]> www.infradead.org Git - users/hch/xfs.git/log
users/hch/xfs.git
8 months agoxfs: check for incompatible features xfs-zoned-2024-08-22
Christoph Hellwig [Sun, 28 Jul 2024 13:50:24 +0000 (06:50 -0700)]
xfs: check for incompatible features

We don't currenty support the rtsb or reflink features with zoned
devices.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: add data placement info to mount stats
Hans Holmberg [Fri, 7 Jun 2024 08:31:43 +0000 (08:31 +0000)]
xfs: add data placement info to mount stats

Add per-rtg active refs, life time hint and data separation score and
an aggregate data separation score as output to the mount stats
to aid debugging and analysis.

Signed-off-by: Hans Holmberg <hans.holmberg@wdc.com>
9 months agoxfs: first stab at write life time data placement
Hans Holmberg [Fri, 7 Jun 2024 08:31:43 +0000 (08:31 +0000)]
xfs: first stab at write life time data placement

Add a file write life time data placement allocation scheme that aims
minimize fragmentation and thereby to do two things:

a) Complete separate file data when possible into diffent zones when
   possible.
b) Colocate file data of similar life times when feasible.

To get best results, average file sizes should align with average
zone capacitity.

Benchmarked with RocksDB using leveled compaction, obeserving ~10%
throughput improvement for overwrite workloads at 80% file system
utilization.

Signed-off-by: Hans Holmberg <hans.holmberg@wdc.com>
9 months agoxfs: add plumbing and mount option for write life time hints
Hans Holmberg [Mon, 22 Jul 2024 13:34:12 +0000 (06:34 -0700)]
xfs: add plumbing and mount option for write life time hints

Add a mount option and some plumbing for enabling usage
of file write life time hints.

Based on a patch from Christoph Hellwig

Signed-off-by: Hans Holmberg <hans.holmberg@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: allow COW forks on zoned file systems in xchk_bmap
Christoph Hellwig [Fri, 10 May 2024 06:51:02 +0000 (08:51 +0200)]
xfs: allow COW forks on zoned file systems in xchk_bmap

zoned file systems can have COW forks even without reflinks.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: support xrep_require_rtext_inuse on zoned file systems
Christoph Hellwig [Mon, 22 Jul 2024 13:31:28 +0000 (06:31 -0700)]
xfs: support xrep_require_rtext_inuse on zoned file systems

Space usage is tracked by the rmap, which already is separately
cross-reference.  But on top of that we have the write pointer and can
do a basic sanity check here that the block is not beyond the write
pointer.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: support xchk_xref_is_used_rt_space on zoned file systems
Christoph Hellwig [Fri, 10 May 2024 06:43:29 +0000 (08:43 +0200)]
xfs: support xchk_xref_is_used_rt_space on zoned file systems

Space usage is tracked by the rmap, which already is separately
cross-reference.  But on top of that we have the write pointer and can
do a basic sanity check here that the block is not beyond the write
pointer.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: disable sb_frextents scrub/repair for zoned file systems
Christoph Hellwig [Sun, 12 May 2024 05:39:45 +0000 (07:39 +0200)]
xfs: disable sb_frextents scrub/repair for zoned file systems

Zoned file systems not only don't use the frextents counter, but the
in-memory percpu couner also includes reservations take before even
allocating delalloc extent records, so it will never match the per-zone
used information.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: ensure we have blocks available before taking the iolock
Christoph Hellwig [Wed, 24 Jul 2024 14:20:34 +0000 (07:20 -0700)]
xfs: ensure we have blocks available before taking the iolock

With the last patch we have the infrastructure in place to have space
reservations before taking the iolock and thus avoid the GC deadlock
in generic/269.  But right now it will happily take space that has
been freed in a used zoned that would still require GC.  Add a new
RTAVAILABLE counter of blocks that are actually directly available to
be written into in addition to the classic free counter.  Only allow
a write to go ahead if it has blocks available to write, and otherwise
wait for GC.  This also requires tweaking the need GC condition a
bit as we now always need to GC if someone is waiting for space.

Because GC always allocates from the reserved pool that gets replenished
first we can also do away with the ratio to favor it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: reserve blocks before taking the iolock
Christoph Hellwig [Mon, 22 Jul 2024 13:29:10 +0000 (06:29 -0700)]
xfs: reserve blocks before taking the iolock

For zoned file systems we reserve the free extents before taking the
ilock so that if we have to force garbage collection it happens before we
take the iolock.  This is done because GC has to take the iolock after it
moved data to a new place, and this could otherwise deadlock.

This unfortunately has to exclude block zeroing, as for truncate we are
called with the iolock (aka i_rwsem) already held.  As zeroing is always
only for a single block at a time, or up to two total for a syscall in
case for free_file_range we deal with that by just stealing the block,
but failing the allocation if we'd have to wait for GC (this will only
be implemented in the following patch).

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: support zoned RT devices
Christoph Hellwig [Wed, 24 Jul 2024 14:10:51 +0000 (07:10 -0700)]
xfs: support zoned RT devices

WARNING: this is early prototype code.

The zoned allocator works by handing out data blocks to the direct or
buffered write code at the place where XFS currently does block
allocations.  It does not actually insert them into the bmap extent tree
at this time, but only after I/O completion when we known the block number.

The zoned allocator works on any kind of device, including conventional
devices or conventional zones by having a crude write pointer emulation.
For zone devices active zone management is fully support, as is
zone capacity < zone size.

The two major limitations are:

 - there is no support for unwritten extents and thus persistent
   file preallocations from fallocate().  This is inherent to an
   always out of place write scheme as there is no way to persistently
   preallocate blocks for an indefinite number of overwrites
 - because the metadata blocks and data blocks are on different
   device you can run out of space for metadata while having plenty
   of space for data and vice versa.  This is inherent to a scheme
   where we use different devices or pools for each.

Thanks to Hans Holmberg <hans.holmberg@wdc.com> for lots of fixes
and improvements.

Co-developed-by: Hans Holmberg <hans.holmberg@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: don't do speculative preallocations for zoned RT allocations
Christoph Hellwig [Mon, 22 Jul 2024 13:24:04 +0000 (06:24 -0700)]
xfs: don't do speculative preallocations for zoned RT allocations

XFS does extensive speculative delalloc preallocation to ensure it can
allocate large contiguous regions on disk.  For the zone write path this
doesn't make sense as we can always just convert what is being written
anyway, and allocating more delalloc space also means that we take more
blocks out of the pool that we have reserved space for before.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: don't lock the RT inodes for zoned devices
Christoph Hellwig [Fri, 23 Feb 2024 06:31:06 +0000 (07:31 +0100)]
xfs: don't lock the RT inodes for zoned devices

Zoned RT devices don't use the bitmap/summary, so don't lock them.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: add a helper to check if an inode sits on a zoned device
Christoph Hellwig [Sun, 25 Feb 2024 05:57:49 +0000 (06:57 +0100)]
xfs: add a helper to check if an inode sits on a zoned device

Add a xfs_is_zoned_inode helper that returns true if an inode has the
RT flag set and the file system is zoned.  This will be used to key
off zoned allocator behavior.

Make xfs_is_always_cow_inode return true for zoned inodes as we always
need to write out of place on zoned devices.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: add an incompat feature bit for zoned RT devices
Christoph Hellwig [Wed, 24 Jul 2024 14:08:48 +0000 (07:08 -0700)]
xfs: add an incompat feature bit for zoned RT devices

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: refine the unaligned check for always COW inodes in xfs_file_dio_write
Christoph Hellwig [Fri, 27 Oct 2023 07:58:24 +0000 (09:58 +0200)]
xfs: refine the unaligned check for always COW inodes in xfs_file_dio_write

For always COW inodes we also must check the alignment of each individual
iovec segment, as they could end up with different I/Os due to the way
bio_iov_iter_get_pages works, and we'd then overwrite an already written
block.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: support XFS_BMAPI_REMAP in xfs_bmap_del_extent_delay
Christoph Hellwig [Fri, 31 May 2024 09:27:32 +0000 (11:27 +0200)]
xfs: support XFS_BMAPI_REMAP in xfs_bmap_del_extent_delay

The zone allocator wants to be able to remove a delalloc mapping in the
COW fork while keeping the block reservation.  To support that pass the
blags argument down to xfs_bmap_del_extent_delay and support the
XFS_BMAPI_REMAP flag to keep the reservation.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: pass an optional startblock argument to xfs_reflink_end_cow
Christoph Hellwig [Tue, 17 Oct 2023 07:15:13 +0000 (09:15 +0200)]
xfs: pass an optional startblock argument to xfs_reflink_end_cow

With the upcoming zoned allocator, extents won't be convered from
delalloc to real allocations before submitting the I/O, as we'll only
know the actual block number on I/O completion.  Add a new argument to
xfs_reflink_end_cow that can pass in the startblock we got from the
I/O completion handler, and convert the completed range from delalloc
to a written extent covering that block range.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: factor our a __xfs_reflink_end_cow_extent helper
Christoph Hellwig [Fri, 31 May 2024 09:25:00 +0000 (11:25 +0200)]
xfs: factor our a __xfs_reflink_end_cow_extent helper

This will allow reusing the code for zoned direct I/O completions, where
the out of place write allocation is passed as private data with the
I/O and not stored in the COW fork.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: skip always_cow inodes in xfs_reflink_trim_around_shared
Christoph Hellwig [Sat, 14 Oct 2023 05:50:35 +0000 (07:50 +0200)]
xfs: skip always_cow inodes in xfs_reflink_trim_around_shared

xfs_reflink_trim_around_shared tries to find shared blocks in the
refcount btree.  Always_cow inodes don't have that tree, so don't
bother.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: preserve RT reservations across remounts
Hans Holmberg [Tue, 9 Jan 2024 18:01:15 +0000 (19:01 +0100)]
xfs: preserve RT reservations across remounts

Introduce a reservation setting for rt devices so that zoned GC
reservations are preserved over remount ro/rw cycles.

Signed-off-by: Hans Holmberg <hans.holmberg@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: generalize the freespace and reserved blocks handling
Christoph Hellwig [Mon, 22 Jul 2024 13:20:19 +0000 (06:20 -0700)]
xfs: generalize the freespace and reserved blocks handling

The main handling of the incore per-cpu freespace counters is already
handled in xfs_mod_freecounter for both the block and RT extent cases,
but the actual counter is passed in an special cases.

Replace both the percpu counters and the resblks counters with arrays,
so that support reserved RT extents can be supported, which will be
needed for garbarge collection on zoned devices.

Use helpers to access the freespace counters everywhere intead of
poking through the abstraction by using the percpu_count helpers
directly.  This also switches the flooring of the frextents counter
to 0 in statfs for the rthinherit case to a manual min_t call to match
the handling of the fdblocks counter for normal file systems.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: convert rtgroup lookup to an xarray
Christoph Hellwig [Wed, 14 Feb 2024 14:17:15 +0000 (15:17 +0100)]
xfs: convert rtgroup lookup to an xarray

The xarray is the modern replacement for the radix-tree.  It is simpler
to use, especially when using marks to find specific entries, something
we will heavily use for the zone allocator implementation.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: pass private data to iomap_file_buffered_write_punch_delalloc
Christoph Hellwig [Wed, 24 Apr 2024 06:53:45 +0000 (08:53 +0200)]
iomap: pass private data to iomap_file_buffered_write_punch_delalloc

Allow the file system to pass private data which then gets passed on
to the punch callback.   Also move the iomap_punch_t typedef to
iomap.h so it doesn't have to be open coded in the declaration.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: pass private data to iomap_page_mkwrite
Christoph Hellwig [Mon, 22 Jul 2024 13:18:17 +0000 (06:18 -0700)]
iomap: pass private data to iomap_page_mkwrite

Allow the file system to pass private data which can be used by the
iomap_begin and iomap_end methods through the private pointer in the
iomap_iter structure.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: pass private data to iomap_file_buffered_write
Christoph Hellwig [Fri, 19 Apr 2024 06:41:28 +0000 (08:41 +0200)]
iomap: pass private data to iomap_file_buffered_write

Allow the file system to pass private data which can be used by the
iomap_begin and iomap_end methods through the private pointer in the
iomap_iter structure.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: optionally use ioends for direct I/O
Christoph Hellwig [Fri, 24 Nov 2023 09:45:35 +0000 (10:45 +0100)]
iomap: optionally use ioends for direct I/O

struct iomap_ioend currently tracks outstanding buffered writes and has
some really nice code in core iomap and XFS to merge contiguous I/Os
an defer them to userspace for completion in a very efficient way.

For zoned writes we'll also need a per-bio user context completion to
record the written blocks, and the infrastructure for that would look
basically like the ioend handling for buffered I/O.

So intead of reinventing the wheel, reuse the existing infrastructure.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: don't merge ioends with mismatching fs private flag
Christoph Hellwig [Fri, 24 Nov 2023 18:51:38 +0000 (19:51 +0100)]
iomap: don't merge ioends with mismatching fs private flag

If the file system set it's private flag on one ioend but not the other
we better don't merge the two.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: export iomap_submit_ioend
Christoph Hellwig [Wed, 14 Feb 2024 14:10:41 +0000 (15:10 +0100)]
iomap: export iomap_submit_ioend

This will let file systems submit the current ioend from ->map_blocks
to free resources for additional allocations.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: support IOMAP_F_ZONE_APPEND for buffered I/O
Christoph Hellwig [Wed, 14 Feb 2024 14:09:44 +0000 (15:09 +0100)]
iomap: support IOMAP_F_ZONE_APPEND for buffered I/O

Add support for Zone Append commands to the iomap buffer writeback code.
This involves selecting the right block layer operation and using the
right helper to add data to the bio, as well as not creating chained
bios inside the iomap for zone append as they could be written
non-contiguously and adjusting the sector based merge criteria.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: make iomap_sector Zone Append aware
Christoph Hellwig [Fri, 20 Oct 2023 13:36:03 +0000 (15:36 +0200)]
iomap: make iomap_sector Zone Append aware

Zone Append commands always point to the zone start sector.  Change
the iomap_sector() helper to not adjust the start block for the position
in the iomap range for Zone Append iomaps.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: reinstate IOMAP_F_ZONE_APPEND support
Christoph Hellwig [Sun, 5 Nov 2023 05:40:52 +0000 (06:40 +0100)]
iomap: reinstate IOMAP_F_ZONE_APPEND support

Add back the support for using Zone Append in the iomap direct I/O code
that was removed a while ago as we'll use it for the XFS zoned device
support.

This is essentially a revert of commit 8e81aa16a421 ("iomap: remove
IOMAP_F_ZONE_APPEND") with an additional comment describing the flag now
that all the other IOMAP_F_* flags have a nice description.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: wait for writeback before allocating new blocks
Christoph Hellwig [Fri, 13 Oct 2023 06:09:39 +0000 (08:09 +0200)]
iomap: wait for writeback before allocating new blocks

This means we are actually forced to allocate new delalloc space for the
new dirtier instead of reusing one that is currently being used for
writeback.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoiomap: factor out a iomap_last_written_block helper
Christoph Hellwig [Fri, 20 Oct 2023 10:57:00 +0000 (12:57 +0200)]
iomap: factor out a iomap_last_written_block helper

Split out a pice of logic from iomap_file_buffered_write_punch_delalloc
that is useful for all iomap_end implementations.

Signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoTEMP: nvme-pci: disable async probe
Christoph Hellwig [Sun, 15 Oct 2023 07:26:20 +0000 (09:26 +0200)]
TEMP: nvme-pci: disable async probe

This keeps getting my ZNS vs ZNS drivers reordered a bit and is annoying
for testing.

Not-really-signed-off-by: Christoph Hellwig <hch@lst.de>
9 months agoxfs: enable realtime reflink
Darrick J. Wong [Wed, 29 May 2024 04:13:30 +0000 (21:13 -0700)]
xfs: enable realtime reflink

Enable reflink for realtime devices, sort of.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: fix CoW forks for realtime files
Darrick J. Wong [Wed, 29 May 2024 04:13:29 +0000 (21:13 -0700)]
xfs: fix CoW forks for realtime files

Port the copy on write fork repair to realtime files.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: check for shared rt extents when rebuilding rt file's data fork
Darrick J. Wong [Wed, 29 May 2024 04:13:29 +0000 (21:13 -0700)]
xfs: check for shared rt extents when rebuilding rt file's data fork

When we're rebuilding the data fork of a realtime file, we need to
cross-reference each mapping with the rt refcount btree to ensure that
the reflink flag is set if there are any shared extents found.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: repair inodes that have a refcount btree in the data fork
Darrick J. Wong [Wed, 29 May 2024 04:13:28 +0000 (21:13 -0700)]
xfs: repair inodes that have a refcount btree in the data fork

Plumb knowledge of refcount btrees into the inode core repair code.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: online repair of the realtime refcount btree
Darrick J. Wong [Wed, 29 May 2024 04:13:28 +0000 (21:13 -0700)]
xfs: online repair of the realtime refcount btree

Port the data device's refcount btree repair code to the realtime
refcount btree.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: capture realtime CoW staging extents when rebuilding rt rmapbt
Darrick J. Wong [Wed, 29 May 2024 04:13:27 +0000 (21:13 -0700)]
xfs: capture realtime CoW staging extents when rebuilding rt rmapbt

Walk the realtime refcount btree to find the CoW staging extents when
we're rebuilding the realtime rmap btree.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: walk the rt reference count tree when rebuilding rmap
Darrick J. Wong [Wed, 29 May 2024 04:13:27 +0000 (21:13 -0700)]
xfs: walk the rt reference count tree when rebuilding rmap

When we're rebuilding the data device rmap, if we encounter a "refcount"
format fork, we have to walk the (realtime) refcount btree inode to
build the appropriate mappings.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: check new rtbitmap records against rt refcount btree
Darrick J. Wong [Wed, 29 May 2024 04:13:26 +0000 (21:13 -0700)]
xfs: check new rtbitmap records against rt refcount btree

When we're rebuilding the realtime bitmap, check the proposed free
extents against the rt refcount btree to make sure we don't commit any
grievous errors.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: don't flag quota rt block usage on rtreflink filesystems
Darrick J. Wong [Wed, 29 May 2024 04:13:25 +0000 (21:13 -0700)]
xfs: don't flag quota rt block usage on rtreflink filesystems

Quota space usage is allowed to exceed the size of the physical storage
when reflink is enabled.  Now that we have reflink for the realtime
volume, apply this same logic to the rtb repair logic.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: scrub the metadir path of rt refcount btree files
Darrick J. Wong [Wed, 29 May 2024 04:13:25 +0000 (21:13 -0700)]
xfs: scrub the metadir path of rt refcount btree files

Add a new XFS_SCRUB_METAPATH subtype so that we can scrub the metadata
directory tree path to the refcount btree file for each rt group.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: detect and repair misaligned rtinherit directory cowextsize hints
Darrick J. Wong [Wed, 29 May 2024 04:13:24 +0000 (21:13 -0700)]
xfs: detect and repair misaligned rtinherit directory cowextsize hints

If we encounter a directory that has been configured to pass on a CoW
extent size hint to a new realtime file and the hint isn't an integer
multiple of the rt extent size, we should flag the hint for
administrative review and/or turn it off because that is a
misconfiguration.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: allow dquot rt block count to exceed rt blocks on reflink fs
Darrick J. Wong [Wed, 29 May 2024 04:13:24 +0000 (21:13 -0700)]
xfs: allow dquot rt block count to exceed rt blocks on reflink fs

Update the quota scrubber to allow dquots where the realtime block count
exceeds the block count of the rt volume if reflink is enabled.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: check reference counts of gaps between rt refcount records
Darrick J. Wong [Wed, 29 May 2024 04:13:23 +0000 (21:13 -0700)]
xfs: check reference counts of gaps between rt refcount records

If there's a gap between records in the rt refcount btree, we ought to
cross-reference the gap with the rtrmap records to make sure that there
aren't any overlapping records for a region that doesn't have any shared
ownership.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: allow overlapping rtrmapbt records for shared data extents
Darrick J. Wong [Wed, 29 May 2024 04:13:23 +0000 (21:13 -0700)]
xfs: allow overlapping rtrmapbt records for shared data extents

Allow overlapping realtime reverse mapping records if they both describe
shared data extents and the fs supports reflink on the realtime volume.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: cross-reference checks with the rt refcount btree
Darrick J. Wong [Wed, 29 May 2024 04:13:22 +0000 (21:13 -0700)]
xfs: cross-reference checks with the rt refcount btree

Use the realtime refcount btree to implement cross-reference checks in
other data structures.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: scrub the realtime refcount btree
Darrick J. Wong [Wed, 29 May 2024 04:13:21 +0000 (21:13 -0700)]
xfs: scrub the realtime refcount btree

Add code to scrub realtime refcount btrees.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: report realtime refcount btree corruption errors to the health system
Darrick J. Wong [Wed, 29 May 2024 04:13:21 +0000 (21:13 -0700)]
xfs: report realtime refcount btree corruption errors to the health system

Whenever we encounter corrupt realtime refcount btree blocks, we should
report that to the health monitoring system for later reporting.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: add realtime refcount btree when adding rt volume
Darrick J. Wong [Wed, 29 May 2024 04:13:20 +0000 (21:13 -0700)]
xfs: add realtime refcount btree when adding rt volume

If we're adding enough space to the realtime section to require the
creation of new realtime groups, create the rt refcount btree inode
before we start adding the space.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: check that the rtrefcount maxlevels doesn't increase when growing fs
Darrick J. Wong [Wed, 29 May 2024 04:13:20 +0000 (21:13 -0700)]
xfs: check that the rtrefcount maxlevels doesn't increase when growing fs

The size of filesystem transaction reservations depends on the maximum
height (maxlevels) of the realtime btrees.  Since we don't want a grow
operation to increase the reservation size enough that we'll fail the
minimum log size checks on the next mount, constrain growfs operations
if they would cause an increase in the rt refcount btree maxlevels.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: enable extent size hints for CoW operations
Darrick J. Wong [Wed, 29 May 2024 04:13:19 +0000 (21:13 -0700)]
xfs: enable extent size hints for CoW operations

Wire up the copy-on-write extent size hint for realtime files, and
connect it to the rt allocator so that we avoid fragmentation on rt
filesystems.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: apply rt extent alignment constraints to CoW extsize hint
Darrick J. Wong [Wed, 29 May 2024 04:13:19 +0000 (21:13 -0700)]
xfs: apply rt extent alignment constraints to CoW extsize hint

The copy-on-write extent size hint is subject to the same alignment
constraints as the regular extent size hint.  Since we're in the process
of adding reflink (and therefore CoW) to the realtime device, we must
apply the same scattered rextsize alignment validation strategies to
both hints to deal with the possibility of rextsize changing.

Therefore, fix the inode validator to perform rextsize alignment checks
on regular realtime files, and to remove misaligned directory hints.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: fix xfs_get_extsz_hint behavior with realtime alwayscow files
Darrick J. Wong [Wed, 29 May 2024 04:13:18 +0000 (21:13 -0700)]
xfs: fix xfs_get_extsz_hint behavior with realtime alwayscow files

Currently, we (ab)use xfs_get_extsz_hint so that it always returns a
nonzero value for realtime files.  This apparently was done to disable
delayed allocation for realtime files.

However, once we enable realtime reflink, we can also turn on the
alwayscow flag to force CoW writes to realtime files.  In this case, the
logic will incorrectly send the write through the delalloc write path.

Fix this by adjusting the logic slightly.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: refcover CoW leftovers in the realtime volume
Darrick J. Wong [Wed, 29 May 2024 04:13:18 +0000 (21:13 -0700)]
xfs: refcover CoW leftovers in the realtime volume

Scan the realtime refcount tree at mount time to get rid of leftover
CoW staging extents.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: allow inodes to have the realtime and reflink flags
Darrick J. Wong [Wed, 29 May 2024 04:13:17 +0000 (21:13 -0700)]
xfs: allow inodes to have the realtime and reflink flags

Now that we can share blocks between realtime files, allow this
combination.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: enable sharing of realtime file blocks
Darrick J. Wong [Wed, 29 May 2024 04:13:16 +0000 (21:13 -0700)]
xfs: enable sharing of realtime file blocks

Update the remapping routines to be able to handle realtime files.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: enable CoW for realtime data
Darrick J. Wong [Wed, 29 May 2024 04:13:16 +0000 (21:13 -0700)]
xfs: enable CoW for realtime data

Update our write paths to support copy on write on the rt volume.  This
works in more or less the same way as it does on the data device, with
the major exception that we never do delalloc on the rt volume.

Because we consider unwritten CoW fork staging extents to be incore
quota reservation, we update xfs_quota_reserve_blkres to support this
case.  Though xfs doesn't allow rt and quota together, the change is
trivial and we shouldn't leave a logic bomb here.

While we're at it, add a missing xfs_mod_delalloc call when we remove
delalloc block reservation from the inode.  This is largely irrelvant
since realtime files do not use delalloc, but we want to avoid leaving
logic bombs.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: refactor reflink quota updates
Darrick J. Wong [Wed, 29 May 2024 04:13:15 +0000 (21:13 -0700)]
xfs: refactor reflink quota updates

Hoist all quota updates for reflink into a helper function, since things
are about to become more complicated.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: compute rtrmap btree max levels when reflink enabled
Darrick J. Wong [Wed, 29 May 2024 04:13:15 +0000 (21:13 -0700)]
xfs: compute rtrmap btree max levels when reflink enabled

Compute the maximum possible height of the realtime rmap btree when
reflink is enabled.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: update rmap to allow cow staging extents in the rt rmap
Darrick J. Wong [Wed, 29 May 2024 04:13:14 +0000 (21:13 -0700)]
xfs: update rmap to allow cow staging extents in the rt rmap

Don't error out on CoW staging extent records when realtime reflink is
enabled.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: create routine to allocate and initialize a realtime refcount btree inode
Darrick J. Wong [Wed, 29 May 2024 04:13:14 +0000 (21:13 -0700)]
xfs: create routine to allocate and initialize a realtime refcount btree inode

Create a library routine to allocate and initialize an empty realtime
refcountbt inode.  We'll use this for growfs, mkfs, and repair.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: wire up realtime refcount btree cursors
Darrick J. Wong [Wed, 29 May 2024 04:13:13 +0000 (21:13 -0700)]
xfs: wire up realtime refcount btree cursors

Wire up realtime refcount btree cursors wherever they're needed
throughout the code base.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoFrom: Christoph Hellwig <hch@lst.de>
Darrick J. Wong [Mon, 1 Jul 2024 21:26:06 +0000 (14:26 -0700)]
From: Christoph Hellwig <hch@lst.de>

xfs: refactor xfs_reflink_find_shared

Move lookup of the perag structure from the callers into the helpers,
and return the offset into the extent of the shared region instead of
the block number that needs post-processing.  This prepares the
callsites for the creation of an rt-specific variant in the next patch.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
[djwong: port to the middle of the rtreflink series for cleanliness]
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: wire up a new inode fork type for the realtime refcount
Darrick J. Wong [Wed, 29 May 2024 04:13:12 +0000 (21:13 -0700)]
xfs: wire up a new inode fork type for the realtime refcount

Plumb in the pieces we need to embed the root of the realtime refcount
btree in an inode's data fork, complete with new fork type and
on-disk interpretation functions.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: add metadata reservations for realtime refcount btree
Darrick J. Wong [Wed, 29 May 2024 04:13:12 +0000 (21:13 -0700)]
xfs: add metadata reservations for realtime refcount btree

Reserve some free blocks so that we will always have enough free blocks
in the data volume to handle expansion of the realtime refcount btree.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: add realtime refcount btree inode to metadata directory
Darrick J. Wong [Wed, 29 May 2024 04:13:11 +0000 (21:13 -0700)]
xfs: add realtime refcount btree inode to metadata directory

Add a metadir path to select the realtime refcount btree inode and load
it at mount time.  The rtrefcountbt inode will have a unique extent format
code, which means that we also have to update the inode validation and
flush routines to look for it.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: add realtime refcount btree block detection to log recovery
Darrick J. Wong [Wed, 29 May 2024 04:13:11 +0000 (21:13 -0700)]
xfs: add realtime refcount btree block detection to log recovery

Identify rt refcount btree blocks in the log correctly so that we can
validate them during log recovery.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: support recovering refcount intent items targetting realtime extents
Darrick J. Wong [Wed, 29 May 2024 04:13:10 +0000 (21:13 -0700)]
xfs: support recovering refcount intent items targetting realtime extents

Now that we have reflink on the realtime device, refcount intent items
have to support remapping extents on the realtime volume.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: add a realtime flag to the refcount update log redo items
Darrick J. Wong [Wed, 29 May 2024 04:13:10 +0000 (21:13 -0700)]
xfs: add a realtime flag to the refcount update log redo items

Extend the refcount update (CUI) log items with a new realtime flag that
indicates that the updates apply against the realtime refcountbt.  We'll
wire up the actual refcount code later.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: prepare refcount functions to deal with rtrefcountbt
Darrick J. Wong [Wed, 29 May 2024 04:13:09 +0000 (21:13 -0700)]
xfs: prepare refcount functions to deal with rtrefcountbt

Prepare the high-level refcount functions to deal with the new realtime
refcountbt and its slightly different conventions.  Provide the ability
to talk to either refcountbt or rtrefcountbt formats from the same high
level code.

Note that we leave the _recover_cow_leftovers functions for a separate
patch so that we can convert it all at once.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: add realtime refcount btree operations
Darrick J. Wong [Wed, 29 May 2024 04:13:09 +0000 (21:13 -0700)]
xfs: add realtime refcount btree operations

Implement the generic btree operations needed to manipulate rtrefcount
btree blocks. This is different from the regular refcountbt in that we
allocate space from the filesystem at large, and are neither constrained
to the free space nor any particular AG.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: realtime refcount btree transaction reservations
Darrick J. Wong [Wed, 29 May 2024 04:13:08 +0000 (21:13 -0700)]
xfs: realtime refcount btree transaction reservations

Make sure that there's enough log reservation to handle mapping
and unmapping realtime extents.  We have to reserve enough space
to handle a split in the rtrefcountbt to add the record and a second
split in the regular refcountbt to record the rtrefcountbt split.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: define the on-disk realtime refcount btree format
Darrick J. Wong [Wed, 29 May 2024 04:13:07 +0000 (21:13 -0700)]
xfs: define the on-disk realtime refcount btree format

Start filling out the rtrefcount btree implementation. Start with the
on-disk btree format; add everything needed to read, write and
manipulate refcount btree blocks. This prepares the way for connecting
the btree operations implementation.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: namespace the maximum length/refcount symbols
Darrick J. Wong [Wed, 29 May 2024 04:13:07 +0000 (21:13 -0700)]
xfs: namespace the maximum length/refcount symbols

Actually namespace these variables properly, so that readers can tell
that this is an XFS symbol, and that it's for the refcount
functionality.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: introduce realtime refcount btree definitions
Darrick J. Wong [Wed, 29 May 2024 04:13:06 +0000 (21:13 -0700)]
xfs: introduce realtime refcount btree definitions

Add new realtime refcount btree definitions. The realtime refcount btree
will be rooted from a hidden inode, but has its own shape and therefore
needs to have most of its own separate types.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: prepare refcount btree cursor tracepoints for realtime
Darrick J. Wong [Wed, 29 May 2024 04:13:06 +0000 (21:13 -0700)]
xfs: prepare refcount btree cursor tracepoints for realtime

Rework the refcount btree cursor tracepoints in preparation to handle the
realtime refcount btree cursor.  Mostly this involves renaming the field to
"refcbno" and extracting the group number from the cursor when possible.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: enable realtime rmap btree
Darrick J. Wong [Wed, 29 May 2024 04:13:00 +0000 (21:13 -0700)]
xfs: enable realtime rmap btree

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: hook live realtime rmap operations during a repair operation
Darrick J. Wong [Wed, 29 May 2024 04:12:59 +0000 (21:12 -0700)]
xfs: hook live realtime rmap operations during a repair operation

Hook the regular realtime rmap code when an rtrmapbt repair operation is
running so that we can unlock the AGF buffer to scan the filesystem and
keep the in-memory btree up to date during the scan.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: create a shadow rmap btree during realtime rmap repair
Darrick J. Wong [Wed, 29 May 2024 04:12:58 +0000 (21:12 -0700)]
xfs: create a shadow rmap btree during realtime rmap repair

Create an in-memory btree of rmap records instead of an array.  This
enables us to do live record collection instead of freezing the fs.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: online repair of the realtime rmap btree
Darrick J. Wong [Wed, 29 May 2024 04:12:58 +0000 (21:12 -0700)]
xfs: online repair of the realtime rmap btree

Repair the realtime rmap btree while mounted.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: support repairing metadata btrees rooted in metadir inodes
Darrick J. Wong [Wed, 29 May 2024 04:12:57 +0000 (21:12 -0700)]
xfs: support repairing metadata btrees rooted in metadir inodes

Adapt the repair code so that we can stage a new btree in the data fork
area of a metadir inode and reap the old blocks.  We already have nearly
all of the infrastructure; the only parts that were missing were the
metadata inode reservation handling.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: online repair of realtime bitmaps for a realtime group
Darrick J. Wong [Wed, 29 May 2024 04:12:57 +0000 (21:12 -0700)]
xfs: online repair of realtime bitmaps for a realtime group

For a given rt group, regenerate the bitmap contents from the group's
realtime rmap btree.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: repair rmap btree inodes
Darrick J. Wong [Wed, 29 May 2024 04:12:56 +0000 (21:12 -0700)]
xfs: repair rmap btree inodes

Teach the inode repair code how to deal with realtime rmap btree inodes
that won't load properly.  This is most likely moot since the filesystem
generally won't mount without the rtrmapbt inodes being usable, but
we'll add this for completeness.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: repair inodes that have realtime extents
Darrick J. Wong [Wed, 29 May 2024 04:12:56 +0000 (21:12 -0700)]
xfs: repair inodes that have realtime extents

Plumb into the inode core repair code the ability to search for extents
on realtime devices.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: online repair of realtime file bmaps
Darrick J. Wong [Wed, 29 May 2024 04:12:55 +0000 (21:12 -0700)]
xfs: online repair of realtime file bmaps

Repair the block mappings of realtime files.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: walk the rt reverse mapping tree when rebuilding rmap
Darrick J. Wong [Wed, 29 May 2024 04:12:55 +0000 (21:12 -0700)]
xfs: walk the rt reverse mapping tree when rebuilding rmap

When we're rebuilding the data device rmap, if we encounter an "rmap"
format fork, we have to walk the (realtime) rmap btree inode to build
the appropriate mappings.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: scrub the metadir path of rt rmap btree files
Darrick J. Wong [Wed, 29 May 2024 04:12:54 +0000 (21:12 -0700)]
xfs: scrub the metadir path of rt rmap btree files

Add a new XFS_SCRUB_METAPATH subtype so that we can scrub the metadata
directory tree path to the rmap btree file for each rt group.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: scan rt rmap when we're doing an intense rmap check of bmbt mappings
Darrick J. Wong [Wed, 29 May 2024 04:12:53 +0000 (21:12 -0700)]
xfs: scan rt rmap when we're doing an intense rmap check of bmbt mappings

Teach the bmbt scrubber how to perform a comprehensive check that the
rmapbt does not contain /any/ mappings that are not described by bmbt
records when it's dealing with a realtime file.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: cross-reference the realtime rmapbt
Darrick J. Wong [Wed, 29 May 2024 04:12:53 +0000 (21:12 -0700)]
xfs: cross-reference the realtime rmapbt

Teach the data fork and realtime bitmap scrubbers to cross-reference
information with the realtime rmap btree.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: cross-reference realtime bitmap to realtime rmapbt scrubber
Darrick J. Wong [Wed, 29 May 2024 04:12:52 +0000 (21:12 -0700)]
xfs: cross-reference realtime bitmap to realtime rmapbt scrubber

When we're checking the realtime rmap btree entries, cross-reference
those entries with the realtime bitmap too.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: scrub the realtime rmapbt
Darrick J. Wong [Wed, 29 May 2024 04:12:52 +0000 (21:12 -0700)]
xfs: scrub the realtime rmapbt

Check the realtime reverse mapping btree against the rtbitmap, and
modify the rtbitmap scrub to check against the rtrmapbt.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: allow queued realtime intents to drain before scrubbing
Darrick J. Wong [Wed, 29 May 2024 04:12:51 +0000 (21:12 -0700)]
xfs: allow queued realtime intents to drain before scrubbing

When a writer thread executes a chain of log intent items for the
realtime volume, the ILOCKs taken during each step are for each rt
metadata file, not the entire rt volume itself.  Although scrub takes
all rt metadata ILOCKs, this isn't sufficient to guard against scrub
checking the rt volume while that writer thread is in the middle of
finishing a chain because there's no higher level locking primitive
guarding the realtime volume.

When there's a collision, cross-referencing between data structures
(e.g. rtrmapbt and rtrefcountbt) yields false corruption events; if
repair is running, this results in incorrect repairs, which is
catastrophic.

Fix this by adding to the mount structure the same drain that we use to
protect scrub against concurrent AG updates, but this time for the
realtime volume.

[Contains a few cleanups from hch]

Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: fix scrub tracepoints when inode-rooted btrees are involved
Darrick J. Wong [Wed, 29 May 2024 04:12:40 +0000 (21:12 -0700)]
xfs: fix scrub tracepoints when inode-rooted btrees are involved

Fix a minor mistakes in the scrub tracepoints that can manifest when
inode-rooted btrees are enabled.  The existing code worked fine for bmap
btrees, but we should tighten the code up to be less sloppy.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: report realtime rmap btree corruption errors to the health system
Darrick J. Wong [Wed, 29 May 2024 04:11:58 +0000 (21:11 -0700)]
xfs: report realtime rmap btree corruption errors to the health system

Whenever we encounter corrupt realtime rmap btree blocks, we should
report that to the health monitoring system for later reporting.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: add realtime rmap btree when adding rt volume
Darrick J. Wong [Wed, 29 May 2024 04:11:57 +0000 (21:11 -0700)]
xfs: add realtime rmap btree when adding rt volume

If we're adding enough space to the realtime section to require the
creation of new realtime groups, create the rt rmap btree inode before
we start adding the space.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
9 months agoxfs: check that the rtrmapbt maxlevels doesn't increase when growing fs
Darrick J. Wong [Wed, 29 May 2024 04:11:57 +0000 (21:11 -0700)]
xfs: check that the rtrmapbt maxlevels doesn't increase when growing fs

The size of filesystem transaction reservations depends on the maximum
height (maxlevels) of the realtime btrees.  Since we don't want a grow
operation to increase the reservation size enough that we'll fail the
minimum log size checks on the next mount, constrain growfs operations
if they would cause an increase in those maxlevels.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>