Use a rhashtable to cache the unlinked list incore.  This should speed
up unlinked processing considerably when there are a lot of inodes on
the unlinked list because iunlink_remove no longer has to traverse an
entire bucket list to find which inode points to the one being removed.

The incore list structure records "X.next_unlinked = Y" relations, with
the rhashtable using Y to index the records.  This makes finding the
inode X that points to a inode Y very quick.  If our cache fails to find
anything we can always fall back on the old method.

FWIW this drastically reduces the amount of time it takes to remove
inodes from the unlinked list.  I wrote a program to open a lot of
O_TMPFILE files and then close them in the same order, which takes
a very long time if we have to traverse the unlinked lists.  With the
ptach, I see:

+ /d/t/tmpfile/tmpfile
Opened 193531 files in 6.33s.
Closed 193531 files in 5.86s

real    0m12.192s
user    0m0.064s
sys     0m11.619s
+ cd /
+ umount /mnt

real    0m0.050s
user    0m0.004s
sys     0m0.030s

And without the patch:

+ /d/t/tmpfile/tmpfile
Opened 193588 files in 6.35s.
Closed 193588 files in 751.61s

real    12m38.853s
user    0m0.084s
sys     12m34.470s
+ cd /
+ umount /mnt

real    0m0.086s
user    0m0.000s
sys     0m0.060s
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

The realtime summary is a two-dimensional array on disk, effectively:

u32 rsum[log2(number of realtime extents) + 1][number of blocks in the bitmap]

rsum[log][bbno] is the number of extents of size 2**log which start in
bitmap block bbno.

xfs_rtallocate_extent_near() uses xfs_rtany_summary() to check whether
rsum[log][bbno] != 0 for any log level. However, the summary array is
stored in row-major order (i.e., like an array in C), so all of these
entries are not adjacent, but rather spread across the entire summary
file. In the worst case (a full bitmap block), xfs_rtany_summary() has
to check every level.

This means that on a moderately-used realtime device, an allocation will
waste a lot of time finding, reading, and releasing buffers for the
realtime summary. In particular, one of our storage services (which runs
on servers with 8 very slow CPUs and 15 8 TB XFS realtime filesystems)
spends almost 5% of its CPU cycles in xfs_rtbuf_get() and
xfs_trans_brelse() called from xfs_rtany_summary().

One solution would be to also store the summary with the dimensions
swapped. However, this would require a disk format change to a very old
component of XFS.

Instead, we can cache the minimum size which contains any extents. We do
so lazily; rather than guaranteeing that the cache contains the precise
minimum, it always contains a loose lower bound which we tighten when we
read or update a summary block. This only uses a few kilobytes of memory
and is already serialized via the realtime bitmap and summary inode
locks, so the cost is minimal. With this change, the same workload only
spends 0.2% of its CPU cycles in the realtime allocator.
Signed-off-by: NOmar Sandoval <osandov@fb.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Store the inode cluster alignment information in units of inodes and
blocks in the mount data so that we don't have to keep recalculating
them.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

Store the number of inodes and blocks per inode cluster in the mount
data so that we don't have to keep recalculating them.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>

The barrier mount options have been no-ops and deprecated since

4cf4573d xfs: deprecate barrier/nobarrier mount option

i.e. kernel 4.10 / December 2016, with a stated deprecation schedule
after v4.15.  Should be fair game to remove them now.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NCarlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Use the "bad summary count" mount flag from the previous patch to skip
writing the unmount record to force log recovery at the next mount,
which will recalculate the summary counters for us.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

Filippo Giunchedi complained that xfs doesn't even perform basic sanity
checks of the fs summary counters at mount time.  Therefore, recalculate
the summary counters from the AGFs after log recovery if the counts were
bad (or we had to recover the fs).  Enhance the recalculation routine to
fail the mount entirely if the new values are also obviously incorrect.

We use a mount state flag to record the "bad summary count" state so
that the (subsequent) online fsck patches can detect subtlely incorrect
counts and set the flag; clear it userspace asks for a repair; or force
a recalculation at the next mount if nobody fixes it by unmount time.
Reported-by: NFilippo Giunchedi <fgiunchedi@wikimedia.org>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

Get rid of the MIN/MAX macros and just use the native min/max macros
directly in the XFS code.
Signed-Off-By: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Remove the verbose license text from XFS files and replace them
with SPDX tags. This does not change the license of any of the code,
merely refers to the common, up-to-date license files in LICENSES/

This change was mostly scripted. fs/xfs/Makefile and
fs/xfs/libxfs/xfs_fs.h were modified by hand, the rest were detected
and modified by the following command:

for f in `git grep -l "GNU General" fs/xfs/` ; do
	echo $f
	cat $f | awk -f hdr.awk > $f.new
	mv -f $f.new $f
done

And the hdr.awk script that did the modification (including
detecting the difference between GPL-2.0 and GPL-2.0+ licenses)
is as follows:

$ cat hdr.awk
BEGIN {
	hdr = 1.0
	tag = "GPL-2.0"
	str = ""
}

/^ \* This program is free software/ {
	hdr = 2.0;
	next
}

/any later version./ {
	tag = "GPL-2.0+"
	next
}

/^ \*\// {
	if (hdr > 0.0) {
		print "// SPDX-License-Identifier: " tag
		print str
		print $0
		str=""
		hdr = 0.0
		next
	}
	print $0
	next
}

/^ \* / {
	if (hdr > 1.0)
		next
	if (hdr > 0.0) {
		if (str != "")
			str = str "\n"
		str = str $0
		next
	}
	print $0
	next
}

/^ \*/ {
	if (hdr > 0.0)
		next
	print $0
	next
}

// {
	if (hdr > 0.0) {
		if (str != "")
			str = str "\n"
		str = str $0
		next
	}
	print $0
}

END { }
$
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The struct xfs_agfl v5 header was originally introduced with
unexpected padding that caused the AGFL to operate with one less
slot than intended. The header has since been packed, but the fix
left an incompatibility for users who upgrade from an old kernel
with the unpacked header to a newer kernel with the packed header
while the AGFL happens to wrap around the end. The newer kernel
recognizes one extra slot at the physical end of the AGFL that the
previous kernel did not. The new kernel will eventually attempt to
allocate a block from that slot, which contains invalid data, and
cause a crash.

This condition can be detected by comparing the active range of the
AGFL to the count. While this detects a padding mismatch, it can
also trigger false positives for unrelated flcount corruption. Since
we cannot distinguish a size mismatch due to padding from unrelated
corruption, we can't trust the AGFL enough to simply repopulate the
empty slot.

Instead, avoid unnecessarily complex detection logic and and use a
solution that can handle any form of flcount corruption that slips
through read verifiers: distrust the entire AGFL and reset it to an
empty state. Any valid blocks within the AGFL are intentionally
leaked. This requires xfs_repair to rectify (which was already
necessary based on the state the AGFL was found in). The reset
mitigates the side effect of the padding mismatch problem from a
filesystem crash to a free space accounting inconsistency. The
generic approach also means that this patch can be safely backported
to kernels with or without a packed struct xfs_agfl.

Check the AGF for an invalid freelist count on initial read from
disk. If detected, set a flag on the xfs_perag to indicate that a
reset is required before the AGFL can be used. In the first
transaction that attempts to use a flagged AGFL, reset it to empty,
warn the user about the inconsistency and allow the freelist fixup
code to repopulate the AGFL with new blocks. The xfs_perag flag is
cleared to eliminate the need for repeated checks on each block
allocation operation.

This allows kernels that include the packing fix commit 96f859d5
("libxfs: pack the agfl header structure so XFS_AGFL_SIZE is correct")
to handle older unpacked AGFL formats without a filesystem crash.
Suggested-by: NDave Chinner <david@fromorbit.com>
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by Dave Chiluk <chiluk+linuxxfs@indeed.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The rmapbt perag metadata reservation reserves blocks for the
reverse mapping btree (rmapbt). Since the rmapbt uses blocks from
the agfl and perag accounting is updated as blocks are allocated
from the allocation btrees, the reservation actually accounts blocks
as they are allocated to (or freed from) the agfl rather than the
rmapbt itself.

While this works for blocks that are eventually used for the rmapbt,
not all agfl blocks are destined for the rmapbt. Blocks that are
allocated to the agfl (and thus "reserved" for the rmapbt) but then
used by another structure leads to a growing inconsistency over time
between the runtime tracking of rmapbt usage vs. actual rmapbt
usage. Since the runtime tracking thinks all agfl blocks are rmapbt
blocks, it essentially believes that less future reservation is
required to satisfy the rmapbt than what is actually necessary.

The inconsistency is rectified across mount cycles because the perag
reservation is initialized based on the actual rmapbt usage at mount
time. The problem, however, is that the excessive drain of the
reservation at runtime opens a window to allocate blocks for other
purposes that might be required for the rmapbt on a subsequent
mount. This problem can be demonstrated by a simple test that runs
an allocation workload to consume agfl blocks over time and then
observe the difference in the agfl reservation requirement across an
unmount/mount cycle:

  mount ...: xfs_ag_resv_init: ... resv 3193 ask 3194 len 3194
  ...
  ...      : xfs_ag_resv_alloc_extent: ... resv 2957 ask 3194 len 1
  umount...: xfs_ag_resv_free: ... resv 2956 ask 3194 len 0
  mount ...: xfs_ag_resv_init: ... resv 3052 ask 3194 len 3194

As the above tracepoints show, the reservation requirement reduces
from 3194 blocks to 2956 blocks as the workload runs.  Without any
other changes in the filesystem, the same reservation requirement
jumps from 2956 to 3052 blocks over a umount/mount cycle.

To address this divergence, update the RMAPBT reservation to account
blocks used for the rmapbt only rather than all blocks filled into
the agfl. This patch makes several high-level changes toward that
end:

1.) Reintroduce an AGFL reservation type to serve as an accounting
    no-op for blocks allocated to (or freed from) the AGFL.
2.) Invoke RMAPBT usage accounting from the actual rmapbt block
    allocation path rather than the AGFL allocation path.

The first change is required because agfl blocks are considered free
blocks throughout their lifetime. The perag reservation subsystem is
invoked unconditionally by the allocation subsystem, so we need a
way to tell the perag subsystem (via the allocation subsystem) to
not make any accounting changes for blocks filled into the AGFL.

The second change causes the in-core RMAPBT reservation usage
accounting to remain consistent with the on-disk state at all times
and eliminates the risk of leaving the rmapbt reservation
underfilled.
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The AGFL perag reservation type accounts all allocations that feed
into (or are released from) the allocation group free list (agfl).
The purpose of the reservation is to support worst case conditions
for the reverse mapping btree (rmapbt). As such, the agfl
reservation usage accounting only considers rmapbt usage when the
in-core counters are initialized at mount time.

This implementation inconsistency leads to divergence of the in-core
and on-disk usage accounting over time. In preparation to resolve
this inconsistency and adjust the AGFL reservation into an rmapbt
specific reservation, rename the AGFL reservation type and
associated accounting fields to something more rmapbt-specific. Also
fix up a couple tracepoints that incorrectly use the AGFL
reservation type to pass the agfl state of the associated extent
where the raw reservation type is expected.

Note that this patch does not change perag reservation behavior.
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The dmevmask structure member is a dmapi leftover; it's
set here and there but never actually used.  Remove it.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NBill O'Donnell <billodo@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

We now have enhanced error injection that can control the frequency
with which errors happen, so convert drop_writes to use this.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NCarlos Maiolino <cmaiolino@redhat.com>

Creates a /sys/fs/xfs/$dev/errortag/ directory to control the errortag
values directly.  This enables us to control the randomness values,
rather than having to accept the defaults.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NCarlos Maiolino <cmaiolino@redhat.com>

Remove the xfs_etest structure in favor of a per-mountpoint structure.
This will give us the flexibility to set as many error injection points
as we want, and later enable us to set up sysfs knobs to set the trigger
frequency as we wish.  This comes at a cost of higher memory use, but
unti we hit 1024 injection points (we're at 29) or a lot of mounts this
shouldn't be a huge issue.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NCarlos Maiolino <cmaiolino@redhat.com>

This is a purely mechanical patch that removes the private
__{u,}int{8,16,32,64}_t typedefs in favor of using the system
{u,}int{8,16,32,64}_t typedefs.  This is the sed script used to perform
the transformation and fix the resulting whitespace and indentation
errors:

s/typedef\t__uint8_t/typedef __uint8_t\t/g
s/typedef\t__uint/typedef __uint/g
s/typedef\t__int\([0-9]*\)_t/typedef int\1_t\t/g
s/__uint8_t\t/__uint8_t\t\t/g
s/__uint/uint/g
s/__int\([0-9]*\)_t\t/__int\1_t\t\t/g
s/__int/int/g
/^typedef.*int[0-9]*_t;$/d
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

On some architectures do_div does the pointer compare
trick to make sure that we've sent it an unsigned 64-bit
number.  (Why unsigned?  I don't know.)

Fix up the few places that squawk about this; in
xfs_bmap_wants_extents() we just used a bare int64_t so change
that to unsigned.

In xfs_adjust_extent_unmap_boundaries() all we wanted was the
mod, and we have an xfs-specific function to handle that w/o
side effects, which includes proper casting for do_div.

In xfs_daddr_to_ag[b]no, we were using the wrong type anyway;
XFS_BB_TO_FSBT returns a block in the filesystem, so use
xfs_rfsblock_t not xfs_daddr_t, and gain the unsignedness
from that type as a bonus.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

The log covering background task used to be part of the xfssyncd
workqueue. That workqueue was removed as of commit 5889608d ("xfs:
syncd workqueue is no more") and the associated work item scheduled
to the xfs-log wq. The latter is used for log buffer I/O completion.

Since xfs_log_worker() can invoke a log flush, a deadlock is
possible between the xfs-log and xfs-cil workqueues. Consider the
following codepath from xfs_log_worker():

xfs_log_worker()
  xfs_log_force()
    _xfs_log_force()
      xlog_cil_force()
        xlog_cil_force_lsn()
          xlog_cil_push_now()
            flush_work()

The above is in xfs-log wq context and blocked waiting on the
completion of an xfs-cil work item. Concurrently, the cil push in
progress can end up blocked here:

xlog_cil_push_work()
  xlog_cil_push()
    xlog_write()
      xlog_state_get_iclog_space()
        xlog_wait(&log->l_flush_wait, ...)

The above is in xfs-cil context waiting on log buffer I/O
completion, which executes in xfs-log wq context. In this scenario
both workqueues are deadlocked waiting on eachother.

Add a new workqueue specifically for the high level log covering and
ail pushing worker, as was the case prior to commit 5889608d.
Diagnosed-by: NDavid Jeffery <djeffery@redhat.com>
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

A debug mode write failure mechanism was introduced to XFS in commit
801cc4e1 ("xfs: debug mode forced buffered write failure") to
facilitate targeted testing of delalloc indirect reservation management
from userspace. This code was subsequently rendered ineffective by the
move to iomap based buffered writes in commit 68a9f5e7 ("xfs:
implement iomap based buffered write path"). This likely went unnoticed
because the associated userspace code had not made it into xfstests.

Resurrect this mechanism to facilitate effective indlen reservation
testing from xfstests. The move to iomap based buffered writes relocated
the hook this mechanism needs to return write failure from XFS to
generic code. The failure trigger must remain in XFS. Given that
limitation, convert this from a write failure mechanism to one that
simply drops writes without returning failure to userspace. Rename all
"fail_writes" references to "drop_writes" to illustrate the point. This
is more hacky than preferred, but still triggers the XFS error handling
behavior required to drive the indlen tests. This is only available in
DEBUG mode and for testing purposes only.
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Currently we force the log and simply try again if we hit a busy extent,
but especially with online discard enabled it might take a while after
the log force for the busy extents to disappear, and we might have
already completed our second pass.

So instead we add a new waitqueue and a generation counter to the pag
structure so that we can do wakeups once we've removed busy extents,
and we replace the single retry with an unconditional one - after
all we hold the AGF buffer lock, so no other allocations or frees
can be racing with us in this AG.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

Currently we try to rely on the global reserved block pool for block
allocations for the free inode btree, but I have customer reports
(fairly complex workload, need to find an easier reproducer) where that
is not enough as the AG where we free an inode that requires a new
finobt block is entirely full.  This causes us to cancel a dirty
transaction and thus a file system shutdown.

I think the right way to guard against this is to treat the finot the same
way as the refcount btree and have a per-AG reservations for the possible
worst case size of it, and the patch below implements that.

Note that this could increase mount times with large finobt trees.  In
an ideal world we would have added a field for the number of finobt
fields to the AGI, similar to what we did for the refcount blocks.
We should do add it next time we rev the AGI or AGF format by adding
new fields.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>

On filesystems with a lot of metadata and in metadata intensive workloads
xfs_buf_find() is showing up at the top of the CPU cycles trace. Most of
the CPU time is spent on CPU cache misses while traversing the rbtree.

As the buffer cache does not need any kind of ordering, but fast lookups
a hashtable is the natural data structure to use. The rhashtable
infrastructure provides a self-scaling hashtable implementation and
allows lookups to proceed while the table is going through a resize
operation.

This reduces the CPU-time spent for the lookups to 1/3 even for small
filesystems with a relatively small number of cached buffers, with
possibly much larger gains on higher loaded filesystems.

[dchinner: reduce minimum hash size to an acceptable size for large
	   filesystems with many AGs with no active use.]
[dchinner: remove stale rbtree asserts.]
[dchinner: use xfs_buf_map for compare function argument.]
[dchinner: make functions static.]
[dchinner: remove redundant comments.]
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NDave Chinner <david@fromorbit.com>

Trim CoW reservations made on behalf of a cowextsz hint if they get too
old or we run low on quota, so long as we don't have dirty data awaiting
writeback or directio operations in progress.

Garbage collection of the cowextsize extents are kept separate from
prealloc extent reaping because setting the CoW prealloc lifetime to a
(much) higher value than the regular prealloc extent lifetime has been
useful for combatting CoW fragmentation on VM hosts where the VMs
experience bursty write behaviors and we can keep the utilization ratios
low enough that we don't start to run out of space.  IOWs, it benefits
us to keep the CoW fork reservations around for as long as we can unless
we run out of blocks or hit inode reclaim.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

Start constructing the refcount btree implementation by establishing
the on-disk format and everything needed to read, write, and
manipulate the refcount btree blocks.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

Add new per-AG refcount btree definitions to the per-AG structures.
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NChristoph Hellwig <hch@lst.de>

One unfortunate quirk of the reference count and reverse mapping
btrees -- they can expand in size when blocks are written to *other*
allocation groups if, say, one large extent becomes a lot of tiny
extents.  Since we don't want to start throwing errors in the middle
of CoWing, we need to reserve some blocks to handle future expansion.
The transaction block reservation counters aren't sufficient here
because we have to have a reserve of blocks in every AG, not just
somewhere in the filesystem.

Therefore, create two per-AG block reservation pools.  One feeds the
AGFL so that rmapbt expansion always succeeds, and the other feeds all
other metadata so that refcountbt expansion never fails.

Use the count of how many reserved blocks we need to have on hand to
create a virtual reservation in the AG.  Through selective clamping of
the maximum length of allocation requests and of the length of the
longest free extent, we can make it look like there's less free space
in the AG unless the reservation owner is asking for blocks.

In other words, play some accounting tricks in-core to make sure that
we always have blocks available.  On the plus side, there's nothing to
clean up if we crash, which is contrast to the strategy that the rough
draft used (actually removing extents from the freespace btrees).
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

As it stands today, the "fail immediately" vs. "retry forever"
values for max_retries and retry_timeout_seconds in the xfs metadata
error configurations are not consistent.

A retry_timeout_seconds of 0 means "retry forever," but a
max_retries of 0 means "fail immediately."

retry_timeout_seconds < 0 is disallowed, while max_retries == -1
means "retry forever."

Make this consistent across the error configs, such that a value of
0 means "fail immediately" (i.e. wait 0 seconds, or retry 0 times),
and a value of -1 always means "retry forever."

This makes retry_timeout a signed long to accommodate the -1, even
though it stores jiffies.  Given our limit of a 1 day maximum
timeout, this should be sufficient even at much higher HZ values
than we have available today.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NDave 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: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

Originally-From: Dave Chinner <dchinner@redhat.com>

Now we have all the surrounding call infrastructure in place, we can
start filling out the rmap btree implementation. Start with the
on-disk btree format; add everything needed to read, write and
manipulate rmap btree blocks. This prepares the way for adding the
btree operations implementation.

[darrick: record owner and offset info in rmap btree]
[darrick: fork, bmbt and unwritten state in rmap btree]
[darrick: flags are a separate field in xfs_rmap_irec]
[darrick: calculate maxlevels separately]
[darrick: move the 'unwritten' bit into unused parts of rm_offset]
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NDarrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

Originally-From: Dave Chinner <dchinner@redhat.com>

XFS reserves a small amount of space in each AG for the minimum
number of free blocks needed for operation. Adding the rmap btree
increases the number of reserved blocks, but it also increases the
complexity of the calculation as the free inode btree is optional
(like the rmbt).

Rather than calculate the prealloc blocks every time we need to
check it, add a function to calculate it at mount time and store it
in the struct xfs_mount, and convert the XFS_PREALLOC_BLOCKS macro
just to use the xfs-mount variable directly.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

If we take "retry forever" literally on metadata IO errors, we can
hang at unmount, once it retries those writes forever. This is the
default behavior, unfortunately.

Add an error configuration option for this behavior and default it
to "fail" so that an unmount will trigger actuall errors, a shutdown
and allow the unmount to succeed. It will be noisy, though, as it
will log the errors and shutdown that occurs.

To fix this, we need to mark the filesystem as being in the process
of unmounting. Do this with a mount flag that is added at the
appropriate time (i.e. before the blocking AIL sync). We also need
to add this flag if mount fails after the initial phase of log
recovery has been run.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NCarlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

now most of the infrastructure is in place, we can start adding
support for configuring specific errors such as ENODEV, ENOSPC, EIO,
etc. Add these error configurations and configure them all to have
appropriate behaviours. That is, all will be configured to retry
forever by default, except for ENODEV, which is an unrecoverable
error, so it will be configured to not retry on error
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NCarlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

On reception of an error, we can fail immediately, perform some
bound amount of retries or retry indefinitely. The current behaviour
we have is to retry forever.

However, we'd like the ability to choose how long the filesystem
should try after an error, it can either fail immediately, retry a
few times, or retry forever. This is implemented by using
max_retries sysfs attribute, to hold the amount of times we allow
the filesystem to retry after an error. Being -1 a special case
where the filesystem will retry indefinitely.

Add both a maximum retry count and a retry timeout so that we can
bound by time and/or physical IO attempts.

Finally, plumb these into xfs_buf_iodone error processing so that
the error behaviour follows the selected configuration.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NCarlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

With the error configuration handle for async metadata write errors
in place, we can now add initial support to the IO error processing
in xfs_buf_iodone_error().

Add an infrastructure function to look up the configuration handle,
and rearrange the error handling to prepare the way for different
error handling conigurations to be used.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NCarlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

Now we have the basic infrastructure, add the first error class so
we can build up the infrastructure in a meaningful way. Add the
metadata async write IO error class and sysfs entry, and introduce a
default configuration that matches the existing "retry forever"
behavior for async write metadata buffers.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NCarlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

We need to be able to change the way XFS behaviours in error
conditions depending on the type of underlying storage. This is
necessary for handling non-traditional block devices with extended
error cases, such as thin provisioned devices that can return ENOSPC
as an IO error.

Introduce the basic sysfs infrastructure needed to define and
configure error behaviours. This is done to be generic enough to
extend to configuring behaviour in other error conditions, such as
ENOMEM, which also has different desired behaviours according to
machine configuration.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NCarlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>

PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.

This promise never materialized.  And unlikely will.

We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE.  And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.

Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.

Let's stop pretending that pages in page cache are special.  They are
not.

The changes are pretty straight-forward:

 - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;

 - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;

 - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};

 - page_cache_get() -> get_page();

 - page_cache_release() -> put_page();

This patch contains automated changes generated with coccinelle using
script below.  For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.

The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.

There are few places in the code where coccinelle didn't reach.  I'll
fix them manually in a separate patch.  Comments and documentation also
will be addressed with the separate patch.

virtual patch

@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E

@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E

@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT

@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE

@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK

@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)

@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)

@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: NKirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: NMichal Hocko <mhocko@suse.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add a DEBUG mode-only sysfs knob to enable forced buffered write
failure. An additional side effect of this mode is brute force killing
of delayed allocation blocks in the range of the write. The latter is
the prime motiviation behind this patch, as userspace test
infrastructure requires a reliable mechanism to create and split
delalloc extents without causing extent conversion.

Certain fallocate operations (i.e., zero range) were used for this in
the past, but the implementations have changed such that delalloc
extents are flushed and converted to real blocks, rendering the test
useless.
Signed-off-by: NBrian Foster <bfoster@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NDave Chinner <david@fromorbit.com>

inode32/inode64 allocator behavior with respect to mount, remount
and growfs is a little tricky.

The inode32 mount option should only enable the inode32 allocator
heuristics if the filesystem is large enough for 64-bit inodes to
exist.  Today, it has this behavior on the initial mount, but a
remount with inode32 unconditionally changes the allocation
heuristics, even for a small fs.

Also, an inode32 mounted small filesystem should transition to the
inode32 allocator if the filesystem is subsequently grown to a
sufficient size.  Today that does not happen.

This patch consolidates xfs_set_inode32 and xfs_set_inode64 into a
single new function, and moves the "is the maximum inode number big
enough to matter" test into that function, so it doesn't rely on the
caller to get it right - which remount did not do, previously.
Signed-off-by: NEric Sandeen <sandeen@redhat.com>
Reviewed-by: NBrian Foster <bfoster@redhat.com>
Signed-off-by: NDave Chinner <david@fromorbit.com>