djwong/xfsprogs

Create a systemd service unit so that we can run the online scrubber
under systemd with (somewhat) appropriate containment.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create an xfs_scrub_all command to find all XFS filesystems
and run an online scrub against them all.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Implement a progress indicator.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Iterate all directory and xattr names to look for name collisions
amongst Unicode normalized names.  This is generally a sign of buggy
programs or malicious duplicate files.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

…stem

If the filesystem scan comes out clean or fixes all the problems, call
fstrim to clean out the free areas (if it's an ssd/thinp/whatever).

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Teach xfs_scrub to remember scrub requests that failed (or indicated
that optimization is a possibility) as repair requests that can be
deferred until later.  Add a new repair phase that deals with the
repair requests.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create the mechanism we need to actually call the kernel's online repair
functionality.  The interface will consume a repair description; the
descriptor management will follow in the next patch.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Make sure the filesystem summary counters are somewhat close to what
we can find by scanning the filesystem.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

…cks on disk

If we sense that we're talking to a raw SCSI disk, use the SCSI READ
VERIFY command to ask the disk to verify a disk internally.  This can
sharply reduce the runtime of the data block verification phase on
devices whose internal bandwidth exceeds their link bandwidth.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Read all data blocks from the disk, hoping to catch IO errors.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create a couple of functions to iterate a file's block map or the
filesystem's space map.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Manage the scheduling, issuance, and reporting of data block
verification reads.  This enables us to combine adjacent (or nearly
adjacent) read requests, and to take advantage of high-IOPS devices by
issuing IO from multiple threads.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create an efficient tree-based bitmap data structure.  We will use this
during the data block scan to record the LBAs of IO errors so that we
can report broken files to userspace.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create a threaded stats counter that we'll use to track scan progress.
This includes things like how much of the disk blocks we've scanned,
or later how much progress we've made in each phase.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Opening directories by file handle will cause the kernel to perform
parent lookups all the way to the root directory.  Take advantage of
this to ensure that directories actually connect to the root.  Some
day we'll have parent pointers and can make this more comprehensive.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Scan all the inodes in the system for problems.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Scrub the filesystem and per-AG metadata.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Discover the geometry of the XFS filesystem that we've been told to
scan, and set up some common functions that will be used by the
scrub phases.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create an abstraction to handle all of our low level disk operations,
then use it to bind to a fs mount point and block device.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create the dispatching routines that we'll use to call out to each
separate phase of the program.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Parse command line options in order to set up the context in which we
will scrub the filesystem.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Standardize how we record and report errors.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create the foundations of a filesystem scrubbing tool that asks the
kernel to inspect all metadata in the filesystem and (ultimately) to
repair anything that's broken.  Also create the man page for the
utility.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Document the XFS-specific metadata scrub/repair ioctl's behavior,
arguments, and side effects.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Create a new xfs_io command to call the new XFS metadata scrub ioctl.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Use rmap records to rebuild corrupt inode forks instead of zapping
the whole inode.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

…ork processing

Currently, the process_dinode* family of functions assume that they have
the buffer backing the inodes locked, and therefore the dinode pointer
won't ever change.  However, the bmbt rebuilding code in the next patch
will violate that assumption, so we must pass pointers to the inobp and
the dinode pointer (that is to say, double pointers) all the way through
to process_inode_{data,attr}_fork so that we can regrab the buffer after
the rebuilding step finishes.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

…g rmpabt

Add a new flag, XFS_BMAPI_NORMAP, which will perform file block
remapping without updating the rmapbt.  This will be used by the repair
code to reconstruct bmbts from the rmapbt, in which case we don't want
the rmapbt update.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Add a couple of functions to the refcount btree and generic btree code
that will be used to repair the refcountbt.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Add a couple of functions to the reverse mapping btree that will be used
to repair the rmapbt.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Expose various helpers that the repair code will want to use.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Add a bunch of helper functions that calculate the sizes of various
btrees.  These will be used to repair btrees and btree headers.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

While we're scrubbing various btrees, cross-reference the records
with the other metadata.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Add a couple of functions to the refcount btrees that will be used
to cross-reference metadata against the refcountbt.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Add a couple of functions to the rmap btrees that will be used
to cross-reference metadata against the rmapbt.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>