The iclog IO completion state processing is somewhat complex, and
because it's inside two nested loops it is highly indented and very
hard to read. Factor it out, flatten the logic flow and clean up the
comments so that it much easier to see what the code is doing both
in processing the individual iclogs and in the over
xlog_state_do_callback() operation.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Simplify the code flow by lifting the iclog callback work out of
the main iclog iteration loop. This isolates the log juggling and
callbacks from the iclog state change logic in the loop.
Note that the loopdidcallbacks variable is not actually tracking
whether callbacks are actually run - it is tracking whether the
icloglock was dropped during the loop and so determines if we
completed the entire iclog scan loop atomically. Hence we know for
certain there are either no more ordered completions to run or
that the next completion will run the remaining ordered iclog
completions. Hence rename that variable appropriately for it's
function.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Start making this function readable by lifting the debug code into
a conditional function.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
generic/530 on a machine with enough ram and a non-preemptible
kernel can run the AGI processing phase of log recovery enitrely out
of cache. This means it never blocks on locks, never waits for IO
and runs entirely through the unlinked lists until it either
completes or blocks and hangs because it has run out of log space.
It runs out of log space because the background CIL push is
scheduled but never runs. queue_work() queues the CIL work on the
current CPU that is busy, and the workqueue code will not run it on
any other CPU. Hence if the unlinked list processing never yields
the CPU voluntarily, the push work is delayed indefinitely. This
results in the CIL aggregating changes until all the log space is
consumed.
When the log recoveyr processing evenutally blocks, the CIL flushes
but because the last iclog isn't submitted for IO because it isn't
full, the CIL flush never completes and nothing ever moves the log
head forwards, or indeed inserts anything into the tail of the log,
and hence nothing is able to get the log moving again and recovery
hangs.
There are several problems here, but the two obvious ones from
the trace are that:
a) log recovery does not yield the CPU for over 4 seconds,
b) binding CIL pushes to a single CPU is a really bad idea.
This patch addresses just these two aspects of the problem, and are
suitable for backporting to work around any issues in older kernels.
The more fundamental problem of preventing the CIL from consuming
more than 50% of the log without committing will take more invasive
and complex work, so will be done as followup work.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
The code in xlog_wait uses the spinlock to make adding the task to
the wait queue, and setting the task state to UNINTERRUPTIBLE atomic
with respect to the waker.
Doing the wakeup after releasing the spinlock opens up the following
race condition:
Task 1 task 2
add task to wait queue
wake up task
set task state to UNINTERRUPTIBLE
This issue was found through code inspection as a result of kworkers
being observed stuck in UNINTERRUPTIBLE state with an empty
wait queue. It is rare and largely unreproducable.
Simply moving the spin_unlock to after the wake_up_all results
in the waker not being able to see a task on the waitqueue before
it has set its state to UNINTERRUPTIBLE.
This bug dates back to the conversion of this code to generic
waitqueue infrastructure from a counting semaphore back in 2008
which didn't place the wakeups consistently w.r.t. to the relevant
spin locks.
[dchinner: Also fix a similar issue in the shutdown path on
xc_commit_wait. Update commit log with more details of the issue.]
Fixes: d748c62367 ("[XFS] Convert l_flushsema to a sv_t")
Reported-by: Chris Mason <clm@fb.com>
Signed-off-by: Rik van Riel <riel@surriel.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
In the situation where the log is full and the CIL has not recently
flushed, the AIL push threshold is throttled back to the where the
last write of the head of the log was completed. This is stored in
log->l_last_sync_lsn. Hence if the CIL holds > 25% of the log space
pinned by flushes and/or aggregation in progress, we can get the
situation where the head of the log lags a long way behind the
reservation grant head.
When this happens, the AIL push target is trimmed back from where
the reservation grant head wants to push the log tail to, back to
where the head of the log currently is. This means the push target
doesn't reach far enough into the log to actually move the tail
before the transaction reservation goes to sleep.
When the CIL push completes, it moves the log head forward such that
the AIL push target can now be moved, but that has no mechanism for
puhsing the log tail. Further, if the next tail movement of the log
is not large enough wake the waiter (i.e. still not enough space for
it to have a reservation granted), we don't wake anything up, and
hence we do not update the AIL push target to take into account the
head of the log moving and allowing the push target to be moved
forwards.
To avoid this particular condition, if we fail to wake the first
waiter on the grant head because we don't have enough space,
push on the AIL again. This will pick up any movement of the log
head and allow the push target to move forward due to completion of
CIL pushing.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
If the CONFIG_BUG is enabled, BUG is executed and then system is crashed.
However, the bailout for mount is no longer proceeding.
Using WARN_ON_ONCE rather than BUG can prevent this situation.
Signed-off-by: Austin Kim <austindh.kim@gmail.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Convert the squashfs filesystem to the new internal mount API as the old
one will be obsoleted and removed. This allows greater flexibility in
communication of mount parameters between userspace, the VFS and the
filesystem.
See Documentation/filesystems/mount_api.txt for more information.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Phillip Lougher <phillip@squashfs.org.uk>
cc: squashfs-devel@lists.sourceforge.net
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Convert the jffs2 filesystem to the new internal mount API as the old
one will be obsoleted and removed. This allows greater flexibility in
communication of mount parameters between userspace, the VFS and the
filesystem.
See Documentation/filesystems/mount_api.txt for more information.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: David Woodhouse <dwmw2@infradead.org>
cc: linux-mtd@lists.infradead.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Convert the cramfs filesystem to the new internal mount API as the old
one will be obsoleted and removed. This allows greater flexibility in
communication of mount parameters between userspace, the VFS and the
filesystem.
See Documentation/filesystems/mount_api.txt for more information.
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Nicolas Pitre <nico@fluxnic.net>
Acked-by: Nicolas Pitre <nico@fluxnic.net>
cc: linux-mtd@lists.infradead.org
cc: linux-block@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Convert the romfs filesystem to the new internal mount API as the old
one will be obsoleted and removed. This allows greater flexibility in
communication of mount parameters between userspace, the VFS and the
filesystem.
See Documentation/filesystems/mount_api.txt for more information.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: linux-mtd@lists.infradead.org
cc: linux-block@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Add an additional keying mode to vfs_get_super() to indicate that only a
single superblock should exist in the system, and that, if it does, further
mounts should invoke reconfiguration upon it.
This allows mount_single() to be replaced.
[Fix by Eric Biggers folded in]
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Create a function, get_tree_bdev(), that is fs_context-aware and a
->get_tree() counterpart of mount_bdev().
It caches the block device pointer in the fs_context struct so that this
information can be passed into sget_fc()'s test and set functions.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Jens Axboe <axboe@kernel.dk>
cc: linux-block@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
As Christoph said [1], "I'd much prefer to just use
read_cache_page_gfp, and live with the fact that this
allocates bufferheads behind you for now. I'll try to
speed up my attempts to get rid of the buffer heads on
the block device mapping instead. "
This simplifies the code a lot and a minor thing is
"no REQ_META (e.g. for blktrace) on metadata at all..."
[1] https://lore.kernel.org/r/20190903153704.GA2201@infradead.org/
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-26-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As Christoph said [1], "This seems to be your only direct
use of buffer heads, which while not deprecated are a bit
of an ugly step child. So if you can easily avoid creating
a buffer_head dependency in a new filesystem I think you
should avoid it. "
[1] https://lore.kernel.org/r/20190902125109.GA9826@infradead.org/
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-24-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add prefix "erofs_" to these functions and print
sb->s_id as a prefix to erofs_{err, info} so that
the user knows which file system is affected.
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-23-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As Christoph said [1],
"vm_map_ram is supposed to generally behave better. So if
it doesn't please report that that to the arch maintainer
and linux-mm so that they can look into the issue. Having
user make choices of deep down kernel internals is just
a horrible interface.
Please talk to maintainers of other bits of the kernel
if you see issues and / or need enhancements. "
Let's redo the previous conclusion and kill the vmap
approach.
[1] https://lore.kernel.org/r/20190830165533.GA10909@infradead.org/
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-21-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As Christoph pointed out [1], "
Why is there __submit_bio which really just obsfucates
what is going on? Also why is __submit_bio using
bio_set_op_attrs instead of opencode it as the comment
right next to it asks you to? "
Let's use submit_bio directly instead.
[1] https://lore.kernel.org/r/20190830162812.GA10694@infradead.org/
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-18-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As Christoph pointed out [1],
"Why is there __erofs_get_meta_page with the two weird
booleans instead of a single erofs_get_meta_page that
gets and gfp_t for additional flags and an unsigned int
for additional bio op flags."
And since all callers can handle errors, let's kill
prio and nofail and erofs_get_inline_page() now.
[1] https://lore.kernel.org/r/20190830162812.GA10694@infradead.org/
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-17-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As Christoph said [1] "having this function seems
entirely pointless", let's kill those.
filesystem function name
ext2,f2fs,ext4,isofs,squashfs,cifs,... init_inodecache
In addition, add a necessary "rcu_barrier()" on exit_fs();
[1] https://lore.kernel.org/r/20190829101545.GC20598@infradead.org/
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-9-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As Christoph said, "This looks like a really obsfucated
way to write:
return datamode == EROFS_INODE_FLAT_COMPRESSION ||
datamode == EROFS_INODE_FLAT_COMPRESSION_LEGACY; "
Although I had my own consideration, it's the right way for now.
[1] https://lore.kernel.org/r/20190829095954.GB20598@infradead.org/
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-6-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As Christoph suggested "Please don't add __packed" [1],
remove all __packed except struct erofs_dirent here.
Note that all on-disk fields except struct erofs_dirent
(12 bytes with a 8-byte nid) in EROFS are naturally aligned.
[1] https://lore.kernel.org/r/20190829095954.GB20598@infradead.org/
Reported-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Link: https://lore.kernel.org/r/20190904020912.63925-5-gaoxiang25@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When ext4 file systems were created intentionally with 128 byte inodes,
the rate-limited warning of eventual possible timestamp overflow are
still emitted rather frequently. Remove the warning for now.
Discussion for whether any warning is needed,
and where it should be emitted, can be found at
https://lore.kernel.org/lkml/1567523922.5576.57.camel@lca.pw/.
I can post a separate follow-up patch after the conclusion.
Reported-by: Qian Cai <cai@lca.pw>
Signed-off-by: Deepa Dinamani <deepa.kernel@gmail.com>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Make sure that attribute methods are not called after the item
has been removed from the tree. To do so, we
* at the point of no return in removals, grab ->frag_sem
exclusive and mark the fragment dead.
* call the methods of attributes with ->frag_sem taken
shared and only after having verified that the fragment is still
alive.
The main benefit is for method instances - they are
guaranteed that the objects they are accessing *and* all ancestors
are still there. Another win is that we don't need to bother
with extra refcount on config_item when opening a file -
the item will be alive for as long as it stays in the tree, and
we won't touch it/attributes/any associated data after it's
been removed from the tree.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Because s_vfs_rename_mutex is not cluster-wide, multiple nodes can
reverse the roles of which directories are "old" and which are "new" for
the purposes of rename. This can cause deadlocks where two nodes end up
waiting for each other.
There can be several layers of directory dependencies across many nodes.
This patch fixes the problem by acquiring all gfs2_rename's inode glocks
asychronously and waiting for all glocks to be acquired. That way all
inodes are locked regardless of the order.
The timeout value for multiple asynchronous glocks is calculated to be
the total of the individual wait times for each glock times two.
Since gfs2_exchange is very similar to gfs2_rename, both functions are
patched in the same way.
A new async glock wait queue, sd_async_glock_wait, keeps a list of
waiters for these events. If gfs2's holder_wake function detects an
async holder, it wakes up any waiters for the event. The waiter only
tests whether any of its requests are still pending.
Since the glocks are sent to dlm asychronously, the wait function needs
to check to see which glocks, if any, were granted.
If a glock is granted by dlm (and therefore held), its minimum hold time
is checked and adjusted as necessary, as other glock grants do.
If the event times out, all glocks held thus far must be dequeued to
resolve any existing deadlocks. Then, if there are any outstanding
locking requests, we need to loop around and wait for dlm to respond to
those requests too. After we release all requests, we return -ESTALE to
the caller (vfs rename) which loops around and retries the request.
Node1 Node2
--------- ---------
1. Enqueue A Enqueue B
2. Enqueue B Enqueue A
3. A granted
6. B granted
7. Wait for B
8. Wait for A
9. A times out (since Node 1 holds A)
10. Dequeue B (since it was granted)
11. Wait for all requests from DLM
12. B Granted (since Node2 released it in step 10)
13. Rename
14. Dequeue A
15. DLM Grants A
16. Dequeue A (due to the timeout and since we
no longer have B held for our task).
17. Dequeue B
18. Return -ESTALE to vfs
19. VFS retries the operation, goto step 1.
This release-all-locks / acquire-all-locks may slow rename / exchange
down as both nodes struggle in the same way and do the same thing.
However, this will only happen when there is contention for the same
inodes, which ought to be rare.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
This patch moves the code that updates glock minimum hold
time to a separate function. This will be called by a future
patch.
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Before this patch, gfs2_rename added a holder for the rgrp glock to
its array of holders, ghs. There's nothing wrong with that, but this
patch separates it into a separate holder. This is done to ensure
it's always locked last as per the proper glock lock ordering,
and also to pave the way for a future patch in which we will
lock the non-rgrp glocks asynchronously.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
The brelse() function tests whether its argument is NULL and then
returns immediately. Thus the test around the call is not needed.
This issue was detected by using the Coccinelle software.
[The same applies to brelse() in gfs2_dir_no_add (which Coccinelle
apparently missed), so fix that as well.]
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
The brelse() function tests whether its argument is NULL
and then returns immediately.
Thus the test around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Link: https://lore.kernel.org/r/a254c1d1-0109-ab51-c67a-edc5c1c4b4cd@web.de
Signed-off-by: Jan Kara <jack@suse.cz>
The brelse() function tests whether its argument is NULL
and then returns immediately.
Thus the test around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Link: https://lore.kernel.org/r/51dea296-2207-ebc0-bac3-13f3e5c3b235@web.de
Signed-off-by: Jan Kara <jack@suse.cz>
