I would like to discuss the idea that the current checks for attribute
timeout using time_after are inadequate for 32bit architectures, since
time_after works correctly only when the two timestamps being compared
are within 2^31 jiffies of each other. The signed overflow caused by
comparing values more than 2^31 jiffies apart will flip the result,
causing incorrect assumptions of validity.
2^31 jiffies is a fairly large period of time (~25 days) when compared
to the lifetime of most kernel data structures, but for long lived NFS
mounts that can sit idle for months (think that for some reason autofs
cannot be used), it is easy to compare inode attribute timestamps with
very disparate or even bogus values (as in when jiffies have wrapped
many times, where the comparison doesn't even make sense).
Currently the code tests for attribute timeout by simply adding the
desired amount of jiffies to the stored timestamp and comparing that
with the current timestamp of obtained attribute data with time_after.
This is incorrect, as it returns true for the desired timeout period
and another full 2^31 range of jiffies.
In testing with artificial jumps (several small jumps, not one big
crank) of the jiffies I was able to reproduce a problem found in a
server with very long lived NFS mounts, where attributes would not be
refreshed even after touching files and directories in the server:
Initial uptime:
03:42:01 up 6 min, 0 users, load average: 0.01, 0.12, 0.07
NFS volume is mounted and time is advanced:
03:38:09 up 25 days, 2 min, 0 users, load average: 1.22, 1.05, 1.08
# ls -l /local/A/foo/bar /nfs/A/foo/bar
-rw-r--r-- 1 root root 0 Dec 17 03:38 /local/A/foo/bar
-rw-r--r-- 1 root root 0 Nov 22 00:36 /nfs/A/foo/bar
# touch /local/A/foo/bar
# ls -l /local/A/foo/bar /nfs/A/foo/bar
-rw-r--r-- 1 root root 0 Dec 17 03:47 /local/A/foo/bar
-rw-r--r-- 1 root root 0 Nov 22 00:36 /nfs/A/foo/bar
We can see the local mtime is updated, but the NFS mount still shows
the old value. The patch below makes it work:
Initial setup...
07:11:02 up 25 days, 1 min, 0 users, load average: 0.15, 0.03, 0.04
# ls -l /local/A/foo/bar /nfs/A/foo/bar
-rw-r--r-- 1 root root 0 Jan 11 07:11 /local/A/foo/bar
-rw-r--r-- 1 root root 0 Jan 11 07:11 /nfs/A/foo/bar
# touch /local/A/foo/bar
# ls -l /local/A/foo/bar /nfs/A/foo/bar
-rw-r--r-- 1 root root 0 Jan 11 07:14 /local/A/foo/bar
-rw-r--r-- 1 root root 0 Jan 11 07:14 /nfs/A/foo/bar
Signed-off-by: Fabio Olive Leite <fleite@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Hi.
Attached is a patch to modify the NFS client code to support
64 bit ino's, as appropriate for the system and the NFS
protocol version.
The code basically just expand the NFS interfaces for routines
which handle ino's from using ino_t to u64 and then uses the
fileid in the nfs_inode instead of i_ino in the inode. The
code paths that were updated are in the getattr method and
the readdir methods.
This should be no real change on 64 bit platforms. Since
the ino_t is an unsigned long, it would already be 64 bits
wide.
Thanx...
ps
Signed-off-by: Peter Staubach <staubach@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
This helps prevent huge queues of background writes from building up
whenever the server runs out of disk or quota space, or if someone changes
the file access modes behind our backs.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Schedule writes using WB_SYNC_NONE first, then come back for a second pass
using WB_SYNC_ALL.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The only user of nfs_sync_mapping_range() is nfs_getattr(), which uses it
to flush out the entire inode without sending a commit. We therefore
replace nfs_sync_mapping_range with a more appropriate helper.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Just call write_cache_pages directly instead of hacking the writeback
control structure in order to find out if we were called from writepages()
or directly from the VM.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The addition of nfs_page_mkwrite means that We should no longer need to
create requests inside nfs_writepage()
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
It doesn't look as if the NFS file name limit is being initialised correctly
in the struct nfs_server. Make sure that we limit whatever is being set in
nfs_probe_fsinfo() and nfs_init_server().
Also ensure that readdirplus and nfs4_path_walk respect our file name
limits.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Ryusuke Konishi says:
The recent truncate_complete_page() clears the dirty flag from a page
before calling a_ops->invalidatepage(),
^^^^^^
static void
truncate_complete_page(struct address_space *mapping, struct page *page)
{
...
cancel_dirty_page(page, PAGE_CACHE_SIZE); <--- Inserted here at
kernel 2.6.20
if (PagePrivate(page))
do_invalidatepage(page, 0); ---> will call
a_ops->invalidatepage()
...
}
and this is disturbing nfs_wb_page_priority() from calling
nfs_writepage_locked() that is expected to handle the pending
request (=nfs_page) associated with the page.
int nfs_wb_page_priority(struct inode *inode, struct page *page, int how)
{
...
if (clear_page_dirty_for_io(page)) {
ret = nfs_writepage_locked(page, &wbc);
if (ret < 0)
goto out;
}
...
}
Since truncate_complete_page() will get rid of the page after
a_ops->invalidatepage() returns, the request (=nfs_page) associated
with the page becomes a garbage in nfs_inode->nfs_page_tree.
------------------------
Fix this by ensuring that nfs_wb_page_priority() recognises that it may
also need to clear out non-dirty pages that have an nfs_page associated
with them.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
According to the mount(2) man page, the proper error return code for the
mount(2) system call when the special device name or the mounted-on
directory name is too long is ENAMETOOLONG.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The hostname was getting truncated in the new text-based NFS mount API.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Don't filter the return code from the in-kernel rpcbind or NFS mount
clients. Return the real error code so that callers of the new NFS
text-based mount API can apply a useful retry strategy.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The new text-based NFS mount option parsing logic doesn't recognize any
valid transport protocols due to a silly mistake in the protocol token
matching logic. This prevents basic mount requests such as:
mount.nfs server:/export /mnt -o proto=tcp
from working with the new text-based NFS mount API.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Doh! We can't use cancel_delayed_work_sync because we may have been called
from an unmount that was being performed by nfs_automount_task.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
This avoids the recent NFS mount regression (returning EBUSY when
mounting the same filesystem twice with different parameters).
The best I can do given the constraints appears to be to have the kernel
first look for a superblock that matches both the fsid and the
user-specified mount options, and then spawn off a new superblock if
that search fails.
Note that this is not the same as specifying nosharecache everywhere
since nosharecache will never attempt to match an existing superblock.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Tested-by: Hua Zhong <hzhong@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Do not allow cached open for O_RDONLY or O_WRONLY unless the file has been
previously opened in these modes.
Also Fix the calculation of the mode in nfs4_close_prepare. We should only
issue an OPEN_DOWNGRADE if we're sure that we will still be holding the
correct open modes. This may not be the case if we've been doing delegated
opens.
Finally, there is no need to adjust the open mode bit flags in
nfs4_close_done(): that has already been done in nfs4_close_prepare().
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
We don't really need to clear &state->inode_states inside
nfs4_set_mode_locked, and doing so without holding the inode->i_lock would
in any case be a bug...
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
We need to grab the inode->i_lock atomically with the last reference put in
order to remove the open context that is being freed from the
nfsi->open_files list.
Fix by converting the kref to a standard atomic counter and then using
atomic_dec_and_lock()...
Thanks to Arnd Bergmann for pointing out the problem.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Obviously broken on little-endian; fortunately, the option is not
frequently used...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
[ Hey, sparse is wonderful, but even better than sparse is having people
like Al that actually _run_ it and fix bugs using it. - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Slab destructors were no longer supported after Christoph's
c59def9f22 change. They've been
BUGs for both slab and slub, and slob never supported them
either.
This rips out support for the dtor pointer from kmem_cache_create()
completely and fixes up every single callsite in the kernel (there were
about 224, not including the slab allocator definitions themselves,
or the documentation references).
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
If a NFSv4 mount is attempted with string based options, and the
option string doesn't contain a clientaddr= option, the kernel will
currently oops. Check for this situation and return a proper error.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
status in nfs client callback xdr code is passed in network order.
print it in host order for better readability.
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Fix a couple of bugs:
- Don't rely on the parent dentry still being valid when the call completes.
Fixes a race with shrink_dcache_for_umount_subtree()
- Don't remove the file if the filehandle has been labelled as stale.
Fix a couple of inefficiencies
- Remove the global list of sillyrenamed files. Instead we can cache the
sillyrename information in the dentry->d_fsdata
- Move common code from unlink_setup/unlink_done into fs/nfs/unlink.c
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
We need a common structure for setting up an unlink() rpc call in order to
fix the asynchronous unlink code.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Ensure that opendata->state is always initialised when we do state
recovery.
Ensure that we set the filehandle in the case where we're doing an
"OPEN_CLAIM_PREVIOUS" call due to a server reboot.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Since posix_test_lock(), like fcntl() and ->lock(), indicates absence or
presence of a conflict lock by setting fl_type to, respectively, F_UNLCK
or something other than F_UNLCK, the return value is no longer needed.
Signed-off-by: "J. Bruce Fields" <bfields@citi.umich.edu>
As Peter Staubach says elsewhere
(http://marc.info/?l=linux-kernel&m=118113649526444&w=2):
> The problem is that some file system such as NFSv2 and NFSv3 do
> not have sufficient support to be able to support leases correctly.
> In particular for these two file systems, there is no over the wire
> protocol support.
>
> Currently, these two file systems fail the fcntl(F_SETLEASE) call
> accidentally, due to a reference counting difference. These file
> systems should fail more consciously, with a proper error to
> indicate that the call is invalid for them.
Define an nfs setlease method that just returns -EINVAL.
If someone can demonstrate a real need, perhaps we could reenable
them in the presence of the "nolock" mount option.
Signed-off-by: "J. Bruce Fields" <bfields@citi.umich.edu>
Cc: Peter Staubach <staubach@redhat.com>
Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
Currently, the freezer treats all tasks as freezable, except for the kernel
threads that explicitly set the PF_NOFREEZE flag for themselves. This
approach is problematic, since it requires every kernel thread to either
set PF_NOFREEZE explicitly, or call try_to_freeze(), even if it doesn't
care for the freezing of tasks at all.
It seems better to only require the kernel threads that want to or need to
be frozen to use some freezer-related code and to remove any
freezer-related code from the other (nonfreezable) kernel threads, which is
done in this patch.
The patch causes all kernel threads to be nonfreezable by default (ie. to
have PF_NOFREEZE set by default) and introduces the set_freezable()
function that should be called by the freezable kernel threads in order to
unset PF_NOFREEZE. It also makes all of the currently freezable kernel
threads call set_freezable(), so it shouldn't cause any (intentional)
change of behaviour to appear. Additionally, it updates documentation to
describe the freezing of tasks more accurately.
[akpm@linux-foundation.org: build fixes]
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Nigel Cunningham <nigel@nigel.suspend2.net>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
I can never remember what the function to register to receive VM pressure
is called. I have to trace down from __alloc_pages() to find it.
It's called "set_shrinker()", and it needs Your Help.
1) Don't hide struct shrinker. It contains no magic.
2) Don't allocate "struct shrinker". It's not helpful.
3) Call them "register_shrinker" and "unregister_shrinker".
4) Call the function "shrink" not "shrinker".
5) Reduce the 17 lines of waffly comments to 13, but document it properly.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: David Chinner <dgc@sgi.com>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This includes /proc/fs/nfsfs/servers and /proc/fs/nfsfs/volumes entries.
Both need to show the header and use the list_head.
Signed-off-by: Pavel Emelianov <xemul@openvz.org>
Acked-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
I ran into a curious issue when a lock is being canceled. The
cancellation results in a lock request to the vfs layer instead of an
unlock request. This is particularly insidious when the process that
owns the lock is exiting. In that case, sometimes the erroneous lock is
applied AFTER the process has entered zombie state, preventing the lock
from ever being released. Eventually other processes block on the lock
causing a slow degredation of the system. In the 2.6.16 kernel this was
investigated on, the problem is compounded by the fact that the cl_sem
is held while blocking on the vfs lock, which results in most processes
accessing the nfs file system in question hanging.
In more detail, here is how the situation occurs:
first _nfs4_do_setlk():
static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int reclaim)
...
ret = nfs4_wait_for_completion_rpc_task(task);
if (ret == 0) {
...
} else
data->cancelled = 1;
then nfs4_lock_release():
static void nfs4_lock_release(void *calldata)
...
if (data->cancelled != 0) {
struct rpc_task *task;
task = nfs4_do_unlck(&data->fl, data->ctx, data->lsp,
data->arg.lock_seqid);
The problem is the same file_lock that was passed in to _nfs4_do_setlk()
gets passed to nfs4_do_unlck() from nfs4_lock_release(). So the type is
still F_RDLCK or FWRLCK, not F_UNLCK. At some point, when cancelling the
lock, the type needs to be changed to F_UNLCK. It seemed easiest to do
that in nfs4_do_unlck(), but it could be done in nfs4_lock_release().
The concern I had with doing it there was if something still needed the
original file_lock, though it turns out the original file_lock still
needs to be modified by nfs4_do_unlck() because nfs4_do_unlck() uses the
original file_lock to pass to the vfs layer, and a copy of the original
file_lock for the RPC request.
It seems like the simplest solution is to force all situations where
nfs4_do_unlck() is being used to result in an unlock, so with that in
mind, I made the following change:
Signed-off-by: Frank Filz <ffilzlnx@us.ibm.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Consider the case where the user has mounted the remote filesystem
server:/foo on the two local directories /bar and /baz using the
nosharedcache mount option. The files /bar/file and /baz/file are
represented by different inodes in the local namespace, but refer to the
same file /foo/file on the server.
Consider the case where a process opens both /bar/file and /baz/file, then
closes /bar/file: because the nfs4_state is not shared between /bar/file
and /baz/file, the kernel will see that the nfs4_state for /bar/file is no
longer referenced, so it will send off a CLOSE rpc call. Unless the
open_owners differ, then that CLOSE call will invalidate the open state on
/baz/file too.
Conclusion: we cannot share open state owners between two different
non-shared mount instances of the same filesystem.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Unless the user sets the NFS_MOUNT_NOSHAREDCACHE mount flag, we should
return EBUSY if the filesystem is already mounted on a superblock that
has set conflicting mount options.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Prior to David Howell's mount changes in 2.6.18, users who mounted
different directories which happened to be from the same filesystem on the
server would get different super blocks, and hence could choose different
mount options. As long as there were no hard linked files that crossed from
one subtree to another, this was quite safe.
Post the changes, if the two directories are on the same filesystem (have
the same 'fsid'), they will share the same super block, and hence the same
mount options.
Add a flag to allow users to elect not to share the NFS super block with
another mount point, even if the fsids are the same. This will allow
users to set different mount options for the two different super blocks, as
was previously possible. It is still up to the user to ensure that there
are no cache coherency issues when doing this, however the default
behaviour will be to share super blocks whenever two paths result in
the same fsid.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Hook in final components required for supporting in-kernel mount option
parsing for NFSv2 and NFSv3 mounts.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
For NFSv2 and v3 mounts, the first step is to contact the server's MOUNTD
and request the file handle for the root of the mounted share. Add a
function to the NFS client that handles this operation.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Fabio Olive Leite