Ceph can in some cases issue an async DIO request, in which case we can
end up calling ceph_end_io_direct before the I/O is actually complete.
That may allow buffered operations to proceed while DIO requests are
still in flight.
Fix this by incrementing the i_dio_count when issuing an async DIO
request, and decrement it when tearing down the aio_req.
Fixes: 321fe13c93 ("ceph: add buffered/direct exclusionary locking for reads and writes")
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Most of the time, we (or the vfs layer) takes the inode_lock and then
acquires caps, but ceph_read_iter does the opposite, and that can lead
to a deadlock.
When there are multiple clients treading over the same data, we can end
up in a situation where a reader takes caps and then tries to acquire
the inode_lock. Another task holds the inode_lock and issues a request
to the MDS which needs to revoke the caps, but that can't happen until
the inode_lock is unwedged.
Fix this by having ceph_read_iter take the inode_lock earlier, before
attempting to acquire caps.
Fixes: 321fe13c93 ("ceph: add buffered/direct exclusionary locking for reads and writes")
Link: https://tracker.ceph.com/issues/36348
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
If someone requests fscache on the mount, and the kernel doesn't
support it, it should fail the mount.
[ Drop ceph prefix -- it's provided by pr_err. ]
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
If ceph_atomic_open is handed a !d_in_lookup dentry, then that means
that it already passed d_revalidate so we *know* that it's negative (or
at least was very recently). Just return -ENOENT in that case.
This also addresses a subtle bug in dentry handling. Non-O_CREAT opens
call atomic_open with the parent's i_rwsem shared, but calling
d_splice_alias on a hashed dentry requires the exclusive lock.
If ceph_atomic_open receives a hashed, negative dentry on a non-O_CREAT
open, and another client were to race in and create the file before we
issue our OPEN, ceph_fill_trace could end up calling d_splice_alias on
the dentry with the new inode with insufficient locks.
Cc: stable@vger.kernel.org
Reported-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
In the future, we're going to want to extend the ceph_reply_info_extra
for create replies. Currently though, the kernel code doesn't accept an
extra blob that is larger than the expected data.
Change the code to skip over any unrecognized fields at the end of the
extra blob, rather than returning -EIO.
Cc: stable@vger.kernel.org
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Most filesystems don't limit what security.* xattrs can be set or
fetched. I see no reason that we need to limit that on cephfs either.
Drop the special xattr handler for "security." xattrs, and allow the
"other" xattr handler to handle security xattrs as well.
In addition to fixing xfstest generic/093, this allows us to support
per-file capabilities (a'la setcap(8)).
Link: https://tracker.ceph.com/issues/41135
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
__ceph_getxattr will set the CEPH_I_SEC_INITED flag whenever it gets
any xattr that starts with "security.". We only want to set that flag
when fetching the MAC label for the currently-active LSM, however.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
No need to do an extra jump here. Also add some comments on the endifs.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
xfstest generic/451 intermittently fails. The test does O_DIRECT writes
to a file, and then reads back the result using buffered I/O, while
running a separate set of tasks that are also doing buffered reads.
The client will invalidate the cache prior to a direct write, but it's
easy for one of the other readers' replies to race in and reinstantiate
the invalidated range with stale data.
To fix this, we must to serialize direct I/O writes and buffered reads.
We could just sprinkle in some shared locks on the i_rwsem for reads,
and increase the exclusive footprint on the write side, but that would
cause O_DIRECT writes to end up serialized vs. other direct requests.
Instead, borrow the scheme used by nfs.ko. Buffered writes take the
i_rwsem exclusively, but buffered reads take a shared lock, allowing
them to run in parallel.
O_DIRECT requests also take a shared lock, but we need for them to not
run in parallel with buffered reads. A flag on the ceph_inode_info is
used to indicate whether it's in direct or buffered I/O mode. When a
conflicting request is submitted, it will block until the inode can be
flipped to the necessary mode.
Link: https://tracker.ceph.com/issues/40985
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Prevent freezing operations during write page faults. This is good
practice for most filesystems, but especially for ceph since we're
monkeying with the signal table here.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
If we have Fx caps, and the we're truncating the size to be larger, then
we'll cache the size attribute change, but the mtime won't be updated.
Move the size handling before the mtime, and add ATTR_MTIME to ia_valid
in that case to make sure the mtime also gets updated.
This fixes xfstest generic/313.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
It doesn't do anything to invalidate the cache when dropping RD caps.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
cap->session is always non-NULL, so we can just do a single test for
equality w/o testing explicitly for a NULL pointer.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Currently, this function returns ci->i_dirty_caps, but the callers have
to check that that isn't 0 before calling this function. Have the
callers grab that value directly out of the inode, and have
__mark_caps_flushing return the flush_tid instead.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
We actually need the ci->i_ceph_lock here. The necessity of the s_mutex
is less clear. Also add a lockdep assertion for the i_ceph_lock.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
It's only used to keep count of caps being trimmed, but that requires
that we hold the session->s_mutex to prevent multiple trimming
operations from running concurrently.
We can achieve the same effect using an integer on the stack, which
allows us to (eventually) not need the s_mutex.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
It's protected by the s_gen_ttl_lock, so we should fetch under it
and ensure that we're using the same generation in both places.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
We already mark the mapping in that case, and doing this can cause
false positives to occur at fsync time, as well as spurious read
errors.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Most filesystems that provide virtual xattrs (e.g. CIFS) don't display
them via listxattr(). Ceph does, and that causes some of the tests in
xfstests to fail.
Have cephfs stop listing vxattrs in listxattr. Userspace can always
query them directly when the name is known.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Acked-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
There is no reason to prevent this. The OSD should be able to handle
this as long as the objects are different, and the existing code falls
back when the offset into the object is different.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Acked-by: Luis Henriques <lhenriques@suse.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
When ceph_mdsc_do_request returns an error, we can't assume that the
filelock_reply pointer will be set. Only try to fetch fields out of
the r_reply_info when it returns success.
Cc: stable@vger.kernel.org
Reported-by: Hector Martin <hector@marcansoft.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Since 9e8925b67a ("locks: Allow disabling mandatory locking at compile
time"), attempts to mount filesystems with "-o mand" will fail.
Unfortunately, there is no other indiciation of the reason for the
failure.
Change how the function is defined for better readability. When
CONFIG_MANDATORY_FILE_LOCKING is disabled, printk a warning when
someone attempts to mount with -o mand.
Also, add a blurb to the mandatory-locking.txt file to explain about
the "mand" option, and the behavior one should expect when it is
disabled.
Reported-by: Jan Kara <jack@suse.cz>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Now that check_conflicting_open uses inode->i_readcount instead of
the dentry->d_count to detect opens for read, revise the tracepoint
to display that value instead.
Also, fl is never NULL, so no need to check for that in the fast
assign section.
Cc: Amir Goldstein <amir73il@gmail.com>
Reported-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Zheng wants to be able to spend more time working on the MDS, so I've
volunteered to take over for him as the CephFS kernel client maintainer.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Acked-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
The convention with xattrs is to not store the termination with string
data, given that it returns the length. This is how setfattr/getfattr
operate.
Most of ceph's virtual xattr routines use snprintf to plop the string
directly into the destination buffer, but snprintf always NULL
terminates the string. This means that if we send the kernel a buffer
that is the exact length needed to hold the string, it'll end up
truncated.
Add a ceph_fmt_xattr helper function to format the string into an
on-stack buffer that should always be large enough to hold the whole
thing and then memcpy the result into the destination buffer. If it does
turn out that the formatted string won't fit in the on-stack buffer,
then return -E2BIG and do a WARN_ONCE().
Change over most of the virtual xattr routines to use the new helper. A
couple of the xattrs are sourced from strings however, and it's
difficult to know how long they'll be. Just have those memcpy the result
in place after verifying the length.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Acked-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
The getxattr manpage states that we should return ERANGE if the
destination buffer size is too small to hold the value.
ceph_vxattrcb_layout does this internally, but we should be doing
this for all vxattrs.
Fix the only caller of getxattr_cb to check the returned size
against the buffer length and return -ERANGE if it doesn't fit.
Drop the same check in ceph_vxattrcb_layout and just rely on the
caller to handle it.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Acked-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
The getxattr_cb functions return size_t, which is unsigned and then
cast that value to int and then ssize_t before returning it. While all
of this works, it relies on implicit casting rules for signed/unsigned
conversions.
Change getxattr_cb to return ssize_t to better conform with what the
caller actually wants. Also, remove some suspicious casts.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Acked-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
We don't set SB_I_VERSION on ceph since we need to manage it ourselves,
so we must increment it whenever we update the file times.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Under ceph, clients can be independently updating iversion themselves,
while working under comprehensive sets of caps on an inode. In that
situation we always want to prefer the largest value of a change
attribute. Add a new function that will update a raw value with a larger
one, but otherwise leave it alone.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Now that the client can handle either address formatting, advertise to
the peer that we can support it.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
...ditto for the decode function. We only use these functions to fix
up banner addresses now, so let's name them more appropriately.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Going forward, we'll have different address types so let's use
the addr2 TYPE_LEGACY for internal tracking rather than TYPE_NONE.
Also, make ceph_pr_addr print the address type value as well.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Given the new format, we have to decode the addresses twice. Once to
skip past the new_up_client field, and a second time to collect the
addresses.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
While we're in there, let's also fix up the decoder to do proper
bounds checking.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Switch the MonMap decoder to use the new decoding routine for
entity_addr_t's.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Add a function for decoding an entity_addr_t. Once
CEPH_FEATURE_MSG_ADDR2 is enabled, the server daemons will start
encoding entity_addr_t differently.
Add a new helper function that can handle either format.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
When ceph_mdsc_build_path is handed a positive dentry, it will return a
zero-length path string with the base set to that dentry. This is not
what we want. Always include at least one path component in the string.
ceph_mdsc_build_path has behaved this way for a long time but it didn't
matter until recent d_name handling rework.
Fixes: 964fff7491 ("ceph: use ceph_mdsc_build_path instead of clone_dentry_name")
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Originally, filemap_write_and_wait took the i_mutex internally, but
commit 02c24a8218 pushed the mutex acquisition into the individual
fsync routines, leaving it up to the subsystem maintainers to remove
it if it wasn't needed.
For cifs, I see no reason to take the inode_lock here. All of the
operations inside that lock are protected in other ways.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Acked-by: Pavel Shilovsky <pshilov@microsoft.com>
xfstest generic/452 was triggering a "Busy inodes after umount" warning.
ceph was allowing the mount to go read-only without first flushing out
dirty inodes in the cache. Ensure we sync out the filesystem before
allowing a remount to proceed.
Cc: stable@vger.kernel.org
Link: http://tracker.ceph.com/issues/39571
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
GCC9 is throwing a lot of warnings about unaligned accesses by
callers of ceph_pr_addr. All of the current callers are passing a
pointer to the sockaddr inside struct ceph_entity_addr.
Fix it to take a pointer to a struct ceph_entity_addr instead,
and then have the function make a copy of the sockaddr before
printing it.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
GCC9 is throwing a lot of warnings about unaligned access. This patch
fixes some of them by changing most of the sockaddr handling functions
to take a pointer to struct ceph_entity_addr instead of struct
sockaddr_storage. The lower functions can then make copies or do
unaligned accesses as needed.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
I originally thought there was a potential race here, but the fact
that this is called with the mdsc->mutex held, ensures that the
last reference to the session can't be put here.
Still, it's clearer to just return the value from get_session here,
and may prevent a bug later if we ever rework this code to be less
reliant on mutexes.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
The return of this function is rather complex. It can return 0 or 1,
and in the case of a 1 return, the "err" pointer will be filled out.
This necessitates a lot of copying of values.
We can achieve the same effect by just returning 0, 1 or a negative
error code, and drop the "err" argument from this function.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
It's not clear what AUTH_RDCACHE means in this context, and we're
clearly just dropping LINK caps here.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Nothing calls ceph_mdsc_submit_request today, but in later patches we'll
need to be able to call this separately.
Have the helper return an int so we can check the r_err under the mutex,
and have the caller just check the error code from the submit. Also move
the acquisition of CEPH_CAP_PIN references into the same function.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
No MDS requests use r_callback today, but that will change in the
future. The OSD client always does r_callback and then completes
r_completion. Let's have the MDS client do the same.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
We make copies of the dentry name in set_request_path_attr, but then
create_request_message re-fetches the lengths out of the dentry. While
we don't currently set the *_drop fields unless the parents are locked,
it's still better not to rely on that sort of implicit assumption.
Use the pathlen values that set_request_path_attr returned instead, as
they will always be correct for the returned paths themselves.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Al suggested we get rid of the kmalloc here and just use __getname
and __putname to get a full PATH_MAX pathname buffer.
Since we build the path in reverse, we continue to return a pointer
to the beginning of the string and the length, and add a new helper
to free the thing at the end.
Suggested-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
While it may be slightly more efficient, it's probably not worthwhile to
optimize for the case that clone_dentry_name handles. We can get the
same result by just calling ceph_mdsc_build_path when the parent isn't
locked, with less code duplication.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
temp is not defined outside of the RCU critical section here. Ensure
we grab that value before we drop the rcu_read_lock.
Reported-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
We have a "caps" file already that gives statistics on the caps
cache as a whole. Add another section to that output and dump a
line for each individual cap record.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
cephfs can benefit from statx. We can have the client just request caps
sufficient for the needed attributes and leave off the rest.
Also, recognize when AT_STATX_DONT_SYNC is set, and just scrape the
inode without doing any call in that case. Force a call to the MDS in
the event that AT_STATX_FORCE_SYNC is set.
Link: http://tracker.ceph.com/issues/39258
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Reviewed-by: David Howells <dhowells@redhat.com>
Reviewed-by: Sage Weil <sage@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Originally, filemap_write_and_wait took the i_mutex internally, but
commit 02c24a8218 pushed the mutex acquisition into the individual
fsync routines, leaving it up to the subsystem maintainers to remove
it if it wasn't needed.
For ceph, I see no reason to take the inode_lock here. All of the
operations inside that lock are protected by their own locking.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
It's possible for us to issue a lookup to revalidate a dentry
concurrently with a rename. If done in the right order, then we could
end up processing dentry info in the reply that no longer reflects the
state of the dentry.
If req->r_dentry->d_name differs from the one in the trace, then just
ignore the trace in the reply. We only need to do this however if the
parent's i_rwsem is not held.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Ben reported tripping the BUG_ON in create_request_message during some
performance testing. Analysis of the vmcore showed that the length of
the r_dentry->d_name string changed after we allocated the buffer, but
before we encoded it.
build_dentry_path returns pointers to d_name in the common case of
non-snapped dentries, but this optimization isn't safe unless the parent
directory is locked. When it isn't, have the code make a copy of the
d_name while holding the d_lock.
Cc: stable@vger.kernel.org
Reported-by: Ben England <bengland@redhat.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
When a blocked NFS lock is "awoken" we send a callback to the server and
then wake any hosts waiting on it. If a client attempts to get a lock
and then drops off the net, we could end up waiting for a long time
until we end up waking locks blocked on that request.
So, wake any other waiting lock requests before sending the callback.
Do this by calling locks_delete_block in a new "prepare" phase for
CB_NOTIFY_LOCK callbacks.
URL: https://bugzilla.kernel.org/show_bug.cgi?id=203363
Fixes: 16306a61d3 ("fs/locks: always delete_block after waiting.")
Reported-by: Slawomir Pryczek <slawek1211@gmail.com>
Cc: Neil Brown <neilb@suse.com>
Cc: stable@vger.kernel.org
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
After a blocked nfsd file_lock request is deleted, knfsd will send a
callback to the client and then free the request. Commit 16306a61d3
("fs/locks: always delete_block after waiting.") changed it such that
locks_delete_block is always called on a request after it is awoken,
but that patch missed fixing up blocked nfsd request handling.
Call locks_delete_block on the block to wake up any locks still blocked
on the nfsd lock request before freeing it. Some of its callers already
do this however, so just remove those calls.
URL: https://bugzilla.kernel.org/show_bug.cgi?id=203363
Fixes: 16306a61d3 ("fs/locks: always delete_block after waiting.")
Reported-by: Slawomir Pryczek <slawek1211@gmail.com>
Cc: Neil Brown <neilb@suse.com>
Cc: stable@vger.kernel.org
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Andreas reported that he was seeing the tdbtorture test fail in some
cases with -EDEADLCK when it wasn't before. Some debugging showed that
deadlock detection was sometimes discovering the caller's lock request
itself in a dependency chain.
While we remove the request from the blocked_lock_hash prior to
reattempting to acquire it, any locks that are blocked on that request
will still be present in the hash and will still have their fl_blocker
pointer set to the current request.
This causes posix_locks_deadlock to find a deadlock dependency chain
when it shouldn't, as a lock request cannot block itself.
We are going to end up waking all of those blocked locks anyway when we
go to reinsert the request back into the blocked_lock_hash, so just do
it prior to checking for deadlocks. This ensures that any lock blocked
on the current request will no longer be part of any blocked request
chain.
URL: https://bugzilla.kernel.org/show_bug.cgi?id=202975
Fixes: 5946c4319e ("fs/locks: allow a lock request to block other requests.")
Cc: stable@vger.kernel.org
Reported-by: Andreas Schneider <asn@redhat.com>
Signed-off-by: Neil Brown <neilb@suse.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
The spinlock handling in this file has changed significantly since this
comment was written, and the file_lock_lock is no more. In addition,
this overall comment no longer applies. Deleting an entry now requires
both locks.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
When reclaiming a delegation via CLAIM_PREVIOUS open, the server can
indicate that the delegation has been recalled since it was issued by
setting the "recalled" flag in the delegation.
Ensure that we respect the flag by initiating a delegation return when
it is set.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
When a loop block device encounters a writeback error, that error will
get propagated to the bd_inode's wb_err field. If we then detach the
backing file from it, attach another and fsync it, we'll get back the
writeback error that we had from the previous backing file.
This is a bit of a grey area as POSIX doesn't cover loop devices, but it
is somewhat counterintuitive.
If we detach a backing file from the loopdev while there are still
unreported errors, take it as a sign that we're no longer interested in
the previous file, and clear out the wb_err in the loop blockdev.
Reported-and-Tested-by: Theodore Y. Ts'o <tytso@mit.edu>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
If we have a RECLAIM_COMPLETE with a populated cl_lock_waitq, then
that implies that a reconnect has occurred. Since we can't expect a
CB_NOTIFY_LOCK callback at that point, just wake up the entire queue
so that all the tasks can re-poll for their locks.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
The task is expected to sleep for a while here, and it's possible that
a new EXCHANGE_ID has occurred in the interim, and we were assigned a
new clientid. Since this is a per-client list, there isn't a lot of
value in vetting the clientid on the incoming request.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
We may get a notification and lose the race to another client. Ensure
that we wait again for a notification in that case.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
On x86_64, it's 1152 bytes, so we can avoid wasting 896 bytes each.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
We already send it for v4.1, but RFC7530 also notes that the stateid in
the close reply is bogus.
Always send the special close stateid, even in v4.0 responses. No client
should put any meaning on it whatsoever. For now, we continue to
increment the stateid value, though that might not be necessary either.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
We had some reports of panics in nfsd4_lm_notify, and that showed a
nfs4_lockowner that had outlived its so_client.
Ensure that we walk any leftover lockowners after tearing down all of
the stateids, and remove any blocked locks that they hold.
With this change, we also don't need to walk the nbl_lru on nfsd_net
shutdown, as that will happen naturally when we tear down the clients.
Fixes: 76d348fadf (nfsd: have nfsd4_lock use blocking locks for v4.1+ locks)
Reported-by: Frank Sorenson <fsorenso@redhat.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Cc: stable@vger.kernel.org # 4.9
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
As Linus points out:
The inode_cmp_iversion{+raw}() functions are pure and utter crap.
Why?
You say that they return 0/negative/positive, but they do so in a
completely broken manner. They return that ternary value as the
sequence number difference in a 's64', which means that if you
actually care about that ternary value, and do the *sane* thing that
the kernel-doc of the function implies is the right thing, you would
do
int cmp = inode_cmp_iversion(inode, old);
if (cmp < 0 ...
and as a result you get code that looks sane, but that doesn't
actually *WORK* right.
Since none of the callers actually care about the ternary value here,
convert the inode_cmp_iversion{+raw} functions to just return a boolean
value (false for matching, true for non-matching).
This matches the existing use of these functions just fine, and makes it
simple to convert them to return a ternary value in the future if we
grow callers that need it.
With this change we can also reimplement inode_cmp_iversion in a simpler
way using inode_peek_iversion.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since i_version is mostly treated as an opaque value, we can exploit that
fact to avoid incrementing it when no one is watching. With that change,
we can avoid incrementing the counter on writes, unless someone has
queried for it since it was last incremented. If the a/c/mtime don't
change, and the i_version hasn't changed, then there's no need to dirty
the inode metadata on a write.
Convert the i_version counter to an atomic64_t, and use the lowest order
bit to hold a flag that will tell whether anyone has queried the value
since it was last incremented.
When we go to maybe increment it, we fetch the value and check the flag
bit. If it's clear then we don't need to do anything if the update
isn't being forced.
If we do need to update, then we increment the counter by 2, and clear
the flag bit, and then use a CAS op to swap it into place. If that
works, we return true. If it doesn't then do it again with the value
that we fetch from the CAS operation.
On the query side, if the flag is already set, then we just shift the
value down by 1 bit and return it. Otherwise, we set the flag in our
on-stack value and again use cmpxchg to swap it into place if it hasn't
changed. If it has, then we use the value from the cmpxchg as the new
"old" value and try again.
This method allows us to avoid incrementing the counter on writes (and
dirtying the metadata) under typical workloads. We only need to increment
if it has been queried since it was last changed.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Dave Chinner <dchinner@redhat.com>
Tested-by: Krzysztof Kozlowski <krzk@kernel.org>
At this point, we know that "now" and the file times may differ, and we
suspect that the i_version has been flagged to be bumped. Attempt to
bump the i_version, and only mark the inode dirty if that actually
occurred or if one of the times was updated.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Acked-by: David Sterba <dsterba@suse.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
We only really need to update i_version if someone has queried for it
since we last incremented it. By doing that, we can avoid having to
update the inode if the times haven't changed.
If the times have changed, then we go ahead and forcibly increment the
counter, under the assumption that we'll be going to the storage
anyway, and the increment itself is relatively cheap.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
For NFS, we just use the "raw" API since the i_version is mostly
managed by the server. The exception there is when the client
holds a write delegation, but we only need to bump it once
there anyway to handle CB_GETATTR.
Tested-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Jeff Layton