Since commit 1a34c8c9a4 ("NFS: Support larger readdir buffers") has
updated dtsize, and with recent improvements to the READDIRPLUS helper
heuristic, the heuristic may not trigger until many dentries are emitted
to userspace. This will cause many thousands of GETATTR calls for "ls
-l" when the directory's pagecache has already been populated. This
manifests as poor performance for long directory listings after an
initially fast "ls -l".
Fix this by emitting only 17 entries for any first pass through the NFS
directory's ->iterate_shared(), which allows userpace to prime the
counters for the heuristic.
Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
POSIX typically only refreshes the user's supplementary group
information upon login. Since NFS servers may often refresh their
concept of the user supplementary group membership at their own cadence,
it is possible for the NFS client's access cache to become stale due to
the user's group membership changing on the server after the user has
already logged in on the client.
While it is reasonable to expect that such group membership changes are
rare, and that we do not want to optimise the cache to accommodate them,
it is also not unreasonable for the user to expect that if they log out
and log back in again, that the staleness would clear up.
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Tested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Fix the following coccicheck warning:
fs/nfs/dir.c:2494:2-7: WARNING:
NULL check before some freeing functions is not needed.
Signed-off-by: Yushan Zhou <katrinzhou@tencent.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
-----BEGIN PGP SIGNATURE-----
iHUEABYIAB0WIQQqUNBr3gm4hGXdBJlZ7Krx/gZQ6wUCYzxj0gAKCRBZ7Krx/gZQ
66/1AQC/KfIAINNOPxozsZaxOaOKo0ouVJ7sJV4ZGsPKpU69gwD/UodJZCtyZ52h
wwkmfzTDjAgGt1QCKj96zk2XFqg4swE=
=u0pv
-----END PGP SIGNATURE-----
Merge tag 'pull-file_inode' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull file_inode() updates from Al Vrio:
"whack-a-mole: cropped up open-coded file_inode() uses..."
* tag 'pull-file_inode' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
orangefs: use ->f_mapping
_nfs42_proc_copy(): use ->f_mapping instead of file_inode()->i_mapping
dma_buf: no need to bother with file_inode()->i_mapping
nfs_finish_open(): don't open-code file_inode()
bprm_fill_uid(): don't open-code file_inode()
sgx: use ->f_mapping...
exfat_iterate(): don't open-code file_inode(file)
ibmvmc: don't open-code file_inode()
nfs_unlink() calls d_delete() twice if it receives ENOENT from the
server - once in nfs_dentry_handle_enoent() from nfs_safe_remove and
once in nfs_dentry_remove_handle_error().
nfs_rmddir() also calls it twice - the nfs_dentry_handle_enoent() call
is direct and inside a region locked with ->rmdir_sem
It is safe to call d_delete() twice if the refcount > 1 as the dentry is
simply unhashed.
If the refcount is 1, the first call sets d_inode to NULL and the second
call crashes.
This patch guards the d_delete() call from nfs_dentry_handle_enoent()
leaving the one under ->remdir_sem in case that is important.
In mainline it would be safe to remove the d_delete() call. However in
older kernels to which this might be backported, that would change the
behaviour of nfs_unlink(). nfs_unlink() used to unhash the dentry which
resulted in nfs_dentry_handle_enoent() not calling d_delete(). So in
older kernels we need the d_delete() in nfs_dentry_remove_handle_error()
when called from nfs_unlink() but not when called from nfs_rmdir().
To make the code work correctly for old and new kernels, and from both
nfs_unlink() and nfs_rmdir(), we protect the d_delete() call with
simple_positive(). This ensures it is never called in a circumstance
where it could crash.
Fixes: 3c59366c20 ("NFS: don't unhash dentry during unlink/rename")
Fixes: 9019fb391d ("NFS: Label the dentry with a verifier in nfs_rmdir() and nfs_unlink()")
Signed-off-by: NeilBrown <neilb@suse.de>
Tested-by: Olga Kornievskaia <aglo@umich.edu>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Add the missing unlock before goto.
Fixes: 3c59366c20 ("NFS: don't unhash dentry during unlink/rename")
Signed-off-by: Sun Ke <sunke32@huawei.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
NFS unlink() (and rename over existing target) must determine if the
file is open, and must perform a "silly rename" instead of an unlink (or
before rename) if it is. Otherwise the client might hold a file open
which has been removed on the server.
Consequently if it determines that the file isn't open, it must block
any subsequent opens until the unlink/rename has been completed on the
server.
This is currently achieved by unhashing the dentry. This forces any
open attempt to the slow-path for lookup which will block on i_rwsem on
the directory until the unlink/rename completes. A future patch will
change the VFS to only get a shared lock on i_rwsem for unlink, so this
will no longer work.
Instead we introduce an explicit interlock. A special value is stored
in dentry->d_fsdata while the unlink/rename is running and
->d_revalidate blocks while that value is present. When ->d_revalidate
unblocks, the dentry will be invalid. This closes the race
without requiring exclusion on i_rwsem.
d_fsdata is already used in two different ways.
1/ an IS_ROOT directory dentry might have a "devname" stored in
d_fsdata. Such a dentry doesn't have a name and so cannot be the
target of unlink or rename. For safety we check if an old devname
is still stored, and remove it if it is.
2/ a dentry with DCACHE_NFSFS_RENAMED set will have a 'struct
nfs_unlinkdata' stored in d_fsdata. While this is set maydelete()
will fail, so an unlink or rename will never proceed on such
a dentry.
Neither of these can be in effect when a dentry is the target of unlink
or rename. So we can expect d_fsdata to be NULL, and store a special
value ((void*)1) which is given the name NFS_FSDATA_BLOCKED to indicate
that any lookup will be blocked.
The d_count() is incremented under d_lock() when a lookup finds the
dentry, so we check d_count() is low, and set NFS_FSDATA_BLOCKED under
the same lock to avoid any races.
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
The use of kmap() is being deprecated in favor of kmap_local_page().
With kmap_local_page(), the mapping is per thread, CPU local and not
globally visible. Furthermore, the mapping can be acquired from any context
(including interrupts).
Therefore, use kmap_local_page() in nfs_do_filldir() because this mapping
is per thread, CPU local, and not globally visible.
Suggested-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
For filesystems that are case insensitive and case preserving, we need
to be able to rename from one case folded variant of the filename to
another.
Currently, if we have looked up the target filename before the call to
rename, then we may have a hashed dentry with that target name in the
dcache, causing the vfs to optimise away the rename.
To avoid that, let's drop the target dentry, and leave it to the server
to optimise away the rename if that is the correct thing to do.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Commit a2ad63daa8 ("VFS: add FMODE_CAN_ODIRECT file flag")
added the FMODE_CAN_ODIRECT flag for NFSv3 but neglected to add
it for NFSv4.x. This causes direct io on NFSv4.x to fail open
with EINVAL:
mount -o vers=4.2 127.0.0.1:/export /mnt/nfs4
dd if=/dev/zero of=/mnt/nfs4/file.bin bs=128k count=1 oflag=direct
dd: failed to open '/mnt/nfs4/file.bin': Invalid argument
dd of=/dev/null if=/mnt/nfs4/file.bin bs=128k count=1 iflag=direct
dd: failed to open '/mnt/dir1/file1.bin': Invalid argument
Fixes: a2ad63daa8 ("VFS: add FMODE_CAN_ODIRECT file flag")
Signed-off-by: Dave Wysochanski <dwysocha@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Add a wrapper that converts back from the folio to the page. This
entire file needs to be converted to use folios, but that's a
task for a different set of patches.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Both xxhash() and hash_64() appear to give similarly low collision
rates with a standard linearly increasing readdir offset. They both give
similarly higher collision rates when applied to ext4's offsets.
So switch to using the standard hash_64().
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
open() with O_ACCMODE|O_DIRECT flags secondly will fail.
Reproducer:
1. mount -t nfs -o vers=4.2 $server_ip:/ /mnt/
2. fd = open("/mnt/file", O_ACCMODE|O_DIRECT|O_CREAT)
3. close(fd)
4. fd = open("/mnt/file", O_ACCMODE|O_DIRECT)
Server nfsd4_decode_share_access() will fail with error nfserr_bad_xdr when
client use incorrect share access mode of 0.
Fix this by using NFS4_SHARE_ACCESS_BOTH share access mode in client,
just like firstly opening.
Fixes: ce4ef7c0a8 ("NFS: Split out NFS v4 file operations")
Signed-off-by: ChenXiaoSong <chenxiaosong2@huawei.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
If the page is empty, we need to check the array->last_cookie instead of
the first entry. Add a helper for the cases where we care.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
In the very rare case where the readdir reply contains multiple cookies
that map to the same hash value, we can end up deadlocking waiting for a
page lock that we already hold. In this case we should fail the page
lock by using grab_cache_page_nowait().
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Even if we're not able to cache all the entries in the readdir buffer,
let's ensure that we do prime the dcache.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Avoid clearing the entire readdir page cache if we're just doing forced
readdirplus for the 'ls -l' heuristic.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Instead of using a linear index to address the pages, use the cookie of
the first entry, since that is what we use to match the page anyway.
This allows us to avoid re-reading the entire cache on a seekdir() type
of operation. The latter is very common when re-exporting NFS, and is a
major performance drain.
The change does affect our duplicate cookie detection, since we can no
longer rely on the page index as a linear offset for detecting whether
we looped backwards. However since we no longer do a linear search
through all the pages on each call to nfs_readdir(), this is less of a
concern than it was previously.
The other downside is that invalidate_mapping_pages() no longer can use
the page index to avoid clearing pages that have been read. A subsequent
patch will restore the functionality this provides to the 'ls -l'
heuristic.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Add tracing to track how often the client goes to the server for updated
readdir information.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
If the revalidation was forced, due to the presence of a LOOKUP_EXCL or
a LOOKUP_REVAL flag, then readdirplus won't help. It also can't help
when we're doing a path component lookup.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
If the filesystem is case insensitive, then readdirplus can't help with
cache misses, since it won't return case folded variants of the filename.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
The heuristic for readdirplus is designed to try to detect 'ls -l' and
similar patterns. It does so by looking for cache hit/miss patterns in
both the attribute cache and in the dcache of the files in a given
directory, and then sets a flag for the readdirplus code to interpret.
The problem with this approach is that a single attribute or dcache miss
can cause the NFS code to force a refresh of the attributes for the
entire set of files contained in the directory.
To be able to make a more nuanced decision, let's sample the number of
hits and misses in the set of open directory descriptors. That allows us
to set thresholds at which we start preferring READDIRPLUS over regular
READDIR, or at which we start to force a re-read of the remaining
readdir cache using READDIRPLUS.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
When reading a very large directory, we want to try to keep the page
cache up to date if doing so is inexpensive. With the change to allow
readdir to continue reading even when the cache is incomplete, we no
longer need to fall back to uncached readdir in order to scale to large
directories.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Recent changes to readdir mean that we can cope with partially filled
page cache entries, so we no longer need to rely on looping in
nfs_readdir_xdr_to_array().
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Ensure that if the cookie verifier changes when we use the zero-valued
cookie, then we invalidate any cached pages.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
The current NFS readdir code will always try to maximise the amount of
readahead it performs on the assumption that we can cache anything that
isn't immediately read by the process.
There are several cases where this assumption breaks down, including
when the 'ls -l' heuristic kicks in to try to force use of readdirplus
as a batch replacement for lookup/getattr.
This patch therefore tries to tone down the amount of readahead we
perform, and adjust it to try to match the amount of data being
requested by user space.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
When we hit the end of the data in the readdir page, we don't want to
start filling a new page, unless this one is full.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
If the page cache entry that was last read gets invalidated for some
reason, then make sure we can re-create it on the next call to readdir.
This, combined with the cache page validation, allows us to reuse the
cached value of page-index on successive calls to nfs_readdir.
Credit is due to Benjamin Coddington for showing that the concept works,
and that it allows for improved cache sharing between processes even in
the case where pages are lost due to LRU or active invalidation.
Suggested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Use the change attribute and the first cookie in a directory page cache
entry to validate that the page is up to date.
Suggested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Instead of relying on counting the page offsets as we walk through the
page cache, switch to calculating them algorithmically.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
For the purpose of ensuring that opendir() followed by seekdir() work as
correctly as possible, try to initialise the readdir verifier in
nfs_opendir().
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
In 4.1+, the server is allowed to set a flag
NFS4_RESULT_PRESERVE_UNLINKED in reply to the OPEN, that tells
the client that it does not need to do a silly rename of an
opened file when it's being removed.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Allow kmemcg to limit the number of open/lock file contexts, in the same
way that it limits the parent file descriptors.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Commit ac795161c9 (NFSv4: Handle case where the lookup of a directory
fails) [1], part of Linux since 5.17-rc2, introduced a regression, where
a symbolic link on an NFS mount to a directory on another NFS does not
resolve(?) the first time it is accessed:
Reported-by: Paul Menzel <pmenzel@molgen.mpg.de>
Fixes: ac795161c9 ("NFSv4: Handle case where the lookup of a directory fails")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Tested-by: Donald Buczek <buczek@molgen.mpg.de>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
If we've reached the end of the directory, then cache that information
in the context so that we don't need to do an uncached readdir in order
to rediscover that fact.
Fixes: 794092c57f ("NFS: Do uncached readdir when we're seeking a cookie in an empty page cache")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
If we're doing an uncached read of the directory, then we ideally want
to read only the exact set of entries that will fit in the buffer
supplied by the getdents() system call. So unlike the case where we're
reading into the page cache, let's send only one READDIR call, before
trying to fill up the buffer.
Fixes: 35df59d3ef ("NFS: Reduce number of RPC calls when doing uncached readdir")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
- New Features:
- Basic handling for case insensitive filesystems
- Initial support for fs_locations and server trunking
- Bugfixes and Cleanups:
- Cleanups to how the "struct cred *" is handled for the nfs_access_entry
- Ensure the server has an up to date ctimes before hardlinking or renaming
- Update 'blocks used' after writeback, fallocate, and clone
- nfs_atomic_open() fixes
- Improvements to sunrpc tracing
- Various null check & indenting related cleanups
- Some improvements to the sunrpc sysfs code
- Use default_groups in kobj_type
- Fix some potential races and reference leaks
- A few tracepoint cleanups in xprtrdma
-----BEGIN PGP SIGNATURE-----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=wcVT
-----END PGP SIGNATURE-----
Merge tag 'nfs-for-5.17-1' of git://git.linux-nfs.org/projects/anna/linux-nfs
Pull NFS client updates from Anna Schumaker:
"New Features:
- Basic handling for case insensitive filesystems
- Initial support for fs_locations and server trunking
Bugfixes and Cleanups:
- Cleanups to how the "struct cred *" is handled for the
nfs_access_entry
- Ensure the server has an up to date ctimes before hardlinking or
renaming
- Update 'blocks used' after writeback, fallocate, and clone
- nfs_atomic_open() fixes
- Improvements to sunrpc tracing
- Various null check & indenting related cleanups
- Some improvements to the sunrpc sysfs code:
- Use default_groups in kobj_type
- Fix some potential races and reference leaks
- A few tracepoint cleanups in xprtrdma"
[ This should have gone in during the merge window, but didn't. The
original pull request - sent during the merge window - had gotten
marked as spam and discarded due missing DKIM headers in the email
from Anna. - Linus ]
* tag 'nfs-for-5.17-1' of git://git.linux-nfs.org/projects/anna/linux-nfs: (35 commits)
SUNRPC: Don't dereference xprt->snd_task if it's a cookie
xprtrdma: Remove definitions of RPCDBG_FACILITY
xprtrdma: Remove final dprintk call sites from xprtrdma
sunrpc: Fix potential race conditions in rpc_sysfs_xprt_state_change()
net/sunrpc: fix reference count leaks in rpc_sysfs_xprt_state_change
NFSv4.1 test and add 4.1 trunking transport
SUNRPC allow for unspecified transport time in rpc_clnt_add_xprt
NFSv4 handle port presence in fs_location server string
NFSv4 expose nfs_parse_server_name function
NFSv4.1 query for fs_location attr on a new file system
NFSv4 store server support for fs_location attribute
NFSv4 remove zero number of fs_locations entries error check
NFSv4: nfs_atomic_open() can race when looking up a non-regular file
NFSv4: Handle case where the lookup of a directory fails
NFSv42: Fallocate and clone should also request 'blocks used'
NFSv4: Allow writebacks to request 'blocks used'
SUNRPC: use default_groups in kobj_type
NFS: use default_groups in kobj_type
NFS: Fix the verifier for case sensitive filesystem in nfs_atomic_open()
NFS: Add a helper to remove case-insensitive aliases
...
If the file type changes back to being a regular file on the server
between the failed OPEN and our LOOKUP, then we need to re-run the OPEN.
Fixes: 0dd2b474d0 ("nfs: implement i_op->atomic_open()")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
If the application sets the O_DIRECTORY flag, and tries to open a
regular file, nfs_atomic_open() will punt to doing a regular lookup.
If the server then returns a regular file, we will happily return a
file descriptor with uninitialised open state.
The fix is to return the expected ENOTDIR error in these cases.
Reported-by: Lyu Tao <tao.lyu@epfl.ch>
Fixes: 0dd2b474d0 ("nfs: implement i_op->atomic_open()")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
When dealing with case insensitive names, the client has no idea how the
server performs the mapping, so cannot collapse the dentries into a
single representative. So both rename and unlink need to deal with the
fact that there could be several dentries representing the file, and
have to somehow force them to be revalidated. Use d_prune_aliases() as a
big hammer approach.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
