05 Nov, 2020
2 commits
-
In commit 7588cbeec6df, we tried to fix a race stemming from the lack of
coordination between higher level code that wants to allocate and remap
CoW fork extents into the data fork. Christoph cites as examples the
always_cow mode, and a directio write completion racing with writeback.According to the comments before the goto retry, we want to restart the
lookup to catch the extent in the data fork, but we don't actually reset
whichfork or cow_fsb, which means the second try executes using stale
information. Up until now I think we've gotten lucky that either
there's something left in the CoW fork to cause cow_fsb to be reset, or
either data/cow fork sequence numbers have advanced enough to force a
fresh lookup from the data fork. However, if we reach the retry with an
empty stable CoW fork and a stable data fork, neither of those things
happens. The retry foolishly re-calls xfs_convert_blocks on the CoW
fork which fails again. This time, we toss the write.I've recently been working on extending reflink to the realtime device.
When the realtime extent size is larger than a single block, we have to
force the page cache to CoW the entire rt extent if a write (or
fallocate) are not aligned with the rt extent size. The strategy I've
chosen to deal with this is derived from Dave's blocksize > pagesize
series: dirtying around the write range, and ensuring that writeback
always starts mapping on an rt extent boundary. This has brought this
race front and center, since generic/522 blows up immediately.However, I'm pretty sure this is a bug outright, independent of that.
Fixes: 7588cbeec6df ("xfs: retry COW fork delalloc conversion when no extent was found")
Signed-off-by: Darrick J. Wong
Reviewed-by: Christoph Hellwig -
iomap writeback mapping failure only calls into ->discard_page() if
the current page has not been added to the ioend. Accordingly, the
XFS callback assumes a full page discard and invalidation. This is
problematic for sub-page block size filesystems where some portion
of a page might have been mapped successfully before a failure to
map a delalloc block occurs. ->discard_page() is not called in that
error scenario and the bio is explicitly failed by iomap via the
error return from ->prepare_ioend(). As a result, the filesystem
leaks delalloc blocks and corrupts the filesystem block counters.Since XFS is the only user of ->discard_page(), tweak the semantics
to invoke the callback unconditionally on mapping errors and provide
the file offset that failed to map. Update xfs_discard_page() to
discard the corresponding portion of the file and pass the range
along to iomap_invalidatepage(). The latter already properly handles
both full and sub-page scenarios by not changing any iomap or page
state on sub-page invalidations.Signed-off-by: Brian Foster
Reviewed-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
21 Sep, 2020
1 commit
-
This helper is useful for both THPs and for supporting block size larger
than page size. Convert all users that I could find (we have a few
different ways of writing this idiom, and I may have missed some).Signed-off-by: Matthew Wilcox (Oracle)
Reviewed-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
Acked-by: Dave Kleikamp
03 Jun, 2020
2 commits
-
Pull xfs updates from Darrick Wong:
"Most of the changes this cycle are refactoring of existing code in
preparation for things landing in the future.We also fixed various problems and deficiencies in the quota
implementation, and (I hope) the last of the stale read vectors by
forcing write allocations to go through the unwritten state until the
write completes.Summary:
- Various cleanups to remove dead code, unnecessary conditionals,
asserts, etc.- Fix a linker warning caused by xfs stuffing '-g' into CFLAGS
redundantly.- Tighten up our dmesg logging to ensure that everything is prefixed
with 'XFS' for easier grepping.- Kill a bunch of typedefs.
- Refactor the deferred ops code to reduce indirect function calls.
- Increase type-safety with the deferred ops code.
- Make the DAX mount options a tri-state.
- Fix some error handling problems in the inode flush code and clean
up other inode flush warts.- Refactor log recovery so that each log item recovery functions now
live with the other log item processing code.- Fix some SPDX forms.
- Fix quota counter corruption if the fs crashes after running
quotacheck but before any dquots get logged.- Don't fail metadata verification on zero-entry attr leaf blocks,
since they're just part of the disk format now due to a historic
lack of log atomicity.- Don't allow SWAPEXT between files with different [ugp]id when
quotas are enabled.- Refactor inode fork reading and verification to run directly from
the inode-from-disk function. This means that we now actually
guarantee that _iget'ted inodes are totally verified and ready to
go.- Move the incore inode fork format and extent counts to the ifork
structure.- Scalability improvements by reducing cacheline pingponging in
struct xfs_mount.- More scalability improvements by removing m_active_trans from the
hot path.- Fix inode counter update sanity checking to run /only/ on debug
kernels.- Fix longstanding inconsistency in what error code we return when a
program hits project quota limits (ENOSPC).- Fix group quota returning the wrong error code when a program hits
group quota limits.- Fix per-type quota limits and grace periods for group and project
quotas so that they actually work.- Allow extension of individual grace periods.
- Refactor the non-reclaim inode radix tree walking code to remove a
bunch of stupid little functions and straighten out the
inconsistent naming schemes.- Fix a bug in speculative preallocation where we measured a new
allocation based on the last extent mapping in the file instead of
looking farther for the last contiguous space allocation.- Force delalloc writes to unwritten extents. This closes a stale
disk contents exposure vector if the system goes down before the
write completes.- More lockdep whackamole"
* tag 'xfs-5.8-merge-8' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux: (129 commits)
xfs: more lockdep whackamole with kmem_alloc*
xfs: force writes to delalloc regions to unwritten
xfs: refactor xfs_iomap_prealloc_size
xfs: measure all contiguous previous extents for prealloc size
xfs: don't fail unwritten extent conversion on writeback due to edquot
xfs: rearrange xfs_inode_walk_ag parameters
xfs: straighten out all the naming around incore inode tree walks
xfs: move xfs_inode_ag_iterator to be closer to the perag walking code
xfs: use bool for done in xfs_inode_ag_walk
xfs: fix inode ag walk predicate function return values
xfs: refactor eofb matching into a single helper
xfs: remove __xfs_icache_free_eofblocks
xfs: remove flags argument from xfs_inode_ag_walk
xfs: remove xfs_inode_ag_iterator_flags
xfs: remove unused xfs_inode_ag_iterator function
xfs: replace open-coded XFS_ICI_NO_TAG
xfs: move eofblocks conversion function to xfs_ioctl.c
xfs: allow individual quota grace period extension
xfs: per-type quota timers and warn limits
xfs: switch xfs_get_defquota to take explicit type
... -
Use the new readahead operation in iomap. Convert XFS and ZoneFS to use
it.Signed-off-by: Matthew Wilcox (Oracle)
Signed-off-by: Andrew Morton
Reviewed-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Reviewed-by: William Kucharski
Cc: Chao Yu
Cc: Cong Wang
Cc: Dave Chinner
Cc: Eric Biggers
Cc: Gao Xiang
Cc: Jaegeuk Kim
Cc: John Hubbard
Cc: Joseph Qi
Cc: Junxiao Bi
Cc: Michal Hocko
Cc: Zi Yan
Cc: Johannes Thumshirn
Cc: Miklos Szeredi
Link: http://lkml.kernel.org/r/20200414150233.24495-26-willy@infradead.org
Signed-off-by: Linus Torvalds
20 May, 2020
1 commit
-
Both the data and attr fork have a format that is stored in the legacy
idinode. Move it into the xfs_ifork structure instead, where it uses
up padding.Signed-off-by: Christoph Hellwig
Reviewed-by: Brian Foster
Reviewed-by: Chandan Babu R
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
03 Mar, 2020
1 commit
-
Use printk_ratelimit() to limit the amount of messages printed from
xfs_discard_page. Without that a failing device causes a large
number of errors that doesn't really help debugging the underling
issue.Signed-off-by: Christoph Hellwig
Reviewed-by: Brian Foster
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
04 Jan, 2020
1 commit
-
As of now dax_writeback_mapping_range() takes "struct block_device" as a
parameter and dax_dev is searched from bdev name. This also involves taking
a fresh reference on dax_dev and putting that reference at the end of
function.We are developing a new filesystem virtio-fs and using dax to access host
page cache directly. But there is no block device. IOW, we want to make
use of dax but want to get rid of this assumption that there is always
a block device associated with dax_dev.So pass in "struct dax_device" as parameter instead of bdev.
ext2/ext4/xfs are current users and they already have a reference on
dax_device. So there is no need to take reference and drop reference to
dax_device on each call of this function.Suggested-by: Christoph Hellwig
Reviewed-by: Christoph Hellwig
Reviewed-by: Jan Kara
Signed-off-by: Vivek Goyal
Link: https://lore.kernel.org/r/20200103183307.GB13350@redhat.com
Signed-off-by: Dan Williams
28 Oct, 2019
1 commit
-
Add a new xfs_inode_buftarg helper that gets the data I/O buftarg for a
given inode. Replace the existing xfs_find_bdev_for_inode and
xfs_find_daxdev_for_inode helpers with this new general one and cleanup
some of the callers.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
22 Oct, 2019
1 commit
-
Start untangling xfs_file_iomap_begin by splitting out the read-only
case into its own set of iomap_ops with a very simply iomap_begin
helper.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
21 Oct, 2019
6 commits
-
Take the xfs writeback code and move it to fs/iomap. A new structure
with three methods is added as the abstraction from the generic writeback
code to the file system. These methods are used to map blocks, submit an
ioend, and cancel a page that encountered an error before it was added to
an ioend.Signed-off-by: Christoph Hellwig
[darrick: rename ->submit_ioend to ->prepare_ioend to clarify what it
does]
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
Reviewed-by: Dave Chinner -
Lift the xfs code for tracing address space operations to the iomap
layer.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
In preparation for moving the writeback code to iomap.c, replace the
XFS-specific COW fork concept with the iomap IOMAP_F_SHARED flag.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
In preparation for moving the ioend structure to common code we need
to get rid of the xfs-specific xfs_trans type. Just make it a file
system private void pointer instead.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
Introduce two nicely abstracted helper, which can be moved to the iomap
code later. Also use list_first_entry_or_null to simplify the code a
bit.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
In preparation for moving the XFS writeback code to fs/iomap.c, switch
it to use struct iomap instead of the XFS-specific struct xfs_bmbt_irec.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Reviewed-by: Carlos Maiolino
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
16 Jul, 2019
1 commit
-
Pull more block updates from Jens Axboe:
"A later pull request with some followup items. I had some vacation
coming up to the merge window, so certain things items were delayed a
bit. This pull request also contains fixes that came in within the
last few days of the merge window, which I didn't want to push right
before sending you a pull request.This contains:
- NVMe pull request, mostly fixes, but also a few minor items on the
feature side that were timing constrained (Christoph et al)- Report zones fixes (Damien)
- Removal of dead code (Damien)
- Turn on cgroup psi memstall (Josef)
- block cgroup MAINTAINERS entry (Konstantin)
- Flush init fix (Josef)
- blk-throttle low iops timing fix (Konstantin)
- nbd resize fixes (Mike)
- nbd 0 blocksize crash fix (Xiubo)
- block integrity error leak fix (Wenwen)
- blk-cgroup writeback and priority inheritance fixes (Tejun)"
* tag 'for-linus-20190715' of git://git.kernel.dk/linux-block: (42 commits)
MAINTAINERS: add entry for block io cgroup
null_blk: fixup ->report_zones() for !CONFIG_BLK_DEV_ZONED
block: Limit zone array allocation size
sd_zbc: Fix report zones buffer allocation
block: Kill gfp_t argument of blkdev_report_zones()
block: Allow mapping of vmalloc-ed buffers
block/bio-integrity: fix a memory leak bug
nvme: fix NULL deref for fabrics options
nbd: add netlink reconfigure resize support
nbd: fix crash when the blksize is zero
block: Disable write plugging for zoned block devices
block: Fix elevator name declaration
block: Remove unused definitions
nvme: fix regression upon hot device removal and insertion
blk-throttle: fix zero wait time for iops throttled group
block: Fix potential overflow in blk_report_zones()
blkcg: implement REQ_CGROUP_PUNT
blkcg, writeback: Implement wbc_blkcg_css()
blkcg, writeback: Add wbc->no_cgroup_owner
blkcg, writeback: Rename wbc_account_io() to wbc_account_cgroup_owner()
...
13 Jul, 2019
1 commit
-
Pull xfs updates from Darrick Wong:
"In this release there are a significant amounts of consolidations and
cleanups in the log code; restructuring of the log to issue struct
bios directly; new bulkstat ioctls to return v5 fs inode information
(and fix all the padding problems of the old ioctl); the beginnings of
multithreaded inode walks (e.g. quotacheck); and a reduction in memory
usage in the online scrub code leading to reduced runtimes.- Refactor inode geometry calculation into a single structure instead
of open-coding pieces everywhere.- Add online repair to build options.
- Remove unnecessary function call flags and functions.
- Claim maintainership of various loose xfs documentation and header
files.- Use struct bio directly for log buffer IOs instead of struct
xfs_buf.- Reduce log item boilerplate code requirements.
- Merge log item code spread across too many files.
- Further distinguish between log item commits and cancellations.
- Various small cleanups to the ag small allocator.
- Support cgroup-aware writeback
- libxfs refactoring for mkfs cleanup
- Remove unneeded #includes
- Fix a memory allocation miscalculation in the new log bio code
- Fix bisection problems
- Fix a crash in ioend processing caused by tripping over freeing of
preallocated transactions- Split out a generic inode walk mechanism from the bulkstat code,
hook up all the internal users to use the walking code, then clean
up bulkstat to serve only the bulkstat ioctls.- Add a multithreaded iwalk implementation to speed up quotacheck on
fast storage with many CPUs.- Remove unnecessary return values in logging teardown functions.
- Supplement the bstat and inogrp structures with new bulkstat and
inumbers structures that have all the fields we need for v5
filesystem features and none of the padding problems of their
predecessors.- Wire up new ioctls that use the new structures with a much simpler
bulk_ireq structure at the head instead of the pointerhappy mess we
had before.- Enable userspace to constrain bulkstat returns to a single AG or a
single special inode so that we can phase out a lot of geometry
guesswork in userspace.- Reduce memory consumption and zeroing overhead in extended
attribute scrub code.- Fix some behavioral regressions in the new bulkstat backend code.
- Fix some behavioral regressions in the new log bio code"
* tag 'xfs-5.3-merge-12' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux: (100 commits)
xfs: chain bios the right way around in xfs_rw_bdev
xfs: bump INUMBERS cursor correctly in xfs_inumbers_walk
xfs: don't update lastino for FSBULKSTAT_SINGLE
xfs: online scrub needn't bother zeroing its temporary buffer
xfs: only allocate memory for scrubbing attributes when we need it
xfs: refactor attr scrub memory allocation function
xfs: refactor extended attribute buffer pointer functions
xfs: attribute scrub should use seen_enough to pass error values
xfs: allow single bulkstat of special inodes
xfs: specify AG in bulk req
xfs: wire up the v5 inumbers ioctl
xfs: wire up new v5 bulkstat ioctls
xfs: introduce v5 inode group structure
xfs: introduce new v5 bulkstat structure
xfs: rename bulkstat functions
xfs: remove various bulk request typedef usage
fs: xfs: xfs_log: Change return type from int to void
xfs: poll waiting for quotacheck
xfs: multithreaded iwalk implementation
xfs: refactor INUMBERS to use iwalk functions
...
01 Jul, 2019
5 commits
-
'bio->bi_iter.bi_size' is 'unsigned int', which at most hold 4G - 1
bytes.Before 07173c3ec276 ("block: enable multipage bvecs"), one bio can
include very limited pages, and usually at most 256, so the fs bio
size won't be bigger than 1M bytes most of times.Since we support multi-page bvec, in theory one fs bio really can
be added > 1M pages, especially in case of hugepage, or big writeback
with too many dirty pages. Then there is chance in which .bi_size
is overflowed.Fixes this issue by using bio_full() to check if the added segment may
overflow .bi_size.Cc: Liu Yiding
Cc: kernel test robot
Cc: "Darrick J. Wong"
Cc: linux-xfs@vger.kernel.org
Cc: linux-fsdevel@vger.kernel.org
Cc: stable@vger.kernel.org
Fixes: 07173c3ec276 ("block: enable multipage bvecs")
Reviewed-by: Christoph Hellwig
Signed-off-by: Ming Lei
Signed-off-by: Jens Axboe -
Instead of a magic flag for xfs_trans_alloc, just ensure all callers
that can't relclaim through the file system use memalloc_nofs_save to
set the per-task nofs flag.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
Compare the block layer status directly instead of converting it to
an errno first.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
There is no real problem merging ioends that go beyond i_size into an
ioend that doesn't. We just need to move the append transaction to the
base ioend. Also use the opportunity to use a real error code instead
of the magic 1 to cancel the transactions, and write a comment
explaining the scheme.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
The fail argument is long gone, update the comment.
Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
29 Jun, 2019
3 commits
-
There are many, many xfs header files which are included but
unneeded (or included twice) in the xfs code, so remove them.nb: xfs_linux.h includes about 9 headers for everyone, so those
explicit includes get removed by this. I'm not sure what the
preference is, but if we wanted explicit includes everywhere,
a followup patch could remove those xfs_*.h includes from
xfs_linux.h and move them into the files that need them.
Or it could be left as-is.Signed-off-by: Eric Sandeen
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
Link every newly allocated writeback bio to cgroup pointed to by the
writeback control structure, and charge every byte written back to it.Tested-by: Stefan Priebe - Profihost AG
Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
Move setting up operation and write hint to xfs_alloc_ioend, and
then just copy over all needed information from the previous bio
in xfs_chain_bio and stop passing various parameters to it.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
17 Jun, 2019
1 commit
-
We currently have an input same_page parameter to __bio_try_merge_page
to prohibit merging in the same page. The rationale for that is that
some callers need to account for every page added to a bio. Instead of
letting these callers call twice into the merge code to account for the
new vs existing page cases, just turn the paramter into an output one that
returns if a merge in the same page occured and let them act accordingly.Signed-off-by: Christoph Hellwig
Reviewed-by: Ming Lei
Signed-off-by: Jens Axboe
08 May, 2019
1 commit
-
Pull block updates from Jens Axboe:
"Nothing major in this series, just fixes and improvements all over the
map. This contains:- Series of fixes for sed-opal (David, Jonas)
- Fixes and performance tweaks for BFQ (via Paolo)
- Set of fixes for bcache (via Coly)
- Set of fixes for md (via Song)
- Enabling multi-page for passthrough requests (Ming)
- Queue release fix series (Ming)
- Device notification improvements (Martin)
- Propagate underlying device rotational status in loop (Holger)
- Removal of mtip32xx trim support, which has been disabled for years
(Christoph)- Improvement and cleanup of nvme command handling (Christoph)
- Add block SPDX tags (Christoph)
- Cleanup/hardening of bio/bvec iteration (Christoph)
- A few NVMe pull requests (Christoph)
- Removal of CONFIG_LBDAF (Christoph)
- Various little fixes here and there"
* tag 'for-5.2/block-20190507' of git://git.kernel.dk/linux-block: (164 commits)
block: fix mismerge in bvec_advance
block: don't drain in-progress dispatch in blk_cleanup_queue()
blk-mq: move cancel of hctx->run_work into blk_mq_hw_sysfs_release
blk-mq: always free hctx after request queue is freed
blk-mq: split blk_mq_alloc_and_init_hctx into two parts
blk-mq: free hw queue's resource in hctx's release handler
blk-mq: move cancel of requeue_work into blk_mq_release
blk-mq: grab .q_usage_counter when queuing request from plug code path
block: fix function name in comment
nvmet: protect discovery change log event list iteration
nvme: mark nvme_core_init and nvme_core_exit static
nvme: move command size checks to the core
nvme-fabrics: check more command sizes
nvme-pci: check more command sizes
nvme-pci: remove an unneeded variable initialization
nvme-pci: unquiesce admin queue on shutdown
nvme-pci: shutdown on timeout during deletion
nvme-pci: fix psdt field for single segment sgls
nvme-multipath: don't print ANA group state by default
nvme-multipath: split bios with the ns_head bio_set before submitting
...
30 Apr, 2019
1 commit
-
We only have two callers that need the integer loop iterator, and they
can easily maintain it themselves.Suggested-by: Matthew Wilcox
Reviewed-by: Johannes Thumshirn
Acked-by: David Sterba
Reviewed-by: Hannes Reinecke
Acked-by: Coly Li
Reviewed-by: Matthew Wilcox
Signed-off-by: Christoph Hellwig
Signed-off-by: Jens Axboe
17 Apr, 2019
2 commits
-
It's possible for pagecache writeback to split up a large amount of work
into smaller pieces for throttling purposes or to reduce the amount of
time a writeback operation is pending. Whatever the reason, XFS can end
up with a bunch of IO completions that call for the same operation to be
performed on a contiguous extent mapping. Since mappings are extent
based in XFS, we'd prefer to run fewer transactions when we can.When we're processing an ioend on the list of io completions, check to
see if the next items on the list are both adjacent and of the same
type. If so, we can merge the completions to reduce transaction
overhead.On fast storage this doesn't seem to make much of a difference in
performance, though the number of transactions for an overnight xfstests
run seems to drop by ~5%.Signed-off-by: Darrick J. Wong
Reviewed-by: Brian Foster -
When scheduling writeback of dirty file data in the page cache, XFS uses
IO completion workqueue items to ensure that filesystem metadata only
updates after the write completes successfully. This is essential for
converting unwritten extents to real extents at the right time and
performing COW remappings.Unfortunately, XFS queues each IO completion work item to an unbounded
workqueue, which means that the kernel can spawn dozens of threads to
try to handle the items quickly. These threads need to take the ILOCK
to update file metadata, which results in heavy ILOCK contention if a
large number of the work items target a single file, which is
inefficient.Worse yet, the writeback completion threads get stuck waiting for the
ILOCK while holding transaction reservations, which can use up all
available log reservation space. When that happens, metadata updates to
other parts of the filesystem grind to a halt, even if the filesystem
could otherwise have handled it.Even worse, if one of the things grinding to a halt happens to be a
thread in the middle of a defer-ops finish holding the same ILOCK and
trying to obtain more log reservation having exhausted the permanent
reservation, we now have an ABBA deadlock - writeback completion has a
transaction reserved and wants the ILOCK, and someone else has the ILOCK
and wants a transaction reservation.Therefore, we create a per-inode writeback io completion queue + work
item. When writeback finishes, it can add the ioend to the per-inode
queue and let the single worker item process that queue. This
dramatically cuts down on the number of kworkers and ILOCK contention in
the system, and seems to have eliminated an occasional deadlock I was
seeing while running generic/476.Testing with a program that simulates a heavy random-write workload to a
single file demonstrates that the number of kworkers drops from
approximately 120 threads per file to 1, without dramatically changing
write bandwidth or pagecache access latency.Note that we leave the xfs-conv workqueue's max_active alone because we
still want to be able to run ioend processing for as many inodes as the
system can handle.Signed-off-by: Darrick J. Wong
Reviewed-by: Brian Foster
09 Mar, 2019
1 commit
-
Pull block layer updates from Jens Axboe:
"Not a huge amount of changes in this round, the biggest one is that we
finally have Mings multi-page bvec support merged. Apart from that,
this pull request contains:- Small series that avoids quiescing the queue for sysfs changes that
match what we currently have (Aleksei)- Series of bcache fixes (via Coly)
- Series of lightnvm fixes (via Mathias)
- NVMe pull request from Christoph. Nothing major, just SPDX/license
cleanups, RR mp policy (Hannes), and little fixes (Bart,
Chaitanya).- BFQ series (Paolo)
- Save blk-mq cpu -> hw queue mapping, removing a pointer indirection
for the fast path (Jianchao)- fops->iopoll() added for async IO polling, this is a feature that
the upcoming io_uring interface will use (Christoph, me)- Partition scan loop fixes (Dongli)
- mtip32xx conversion from managed resource API (Christoph)
- cdrom registration race fix (Guenter)
- MD pull from Song, two minor fixes.
- Various documentation fixes (Marcos)
- Multi-page bvec feature. This brings a lot of nice improvements
with it, like more efficient splitting, larger IOs can be supported
without growing the bvec table size, and so on. (Ming)- Various little fixes to core and drivers"
* tag 'for-5.1/block-20190302' of git://git.kernel.dk/linux-block: (117 commits)
block: fix updating bio's front segment size
block: Replace function name in string with __func__
nbd: propagate genlmsg_reply return code
floppy: remove set but not used variable 'q'
null_blk: fix checking for REQ_FUA
block: fix NULL pointer dereference in register_disk
fs: fix guard_bio_eod to check for real EOD errors
blk-mq: use HCTX_TYPE_DEFAULT but not 0 to index blk_mq_tag_set->map
block: optimize bvec iteration in bvec_iter_advance
block: introduce mp_bvec_for_each_page() for iterating over page
block: optimize blk_bio_segment_split for single-page bvec
block: optimize __blk_segment_map_sg() for single-page bvec
block: introduce bvec_nth_page()
iomap: wire up the iopoll method
block: add bio_set_polled() helper
block: wire up block device iopoll method
fs: add an iopoll method to struct file_operations
loop: set GENHD_FL_NO_PART_SCAN after blkdev_reread_part()
loop: do not print warn message if partition scan is successful
block: bounce: make sure that bvec table is updated
...
21 Feb, 2019
2 commits
-
Add a mode where XFS never overwrites existing blocks in place. This
is to aid debugging our COW code, and also put infatructure in place
for things like possible future support for zoned block devices, which
can't support overwrites.This mode is enabled globally by doing a:
echo 1 > /sys/fs/xfs/debug/always_cow
Note that the parameter is global to allow running all tests in xfstests
easily in this mode, which would not easily be possible with a per-fs
sysfs file.In always_cow mode persistent preallocations are disabled, and fallocate
will fail when called with a 0 mode (with our without
FALLOC_FL_KEEP_SIZE), and not create unwritten extent for zeroed space
when called with FALLOC_FL_ZERO_RANGE or FALLOC_FL_UNSHARE_RANGE.There are a few interesting xfstests failures when run in always_cow
mode:- generic/392 fails because the bytes used in the file used to test
hole punch recovery are less after the log replay. This is
because the blocks written and then punched out are only freed
with a delay due to the logging mechanism.
- xfs/170 will fail as the already fragile file streams mechanism
doesn't seem to interact well with the COW allocator
- xfs/180 xfs/182 xfs/192 xfs/198 xfs/204 and xfs/208 will claim
the file system is badly fragmented, but there is not much we
can do to avoid that when always writing out of place
- xfs/205 fails because overwriting a file in always_cow mode
will require new space allocation and the assumption in the
test thus don't work anymore.
- xfs/326 fails to modify the file at all in always_cow mode after
injecting the refcount error, leading to an unexpected md5sum
after the remount, but that again is expectedSigned-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
This only matters if we want to write data through the COW fork that is
not actually an overwrite of existing data. Reasons for that are
speculative COW fork allocations using the cowextsize, or a mode where
we always write through the COW fork. Currently both can't actually
happen, but I plan to enable them.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
18 Feb, 2019
5 commits
-
While we can only truncate a block under the page lock for the current
page, there is no high-level synchronization for moving extents from the
COW to the data fork. This means that for example we can have another
thread doing a direct I/O completion that moves extents from the COW to
the data fork race with writeback. While this race is very hard to hit
the always_cow seems to reproduce it reasonably well, and it also exists
without that. Because of that there is a chance that a delalloc
conversion for the COW fork might not find any extents to convert. In
that case we should retry the whole block lookup and now find the blocks
in the data fork.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
Now that we properly handle the race with truncate in the delalloc
allocator there is no need to short cut this exceptional case earlier
on.Signed-off-by: Christoph Hellwig
Reviewed-by: Brian Foster
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
This function is a small wrapper only used by the writeback code, so
move it together with the writeback code and simplify it down to the
glorified do { } while loop that is now is.A few bits intentionally got lost here: no need to call xfs_qm_dqattach
because quotas are always attached when we create the delalloc
reservation, and no need for the imap->br_startblock == 0 check given
that xfs_bmapi_convert_delalloc already has a WARN_ON_ONCE for exactly
that condition.Signed-off-by: Christoph Hellwig
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
We already ensure all data fits into s_maxbytes in the write / fault
path. The only reason we have them here is that they were copy and
pasted from xfs_bmapi_read when we stopped using that function.Signed-off-by: Christoph Hellwig
Reviewed-by: Brian Foster
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong -
The io_type field contains what is basically a summary of information
from the inode fork and the imap. But we can just as easily use that
information directly, simplifying a few bits here and there and
improving the trace points.Signed-off-by: Christoph Hellwig
Reviewed-by: Brian Foster
Reviewed-by: Darrick J. Wong
Signed-off-by: Darrick J. Wong
15 Feb, 2019
1 commit
-
Pull in 5.0-rc6 to avoid a dumb merge conflict with fs/iomap.c.
This is needed since io_uring is now based on the block branch,
to avoid a conflict between the multi-page bvecs and the bits
of io_uring that touch the core block parts.* tag 'v5.0-rc6': (525 commits)
Linux 5.0-rc6
x86/mm: Make set_pmd_at() paravirt aware
MAINTAINERS: Update the ocores i2c bus driver maintainer, etc
blk-mq: remove duplicated definition of blk_mq_freeze_queue
Blk-iolatency: warn on negative inflight IO counter
blk-iolatency: fix IO hang due to negative inflight counter
MAINTAINERS: unify reference to xen-devel list
x86/mm/cpa: Fix set_mce_nospec()
futex: Handle early deadlock return correctly
futex: Fix barrier comment
net: dsa: b53: Fix for failure when irq is not defined in dt
blktrace: Show requests without sector
mips: cm: reprime error cause
mips: loongson64: remove unreachable(), fix loongson_poweroff().
sit: check if IPv6 enabled before calling ip6_err_gen_icmpv6_unreach()
geneve: should not call rt6_lookup() when ipv6 was disabled
KVM: nVMX: unconditionally cancel preemption timer in free_nested (CVE-2019-7221)
KVM: x86: work around leak of uninitialized stack contents (CVE-2019-7222)
kvm: fix kvm_ioctl_create_device() reference counting (CVE-2019-6974)
signal: Better detection of synchronous signals
...
