The collapse range operation currently writes the entire file before
starting the collapse to avoid changes in the in-core extent list due to
writeback causing the extent count to change. Now that collapse range is
fsb based rather than extent index based it can sustain changes in the
extent list during the shift sequence without disruption.
Modify xfs_collapse_file_space() to writeback and invalidate pages
associated with the range of the file to be shifted.
xfs_free_file_space() currently has similar behavior, but the space free
need only affect the region of the file that is freed and this could
change in the future.
Also update the comments to reflect the current implementation. We
retain the eofblocks trim permanently as a best option for dealing with
delalloc extents. We don't shift delalloc extents because this scenario
only occurs with post-eof preallocation (since data must be flushed such
that the cache can be invalidated and data can be shifted). That means
said space must also be initialized before being shifted into the
accessible region of the file only to be immediately truncated off as
the last part of the collapse. In other words, the eofblocks trim will
happen anyways, we just run it first to ensure the file remains in a
consistent state throughout the collapse.
Finally, detect and fail explicitly in the event of a delalloc extent
during the extent shift. The implementation does not support delalloc
extents and the caller is expected to prevent this scenario in advance
as is done by collapse.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_bmap_shift_extents() has a variety of conditions and error checks
that make the logic difficult to follow and indent heavy. Refactor the
loop body of this function into a new xfs_bmse_shift_one() helper. This
simplifies the error checks, eliminates index decrement on merge hack by
pushing the index increment down into the helper, and makes the code
more readable by reducing multiple levels of indentation.
This is a code refactor only. The behavior of extent shift and collapse
range is not modified.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The extent shift mechanism in xfs_bmap_shift_extents() is complicated
and handles several different, non-deterministic scenarios. These
include extent shifts, extent merges and potential btree updates in
either of the former scenarios.
Refactor the code to be more linear and readable. The loop logic in
xfs_bmap_shift_extents() and some initial error checking is adjusted
slightly. The associated btree lookup and update/delete operations are
condensed into single blocks of code. This reduces the number of
btree-specific blocks and facilitates the separation of the merge
operation into a new xfs_bmse_merge() and xfs_bmse_can_merge() helpers.
This is a code refactor only. The behavior of extent shift and collapse
range is not modified.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The collapse range implementation uses a transaction per extent shift.
The progress of the overall operation is tracked via the current extent
index of the in-core extent list. This is racy because the ilock must be
dropped and reacquired for each transaction according to locking and log
reservation rules. Therefore, writeback to prior regions of the file is
possible and can change the extent count. This changes the extent to
which the current index refers and causes the collapse to fail mid
operation. To avoid this problem, the entire file is currently written
back before the collapse operation starts.
To eliminate the need to flush the entire file, use the file offset
(fsb) to track the progress of the overall extent shift operation rather
than the extent index. Modify xfs_bmap_shift_extents() to
unconditionally convert the start_fsb parameter to an extent index and
return the file offset of the extent where the shift left off, if
further extents exist. The bulk of ths function can remain based on
extent index as ilock is held by the caller. xfs_collapse_file_space()
now uses the fsb output as the starting point for the subsequent shift.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
XFS has been having trouble with stray delayed allocation extents
beyond EOF for a long time. Recent changes to the collapse range
code has triggered erroneous EBUSY errors on page invalidtion for
block size smaller than page size filesystems. These
have been caused by dirty buffers beyond EOF on a partial page which
do not get written to disk during a sync.
The issue is that write-ahead in xfs_cluster_write() finds such a
partial page and handles it by leaving the page dirty but pushing it
into a writeback state. This used to work just fine, as the
write_cache_pages() code would then find the dirty partial page in
the next mapping tree lookup as the dirty tag is still set.
Unfortunately, when we moved to a mark and sweep approach to
writeback to fix other writeback sync issues, we broken this. THe
act of marking the page as under writeback now clears the TOWRITE
tag in the radix tree, even though the page is still dirty. This
causes the TOWRITE tag to be cleared, and hence the next lookup on
the mapping tree does not find the dirty partial page and so doesn't
try to write it again.
This same writeback bug was found recently in ext4 and fixed in
commit 1c8349a ("ext4: fix data integrity sync in ordered mode")
without communication to the wider filesystem community. We can use
exactly the same fix here so the TOWRITE flag is not cleared on
partial page writes.
cc: stable@vger.kernel.org # dependent on 1c8349a171
Root-cause-found-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
rbpp is always passed into xfs_rtmodify_summary
and xfs_rtget_summary, so there is no need to
test for it in xfs_rtmodify_summary_int.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_rtmodify_summary and xfs_rtget_summary are almost identical;
fold them into xfs_rtmodify_summary_int(), with wrappers for each of
the original calls.
The _int function modifies if a delta is passed, and returns a
summary pointer if *sum is passed.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_dir_canenter and xfs_dir_createname are
almost identical.
Fold the former into the latter, with a helpful
wrapper for the former. If createname is called without
an inode number, it now only checks for space, and does
not actually add the entry.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Move the resblks test out of the xfs_dir_canenter,
and into the caller.
This makes a little more sense on the face of it;
xfs_dir_canenter immediately returns if resblks !=0;
and given some of the comments preceding the calls:
* Check for ability to enter directory entry, if no space reserved.
even more so.
It also facilitates the next patch.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
In xlog_do_recovery_pass(), there are 2 distinct cases:
non-wrapped and wrapped log recovery.
If we find a wrapped log, we recover around the end
of the log, and then handle the rest of recovery
exactly as in the non-wrapped case - using exactly the same
(duplicated) code.
Rather than having the same code in both cases, we can
get the wrapped portion out of the way first if needed,
and then recover the non-wrapped portion of the log.
There should be no functional change here, just code
reorganization & deduplication.
The patch looks a bit bigger than it really is; the last
hunk is whitespace changes (un-indenting).
Tested with xfstests "check -g log" on a stock configuration.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
For some reason, the older commit:
965c8e5 lseek: the "whence" argument is called "whence"
lseek: the "whence" argument is called "whence"
But the kernel decided to call it "origin" instead.
Fix most of the sites.
left out xfs. So fix xfs.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_seek_hole & xfs_seek_data are remarkably similar;
so much so that they can be combined, saving a fair
bit of semi-complex code duplication.
The following patch passes generic/285 and generic/286,
which specifically test seek behavior.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
XFS log recovery has been discovered to have race conditions with
buffers when I/O errors occur. External tools are available to simulate
I/O errors to XFS, but this alone is not sufficient for testing log
recovery. XFS unconditionally resets the inactive region of the log
prior to log recovery to avoid confusion over processing any partially
written log records that might have been written before an unclean
shutdown. Therefore, unconditional write I/O failures at mount time are
caught by the reset sequence rather than log recovery and hinder the
ability to test the latter.
The device-mapper dm-flakey module uses an up/down timer to define a
cycle for when to fail I/Os. Create a pre log recovery delay tunable
that can be used to coordinate XFS log recovery with I/O errors
simulated by dm-flakey. This facilitates coordination in userspace that
allows the reset of stale log blocks to succeed and writes due to log
recovery to fail. For example, define a dm-flakey instance with an
uptime long enough to allow log reset to succeed and a log recovery
delay long enough to allow the dm-flakey uptime to expire.
The 'log_recovery_delay' sysfs tunable is exported under
/sys/fs/xfs/debug and is only enabled for kernels compiled in XFS debug
mode. The value is exported in units of seconds and allows for a delay
of up to 60 seconds. Note that this is for XFS debug and test
instrumentation purposes only and should not be used by applications. No
delay is enabled by default.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Create a top-level debug directory for global debug sysfs attributes.
This directory is added and removed on XFS module initialization and
removal respectively for DEBUG mode kernels only. It typically resides
at /sys/fs/xfs/debug. It is located at the top level of the xfs sysfs
hierarchy as attributes might define global behavior or behavior that
must be configured before an xfs mount is available (e.g., log recovery
behavior).
Define the global debug kobject that represents the debug sysfs
directory and add generic attribute show/store helpers to support future
attributes. No debug attributes are exported as of yet.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
These were exposed by fsfuzzer runs; without them we fail
in various exciting and sometimes convoluted ways when we
encounter disk corruption.
Without the MAXLEVELS tests we tend to walk off the end of
an array in a loop like this:
for (i = 0; i < cur->bc_nlevels; i++) {
if (cur->bc_bufs[i])
Without the dirblklog test we try to allocate more memory
than we could possibly hope for and loop forever:
xfs_dabuf_map()
nfsb = mp->m_dir_geo->fsbcount;
irecs = kmem_zalloc(sizeof(irec) * nfsb, KM_SLEEP...
As for the logbsize check, that's the convoluted one.
If logbsize is specified at mount time, it's sanitized
in xfs_parseargs; in particular it makes sure that it's
not > XLOG_MAX_RECORD_BSIZE.
If not specified at mount time, it comes from the superblock
via sb_logsunit; this is limited to 256k at mkfs time as well;
it's copied into m_logbsize in xfs_finish_flags().
However, if for some reason the on-disk value is corrupt and
too large, nothing catches it. It's a circuitous path, but
that size eventually finds its way to places that make the kernel
very unhappy, leading to oopses in xlog_pack_data() because we
use the size as an index into iclog->ic_data, but the array
is not necessarily that big.
Anyway - bounds checking when we read from disk is a good thing!
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Workqueues must be explicitly set as freezable to ensure they are frozen
in the assocated part of the hibernation/suspend sequence. Freezing of
workqueues and kernel threads is important to ensure that modifications
are not made on-disk after the hibernation image has been created.
Otherwise, the in-memory state can become inconsistent with what is on
disk and eventually lead to filesystem corruption. We have reports of
free space btree corruptions that occur immediately after restore from
hibernate that suggest the xfs-eofblocks workqueue could be causing
such problems if it races with hibernation.
Mark all of the internal XFS workqueues as freezable to ensure nothing
changes on-disk once the freezer infrastructure freezes kernel threads
and creates the hibernation image.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reported-by: Carlos E. R. <carlos.e.r@opensuse.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
bdev_get_queue() returns the request_queue associated with the
specified block_device. blk_get_backing_dev_info() makes use of
bdev_get_queue() to determine the associated bdi given a block_device.
All the callers of bdev_get_queue() including
blk_get_backing_dev_info() assume that bdev_get_queue() may return
NULL and implement NULL handling; however, bdev_get_queue() requires
the passed in block_device is opened and attached to its gendisk.
Because an active gendisk always has a valid request_queue associated
with it, bdev_get_queue() can never return NULL and neither can
blk_get_backing_dev_info().
Make it clear that neither of the two functions can return NULL and
remove NULL handling from all the callers.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Chris Mason <clm@fb.com>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
xfs_collapse_file_space() currently writes back the entire file
undergoing collapse range to settle things down for the extent shift
algorithm. While this prevents changes to the extent list during the
collapse operation, the writeback itself is not enough to prevent
unnecessary collapse failures.
The current shift algorithm uses the extent index to iterate the in-core
extent list. If a post-eof delalloc extent persists after the writeback
(e.g., a prior zero range op where the end of the range aligns with eof
can separate the post-eof blocks such that they are not written back and
converted), xfs_bmap_shift_extents() becomes confused over the encoded
br_startblock value and fails the collapse.
As with the full writeback, this is a temporary fix until the algorithm
is improved to cope with a volatile extent list and avoid attempts to
shift post-eof extents.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
If we have delalloc extents on a file before we run a collapse range
opertaion, we sync the range that we are going to collapse to
convert delalloc extents in that region to real extents to simplify
the shift operation.
However, the shift operation then assumes that the extent list is
not going to change as it iterates over the extent list moving
things about. Unfortunately, this isn't true because we can't hold
the ILOCK over all the operations. We can prevent new IO from
modifying the extent list by holding the IOLOCK, but that doesn't
prevent writeback from running....
And when writeback runs, it can convert delalloc extents is the
range of the file prior to the region being collapsed, and this
changes the indexes of all the extents in the file. That causes the
collapse range operation to Go Bad.
The right fix is to rewrite the extent shift operation not to be
dependent on the extent list not changing across the entire
operation, but this is a fairly significant piece of work to do.
Hence, as a short-term workaround for the problem, sync the entire
file before starting a collapse operation to remove all delalloc
ranges from the file and so avoid the problem of concurrent
writeback changing the extent list.
Diagnosed-and-Reported-by: Brian Foster <bfoster@redhat.com>
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 file collapse mechanism uses xfs_bmap_shift_extents() to collapse
all subsequent extents down into the specified, previously punched out,
region. This function performs some validation, such as whether a
sufficient hole exists in the target region of the collapse, then shifts
the remaining exents downward.
The exit path of the function currently logs the inode unconditionally.
While we must log the inode (and abort) if an error occurs and the
transaction is dirty, the initial validation paths can generate errors
before the transaction has been dirtied. This creates an unnecessary
filesystem shutdown scenario, as the caller will cancel a transaction
that has been marked dirty.
Modify xfs_bmap_shift_extents() to OR the logflags bits as modifications
are made to the inode bmap. Only log the inode in the exit path if
logflags has been set. This ensures we only have to cancel a dirty
transaction if modifications have been made and prevents an unnecessary
filesystem shutdown otherwise.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Now we are not doing silly things with dirtying buffers beyond EOF
and using invalidation correctly, we can finally reduce the ranges of
writeback and invalidation used by direct IO to match that of the IO
being issued.
Bring the writeback and invalidation ranges back to match the
generic direct IO code - this will greatly reduce the perturbation
of cached data when direct IO and buffered IO are mixed, but still
provide the same buffered vs direct IO coherency behaviour we
currently have.
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>
Similar to direct IO reads, direct IO writes are using
truncate_pagecache_range to invalidate the page cache. This is
incorrect due to the sub-block zeroing in the page cache that
truncate_pagecache_range() triggers.
This patch fixes things by using invalidate_inode_pages2_range
instead. It preserves the page cache invalidation, but won't zero
any pages.
cc: stable@vger.kernel.org
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>
xfs is using truncate_pagecache_range to invalidate the page cache
during DIO reads. This is different from the other filesystems who
only invalidate pages during DIO writes.
truncate_pagecache_range is meant to be used when we are freeing the
underlying data structs from disk, so it will zero any partial
ranges in the page. This means a DIO read can zero out part of the
page cache page, and it is possible the page will stay in cache.
buffered reads will find an up to date page with zeros instead of
the data actually on disk.
This patch fixes things by using invalidate_inode_pages2_range
instead. It preserves the page cache invalidation, but won't zero
any pages.
[dchinner: catch error and warn if it fails. Comment.]
cc: stable@vger.kernel.org
Signed-off-by: Chris Mason <clm@fb.com>
Reviewed-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>
generic/263 is failing fsx at this point with a page spanning
EOF that cannot be invalidated. The operations are:
1190 mapwrite 0x52c00 thru 0x5e569 (0xb96a bytes)
1191 mapread 0x5c000 thru 0x5d636 (0x1637 bytes)
1192 write 0x5b600 thru 0x771ff (0x1bc00 bytes)
where 1190 extents EOF from 0x54000 to 0x5e569. When the direct IO
write attempts to invalidate the cached page over this range, it
fails with -EBUSY and so any attempt to do page invalidation fails.
The real question is this: Why can't that page be invalidated after
it has been written to disk and cleaned?
Well, there's data on the first two buffers in the page (1k block
size, 4k page), but the third buffer on the page (i.e. beyond EOF)
is failing drop_buffers because it's bh->b_state == 0x3, which is
BH_Uptodate | BH_Dirty. IOWs, there's dirty buffers beyond EOF. Say
what?
OK, set_buffer_dirty() is called on all buffers from
__set_page_buffers_dirty(), regardless of whether the buffer is
beyond EOF or not, which means that when we get to ->writepage,
we have buffers marked dirty beyond EOF that we need to clean.
So, we need to implement our own .set_page_dirty method that
doesn't dirty buffers beyond EOF.
This is messy because the buffer code is not meant to be shared
and it has interesting locking issues on the buffer dirty bits.
So just copy and paste it and then modify it to suit what we need.
Note: the solutions the other filesystems and generic block code use
of marking the buffers clean in ->writepage does not work for XFS.
It still leaves dirty buffers beyond EOF and invalidations still
fail. Hence rather than play whack-a-mole, this patch simply
prevents those buffers from being dirtied in the first place.
cc: <stable@kernel.org>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
We need to treat both inodes identically from a page cache point of
view when prepareing them for extent swapping. We don't do this
right now - we assume that one of the inodes empty, because that's
what xfs_fsr currently does. Remove this assumption from the code.
While factoring out the flushing and related checks, move the
transactions reservation to immeidately after the flushes so that we
don't need to pick up and then drop the ilock to do the transaction
reservation. There are no issues with aborting the transaction it if
the checks fail before we join the inodes to the transaction and
dirty them, so this is a safe change to make.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_swap_extents() holds the ilock over a call to
filemap_write_and_wait(), which can then try to write data and take
the ilock. That causes a self-deadlock.
Fix the deadlock and clean up the code by separating the locking
appropriately. Add a lockflags variable to track what locks we are
holding as we gain and drop them and cleanup the error handling to
always use "out_unlock" with the lockflags variable.
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 IO flag definitions to xfs_inode.h and kill the header file
as it is now empty.
Removing the xfs_vnode.h file showed up an implicit header include
path:
xfs_linux.h -> xfs_vnode.h -> xfs_fs.h
And so every xfs header file has been inplicitly been including
xfs_fs.h where it is needed or not. Hence the removal of xfs_vnode.h
causes all sorts of build issues because BBTOB() and friends are no
longer automatically included in the build. This also gets fixed.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Only one user, no longer needed.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Only has 2 users, has outlived it's usefulness.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Only one user of the macro and the dirty mapping check is redundant
so just get rid of it.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
dquot recovery should add verifiers to the dquot buffers that it
recovers changes into. Unfortunately, it doesn't attached the
verifiers to the buffers in a consistent manner. For example,
xlog_recover_dquot_pass2() reads dquot buffers without a verifier
and then writes it without ever having attached a verifier to the
buffer.
Further, dquot buffer recovery may write a dquot buffer that has not
been modified, or indeed, shoul dbe written because quotas are not
enabled and hence changes to the buffer were not replayed. In this
case, we again write buffers without verifiers attached because that
doesn't happen until after the buffer changes have been replayed.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
When running xfs/305, I noticed that quotacheck was flushing dquot
buffers that did not have the xfs_dquot_buf_ops verifiers attached:
XFS (vdb): _xfs_buf_ioapply: no ops on block 0x1dc8/0x1dc8
ffff880052489000: 44 51 01 04 00 00 65 b8 00 00 00 00 00 00 00 00 DQ....e.........
ffff880052489010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
ffff880052489020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
ffff880052489030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
CPU: 1 PID: 2376 Comm: mount Not tainted 3.16.0-rc2-dgc+ #306
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
ffff88006fe38000 ffff88004a0ffae8 ffffffff81cf1cca 0000000000000001
ffff88004a0ffb88 ffffffff814d50ca 000010004a0ffc70 0000000000000000
ffff88006be56dc4 0000000000000021 0000000000001dc8 ffff88007c773d80
Call Trace:
[<ffffffff81cf1cca>] dump_stack+0x45/0x56
[<ffffffff814d50ca>] _xfs_buf_ioapply+0x3ca/0x3d0
[<ffffffff810db520>] ? wake_up_state+0x20/0x20
[<ffffffff814d51f5>] ? xfs_bdstrat_cb+0x55/0xb0
[<ffffffff814d513b>] xfs_buf_iorequest+0x6b/0xd0
[<ffffffff814d51f5>] xfs_bdstrat_cb+0x55/0xb0
[<ffffffff814d53ab>] __xfs_buf_delwri_submit+0x15b/0x220
[<ffffffff814d6040>] ? xfs_buf_delwri_submit+0x30/0x90
[<ffffffff814d6040>] xfs_buf_delwri_submit+0x30/0x90
[<ffffffff8150f89d>] xfs_qm_quotacheck+0x17d/0x3c0
[<ffffffff81510591>] xfs_qm_mount_quotas+0x151/0x1e0
[<ffffffff814ed01c>] xfs_mountfs+0x56c/0x7d0
[<ffffffff814f0f12>] xfs_fs_fill_super+0x2c2/0x340
[<ffffffff811c9fe4>] mount_bdev+0x194/0x1d0
[<ffffffff814f0c50>] ? xfs_finish_flags+0x170/0x170
[<ffffffff814ef0f5>] xfs_fs_mount+0x15/0x20
[<ffffffff811ca8c9>] mount_fs+0x39/0x1b0
[<ffffffff811e4d67>] vfs_kern_mount+0x67/0x120
[<ffffffff811e757e>] do_mount+0x23e/0xad0
[<ffffffff8117abde>] ? __get_free_pages+0xe/0x50
[<ffffffff811e71e6>] ? copy_mount_options+0x36/0x150
[<ffffffff811e8103>] SyS_mount+0x83/0xc0
[<ffffffff81cfd40b>] tracesys+0xdd/0xe2
This was caused by dquot buffer readahead not attaching a verifier
structure to the buffer when readahead was issued, resulting in the
followup read of the buffer finding a valid buffer and so not
attaching new verifiers to the buffer as part of the read.
Also, when a verifier failure occurs, we then read the buffer
without verifiers. Attach the verifiers manually after this read so
that if the buffer is then written it will be verified that the
corruption has been repaired.
Further, when flushing a dquot we don't ask for a verifier when
reading in the dquot buffer the dquot belongs to. Most of the time
this isn't an issue because the buffer is still cached, but when it
is not cached it will result in writing the dquot buffer without
having the verfier attached.
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>
Crash testing of CRC enabled filesystems has resulted in a number of
reports of bad CRCs being detected after the filesystem was mounted.
Errors such as the following were being seen:
XFS (sdb3): Mounting V5 Filesystem
XFS (sdb3): Starting recovery (logdev: internal)
XFS (sdb3): Metadata CRC error detected at xfs_agf_read_verify+0x5a/0x100 [xfs], block 0x1
XFS (sdb3): Unmount and run xfs_repair
XFS (sdb3): First 64 bytes of corrupted metadata buffer:
ffff880136ffd600: 58 41 47 46 00 00 00 01 00 00 00 00 00 0f aa 40 XAGF...........@
ffff880136ffd610: 00 02 6d 53 00 02 77 f8 00 00 00 00 00 00 00 01 ..mS..w.........
ffff880136ffd620: 00 00 00 01 00 00 00 00 00 00 00 00 00 00 00 03 ................
ffff880136ffd630: 00 00 00 04 00 08 81 d0 00 08 81 a7 00 00 00 00 ................
XFS (sdb3): metadata I/O error: block 0x1 ("xfs_trans_read_buf_map") error 74 numblks 1
The errors were typically being seen in AGF, AGI and their related
btree block buffers some time after log recovery had run. Often it
wasn't until later subsequent mounts that the problem was
discovered. The common symptom was a buffer with the correct
contents, but a CRC and an LSN that matched an older version of the
contents.
Some debug added to _xfs_buf_ioapply() indicated that buffers were
being written without verifiers attached to them from log recovery,
and Jan Kara isolated the cause to log recovery readahead an dit's
interactions with buffers that had a more recent LSN on disk than
the transaction being recovered. In this case, the buffer did not
get a verifier attached, and os when the second phase of log
recovery ran and recovered EFIs and unlinked inodes, the buffers
were modified and written without the verifier running. Hence they
had up to date contents, but stale LSNs and CRCs.
Fix it by attaching verifiers to buffers we skip due to future LSN
values so they don't escape into the buffer cache without the
correct verifier attached.
This patch is based on analysis and a patch from Jan Kara.
cc: <stable@vger.kernel.org>
Reported-by: Jan Kara <jack@suse.cz>
Reported-by: Fanael Linithien <fanael4@gmail.com>
Reported-by: Grozdan <neutrino8@gmail.com>
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 recently had a bug where buffers were slipping through log
recovery without any verifier attached to them. This was resulting
in on-disk CRC mismatches for valid data. Add some warning code to
catch this occurrence so that we catch such bugs during development
rather than not being aware they exist.
Note that we cannot do this verification unconditionally as non-CRC
filesystems don't always attach verifiers to the buffers being
written. e.g. during log recovery we cannot identify all the
different types of buffers correctly on non-CRC filesystems, so we
can't attach the correct verifiers in all cases and so we don't
attach any. Hence we don't want on non-CRC filesystems to avoid
spamming the logs with false indications.
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 commit
83e782e xfs: Remove incore use of XFS_OQUOTA_ENFD and XFS_OQUOTA_CHKD
added a new function xfs_sb_quota_from_disk() which swaps
on-disk XFS_OQUOTA_* flags for in-core XFS_GQUOTA_* and XFS_PQUOTA_*
flags after the superblock is read.
However, if log recovery is required, the superblock is read again,
and the modified in-core flags are re-read from disk, so we have
XFS_OQUOTA_* flags in memory again. This causes the
XFS_QM_NEED_QUOTACHECK() test to be true, because the XFS_OQUOTA_CHKD
is still set, and not XFS_GQUOTA_CHKD or XFS_PQUOTA_CHKD.
Change xfs_sb_from_disk to call xfs_sb_quota_from disk and always
convert the disk flags to in-memory flags.
Add a lower-level function which can be called with "false" to
not convert the flags, so that the sb verifier can verify
exactly what was on disk, per Brian Foster's suggestion.
Reported-by: Cyril B. <cbay@excellency.fr>
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
The offset and length parameters are converted from bytes to basic
blocks by xfs_vn_fiemap(). The BTOBB() converter rounds the value up to
the nearest basic block. This leads to unexpected behavior when
unaligned offsets are provided to FIEMAP.
Fix the conversions of byte values to block values to cover the provided
offsets. Round down the start offset to the nearest basic block.
Calculate the end offset based on the provided values, round up and
calculate length based on the start block offset.
Reported-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Introduce xfs_bulkstat_ag_ichunk() to process inodes in chunk with a
pointer to a formatter function that will iget the inode and fill in
the appropriate structure.
Refactor xfs_bulkstat() with it.
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Trying to support tiny disks only and saving a bit memory might have
made sense on an SGI O2 15 years ago, but is pretty pointless today.
Remove the rarely tested codepath that uses various smaller in-memory
types to reduce our test matrix and make the codebase a little bit
smaller and less complicated.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ben Myers <bpm@sgi.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
We are intended to check up uflags against FS_PROJ_QUOTA rather than
FS_USER_UQUOTA once more, it looks to me like a typo, but might cause
the project quota metadata space can not be removed.
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Remove the XFS_IS_OQUOTA_ON macros as it is obsoleted.
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_set_inode32() caught my eye because it had weird spacing around
the "-1's". In cleaning that up, I realized that the assignment in
the declaration of "ino" is never used; it's rewritten before it
gets read.
Drop the ino initializer from its declaration since it's not used,
and move the agino initialization into the body of the function,
mostly so that we can have pretty whitespace and not exceed 80
columns. :)
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Today, if we perform an xfs_growfs which adds allocation groups,
mp->m_maxagi is not properly updated when the growfs is complete.
Therefore inodes will continue to be allocated only in the
AGs which existed prior to the growfs, and the new space
won't be utilized.
This is because of this path in xfs_growfs_data_private():
xfs_growfs_data_private
xfs_initialize_perag(mp, nagcount, &nagimax);
if (mp->m_flags & XFS_MOUNT_32BITINODES)
index = xfs_set_inode32(mp);
else
index = xfs_set_inode64(mp);
if (maxagi)
*maxagi = index;
where xfs_set_inode* iterates over the (old) agcount in
mp->m_sb.sb_agblocks, which has not yet been updated
in the growfs path. So "index" will be returned based on
the old agcount, not the new one, and new AGs are not available
for inode allocation.
Fix this by explicitly passing the proper AG count (which
xfs_initialize_perag() already has) down another level,
so that xfs_set_inode* can make the proper decision about
acceptable AGs for inode allocation in the potentially
newly-added AGs.
This has been broken since 3.7, when these two
xfs_set_inode* functions were added in commit 2d2194f.
Prior to that, we looped over "agcount" not sb_agblocks
in these calculations.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_qm_quotacheck() is not used outside of xfs_qm.c. Mark it static
and move it around in the file to avoid a forward declaration.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
When the CIL checkpoint is fully written to the log, the LSN of the checkpoint
commit record is written into the CIL context structure. This allows log force
waiters to correctly detect when the checkpoint they are waiting on have been
fully written into the log buffers.
However, the initial context after mount is initialised with a non-zero commit
LSN, so appears to waiters as though it is complete even though it may not have
even been pushed, let alone written to the log buffers. Hence a log force
immediately after a filesystem is mounted may not behave correctly, nor does
commit record ordering if multiple CIL pushes interleave immediately after
mount.
To fix this, make sure the initial context commit LSN is not touched until the
first checkpointis actually pushed.
[dchinner: rewrite commit message]
Signed-off-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Brian Foster <bfoster@redhat.com>
Commit 4d559a3b introduced heavy prealloc. squashing to catch the case
of requesting too large a prealloc on smaller filesystems, leading to
repeated flush and retry cycles that occur on ENOSPC. Now that we issue
eofblocks scans on EDQUOT/ENOSPC, squash the prealloc against the
minimum available free space across all applicable quotas as well to
avoid a similar problem of repeated eofblocks scans.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Brian Foster <bfoster@redhat.com>
Speculative preallocation and and the associated throttling metrics
assume we're working with large files on large filesystems. Users have
reported inefficiencies in these mechanisms when we happen to be dealing
with large files on smaller filesystems. This can occur because while
prealloc throttling is aggressive under low free space conditions, it is
not active until we reach 5% free space or less.
For example, a 40GB filesystem has enough space for several files large
enough to have multi-GB preallocations at any given time. If those files
are slow growing, they might reserve preallocation for long periods of
time as well as avoid the background scanner due to frequent
modification. If a new file is written under these conditions, said file
has no access to this already reserved space and premature ENOSPC is
imminent.
To handle this scenario, modify the buffered write ENOSPC handling and
retry sequence to invoke an eofblocks scan. In the smaller filesystem
scenario, the eofblocks scan resets the usage of preallocation such that
when the 5% free space threshold is met, throttling effectively takes
over to provide fair and efficient preallocation until legitimate
ENOSPC.
The eofblocks scan is selective based on the nature of the failure. For
example, an EDQUOT failure in a particular quota will use a filtered
scan for that quota. Because we don't know which quota might have caused
an allocation failure at any given time, we include each applicable
quota determined to be under low free space conditions in the scan.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Brian Foster <bfoster@redhat.com>
The eofblocks scan inode filter uses intersection logic by default.
E.g., specifying both user and group quota ids filters out inodes that
are not covered by both the specified user and group quotas. This is
suitable for behavior exposed to userspace.
Scans that are initiated from within the kernel might require more broad
semantics, such as scanning all inodes under each quota associated with
an inode to alleviate low free space conditions in each.
Create the XFS_EOF_FLAGS_UNION flag to support a conditional union-based
filtering algorithm for eofblocks scans. This flag is intentionally left
out of the valid mask as it is not supported for scans initiated from
userspace.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Brian Foster <bfoster@redhat.com>
The scan owner field represents an optional inode number that is
responsible for the current scan. The purpose is to identify that an
inode is under iolock and as such, the iolock shouldn't be attempted
when trimming eofblocks. This is an internal only field.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Jie Liu <jeff.liu@oracle.com>
Introduce xfs_bulkstat_grab_ichunk() to look up an inode chunk in where
the given inode resides, then grab the record. Update the data for the
pointed-to record if the inode was not the last in the chunk and there
are some left allocated, return the grabbed inode count on success.
Refactor xfs_bulkstat() with it.
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Jie Liu <jeff.liu@oracle.com>
Introduce xfs_bulkstat_ichunk_ra() to loop over all clusters in the
next inode chunk, then performs readahead if there are any allocated
inodes in that cluster.
Refactor xfs_bulkstat() with it.
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Jie Liu <jeff.liu@oracle.com>
We should not ignore the btree operation errors at xfs_bulkstat() but
to propagate them if any. This patch fix two places in this function
and the remaining things will be fixed with code refactoring thereafter.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Jie Liu <jeff.liu@oracle.com>
Remove the redundant user buffer and count checks as it has already
been validated at xfs_ioc_bulkstat().
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Jie Liu <jeff.liu@oracle.com>
To fetch the file system number tables, we currently just ignore the
errors and proceed to loop over the next AG or bump agino to the next
chunk in case of btree operations failed, that is not properly because
those errors might hint us potential file system problems.
This patch rework xfs_inumbers() to handle the btree operation errors
as well as the loop conditions.
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Jie Liu <jeff.liu@oracle.com>
Consolidate xfs_inumbers() to make the formatter function return correct
error and make the source code looks a bit neat.
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Christoph Hellwig <hch@lst.de>
xfs_bukstat_one doesn't have any failure case that would go away when
called through xfs_bulkstat, so remove the fallback and the now unessecary
xfs_bulkstat_single function.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
From: Jie Liu <jeff.liu@oracle.com>
Remove the redundant BULKSTAT_RV_NOTHING assignment in case of call
xfs_iget() failed at xfs_bulkstat_one_int().
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Create log attributes to export the current runtime state of the log to
sysfs. Note that the filesystem should be frozen for consistency across
attributes.
The following per-mount attributes are created: log_head_lsn,
log_tail_lsn, reserve_grant_head and write_grant_head. These represent
the physical log head, tail and reserve and write grant heads
respectively. Attribute values are exported in the following format:
"cycle:[block,byte]"
... where cycle represents the log cycle and [block,bytes] represents
either the basic block or byte offset of the log, depending on the
attribute. Log sequence number (LSN) values are encoded in basic blocks
and grant heads are encoded in bytes. All values are in decimal format.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Embed a kobject into the xfs log data structure (xlog). This creates a
'log' subdirectory for every XFS mount instance in sysfs. The lifecycle
of the log kobject is tied to the lifecycle of the log.
Also define a set of generic attribute handlers associated with the log
kobject in preparation for the addition of attributes.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Embed a base kobject into xfs_mount. This creates a kobject associated
with each XFS mount and a subdirectory in sysfs with the name of the
filesystem. The subdirectory lifecycle matches that of the mount. Also
add the new xfs_sysfs.[c,h] source files with some XFS sysfs
infrastructure to facilitate attribute creation.
Note that there are currently no attributes exported as part of the
xfs_mount kobject. It exists solely to serve as a per-mount container
for child objects.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Create a sysfs kset to contain all sub-objects associated with the XFS
module. The kset is created and removed on module initialization and
removal respectively. The kset uses fs_obj as a parent. This leads to
the creation of a /sys/fs/xfs directory when the kset exists.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_mountfs() has a couple failure conditions that do not jump to the
correct labels. Specifically:
- xfs_initialize_perag_data() failure does not deallocate the log even
though it occurs after log initialization
- xfs_mount_reset_sbqflags() failure returns the error directly rather
than jump to the error sequence
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
When quota is on, it is expected that unused quota inodes have a
value of NULLFSINO. The changes to support a separate project quota
in 3.12 broken this rule for non-project quota inode enabled
filesystem, as the code now refuses to write the group quota inode
if neither group or project quotas are enabled. This regression was
introduced by commit d892d58 ("xfs: Start using pquotaino from the
superblock").
In this case, we should be writing NULLFSINO rather than nothing to
ensure that we leave the group quota inode in a valid state while
quotas are enabled.
Failure to do so doesn't cause a current kernel to break - the
separate project quota inodes introduced translation code to always
treat a zero inode as NULLFSINO. This was introduced by commit
0102629 ("xfs: Initialize all quota inodes to be NULLFSINO") with is
also in 3.12 but older kernels do not do this and hence taking a
filesystem back to an older kernel can result in quotas failing
initialisation at mount time. When that happens, we see this in
dmesg:
[ 1649.215390] XFS (sdb): Mounting Filesystem
[ 1649.316894] XFS (sdb): Failed to initialize disk quotas.
[ 1649.316902] XFS (sdb): Ending clean mount
By ensuring that we write NULLFSINO to quota inodes that aren't
active, we avoid this problem. We have to be really careful when
determining if the quota inodes are active or not, because we don't
want to write a NULLFSINO if the quota inodes are active and we
simply aren't updating them.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The allocation stack switch at xfs_bmapi_allocate() has served it's
purpose, but is no longer a sufficient solution to the stack usage
problem we have in the XFS allocation path.
Whilst the kernel stack size is now 16k, that is not a valid reason
for undoing all our "keep stack usage down" modifications. What it
does allow us to do is have the freedom to refine and perfect the
modifications knowing that if we get it wrong it won't blow up in
our faces - we have a safety net now.
This is important because we still have the issue of older kernels
having smaller stacks and that they are still supported and are
demonstrating a wide range of different stack overflows. Red Hat
has several open bugs for allocation based stack overflows from
directory modifications and direct IO block allocation and these
problems still need to be solved. If we can solve them upstream,
then distro's won't need to bake their own unique solutions.
To that end, I've observed that every allocation based stack
overflow report has had a specific characteristic - it has happened
during or directly after a bmap btree block split. That event
requires a new block to be allocated to the tree, and so we
effectively stack one allocation stack on top of another, and that's
when we get into trouble.
A further observation is that bmap btree block splits are much rarer
than writeback allocation - over a range of different workloads I've
observed the ratio of bmap btree inserts to splits ranges from 100:1
(xfstests run) to 10000:1 (local VM image server with sparse files
that range in the hundreds of thousands to millions of extents).
Either way, bmap btree split events are much, much rarer than
allocation events.
Finally, we have to move the kswapd state to the allocation workqueue
work when allocation is done on behalf of kswapd. This is proving to
cause significant perturbation in performance under memory pressure
and appears to be generating allocation deadlock warnings under some
workloads, so avoiding the use of a workqueue for the majority of
kswapd writeback allocation will minimise the impact of such
behaviour.
Hence it makes sense to move the stack switch to xfs_btree_split()
and only do it for bmap btree splits. Stack switches during
allocation will be much rarer, so there won't be significant
performacne overhead caused by switching stacks. The worse case
stack from all allocation paths will be split, not just writeback.
And the majority of memory allocations will be done in the correct
context (e.g. kswapd) without causing additional latency, and so we
simplify the memory reclaim interactions between processes,
workqueues and kswapd.
The worst stack I've been able to generate with this patch in place
is 5600 bytes deep. It's very revealing because we exit XFS at:
37) 1768 64 kmem_cache_alloc+0x13b/0x170
about 1800 bytes of stack consumed, and the remaining 3800 bytes
(and 36 functions) is memory reclaim, swap and the IO stack. And
this occurs in the inode allocation from an open(O_CREAT) syscall,
not writeback.
The amount of stack being used is much less than I've previously be
able to generate - fs_mark testing has been able to generate stack
usage of around 7k without too much trouble; with this patch it's
only just getting to 5.5k. This is primarily because the metadata
allocation paths (e.g. directory blocks) are no longer causing
double splits on the same stack, and hence now stack tracing is
showing swapping being the worst stack consumer rather than XFS.
Performance of fs_mark inode create workloads is unchanged.
Performance of fs_mark async fsync workloads is consistently good
with context switches reduced by around 150,000/s (30%).
Performance of dbench, streaming IO and postmark is unchanged.
Allocation deadlock warnings have not been seen on the workloads
that generated them since adding this patch.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
This reverts commit 1f6d64829d.
This commit resulted in regressions in performance in low
memory situations where kswapd was doing writeback of delayed
allocation blocks. It resulted in significant parallelism of the
kswapd work and with the special kswapd flags meant that hundreds of
active allocation could dip into kswapd specific memory reserves and
avoid being throttled. This cause a large amount of performance
variation, as well as random OOM-killer invocations that didn't
previously exist.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Convert all the errors the core XFs code to negative error signs
like the rest of the kernel and remove all the sign conversion we
do in the interface layers.
Errors for conversion (and comparison) found via searches like:
$ git grep " E" fs/xfs
$ git grep "return E" fs/xfs
$ git grep " E[A-Z].*;$" fs/xfs
Negation points found via searches like:
$ git grep "= -[a-z,A-Z]" fs/xfs
$ git grep "return -[a-z,A-D,F-Z]" fs/xfs
$ git grep " -[a-z].*;" fs/xfs
[ with some bits I missed from Brian Foster ]
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Move all the source files that are shared with userspace into
libxfs/. This is done as one big chunk simpy to get it done
quickly
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Move all the header files that are shared with userspace into
libxfs. This is done as one big chunk simpy to get it done quickly.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
To minimise the differences between kernel and userspace code,
split the kernel code into the same structure as the userspace code.
That is, the gneric core functionality of XFS is moved to a libxfs/
directory and treat it as a layering barrier in the XFS code.
This patch introduces the libxfs directory, the build infrastructure
and an initial source and header file to build. The libxfs directory
will contain the header files that are needed to build libxfs - most
of userspace does not care about the location of these header files
as they are accessed indirectly. Hence keeping them inside libxfs
makes it easy to track the changes and script the sync process as
the directory structure will be identical.
To allow this changeover to occur in the kernel code, there are some
temporary infrastructure in the makefiles to grab the header
filesystem from both locations. Once all the files are moved,
modifications will be made in the source code that will make the
need for these include directives go away.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
XFS_ERROR was designed long ago to trap return values, but it's not
runtime configurable, it's not consistently used, and we can do
similar error trapping with ftrace scripts and triggers from
userspace.
Just nuke XFS_ERROR and associated bits.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
return is not a function. "return(EIO);" is silly;
"return (EIO);" moreso. return is not a function.
Nuke the pointless parens.
[dchinner: catch a couple of extra cases in xfs_attr_list.c,
xfs_acl.c and xfs_linux.h.]
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Pull vfs updates from Al Viro:
"This the bunch that sat in -next + lock_parent() fix. This is the
minimal set; there's more pending stuff.
In particular, I really hope to get acct.c fixes merged this cycle -
we need that to deal sanely with delayed-mntput stuff. In the next
pile, hopefully - that series is fairly short and localized
(kernel/acct.c, fs/super.c and fs/namespace.c). In this pile: more
iov_iter work. Most of prereqs for ->splice_write with sane locking
order are there and Kent's dio rewrite would also fit nicely on top of
this pile"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (70 commits)
lock_parent: don't step on stale ->d_parent of all-but-freed one
kill generic_file_splice_write()
ceph: switch to iter_file_splice_write()
shmem: switch to iter_file_splice_write()
nfs: switch to iter_splice_write_file()
fs/splice.c: remove unneeded exports
ocfs2: switch to iter_file_splice_write()
->splice_write() via ->write_iter()
bio_vec-backed iov_iter
optimize copy_page_{to,from}_iter()
bury generic_file_aio_{read,write}
lustre: get rid of messing with iovecs
ceph: switch to ->write_iter()
ceph_sync_direct_write: stop poking into iov_iter guts
ceph_sync_read: stop poking into iov_iter guts
new helper: copy_page_from_iter()
fuse: switch to ->write_iter()
btrfs: switch to ->write_iter()
ocfs2: switch to ->write_iter()
xfs: switch to ->write_iter()
...
iter_file_splice_write() - a ->splice_write() instance that gathers the
pipe buffers, builds a bio_vec-based iov_iter covering those and feeds
it to ->write_iter(). A bunch of simple cases coverted to that...
[AV: fixed the braino spotted by Cyrill]
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
This update contains:
o cleanup removing unused function args
o rework of the filestreams allocator to use dentry cache parent lookups
o new on-disk free inode btree and optimised inode allocator
o various bug fixes
o rework of internal attribute API
o cleanup of superblock feature bit support to remove historic cruft
o more fixes and minor cleanups
o added a new directory/attribute geometry abstraction
o yet more fixes and minor cleanups.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)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=qnz2
-----END PGP SIGNATURE-----
Merge tag 'xfs-for-linus-3.16-rc1' of git://oss.sgi.com/xfs/xfs
Pull xfs updates from Dave Chinner:
"This update contains:
- cleanup removing unused function args
- rework of the filestreams allocator to use dentry cache parent
lookups
- new on-disk free inode btree and optimised inode allocator
- various bug fixes
- rework of internal attribute API
- cleanup of superblock feature bit support to remove historic cruft
- more fixes and minor cleanups
- added a new directory/attribute geometry abstraction
- yet more fixes and minor cleanups"
* tag 'xfs-for-linus-3.16-rc1' of git://oss.sgi.com/xfs/xfs: (86 commits)
xfs: fix xfs_da_args sparse warning in xfs_readdir
xfs: Fix rounding in xfs_alloc_fix_len()
xfs: tone down writepage/releasepage WARN_ONs
xfs: small cleanup in xfs_lowbit64()
xfs: kill xfs_buf_geterror()
xfs: xfs_readsb needs to check for magic numbers
xfs: block allocation work needs to be kswapd aware
xfs: remove redundant geometry information from xfs_da_state
xfs: replace attr LBSIZE with xfs_da_geometry
xfs: pass xfs_da_args to xfs_attr_leaf_newentsize
xfs: use xfs_da_geometry for block size in attr code
xfs: remove mp->m_dir_geo from directory logging
xfs: reduce direct usage of mp->m_dir_geo
xfs: move node entry counts to xfs_da_geometry
xfs: convert dir/attr btree threshold to xfs_da_geometry
xfs: convert m_dirblksize to xfs_da_geometry
xfs: convert m_dirblkfsbs to xfs_da_geometry
xfs: convert directory segment limits to xfs_da_geometry
xfs: convert directory db conversion to xfs_da_geometry
xfs: convert directory dablk conversion to xfs_da_geometry
...
The kernel has no concept of capabilities with respect to inodes; inodes
exist independently of namespaces. For example, inode_capable(inode,
CAP_LINUX_IMMUTABLE) would be nonsense.
This patch changes inode_capable to check for uid and gid mappings and
renames it to capable_wrt_inode_uidgid, which should make it more
obvious what it does.
Fixes CVE-2014-4014.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Serge Hallyn <serge.hallyn@ubuntu.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: stable@vger.kernel.org
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The kbuild test robot reported:
>> fs/xfs/xfs_dir2_readdir.c:672:41: sparse: Using plain integer as NULL pointer
Fix it.
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Rounding in xfs_alloc_fix_len() is wrong. As the comment states, the
result should be a number of a form (k*prod+mod) however due to sign
mistake the result is different. As a result allocations on raid arrays
could be misaligned in some cases.
This also seems to fix occasional assertion failure:
XFS_WANT_CORRUPTED_GOTO(rlen <= flen, error0)
in xfs_alloc_ag_vextent_size().
Also add an assertion that the result of xfs_alloc_fix_len() is of
expected form.
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
I recently ran into the issue fixed by
"xfs: kill buffers over failed write ranges properly"
which spams the log with lots of backtraces. Make debugging any
issues like that easier by using WARN_ON_ONCE in the writeback code.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
There are two checkpatch.pl complaints here because of the bad
indenting and because of the assignment inside the condition.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Most of the callers are just calling ASSERT(!xfs_buf_geterror())
which means they are checking for bp->b_error == 0. If bp is null in
this case, we will assert fail, and hence it's no different in
result to oopsing because of a null bp. In some cases, errors have
already been checked for or the function returning the buffer can't
return a buffer with an error, so it's just a redundant assert.
Either way, the assert can either be removed.
The other two non-assert callers can just test for a buffer and
error properly.
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>
Commit daba542 ("xfs: skip verification on initial "guess"
superblock read") dropped the use of a verifier for the initial
superblock read so we can probe the sector size of the filesystem
stored in the superblock. It, however, now fails to validate that
what was read initially is actually an XFS superblock and hence will
fail the sector size check and return ENOSYS.
This causes probe-based mounts to fail because it expects XFS to
return EINVAL when it doesn't recognise the superblock format.
cc: <stable@vger.kernel.org>
Reported-by: Plamen Petrov <plamen.sisi@gmail.com>
Tested-by: Plamen Petrov <plamen.sisi@gmail.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Upon memory pressure, kswapd calls xfs_vm_writepage() from
shrink_page_list(). This can result in delayed allocation occurring
and that gets deferred to the the allocation workqueue.
The allocation then runs outside kswapd context, which means if it
needs memory (and it does to demand page metadata from disk) it can
block in shrink_inactive_list() waiting for IO congestion. These
blocking waits are normally avoiding in kswapd context, so under
memory pressure writeback from kswapd can be arbitrarily delayed by
memory reclaim.
To avoid this, pass the kswapd context to the allocation being done
by the workqueue, so that memory reclaim understands correctly that
the work is being done for kswapd and therefore it is not blocked
and does not delay memory reclaim.
To avoid issues with int->char conversion of flag fields (as noticed
in v1 of this patch) convert the flag fields in the struct
xfs_bmalloca to bool types. pahole indicates these variables are
still single byte variables, so no extra space is consumed by this
change.
cc: <stable@vger.kernel.org>
Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
It's carried in state->args->geo, so there's no need to duplicate it
and use more stack space than necessary.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
As it's only ever called from contexts where the xfs_da_args is
present and contains all the information needed inside the args
structure.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Rather than using the superblock value obtained through the
xfs_mount.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
We don't pass the xfs_da_args or the geometry all the way down to
the directory buffer logging code, hence we have to use
mp->m_dir_geo here. Fix this to use the geometry passed via the
xfs_da_args, and convert all the directory logging functions for
consistency.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
There are many places in the directory code were we don't pass the
args into and so have to extract the geometry direct from the mount
structure. Push the args or the geometry into these leaf functions
so that we don't need to grab it from the struct xfs_mount.
This, in turn, brings use to the point where directory geometry is
no longer a property of the struct xfs_mount; it is not a global
property anymore, and hence we can start to consider per-directory
configuration of physical geometries.
Start by converting the xfs_dir_isblock/leaf code - pass in the
xfs_da_args and convert the readdir code to use xfs_da_args like
the rest of the directory code to pass information around.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
