- 10 Mar, 2005 40 commits
-
-
Neil Brown authored
If the client was given a delegation, but our callback path to the client has since failed for some reason, then our only chance to inform the client of this is to return the cb_path_down error next time the client attempts to renew its state. Signed-off-by:
Andy Adamson <andros@citi.umich.edu> Signed-off-by:
J. Bruce Fields <bfields@citi.umich.edu> Signed-off-by:
Neil Brown <neilb@cse.unsw.edu.au> Signed-off-by:
Andrew Morton <akpm@osdl.org> Signed-off-by:
Linus Torvalds <torvalds@osdl.org>
-
Neil Brown authored
Don't bother searching for unconfirmed clients when handling RENEW; a client is pretty sick if it's trying to renew state it hasn't even finished acquiring yet. Signed-off-by:
-
Neil Brown authored
Simplify verify_clientid out of existance. Signed-off-by:
-
Neil Brown authored
We repeat the same hash table searches in a bunch of places, and keep making the mistake of assuming the variable iterating over the list will be NULL when the search fails. Some helper functions here simplify things a bit. While we're at it, make move_to_confirmed take just a clp instead of making the caller compute the hash. This means we sometimes have to compute the hash multiple times, but it's only an &, so no big deal. Signed-off-by:
-
Neil Brown authored
Inline nfs4_check_delegmode, replace & by &&, remove incorrect check. (Note that it's fine to read using the stateid for a write delegation.) While we're at it, the status return logic seems like overkill for one simple check. Signed-off-by:
-
Neil Brown authored
We would have oopsed some time ago if either of these parameters were NULL. Signed-off-by:
-
Neil Brown authored
Don't hand out a delegation to a client that already has one. Minor simplification and reformatting. Signed-off-by:
-
Neil Brown authored
Simplify logic in nfs4_open_delegation, fix a small bug on allocation failure. Signed-off-by:
-
Neil Brown authored
The DONT_DELEGATE flag is superfluous. Signed-off-by:
-
Neil Brown authored
Turn off rpc layer timeout replay for the NFSv4 server callback client. Rely on manual replay in nfs4_cb_recall_done instead. Signed-off-by:
-
Neil Brown authored
Only release a struct nfs4_file if there are no associated delegations. Signed-off-by:
-
Neil Brown authored
More minor logic cleanup. Signed-off-by:
-
Neil Brown authored
Replace sequence of if {... goto} 's by if-elseif's, other minor simplification of logic. Signed-off-by: -
Neil Brown authored
Minor trivial cleanup, mostly whitespace. Signed-off-by:
-
Neil Brown authored
Use the struct file * obtained in nfsd4_open for nfsd_read and nfsd_write when available. To do this we add a struct file * argument to nfsd_read and nfsd_write. If a struct file is passed in, nfsd_read and nfsd_write will use it, doing just an access check instead of an open and close. If the new argument is NULL, they will fall back on the old behaviour. Signed-off-by:
-
Neil Brown authored
Allow READ and WRITE during the grace period on OPEN or LOCK state that has been reclaimed, for files for which mandatory locking is not in force. (Reads and writes to files with mandatory locking must be delayed until we're sure all the other clients have had a chance to recover locks.) Signed-off-by:
-
Neil Brown authored
Move special stateid processing to nfs4_preprocess_stateid_op(). Also make stateid processing for setattr the same as that for write. Signed-off-by:
-
Neil Brown authored
Remove unnecessary fields in the nfs4_client_reclaim structure. Initialize reclaim_str_hashtbl_size and increment it inside of nfs4_client_to_reclaim. Modify client_to_reclaim() in preparation for reboot recovery. Signed-off-by:
-
Neil Brown authored
Properly decrement refcount on failure in nfsd4_probe_callback. Signed-off-by:
-
Neil Brown authored
Fix Oops at nfsd shutdown. Reap any open nfs4_stateowners remaining on the close_lru hash list prior to reaping nfs4_clients at shutdown. Signed-off-by:
-
Neil Brown authored
Get rid of these debug counters. Since they're no longer necessarily incremented under state lock, we'd need to convert them to atomic_t to guarantee them correct. It doesn't seem worth it. Also, the debug checks in kmem_cache_destroy() should be sufficient for at least one of the purposes the debugging counters existed for--we'll get a dump_stack() if there are still stateowners left when we shut down. Signed-off-by:
-
Neil Brown authored
Create a slab cache for nfsv4 stateowners. Signed-off-by:
-
Neil Brown authored
RFC3530 says NFSv4 filenames are in UTF-8, and that servers "should" reject filenames that aren't valid UTF-8. Currently we do some incomplete UTF-8 checking, but this turns out to be more of a hindrance than a help--if we export a filesystem with non-UTF8 names, things break. We might be able to help things if we had some idea of the exported filesystem's charset and could do some translation. For now, though, the simplest solution by far seems to be just to remove the UTF-8 checks, making NFSv4's behaviour just like NFSv2/v3's. Similar problems exist with usernames. Signed-off-by:
-
Mingming Cao authored
This patch fixed a race between ext3_discard_reservation() and ext3_try_to_allocate_with_rsv(). There is a window where ext3_discard_reservation will remove an already unlinked reservation window node from the filesystem reservation tree: It thinks the reservation is still linked in the filesystem reservation tree, but it is actually temperately removed from the tree by allocate_new_reservation() when it failed to make a new reservation from the current group and try to make a new reservation from next block group. Here is how it could happen: CPU 1 try to allocate a block in group1 with given reservation window my_rsv ext3_try_to_allocate_with_rsv(group ----copy reservation window my_rsv into local rsv_copy ext3_try_to_allocate(...rsv_copy) ----no free block in existing reservation window, ----need a new reservation window spin_lock(&rsv_lock); CPU 2 ext3_discard_reservation if (!rsv_is_empty() ----this is true spin_lock(&rsv_lock) ----waiting for thread 1 CPU 1: allocate_new_reservation failed to reserve blocks in this group remove the window from the tree rsv_window_remove(my_rsv) ----window node is unlinked from the tree here return -1 spin_unlock(&rsv_lock) ext3_try_to_allocate_with_rsv() failed in this group group++ CPU 2 spin_lock(&rsv_lock) succeed rsv_remove_window () ---------------break, trying to remove a unlinked node from the tree .... CPU 1: ext3_try_to_allocate_with_rsv(group, my_rsv) rsv_is_empty is true, need a new reservation window spin_lock(&rsv_lock); ^--------------- spinning forever We need to re-check whether the reservation window is still linked to the tree after grab the rsv_lock spin lock in ext3_discard_reservation, to prevent panic in rsv_remove_window->rb_erase. Signed-off-by:
-
Mike Miller authored
This patch adds per disk queue functionality. It seems that the 2.6 kernel expects a queue per disk. If you have multiple logical drives on a controller all of the queues actually point back to the same queue. If a drive is deleted it blows us out of the water. We hold the lock during any queue operations and have added what we call a "fair-enough" algorithm to prevent starving out any drive. Signed-off-by:
Mike Miller <mike.miller@hp.com> Signed-off-by:
-
Mike Miller authored
This patch adds support for more than 8 controllers. If we run out of preallocated major numbers we dynamically allocate more. Signed-off-by:
-
Mike Miller authored
This patch adds support for 2 new SAS controllers due out this summer. It also bumps the version to 2.6.6. Signed-off-by:
-
Russell King authored
The Mobility docking station provides a PCI-based parallel port. Since the docking station connects via Cardbus, such devices are removable. Therefore, track which parallel ports are registered to each PCI device, and remove them when the PCI device is removed. Signed-off-by:
Russell King <rmk@arm.linux.org.uk> Signed-off-by:
-
David Howells authored
The attached patch causes process and session keyrings to be shared properly when CLONE_THREAD is in force. It does this by moving the keyring pointers into struct signal_struct[*]. [*] I have a patch to rename this to struct thread_group that I'll revisit after the advent of 2.6.11. Furthermore, once this patch is applied, process keyrings will no longer be allocated at fork, but will instead only be allocated when needed. Allocating them at fork was a way of half getting around the sharing across threads problem, but that's no longer necessary. This revision of the patch has the documentation changes patch rolled into it and no longer abstracts the locking for signal_struct into a pair of macros. Signed-Off-By:
David Howells <dhowells@redhat.com> Signed-off-by:
-
Alexander Nyberg authored
I looked at some of the oops reports against keyrings, I think the problem is that the search isn't restarted after dropping the key_user_lock, *p will still be NULL when we get back to try_again and look through the tree. It looks like the intention was that the search start over from scratch. Signed-off-by:
Alexander Nyberg <alexn@dsv.su.se> Cc: David Howells <dhowells@redhat.com> Signed-off-by:
-
Chris Wright authored
Add security contact info and relevant documentation. Signed-off-by:
Chris Wright <chrisw@osdl.org> Signed-off-by:
-
Coywolf Qi Hunt authored
At the very beginning in 2.4 days, in mnt_init(), mount_hashtable allocation page order was determined at runtime. Later the page order got fixed to 0. This patch cleanups it. Signed-off-by:
Coywolf Qi Hunt <coywolf@gmail.com> Signed-off-by:
-
Yoichi Yuasa authored
Signed-off-by:
Yoichi Yuasa <yuasa@hh.iij4u.or.jp> Signed-off-by:
-
Matt Mackall authored
The current cond_syscall #defines add a semicolon on the end, and then folks leave the semicolons off in kernel/sys_ni.c, which confuses editors that are language-aware and is just generally bad style. This sweeps all the users and makes sys_ni.c look like normal C code. Signed-off-by:
Matt Mackall <mpm@selenic.com> Signed-off-by:
-
Pavel Machek authored
Signed-off-by:
-
Jörn Engel authored
Randy Dunlap found one case where checkstack reported a false positive. From objdump of efi_stub.o: 5: 81 ea 00 00 00 c0 sub $0xc0000000,%edx With the old code, this was interpreted as a negative number. The output line was: 0xc0116f5d efi_call_phys: 1073741824 Randy wanted a change to return the correct number, like this: 0xc0116f5d efi_call_phys: 3221225472 adding "or I can just ignore it, like I've been doing for awhile..." Let's help him with the most sophisticated electronic tools and have the script actively ignore this case for him. Signed-off-by:
Jörn Engel <joern@wohnheim.fh-wedel.de> Signed-off-by:
-
Stephen Rothwell authored
This patch pulls together the compat_sigevent structs. It also consolidates the copying of these structures into the kernel. The only part of the second union in sigevent that the kernel looks at currently is the _tid, so that is the only bit we copy. This patch depends on my previous two patches "add and use COMPAT_SIGEV_PAD_SIZE" and "Consolidate the last compat sigvals". Signed-off-by:
Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by:
-
Stephen Rothwell authored
This patch just consolidates the last of the (what should have been) compat_sigval_ts. It also fixes S390 that had a sigval_t in its struct compat_siginfo. Signed-off-by:
-
Stephen Rothwell authored
All the 32 bit architectures (effectively) define SIGEV_PAD_SIZE to be ((SIGEV_MAX_SIZE/sizeof(int)) - 3). So define COMPAT_SIGEV_PAD_SIZE to be this and replace SIGEV_PAD_SIZE32 where it is used. It also needs to be used in the definition of struct compat_sigevent as most of the architectures would have had it 4 bytes too small in the kernel (since we were using SIGEV_PAD_SIZE). Signed-off-by:
-
Stephen C. Tweedie authored
With Andrew Morton invalidate_inode_pages2_range() can livelock if all the pages in the pagevec are outside the end of teh range which we're trying to invalidate. Fix that up in a similar manner to the truncate_inode_pages_range() handling. Signed-off-by:
-
