13 Sep, 2005
4 commits
-
We could try to unlock the state lock here without having first locked it.
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
In the case of a lock which introduces a new lockowner, the openowner's
sequence id should be incremented, even when the operation fails, if the
error is a sequence-id-mutating error. The current code fails to do that
in some cases. Fix this by using the same sequence-id-incrementing
mechanism that all other such operations use.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
It seems more natural to move the setting of the replay_owner into the
relevant procedure instead of doing it in nfsv4_proc_compound.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Demote some printk's that look like they could be triggered by non-buggy
clients to dprintk's. (For example, stale clientid's are normal
occurrences on reboot, and on a server with a lot of clients these messages
could become annoying.)Also remove some redundant dprintk's (e.g. no need for both STALE_CLIENTID
and its callers to do dprintks).Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
08 Sep, 2005
1 commit
-
When registering an RPC cache, cache_register() always sets the owner as the
sunrpc module. However, there are RPC caches owned by other modules. With
the incorrect owner setting, the real owning module can be removed potentially
with an open reference to the cache from userspace.For example, if one were to stop the nfs server and unmount the nfsd
filesystem, the nfsd module could be removed eventhough rpc.idmapd had
references to the idtoname and nametoid caches (i.e.
/proc/net/rpc/nfs4./channel is still open). This resulted in a
system panic on one of our machines when attempting to restart the nfs
services after reloading the nfsd module.The following patch adds a 'struct module *owner' field in struct
cache_detail. The owner is further assigned to the struct proc_dir_entry
in cache_register() so that the module cannot be unloaded while user-space
daemons have an open reference on the associated file under /proc.Signed-off-by: Bruce Allan
Cc: Trond Myklebust
Cc: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
02 Sep, 2005
2 commits
-
Since the patch to add a NULL short-circuit to crypto_free_tfm() went in,
there's no longer any need for callers of that function to check for NULL.
This patch removes the redundant NULL checks and also a few similar checks
for NULL before calls to kfree() that I ran into while doing the
crypto_free_tfm bits.I've succesfuly compile tested this patch, and a kernel with the patch
applied boots and runs just fine.When I posted the patch to LKML (and other lists/people on Cc) it drew the
following comments :J. Bruce Fields commented
"I've no problem with the auth_gss or nfsv4 bits.--b."Sridhar Samudrala said
"sctp change looks fine."Herbert Xu signed off on the patch.
So, I guess this is ready to be dropped into -mm and eventually mainline.
Signed-off-by: Jesper Juhl
Signed-off-by: Herbert Xu
Signed-off-by: David S. Miller -
This patch goes through the current users of the crypto layer and sets
CRYPTO_TFM_REQ_MAY_SLEEP at crypto_alloc_tfm() where all crypto operations
are performed in process context.Signed-off-by: Herbert Xu
Signed-off-by: David S. Miller
18 Aug, 2005
1 commit
-
The nfsd holds the big kernel lock upon exit, when it really shouldn't.
Not to mention that this breaks Ingo's RT patch. This is a trivial fix
to release the lock.Ingo, this patch also works with your kernel, and stops the problem with
nfsd.Note, there's a "goto out;" where "out:" is right above svc_exit_thread.
The point of the goto also holds the kernel_lock, so I don't see any
problem here in releasing it.Signed-off-by: Steven Rostedt
Signed-off-by: Linus Torvalds
13 Jul, 2005
1 commit
-
inotify is intended to correct the deficiencies of dnotify, particularly
its inability to scale and its terrible user interface:* dnotify requires the opening of one fd per each directory
that you intend to watch. This quickly results in too many
open files and pins removable media, preventing unmount.
* dnotify is directory-based. You only learn about changes to
directories. Sure, a change to a file in a directory affects
the directory, but you are then forced to keep a cache of
stat structures.
* dnotify's interface to user-space is awful. Signals?inotify provides a more usable, simple, powerful solution to file change
notification:* inotify's interface is a system call that returns a fd, not SIGIO.
You get a single fd, which is select()-able.
* inotify has an event that says "the filesystem that the item
you were watching is on was unmounted."
* inotify can watch directories or files.Inotify is currently used by Beagle (a desktop search infrastructure),
Gamin (a FAM replacement), and other projects.See Documentation/filesystems/inotify.txt.
Signed-off-by: Robert Love
Cc: John McCutchan
Cc: Christoph Hellwig
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
08 Jul, 2005
19 commits
-
After discussion at the recent NFSv4 bake-a-thon, I realized that my
assumption that NFS4_FH_PERSISTENT required filehandles to persist was a
misreading of the spec. This also fixes an interoperability problem with the
Solaris client.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
We shouldn't be allowing, e.g., write locks on files not open for read. To
enforce this, we add a pointer from the lock stateid back to the open stateid
it came from, so that the check will continue to be correct even after the
open is upgraded or downgraded.Signed-off-by: Andy Adamson
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
As long as we're here, do some miscellaneous cleanup.
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
The handling of close_lru in preprocess_stateid_op was a source of some
confusion here recently. Try to make the logic a little clearer, by renaming
find_openstateowner_id to make its purpose clearer and untangling some
unnecessarily complicated goto's.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
nfs4_preprocess_seqid_op is called by NFSv4 operations that imply an implicit
renewal of the client lease.Signed-off-by: Andy Adamson
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
from RFC 3530:
"Share reservations are established by OPEN operations and by their
nature are mandatory in that when the OPEN denies READ or WRITE
operations, that denial results in such operations being rejected
with error NFS4ERR_LOCKED."(Note that share_denied is really only a legal error for OPEN.)
Signed-off-by: Andy Adamson
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
An OPEN from the same client/open stateowner requires a stateid update because
of the share/deny access update.Signed-off-by: Andy Adamson
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
We're insisting that the lock sequence id field passed in the
open_to_lockowner struct always be zero. This is probably thanks to the
sentence in rfc3530: "The first request issued for any given lock_owner is
issued with a sequence number of zero."But there doesn't seem to be any problem with allowing initial sequence
numbers other than zero. And currently this is causing lock reclaims from the
Linux client to fail.In the spirit of "be liberal in what you accept, conservative in what you
send", we'll relax the check (and patch the Linux client as well).Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Add some comments on the use of so_seqid, in an attempt to avoid some of the
confusion outlined in the previous patch....Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
The sequence number we store in the sequence id is the last one we received
from the client. So on the next operation we'll check that the client gives
us the next higher number.We increment sequence id's at the last moment, in encode, so that we're sure
of knowing the right error return. (The decision to increment the sequence id
depends on the exact error returned.)However on the *first* use of a sequence number, if we set the sequence number
to the one received from the client and then let the increment happen on
encode, we'll be left with a sequence number one to high.For that reason, ENCODE_SEQID_OP_TAIL only increments the sequence id on
*confirmed* stateowners.This creates a problem for open reclaims, which are confirmed on first use.
Therefore the open reclaim code, as a special exception, *decrements* the
sequence id, cancelling out the undesired increment on encode. But this
prevents the sequence id from ever being incremented in the case where
multiple reclaims are sent with the same openowner. Yuch!We could add another exception to the open reclaim code, decrementing the
sequence id only if this is the first use of the open owner.But it's simpler by far to modify the meaning of the op_seqid field: instead
of representing the previous value sent by the client, we take op_seqid, after
encoding, to represent the *next* sequence id that we expect from the client.
This eliminates the need for special-case handling of the first use of a
stateowner.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Yeah, it's trivial, but this drives me up the wall....
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
A misreading of the spec lead us to convert all errors on open and lock
reclaims to RECLAIM_BAD. This causes problems--for example, a reboot within
the grace period could lead to reclaims with stale stateid's, and we'd like to
return STALE errors in those cases.What rfc3530 actually says about RECLAIM_BAD: "The reclaim provided by the
client does not match any of the server's state consistency checks and is
bad." I'm assuming that "state consistency checks" refers to checks for
consistency with the state recorded to stable storage, and that the error
should be reserved for that case.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
A GRACE or NOGRACE response to a lock request should also bump the sequence
id. So we delay the handling of grace period errors till after we've found
the relevant owner.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
The GRACE and NOGRACE errors should bump the sequence id on open. So we delay
the handling of these errors until nfsd4_process_open2, at which point we've
set the open owner, so the encode routine will be able to bump the sequence
id.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
We oops in list_for_each_entry(), because release_stateowner frees something
on the list we're traversing.Signed-off-by: Andy Adamson
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Make sure we don't try to delete client recovery directories multiple times;
fixes some spurious error messages.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Oops, this lookup_one_len needs the i_sem.
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
We need to fsync the recovery directory after writing to it, but we weren't
doing this correctly. (For example, we weren't taking the i_sem when calling
->fsync().)Just reuse the existing nfsd fsync code instead.
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
We need to remove the recovery directory here too. (This chunk just got lost
somehow in the process of commuting the reboot recovery patches past the other
patches.)Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
29 Jun, 2005
1 commit
-
Looks like it sneaked back with the NFS ACL merge..
Signed-off-by: Christoph Hellwig
Cc: Trond Myklebust
Cc: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
24 Jun, 2005
11 commits
-
Set the recovery directory via /proc/fs/nfsd/nfs4recoverydir.
It may be changed any time, but is used only on startup.
Signed-off-by: Andy Adamson
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This patch adds the code to create and remove client subdirectories from the
recovery directory, as described in the previous patch comment.Signed-off-by: Andy Adamson
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
NFSv4 clients are required to know what state they have on the server so that
they can reclaim it on server reboot. However, it is possible for
pathalogical combinations of server reboots and network partitions to leave a
client in a state where it cannot know whether it has lost its state on the
server.For this reason, rfc3530 requires that we store some information about clients
to stable storage.So we maintain a directory /var/lib/nfs/v4recovery with a subdirectory for
each client with active state. We leave open the possibility of including
files underneath each such subdirectory with information about the client, but
for now the subdirectories are empty.We create a client subdirectory whenever a client makes its first non-reclaim
open_confirm.We remove a client subdirectory whenever either
a) its lease expires, or
b) the grace period ends without it reclaiming anything.
When handling reclaims, we allow the reclaim if and only if the client doing
the reclaim has a subdirectory.This patch adds just the code to scan the recovery directory on nfsd startup.
Signed-off-by: Andy Adamson
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
The cb_parsed field is only used by probe_callback, to determine whether the
callback information has been filled in by setclientid. But there is no way
that probe_callback() can be called without that having already happened, so
that check is superfluous, as is cb_parsed.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
>From the language of rfc3530 section 8.1.3 (e.g., the suggestion that a
"process id" might be a reasonable lockowner value) it's conceivable that a
client might want to use the same lockowner string on multiple files, so we may
as well allow that. We expect each use of open_to_lockowner to create a
distinct seqid stream, though.For now we're also allowing multiple uses of open_to_lockowner with the same
open, though it seems unlikely clients would actually do that.Also add a comment reminding myself of some very non-scalable data structures.
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Trivial renaming patch:
I can never remember, while looking at various lists relating the nfsd4 state
structures, which are the "heads" and which are items on other lists, or which
structures are actually on the various lists. The following convention helps
me: given structures foo and bar, with foo containing the head of a list of
bars, use "bars" for the name of the head of the list contained in the struct
foo, and use "per_foo" for the entries in the struct bars.Already done for struct nfs4_file; go ahead and do it for the other nfsd4
state structures.Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Minor cleanup, remove some unnecessary printk's.
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Trivial whitespace and comment fixes.
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Change from "goto" to "else if" format in setclientid_confirm.
From: Fred Isaman
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
NFS4_INVAL is not a valid error for setclientid_confirm, and INUSE is the more
logical error here anyway.From: Fred Isaman
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Setclientid_confirm code confused states 1 and 3 (numbering from the
IMPLEMENTATION section of rfc3530, section 14.2.33). Fix this.State 1 allows the client to change the callback channel on the fly. We don't
implement this currently, so just turn off the callback channel in this case.From: Fred Isaman
Signed-off-by: J. Bruce Fields
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds