Commit Graph

1025 Commits

Author SHA1 Message Date
Mark Johnston 6e2a59181e
Avoid memory allocations in the ARC eviction thread
When the eviction thread goes to shrink an ARC state, it allocates a set
of marker buffers used to hold its place in the state's sublists.

This can be problematic in low memory conditions, since
1) the allocation can be substantial, as we allocate NCPU markers;
2) on at least FreeBSD, page reclamation can block in
   arc_wait_for_eviction()

In particular, in stress tests it's possible to hit a deadlock on
FreeBSD when the number of free pages is very low, wherein the system is
waiting for the page daemon to reclaim memory, the page daemon is
waiting for the ARC eviction thread to finish, and the ARC eviction
thread is blocked waiting for more memory.

Try to reduce the likelihood of such deadlocks by pre-allocating markers
for the eviction thread at ARC initialization time.  When evicting
buffers from an ARC state, check to see if the current thread is the ARC
eviction thread, and use the pre-allocated markers for that purpose
rather than dynamically allocating them.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: George Amanakis <gamanakis@gmail.com>
Signed-off-by: Mark Johnston <markj@FreeBSD.org>
Closes #12985
2022-01-21 10:28:13 -08:00
наб e1c720de7d
libefi: remove efi_type()
All it is right now is some #if 0ed Solaris code that returns ENOSYS,
and is only applicable for the Solaris blockdev layer.
In the Illumos gate, there's a single user: rmformat(1);
I recommend a read of the manual as a blast from the past, but, well

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Issue #12844
Closes #12969
2022-01-18 14:40:43 -08:00
наб 18168da727
module/*.ko: prune .data, global .rodata
Evaluated every variable that lives in .data (and globals in .rodata)
in the kernel modules, and constified/eliminated/localised them
appropriately. This means that all read-only data is now actually
read-only data, and, if possible, at file scope. A lot of previously-
global-symbols became inlinable (and inlined!) constants. Probably
not in a big Wowee Performance Moment, but hey.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #12899
2022-01-14 15:37:55 -08:00
Tino Reichardt a798b485ae Remove sha1 hashing from OpenZFS, it's not used anywhere.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Attila Fülöp <attila@fueloep.org>
Signed-off-by: Tino Reichardt <milky-zfs@mcmilk.de>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #12895
Closes #12902
2022-01-06 16:16:28 -08:00
наб 5c8389a8cb module: icp: rip out the Solaris loadable module architecture
After progressively folding away null cases, it turns out there's
/literally/ nothing there, even if some things are part of the
Solaris SPARC DDI/DKI or the seventeen module types (some doubled for
32-bit userland), or the entire modctl syscall definition.
Nothing.

Initialisation is handled in illumos-crypto.c,
which calls all the initialisers directly

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Attila Fülöp <attila@fueloep.org>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #12895
Closes #12902
2022-01-06 16:14:04 -08:00
Brian Behlendorf 3c80e0742a
Verify dRAID empty sectors
Verify that all empty sectors are zero filled before using them to
calculate parity.  Failure to do so can result in incorrect parity
columns being generated and written to disk if the contents of an
empty sector are non-zero.  This was possible because the checksum
only protects the data portions of the buffer, not the empty sector
padding.

This issue has been addressed by updating raidz_parity_verify() to
check that all dRAID empty sectors are zero filled.  Any sectors
which are non-zero will be fixed, repair IO issued, and a checksum
error logged.  They can then be safely used to verify the parity.

This specific type of damage is unlikely to occur since it requires
a disk to have silently returned bad data, for an empty sector, while
performing a scrub.  However, if a pool were to have been damaged
in this way, scrubbing the pool with this change applied will repair
both the empty sector and parity columns as long as the data checksum
is valid.  Checksum errors will be reported in the `zpool status`
output for any repairs which are made.

Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #12857
2022-01-04 16:46:32 -08:00
наб c2f94afa0e include: dmu.h: fix unused, remove argsused
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #12844
2021-12-23 09:42:47 -08:00
наб 83719bd68c include: sys/arc.h: shim out arc_referenced()
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #12844
2021-12-23 09:36:26 -08:00
наб 29d033c6b5 include: qat.h: mark unused macro arguments as used
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #12844
2021-12-23 09:36:21 -08:00
наб eb51a9d747
zcommon: pre-iterate over sysfs instead of statting every feature
If sufficient memory (<2K, realistically) is available, libzfs_init()
can be significantly shorted by iterating over the correct sysfs
directory before registrations, we can turn 168 stats into 15/18
syscalls (3 opens (6 if built in), 3 fstats, 6 getdentses, and 3
closes), a tenfoldish reduction; this is probably a bit faster, too.

The list is always optional, and registration functions (and one-off
users) can simply pass NULL, which will fall back to the previous
mechanism

Also, don't allocate in zfs_mod_supported_impl, and use use access()
instead of stat(), since existence is really what we care about

Also, fix pre-prop-checking compat in fallback for built-in ZFS

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #12089
2021-12-16 16:43:10 -08:00
Paul Dagnelie 795075e638
Add `const` to nvlist functions to properly expose their real behavior
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Paul Dagnelie <pcd@delphix.com>
Closes #12728
2021-12-06 18:19:13 -07:00
Allan Jude 2a673e76a9
Vdev Properties Feature
Add properties, similar to pool properties, to each vdev.
This makes use of the existing per-vdev ZAP that was added as
part of device evacuation/removal.

A large number of read-only properties are exposed,
many of the members of struct vdev_t, that provide useful
statistics.

Adds support for read-only "removing" vdev property.
Adds the "allocating" property that defaults to "on" and
can be set to "off" to prevent future allocations from that
top-level vdev.

Supports user-defined vdev properties.
Includes support for properties.vdev in SYSFS.

Co-authored-by: Allan Jude <allan@klarasystems.com>
Co-authored-by: Mark Maybee <mark.maybee@delphix.com>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Allan Jude <allan@klarasystems.com>
Closes #11711
2021-11-30 07:46:25 -07:00
Rich Ercolani 269b5dadcf
Enable edonr in FreeBSD
The code is integrated, builds fine, runs fine, there's not really
any reason not to.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@ixsystems.com>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes #12735
2021-11-16 12:40:10 -07:00
George Amanakis c9d62d1356
Introduce a tunable to exclude special class buffers from L2ARC
Special allocation class or dedup vdevs may have roughly the same
performance as L2ARC vdevs. Introduce a new tunable to exclude those
buffers from being cacheable on L2ARC.

Reviewed-by: Don Brady <don.brady@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes #11761 
Closes #12285
2021-11-11 12:52:16 -08:00
Fedor Uporov 49d42425d6
Check l2cache vdevs pending list inside the vdev_inuse()
The l2cache device could be added twice because vdev_inuse() does not
check spa_l2cache for added devices. Make l2cache vdevs inuse checking
logic more closer to spare vdevs.

Reviewed-by: George Amanakis <gamanakis@gmail.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Fedor Uporov <fuporov.vstack@gmail.com>
Closes #9153 
Closes #12689
2021-11-11 11:54:15 -08:00
Fedor Uporov e39fe05b69
Skip spacemaps reading in case of pool readonly import
The only zdb utility require to read metaslab-related data during
read-only pool import because of spacemaps validation. Add global
variable which will allow zdb read spacemaps in case of readonly
import mode.

Reviewed-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Fedor Uporov <fuporov.vstack@gmail.com>
Closes #9095
Closes #12687
2021-11-09 12:50:39 -08:00
Brian Behlendorf de198f2d95
Fix lseek(SEEK_DATA/SEEK_HOLE) mmap consistency
When using lseek(2) to report data/holes memory mapped regions of
the file were ignored.  This could result in incorrect results.
To handle this zfs_holey_common() was updated to asynchronously
writeback any dirty mmap(2) regions prior to reporting holes.

Additionally, while not strictly required, the dn_struct_rwlock is
now held over the dirty check to prevent the dnode structure from
changing.  This ensures that a clean dnode can't be dirtied before
the data/hole is located.  The range lock is now also taken to
ensure the call cannot race with zfs_write().

Furthermore, the code was refactored to provide a dnode_is_dirty()
helper function which checks the dnode for any dirty records to
determine its dirtiness.

Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Rich Ercolani <rincebrain@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #11900
Closes #12724
2021-11-07 14:27:44 -07:00
Damian Szuberski 6d680e61ef
Update `checkstyle` workflow env to ubuntu-20.04
- `checkstyle` workflow uses ubuntu-20.04 environment
- improved `mancheck.sh` readability

Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: szubersk <szuberskidamian@gmail.com>
Closes #12713
2021-11-02 14:02:57 -06:00
Fedor Uporov d5a5ec4693
Remove unused function zvol_set_volblocksize()
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Fedor Uporov <fuporov.vstack@gmail.com>
Closes #12688
2021-10-26 17:07:53 -07:00
Pawel Jakub Dawidek afbc617921
Remove FreeBSD's local copy of the dmu_buf_hold_array() function
Make the main dmu_buf_hold_array() function non-static.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@ixsystems.com>
Signed-off-by: Pawel Jakub Dawidek <pawel@dawidek.net>
Closes #12628
2021-10-13 11:01:01 -07:00
Teodor Spæren d785245857
zio: use unsigned values for enum
cppcheck complains about the use of 1 << 31, because enums are signed
ints which cannot represent this. As discussed in issue #12611, it
appears that with C99, we can use an unsiged int for the enum, on most
platforms.

I've crafted this commit for just the include/sys/zio.h header, as it's
the only one with a shift of 31. If this is something we want to adopt
in the rest of the project, I will go through and apply it to the rest
of the project.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Signed-off-by: Teodor Spæren <teodor@sparen.no>
Closes #12611 
Closes #12615
2021-10-11 10:58:06 -07:00
Rich Ercolani 9d1407e8f2
Correct refcount_add in dmu_zfetch
refcount_add_many(foo,N) is not the same as
for (i=0; i < N; i++) { refcount_add(foo); }

Unfortunately, this is only actually true with debug kernels and
reference_tracking_enable=1.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes #12589 
Closes #12602
2021-10-08 11:10:34 -07:00
Brian Behlendorf 6954c22f35
Use fallthrough macro
As of the Linux 5.9 kernel a fallthrough macro has been added which
should be used to anotate all intentional fallthrough paths.  Once
all of the kernel code paths have been updated to use fallthrough
the -Wimplicit-fallthrough option will because the default.  To
avoid warnings in the OpenZFS code base when this happens apply
the fallthrough macro.

Additional reading: https://lwn.net/Articles/794944/

Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #12441
2021-09-14 10:17:54 -06:00
Jorgen Lundman 5a54a4e051
Upstream: Add snapshot and zvol events
For kernel to send snapshot mount/unmount events to zed.

For kernel to send symlink creates/removes on zvol plumbing.
(/dev/run/dsk/zvol/$pool/$zvol -> /dev/diskX)

If zed misses the ENODEV, all errors after are EINVAL. Treat any error
as kernel module failure.

Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Jorgen Lundman <lundman@lundman.net>
Closes #12416
2021-09-09 10:44:21 -07:00
Rich Ercolani b1a1c64313
Fix cross-endian interoperability of zstd
It turns out that layouts of union bitfields are a pain, and the
current code results in an inconsistent layout between BE and LE
systems, leading to zstd-active datasets on one erroring out on
the other.

Switch everyone over to the LE layout, and add compatibility code
to read both.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes #12008
Closes #12022
2021-08-30 14:13:46 -07:00
Trevor Bautista 00888c0898
Extend zpool-iostat to account for ZIO_PRIORITY_REBUILD (#12319)
Previously, zpool-iostat did not display any data regarding rebuild I/Os
in either the latency/size histograms (-w/-l/-r) or the queue data (-q).
This fix essentially utilizes the existing infrastructure for tracking
rebuild queue data and displays this data in the proper places within
zpool-iostat's output.

Signed-off-by: Trevor Bautista <tbautista@newmexicoconsortium.org>
Signed-off-by: Trevor Bautista <tbautista@lanl.gov>
Co-authored-by: Trevor Bautista <tbautista@newmexicoconsortium.org>
Reviewed-by: Richard Elling <Richard.Elling@RichardElling.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
2021-08-26 11:26:49 -07:00
Alexander Motin 6b88b4b501
Remove b_pabd/b_rabd allocation from arc_hdr_alloc()
When a header is allocated for full overwrite it is a waste of time
to allocate b_pabd/b_rabd for it, since arc_write() will free them
without ever being touched.  If it is a read or a partial overwrite
then arc_read() and arc_hdr_decrypt() allocate them explicitly.

Reduced memory allocation in user threads also reduces ARC eviction
throttling there, proportionally increasing it in ZIO threads, that
is not good.  To minimize or even avoid it introduce ARC allocation
reserve, allowing certain arc_get_data_abd() callers to allocate a
bit longer in situations where user threads will already throttle.

Reviewed-by: George Wilson <gwilson@delphix.com>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12398
2021-08-17 10:15:54 -06:00
Alexander Motin 72f0521aba
Increase default volblocksize from 8KB to 16KB
Many things has changed since previous default was set many years ago.
Nowadays 8KB does not allow adequate compression or even decent space
efficiency on many of pools due to 4KB disk physical block rounding,
especially on RAIDZ and DRAID.  It effectively limits write throughput
to only 2-3GB/s (250-350K blocks/s) due to sync thread, allocation,
vdev queue and other block rate bottlenecks.  It keeps L2ARC expensive
despite many optimizations and dedup just unrealistic.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes #12406
2021-08-17 09:59:46 -06:00
Alexander Motin cfe8e960f1
Fix/improve dbuf hits accounting
Instead of clearing stats inside arc_buf_alloc_impl() do it inside
arc_hdr_alloc() and arc_release().  It fixes statistics being wiped
every time a new dbuf is filled from the ARC.

Remove b_l1hdr.b_l2_hits. L2ARC hits are accounted at b_l2hdr.b_hits.
Since the hits are accounted under hash lock, replace atomics with
simple increments.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Wilson <george.wilson@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12422
2021-08-17 09:50:31 -06:00
Alexander Motin e829a865bf
Use more atomics in refcounts
Use atomic_load_64() for zfs_refcount_count() to prevent torn reads
on 32-bit platforms.  On 64-bit ones it should not change anything.

When built with ZFS_DEBUG but running without tracking enabled use
atomics instead of mutexes same as for builds without ZFS_DEBUG.
Since rc_tracked can't change live we can check it without lock.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12420
2021-08-17 09:44:34 -06:00
Allan Jude e945e8d7f4
Restore FreeBSD sysctl processing for arc.min and arc.max
Before OpenZFS 2.0, trying to set the FreeBSD sysctl vfs.zfs.arc_max
to a disallowed value would return an error.
Since the switch, it instead only generates WARN_IF_TUNING_IGNORED

Keep the ability to set the sysctl's specifically to 0, even though
that is less than the minimum, because some tests depend on this.

Also lost, was the ability to set vfs.zfs.arc_max to a value less
than the default vfs.zfs.arc_min at boot time. Restore this as well.

Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Reviewed-by: Ryan Moeller <ryan@ixsystems.com>
Signed-off-by: Allan Jude <allan@klarasystems.com>
Closes #12161
2021-08-16 09:35:19 -06:00
Tony Nguyen 6bc61d22c4
Run arc_evict thread at higher priority
Run arc_evict thread at higher priority, nice=0, to give it more CPU
time which can improve performance for workload with high ARC evict
activities.

On mixed read/write and sequential read workloads, I've seen between
10-40% better performance.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Tony Nguyen <tony.nguyen@delphix.com>
Closes #12397
2021-08-10 11:36:26 -06:00
Alexander Motin 7eebcd2be6
Avoid small buffer copying on write
It is wrong for arc_write_ready() to use zfs_abd_scatter_enabled to
decide whether to reallocate/copy the buffer, because the answer is
OS-specific and depends on the buffer size.  Instead of that use
abd_size_alloc_linear(), moved into public header.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes #12425
2021-07-27 16:05:47 -07:00
наб 037af3e0d4 Remove NOTE(CONSTCOND) and note.h
These were mostly used to annotate do {} while(0)s

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Issue #12201
2021-07-26 12:07:53 -07:00
наб 5dbf6c5a66 Replace /*PRINTFLIKEn*/ with attribute(printf)
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Issue #12201
2021-07-26 12:07:15 -07:00
Alexander Motin 1b50749ce9
Optimize allocation throttling
Remove mc_lock use from metaslab_class_throttle_*().  The math there
is based on refcounts and so atomic, so the only race possible there
is between zfs_refcount_count() and zfs_refcount_add().  But in most
cases metaslab_class_throttle_reserve() is called with the allocator
lock held, which covers the race.  In cases where the lock is not
held, GANG_ALLOCATION() or METASLAB_MUST_RESERVE are set, and so we
do not use zfs_refcount_count().  And even if we assume some other
non-existing scenario, the worst that may happen from this race is
few more I/Os get to allocation earlier, that is not a problem.

Move locks and data of different allocators into different cache
lines to avoid false sharing.  Group spa_alloc_* arrays together
into single array of aligned struct spa_alloc spa_allocs.  Align
struct metaslab_class_allocator.

Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Don Brady <don.brady@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12314
2021-07-21 06:40:36 -06:00
Kevin Jin a7bd20e309
Add Module Parameter Regarding Log Size Limit
* Add Module Parameters Regarding Log Size Limit

zfs_wrlog_data_max
The upper limit of TX_WRITE log data. Once it is reached,
write operation is blocked, until log data is cleared out
after txg sync. It only counts TX_WRITE log with WR_COPIED
or WR_NEED_COPY.

Reviewed-by: Prakash Surya <prakash.surya@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: jxdking <lostking2008@hotmail.com>
Closes #12284
2021-07-20 09:40:24 -06:00
Alexander Motin 8172df643b
Minor ARC optimizations
Remove unneeded global, practically constant, state pointer variables
(arc_anon, arc_mru, etc.), replacing them with macros of real state
variables addresses (&ARC_anon, &ARC_mru, etc.). 

Change ARC_EVICT_ALL from -1ULL to UINT64_MAX, not requiring special
handling in inner loop of ARC reclamation.  Respectively change bytes
argument of arc_evict_state() from int64_t to uint64_t.

Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes #12348
2021-07-20 08:13:21 -06:00
Alexander 23c13c7e80
A few fixes of callback typecasting (for the upcoming ClangCFI)
* zio: avoid callback typecasting
* zil: avoid zil_itxg_clean() callback typecasting
* zpl: decouple zpl_readpage() into two separate callbacks
* nvpair: explicitly declare callbacks for xdr_array()
* linux/zfs_nvops: don't use external iput() as a callback
* zcp_synctask: don't use fnvlist_free() as a callback
* zvol: don't use ops->zv_free() as a callback for taskq_dispatch()

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Alexander Lobakin <alobakin@pm.me>
Closes #12260
2021-07-20 08:03:33 -06:00
Alexander Motin c1b5869bab
Introduce dsl_dir_diduse_transfer_space()
Most of dsl_dir_diduse_space() and dsl_dir_transfer_space() CPU time
is a dd_lock overhead and time spent in dmu_buf_will_dirty(). Calling
them one after another is a waste of time and even more contention.
Doing that twice for each rewritten block within dbuf_write_done()
via dsl_dataset_block_kill() and dsl_dataset_block_born() created one
of the biggest CPU overheads in case of small blocks rewrite.

dsl_dir_diduse_transfer_space() combines functionality of these two
functions for cases where it is needed, but without double overhead,
practically for the cost of dsl_dir_diduse_space() or even cheaper.

While there, optimize dsl_dir_phys() calls in dsl_dir_diduse_space()
and dsl_dir_transfer_space().  It seems Clang detects some aliasing
there, repeating dd->dd_dbuf->db_data dereference multiple times,
increasing dd_lock scope and contention.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Author: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12300
2021-07-16 13:39:24 -06:00
Alexander Motin f7de776da2
Fix ARC ghost states eviction accounting
arc_evict_hdr() returns number of evicted bytes in scope of specific
state.  For ghost states it does not mean the amount of really freed
memory, but the logical buffer size.  It is correct for the eviction
process, but not for waking up threads waiting for ARC size reduction,
as added in "Revise ARC shrinker algorithm" commit, causing premature
wakeups while ARC is still overflowed, allowing even bigger overflow,
plus processing overhead when next allocation will also get blocked,
probably also for too short time.

To fix that make arc_evict_hdr() also return the amount of really
freed memory, which for the ghost states is only the header, and use
it to update arc_evict_count instead.  Originally I was thinking to
not return it at all, since arc_get_data_impl() does not account for
the headers, but decided that some slow allocation progress is better
than long waits, reaching on my tests up to 100ms.

To reduce negative latency effects of long time periods when reclaim
thread can free little real memory, start reclamation process earlier,
before we actually reached the overflow threshold, when we have to
throttle new allocations.  We can also do it without taking global
arc_evict_lock, reducing the contention.

Reviewed-by: George Wilson <gwilson@delphix.com>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12279
2021-07-13 09:41:59 -06:00
George Wilson 958826be7a
file reference counts can get corrupted
Callers of zfs_file_get and zfs_file_put can corrupt the reference
counts for the file structure resulting in a panic or a soft lockup.
When zfs send/recv runs, it will add a reference count to the
open file, and begin to send or recv the stream. If the file descriptor
is closed, then when dmu_recv_stream() or dmu_send() return we will
call zfs_file_put to remove the reference we placed on the file
structure. Unfortunately, because zfs_file_put() uses the file
descriptor to lookup the file structure, it may end up finding that
the file descriptor table no longer contains the file struct, thus
leaking the file structure. Or it might end up finding a file
descriptor for a different file and blindly updating its reference
counts. Other failure modes probably exists.

This change reworks the zfs_file_[get|put] interface to not rely
on the file descriptor but instead pass the zfs_file_t pointer around.

Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Co-authored-by: Allan Jude <allan@klarasystems.com>
Signed-off-by: George Wilson <gwilson@delphix.com>
External-issue: DLPX-76119
Closes #12299
2021-07-10 19:00:37 -06:00
Jorgen Lundman 03dba7ae31
dprintf_dnode: strcpy -> strlcpy
Missed a couple of strcpy() in earlier commit, this is only used with
--enable-debug.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: Jorgen Lundman <lundman@lundman.net>
Closes #12311
2021-07-07 21:13:40 -06:00
Alexander Motin bdd11cbb90
FreeBSD: Hardcode abd_chunk_size to PAGE_SIZE
It makes no sense to set it below PAGE_SIZE, since it increases all
overheads and makes returning memory to OS problematic.  It makes no
sense to set it above PAGE_SIZE, since such allocations and especially
frees are too expensive and cause KVA fragmentation to benefit from
fewer chunks.  After that it makes no sense to keep more complicated
math here.

What may have sense though is just a tunable border between linear and
scatter ABDs, previously also controlled by this tunable.  Retain that
functionality by taking abd_scatter_min_size tunable from Linux, just
with different default value.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes #12328
2021-07-06 17:39:23 -07:00
Alexander Motin b192a2c0a1
Remove avl_size field from struct avl_tree
This field is used only by illumos mdb.  On other platforms it only
increases the struct size from 32 to 40 bytes.  For struct vdev_queue
including 13 instances of avl_tree_t size means active cache lines.

Keep the padding in user-space for now to not break the ABI.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12290
2021-07-01 09:32:31 -06:00
Alexander Motin 490c845efe
Compact dbuf/buf hashes and lock arrays
With default dbuf cache size of 1/32 of ARC, it makes no sense to have
hash table of the same size (or even bigger on Linux).  Reduce it to
1/8 of ARC's one, still leaving some slack, assuming higher I/O rate
via dbuf cache than via ARC.

Remove padding from ARC hash locks array.  The idea behind padding
is to avoid false sharing between locks.  It would have sense if
there would be a limited number of very busy locks.  But since we
have no limit on the number, using the same memory for more locks we
can achieve even lower lock contention with the same false sharing,
or we can use less memory for the same contention level.

Reduce number of hash locks from 8192 to 2048.  The number is still
big enough to not cause contention, but reduced memory size improves
cache hit rate for mutex_tryenter() in ARC eviction thread, saving
about 1% of the thread time.

Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12289
2021-07-01 09:30:31 -06:00
Jorgen Lundman c6d1112bf4
Fix abd leak, kmem_free correct size of abd_t
Fix a leak of abd_t that manifested mostly when using
raidzN with at least as many columns as N (e.g. a
four-disk raidz2 but not a three-disk raidz2).
Sufficiently heavy raidz use would eventually run a system
out of memory.

Additionally:

* Switch abd_cache arena to FIRSTFIT, which empirically
improves perofrmance.

* Make abd_chunk_cache more performant and debuggable.

* Allocate the abd_zero_buf from abd_chunk_cache rather
than the heap.

* Don't try to reap non-existent qcaches in abd_cache arena.

* KM_PUSHPAGE->KM_SLEEP when allocating chunks from their
own arena

Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Jorgen Lundman <lundman@lundman.net>
Co-authored-by: Sean Doran <smd@use.net>
Closes #12295
2021-07-01 09:28:15 -06:00
Kevin Jin 50e09eddd0
Optimize txg_kick() process (#12274)
Use dp_dirty_pertxg[] for txg_kick(), instead of dp_dirty_total in
original code. Extra parameter "txg" is added for txg_kick(), thus it
knows which txg to kick. Also txg_kick() call is moved from
dsl_pool_need_dirty_delay() to dsl_pool_dirty_space() so that we can
know the txg number assigned for txg_kick().

Some unnecessary code regarding dp_dirty_total in txg_sync_thread() is
also cleaned up.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: jxdking <lostking2008@hotmail.com>
Closes #12274
2021-07-01 09:20:27 -06:00
Alexander Motin 42afb12da7
Remove refcount from spa_config_*()
The only reason for spa_config_*() to use refcount instead of simple
non-atomic (thanks to scl_lock) variable for scl_count is tracking,
hard disabled for the last 8 years.  Switch to simple int scl_count
reduces the lock hold time by avoiding atomic, plus makes structure
fit into single cache line, reducing the locks contention.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes #12287
2021-07-01 09:16:54 -06:00
Martin Matuška 14d2841b53
FreeBSD: fix compilation of FreeBSD world after 29274c9f6
prng32_bounded() is available to kernel only on FreeBSD 13+.

Call inline random_get_pseudo_bytes() with correct pointer type.
To be consistent, apply to Linux as well.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Martin Matuska <mm@FreeBSD.org>
Closes #12282
2021-06-25 10:28:51 -07:00