| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
folio_file_pos is only needed for mixed usage of page cache and swap
cache, for pure page cache usage, the caller can just use folio_pos
instead.
It can't be a swap cache page here. Swap mapping may only call into fs
through swap_rw and that is not supported for netfs. So just drop it and
use folio_pos instead.
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Kairui Song <[email protected]>
Cc: David Howells <[email protected]>
Cc: Jeff Layton <[email protected]>
Cc: Anna Schumaker <[email protected]>
Cc: Barry Song <[email protected]>
Cc: Chao Yu <[email protected]>
Cc: Chris Li <[email protected]>
Cc: David Hildenbrand <[email protected]>
Cc: "Huang, Ying" <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Ilya Dryomov <[email protected]>
Cc: Jaegeuk Kim <[email protected]>
Cc: Marc Dionne <[email protected]>
Cc: Matthew Wilcox (Oracle) <[email protected]>
Cc: Minchan Kim <[email protected]>
Cc: NeilBrown <[email protected]>
Cc: Ryan Roberts <[email protected]>
Cc: Ryusuke Konishi <[email protected]>
Cc: Trond Myklebust <[email protected]>
Cc: Xiubo Li <[email protected]>
Cc: Yosry Ahmed <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
|
|
Export fscache_put_volume() and add fscache_try_get_volume()
helper function to allow cachefiles to get/put fscache_volume
via linux/fscache-cache.h.
Signed-off-by: Baokun Li <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
Fix netfs_page_mkwrite() to use filemap_fdatawrite_range(), not
filemap_fdatawait_range() to flush conflicting data.
Fixes: 102a7e2c598c ("netfs: Allow buffered shared-writeable mmap through netfs_page_mkwrite()")
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
cc: Matthew Wilcox <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
Fix netfs_page_mkwrite() to check that folio->mapping is valid once it has
taken the folio lock (as filemap_page_mkwrite() does). Without this,
generic/247 occasionally oopses with something like the following:
BUG: kernel NULL pointer dereference, address: 0000000000000000
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
RIP: 0010:trace_event_raw_event_netfs_folio+0x61/0xc0
...
Call Trace:
<TASK>
? __die_body+0x1a/0x60
? page_fault_oops+0x6e/0xa0
? exc_page_fault+0xc2/0xe0
? asm_exc_page_fault+0x22/0x30
? trace_event_raw_event_netfs_folio+0x61/0xc0
trace_netfs_folio+0x39/0x40
netfs_page_mkwrite+0x14c/0x1d0
do_page_mkwrite+0x50/0x90
do_pte_missing+0x184/0x200
__handle_mm_fault+0x42d/0x500
handle_mm_fault+0x121/0x1f0
do_user_addr_fault+0x23e/0x3c0
exc_page_fault+0xc2/0xe0
asm_exc_page_fault+0x22/0x30
This is due to the invalidate_inode_pages2_range() issued at the end of the
DIO write interfering with the mmap'd writes.
Fixes: 102a7e2c598c ("netfs: Allow buffered shared-writeable mmap through netfs_page_mkwrite()")
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Reviewed-by: Jeff Layton <[email protected]>
cc: Matthew Wilcox <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
Delete some xarray-based buffer wangling functions that are intended for
use with bounce buffering, but aren't used because bounce-buffering got
deferred to a later patch series. Now, however, the intention is to use
something other than an xarray to do this.
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
During the writeback procedure, at the end of netfs_write_folio(), pending
write operations are flushed if the amount of write-streaming data stored
in a page is less than the size of the folio because if we haven't modified
a folio to the end, it cannot be contiguous with the following folio...
except if the dirty region of the folio is right at the end of the folio
space.
Fix the test to take the offset into the folio into account as well, such
that if the dirty region runs right up to the end of the folio, we leave
the flushing for later.
Fixes: 288ace2f57c9 ("netfs: New writeback implementation")
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: Marc Dionne <[email protected]>
cc: Steve French <[email protected]>
cc: Paulo Alcantara <[email protected]> (DFS, global name space)
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
[This was included in v2 of 9b038d004ce95551cb35381c49fe896c5bc11ffe, but
v1 got pushed instead]
Fix netfs_unbuffered_write_iter_locked() to set the total request length in
the netfs_io_request struct rather than leaving it as zero.
Fixes: 288ace2f57c9 ("netfs: New writeback implementation")
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: Steve French <[email protected]>
cc: Enzo Matsumiya <[email protected]>
cc: Christian Brauner <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
Instead of inventing a custom way to conditionally enable debugging,
just make use of pr_debug(), which also has dynamic debugging facilities
and is more likely known to someone who hunts a problem in the netfs
code. Also drop the module parameter netfs_debug which didn't have any
effect without further source changes. (The variable netfs_debug was
only used in #ifdef blocks for cpp vars that don't exist; Note that
CONFIG_NETFS_DEBUG isn't settable via kconfig, a variable with that name
never existed in the mainline and is probably just taken over (and
renamed) from similar custom debug logging implementations.)
Signed-off-by: Uwe Kleine-König <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs
Pull vfs fixes from Christian Brauner:
- Fix io_uring based write-through after converting cifs to use the
netfs library
- Fix aio error handling when doing write-through via netfs library
- Fix performance regression in iomap when used with non-large folio
mappings
- Fix signalfd error code
- Remove obsolete comment in signalfd code
- Fix async request indication in netfs_perform_write() by raising
BDP_ASYNC when IOCB_NOWAIT is set
- Yield swap device immediately to prevent spurious EBUSY errors
- Don't cross a .backup mountpoint from backup volumes in afs to avoid
infinite loops
- Fix a race between umount and async request completion in 9p after 9p
was converted to use the netfs library
* tag 'vfs-6.10-rc2.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
netfs, 9p: Fix race between umount and async request completion
afs: Don't cross .backup mountpoint from backup volume
swap: yield device immediately
netfs: Fix setting of BDP_ASYNC from iocb flags
signalfd: drop an obsolete comment
signalfd: fix error return code
iomap: fault in smaller chunks for non-large folio mappings
filemap: add helper mapping_max_folio_size()
netfs: Fix AIO error handling when doing write-through
netfs: Fix io_uring based write-through
|
|
There's a problem in 9p's interaction with netfslib whereby a crash occurs
because the 9p_fid structs get forcibly destroyed during client teardown
(without paying attention to their refcounts) before netfslib has finished
with them. However, it's not a simple case of deferring the clunking that
p9_fid_put() does as that requires the p9_client record to still be
present.
The problem is that netfslib has to unlock pages and clear the IN_PROGRESS
flag before destroying the objects involved - including the fid - and, in
any case, nothing checks to see if writeback completed barring looking at
the page flags.
Fix this by keeping a count of outstanding I/O requests (of any type) and
waiting for it to quiesce during inode eviction.
Reported-by: [email protected]
Link: https://lore.kernel.org/all/[email protected]/
Reported-by: [email protected]
Link: https://lore.kernel.org/all/[email protected]/
Reported-by: [email protected]
Link: https://lore.kernel.org/all/[email protected]/
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Tested-by: [email protected]
Reviewed-by: Dominique Martinet <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: Jeff Layton <[email protected]>
cc: Steve French <[email protected]>
cc: Hillf Danton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Reported-and-tested-by: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
Fix netfs_perform_write() to set BDP_ASYNC if IOCB_NOWAIT is set rather
than if IOCB_SYNC is not set. It reflects asynchronicity in the sense of
not waiting rather than synchronicity in the sense of not returning until
the op is complete.
Without this, generic/590 fails on cifs in strict caching mode with a
complaint that one of the writes fails with EAGAIN. The test can be
distilled down to:
mount -t cifs /my/share /mnt -ostuff
xfs_io -i -c 'falloc 0 8191M -c fsync -f /mnt/file
xfs_io -i -c 'pwrite -b 1M -W 0 8191M' /mnt/file
Fixes: c38f4e96e605 ("netfs: Provide func to copy data to pagecache for buffered write")
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Reviewed-by: Jens Axboe <[email protected]>
cc: Jeff Layton <[email protected]>
cc: Enzo Matsumiya <[email protected]>
cc: Jens Axboe <[email protected]>
cc: Matthew Wilcox <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
If an error occurs whilst we're doing an AIO write in write-through mode,
we may end up calling ->ki_complete() *and* returning an error from
->write_iter(). This can result in either a UAF (the ->ki_complete() func
pointer may get overwritten, for example) or a refcount underflow in
io_submit() as ->ki_complete is called twice.
Fix this by making netfs_end_writethrough() - and thus
netfs_perform_write() - unconditionally return -EIOCBQUEUED if we're doing
an AIO write and wait for completion if we're not.
Fixes: 288ace2f57c9 ("netfs: New writeback implementation")
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
cc: Jeff Layton <[email protected]>
cc: Enzo Matsumiya <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
This can be triggered by mounting a cifs filesystem with a cache=strict
mount option and then, using the fsx program from xfstests, doing:
ltp/fsx -A -d -N 1000 -S 11463 -P /tmp /cifs-mount/foo \
--replay-ops=gen112-fsxops
Where gen112-fsxops holds:
fallocate 0x6be7 0x8fc5 0x377d3
copy_range 0x9c71 0x77e8 0x2edaf 0x377d3
write 0x2776d 0x8f65 0x377d3
The problem is that netfs_io_request::len is being used for two purposes
and ends up getting set to the amount of data we transferred, not the
amount of data the caller asked to be transferred (for various reasons,
such as mmap'd writes, we might end up rounding out the data written to the
server to include the entire folio at each end).
Fix this by keeping the amount we were asked to write in ->len and using
->submitted to track what we issued ops for. Then, when we come to calling
->ki_complete(), ->len is the right size.
This also required netfs_cleanup_dio_write() to change since we're no
longer advancing wreq->len. Use wreq->transferred instead as we might have
done a short read.
With this, the generic/112 xfstest passes if cifs is forced to put all
non-DIO opens into write-through mode.
Fixes: 288ace2f57c9 ("netfs: New writeback implementation")
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
cc: Jeff Layton <[email protected]>
cc: Steve French <[email protected]>
cc: Enzo Matsumiya <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
With the changes to folios/netfs it is now easier to reenable
swapfile support over SMB3 which fixes various xfstests
Reviewed-by: David Howells <[email protected]>
Suggested-by: David Howells <[email protected]>
Fixes: e1209d3a7a67 ("mm: introduce ->swap_rw and use it for reads from SWP_FS_OPS swap-space")
Signed-off-by: Steve French <[email protected]>
|
|
Fix to take the i_rwsem (through the netfs locking wrappers) before taking
cinode->lock_sem.
Fixes: 3ee1a1fc3981 ("cifs: Cut over to using netfslib")
Reported-by: Enzo Matsumiya <[email protected]>
Signed-off-by: David Howells <[email protected]>
Signed-off-by: Steve French <[email protected]>
|
|
Make the cifs filesystem use netfslib to handle reading and writing on
behalf of cifs. The changes include:
(1) Various read_iter/write_iter type functions are turned into wrappers
around netfslib API functions or are pointed directly at those
functions:
cifs_file_direct{,_nobrl}_ops switch to use
netfs_unbuffered_read_iter and netfs_unbuffered_write_iter.
Large pieces of code that will be removed are #if'd out and will be removed
in subsequent patches.
[?] Why does cifs mark the page dirty in the destination buffer of a DIO
read? Should that happen automatically? Does netfs need to do that?
Signed-off-by: David Howells <[email protected]>
cc: Steve French <[email protected]>
cc: Shyam Prasad N <[email protected]>
cc: Rohith Surabattula <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Provide implementation of the netfslib hooks that will be used by netfslib
to ask cifs to set up and perform operations. Of particular note are
(*) cifs_clamp_length() - This is used to negotiate the size of the next
subrequest in a read request, taking into account the credit available
and the rsize. The credits are attached to the subrequest.
(*) cifs_req_issue_read() - This is used to issue a subrequest that has
been set up and clamped.
(*) cifs_prepare_write() - This prepares to fill a subrequest by picking a
channel, reopening the file and requesting credits so that we can set
the maximum size of the subrequest and also sets the maximum number of
segments if we're doing RDMA.
(*) cifs_issue_write() - This releases any unneeded credits and issues an
asynchronous data write for the contiguous slice of file covered by
the subrequest. This should possibly be folded in to all
->async_writev() ops and that called directly.
(*) cifs_begin_writeback() - This gets the cached writable handle through
which we do writeback (this does not affect writethrough, unbuffered
or direct writes).
At this point, cifs is not wired up to actually *use* netfslib; that will
be done in a subsequent patch.
Signed-off-by: David Howells <[email protected]>
cc: Steve French <[email protected]>
cc: Shyam Prasad N <[email protected]>
cc: Rohith Surabattula <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Use a hook in the new writeback code's retry algorithm to rotate the keys
once all the outstanding subreqs have failed rather than doing it
separately on each subreq.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Marc Dionne <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Do a couple of miscellaneous tidy ups:
(1) Add a qualifier into a file banner comment.
(2) Put the writeback folio traces back into alphabetical order.
(3) Remove some unused folio traces.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
|
|
Remove the old writeback code.
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: Marc Dionne <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Cut over to using the new writeback code. The old code is #ifdef'd out or
otherwise removed from compilation to avoid conflicts and will be removed
in a future patch.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: Marc Dionne <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Add some write-side stats to count buffered writes, buffered writethrough,
and writepages calls.
Whilst we're at it, clean up the naming on some of the existing stats
counters and organise the output into two sets.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
|
|
The current netfslib writeback implementation creates writeback requests of
contiguous folio data and then separately tiles subrequests over the space
twice, once for the server and once for the cache. This creates a few
issues:
(1) Every time there's a discontiguity or a change between writing to only
one destination or writing to both, it must create a new request.
This makes it harder to do vectored writes.
(2) The folios don't have the writeback mark removed until the end of the
request - and a request could be hundreds of megabytes.
(3) In future, I want to support a larger cache granularity, which will
require aggregation of some folios that contain unmodified data (which
only need to go to the cache) and some which contain modifications
(which need to be uploaded and stored to the cache) - but, currently,
these are treated as discontiguous.
There's also a move to get everyone to use writeback_iter() to extract
writable folios from the pagecache. That said, currently writeback_iter()
has some issues that make it less than ideal:
(1) there's no way to cancel the iteration, even if you find a "temporary"
error that means the current folio and all subsequent folios are going
to fail;
(2) there's no way to filter the folios being written back - something
that will impact Ceph with it's ordered snap system;
(3) and if you get a folio you can't immediately deal with (say you need
to flush the preceding writes), you are left with a folio hanging in
the locked state for the duration, when really we should unlock it and
relock it later.
In this new implementation, I use writeback_iter() to pump folios,
progressively creating two parallel, but separate streams and cleaning up
the finished folios as the subrequests complete. Either or both streams
can contain gaps, and the subrequests in each stream can be of variable
size, don't need to align with each other and don't need to align with the
folios.
Indeed, subrequests can cross folio boundaries, may cover several folios or
a folio may be spanned by multiple folios, e.g.:
+---+---+-----+-----+---+----------+
Folios: | | | | | | |
+---+---+-----+-----+---+----------+
+------+------+ +----+----+
Upload: | | |.....| | |
+------+------+ +----+----+
+------+------+------+------+------+
Cache: | | | | | |
+------+------+------+------+------+
The progressive subrequest construction permits the algorithm to be
preparing both the next upload to the server and the next write to the
cache whilst the previous ones are already in progress. Throttling can be
applied to control the rate of production of subrequests - and, in any
case, we probably want to write them to the server in ascending order,
particularly if the file will be extended.
Content crypto can also be prepared at the same time as the subrequests and
run asynchronously, with the prepped requests being stalled until the
crypto catches up with them. This might also be useful for transport
crypto, but that happens at a lower layer, so probably would be harder to
pull off.
The algorithm is split into three parts:
(1) The issuer. This walks through the data, packaging it up, encrypting
it and creating subrequests. The part of this that generates
subrequests only deals with file positions and spans and so is usable
for DIO/unbuffered writes as well as buffered writes.
(2) The collector. This asynchronously collects completed subrequests,
unlocks folios, frees crypto buffers and performs any retries. This
runs in a work queue so that the issuer can return to the caller for
writeback (so that the VM can have its kswapd thread back) or async
writes.
(3) The retryer. This pauses the issuer, waits for all outstanding
subrequests to complete and then goes through the failed subrequests
to reissue them. This may involve reprepping them (with cifs, the
credits must be renegotiated, and a subrequest may need splitting),
and doing RMW for content crypto if there's a conflicting change on
the server.
[!] Note that some of the functions are prefixed with "new_" to avoid
clashes with existing functions. These will be renamed in a later patch
that cuts over to the new algorithm.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: Marc Dionne <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Switch to using unsigned long long rather than loff_t in netfslib to avoid
problems with the sign flipping in the maths when we're dealing with the
byte at position 0x7fffffffffffffff.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Ilya Dryomov <[email protected]>
cc: Xiubo Li <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Use mempools for allocating requests and subrequests in an effort to make
sure that allocation always succeeds so that when performing writeback we
can always make progress.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Remove support for ->launder_folio() from netfslib and expect filesystems
to use filemap_invalidate_inode() instead. netfs_launder_folio() can then
be got rid of.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: David Howells <[email protected]>
cc: Marc Dionne <[email protected]>
cc: Steve French <[email protected]>
cc: Matthew Wilcox <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Implement a replacement for launder_folio. The key feature of
invalidate_inode_pages2() is that it locks each folio individually, unmaps
it to prevent mmap'd accesses interfering and calls the ->launder_folio()
address_space op to flush it. This has problems: firstly, each folio is
written individually as one or more small writes; secondly, adjacent folios
cannot be added so easily into the laundry; thirdly, it's yet another op to
implement.
Instead, use the invalidate lock to cause anyone wanting to add a folio to
the inode to wait, then unmap all the folios if we have mmaps, then,
conditionally, use ->writepages() to flush any dirty data back and then
discard all pages.
The invalidate lock prevents ->read_iter(), ->write_iter() and faulting
through mmap all from adding pages for the duration.
This is then used from netfslib to handle the flusing in unbuffered and
direct writes.
Signed-off-by: David Howells <[email protected]>
cc: Matthew Wilcox <[email protected]>
cc: Miklos Szeredi <[email protected]>
cc: Trond Myklebust <[email protected]>
cc: Christoph Hellwig <[email protected]>
cc: Andrew Morton <[email protected]>
cc: Alexander Viro <[email protected]>
cc: Christian Brauner <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Use the subreq_counter in netfs_io_request to allocate subrequest
debug_index values in read ops as well as write ops.
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
|
|
Make the netfs_io_request::subreq_counter, used to generate values for
netfs_io_subrequest::debug_index, into an atomic_t so that it can be called
from the retry thread at the same time as the app thread issuing writes.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
|
|
Remove the deprecated use of PG_private_2 in netfslib.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Matthew Wilcox (Oracle) <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Remove the PG_fscache alias for PG_private_2 and use the latter directly.
Use of this flag for marking pages undergoing writing to the cache should
be considered deprecated and the folios should be marked dirty instead and
the write done in ->writepages().
Note that PG_private_2 itself should be considered deprecated and up for
future removal by the MM folks too.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Matthew Wilcox (Oracle) <[email protected]>
cc: Ilya Dryomov <[email protected]>
cc: Xiubo Li <[email protected]>
cc: Steve French <[email protected]>
cc: Paulo Alcantara <[email protected]>
cc: Ronnie Sahlberg <[email protected]>
cc: Shyam Prasad N <[email protected]>
cc: Tom Talpey <[email protected]>
cc: Bharath SM <[email protected]>
cc: Trond Myklebust <[email protected]>
cc: Anna Schumaker <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
When dirty data is being written to the cache, setting/waiting on/clearing
the fscache flag is always done in tandem with setting/waiting on/clearing
the writeback flag. The netfslib buffered write routines wait on and set
both flags and the write request cleanup clears both flags, so the fscache
flag is almost superfluous.
The reason it isn't superfluous is because the fscache flag is also used to
indicate that data just read from the server is being written to the cache.
The flag is used to prevent a race involving overlapping direct-I/O writes
to the cache.
Change this to indicate that a page is in need of being copied to the cache
by placing a magic value in folio->private and marking the folios dirty.
Then when the writeback code sees a folio marked in this way, it only
writes it to the cache and not to the server.
If a folio that has this magic value set is modified, the value is just
replaced and the folio will then be uplodaded too.
With this, PG_fscache is no longer required by the netfslib core, 9p and
afs.
Ceph and nfs, however, still need to use the old PG_fscache-based tracking.
To deal with this, a flag, NETFS_ICTX_USE_PGPRIV2, now has to be set on the
flags in the netfs_inode struct for those filesystems. This reenables the
use of PG_fscache in that inode. 9p and afs use the netfslib write helpers
so get switched over; cifs, for the moment, does page-by-page manual access
to the cache, so doesn't use PG_fscache and is unaffected.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Matthew Wilcox (Oracle) <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: Marc Dionne <[email protected]>
cc: Ilya Dryomov <[email protected]>
cc: Xiubo Li <[email protected]>
cc: Steve French <[email protected]>
cc: Paulo Alcantara <[email protected]>
cc: Ronnie Sahlberg <[email protected]>
cc: Shyam Prasad N <[email protected]>
cc: Tom Talpey <[email protected]>
cc: Bharath SM <[email protected]>
cc: Trond Myklebust <[email protected]>
cc: Anna Schumaker <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Update i_blocks when i_size is updated when we finish making a write to the
pagecache to reflect the amount of space we think will be consumed.
This maintains cifs commit dbfdff402d89854126658376cbcb08363194d3cd ("smb3:
update allocation size more accurately on write completion") which would
otherwise be removed by the cifs part of the netfs writeback rewrite.
Signed-off-by: David Howells <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Steve French <[email protected]>
cc: Shyam Prasad N <[email protected]>
cc: Rohith Surabattula <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
In netfs_perform_write(), when the file is marked NETFS_ICTX_WRITETHROUGH
or O_*SYNC or RWF_*SYNC was specified, write-through caching is performed
on a buffered file. When setting up for write-through, we flush any
conflicting writes in the region and wait for the write to complete,
failing if there's a write error to return.
The issue arises if we're writing at or above the EOF position because we
skip the flush and - more importantly - the wait. This becomes a problem
if there's a partial folio at the end of the file that is being written out
and we want to make a write to it too. Both the already-running write and
the write we start both want to clear the writeback mark, but whoever is
second causes a warning looking something like:
------------[ cut here ]------------
R=00000012: folio 11 is not under writeback
WARNING: CPU: 34 PID: 654 at fs/netfs/write_collect.c:105
...
CPU: 34 PID: 654 Comm: kworker/u386:27 Tainted: G S ...
...
Workqueue: events_unbound netfs_write_collection_worker
...
RIP: 0010:netfs_writeback_lookup_folio
Fix this by making the flush-and-wait unconditional. It will do nothing if
there are no folios in the pagecache and will return quickly if there are
no folios in the region specified.
Further, move the WBC attachment above the flush call as the flush is going
to attach a WBC and detach it again if it is not present - and since we
need one anyway we might as well share it.
Fixes: 41d8e7673a77 ("netfs: Implement a write-through caching option")
Reported-by: kernel test robot <[email protected]>
Closes: https://lore.kernel.org/oe-lkp/[email protected]
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Reviewed-by: Jeffrey Layton <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: Marc Dionne <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
Fix the error return in netfs_perform_write() acting in writethrough-mode
to return any cached error in the case that netfs_end_writethrough()
returns 0.
This can affect the use of O_SYNC/O_DSYNC/RWF_SYNC/RWF_DSYNC in 9p and afs.
Fixes: 41d8e7673a77 ("netfs: Implement a write-through caching option")
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Reviewed-by: Jeff Layton <[email protected]>
cc: Eric Van Hensbergen <[email protected]>
cc: Latchesar Ionkov <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Christian Schoenebeck <[email protected]>
cc: Marc Dionne <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
Fix fscache_begin_operation() to clear cres->cache_priv on error, otherwise
fscache_resources_valid() will report it as being valid.
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Reviewed-by: Jeff Layton <[email protected]>
Reported-by: Marc Dionne <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
Signed-off-by: Christian Brauner <[email protected]>
|
|
Fix netfs_unbuffered_write_iter() to return immediately if
generic_write_checks() returns 0, indicating there's nothing to write.
Note that netfs_file_write_iter() already does this.
Also, whilst we're at it, put in checks for the size being zero before we
even take the locks. Note that generic_write_checks() can still reduce the
size to zero, so we still need that check.
Without this, a warning similar to the following is logged to dmesg:
netfs: Zero-sized write [R=1b6da]
and the syscall fails with EIO, e.g.:
/sbin/ldconfig.real: Writing of cache extension data failed: Input/output error
This can be reproduced on 9p by:
xfs_io -f -c 'pwrite 0 0' /xfstest.test/foo
Fixes: 153a9961b551 ("netfs: Implement unbuffered/DIO write support")
Reported-by: Eric Van Hensbergen <[email protected]>
Link: https://lore.kernel.org/r/ZbQUU6QKmIftKsmo@FV7GG9FTHL/
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Tested-by: Dominique Martinet <[email protected]>
Reviewed-by: Jeff Layton <[email protected]>
cc: Dominique Martinet <[email protected]>
cc: Jeff Layton <[email protected]>
cc: <[email protected]>
cc: <[email protected]>
cc: <[email protected]>
cc: <[email protected]>
Signed-off-by: Christian Brauner <[email protected]>
|
|
If netfs_begin_read gets a NETFS_DIO_READ request that begins
past i_size, it won't perform any i/o and just return 0. This
will leak an increment to i_dio_count that is done at the top
of the function.
This can cause subsequent buffered read requests to block
indefinitely, waiting for a non existing dio operation to complete.
Add a inode_dio_end() for the NETFS_DIO_READ case, before returning.
Signed-off-by: Marc Dionne <[email protected]>
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Reviewed-by: Jeff Layton <[email protected]>
cc: Jeff Layton <[email protected]>
cc: <[email protected]>
cc: <[email protected]>
cc: <[email protected]>
Signed-off-by: Christian Brauner <[email protected]>
|
|
The netfs_grab_folio_for_write() function doesn't return NULL, it returns
error pointers. Update the check accordingly.
Fixes: c38f4e96e605 ("netfs: Provide func to copy data to pagecache for buffered write")
Signed-off-by: Dan Carpenter <[email protected]>
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]/
|
|
This function dereferences "cache" and then checks if it's
IS_ERR_OR_NULL(). Check first, then dereference.
Fixes: 9549332df4ed ("fscache: Implement cache registration")
Signed-off-by: Dan Carpenter <[email protected]>
Signed-off-by: David Howells <[email protected]>
Link: https://lore.kernel.org/r/[email protected]/ # v2
|
|
Filesystems should use folio->index and folio->mapping, instead of
folio_index(folio), folio_mapping() and folio_file_mapping() since
they know that it's in the pagecache.
Change this automagically with:
perl -p -i -e 's/folio_mapping[(]([^)]*)[)]/\1->mapping/g' fs/netfs/*.c
perl -p -i -e 's/folio_file_mapping[(]([^)]*)[)]/\1->mapping/g' fs/netfs/*.c
perl -p -i -e 's/folio_index[(]([^)]*)[)]/\1->index/g' fs/netfs/*.c
Reported-by: Matthew Wilcox <[email protected]>
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
gitolite.kernel.org:pub/scm/linux/kernel/git/vfs/vfs
Pull netfs updates from Christian Brauner:
"This extends the netfs helper library that network filesystems can use
to replace their own implementations. Both afs and 9p are ported. cifs
is ready as well but the patches are way bigger and will be routed
separately once this is merged. That will remove lots of code as well.
The overal goal is to get high-level I/O and knowledge of the page
cache and ouf of the filesystem drivers. This includes knowledge about
the existence of pages and folios
The pull request converts afs and 9p. This removes about 800 lines of
code from afs and 300 from 9p. For 9p it is now possible to do writes
in larger than a page chunks. Additionally, multipage folio support
can be turned on for 9p. Separate patches exist for cifs removing
another 2000+ lines. I've included detailed information in the
individual pulls I took.
Summary:
- Add NFS-style (and Ceph-style) locking around DIO vs buffered I/O
calls to prevent these from happening at the same time.
- Support for direct and unbuffered I/O.
- Support for write-through caching in the page cache.
- O_*SYNC and RWF_*SYNC writes use write-through rather than writing
to the page cache and then flushing afterwards.
- Support for write-streaming.
- Support for write grouping.
- Skip reads for which the server could only return zeros or EOF.
- The fscache module is now part of the netfs library and the
corresponding maintainer entry is updated.
- Some helpers from the fscache subsystem are renamed to mark them as
belonging to the netfs library.
- Follow-up fixes for the netfs library.
- Follow-up fixes for the 9p conversion"
* tag 'vfs-6.8.netfs' of gitolite.kernel.org:pub/scm/linux/kernel/git/vfs/vfs: (50 commits)
netfs: Fix wrong #ifdef hiding wait
cachefiles: Fix signed/unsigned mixup
netfs: Fix the loop that unmarks folios after writing to the cache
netfs: Fix interaction between write-streaming and cachefiles culling
netfs: Count DIO writes
netfs: Mark netfs_unbuffered_write_iter_locked() static
netfs: Fix proc/fs/fscache symlink to point to "netfs" not "../netfs"
netfs: Rearrange netfs_io_subrequest to put request pointer first
9p: Use length of data written to the server in preference to error
9p: Do a couple of cleanups
9p: Fix initialisation of netfs_inode for 9p
cachefiles: Fix __cachefiles_prepare_write()
9p: Use netfslib read/write_iter
afs: Use the netfs write helpers
netfs: Export the netfs_sreq tracepoint
netfs: Optimise away reads above the point at which there can be no data
netfs: Implement a write-through caching option
netfs: Provide a launder_folio implementation
netfs: Provide a writepages implementation
netfs, cachefiles: Pass upper bound length to allow expansion
...
|
|
netfs_writepages_begin() has the wait on the fscache folio conditional on
CONFIG_NETFS_FSCACHE - which doesn't exist.
Fix it to be conditional on CONFIG_FSCACHE instead.
Fixes: 62c3b7481b9a ("netfs: Provide a writepages implementation")
Reported-by: Lukas Bulwahn <[email protected]>
Signed-off-by: David Howells <[email protected]>
cc: Matthew Wilcox <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]/
|
|
In the loop in netfs_rreq_unmark_after_write() that removes the PG_fscache
from folios after they've been written to the cache, as soon as we remove
the mark from a multipage folio, it can get split - and then we might see a
fragment of folio again.
Guard against this by advancing the 'unlocked' tracker to the index of the
last page in the folio to avoid a double removal of the PG_fscache mark.
Reported-by: Marc Dionne <[email protected]>
Signed-off-by: David Howells <[email protected]>
cc: Matthew Wilcox <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
An issue can occur between write-streaming (storing dirty data in partial
non-uptodate pages) and a cachefiles object being culled to make space.
The problem occurs because the cache object is only marked in use while
there are files open using it. Once it has been released, it can be culled
and the cookie marked disabled.
At this point, a streaming write is permitted to occur (if the cache is
active, we require pages to be prefetched and cached), but the cache can
become active again before this gets flushed out - and then two effects can
occur:
(1) The cache may be asked to write out a region that's less than its DIO
block size (assumed by cachefiles to be PAGE_SIZE) - and this causes
one of two debugging statements to be emitted.
(2) netfs_how_to_modify() gets confused because it sees a page that isn't
allowed to be non-uptodate being uptodate and tries to prefetch it -
leading to a warning that PG_fscache is set twice.
Fix this by the following means:
(1) Add a netfs_inode flag to disallow write-streaming to an inode and set
it if we ever do local caching of that inode. It remains set for the
lifetime of that inode - even if the cookie becomes disabled.
(2) If the no-write-streaming flag is set, then make netfs_how_to_modify()
always want to prefetch instead.
(3) If netfs_how_to_modify() decides it wants to prefetch a folio, but
that folio has write-streamed data in it, then it requires the folio
be flushed first.
(4) Export a counter of the number of times we wanted to prefetch a
non-uptodate page, but found it had write-streamed data in it.
(5) Export a counter of the number of times we cancelled a write to the
cache because it didn't DIO align and remove the debug statements.
Reported-by: Marc Dionne <[email protected]>
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Provide a counter for DIO writes to match that for DIO reads.
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Mark netfs_unbuffered_write_iter_locked() static as it's only called from
the file in which it is defined.
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Fix the proc/fs/fscache symlink to point to "netfs" not "../netfs".
Reported-by: Marc Dionne <[email protected]>
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: Christian Brauner <[email protected]>
cc: [email protected]
cc: [email protected]
|
|
Export the netfs_sreq tracepoint so that it can be called directly from
client filesystems/cache backend modules.
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
|
|
Track the file position above which the server is not expected to have any
data (the "zero point") and preemptively assume that we can satisfy
requests by filling them with zeroes locally rather than attempting to
download them if they're over that line - even if we've written data back
to the server. Assume that any data that was written back above that
position is held in the local cache. Note that we have to split requests
that straddle the line.
Make use of this to optimise away some reads from the server. We need to
set the zero point in the following circumstances:
(1) When we see an extant remote inode and have no cache for it, we set
the zero_point to i_size.
(2) On local inode creation, we set zero_point to 0.
(3) On local truncation down, we reduce zero_point to the new i_size if
the new i_size is lower.
(4) On local truncation up, we don't change zero_point.
(5) On local modification, we don't change zero_point.
(6) On remote invalidation, we set zero_point to the new i_size.
(7) If stored data is discarded from the pagecache or culled from fscache,
we must set zero_point above that if the data also got written to the
server.
(8) If dirty data is written back to the server, but not fscache, we must
set zero_point above that.
(9) If a direct I/O write is made, set zero_point above that.
Assuming the above, any read from the server at or above the zero_point
position will return all zeroes.
The zero_point value can be stored in the cache, provided the above rules
are applied to it by any code that culls part of the local cache.
Signed-off-by: David Howells <[email protected]>
cc: Jeff Layton <[email protected]>
cc: [email protected]
cc: [email protected]
cc: [email protected]
|