- 28 Nov, 2012 2 commits
-
-
Stanislav Kinsbursky authored
Protection of __nfs4_state_shutdown() with nfs4_lock_state() looks redundant. This function is called by the last NFSd thread on it's exit and state lock protects actually two functions (del_recall_lru is protected by recall_lock): 1) nfsd4_client_tracking_exit 2) __nfs4_state_shutdown_net "nfsd4_client_tracking_exit" doesn't require state lock protection, because it's state can be modified only by tracker callbacks. Here a re they: 1) create: is called only from nfsd4_proc_compound. 2) remove: is called from either nfsd4_proc_compound or nfs4_laundromat. 3) check: is called only from nfsd4_proc_compound. 4) grace_done; called only from nfs4_laundromat. nfsd4_proc_compound is called onll by NFSd kthread, which is exiting right now. nfs4_laundromat is called by laundry_wq. But laundromat_work was canceled already. "__nfs4_state_shutdown_net" also doesn't require state lock protection, because all NFSd kthreads are dead, and no race can happen with NFSd start, because "nfsd_up" flag is still set. Moreover, all Nfsd shutdown is protected with global nfsd_mutex. Signed-off-by:
Stanislav Kinsbursky <skinsbursky@parallels.com> Signed-off-by:
J. Bruce Fields <bfields@redhat.com>
-
J. Bruce Fields authored
That function is only called under nfsd_mutex: we know that because the only caller is nfsd_svc, via nfsd_svc nfsd_startup nfs4_state_start nfsd4_client_tracking_init client_tracking_ops->init == nfsd4_load_reboot_recovery_data The shared state accessed here includes: - user_recovery_dirname: used here, modified only by nfs4_reset_recoverydir, which can be verified to only be called under nfsd_mutex. - filesystem state, protected by i_mutex (handwaving slightly here) - rec_file, reclaim_str_hashtbl, reclaim_str_hashtbl_size: other than here, used only from code called from nfsd or laundromat threads, both of which should be started only after this runs (see nfsd_svc) and stopped before this could run again (see nfsd_shutdown, called from nfsd_last_thread). Signed-off-by:
-
- 27 Nov, 2012 1 commit
-
-
J. Bruce Fields authored
The spec requires badname, not inval, in these cases. Some callers want us to return enoent, but I can see no justification for that. Signed-off-by:
-
- 26 Nov, 2012 7 commits
-
-
J. Bruce Fields authored
Linus has pointed out that indiscriminate use of BUG's can make it harder to diagnose bugs because they can bring a machine down, often before we manage to get any useful debugging information to the logs. (Consider, for example, a BUG() that fires in a workqueue, or while holding a spinlock). Most of these BUG's won't do much more than kill an nfsd thread, but it would still probably be safer to get out the warning without dying. There's still more of this to do in nfsd/. Signed-off-by: -
J. Bruce Fields authored
Our server rejects compounds containing more than one write operation. It's unclear whether this is really permitted by the spec; with 4.0, it's possibly OK, with 4.1 (which has clearer limits on compound parameters), it's probably not OK. No client that we're aware of has ever done this, but in theory it could be useful. The source of the limitation: we need an array of iovecs to pass to the write operation. In the worst case that array of iovecs could have hundreds of elements (the maximum rwsize divided by the page size), so it's too big to put on the stack, or in each compound op. So we instead keep a single such array in the compound argument. We fill in that array at the time we decode the xdr operation. But we decode every op in the compound before executing any of them. So once we've used that array we can't decode another write. If we instead delay filling in that array till the time we actually perform the write, we can reuse it. Another option might be to switch to decoding compound ops one at a time. I considered doing that, but it has a number of other side effects, and I'd rather fix just this one problem for now. Signed-off-by: -
J. Bruce Fields authored
In preparation for moving some of this elsewhere. Signed-off-by: -
J. Bruce Fields authored
In preparation for moving some of it elsewhere. Signed-off-by: -
J. Bruce Fields authored
The comment here is totally bogus: - OP_WRITE + 1 is RELEASE_LOCKOWNER. Maybe there was some older version of the spec in which that served as a sort of OP_ILLEGAL? No idea, but it's clearly wrong now. - In any case, I can't see that the spec says anything about what to do if the client sends us less ops than promised. It's clearly nutty client behavior, and we should do whatever's easiest: returning an xdr error (even though it won't be consistent with the error on the last op returned) seems fine to me. Signed-off-by: -
J. Bruce Fields authored
Signed-off-by: -
J. Bruce Fields authored
Very embarassing: 1091006c "nfsd: turn on reply cache for NFSv4" missed a line, effectively leaving the reply cache off in the v4 case. I thought I'd tested that, but I guess not. This time, wrote a pynfs test to confirm it works. Cc: stable@vger.kernel.org Signed-off-by:
-
- 15 Nov, 2012 16 commits
-
-
Stanislav Kinsbursky authored
This patch moves laundromat_work to nfsd per-net context, thus allowing to run multiple laundries. Signed-off-by:
-
Stanislav Kinsbursky authored
Passing net context looks as overkill. Signed-off-by:
-
Stanislav Kinsbursky authored
This patch replaces init_net by SVC_NET(), where possible and also passes proper context to nested functions where required. Signed-off-by:
-
Stanislav Kinsbursky authored
This list holds nfs4 clients (open) stateowner queue for last close replay, which are network namespace aware. So let's make this list per network namespace too. Signed-off-by:
-
Stanislav Kinsbursky authored
This list holds nfs4 clients queue for lease renewal, which are network namespace aware. So let's make this list per network namespace too. Signed-off-by:
-
Stanislav Kinsbursky authored
This hash holds established sessions state and closely associated with nfs4_clients info, which are network namespace aware. So let's make it allocated per network namespace too. Note: this hash can be allocated in per-net operations. But it looks better to allocate it on nfsd state start and thus don't waste resources if server is not running. Signed-off-by:
-
Stanislav Kinsbursky authored
This hash holds file lock owners and closely associated with nfs4_clients info, which are network namespace aware. So let's make it allocated per network namespace too. Note: this hash can be allocated in per-net operations. But it looks better to allocate it on nfsd state start and thus don't waste resources if server is not running. Signed-off-by:
-
Stanislav Kinsbursky authored
This hash holds open owner state and closely associated with nfs4_clients info, which are network namespace aware. So let's make it allocated per network namespace too. Note: this hash can be allocated in per-net operations. But it looks better to allocate it on nfsd state start and thus don't waste resources if server is not running. Signed-off-by:
-
Stanislav Kinsbursky authored
This hash holds nfs4_clients info, which are network namespace aware. So let's make it allocated per network namespace. Signed-off-by:
-
Stanislav Kinsbursky authored
This hash holds nfs4_clients info, which are network namespace aware. So let's make it allocated per network namespace. Note: this hash can be allocated in per-net operations. But it looks better to allocate it on nfsd state start and thus don't waste resources if server is not running. Signed-off-by:
-
Stanislav Kinsbursky authored
This tree holds nfs4_clients info, which are network namespace aware. So let's make it per network namespace. Signed-off-by:
-
Stanislav Kinsbursky authored
This hash holds nfs4_clients info, which are network namespace aware. So let's make it allocated per network namespace. Note: this hash can be allocated in per-net operations. But it looks better to allocate it on nfsd state start and thus don't waste resources if server is not running. Signed-off-by:
-
Stanislav Kinsbursky authored
This hash holds nfs4_clients info, which are network namespace aware. So let's make it allocated per network namespace. Note: this hash is used only by legacy tracker. So let's allocate hash in tracker init. Signed-off-by:
-
Stanislav Kinsbursky authored
And use it's net where possible. Signed-off-by:
-
Stanislav Kinsbursky authored
Signed-off-by:
-
J. Bruce Fields authored
Commit bbf43dc8 "sunrpc/cache.h: replace simple_strtoul" introduced new range-checking which could cause get_int to fail on unsigned integers too large to be represented as an int. We could parse them as unsigned instead--but it turns out svcgssd is actually passing down "-1" in some cases. Which is perhaps stupid, but there's nothing we can do about it now. So just revert back to the previous "sloppy" behavior that accepts either representation. Cc: stable@vger.kernel.org Reported-by:
Sven Geggus <lists@fuchsschwanzdomain.de> Signed-off-by:
-
- 14 Nov, 2012 2 commits
-
-
Fengguang Wu authored
Signed-off-by: -
Fengguang Wu authored
Signed-off-by:
-
- 12 Nov, 2012 11 commits
-
-
Jeff Layton authored
The current code holds on to this list until nfsd is shut down, but it's never touched once the grace period ends. Release that memory back into the wild when the grace period ends. Signed-off-by:
Jeff Layton <jlayton@redhat.com> Signed-off-by:
-
Jeff Layton authored
Remove the cl_recdir field from the nfs4_client struct. Instead, just compute it on the fly when and if it's needed, which is now only when the legacy client tracking code is in effect. The error handling in the legacy client tracker is also changed to handle the case where md5 is unavailable. In that case, we'll warn the admin with a KERN_ERR message and disable the client tracking. Signed-off-by:
-
Jeff Layton authored
The current code requires that we md5 hash the name in order to store the client in the confirmed and unconfirmed trees. Change it instead to store the clients in a pair of rbtrees, and simply compare the cl_names directly instead of hashing them. This also necessitates that we add a new flag to the clp->cl_flags field to indicate which tree the client is currently in. Signed-off-by:
-
Jeff Layton authored
When nfsd starts, the legacy reboot recovery code creates a tracking struct for each directory in the v4recoverydir. When the grace period ends, it basically does a "readdir" on the directory again, and matches each dentry in there to an existing client id to see if it should be removed or not. If the matching client doesn't exist, or hasn't reclaimed its state then it will remove that dentry. This is pretty inefficient since it involves doing a lot of hash-bucket searching. It also means that we have to keep relying on being able to search for a nfs4_client by md5 hashed cl_recdir name. Instead, add a pointer to the nfs4_client that indicates the association between the nfs4_client_reclaim and nfs4_client. When a reclaim operation comes in, we set the pointer to make that association. On gracedone, the legacy client tracker will keep the recdir around iff: 1/ there is a reclaim record for the directory ...and... 2/ there's an association between the reclaim record and a client record -- that is, a create or check operation was performed on the client that matches that directory. Signed-off-by:
-
Jeff Layton authored
Later callers will need to make changes to the record. Signed-off-by:
-
Jeff Layton authored
We'll need to be able to call this from nfs4recover.c eventually. Signed-off-by:
-
Jeff Layton authored
Currently, it takes a client pointer, but later we're going to need to search for these records without knowing whether a matching client even exists. Signed-off-by:
-
Jeff Layton authored
Let's shoot for removing the nfsdcld upcall in 3.10. Most likely, no one is actually using it so I don't expect this warning to fire often (except maybe on misconfigured systems). Signed-off-by:
-
Jeff Layton authored
The usermodehelper upcall program can then decide to use this info as a (one-way) transition mechanism to the new scheme. When a "check" upcall occurs and the client doesn't exist in the database, we can look to see whether the directory exists. If it does, then we'd add the client to the database, remove the legacy recdir, and return success to the kernel to allow the recovery to proceed. For gracedone, we simply pass the v4recovery "topdir" so that the upcall can clean it out prior to returning to the kernel. A module parm is also added to disable the legacy conversion if the admin chooses. Signed-off-by:
-
Jeff Layton authored
First, try to use the new usermodehelper upcall. It should succeed or fail quickly, so there's little cost to doing so. If it fails, and the legacy tracking dir exists, use that. If it doesn't exist then fall back to using nfsdcld. Signed-off-by:
-
Jeff Layton authored
Add a new client tracker upcall type that uses call_usermodehelper to call out to a program. This seems to be the preferred method of calling out to usermode these days for seldom-called upcalls. It's simple and doesn't require a running daemon, so it should "just work" as long as the binary is installed. The client tracking exit operation is also changed to check for a NULL pointer before running. The UMH upcall doesn't need to do anything at module teardown time. Signed-off-by:
-
- 10 Nov, 2012 1 commit
-
-
Jeff Layton authored
Signed-off-by:
-
