aboutsummaryrefslogtreecommitdiffstats
path: root/module/zfs
Commit message (Collapse)AuthorAgeFilesLines
* nowait synctask must succeedMatthew Ahrens2020-09-049-39/+24
| | | | | | | | | | | | | | If a `zfs_space_check_t` other than `ZFS_SPACE_CHECK_NONE` is used with `dsl_sync_task_nowait()`, the sync task may fail due to ENOSPC. However, there is no way to notice or communicate this failure, so it's extremely difficult to use this functionality correctly, and in fact almost all callers use `ZFS_SPACE_CHECK_NONE`. This commit removes the `zfs_space_check_t` argument from `dsl_sync_task_nowait()`, and always uses `ZFS_SPACE_CHECK_NONE`. Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matthew Ahrens <[email protected]> Closes #10855
* Retain thread name when resuming a zthrRyan Moeller2020-09-031-3/+8
| | | | | | | | | | | | When created, a zthr is given a name to identify it by. This name is lost when a cancelled zthr is resumed. Retain the name of a zthr so it can be used when resuming. Reviewed-by: Serapheim Dimitropoulos <[email protected]> Reviewed-by: Alexander Motin <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Ryan Moeller <[email protected]> Closes #10881
* Replace cv_{timed}wait_sig with cv_{timed}wait_idle where appropriateMatthew Macy2020-09-036-19/+10
| | | | | | | | | | | | | | | | There are a number of places where cv_?_sig is used simply for accounting purposes but the surrounding code has no ability to cope with actually receiving a signal. On FreeBSD it is possible to send signals to individual kernel threads so this could enable undesirable behavior. This patch adds routines on Linux that will do the same idle accounting as _sig without making the task interruptible. On FreeBSD cv_*_idle are all aliases for cv_* Reviewed-by: Alexander Motin <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10843
* Make spa_stats.c tunables visible on FreeBSDRyan Moeller2020-09-011-18/+12
| | | | | | | | Use ZFS_MODULE_PARAM for cross-platform tunables in spa_stats.c, and add update tunables.cfg in tests for the newly supported ones. Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Ryan Moeller <[email protected]> Closes #10858
* FreeBSD: Fix up after spa_stats.c moveMatthew Macy2020-09-011-1/+3
| | | | | | | | | | | | | | Moving spa_stats added the additional burden of supporting KSTAT_TYPE_IO. spa_state_addr will always return a valid value regardless of the value of 'n'. On FreeBSD this will cause an infinite loop as it relies on the raw ops addr routine to indicate that there is no more data. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Ryan Moeller <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10860
* Add 'zfs rename -u' to rename without remountingRyan Moeller2020-09-012-4/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | Allow to rename file systems without remounting if it is possible. It is possible for file systems with 'mountpoint' property set to 'legacy' or 'none' - we don't have to change mount directory for them. Currently such file systems are unmounted on rename and not even mounted back. This introduces layering violation, as we need to update 'f_mntfromname' field in statfs structure related to mountpoint (for the dataset we are renaming and all its children). In my opinion it is worth it, as it allow to update FreeBSD in even cleaner way - in ZFS-only configuration root file system is ZFS file system with 'mountpoint' property set to 'legacy'. If root dataset is named system/rootfs, we can snapshot it (system/rootfs@upgrade), clone it (system/oldrootfs), update FreeBSD and if it doesn't boot we can boot back from system/oldrootfs and rename it back to system/rootfs while it is mounted as /. Before it was not possible, because unmounting / was not possible. Authored by: Pawel Jakub Dawidek <[email protected]> Reviewed-by: Allan Jude <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Ported by: Matt Macy <[email protected]> Signed-off-by: Ryan Moeller <[email protected]> Closes #10839
* zio_ereport_post() and zio_ereport_start() return values are ignoredToomas Soome2020-08-318-20/+24
| | | | | | | | use (void) to silence analyzers. Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Toomas Soome <[email protected]> Closes #10857
* Move spa_stats.c to common codeMatthew Macy2020-08-302-0/+1048
| | | | | | | | | | | Initially it was considered simplest to stub out all of the functions on FreeBSD. Now that FreeBSD supports KSTAT_TYPE_RAW at least some of the functionality should be made available. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Richard Elling <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10842
* dnode_sync is careless with range treePatrick Mooney2020-08-261-2/+12
| | | | | | | | | | | | | | Because dnode_sync_free_range() must drop dn_mtx during its processing, using it as a callback to range_tree_vacate() is not safe. No other operations (besides destroy) are allowed once range_tree_vacate() has begun, and dropping dn_mtx would leave a window open for another thread to observe that invalid (and unsafe) state via dnode_block_freed(). Reviewed-by: Matthew Ahrens <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Igor Kozhukhov <[email protected]> Signed-off-by: Patrick Mooney <[email protected]> Closes #10708 Closes #10823
* zpool: Change base URL for ZFS messages to openzfs-docsRyan Moeller2020-08-261-1/+2
| | | | | | | Reviewed-by: George Melikov <[email protected]> Reviewed-by: Kjeld Schouten <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Ryan Moeller <[email protected]> Closes #10820
* Remove duplicate dnode.h includeBrian Behlendorf2020-08-261-1/+0
| | | | | | | | | The zfs/sa.c source file accidentally includes sys/dnode.h twice. Remove the second occurrence. Reviewed-by: Matthew Ahrens <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #10816 Closes #10819
* Always track temporary fses and snapshots for accountingPaul Dagnelie2020-08-261-17/+6
| | | | | | | | | | | | | | | | The root cause of the issue is that we only occasionally do as the comments in the code suggest and actually ignore the %recv dataset when it comes to filesystem limit tracking. Specifically, the only time we ignore it is when initializing the filesystem and snapshot limit values; when creating a new %recv dataset or deleting one, we always update the bookkeeping. This causes a problem if you init the fs count on a filesystem that already has a %recv dataset, since the bookmarking will be decremented but not incremented. This is resolved in this patch by simply always tracking the %recv dataset as a child. Reviewed-by: Matt Ahrens <[email protected]> Reviewed by: Jerry Jelinek <[email protected]> Signed-off-by: Paul Dagnelie <[email protected]> Closes #10791
* FreeBSD: disable neon usageMatthew Macy2020-08-261-1/+1
| | | | | | | | | | The neon support code does not build on FreeBSD, ifdef out references to fix linker issues on arm64. Reviewed-by: George Melikov <[email protected]> Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10809
* Introduce limit on size of L2ARC headersAlexander Motin2020-08-251-5/+22
| | | | | | | | | | | | | | | | | Since L2ARC buffers are not evicted on memory pressure, too large amount of headers on system with irrationally large L2ARC can render it slow or even unusable. This change limits L2ARC writes and rebuild if unevictable L2ARC-only headers reach dangerous level. While there, call arc_adapt() on L2ARC rebuild, so that it could properly grow arc_c, reflecting potentially significant ARC size increase and avoiding slow growth with hopeless eviction attempts later when "overflow" is detected. Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Reported-by: Richard Elling <[email protected]> Signed-off-by: Alexander Motin <[email protected]> Closes #10765
* Export dmu_offset_next() symbolBrian Behlendorf2020-08-251-0/+1
| | | | | | | | Export the dmu_offset_next() symbol for use by Lustre. Reviewed-by: George Melikov <[email protected]> Reviewed-by: Ryan Moeller <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #10796
* Avoid symbol collision with in-kernel zstdlibSebastian Gottschall2020-08-241-2/+2
| | | | | | | | | | | For Linux, when zfs is compiled as an in kernel static variant and the in kernel zstd library is compiled statically into the kernel a symbol collision will occur. This wrapper header renames all of the relevant zstd functions to avoid this problem. Reviewed-by: Kjeld Schouten <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Sebastian Gottschall <[email protected]> Closes #10775
* Import vdev ashift optimization from FreeBSDRyan Moeller2020-08-219-21/+92
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Many modern devices use physical allocation units that are much larger than the minimum logical allocation size accessible by external commands. Two prevalent examples of this are 512e disk drives (512b logical sector, 4K physical sector) and flash devices (512b logical sector, 4K or larger allocation block size, and 128k or larger erase block size). Operations that modify less than the physical sector size result in a costly read-modify-write or garbage collection sequence on these devices. Simply exporting the true physical sector of the device to ZFS would yield optimal performance, but has two serious drawbacks: 1. Existing pools created with devices that have different logical and physical block sizes, but were configured to use the logical block size (e.g. because the OS version used for pool construction reported the logical block size instead of the physical block size) will suddenly find that the vdev allocation size has increased. This can be easily tolerated for active members of the array, but ZFS would prevent replacement of a vdev with another identical device because it now appears that the smaller allocation size required by the pool is not supported by the new device. 2. The device's physical block size may be too large to be supported by ZFS. The optimal allocation size for the vdev may be quite large. For example, a RAID controller may export a vdev that requires read-modify-write cycles unless accessed using 64k aligned/sized requests. ZFS currently has an 8k minimum block size limit. Reporting both the logical and physical allocation sizes for vdevs solves these problems. A device may be used so long as the logical block size is compatible with the configuration. By comparing the logical and physical block sizes, new configurations can be optimized and administrators can be notified of any existing pools that are sub-optimal. Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Co-authored-by: Matthew Macy <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10619
* Fix indentation in dnode_free_range()Matthew Ahrens2020-08-201-7/+7
| | | | | | Reviewed-by: George Melikov <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matthew Ahrens <[email protected]> Closes #10744
* FreeBSD: 11.x arc_stats compatibilityMatthew Macy2020-08-201-0/+8
| | | | | | | | | Removing other_size from arc_stats breaks top in 11.x jails running on HEAD. Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10745
* Add zstd support to zfsMichael Niewöhner2020-08-2012-67/+311
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This PR adds two new compression types, based on ZStandard: - zstd: A basic ZStandard compression algorithm Available compression. Levels for zstd are zstd-1 through zstd-19, where the compression increases with every level, but speed decreases. - zstd-fast: A faster version of the ZStandard compression algorithm zstd-fast is basically a "negative" level of zstd. The compression decreases with every level, but speed increases. Available compression levels for zstd-fast: - zstd-fast-1 through zstd-fast-10 - zstd-fast-20 through zstd-fast-100 (in increments of 10) - zstd-fast-500 and zstd-fast-1000 For more information check the man page. Implementation details: Rather than treat each level of zstd as a different algorithm (as was done historically with gzip), the block pointer `enum zio_compress` value is simply zstd for all levels, including zstd-fast, since they all use the same decompression function. The compress= property (a 64bit unsigned integer) uses the lower 7 bits to store the compression algorithm (matching the number of bits used in a block pointer, as the 8th bit was borrowed for embedded block pointers). The upper bits are used to store the compression level. It is necessary to be able to determine what compression level was used when later reading a block back, so the concept used in LZ4, where the first 32bits of the on-disk value are the size of the compressed data (since the allocation is rounded up to the nearest ashift), was extended, and we store the version of ZSTD and the level as well as the compressed size. This value is returned when decompressing a block, so that if the block needs to be recompressed (L2ARC, nop-write, etc), that the same parameters will be used to result in the matching checksum. All of the internal ZFS code ( `arc_buf_hdr_t`, `objset_t`, `zio_prop_t`, etc.) uses the separated _compress and _complevel variables. Only the properties ZAP contains the combined/bit-shifted value. The combined value is split when the compression_changed_cb() callback is called, and sets both objset members (os_compress and os_complevel). The userspace tools all use the combined/bit-shifted value. Additional notes: zdb can now also decode the ZSTD compression header (flag -Z) and inspect the size, version and compression level saved in that header. For each record, if it is ZSTD compressed, the parameters of the decoded compression header get printed. ZSTD is included with all current tests and new tests are added as-needed. Per-dataset feature flags now get activated when the property is set. If a compression algorithm requires a feature flag, zfs activates the feature when the property is set, rather than waiting for the first block to be born. This is currently only used by zstd but can be extended as needed. Portions-Sponsored-By: The FreeBSD Foundation Co-authored-by: Allan Jude <[email protected]> Co-authored-by: Brian Behlendorf <[email protected]> Co-authored-by: Sebastian Gottschall <[email protected]> Co-authored-by: Kjeld Schouten-Lebbing <[email protected]> Co-authored-by: Michael Niewöhner <[email protected]> Signed-off-by: Allan Jude <[email protected]> Signed-off-by: Allan Jude <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Signed-off-by: Sebastian Gottschall <[email protected]> Signed-off-by: Kjeld Schouten-Lebbing <[email protected]> Signed-off-by: Michael Niewöhner <[email protected]> Closes #6247 Closes #9024 Closes #10277 Closes #10278
* Fix ARC aggsum access after arc_state_fini()Brian Behlendorf2020-08-181-8/+8
| | | | | | | | | | | | | | Commit 85ec5cbae updated abd_update_scatter_stats() such that it calls arc_space_consume() and arc_space_return() when updating the scatter stats. This requires that the global aggsum value for the ARC be initialized. Normally this is not an issue, however during module unload the l2arc_do_free_on_write() function was called in l2arc_cleanup() after arc_state_fini() destroyed the aggsum values. We can resolve this issue by performing l2arc_do_free_on_write() slightly earlier in arc_fini(). Reviewed-by: Matthew Ahrens <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #10739
* FreeBSD: Fix UNIX permissions checkingMatthew Macy2020-08-181-10/+31
| | | | | | Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10727
* Make zc_nvlist_src_size limit tunableRyan Moeller2020-08-181-2/+10
| | | | | | | | | | | | | | | | | | We limit the size of nvlists passed to the kernel so a user cannot make the kernel do an unreasonably large allocation. On FreeBSD this limit was 128 kiB, which turns out to be a bit too small when doing some operations involving a large number of datasets or snapshots, for example replication. Make this limit tunable, with a platform-specific auto default. Linux keeps its limit at KMALLOC_MAX_SIZE. FreeBSD uses 1/4 of the system limit on user wired memory, which allows it to scale depending on system configuration. Reviewed-by: Matt Macy <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Ryan Moeller <[email protected]> Issue #6572 Closes #10706
* Remove GRUB restrictionsRichard Laager2020-08-172-59/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The GRUB restrictions are based around the pool's bootfs property. Given the current situation where GRUB is not staying current with OpenZFS pool features, having either a non-ZFS /boot or a separate pool with limited features are pretty much the only long-term answers for GRUB support. Only the second case matters in this context. For the restrictions to be useful, the bootfs property would have to be set on the boot pool, because that is where we need the restrictions, as that is the pool that GRUB reads from. The documentation for bootfs describes it as pointing to the root pool. That's also how it's used in the initramfs. ZFS does not allow setting bootfs to point to a dataset in another pool. (If it did, it'd be difficult-to-impossible to enforce these restrictions cross-pool). Accordingly, bootfs is pretty much useless for GRUB scenarios moving forward. Even for users who have only one pool, the existing restrictions for GRUB are incomplete. They don't prevent you from enabling the unsupported checksums, for example. For that reason, I have ripped out all the GRUB restrictions. A little longer-term, I think extending the proposed features=portable system to define a features=grub is a much more useful approach. The user could set that on the boot pool at creation, and things would Just Work. Reviewed-by: Paul Dagnelie <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Richard Laager <[email protected]> Closes #8627
* Include scatter_chunk_waste in arc_sizeMatthew Ahrens2020-08-171-2/+20
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The ARC caches data in scatter ABD's, which are collections of pages, which are typically 4K. Therefore, the space used to cache each block is rounded up to a multiple of 4K. The ABD subsystem tracks this wasted memory in the `scatter_chunk_waste` kstat. However, the ARC's `size` is not aware of the memory used by this round-up, it only accounts for the size that it requested from the ABD subsystem. Therefore, the ARC is effectively using more memory than it is aware of, due to the `scatter_chunk_waste`. This impacts observability, e.g. `arcstat` will show that the ARC is using less memory than it effectively is. It also impacts how the ARC responds to memory pressure. As the amount of `scatter_chunk_waste` changes, it appears to the ARC as memory pressure, so it needs to resize `arc_c`. If the sector size (`1<<ashift`) is the same as the page size (or larger), there won't be any waste. If the (compressed) block size is relatively large compared to the page size, the amount of `scatter_chunk_waste` will be small, so the problematic effects are minimal. However, if using 512B sectors (`ashift=9`), and the (compressed) block size is small (e.g. `compression=on` with the default `volblocksize=8k` or a decreased `recordsize`), the amount of `scatter_chunk_waste` can be very large. On a production system, with `arc_size` at a constant 50% of memory, `scatter_chunk_waste` has been been observed to be 10-30% of memory. This commit adds `scatter_chunk_waste` to `arc_size`, and adds a new `waste` field to `arcstat`. As a result, the ARC's memory usage is more observable, and `arc_c` does not need to be adjusted as frequently. Reviewed-by: Pavel Zakharov <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: George Wilson <[email protected]> Reviewed-by: Ryan Moeller <[email protected]> Signed-off-by: Matthew Ahrens <[email protected]> Closes #10701
* FreeBSD: fix the build with Clang 11Ryan Moeller2020-08-176-2/+12
| | | | | | | | | | | | * Cast void * to uintptr_t before casting to boolean_t. * Avoid clashing definition of __asm when not on Linux to prevent duplicate __volatile__. This was already done in some places but not all. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Matt Macy <[email protected]> Signed-off-by: Ryan Moeller <[email protected]> Closes #10723
* Fix typo in btree.cSerapheim Dimitropoulos2020-08-171-2/+2
| | | | | | | Reviewed-by: George Melikov <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Matthew Ahrens <[email protected]> Signed-off-by: Serapheim Dimitropoulos <[email protected]> Closes #10725
* FreeBSD: fallback to /boot/ to look for zpool.cacheMatthew Macy2020-08-171-0/+4
| | | | | | | | | Up until now zpool.cache has always lived in /boot on FreeBSD. For the sake of compatibility fallback to /boot if zpool.cache isn't found in /etc/zfs. Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10720
* Fix l2arc_dev_rebuild_start thread nameRyan Moeller2020-08-171-4/+5
| | | | | | | | | | | | | | | `thread_create` on FreeBSD stringifies the argument passed as the thread function to create a name for the thread. The thread name for `l2arc_dev_rebuild_start` ended up with `(void (*)(void *))` in it. Change the type signature so the function does not need to be cast when creating the thread. Rename the function to `l2arc_dev_rebuild_thread` for clarity and consistency, as well. Reviewed-by: Alexander Motin <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: George Amanakis <[email protected]> Signed-off-by: Ryan Moeller <[email protected]> Closes #10716
* Fix L2ARC reads when compressed ARC disabledAllan Jude2020-08-131-0/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When reading compressed blocks from the L2ARC, with compressed ARC disabled, arc_hdr_size() returns LSIZE rather than PSIZE, but the actual read is PSIZE. This causes l2arc_read_done() to compare the checksum against the wrong size, resulting in checksum failure. This manifests as an increase in the kstat l2_cksum_bad and the read being retried from the main pool, making the L2ARC ineffective. Add new L2ARC tests with Compressed ARC enabled/disabled Blocks are handled differently depending on the state of the zfs_compressed_arc_enabled tunable. If a block is compressed on-disk, and compressed_arc is enabled: - the block is read from disk - It is NOT decompressed - It is added to the ARC in its compressed form - l2arc_write_buffers() may write it to the L2ARC (as is) - l2arc_read_done() compares the checksum to the BP (compressed) However, if compressed_arc is disabled: - the block is read from disk - It is decompressed - It is added to the ARC (uncompressed) - l2arc_write_buffers() will use l2arc_apply_transforms() to recompress the block, before writing it to the L2ARC - l2arc_read_done() compares the checksum to the BP (compressed) - l2arc_read_done() will use l2arc_untransform() to uncompress it This test writes out a test file to a pool consisting of one disk and one cache device, then randomly reads from it. Since the arc_max in the tests is low, this will feed the L2ARC, and result in reads from the L2ARC. We compare the value of the kstat l2_cksum_bad before and after to determine if any blocks failed to survive the trip through the L2ARC. Sponsored-by: The FreeBSD Foundation Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Allan Jude <[email protected]> Closes #10693
* Release onexit/events with any missed zfsdev_stateJorgen Lundman2020-08-131-7/+8
| | | | | | | | | | | | | | Linux and FreeBSD will most likely never see this issue. On macOS when kext is unloaded, but zed is still connected, zed will be issued ENODEV. As the cdevsw is released, the kernel will not have zfsdev_release() called to release minor/onexit/events, and it "leaks". This ensures it is cleaned up before unload. Changed the for loop from zsprev, to zsnext style, for less code duplication. Signed-off-by: Brian Behlendorf <[email protected]> Signed-off-by: Jorgen Lundman <[email protected]> Closes #10700
* Use zfs_dbgmsg to log metaslab_load/unloadMatthew Ahrens2020-08-121-19/+34
| | | | | | | | | | | | | Metaslabs are now (usually) loaded and unloaded infrequently, but when that is not the case, it is useful to have a log of when and why these events happened. This commit enables the zfs_dbgmsg() in metaslab_load(), and adds a zfs_dbgmsg() in metaslab_unload(). Reviewed-by: Serapheim Dimitropoulos <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matthew Ahrens <[email protected]> Closes #10683
* Restore ARC MFU/MRU pressureMatthew Macy2020-08-121-22/+42
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | The arc_adapt() function tunes LRU/MLU balance according to 4 types of cache hits (which is passed as state agrument): ghost LRU, LRU, MRU, ghost MRU. If this function is called with wrong cache hit (state), adaptation will be sub-optimal and performance will suffer. Some time ago upstream received this commit: 6950 ARC should cache compressed data) in arc_read() do next sequence (access to ghost buffer) Before this commit, hit to any ghost list was passed arc_adapt() before call to arc_access() which revive element in cache and change state from ghost to real hit. After this commit, the order of calls was reverted and arc_adapt() is now called only with «real» hits even if hit was in one of two ghost lists, which renders ghost lists useless and breaks the ARC algorithm. FreeBSD fixed this problem locally in Change D19094 / Commit r348772. This change is an adaptation of the above commit to the current arc code. Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10548 Closes #10618
* Fix typoAllan Jude2020-08-111-1/+1
| | | | | | Reviewed-by: George Melikov <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Allan Jude <[email protected]> Closes #10694
* Clarify error message when a range-tree double-add occursPaul Dagnelie2020-08-071-8/+22
| | | | | | | | | | | | | In various other pieces of logic have resulted in situations where we double-free space in ZFS. This in turn results in a double-add to the range trees. These issues have been much more difficult to diagnose than they should have been, because the error handling around this case is much weaker than around the double remove case. Reviewed-by: Matt Ahrens <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: George Wilson <[email protected]> Signed-off-by: Paul Dagnelie <[email protected]> Closes #10654
* Fix i/o error handling of livelists and zap iterationMatthew Ahrens2020-08-054-69/+91
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Pool-wide metadata is stored in the MOS (Meta Object Set). This metadata is stored in triplicate, in addition to any pool-level reduncancy (e.g. RAIDZ). However, if all 3+ copies of this metadata are not available, we can still get EIO/ECKSUM when reading from the MOS. If we encounter such an error in syncing context, we have typically already committed to making a change that we now can't do because of the corrupt/missing metadata. We typically "handle" this with a `VERIFY()` or `zfs_panic_recover()`. This prevents the system from continuing on in an undefined state, while minimizing the amount of error-handling code. However, there are some code paths that ignore these i/o errors, or `ASSERT()` that they don't happen. Since assertions are disabled on non-debug builds, they effectively ignore them as well. This can lead to ZFS continuing on in an incorrect state, potentially leading to on-disk inconsistencies. This commit adds handling for these i/o errors on MOS metadata, typically with a `VERIFY()`: * Handle error return from `zap_cursor_retrieve()` in 4 places in `dsl_deadlist.c`. * Handle error return from `zap_contains()` in `dsl_dir_hold_obj()`. Turns out this call isn't necessary because we can always call `zap_lookup()`. * Handle error return from `zap_lookup()` in `dsl_fs_ss_limit_check()`. * Handle error return from `zap_remove()` in `dsl_dir_rename_sync()`. * Handle error return from `zap_lookup()` in `dsl_dir_remove_livelist()`. * Handle error return from `dsl_process_sub_livelist()` in `spa_livelist_delete_cb()`. Additionally: * Augment the internal history log message for `zfs destroy` to note which method is used (e.g. bptree, livelist, or, synchronous) and the mintxg. * Correct a comment in `dbuf_init()`. * Correct indentation in `dsl_dir_remove_livelist()`. Reviewed by: Sara Hartse <[email protected]> Reviewed-by: George Wilson <[email protected]> Signed-off-by: Matthew Ahrens <[email protected]> Closes #10643
* Add missed thread_exit() to vdev_{autotrim,rebuild}_threadMatthew Macy2020-08-052-0/+4
| | | | | | | Reviewed-by: Jorgen Lundman <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Co-authored-by: Ryan Moeller <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10668
* Fix logging in l2arc_rebuild()George Amanakis2020-08-011-0/+7
| | | | | | | | | | | | | | | | | | | | | In case the L2ARC rebuild was canceled, do not log to spa history log as the pool may be in the process of being removed and a panic may occur: BUG: kernel NULL pointer dereference, address: 0000000000000018 RIP: 0010:spa_history_log_internal+0xb1/0x120 [zfs] Call Trace: l2arc_rebuild+0x464/0x7c0 [zfs] l2arc_dev_rebuild_start+0x2d/0x130 [zfs] ? l2arc_rebuild+0x7c0/0x7c0 [zfs] thread_generic_wrapper+0x78/0xb0 [spl] kthread+0xfb/0x130 ? IS_ERR+0x10/0x10 [spl] ? kthread_park+0x90/0x90 ret_from_fork+0x35/0x40 Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: George Amanakis <[email protected]> Closes #10659
* Change the error handling for invalid property valuesAllan Jude2020-08-011-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | ZFS recv should return a useful error message when an invalid index property value is provided in the send stream properties nvlist With a compression= property outside of the understood range: Before: ``` receiving full stream of zof/zstd_send@send2 into testpool/recv@send2 internal error: Invalid argument Aborted (core dumped) ``` Note: the recv completes successfully, the abort() is likely just to make it easier to track the unexpected error code. After: ``` receiving full stream of zof/zstd_send@send2 into testpool/recv@send2 cannot receive compression property on testpool/recv: invalid property value received 28.9M stream in 1 seconds (28.9M/sec) ``` Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Allan Jude <[email protected]> Closes #10631
* Changes to make openzfs build within FreeBSD buildworldMatthew Macy2020-07-311-0/+6
| | | | | | | | | A collection of header changes to enable FreeBSD to build with vendored OpenZFS. Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10635
* Revise ARC shrinker algorithmMatthew Ahrens2020-07-311-78/+174
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The ARC shrinker callback `arc_shrinker_count/_scan()` is invoked by the kernel's shrinker mechanism when the system is running low on free pages. This happens via 2 code paths: 1. "direct reclaim": The system is attempting to allocate a page, but we are low on memory. The ARC shrinker callback is invoked from the page-allocation code path. 2. "indirect reclaim": kswapd notices that there aren't many free pages, so it invokes the ARC shrinker callback. In both cases, the kernel's shrinker code requests that the ARC shrinker callback release some of its cache, and then it measures how many pages were released. However, it's measurement of released pages does not include pages that are freed via `__free_pages()`, which is how the ARC releases memory (via `abd_free_chunks()`). Rather, the kernel shrinker code is looking for pages to be placed on the lists of reclaimable pages (which is separate from actually-free pages). Because the kernel shrinker code doesn't detect that the ARC has released pages, it may call the ARC shrinker callback many times, resulting in the ARC "collapsing" down to `arc_c_min`. This has several negative impacts: 1. ZFS doesn't use RAM to cache data effectively. 2. In the direct reclaim case, a single page allocation may wait a long time (e.g. more than a minute) while we evict the entire ARC. 3. Even with the improvements made in 67c0f0dedc5 ("ARC shrinking blocks reads/writes"), occasionally `arc_size` may stay above `arc_c` for the entire time of the ARC collapse, thus blocking ZFS read/write operations in `arc_get_data_impl()`. To address these issues, this commit limits the ways that the ARC shrinker callback can be used by the kernel shrinker code, and mitigates the impact of arc_is_overflowing() on ZFS read/write operations. With this commit: 1. We limit the amount of data that can be reclaimed from the ARC via the "direct reclaim" shrinker. This limits the amount of time it takes to allocate a single page. 2. We do not allow the ARC to shrink via kswapd (indirect reclaim). Instead we rely on `arc_evict_zthr` to monitor free memory and reduce the ARC target size to keep sufficient free memory in the system. Note that we can't simply rely on limiting the amount that we reclaim at once (as for the direct reclaim case), because kswapd's "boosted" logic can invoke the callback an unlimited number of times (see `balance_pgdat()`). 3. When `arc_is_overflowing()` and we want to allocate memory, `arc_get_data_impl()` will wait only for a multiple of the requested amount of data to be evicted, rather than waiting for the ARC to no longer be overflowing. This allows ZFS reads/writes to make progress even while the ARC is overflowing, while also ensuring that the eviction thread makes progress towards reducing the total amount of memory used by the ARC. 4. The amount of memory that the ARC always tries to keep free for the rest of the system, `arc_sys_free` is increased. 5. Now that the shrinker callback is able to provide feedback to the kernel's shrinker code about our progress, we can safely enable the kswapd hook. This will allow the arc to receive notifications when memory pressure is first detected by the kernel. We also re-enable the appropriate kstats to track these callbacks. Reviewed-by: Alexander Motin <[email protected]> Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Co-authored-by: George Wilson <[email protected]> Signed-off-by: Matthew Ahrens <[email protected]> Closes #10600
* Remove duplicate include of sys/zfeature.h in dmu_objset.cAllan Jude2020-07-311-1/+0
| | | | | | Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Allan Jude <[email protected]> Closes #10636
* zfs promote does not delete livelist of originMatthew Ahrens2020-07-311-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When a clone is promoted, its livelist is no longer accurate, so it is discarded. If the clone's origin is also a clone (i.e. we are promoting a clone of a clone), then the origin's livelist is also no longer accurate, so it should be discarded, but the code doesn't actually do that. Consider a pool with: * Filesystem A * Clone B, a clone of A * Clone C, a clone of B If we promote C, it discards C's livelist. It should discard B's livelist, but that is not happening. The impact is that when B is destroyed, we use the livelist to find the blocks to free, but the livelist is no longer correct so we end up freeing blocks that are still in use by C. The incorrectly-freed blocks can be reallocated causing checksum errors. And when C is destroyed it can double-free the incorrectly-freed blocks. The problem is that we remove the livelist of `origin_ds->ds_dir`, but the origin snapshot has already been moved to the promoted dsl_dir. So this is actually trying to remove the livelist of the promoted dsl_dir, which was already removed. As explained in a comment in the beginning of `dsl_dataset_promote_sync()`, we need to use the saved `odd` for the origin's dsl_dir. Reviewed-by: Pavel Zakharov <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: George Wilson <[email protected]> Reviewed by: Sara Hartse <[email protected]> Signed-off-by: Matthew Ahrens <[email protected]> Closes #10652
* Fix error handling of vdev_top_zapMatthew Ahrens2020-07-291-2/+10
| | | | | | | | | | | | | | | | | | | | | | | In `vdev_load()`, we look up several entries in the `vdev_top_zap` object. In most cases, if we encounter an i/o error, it will be returned to the caller. However, when handling `VDEV_TOP_ZAP_ALLOCATION_BIAS`, if we get an i/o error, we may continue on, which in theory could cause us to not realize that a vdev should be used only for `special` allocations. In practice, if we encountered an i/o error while looking for `VDEV_TOP_ZAP_ALLOCATION_BIAS` in the `vdev_top_zap`, we'd also get an i/o error while looking for other entries in the same object, and thus the zpool open/import would fail. Therefore the impact of this problem is negligible. This commit adds error handling for i/o errors while accessing the `vdev_top_zap`, so that we aren't relying on unrelated code to fail for us. Reviewed-by: Don Brady <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matthew Ahrens <[email protected]> Closes #10637
* Rename refcount.h to zfs_refcount.hMatthew Macy2020-07-2913-13/+3
| | | | | | | | | Renamed to avoid conflicting with refcount.h when a different implementation is already provided by the platform. Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10620
* Introduce names for ZTHRsSerapheim Dimitropoulos2020-07-294-16/+24
| | | | | | | | | | | | When debugging issues or generally analyzing the runtime of a system it would be nice to be able to tell the different ZTHRs running by name rather than having to analyze their stack. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Matthew Ahrens <[email protected]> Co-authored-by: Ryan Moeller <[email protected]> Signed-off-by: Serapheim Dimitropoulos <[email protected]> Closes #10630
* Prefix zfs internal endian checks with _ZFSMatthew Macy2020-07-282-2/+2
| | | | | | | | | | | FreeBSD defines _BIG_ENDIAN BIG_ENDIAN _LITTLE_ENDIAN LITTLE_ENDIAN on every architecture. Trying to do cross builds whilst hiding this from ZFS has proven extremely cumbersome. Reviewed-by: Ryan Moeller <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10621
* Refactor ccompile.h to not include system headersMatthew Macy2020-07-252-0/+6
| | | | | | | | This is a step toward being able to vendor the OpenZFS code in FreeBSD. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Ryan Moeller <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10625
* Make use of ZFS_DEBUG consistent within kmod sourcesMatthew Macy2020-07-258-13/+13
| | | | | | Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Ryan Moeller <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10623
* FreeBSD: Fixes required to build ZFS on PowerPCMatthew Macy2020-07-251-1/+1
| | | | | Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Matt Macy <[email protected]> Closes #10622
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-12 19:45:10 +0000