IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
Nick Piggin reported that the CRC overhead in an fsync heavy
workload was higher than expected on a Power8 machine. Part of this
was to do with the fact that the power8 CRC implementation is not
efficient for CRC lengths of less than 512 bytes, and so the way we
split the CRCs over the CRC field means a lot of the CRCs are
reduced to being less than than optimal size.
To optimise this, change the CRC update mechanism to zero the CRC
field first, and then compute the CRC in one pass over the buffer
and write the result back into the buffer. We can do this safely
because anything writing a CRC has exclusive access to the buffer
the CRC is being calculated over.
We leave the CRC verify code the same - it still splits the CRC
calculation - because we do not want read-only operations modifying
the underlying buffer. This is because read-only operations may not
have an exclusive access to the buffer guaranteed, and so temporary
modifications could leak out to to other processes accessing the
buffer concurrently.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
We've missed properly setting the buffer type for
an AGI transaction in 3 spots now, so just move it
into xfs_read_agi() and set it if we are in a transaction
to avoid the problem in the future.
This is similar to how it is done in i.e. the dir3
and attr3 read functions.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xlog_recover_clear_agi_bucket didn't set the
type to XFS_BLFT_AGI_BUF, so we got a warning during log
replay (or an ASSERT on a debug build).
XFS (md0): Unknown buffer type 0!
XFS (md0): _xfs_buf_ioapply: no ops on block 0xaea8802/0x1
Fix this, as was done in f19b872b for 2 other locations
with the same problem.
cc: <stable@vger.kernel.org> # 3.10 to current
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Since xfsprogs dropped ushort in favor of unsigned short, do that
here too.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Log recovery will iget an inode to replay BUI items and iput the inode
when it's done. Unfortunately, if the inode was unlinked, the iput
will see that i_nlink == 0 and decide to truncate & free the inode,
which prevents us from replaying subsequent BUIs. We can't skip the
BUIs because we have to replay all the redo items to ensure that
atomic operations complete.
Since unlinked inode recovery will reap the inode anyway, we can
safely introduce a new inode flag to indicate that an inode is in this
'unlinked recovery' state and should not be auto-reaped in the
drop_inode path.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Provide a mechanism for higher levels to create BUI/BUD items, submit
them to the log, and a stub function to deal with recovered BUI items.
These parts will be connected to the rmapbt in a later patch.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Identify refcountbt blocks in the log correctly so that we can
validate them during log recovery.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Provide a mechanism for higher levels to create CUI/CUD items, submit
them to the log, and a stub function to deal with recovered CUI items.
These parts will be connected to the refcountbt in a later patch.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Log recovery has particular rules around buffer submission along with
tricky corner cases where independent transactions can share an LSN. As
such, it can be difficult to follow when/why buffers are submitted
during recovery.
Add a couple tracepoints to post the current LSN of a record when a new
record is being processed and when a buffer is being skipped due to LSN
ordering. Also, update the recover item class to include the LSN of the
current transaction for the item being processed.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Log recovery is currently broken for v5 superblocks in that it never
updates the metadata LSN of buffers written out during recovery. The
metadata LSN is recorded in various bits of metadata to provide recovery
ordering criteria that prevents transient corruption states reported by
buffer write verifiers. Without such ordering logic, buffer updates can
be replayed out of order and lead to false positive transient corruption
states. This is generally not a corruption vector on its own, but
corruption detection shuts down the filesystem and ultimately prevents a
mount if it occurs during log recovery. This requires an xfs_repair run
that clears the log and potentially loses filesystem updates.
This problem is avoided in most cases as metadata writes during normal
filesystem operation update the metadata LSN appropriately. The problem
with log recovery not updating metadata LSNs manifests if the system
happens to crash shortly after log recovery itself. In this scenario, it
is possible for log recovery to complete all metadata I/O such that the
filesystem is consistent. If a crash occurs after that point but before
the log tail is pushed forward by subsequent operations, however, the
next mount performs the same log recovery over again. If a buffer is
updated multiple times in the dirty range of the log, an earlier update
in the log might not be valid based on the current state of the
associated buffer after all of the updates in the log had been replayed
(before the previous crash). If a verifier happens to detect such a
problem, the filesystem claims corruption and immediately shuts down.
This commonly manifests in practice as directory block verifier failures
such as the following, likely due to directory verifiers being
particularly detailed in their checks as compared to most others:
...
Mounting V5 Filesystem
XFS (dm-0): Starting recovery (logdev: internal)
XFS (dm-0): Internal error XFS_WANT_CORRUPTED_RETURN at line ... of \
file fs/xfs/libxfs/xfs_dir2_data.c. Caller xfs_dir3_data_verify ...
...
Update log recovery to update the metadata LSN of recovered buffers.
Since metadata LSNs are already updated by write verifer functions via
attached log items, attach a dummy log item to the buffer during
validation and explicitly set the LSN of the current transaction. This
ensures that the metadata LSN of a buffer is updated based on whether
the recovery I/O actually completes, and if so, that subsequent recovery
attempts identify that the buffer is already up to date with respect to
the current transaction.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The log recovery buffer validation function is invoked in cases where a
buffer update may be skipped due to LSN ordering. If the validation
function happens to come across directory conversion situations (e.g., a
dir3 block to data conversion), it may warn about seeing a buffer log
format of one type and a buffer with a magic number of another.
This warning is not valid as the buffer update is ultimately skipped.
This is indicated by a current_lsn of NULLCOMMITLSN provided by the
caller. As such, update xlog_recover_validate_buf_type() to only warn in
such cases when a buffer update is expected.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The current LSN must be available to the buffer validation function to
provide the ability to update the metadata LSN of the buffer. Pass the
current_lsn value down to xlog_recover_validate_buf_type() in
preparation.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The fix to log recovery to update the metadata LSN in recovered buffers
introduces the requirement that a buffer is submitted only once per
current LSN. Log recovery currently submits buffers on transaction
boundaries. This is not sufficient as the abstraction between log
records and transactions allows for various scenarios where multiple
transactions can share the same current LSN. If independent transactions
share an LSN and both modify the same buffer, log recovery can
incorrectly skip updates and leave the filesystem in an inconsisent
state.
In preparation for proper metadata LSN updates during log recovery,
update log recovery to submit buffers for write on LSN change boundaries
rather than transaction boundaries. Explicitly track the current LSN in
a new struct xlog field to handle the various corner cases of when the
current LSN may or may not change.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Nothing ever uses the extent array in the rmap update done redo
item, so remove it before it is fixed in the on-disk log format.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Originally-From: Dave Chinner <dchinner@redhat.com>
So such blocks can be correctly identified and have their operations
structures attached to validate recovery has not resulted in a
correct block.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Provide a mechanism for higher levels to create RUI/RUD items, submit
them to the log, and a stub function to deal with recovered RUI items.
These parts will be connected to the rmapbt in a later patch.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Originally-From: Dave Chinner <dchinner@redhat.com>
The rmap btree is allocated from the AGFL, which means we have to
ensure ENOSPC is reported to userspace before we run out of free
space in each AG. The last allocation in an AG can cause a full
height rmap btree split, and that means we have to reserve at least
this many blocks *in each AG* to be placed on the AGFL at ENOSPC.
Update the various space calculation functions to handle this.
Also, because the macros are now executing conditional code and are
called quite frequently, convert them to functions that initialise
variables in the struct xfs_mount, use the new variables everywhere
and document the calculations better.
[darrick.wong@oracle.com: don't reserve blocks if !rmap]
[dchinner@redhat.com: update m_ag_max_usable after growfs]
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Refactor the EFI intent item recovery (and cancellation) functions
into a general function that scans the AIL and an intent item type
specific handler. Move the function that recovers a single EFI item
into the extent free item code. We'll want the generalized function
when we start wiring up more redo item types.
Furthermore, ensure that log recovery only replays the redo items
that were in the AIL prior to recovery by checking the item LSN
against the largest LSN seen during log scanning. As written this
should never happen, but we can be defensive anyway.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Restructure everything that used xfs_bmap_free to use xfs_defer_ops
instead. For now we'll just remove the old symbols and play some
cpp magic to make it work; in the next patch we'll actually rename
everything.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Use krealloc to implement our realloc function. This helps to avoid
new allocations if we are still in the slab bucket. At least for the
bmap btree root that's actually the common case.
This also allows removing the now unused oldsize argument.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Merge xfs_trans_reserve and xfs_trans_alloc into a single function call
that returns a transaction with all the required log and block reservations,
and which allows passing transaction flags directly to avoid the cumbersome
_xfs_trans_alloc interface.
While we're at it we also get rid of the transaction type argument that has
been superflous since we stopped supporting the non-CIL logging mode. The
guts of it will be removed in another patch.
[dchinner: fixed transaction leak in error path in xfs_setattr_nonsize]
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
If a crash occurs immediately after a filesystem grow operation, the
updated superblock geometry is found only in the log. After we
recover the log, the superblock is reread and re-initialised and so
has the new geometry in memory. If the new geometry has more AGs
than prior to the grow operation, then the new AGs will not have
in-memory xfs_perag structurea associated with them.
This will result in an oops when the first metadata buffer from a
new AG is looked up in the buffer cache, as the block lies within
the new geometry but then fails to find a perag structure on lookup.
This is easily fixed by simply re-initialising the perag structure
after re-reading the superblock at the conclusion of the first pahse
of log recovery.
This, however, does not fix the case of log recovery requiring
access to metadata in the newly grown space. Fortunately for us,
because the in-core superblock has not been updated, this will
result in detection of access beyond the end of the filesystem
and so recovery will fail at that point. If this proves to be
a problem, then we can address it separately to the current
reported issue.
Reported-by: Alex Lyakas <alex@zadarastorage.com>
Tested-by: Alex Lyakas <alex@zadarastorage.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
XFS uses CRC verification over a sub-range of the head of the log to
detect and handle torn writes. This torn log write detection currently
runs unconditionally at mount time, regardless of whether the log is
dirty or clean. This is problematic in cases where a filesystem might
end up being moved across different, incompatible (i.e., opposite
byte-endianness) architectures.
The problem lies in the fact that log data is not necessarily written in
an architecture independent format. For example, certain bits of data
are written in native endian format. Further, the size of certain log
data structures differs (i.e., struct xlog_rec_header) depending on the
word size of the cpu. This leads to false positive crc verification
errors and ultimately failed mounts when a cleanly unmounted filesystem
is mounted on a system with an incompatible architecture from data that
was written near the head of the log.
Update the log head/tail discovery code to run torn write detection only
when the log is not clean. This means something other than an unmount
record resides at the head of the log and log recovery is imminent. It
is a requirement to run log recovery on the same type of host that had
written the content of the dirty log and therefore CRC failures are
legitimate corruptions in that scenario.
Reported-by: Jan Beulich <JBeulich@suse.com>
Tested-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Once the record at the head of the log is identified and verified, the
in-core log state is updated based on the record. This includes
information such as the current head block and cycle, the start block of
the last record written to the log, the tail lsn, etc.
Once torn write detection is conditional, this logic will need to be
reused. Factor the code to update the in-core log data structures into a
new helper function. This patch does not change behavior.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Once the mount sequence has identified the head and tail blocks of the
physical log, the record at the head of the log is located and examined
for an unmount record to determine if the log is clean. This currently
occurs after torn write verification of the head region of the log.
This must ultimately be separated from torn write verification and may
need to be called again if the log head is walked back due to a torn
write (to determine whether the new head record is an unmount record).
Separate this logic into a new helper function. This patch does not
change behavior.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The code that locates the log record at the head of the log is buried in
the log head verification function. This is fine when torn write
verification occurs unconditionally, but this behavior is problematic
for filesystems that might be moved across systems with different
architectures.
In preparation for separating examination of the log head for unmount
records from torn write detection, lift the record location logic out of
the log verification function and into the caller. This patch does not
change behavior.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The places where we use this macro already clear unnecessary IO
flags (e.g. through xfs_bwrite()) or never have unexpected IO flags
set on them in the first place (e.g. iclog buffers). Remove the
macro from these locations, and where necessary clear only the
specific flags that are conditional in the current buffer context.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
They only set/clear/check a flag, no need for obfuscating this
with a macro.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
They only set/clear/check a flag, no need for obfuscating this
with a macro.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
They only set/clear/check a flag, no need for obfuscating this
with a macro.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
They only set/clear/check a flag, no need for obfuscating this
with a macro.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Move the di_mode value from the xfs_icdinode to the VFS inode, reducing
the xfs_icdinode byte another 2 bytes and collapsing another 2 byte hole
in the structure.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The VFS tracks the inode nlink just like the xfs_icdinode. We can
remove the variable from the icdinode and use the VFS inode variable
everywhere, reducing the size of the xfs_icdinode by a further 4
bytes.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The struct xfs_inode has two copies of the current timestamps in it,
one in the vfs inode and one in the struct xfs_icdinode. Now that we
no longer log the struct xfs_icdinode directly, we don't need to
keep the timestamps in this structure. instead we can copy them
straight out of the VFS inode when formatting the inode log item or
the on-disk inode.
This reduces the struct xfs_inode in size by 24 bytes.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
We currently carry around and log an entire inode core in the
struct xfs_inode. A lot of the information in the inode core is
duplicated in the VFS inode, but we cannot remove this duplication
of infomration because the inode core is logged directly in
xfs_inode_item_format().
Add a new function xfs_inode_item_format_core() that copies the
inode core data into a struct xfs_icdinode that is pulled directly
from the log vector buffer. This means we no longer directly
copy the inode core, but copy the structures one member at a time.
This will be slightly less efficient than copying, but will allow us
to remove duplicate and unnecessary items from the struct xfs_inode.
To enable us to do this, call the new structure a xfs_log_dinode,
so that we know it's different to the physical xfs_dinode and the
in-core xfs_icdinode.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Buffers without verifiers issue runtime warnings on XFS. We don't
have anything we can actually verify in the RT buffers (no CRCs, not
magic numbers, etc), but we still need verifiers to avoid the
warnings.
Add a set of dummy verifier operations for the realtime buffers and
apply them in the appropriate places.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Tested-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
When logging buffers, we attach a type to them that follows the
buffer all the way into the log and is used to identify the buffer
contents in log recovery. Both the realtime summary buffers and the
bitmap buffers do not have types defined or set, so when we try to
log them we see assert failure:
XFS: Assertion failed: (bip->bli_flags & XFS_BLI_STALE) || (xfs_blft_from_flags(&bip->__bli_format) > XFS_BLFT_UNKNOWN_BUF && xfs_blft_from_flags(&bip->__bli_format) < XFS_BLFT_MAX_BUF), file: fs/xfs/xfs_buf_item.c, line: 294
Fix this by adding buffer log format types for these buffers, and
add identification support into log recovery for them. Only build the log
recovery support if CONFIG_XFS_RT=y - we can't get into log recovery for real
time filesystems if support is not built into the kernel, and this avoids
potential build problems.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Tested-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Since the checksum function and the field are both __le32, don't
perform endian conversion when comparing the two. This fixes mount
failures on ppc64.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
When we do dquot readahead in log recovery, we do not use a verifier
as the underlying buffer may not have dquots in it. e.g. the
allocation operation hasn't yet been replayed. Hence we do not want
to fail recovery because we detect an operation to be replayed has
not been run yet. This problem was addressed for inodes in commit
d891400 ("xfs: inode buffers may not be valid during recovery
readahead") but the problem was not recognised to exist for dquots
and their buffers as the dquot readahead did not have a verifier.
The result of not using a verifier is that when the buffer is then
next read to replay a dquot modification, the dquot buffer verifier
will only be attached to the buffer if *readahead is not complete*.
Hence we can read the buffer, replay the dquot changes and then add
it to the delwri submission list without it having a verifier
attached to it. This then generates warnings in xfs_buf_ioapply(),
which catches and warns about this case.
Fix this and make it handle the same readahead verifier error cases
as for inode buffers by adding a new readahead verifier that has a
write operation as well as a read operation that marks the buffer as
not done if any corruption is detected. Also make sure we don't run
readahead if the dquot buffer has been marked as cancelled by
recovery.
This will result in readahead either succeeding and the buffer
having a valid write verifier, or readahead failing and the buffer
state requiring the subsequent read to resubmit the IO with the new
verifier. In either case, this will result in the buffer always
ending up with a valid write verifier on it.
Note: we also need to fix the inode buffer readahead error handling
to mark the buffer with EIO. Brian noticed the code I copied from
there wrong during review, so fix it at the same time. Add comments
linking the two functions that handle readahead verifier errors
together so we don't forget this behavioural link in future.
cc: <stable@vger.kernel.org> # 3.12 - current
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Certain types of storage, such as persistent memory, do not provide
sector atomicity for writes. This means that if a crash occurs while XFS
is writing log records, only part of those records might make it to the
storage. This is problematic because log recovery uses the cycle value
packed at the top of each log block to locate the head/tail of the log.
This can lead to CRC verification failures during log recovery and an
unmountable fs for a filesystem that is otherwise consistent.
Update log recovery to incorporate log record CRC verification as part
of the head/tail discovery process. Once the head is located via the
traditional algorithm, run a CRC-only pass over the records up to the
head of the log. If CRC verification fails, assume that the records are
torn as a matter of policy and trim the head block back to the start of
the first bad record.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
As part of the head/tail discovery process, log recovery locates the
head block and then reverse seeks to find the start of the last active
record in the log. This is non-trivial as the record itself could have
wrapped around the end of the physical log. Log recovery torn write
detection potentially needs to walk further behind the last record in
the log, as multiple log I/Os can be in-flight at one time during a
crash event.
Therefore, refactor the reverse log record header search mechanism into
a new helper that supports the ability to seek past an arbitrary number
of log records (or until the tail is hit). Update the head/tail search
mechanism to call the new helper, but otherwise there is no change in
log recovery behavior.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Log recovery torn write detection uses CRC verification over a range of
the active log to identify torn writes. Since the generic log recovery
pass code implements a superset of the functionality required for CRC
verification, it can be easily modified to support a CRC verification
only pass.
Create a new CRC pass type and update the log record processing helper
to skip everything beyond CRC verification when in this mode. This pass
will be invoked in subsequent patches to implement torn write detection.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
