher than the Linux
`get_random_bytes()` function that we use now, which runs at 16.3 MB/s
on my Intel Xeon E3-1276v3 processor when measured by using dd on
/dev/urandom.
Also, putting this generator's seed into per-CPU variables allows us to
eliminate overhead from both spin locks and CPU memory barriers, which
is NUMA friendly.
We could have alternatively modified consumers to use something like
`gethrtime() % 3` as suggested by both Matthew Ahrens and Tim Chase, but
that has a few potential problems that this approach avoids:
1. Switching to `gethrtime() % 3` in hot code paths today requires
diverging from illumos-gate and does nothing about potential future
patches from illumos-gate that call our slow `random_get_pseudo_bytes()`
in different hot code paths. Reimplementing `random_get_pseudo_bytes()`
with a per-CPU PRNG avoids both of those things entirely, which means
less work for us in the future.
2. Looking at the code that implements `gethrtime()`, I think it is
unlikely to be faster than this per-CPU PRNG implementation of
`random_get_pseudo_bytes()`. It would be best to go with something fast
now so that there is no point in revisiting this from a performance
perspective.
3. `gethrtime() % 3` can vary in behavior from system to system based on
kernel version, architecture and clock source. In comparison, this
per-CPU PRNG is about ~40 lines of code in `random_get_pseudo_bytes()`
that should behave consistently across all systems regardless of kernel
version, system architecture or machine clock source. It is unlikely
that we would ever need to revisit this per-CPU PRNG while the same
cannot be said for `gethrtime() % 3`.
4. `gethrtime()` uses CPU memory barriers and maybe atomic instructions
depending on the clock source, so replacing `random_get_pseudo_bytes()`
with `gethrtime()` in hot code paths could still require a future person
working on NUMA scalability to reimplement it anyway while this per-CPU
PRNG would not by virtue of using neither CPU memory barriers nor atomic
instructions. Note that I did not check various clock sources for the
presence of atomic instructions. There is simply too much code to read
and given the drawbacks versus this per-cpu PRNG, there is no point in
being certain.
5. I have heard of instances where poor quality pseudo-random numbers
caused problems for HPC code in ways that took more than a year to
identify and were remedied by switching to a higher quality source of
pseudo-random numbers. While filesystems are different than HPC code, I
do not think it is impossible for us to have instances where poor
quality pseudo-random numbers can cause problems. Opting for a well
studied PRNG algorithm that passes tests for statistical randomness over
changing callers to use `gethrtime() % 3` bypasses the need to think
about both whether poor quality pseudo-random numbers can cause problems
and the statistical quality of numbers from `gethrtime() % 3`.
6. `gethrtime()` calls `getrawmonotonic()`, which uses seqlocks. This is
probably not a huge issue, but anyone using kgdb would never be able to
step through a seqlock critical section, which is not a problem either
now or with the per-CPU PRNG:
https://en.wikipedia.org/wiki/Seqlock
The only downside that I can see is that this code's memory requirement
is O(N) where N is NR_CPUS, versus the current code and `gethrtime() %
3`, which are O(1), but that should not be a problem. The seeds will use
64KB of memory at the high end (i.e `NR_CPU == 4096`) and 16 bytes of
memory at the low end (i.e. `NR_CPU == 1`). In either case, we should
only use a few hundred bytes of code for text, especially since
`spl_rand_jump()` should be inlined into `spl_random_init()`, which
should be removed during early boot as part of "Freeing unused kernel
memory". In either case, the memory requirements are minuscule.
Signed-off-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tim Chase <tim@chase2k.com>
Closes#372
This patch add a module parameter spl_taskq_kick. When writing non-zero value
to it, it will scan all the taskq, if a taskq contains a task pending for more
than 5 seconds, it will be forced to spawn a new thread. This is use as an
emergency recovery from deadlock, not a general solution.
Signed-off-by: Chunwei Chen <david.chen@osnexus.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#529
I noticed that the SPL implementation of kobj_read_file is not correct
after comparing it with the userland implementation of kobj_read_file()
in zfsonlinux/zfs#4104.
Note that we no longer pass RLIM64_INFINITY with this, but our vn_rdwr
implementation did not support it anyway, so there is no difference.
Signed-off-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#513
6815179 zpool import with a large number of LUNs is too slow
6844191 zpool import, scanning of disks should be multi-threaded
References:
https://github.com/illumos/illumos-gate/commit/4f67d75
Porting notes:
- This change was originally never ported to Linux due to it
dependence on the thread pool interface. This patch solves
that issue by switching the code to use the existing taskq
implementation which provides the same basic functionality.
However, in order for this to work properly thread_init()
and thread_fini() must be called around to taskq consumer
to perform the needed thread initialization.
- The check_one_slice, nozpool_all_slices, and check_slices
functions have been disabled for Linux. They are difficult,
but possible, to implement for Linux due to how partitions
are get names. Since this is only an optimization this code
can be added at a latter date.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
4950 files sometimes can't be removed from a full filesystem
Reviewed by: Adam Leventhal <adam.leventhal@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Reviewed by: Boris Protopopov <bprotopopov@hotmail.com>
Approved by: Dan McDonald <danmcd@omniti.com>
References:
https://www.illumos.org/issues/4950https://github.com/illumos/illumos-gate/commit/4bb7380
Porting notes:
- ZoL currently does not log discards to zvols, so the portion of
this patch that modifies the discard logging to mark it as
freeing space has been discarded.
2. may_delete_now had been removed from zfs_remove() in ZoL.
It has been reintroduced.
3. We do not try to emulate vnodes, so the following lines are
not valid on Linux:
mutex_enter(&vp->v_lock);
may_delete_now = vp->v_count == 1 && !vn_has_cached_data(vp);
mutex_exit(&vp->v_lock);
This has been replaced with:
mutex_enter(&zp->z_lock);
may_delete_now = atomic_read(&ip->i_count) == 1 && !(zp->z_is_mapped);
mutex_exit(&zp->z_lock);
Ported-by: Richard Yao <richard.yao@clusterhq.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf