]> www.infradead.org Git - users/hch/xfs.git/log
users/hch/xfs.git
5 months agoxfs: enable realtime quota again
Darrick J. Wong [Tue, 15 Oct 2024 19:39:35 +0000 (12:39 -0700)]
xfs: enable realtime quota again

Enable quotas for the realtime device.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: reserve quota for realtime files correctly
Darrick J. Wong [Tue, 15 Oct 2024 19:39:34 +0000 (12:39 -0700)]
xfs: reserve quota for realtime files correctly

Fix xfs_quota_reserve_blkres to reserve rt block quota whenever we're
dealing with a realtime file.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: create quota preallocation watermarks for realtime quota
Darrick J. Wong [Tue, 15 Oct 2024 19:39:33 +0000 (12:39 -0700)]
xfs: create quota preallocation watermarks for realtime quota

Refactor the quota preallocation watermarking code so that it'll work
for realtime quota too.  Convert the do_div calls into div_u64 for
compactness.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: report realtime block quota limits on realtime directories
Darrick J. Wong [Tue, 15 Oct 2024 19:39:32 +0000 (12:39 -0700)]
xfs: report realtime block quota limits on realtime directories

On the data device, calling statvfs on a projinherit directory results
in the block and avail counts being curtailed to the project quota block
limits, if any are set.  Do the same for realtime files or directories,
only use the project quota rt block limits.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: advertise realtime quota support in the xqm stat files
Darrick J. Wong [Tue, 15 Oct 2024 19:39:32 +0000 (12:39 -0700)]
xfs: advertise realtime quota support in the xqm stat files

Add a fifth column to this (really old) stat file to advertise that the
kernel supports quota for realtime volumes.  This will be used by
fstests to detect kernel support.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: fix chown with rt quota
Darrick J. Wong [Tue, 15 Oct 2024 19:39:31 +0000 (12:39 -0700)]
xfs: fix chown with rt quota

Make chown's quota adjustments work with realtime files.  This is mostly
a matter of calling xfs_inode_count_blocks on a given file to figure out
the number of blocks allocated to the data device and to the realtime
device, and using those quantities to update the quota accounting when
the id changes.  Delayed allocation reservations are moved from the old
dquot's incore reservation to the new dquot's incore reservation.

Note that there was a missing ILOCK bug in xfs_qm_dqusage_adjust that we
must fix before calling xfs_iread_extents.  Prior to 2.6.37 the locking
was correct, but then someone removed the ILOCK as part of a cleanup.
Nobody noticed because nowhere in the git history have we ever supported
rt+quota so nobody can use this.

I'm leaving git breadcrumbs in case anyone is desperate enough to try to
backport the rtquota code to old kernels.

Not-Cc: <stable@vger.kernel.org> # v2.6.37
Fixes: 52fda114249578 ("xfs: simplify xfs_qm_dqusage_adjust")
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: persist quota flags with metadir
Darrick J. Wong [Tue, 15 Oct 2024 19:39:30 +0000 (12:39 -0700)]
xfs: persist quota flags with metadir

It's annoying that one has to keep reminding XFS about what quota
options it should mount with, since the quota flags recording the
previous state are sitting right there in the primary superblock.  Even
more strangely, there exists a noquota option to disable quotas
completely, so it's odder still that providing no options is the same as
noquota.

Starting with metadir, let's change the behavior so that if the user
does not specify any quota-related mount options at all, the ondisk
quota flags will be used to bring up quota.  In other words, the
filesystem will mount in the same state and with the same functionality
as it had during the last mount.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: scrub quota file metapaths
Darrick J. Wong [Tue, 15 Oct 2024 19:39:29 +0000 (12:39 -0700)]
xfs: scrub quota file metapaths

Enable online fsck for quota file metadata directory paths.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: use metadir for quota inodes
Darrick J. Wong [Tue, 15 Oct 2024 19:39:28 +0000 (12:39 -0700)]
xfs: use metadir for quota inodes

Store the quota inodes in the /quota metadata directory if metadir is
enabled.  This enables us to stop using the sb_[ugp]uotino fields in the
superblock.  From this point on, all metadata files will be children of
the metadata directory tree root.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: refactor xfs_qm_destroy_quotainos
Darrick J. Wong [Tue, 15 Oct 2024 19:39:28 +0000 (12:39 -0700)]
xfs: refactor xfs_qm_destroy_quotainos

Reuse this function instead of open-coding the logic.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: use rtgroup busy extent list for FITRIM
Darrick J. Wong [Tue, 15 Oct 2024 19:39:27 +0000 (12:39 -0700)]
xfs: use rtgroup busy extent list for FITRIM

For filesystems that have rtgroups and hence use the busy extent list
for freed rt space, use that busy extent list so that FITRIM can issue
discard commands asynchronously without worrying about other callers
accidentally allocating and using space that is being discarded.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: implement busy extent tracking for rtgroups
Darrick J. Wong [Tue, 15 Oct 2024 19:39:26 +0000 (12:39 -0700)]
xfs: implement busy extent tracking for rtgroups

For rtgroups filesystems, track newly freed (rt) space through the log
until the rt EFIs have been committed to disk.  This way we ensure that
space cannot be reused until all traces of the old owner are gone.

As a fringe benefit, we now support -o discard on the realtime device.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: port the perag discard code to handle generic groups
Darrick J. Wong [Tue, 15 Oct 2024 19:39:25 +0000 (12:39 -0700)]
xfs: port the perag discard code to handle generic groups

Port xfs_discard_extents and its tracepoints to handle generic groups
instead of just perags.  This is needed to enable busy extent tracking
for rtgroups.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: move the min and max group block numbers to xfs_group
Darrick J. Wong [Tue, 15 Oct 2024 19:39:24 +0000 (12:39 -0700)]
xfs: move the min and max group block numbers to xfs_group

Move the min and max agblock numbers to the generic xfs_group structure
so that we can start building validators for extents within an rtgroup.
While we're at it, use check_add_overflow for the extent length
computation because that has much better overflow checking.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: adjust min_block usage in xfs_verify_agbno
Darrick J. Wong [Tue, 15 Oct 2024 19:39:24 +0000 (12:39 -0700)]
xfs: adjust min_block usage in xfs_verify_agbno

There's some weird logic in xfs_verify_agbno -- min_block ought to be
the first agblock number in the AG that can be used by non-static
metadata.  However, we initialize it to the last agblock of the static
metadata, which works due to the <= check, even though this isn't
technically correct.

Change the check to < and set min_block to the next agblock past the
static metadata.  This hasn't been an issue up to now, but we're going
to move these things into the generic group struct, and this will cause
problems with rtgroups, where min_block can be zero for an rtgroup that
doesn't have a rt superblock.

Note that there's no user-visible impact with the old logic, so this
isn't a bug fix.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: make xfs_rtblock_t a segmented address like xfs_fsblock_t
Darrick J. Wong [Tue, 15 Oct 2024 23:44:17 +0000 (16:44 -0700)]
xfs: make xfs_rtblock_t a segmented address like xfs_fsblock_t

Now that we've finished adding allocation groups to the realtime volume,
let's make the file block mapping address (xfs_rtblock_t) a segmented
value just like we do on the data device.  This means that group number
and block number conversions can be done with shifting and masking
instead of integer division.

While in theory we could continue caching the rgno shift value in
m_rgblklog, the fact that we now always use the shift value means that
we have an opportunity to increase the redundancy of the rt geometry by
storing it in the ondisk superblock and adding more sb verifier code.
Extend the sueprblock to store the rgblklog value.

Now that we have segmented addresses, set the correct values in
m_groups[XG_TYPE_RTG] so that the xfs_group helpers work correctly.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: create helpers to deal with rounding xfs_filblks_t to rtx boundaries
Darrick J. Wong [Tue, 15 Oct 2024 19:39:20 +0000 (12:39 -0700)]
xfs: create helpers to deal with rounding xfs_filblks_t to rtx boundaries

We're about to segment xfs_rtblock_t addresses, so we must create
type-specific helpers to do rt extent rounding of file mapping block
lengths because the rtb helpers soon will not do the right thing there.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: create helpers to deal with rounding xfs_fileoff_t to rtx boundaries
Darrick J. Wong [Tue, 15 Oct 2024 19:39:20 +0000 (12:39 -0700)]
xfs: create helpers to deal with rounding xfs_fileoff_t to rtx boundaries

We're about to segment xfs_rtblock_t addresses, so we must create
type-specific helpers to do rt extent rounding of file block offsets
because the rtb helpers soon will not do the right thing there.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: mask off the rtbitmap and summary inodes when metadir in use
Darrick J. Wong [Tue, 15 Oct 2024 19:39:19 +0000 (12:39 -0700)]
xfs: mask off the rtbitmap and summary inodes when metadir in use

Set the rtbitmap and summary file inumbers to NULLFSINO in the
superblock and make sure they're zeroed whenever we write the superblock
to disk, to mimic mkfs behavior.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: scrub metadir paths for rtgroup metadata
Darrick J. Wong [Tue, 15 Oct 2024 19:39:18 +0000 (12:39 -0700)]
xfs: scrub metadir paths for rtgroup metadata

Add the code we need to scan the metadata directory paths of rt group
metadata files.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: repair realtime group superblock
Darrick J. Wong [Tue, 15 Oct 2024 19:39:17 +0000 (12:39 -0700)]
xfs: repair realtime group superblock

Repair the realtime superblock if it has become out of date with the
primary superblock.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: scrub the realtime group superblock
Darrick J. Wong [Tue, 15 Oct 2024 19:39:17 +0000 (12:39 -0700)]
xfs: scrub the realtime group superblock

Enable scrubbing of realtime group superblocks.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: don't coalesce file mappings that cross rtgroup boundaries in scrub
Darrick J. Wong [Tue, 15 Oct 2024 19:39:16 +0000 (12:39 -0700)]
xfs: don't coalesce file mappings that cross rtgroup boundaries in scrub

The bmbt scrubber will combine file mappings if they are mergeable to
reduce the number of cross-referencing checks.  However, we shouldn't
combine mappings that cross rt group boundaries because that will cause
verifiers to trip incorrectly.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: make the RT allocator rtgroup aware
Christoph Hellwig [Tue, 15 Oct 2024 19:39:15 +0000 (12:39 -0700)]
xfs: make the RT allocator rtgroup aware

Make the allocator rtgroup aware by either picking a specific group if
there is a hint, or loop over all groups otherwise.  A simple rotor is
provided to pick the placement for initial allocations.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: don't merge ioends across RTGs
Darrick J. Wong [Tue, 15 Oct 2024 21:05:26 +0000 (14:05 -0700)]
xfs: don't merge ioends across RTGs

Unlike AGs, RTGs don't always have metadata in their first blocks, and
thus we don't get automatic protection from merging I/O completions
across RTG boundaries.  Add code to set the IOMAP_F_BOUNDARY flag for
ioends that start at the first block of a RTG so that they never get
merged into the previous ioend.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: use realtime EFI to free extents when rtgroups are enabled
Darrick J. Wong [Tue, 15 Oct 2024 19:39:14 +0000 (12:39 -0700)]
xfs: use realtime EFI to free extents when rtgroups are enabled

When rmap is enabled, XFS expects a certain order of operations, which
is: 1) remove the file mapping, 2) remove the reverse mapping, and then
3) free the blocks.  When reflink is enabled, XFS replaces (3) with a
deferred refcount decrement operation that can schedule freeing the
blocks if that was the last refcount.

For realtime files, xfs_bmap_del_extent_real tries to do 1 and 3 in the
same transaction, which will break both rmap and reflink unless we
switch it to use realtime EFIs.  Both rmap and reflink depend on the
rtgroups feature, so let's turn on EFIs for all rtgroups filesystems.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: support error injection when freeing rt extents
Darrick J. Wong [Tue, 15 Oct 2024 19:39:13 +0000 (12:39 -0700)]
xfs: support error injection when freeing rt extents

A handful of fstests expect to be able to test what happens when extent
free intents fail to actually free the extent.  Now that we're
supporting EFIs for realtime extents, add to xfs_rtfree_extent the same
injection point that exists in the regular extent freeing code.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: support logging EFIs for realtime extents
Darrick J. Wong [Tue, 15 Oct 2024 19:39:12 +0000 (12:39 -0700)]
xfs: support logging EFIs for realtime extents

Teach the EFI mechanism how to free realtime extents.  We're going to
need this to enforce proper ordering of operations when we enable
realtime rmap.

Declare a new log intent item type (XFS_LI_EFI_RT) and a separate defer
ops for rt extents.  This keeps the ondisk artifacts and processing code
completely separate between the rt and non-rt cases.  Hopefully this
will make it easier to debug filesystem problems.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: force swapext to a realtime file to use the file content exchange ioctl
Darrick J. Wong [Tue, 15 Oct 2024 19:39:11 +0000 (12:39 -0700)]
xfs: force swapext to a realtime file to use the file content exchange ioctl

xfs_swap_extent_rmap does not use log items to track the overall
progress of an attempt to swap the extent mappings between two files.
If the system crashes in the middle of swapping a partially written
realtime extent, the mapping will be left in an inconsistent state
wherein a file can point to multiple extents on the rt volume.

The new file range exchange functionality handles this correctly, so all
callers must upgrade to that.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: store rtgroup information with a bmap intent
Darrick J. Wong [Tue, 15 Oct 2024 19:39:11 +0000 (12:39 -0700)]
xfs: store rtgroup information with a bmap intent

Make the bmap intent items take an active reference to the rtgroup
containing the space that is being mapped or unmapped.  We will need
this functionality once we start enabling rmap and reflink on the rt
volume.  Technically speaking we need it even for !rtgroups filesystems
to prevent the (dummy) rtgroup 0 from going away, even though this will
never happen.

As a bonus, we can rework the xfs_bmap_deferred_class tracepoint to use
the xfs_group object to figure out the type and group number, widen the
group block number field to fit 64-bit quantities, and get rid of the
now redundant opdev and rtblock fields.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: grow the realtime section when realtime groups are enabled
Darrick J. Wong [Tue, 15 Oct 2024 19:39:10 +0000 (12:39 -0700)]
xfs: grow the realtime section when realtime groups are enabled

Enable growing the rt section when realtime groups are enabled.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: encode the rtsummary in big endian format
Darrick J. Wong [Tue, 15 Oct 2024 19:39:09 +0000 (12:39 -0700)]
xfs: encode the rtsummary in big endian format

Currently, the ondisk realtime summary file counters are accessed in
units of 32-bit words.  There's no endian translation of the contents of
this file, which means that the Bad Things Happen(tm) if you go from
(say) x86 to powerpc.  Since we have a new feature flag, let's take the
opportunity to enforce an endianness on the file.  Encode the summary
information in big endian format, like most of the rest of the
filesystem.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: encode the rtbitmap in big endian format
Darrick J. Wong [Tue, 15 Oct 2024 19:39:08 +0000 (12:39 -0700)]
xfs: encode the rtbitmap in big endian format

Currently, the ondisk realtime bitmap file is accessed in units of
32-bit words.  There's no endian translation of the contents of this
file, which means that the Bad Things Happen(tm) if you go from (say)
x86 to powerpc.  Since we have a new feature flag, let's take the
opportunity to enforce an endianness on the file.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: add block headers to realtime bitmap and summary blocks
Darrick J. Wong [Tue, 15 Oct 2024 19:39:07 +0000 (12:39 -0700)]
xfs: add block headers to realtime bitmap and summary blocks

Upgrade rtbitmap and rtsummary blocks to have self describing metadata
like most every other thing in XFS.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: export the geometry of realtime groups to userspace
Darrick J. Wong [Tue, 15 Oct 2024 19:39:07 +0000 (12:39 -0700)]
xfs: export the geometry of realtime groups to userspace

Create an ioctl so that the kernel can report the status of realtime
groups to userspace.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: record rt group metadata errors in the health system
Darrick J. Wong [Tue, 15 Oct 2024 19:39:06 +0000 (12:39 -0700)]
xfs: record rt group metadata errors in the health system

Record the state of per-rtgroup metadata sickness in the rtgroup
structure for later reporting.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: convert sick_map loops to use ARRAY_SIZE
Darrick J. Wong [Tue, 15 Oct 2024 19:39:05 +0000 (12:39 -0700)]
xfs: convert sick_map loops to use ARRAY_SIZE

Convert these arrays to use ARRAY_SIZE insteead of requiring an empty
sentinel array element at the end.  This saves memory and would have
avoided a bug that worked its way into the next patch.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: add frextents to the lazysbcounters when rtgroups enabled
Darrick J. Wong [Tue, 15 Oct 2024 19:39:04 +0000 (12:39 -0700)]
xfs: add frextents to the lazysbcounters when rtgroups enabled

Make the free rt extent count a part of the lazy sb counters when the
realtime groups feature is enabled.  This is possible because the patch
to recompute frextents from the rtbitmap during log recovery predates
the code adding rtgroup support, hence we know that the value will
always be correct during runtime.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: add a helper to prevent bmap merges across rtgroup boundaries
Christoph Hellwig [Tue, 15 Oct 2024 19:39:04 +0000 (12:39 -0700)]
xfs: add a helper to prevent bmap merges across rtgroup boundaries

Except for the rt superblock, realtime groups do not store any metadata
at the start (or end) of the group.  There is nothing to prevent the
bmap code from merging allocations from multiple groups into a single
bmap record.  Add a helper to check for this case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
[djwong: massage the commit message after pulling this into rtgroups]
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: check that rtblock extents do not break rtsupers or rtgroups
Darrick J. Wong [Tue, 15 Oct 2024 19:39:03 +0000 (12:39 -0700)]
xfs: check that rtblock extents do not break rtsupers or rtgroups

Check that rt block pointers do not point to the realtime superblock and
that allocated rt space extents do not cross rtgroup boundaries.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: export realtime group geometry via XFS_FSOP_GEOM
Darrick J. Wong [Tue, 15 Oct 2024 19:39:02 +0000 (12:39 -0700)]
xfs: export realtime group geometry via XFS_FSOP_GEOM

Export the realtime geometry information so that userspace can query it.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: update realtime super every time we update the primary fs super
Darrick J. Wong [Tue, 15 Oct 2024 19:39:01 +0000 (12:39 -0700)]
xfs: update realtime super every time we update the primary fs super

Every time we update parts of the primary filesystem superblock that are
echoed in the rt superblock, we must update the rt super.  Avoid
changing the log to support logging to the rt device by using ordered
buffers.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: check the realtime superblock at mount time
Darrick J. Wong [Tue, 15 Oct 2024 19:39:00 +0000 (12:39 -0700)]
xfs: check the realtime superblock at mount time

Check the realtime superblock at mount time, to ensure that the label
and uuids actually match the primary superblock on the data device.  If
the rt superblock is good, attach it to the xfs_mount so that the log
can use ordered buffers to keep this primary in sync with the primary
super on the data device.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: define the format of rt groups
Darrick J. Wong [Tue, 15 Oct 2024 19:39:00 +0000 (12:39 -0700)]
xfs: define the format of rt groups

Define the ondisk format of realtime group metadata, and a superblock
for realtime volumes.  rt supers are conditionally enabled by a
predicate function so that they can be disabled if we ever implement
zoned storage support for the realtime volume.

For rt group enabled file systems there is a separate bitmap and summary
file for each group and thus the number of bitmap and summary blocks
needs to be calculated differently.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoiomap: add a merge boundary flag
Christoph Hellwig [Tue, 15 Oct 2024 19:38:59 +0000 (12:38 -0700)]
iomap: add a merge boundary flag

File systems might have boundaries over which merges aren't possible.
In fact these are very common, although most of the time some kind of
header at the beginning of this region (e.g. XFS alloation groups, ext4
block groups) automatically create a merge barrier.  But if that is
not present, say for a device purely used for data we need to manually
communicate that to iomap.

Add a IOMAP_F_BOUNDARY flag to never merge I/O into a previous mapping.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: fix rt device offset calculations for FITRIM
Darrick J. Wong [Tue, 15 Oct 2024 19:38:58 +0000 (12:38 -0700)]
xfs: fix rt device offset calculations for FITRIM

FITRIM on xfs has this bizarro uapi where we flatten all the physically
addressable storage across two block devices into a linear address
space.  In this address space, the realtime device comes immediately
after the data device.  Therefore, the xfs_trim_rtdev_extents has to
convert its input parameters from the linear address space to actual
rtdev block addresses on the realtime volume.

Right now the address space conversion is done in units of rtblocks.
However, a future patchset will convert xfs_rtblock_t to be a segmented
address space (group:blkno) like the data device.  Change the conversion
code to be done in units of daddrs since those will never be segmented.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: make RT extent numbers relative to the rtgroup
Christoph Hellwig [Tue, 15 Oct 2024 19:38:57 +0000 (12:38 -0700)]
xfs: make RT extent numbers relative to the rtgroup

To prepare for adding per-rtgroup bitmap files, make the xfs_rtxnum_t
type encode the RT extent number relative to the rtgroup.  The biggest
part of this to clearly distinguish between the relative extent number
that gets masked when converting from a global block number and length
values that just have a factor applied to them when converting from
file system blocks.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: refactor xfs_rtsummary_blockcount
Christoph Hellwig [Tue, 15 Oct 2024 19:38:57 +0000 (12:38 -0700)]
xfs: refactor xfs_rtsummary_blockcount

Make xfs_rtsummary_blockcount take all the required information from
the mount structure and return the number of summary levels from it
as well.  This cleans up many of the callers and prepares for making the
rtsummary files per-rtgroup where they need to look at different value.

This means we recalculate some values in some callers, but as all these
calculations are outside the fast path and cheap, which seems like a
price worth paying.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: refactor xfs_rtbitmap_blockcount
Christoph Hellwig [Tue, 15 Oct 2024 19:38:56 +0000 (12:38 -0700)]
xfs: refactor xfs_rtbitmap_blockcount

Rename the existing xfs_rtbitmap_blockcount to
xfs_rtbitmap_blockcount_len and add a new xfs_rtbitmap_blockcount wrapper
around it that takes the number of extents from the mount structure.

This will simplify the move to per-rtgroup bitmaps as those will need to
pass in the number of extents per rtgroup instead.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: factor out a xfs_growfs_check_rtgeom helper
Christoph Hellwig [Tue, 15 Oct 2024 19:38:55 +0000 (12:38 -0700)]
xfs: factor out a xfs_growfs_check_rtgeom helper

Split the check that the rtsummary fits into the log into a separate
helper, and use xfs_growfs_rt_alloc_fake_mount to calculate the new RT
geometry.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
[djwong: avoid division for the 0-rtx growfs check]
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: use xfs_growfs_rt_alloc_fake_mount in xfs_growfs_rt_alloc_blocks
Christoph Hellwig [Tue, 15 Oct 2024 19:38:54 +0000 (12:38 -0700)]
xfs: use xfs_growfs_rt_alloc_fake_mount in xfs_growfs_rt_alloc_blocks

Use xfs_growfs_rt_alloc_fake_mount instead of manually recalculating
the RT bitmap geometry.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: factor out a xfs_growfs_rt_alloc_fake_mount helper
Darrick J. Wong [Tue, 15 Oct 2024 19:38:54 +0000 (12:38 -0700)]
xfs: factor out a xfs_growfs_rt_alloc_fake_mount helper

Split the code to set up a fake mount point to calculate new RT
geometry out of xfs_growfs_rt_bmblock so that it can be reused.

Note that this changes the rmblocks calculation method to be based
on the passed in rblocks and extsize and not the explicitly passed
one, but both methods will always lead to the same result.  The new
version just does a little bit more math while being more general.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: calculate RT bitmap and summary blocks based on sb_rextents
Christoph Hellwig [Tue, 15 Oct 2024 19:38:53 +0000 (12:38 -0700)]
xfs: calculate RT bitmap and summary blocks based on sb_rextents

Use the on-disk rextents to calculate the bitmap and summary blocks
instead of the calculated one so that we can refactor the helpers for
calculating them.

As the RT bitmap and summary scrubbers already check that sb_rextents
match the block count this does not change coverage of the scrubber.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: remove XFS_ILOCK_RT*
Darrick J. Wong [Tue, 15 Oct 2024 19:38:52 +0000 (12:38 -0700)]
xfs: remove XFS_ILOCK_RT*

Now that we've centralized the realtime metadata locking routines, get
rid of the ILOCK subclasses since we now use explicit lockdep classes.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: support creating per-RTG files in growfs
Christoph Hellwig [Tue, 15 Oct 2024 19:38:51 +0000 (12:38 -0700)]
xfs: support creating per-RTG files in growfs

To support adding new RT groups in growfs, we need to be able to create
the per-RT group files.  Add a new xfs_rtginode_create helper to create
a given per-RTG file.  Most of the code for that is shared, but the
details of the actual file are abstracted out using a new create method
in struct xfs_rtginode_ops.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: move RT bitmap and summary information to the rtgroup
Christoph Hellwig [Tue, 15 Oct 2024 19:38:50 +0000 (12:38 -0700)]
xfs: move RT bitmap and summary information to the rtgroup

Move the pointers to the RT bitmap and summary inodes as well as the
summary cache to the rtgroups structure to prepare for having a
separate bitmap and summary inodes for each rtgroup.

Code using the inodes now needs to operate on a rtgroup.  Where easily
possible such code is converted to iterate over all rtgroups, else
rtgroup 0 (the only one that can currently exist) is hardcoded.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: split xfs_trim_rtdev_extents
Christoph Hellwig [Tue, 15 Oct 2024 19:38:50 +0000 (12:38 -0700)]
xfs: split xfs_trim_rtdev_extents

Split xfs_trim_rtdev_extents into two parts to prepare for reusing the
main validation also for RT group aware file systems.

Use the fully features xfs_daddr_to_rtb helper to convert from a daddr
to a xfs_rtblock_t to prepare for segmented addressing in RT groups.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: cleanup xfs_getfsmap_rtdev_rtbitmap
Christoph Hellwig [Tue, 15 Oct 2024 19:38:49 +0000 (12:38 -0700)]
xfs: cleanup xfs_getfsmap_rtdev_rtbitmap

Use mp->m_sb.sb_rblocks to calculate the end instead of sb_rextents that
needs a conversion, use consistent names to xfs_rtblock_t types, and
only calculated them by the time they are needed.  Remove the pointless
"high" local variable that only has a single user.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: factor out a xfs_growfs_rt_alloc_blocks helper
Christoph Hellwig [Tue, 15 Oct 2024 19:38:48 +0000 (12:38 -0700)]
xfs: factor out a xfs_growfs_rt_alloc_blocks helper

Split out a helper to allocate or grow the rtbitmap and rtsummary files
in preparation of per-RT group bitmap and summary files.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: add a xfs_qm_unmount_rt helper
Christoph Hellwig [Tue, 15 Oct 2024 19:38:47 +0000 (12:38 -0700)]
xfs: add a xfs_qm_unmount_rt helper

RT group enabled file systems fix the bug where we pointlessly attach
quotas to the RT bitmap and summary files.  Split the code to detach the
quotas into a helper, make it conditional and document the differing
behavior for RT group and pre-RT group file systems.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: add a xfs_bmap_free_rtblocks helper
Christoph Hellwig [Tue, 15 Oct 2024 19:38:47 +0000 (12:38 -0700)]
xfs: add a xfs_bmap_free_rtblocks helper

Split the RT extent freeing logic from xfs_bmap_del_extent_real because
it will become more complicated when adding RT group.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: add rtgroup-based realtime scrubbing context management
Darrick J. Wong [Tue, 15 Oct 2024 19:38:46 +0000 (12:38 -0700)]
xfs: add rtgroup-based realtime scrubbing context management

Create a state tracking structure and helpers to initialize the tracking
structure so that we can check metadata records against the realtime
space management metadata.  Right now this is limited to grabbing the
incore rtgroup object, but we'll eventually add to the tracking
structure the ILOCK state and btree cursors.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: support caching rtgroup metadata inodes
Darrick J. Wong [Tue, 15 Oct 2024 19:38:45 +0000 (12:38 -0700)]
xfs: support caching rtgroup metadata inodes

Create the necessary per-rtgroup infrastructure that we need to load
metadata inodes into memory.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: add a lockdep class key for rtgroup inodes
Darrick J. Wong [Tue, 15 Oct 2024 19:38:44 +0000 (12:38 -0700)]
xfs: add a lockdep class key for rtgroup inodes

Add a dynamic lockdep class key for rtgroup inodes.  This will enable
lockdep to deduce inconsistencies in the rtgroup metadata ILOCK locking
order.  Each class can have 8 subclasses, and for now we will only have
2 inodes per group.  This enables rtgroup order and inode order checks
when nesting ILOCKs.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: define locking primitives for realtime groups
Darrick J. Wong [Tue, 15 Oct 2024 19:38:43 +0000 (12:38 -0700)]
xfs: define locking primitives for realtime groups

Define helper functions to lock all metadata inodes related to a
realtime group.  There's not much to look at now, but this will become
important when we add per-rtgroup metadata files and online fsck code
for them.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: create incore realtime group structures
Darrick J. Wong [Tue, 15 Oct 2024 19:38:43 +0000 (12:38 -0700)]
xfs: create incore realtime group structures

Create an incore object that will contain information about a realtime
allocation group.  This will eventually enable us to shard the realtime
section in a similar manner to how we shard the data section, but for
now just a single object for the entire RT subvolume is created.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: clean up xfs_getfsmap_helper arguments
Christoph Hellwig [Tue, 15 Oct 2024 19:38:42 +0000 (12:38 -0700)]
xfs: clean up xfs_getfsmap_helper arguments

The calling conventions for xfs_getfsmap_helper are confusing -- callers
pass in an rmap record, but they must also supply startblock and
blockcount in daddr units.  This was bolted onto the original fsmap
implementation so that we could report *something* for realtime
volumes, which do not support rmap and hence can draw only from the rt
free space bitmap.  Free space on the rt volume can be more than 2^32
fsblocks long, which means that we can't use the rmap startblock or
blockcount fields.

This is confusing for callers, because they must supplying redundant
data, but not all of it is used.  Streamline this by creating a separate
fsmap irec structure that contains exactly the data we need, once.

Note that we actually do need rm_startblock for rmap key comparisons
when we're actually querying an rmap btree, so leave that field but
document why it's there.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: repair metadata directory file path connectivity
Darrick J. Wong [Tue, 15 Oct 2024 19:38:41 +0000 (12:38 -0700)]
xfs: repair metadata directory file path connectivity

Fix disconnected or incorrect metadata directory paths.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: confirm dotdot target before replacing it during a repair
Darrick J. Wong [Tue, 15 Oct 2024 19:38:40 +0000 (12:38 -0700)]
xfs: confirm dotdot target before replacing it during a repair

xfs_dir_replace trips an assertion if you tell it to change a dirent to
point to an inumber that it already points at.  Look up the dotdot entry
directly to confirm that we need to make a change.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: check metadata directory file path connectivity
Darrick J. Wong [Tue, 15 Oct 2024 19:38:40 +0000 (12:38 -0700)]
xfs: check metadata directory file path connectivity

Create a new scrubber type that checks that well known metadata
directory paths are connected to the metadata inode that the incore
structures think is in use.  For example, check that "/quota/user" in
the metadata directory tree actually points to
mp->m_quotainfo->qi_uquotaip->i_ino.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: move repair temporary files to the metadata directory tree
Darrick J. Wong [Tue, 15 Oct 2024 19:38:39 +0000 (12:38 -0700)]
xfs: move repair temporary files to the metadata directory tree

Due to resource acquisition rules, we have to create the ondisk
temporary files used to stage a filesystem repair before we can acquire
a reference to the inode that we actually want to repair.  Therefore,
we do not know at tempfile creation time whether the tempfile will
belong to the regular directory tree or the metadata directory tree.

This distinction becomes important when the swapext code tries to figure
out the quota accounting of the two files whose mappings are being
swapped.  The swapext code assumes that accounting updates are required
for a file if dqattach attaches dquots.  Metadir files are never
accounted in quota, which means that swapext must not update the quota
accounting when swapping in a repaired directory/xattr/rtbitmap structure.

Prior to the swapext call, therefore, both files must be marked as
METADIR for dqattach so that dqattach will ignore them.  Add support for
a repair tempfile to be switched to the metadir tree and switched back
before being released so that ifree will just free the file.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: check the metadata directory inumber in superblocks
Darrick J. Wong [Tue, 15 Oct 2024 19:38:38 +0000 (12:38 -0700)]
xfs: check the metadata directory inumber in superblocks

When metadata directories are enabled, make sure that the secondary
superblocks point to the metadata directory.  This isn't strictly
required because the secondaries are only used to recover damaged
filesystems, and the metadir root inumber is fixed.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: scrub metadata directories
Darrick J. Wong [Tue, 15 Oct 2024 19:38:37 +0000 (12:38 -0700)]
xfs: scrub metadata directories

Teach online scrub about the metadata directory tree.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: fix di_metatype field of inodes that won't load
Darrick J. Wong [Tue, 15 Oct 2024 19:38:37 +0000 (12:38 -0700)]
xfs: fix di_metatype field of inodes that won't load

Make sure that the di_metatype field is at least set plausibly so that
later scrubbers could set the real type.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: adjust parent pointer scrubber for sb-rooted metadata files
Darrick J. Wong [Tue, 15 Oct 2024 19:38:36 +0000 (12:38 -0700)]
xfs: adjust parent pointer scrubber for sb-rooted metadata files

Starting with the metadata directory feature, we're allowed to call the
directory and parent pointer scrubbers for every metadata file,
including the ones that are children of the superblock.

For these children, checking the link count against the number of parent
pointers is a bit funny -- there's no such thing as a parent pointer for
a child of the superblock since there's no corresponding dirent.  For
purposes of validating nlink, we pretend that there is a parent pointer.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: metadata files can have xattrs if metadir is enabled
Darrick J. Wong [Tue, 15 Oct 2024 19:38:35 +0000 (12:38 -0700)]
xfs: metadata files can have xattrs if metadir is enabled

If parent pointers are enabled, then metadata files will store parent
pointers in xattrs, just like files in the user visible directory tree.
Therefore, scrub and repair need to handle attr forks for metadata files
on metadir filesystems.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: do not count metadata directory files when doing online quotacheck
Darrick J. Wong [Tue, 15 Oct 2024 19:38:34 +0000 (12:38 -0700)]
xfs: do not count metadata directory files when doing online quotacheck

Previously, we stated that files in the metadata directory tree are not
counted in the dquot information.  Fix the online quotacheck code to
reflect this.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: refactor directory tree root predicates
Darrick J. Wong [Tue, 15 Oct 2024 19:38:33 +0000 (12:38 -0700)]
xfs: refactor directory tree root predicates

Metadata directory trees make reasoning about the parent of a file more
difficult.  Traditionally, user files are children of sb_rootino, and
metadata files are "children" of the superblock.  Now, we add a third
possibility -- some metadata files can be children of sb_metadirino, but
the classic ones (rt free space data and quotas) are left alone.

Let's add some helper functions (instead of open-coding the logic
everywhere) to make scrub logic easier to understand.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: record health problems with the metadata directory
Darrick J. Wong [Tue, 15 Oct 2024 19:38:32 +0000 (12:38 -0700)]
xfs: record health problems with the metadata directory

Make a report to the health monitoring subsystem any time we encounter
something in the metadata directory tree that looks like corruption.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: adjust xfs_bmap_add_attrfork for metadir
Darrick J. Wong [Tue, 15 Oct 2024 19:38:31 +0000 (12:38 -0700)]
xfs: adjust xfs_bmap_add_attrfork for metadir

Online repair might use the xfs_bmap_add_attrfork to repair a file in
the metadata directory tree if (say) the metadata file lacks the correct
parent pointers.  In that case, it is not correct to check that the file
is dqattached -- metadata files must be not have /any/ dquot attached at
all.  Adjust the assertions appropriately.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: mark quota inodes as metadata files
Darrick J. Wong [Tue, 15 Oct 2024 19:38:31 +0000 (12:38 -0700)]
xfs: mark quota inodes as metadata files

When we're creating quota files at mount time, make sure to mark them as
metadir inodes if appropriate.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: don't count metadata directory files to quota
Darrick J. Wong [Tue, 15 Oct 2024 19:38:30 +0000 (12:38 -0700)]
xfs: don't count metadata directory files to quota

Files in the metadata directory tree are internal to the filesystem.
Don't count the inodes or the blocks they use in the root dquot because
users do not need to know about their resource usage.  This will also
quiet down complaints about dquot usage not matching du output.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: allow bulkstat to return metadata directories
Darrick J. Wong [Tue, 15 Oct 2024 19:38:29 +0000 (12:38 -0700)]
xfs: allow bulkstat to return metadata directories

Allow the V5 bulkstat ioctl to return information about metadata
directory files so that xfs_scrub can find and scrub them, since they
are otherwise ordinary directories.

(Metadata files of course require per-file scrub code and hence do not
need exposure.)

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: advertise metadata directory feature
Darrick J. Wong [Tue, 15 Oct 2024 19:38:28 +0000 (12:38 -0700)]
xfs: advertise metadata directory feature

Advertise the existence of the metadata directory feature; this will be
used by scrub to decide if it needs to scan the metadir too.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: hide metadata inodes from everyone because they are special
Darrick J. Wong [Tue, 15 Oct 2024 19:38:28 +0000 (12:38 -0700)]
xfs: hide metadata inodes from everyone because they are special

Metadata inodes are private files and therefore cannot be exposed to
userspace.  This means no bulkstat, no open-by-handle, no linking them
into the directory tree, and no feeding them to LSMs.  As such, we mark
them S_PRIVATE, which stops all that.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: disable the agi rotor for metadata inodes
Darrick J. Wong [Tue, 15 Oct 2024 19:38:27 +0000 (12:38 -0700)]
xfs: disable the agi rotor for metadata inodes

Ideally, we'd put all the metadata inodes in one place if we could, so
that the metadata all stay reasonably close together instead of
spreading out over the disk.  Furthermore, if the log is internal we'd
probably prefer to keep the metadata near the log.  Therefore, disable
AGI rotoring for metadata inode allocations.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: read and write metadata inode directory tree
Darrick J. Wong [Tue, 15 Oct 2024 19:38:26 +0000 (12:38 -0700)]
xfs: read and write metadata inode directory tree

Plumb in the bits we need to load metadata inodes from a named entry in
a metadir directory, create (or hardlink) inodes into a metadir
directory, create metadir directories, and flag inodes as being metadata
files.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: enforce metadata inode flag
Darrick J. Wong [Tue, 15 Oct 2024 19:38:25 +0000 (12:38 -0700)]
xfs: enforce metadata inode flag

Add checks for the metadata inode flag so that we don't ever leak
metadata inodes out to userspace, and we don't ever try to read a
regular inode as metadata.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: load metadata directory root at mount time
Darrick J. Wong [Tue, 15 Oct 2024 19:38:24 +0000 (12:38 -0700)]
xfs: load metadata directory root at mount time

Load the metadata directory root inode into memory at mount time and
release it at unmount time.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: iget for metadata inodes
Darrick J. Wong [Tue, 15 Oct 2024 19:38:24 +0000 (12:38 -0700)]
xfs: iget for metadata inodes

Create a xfs_trans_metafile_iget function for metadata inodes to ensure
that when we try to iget a metadata file, the inode is allocated and its
file mode matches the metadata file type the caller expects.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: define the on-disk format for the metadir feature
Darrick J. Wong [Tue, 15 Oct 2024 19:38:22 +0000 (12:38 -0700)]
xfs: define the on-disk format for the metadir feature

Define the on-disk layout and feature flags for the metadata inode
directory feature.  Add a xfs_sb_version_hasmetadir for benefit of
xfs_repair, which needs to know where the new end of the superblock
lies.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: standardize EXPERIMENTAL warning generation
Darrick J. Wong [Tue, 15 Oct 2024 22:00:10 +0000 (15:00 -0700)]
xfs: standardize EXPERIMENTAL warning generation

Refactor the open-coded warnings about EXPERIMENTAL feature use into a
standard helper before we go adding more experimental features.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: rename metadata inode predicates
Darrick J. Wong [Tue, 15 Oct 2024 21:50:14 +0000 (14:50 -0700)]
xfs: rename metadata inode predicates

The predicate xfs_internal_inum tells us if an inumber refers to one of
the inodes rooted in the superblock.  Soon we're going to have internal
inodes in a metadata directory tree, so this helper should be renamed
to capture its limited scope.

Ondisk inodes will soon have a flag to indicate that they're metadata
inodes.  Head off some confusion by renaming the xfs_is_metadata_inode
predicate to xfs_is_internal_inode.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: constify the xfs_inode predicates
Darrick J. Wong [Tue, 15 Oct 2024 19:38:21 +0000 (12:38 -0700)]
xfs: constify the xfs_inode predicates

Change the xfs_inode predicates to take a const struct xfs_inode pointer
because they do not change the inode.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: constify the xfs_sb predicates
Darrick J. Wong [Tue, 15 Oct 2024 19:38:21 +0000 (12:38 -0700)]
xfs: constify the xfs_sb predicates

Change the xfs_sb predicates to take a const struct xfs_sb pointer
because they do not change the superblock.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
5 months agoxfs: store a generic group structure in the intents
Christoph Hellwig [Tue, 15 Oct 2024 19:38:20 +0000 (12:38 -0700)]
xfs: store a generic group structure in the intents

Replace the pag pointers in the extent free, bmap, rmap and refcount
intent structures with a pointer to the generic group to prepare
for adding intents for realtime groups.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: remove xfs_group_intent_hold and xfs_group_intent_rele
Christoph Hellwig [Tue, 15 Oct 2024 19:38:19 +0000 (12:38 -0700)]
xfs: remove xfs_group_intent_hold and xfs_group_intent_rele

Each of them just has a single caller, so fold them.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: add group based bno conversion helpers
Christoph Hellwig [Tue, 15 Oct 2024 19:38:18 +0000 (12:38 -0700)]
xfs: add group based bno conversion helpers

Add/move the blocks, blklog and blkmask fields to the generic groups
structure so that code can work with AGs and RTGs by just using the
right index into the array.

Then, add convenience helpers to convert block numbers based on the
generic group.  This will allow writing code that doesn't care if it is
used on AGs or the upcoming realtime groups.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: store a generic xfs_group pointer in xfs_getfsmap_info
Christoph Hellwig [Tue, 15 Oct 2024 19:38:18 +0000 (12:38 -0700)]
xfs: store a generic xfs_group pointer in xfs_getfsmap_info

Replace the pag and rtg pointers with a generic group pointer.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
5 months agoxfs: add a generic group pointer to the btree cursor
Christoph Hellwig [Tue, 15 Oct 2024 19:38:17 +0000 (12:38 -0700)]
xfs: add a generic group pointer to the btree cursor

Replace the pag pointers in the type specific union with a generic
xfs_group pointer.  This prepares for adding realtime group support.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>