Whamcloud - gitweb
fs/lustre-release.git
16 years agoVM protocol want old IO finished before start new, in this case
shadow [Fri, 16 May 2008 06:50:00 +0000 (06:50 +0000)]
VM protocol want old IO finished before start new, in this case
need wait until PG_writeback is cleared until check dirty flag and
call writepages in lock cancel callback.
b=15443
i=nikita
i=zam

16 years agob=15660
walter [Fri, 16 May 2008 02:18:53 +0000 (02:18 +0000)]
b=15660
i=adilger

disable LIBLUSTRE test for NETTYPE != tcp && NETTYPE != ptl

16 years agoBranch b1_6
adilger [Thu, 15 May 2008 16:33:22 +0000 (16:33 +0000)]
Branch b1_6
Ignore compiled file.

16 years agoBranch b1_6
adilger [Wed, 14 May 2008 22:24:20 +0000 (22:24 +0000)]
Branch b1_6
Description: MDS may lose file striping (and hence file data) in some cases
Details    : If there are additional extended attributes stored on the MDS,
     in particular ACLs, SELinux, or user attributes (if user_xattr
     is specified for the client mount options) then there is a risk
     of attribute loss.  Additionally, the Lustre file striping
     needs to be larger than default (e.g. striped over all OSTs),
     and an additional attribute must be stored initially in the
     inode and then increase in size enough to be moved to the
     external attribute block (e.g. ACL growing in size) for file
     data to be lost.
i=kalpak.shah
i=alex.zhuravlev
b=15777

Sync ChangeLog from b_release_1_6_5.

16 years agoRefactor start to encapsulate facet specific data.
deen [Wed, 14 May 2008 21:21:39 +0000 (21:21 +0000)]
Refactor start to encapsulate facet specific data.

b=12375
i=deen
i=grev

16 years agoBranch b1_6
adilger [Wed, 14 May 2008 20:59:31 +0000 (20:59 +0000)]
Branch b1_6
Merge ChangeLog entries from b_release_1_6_5 to b1_6.

16 years agob=15738
grev [Tue, 13 May 2008 13:20:02 +0000 (13:20 +0000)]
b=15738
i=Adilger

sanity test_118* fixes; set fail loc on all oss nodes

16 years agoBranch b1_6
anserper [Tue, 13 May 2008 08:08:48 +0000 (08:08 +0000)]
Branch b1_6
b=15238
i=shadow
i=tianzy

refactoring of quota code

16 years agoBranch b1_6
bwzhou [Tue, 13 May 2008 05:56:39 +0000 (05:56 +0000)]
Branch b1_6
b=14740
r=adilger, bobijam

break a single big chunk of memory allocation for lprocfs_stats into separately
allocated small chunks to prevent kmalloc failure in a huge cluster

16 years agoBranch b1_6
bobijam [Tue, 13 May 2008 01:28:34 +0000 (01:28 +0000)]
Branch b1_6
b=12888
i=johann, green

Description: mds_mfd_close() ASSERTION(rc == 0)
Details    : In mds_mfd_close(), we need protect inode's writecount change
             within its orphan write semaphore to prevent possible races.

16 years agob=15745
grev [Mon, 12 May 2008 20:14:44 +0000 (20:14 +0000)]
b=15745
i=Adilger

the minor cleanup of the previous fix:
-- after we get rid of $tdir we do not need to create $tmpdir, because
   unzip -d <dir> does not need <dir> exists, it needs the parent directory
   of <dir> exists; unzip creates <dir> if it does not exists.

16 years agoBranch b1_6
anserper [Mon, 12 May 2008 12:18:48 +0000 (12:18 +0000)]
Branch b1_6
b=14790
i=Johann
i=ZhiYong

additional test. check if setting limits twice doesn't break oplimits.

16 years agoReinitialize optind to 0 on each lfs_* call, in case lfs is being used interactively...
kalpak [Mon, 12 May 2008 11:31:05 +0000 (11:31 +0000)]
Reinitialize optind to 0 on each lfs_* call, in case lfs is being used interactively.  Most lfs_* functions do this already, but a few (find, setstripe, quotachown) do not.
 b=15212
 i=rupesh
 i=johann

16 years agoBranch b1_6
tianzy [Mon, 12 May 2008 05:15:36 +0000 (05:15 +0000)]
Branch b1_6
change the time of ost informing osc out of quota in order to get more
accurate quota
b=14909
i=johann, andrew

16 years agoBranch b1_6
adilger [Mon, 12 May 2008 03:45:46 +0000 (03:45 +0000)]
Branch b1_6
Fix typo in error message.

16 years agoBranch b1_6
bwzhou [Sun, 11 May 2008 10:39:11 +0000 (10:39 +0000)]
Branch b1_6
b=15521
r=adilger, grev

the test case for bug 15521

16 years agob=15732
walter [Sat, 10 May 2008 15:24:50 +0000 (15:24 +0000)]
b=15732
i=walter

Land andreas' fix for catamount build errors

16 years agob=15732
walter [Sat, 10 May 2008 15:10:59 +0000 (15:10 +0000)]
b=15732
i=adilger

fix build warnings which are treated as errors
during Cray XT3 catamount build

16 years agoBranch b1_6
bwzhou [Fri, 9 May 2008 16:44:07 +0000 (16:44 +0000)]
Branch b1_6
b=14034
r=johann, bobijam

error handling for ENOMEM case when allocating a ldlm resource

16 years agodon't hit live lock with umount ost.
shadow [Fri, 9 May 2008 10:19:15 +0000 (10:19 +0000)]
don't hit live lock with umount ost.
b=14645
i=johan
i=alex

16 years agob=15745
grev [Fri, 9 May 2008 08:36:05 +0000 (08:36 +0000)]
b=15745
i=Adilger

conf-sanity test_32a, test_32b: create tmp dir correctly

16 years agob=15660
grev [Fri, 9 May 2008 08:01:41 +0000 (08:01 +0000)]
b=15660
i=Adilger

disable LIBLUSTRE test for NETTYPE != tcp

16 years agorollback bug 15575, because at any run acc-sm i hit
shadow [Thu, 8 May 2008 11:54:07 +0000 (11:54 +0000)]
rollback bug 15575, because at any run acc-sm i hit
LustreError: 29518:0:(lvfs_linux.c:194:pop_ctxt()) ASSERTION(current->fs->pwd == new_ctx->pwd)
after run-tests.

16 years agodon't panic with use echo client
shadow [Thu, 8 May 2008 10:18:09 +0000 (10:18 +0000)]
don't panic with use echo client
b=14949
i=johan
i=deen

16 years agob=11270
zam [Thu, 8 May 2008 05:46:55 +0000 (05:46 +0000)]
b=11270
i=vitaly.vertman
i=oleg.drokin

Lockless i/o code fixes and improvements:
 (1) lockless truncate checks for OBD_CONNECT_TRUNCLOCK and ll_file_punch
     sets OBD_FL_TRUNCLOCK correctly.
 (2) an lproc control for lockless truncate,
     lproc statistics for lockless truncate.
 (3) sanityN tests for lockless code

16 years agoBranch b_release_1_6_5
adilger [Wed, 7 May 2008 22:23:22 +0000 (22:23 +0000)]
Branch b_release_1_6_5
Update client side RPC messages to match those on the server, as changed
previously on the server in bug 13537, allowing the same :pid:xid: string
to be searched in both client and server logs.
b=13537

16 years agoBranch b1_6
bobijam [Wed, 7 May 2008 01:50:50 +0000 (01:50 +0000)]
Branch b1_6
b=15575
i=wangdi, johann

Description: Stack overflow during MDS log replay
Details    : ease stack pressure by using a thread dealing llog_process.

16 years agocompile with -m64 flag produce wrong object file for ppc32.
shadow [Tue, 6 May 2008 18:16:09 +0000 (18:16 +0000)]
compile with -m64 flag produce wrong object file for ppc32.
b=15278
i=johann
i=deen

16 years agoBranch b1_6
bwzhou [Mon, 5 May 2008 15:15:28 +0000 (15:15 +0000)]
Branch b1_6
b=14219
r=adilger, shadow

use punch instead of setattr to change file size

16 years agoBranch b1_6
johann [Fri, 2 May 2008 20:26:35 +0000 (20:26 +0000)]
Branch b1_6
b=15517
i=johann
i=umka

Fix compile warnings on i686.

16 years agoBranch b1_6
johann [Fri, 2 May 2008 19:41:01 +0000 (19:41 +0000)]
Branch b1_6
b=15069
i=rread
i=wangdi

remove LASSERT in ptlrpc_invalidate_import() since ping req can be queued even
imp_invalid is set. See also bug 15716.

16 years agoBranch b1_6
johann [Fri, 2 May 2008 19:18:39 +0000 (19:18 +0000)]
Branch b1_6
b=15069
i=rread
i=wangdi

Skip the "sleep until imp_inflight == 0" part when invalidating an mgc import
which is already invalid.

16 years agoBranch b1_6
johann [Fri, 2 May 2008 19:01:34 +0000 (19:01 +0000)]
Branch b1_6
b=13904
i=johann
i=tianzy

Quota data should not be stored in a 32-bit variable. Use __u64 instead.

16 years agoDon't disable LIBLUSTRE_POSIX_ACL if --disable-liblustre given and
deen [Wed, 30 Apr 2008 20:55:46 +0000 (20:55 +0000)]
Don't disable LIBLUSTRE_POSIX_ACL if --disable-liblustre given and
don't build wiretest with build utils.

b=15642
i=adilger
i=shadow

16 years agoBranch b1_6
fanyong [Wed, 30 Apr 2008 17:23:56 +0000 (17:23 +0000)]
Branch b1_6
b=15691
i=huanghua
i=tappro

1) Prevent "ll_sa_wrong" from being increased repeatedly when hit ratio is low.
2) Remount client before sanity 123a to eliminate previous test effect.

16 years agobranch: b1_6
ericm [Wed, 30 Apr 2008 17:18:03 +0000 (17:18 +0000)]
branch: b1_6
set msghdr flags right before rpc sent out.
b=15487
r=rread
r=shadow

16 years agoBranch b1_6
tianzy [Wed, 30 Apr 2008 09:37:06 +0000 (09:37 +0000)]
Branch b1_6
b=15643
i=johann

Prolong the waiting time in test_12 of sanity-quota.sh

16 years agoBranch b1_6
bwzhou [Wed, 30 Apr 2008 05:32:57 +0000 (05:32 +0000)]
Branch b1_6

Add ChangeLog entries.

16 years agoBranch b1_6
bobijam [Wed, 30 Apr 2008 01:44:14 +0000 (01:44 +0000)]
Branch b1_6
b=14995
i=wangdi, rread

send activate event only for evicted imports.

16 years agob=13539
walter [Tue, 29 Apr 2008 22:49:51 +0000 (22:49 +0000)]
b=13539
i=johann
i=adilger

fix for configure: WARNING: linux/random.h: present but cannot be compiled

16 years agob=15689
walter [Tue, 29 Apr 2008 16:25:58 +0000 (16:25 +0000)]
b=15689
i=johann
i=alexey.lyashkov

fix error building lustre/liblustre/tests for Cray XT3 catamount

16 years agoBranch b1_6
fanyong [Tue, 29 Apr 2008 02:44:39 +0000 (02:44 +0000)]
Branch b1_6
b=15406

Add ChangeLog entry.

16 years agoBranch b1_6
anserper [Mon, 28 Apr 2008 20:51:51 +0000 (20:51 +0000)]
Branch b1_6
b=15441
i=johann
i=tianzy

quota might appear on quota files causing deadlock, this is fixed with this patch

16 years agob=15316,13969
brian [Mon, 28 Apr 2008 19:21:02 +0000 (19:21 +0000)]
b=15316,13969

ChangeLog entries.

16 years agoBranch b1_6
anserper [Mon, 28 Apr 2008 14:04:32 +0000 (14:04 +0000)]
Branch b1_6
b=15036

updated changelog with respect to 15036 updates

16 years agoBranch b1_6
bwzhou [Mon, 28 Apr 2008 04:21:59 +0000 (04:21 +0000)]
Branch b1_6
b=11777
r=green, shadow

invalidate the import to halt request handling so the namespace can be safely
cleaned up

16 years agoBranch b1_6
bobijam [Mon, 28 Apr 2008 02:17:23 +0000 (02:17 +0000)]
Branch b1_6
b=14803
i=johann, adilger

Description: Don't update lov_desc members until making sure they are valid
Details    : When updating lov_desc members via proc fs, need fix their
             validities before doing the real update.

16 years agoUpdated for future beta release.
stinson1947 [Sun, 27 Apr 2008 02:48:57 +0000 (02:48 +0000)]
Updated for future beta release.

16 years agoAdded header for the next release 1.6.6.
stinson1947 [Sun, 27 Apr 2008 02:47:15 +0000 (02:47 +0000)]
Added header for the next release 1.6.6.

16 years ago*** empty log message ***
stinson1947 [Sun, 27 Apr 2008 02:45:41 +0000 (02:45 +0000)]
*** empty log message ***

16 years agoChanged version numbers for 1.6.5.
stinson1947 [Sun, 27 Apr 2008 01:51:13 +0000 (01:51 +0000)]
Changed version numbers for 1.6.5.

16 years agoAdded date for the 1.6.5 release.
stinson1947 [Sun, 27 Apr 2008 01:48:47 +0000 (01:48 +0000)]
Added date for the 1.6.5 release.

16 years agoBranch b1_6
fanyong [Sat, 26 Apr 2008 14:30:54 +0000 (14:30 +0000)]
Branch b1_6
b=15658
i=huanghua
i=tappro

Support directory read ahead under patchless client mode.
When "ls" a dentry, the system trigger more than once "revalidate" or
"lookup", for "getattr", for "getxattr", and maybe for others.
Under patchless client mode, the operation intent is not accurate,
it maybe misguide the statahead thread. For example:
The "revalidate" call for "getattr" and "getxattr" of a dentry maybe
have the same operation intent -- "IT_GETATTR".
In fact, one dentry should has only one chance to interact with the
statahead thread, otherwise the statahead windows will be confused.

16 years agoIt is be possible to send the lock handle along with each read
deen [Sat, 26 Apr 2008 12:39:55 +0000 (12:39 +0000)]
It is be possible to send the lock handle along with each read
or write request because the client is already doing a lock match
itself so there isn't any reason the OST should have to re-do that
search.

b=15198
i=adilger
i=shadow

16 years agob=15277
walter [Sat, 26 Apr 2008 02:23:43 +0000 (02:23 +0000)]
b=15277
i=adilger
i=rread

fixes error linking lustre/liblustre/tests/echo_test
during catamount build of b1_6 for Cray XT3 systems

16 years agoBranch b1_6
adilger [Sat, 26 Apr 2008 00:10:33 +0000 (00:10 +0000)]
Branch b1_6
Clarify instructions.

16 years agoBranch b1_6
adilger [Sat, 26 Apr 2008 00:08:57 +0000 (00:08 +0000)]
Branch b1_6
Fix line wrap.

16 years agoBranch b1_6
adilger [Fri, 25 Apr 2008 23:22:02 +0000 (23:22 +0000)]
Branch b1_6
Minor cleanup.

16 years agob=15316
brian [Fri, 25 Apr 2008 21:51:22 +0000 (21:51 +0000)]
b=15316
i=cliffw
i=wangyb

Build OFED 1.3's kernel-ib and kernel-ib-devel with our patched lustre
server kernel for RHEL 4 and 5 and SLES 10.  SLES 9 is not supported by
OFED 1.3.
Since OFED 1.3 provides a newer open-iscsi stack, prefer those modules
from the kernel-ib over those provided by the vendor kernels.
Additionally, build Lustre with o2iblnd support.

16 years agob=14609
walter [Fri, 25 Apr 2008 19:31:09 +0000 (19:31 +0000)]
b=14609
i=grev

fix for test_16 failure due to trying to create mntpt dir on Read-Only fs.

16 years agob=15670
walter [Fri, 25 Apr 2008 19:22:30 +0000 (19:22 +0000)]
b=15670
i=adilger
i=johann

land attachment 16523

fixes this compile warning (which is treated as an error -Werror on XT3's):
during lustrefs-1.6-ss builds:

lustrefs-1.6-ss/lustre/mds/mds_lov.c: In function `mds_lov_read_objids':
mds_lov.c:198: warning: int format, different type arg (arg 10)

16 years agoBranch b1_6
johann [Fri, 25 Apr 2008 17:42:29 +0000 (17:42 +0000)]
Branch b1_6
b=14909
i=tianzy

use cancel_lru_locks instead of "sync; sleep 1; sync" in sanity-quota.sh.

16 years agoBranch b1_6
johann [Fri, 25 Apr 2008 14:38:30 +0000 (14:38 +0000)]
Branch b1_6
b=14410
i=bzzz
i=kalpak

Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
sure that 1MB requests are not fragmented by the block layer.

16 years agob=13539
walter [Fri, 25 Apr 2008 14:00:05 +0000 (14:00 +0000)]
b=13539
i=adilger
i=rread

add linux/random.h to AC_CHECK_HEADERS
add #if defined(HAVE_LINUX_RANDOM_H) to lustre/lvfs/prng.c

16 years agodon't put request into delay list while invalidate in flight.
shadow [Thu, 24 Apr 2008 11:54:19 +0000 (11:54 +0000)]
don't put request into delay list while invalidate in flight.
normalize imp_infligh calculation.

b=15069
i=johann
i=rread

16 years agoBranch b1_6
yangsheng [Thu, 24 Apr 2008 08:15:57 +0000 (08:15 +0000)]
Branch b1_6
b=15416
i=shadow, johann

Update kernel for SLES9 2.6.5-7.311.

16 years agoBranch b1_6
yangsheng [Thu, 24 Apr 2008 08:08:48 +0000 (08:08 +0000)]
Branch b1_6
b=15240
i=shadow, green

Update to RHEL4 latest kernel.

16 years agouse generic CFS_LIST_HEAD macro instead of LIST_HEAD.
shadow [Thu, 24 Apr 2008 07:07:59 +0000 (07:07 +0000)]
use generic CFS_LIST_HEAD macro instead of LIST_HEAD.
b=14929
i=zam

16 years agob=14712
green [Thu, 24 Apr 2008 05:58:37 +0000 (05:58 +0000)]
b=14712
r=vitaly,shadow

Do not free request that we do not have refcount on on open error

16 years agoBranch b1_6
fanyong [Thu, 24 Apr 2008 04:41:11 +0000 (04:41 +0000)]
Branch b1_6
b=15406
i=huanghua
i=tappro
i=vitaly

Back port dir_SA fixes from HEAD (b1_8_dir_ra) to b1_6.

16 years agoBranch b1_6
adilger [Thu, 24 Apr 2008 02:43:04 +0000 (02:43 +0000)]
Branch b1_6
Document the maximum filesysyem name length.
b=15486

16 years agobranch: b1_6
ericm [Wed, 23 Apr 2008 18:00:48 +0000 (18:00 +0000)]
branch: b1_6
reset AT flag on reconnects.
a=nathan (original author)
b=15487
r=rread
r=ericm

16 years agoBranch b1_6
yangsheng [Wed, 23 Apr 2008 13:24:40 +0000 (13:24 +0000)]
Branch b1_6
b=14714
i=adilger, johann

Fixed previous error.

16 years agonot swab ost objects for directory because it's not exist.
shadow [Tue, 22 Apr 2008 11:39:53 +0000 (11:39 +0000)]
not swab ost objects for directory because it's not exist.
b=14856
i=green
i=johan

16 years agoSet proper locale when generating wiretest.c.
deen [Tue, 22 Apr 2008 08:18:45 +0000 (08:18 +0000)]
Set proper locale when generating wiretest.c.

b=15566

16 years agoUpdate wirecheck.c/wirehdr.c and regenerate wiretest.c
deen [Tue, 22 Apr 2008 07:53:33 +0000 (07:53 +0000)]
Update wirecheck.c/wirehdr.c and regenerate wiretest.c

b=15566
i=adilger
i=shadow

16 years agob=15273
grev [Mon, 21 Apr 2008 18:05:49 +0000 (18:05 +0000)]
b=15273
i=Adilger
skip test_37 if CLIENTMODSONLY is set

16 years agobranch: b1_6
ericm [Mon, 21 Apr 2008 17:11:38 +0000 (17:11 +0000)]
branch: b1_6
change at_max according to AT status.
b=13659
r=nathan
r=rread
r=grev

16 years agoBranch b1_6
anserper [Mon, 21 Apr 2008 11:20:39 +0000 (11:20 +0000)]
Branch b1_6
b=15218
i=johann
i=tianzy

additional admin quota file update

16 years agob=15612
grev [Fri, 18 Apr 2008 10:35:14 +0000 (10:35 +0000)]
b=15612
i=Adilger
force test_20a and test_20b to do write on the correct ost

16 years agoFix warnings with compile liblustre at sles10/rhel5 which have
shadow [Fri, 18 Apr 2008 09:45:38 +0000 (09:45 +0000)]
Fix warnings with compile liblustre at sles10/rhel5 which have
__u64 as usingied long long type.
b=15517
i=johan
i=deen

16 years agoremoving callback before disconnect import is open race with processing
shadow [Fri, 18 Apr 2008 07:16:30 +0000 (07:16 +0000)]
removing callback before disconnect import is open race with processing
callback.
b=15210
i=green
i=deen

16 years agoBranch b1_6
bwzhou [Fri, 18 Apr 2008 04:47:53 +0000 (04:47 +0000)]
Branch b1_6
b=15521
r=adilger, shadow

correct the initialization of lc_min of struct lprocfs_counter

16 years agoinit rq_history_list before use.
shadow [Thu, 17 Apr 2008 13:41:42 +0000 (13:41 +0000)]
init rq_history_list before use.
b=15412
i=johann
i=deen

16 years agofix typo.
shadow [Thu, 17 Apr 2008 12:30:05 +0000 (12:30 +0000)]
fix typo.
i=nikita
i=johann

16 years agob=15440
yury [Wed, 16 Apr 2008 20:49:35 +0000 (20:49 +0000)]
b=15440
r=shadow,adilger
- remove @ns from list using list_del_init() to make sure that consequnet list_empty() will really work right on removed listy_head.

16 years agob=15482
brian [Wed, 16 Apr 2008 14:41:07 +0000 (14:41 +0000)]
b=15482
i=cliffw
i=wangyb

Allow a kernel target to specify a delimiter between the version and the
"flavor" (i.e. "smp")
 - specifically this allows RH kernels to be <version>smp and SLES10 kernels
   to be <version>-smp.
 - maintaining the vendors versioning specification means that when software
   is trying to determine a vendors kernel using patterns (like OFED does),
   it properly detects the SLES10 kernel.
This all lays the ground work for bug 15316.

16 years agob=15273
grev [Wed, 16 Apr 2008 14:07:44 +0000 (14:07 +0000)]
b=15273
i=Nathan.Rutman
skip test_37 for client only testing

16 years agob=15526
grev [Wed, 16 Apr 2008 12:13:07 +0000 (12:13 +0000)]
b=15526
i=Adilger
allow both "cfs" and "sun" e2fsck

16 years agob=15513
grev [Wed, 16 Apr 2008 12:00:54 +0000 (12:00 +0000)]
b=15513
i=Yury.Umanets
i=Alexey.Lyashkov
force test_69 to fail if directio write failed

16 years agob=14901
yury [Wed, 16 Apr 2008 08:38:20 +0000 (08:38 +0000)]
b=14901
r=grev
- fixes using check_runas_id(). Adds check for used runas id in test_72.

16 years agobranch: b1_6
ericm [Tue, 15 Apr 2008 21:11:16 +0000 (21:11 +0000)]
branch: b1_6
fix AT time calculation in AT_OFF mode.
b=3055
r=nathan
r=rread

16 years agoBranch b1_6
yangsheng [Tue, 15 Apr 2008 16:02:37 +0000 (16:02 +0000)]
Branch b1_6
b=12652
i=shadow, johann

Add FMODE_EXEC flag to SLES10 SP1.

16 years agobranch: b1_6
ericm [Mon, 14 Apr 2008 18:56:17 +0000 (18:56 +0000)]
branch: b1_6
increase the version to 1.6.4.56

16 years agobranch: b1_6
ericm [Mon, 14 Apr 2008 18:10:22 +0000 (18:10 +0000)]
branch: b1_6
fix a race that early reply right after realy reply.
b=15178
Christopher Morrone, LLNL (original author)
r=nathan
r=ericm

16 years agobranch: b1_6
ericm [Fri, 11 Apr 2008 19:03:15 +0000 (19:03 +0000)]
branch: b1_6
Do not skip at_check_timed in AT_OFF mode, early replies might still be
sent out until affected RPCs will drain eventually.
b=15341
r=nathan
r=rread

16 years agoUse fsfilt_sync() for both the whole filesystem sync and
deen [Thu, 10 Apr 2008 12:50:35 +0000 (12:50 +0000)]
Use fsfilt_sync() for both the whole filesystem sync and
individual file sync to eliminate dangerous inode locking
with I_LOCK that can lead to a deadlock.

b=15188
i=green
i=shadow

16 years agoBranch b1_6
vitaly [Thu, 10 Apr 2008 11:55:07 +0000 (11:55 +0000)]
Branch b1_6
b=14661
i=green
i=shadow

-- remove BL_AST lock from lru once we mark the lock as a such;
-- do not place BL_AST lock into lru even if the lock is matched and put later;
   => no BL_AST locks in lru at all;
-- CANCELING locks are not so numerous, but to avoid its traversal by another
   thread another time, drop them from the unused list once come them across.

16 years agofix iov handling.
shadow [Thu, 10 Apr 2008 06:27:13 +0000 (06:27 +0000)]
fix iov handling.
b=15483
i=green
i=johann

16 years agoBranch b1_6
yangsheng [Thu, 10 Apr 2008 03:09:56 +0000 (03:09 +0000)]
Branch b1_6
b=13397
i=johann
i=shadow

Split the export journal-api part as a separate patch.