Sage Weil
535bc3f0aa
mds: avoid journal flush on max_size decrease; clean up some flush checks
2009-04-20 10:38:15 -07:00
Sage Weil
b6bdac0bcc
mds: make is_waiter_for() efficient, maskable
2009-04-20 10:38:15 -07:00
Sage Weil
bafb499541
mds: kill unused purging xlist
2009-04-20 10:38:15 -07:00
Sage Weil
cfb35839f6
mds: move leases to dentry only; use map instead of hash_map
...
This lowers mds memory usage by about 20%
2009-04-20 10:38:15 -07:00
Sage Weil
f602d4fef8
debug: fix memory leak
2009-04-20 07:23:41 -07:00
Sage Weil
9dee08715c
logger: fix memory leak (kill timer on shutdown)
2009-04-20 07:21:21 -07:00
Sage Weil
b7ff29c6cb
kclient: move igrab, __mark_inode_dirty into __ceph_mark_dirty_caps
...
Also be sure to __mark_inode_dirty(inode, I_DIRTY_DATASYNC) when
we dirty FILE_WRBUFFER.
2009-04-20 07:10:23 -07:00
Sage Weil
d3ece53710
buffer: cleanups
2009-04-20 07:00:12 -07:00
Sage Weil
47045f7375
atomic: use a spinlock instead of a mutex
2009-04-20 06:59:25 -07:00
Sage Weil
644249407d
spinlock: add support for pthread spinlocks
2009-04-20 06:59:01 -07:00
Sage Weil
d0b7c57a4e
vstart: valgrind select components; make tool + args selectable
2009-04-18 21:13:02 -07:00
Sage Weil
c4d1f15303
mds: kill unused cap updating field
...
It was used to avoid cap release from occuring while cap flush/writeback
is happening, bc there is a cap pointer in the completion. But the client
shouldn't be releasing a cap that is being written back in the first place.
2009-04-18 21:02:35 -07:00
Sage Weil
ceb2a3ba46
mds: delete mds on shutdown
...
Aids memory leak detection.
2009-04-18 20:11:55 -07:00
Sage Weil
5ea5e67528
journaler: put magic in journal headers
2009-04-18 16:58:26 -07:00
Sage Weil
380605530b
mds: fix root fullbit memory leak
2009-04-18 16:57:53 -07:00
Sage Weil
20d68a2ee7
buffer: make hexdump non-inline
2009-04-18 16:57:31 -07:00
Yehuda Sadeh
8207f9cd81
kclient: fix blob size initalization
2009-04-17 16:41:59 -07:00
Sage Weil
9f456a39e7
mds: add/remove inodes to/from open file list based on wanted caps changes
...
We only want to track "open" files for things the client will have open
file handles for. The rest we will recover opportunistically...
2009-04-17 16:01:20 -07:00
Sage Weil
f24baf5fc8
mds: fix file_excl() call preconditions
...
We could have filelock in LOCK_EXCL but no loner.. maybe..
2009-04-17 16:01:19 -07:00
Sage Weil
ee5a63c977
mds: avoid dup locallock wrlocks
2009-04-17 16:01:19 -07:00
Yehuda Sadeh
83c880c352
kclient: reorder setxattr to avoid race
2009-04-17 16:00:36 -07:00
Yehuda Sadeh
a68bfa1977
kclient: removexattr marks inode dirty when required
2009-04-17 12:46:41 -07:00
Sage Weil
4bc6291fc8
kclient: break I_NODELAY into NODELAY and FLUSH
...
FLUSH indicates the inode should be flushed asap. That is, we
should send a cap message with the dirty metadata. It's set
by write_inode.
NODELAY means we shouldn't delay our cap/wanted release when we
do send a message. It's set for unlinked files.
2009-04-17 12:26:44 -07:00
Sage Weil
2d1e6dcada
mds: rewrite scatter_tempsync
...
Recall caps, among other things.
2009-04-17 11:28:28 -07:00
Sage Weil
74907b25ad
vstart: selectively run mds through valgrind
2009-04-17 11:10:19 -07:00
Sage Weil
5fc8205150
mds: drop unused mdr->cap
2009-04-17 11:10:10 -07:00
Sage Weil
27b19722b3
mds: remove debug msg
2009-04-17 11:02:53 -07:00
Sage Weil
4b8036d86e
kclient: helpers for mount options
2009-04-17 10:30:27 -07:00
Sage Weil
a8deb11ca7
kclient: fix dcache readdir bug; add mount option to turn it off
2009-04-17 10:23:21 -07:00
Sage Weil
7d04390265
mds: init var
2009-04-17 10:17:30 -07:00
Sage Weil
939787bdaf
kclient: setxattr cleanups
...
Do dirty stuff inside i_lock.
Reuse retry label after successful allocation.
2009-04-17 10:10:04 -07:00
Sage Weil
0af0f452f8
mds: issue all caps we can (except RD|WR on dirs)
...
Might as well.
Most importantly, it means we issue EXCL caps on directories that
clients are using.
2009-04-17 09:42:56 -07:00
Sage Weil
95d4d52f99
mds: fix snap.c debug prefix
2009-04-17 09:36:44 -07:00
Sage Weil
7ae4ddd196
kclient: drop RD, WR caps on file lock setattr
...
The mds will move to LOCK state, which will revoke them anyway.
2009-04-17 09:36:35 -07:00
Sage Weil
2c3690dc89
mds: use actual cap wanted, not requested
...
(The wanted update doesn't always succeed.)
2009-04-17 09:28:11 -07:00
Sage Weil
14f47c59ef
mds: allow request_release to update wanted; do so if I_NODELAY
2009-04-17 09:15:36 -07:00
Sage Weil
5bb63d1160
kclient: fix up cap timeouts
2009-04-17 08:44:46 -07:00
Sage Weil
1ef1316c9b
mds, kclient: fix issue_seq vs seq, limit dup cap messages
2009-04-17 08:44:31 -07:00
Sage Weil
6ccb97f43b
mds: consolidate calls to issue_caps
2009-04-17 07:35:52 -07:00
Yehuda Sadeh
91f52b9697
kclient: silence down log messages
2009-04-16 16:01:50 -07:00
Yehuda Sadeh
f46db4fc98
kclient: add missing initialization
2009-04-16 15:48:45 -07:00
Yehuda Sadeh
fee4d7f8ad
kclient: fix xattrs cap message
2009-04-16 15:45:14 -07:00
Yehuda Sadeh
9b745ec376
kclient: merge asyncxattr
2009-04-16 14:46:35 -07:00
Yehuda Sadeh
4767269678
kclient: refactor setxattr, listxattr
2009-04-16 14:46:35 -07:00
Yehuda Sadeh
48d9fc9333
kclient: fix removexattr
2009-04-16 14:46:35 -07:00
Yehuda Sadeh
fadaa7af0a
kclient: fix some leaks
2009-04-16 14:46:34 -07:00
Yehuda Sadeh
d18b3544c1
kclient: xattrs, selective rb erasing, fix tree tear down
2009-04-16 14:46:34 -07:00
Yehuda Sadeh
6de31cb2e5
kclient: make listxattr actually work
2009-04-16 14:46:34 -07:00
Yehuda Sadeh
8e44f03c84
kclient: xattr name length fixes
2009-04-16 14:46:34 -07:00
Yehuda Sadeh
bb9e410c39
kclient: initialize xattrs version
2009-04-16 14:46:34 -07:00
