Whamcloud - gitweb
fs/lustre-release.git
14 years agob=21746 compare full filename when writeconf'ing
Nathan Rutman [Mon, 25 Jan 2010 20:01:21 +0000 (12:01 -0800)]
b=21746 compare full filename when writeconf'ing

When erasing MGS logs, we could confuse fsname e.g. test with test2 and erase both sets of logs.

i=breitz
i=brian

14 years agob=21889 Rename lustre-resync-test
Robert Read [Mon, 25 Jan 2010 19:36:28 +0000 (11:36 -0800)]
b=21889 Rename lustre-resync-test

Fix Makefile.

14 years agob=20918 improve log warning
Elena Gryaznova [Mon, 25 Jan 2010 11:42:11 +0000 (14:42 +0300)]
b=20918 improve log warning

i=Brian

14 years agob=20866 DEPS assignment needs quotes
Elena Gryaznova [Mon, 25 Jan 2010 11:28:36 +0000 (14:28 +0300)]
b=20866 DEPS assignment needs quotes

o=Brian
i=grev

14 years agob=21828 drop number of active requests when queued for recovery
Dmitry Zogin [Mon, 25 Jan 2010 05:15:53 +0000 (00:15 -0500)]
b=21828 drop number of active requests when queued for recovery

Now that we take a reference on the original request instead of
making a copy of it for recovery.  We need to drop the number of
active requests or the queued requests will prevent all request
processing when they exceed (srv->srv_threads_running - 1).

i=nathaniel.ruthman
i=tappro

14 years agob=21565 filter_last_id() NULL dereference
Dmitry Zogin [Mon, 25 Jan 2010 05:13:27 +0000 (00:13 -0500)]
b=21565 filter_last_id() NULL dereference

lprocfs_filter_rd_last_id() should check for the fully setup obd device
 before proceeding further.

 i=johann
 i=andrew.perepechko

14 years agob=20139 prevent parent thread to be killed before its child becoming daemon
Fan Yong [Sat, 23 Jan 2010 06:08:13 +0000 (14:08 +0800)]
b=20139 prevent parent thread to be killed before its child becoming daemon

Prevent parent thread to be killed before its child becoming daemon.

i=tappro
i=robert

14 years agob=17682 fix time unit in message
Dmitry Zogin [Fri, 22 Jan 2010 23:16:02 +0000 (18:16 -0500)]
b=17682 fix time unit in message

14 years agob=19557 actually make lustre_hash_for_each_empty() more efficient
Robert Read [Wed, 13 Jan 2010 02:25:03 +0000 (18:25 -0800)]
b=19557 actually make lustre_hash_for_each_empty() more efficient

a=johann
i=adilger
i=herring4

14 years agob=19557 Add cond_resched calls to avoid hogging the cpu
Robert Read [Wed, 13 Jan 2010 01:48:48 +0000 (17:48 -0800)]
b=19557 Add cond_resched calls to avoid hogging the cpu

a=johann
i=adilger
i=andrew.perepechko

14 years agob=21889 Rename lustre_rsync-test.sh to lustre-rsync-test.sh.
Robert Read [Mon, 25 Jan 2010 16:38:48 +0000 (08:38 -0800)]
b=21889 Rename lustre_rsync-test.sh to lustre-rsync-test.sh.

14 years agob=21370 Double qos maxage timeout
Robert Read [Sat, 23 Jan 2010 00:52:51 +0000 (16:52 -0800)]
b=21370 Double qos maxage timeout

a=simmonsja@ornl.gov
i=adilger

14 years agoPrepare for Build 35. 1.10.0.35 v1_10_0_35
Robert Read [Fri, 22 Jan 2010 18:48:15 +0000 (10:48 -0800)]
Prepare for Build 35.

14 years agojbd2 spinlock hang fix for rhel5 and added error check
Rahul Deshmukh [Fri, 22 Jan 2010 02:42:24 +0000 (08:12 +0530)]
jbd2 spinlock hang fix for rhel5 and added error check

b=21595 mdsrate-create-large.sh, BUG: soft lockup - CPU#0 stuck for 10s!
[ll_ost_io_04:8153]

jbd2 spinlock hang fix for rhel5 and added error check

i=johann
i=girish

14 years agob=16715 NFS cannot files in HEAD, MDS crashing
LiuYing [Thu, 21 Jan 2010 09:03:16 +0000 (17:03 +0800)]
b=16715 NFS cannot files in HEAD, MDS crashing

For NFS open, return after anon fid lookup even for ENOENT

i=oleg.drokin
i=eric.mei
o=wangdi

14 years agob=20057 Autovetting and test-framework enhancements
Manoj Joseph [Wed, 20 Jan 2010 09:06:26 +0000 (02:06 -0700)]
b=20057 Autovetting and test-framework enhancements

Test-framework and script changes to support autovetting and buffalo V2

i=rread
i=grev

14 years agoRevert "b=21489 fix several write+utimes race conditions"
Robert Read [Fri, 22 Jan 2010 17:54:06 +0000 (09:54 -0800)]
Revert "b=21489 fix several write+utimes race conditions"

This reverts commit af0e77badba6094e792e5ebf7857f7862dcd4046.

14 years agoRevert "b=15692 statahead should hold parent dir's i_mutex..."
Robert Read [Thu, 21 Jan 2010 20:06:08 +0000 (12:06 -0800)]
Revert "b=15692 statahead should hold parent dir's i_mutex..."

This reverts most of e99ff8c12711281dc7e7e6bfd02c96eb956e4a33. The #ifdef
DCACHE_LUSTRE_INVALID changes remain, but everything else is reverted. This also
normalizes the ll_lookup_lock.

o=ericm
i=rread

14 years agob=21448 send recovery rpc ASAP
Fan Yong [Wed, 20 Jan 2010 18:41:39 +0000 (10:41 -0800)]
b=21448 send recovery rpc ASAP

1) send recovery rpc ASAP.
2) add more diagnostic information for RPC failure.

i=robert.read
i=tappro

14 years agob=21716 Reduce memory consumptions in directio utility
Alexander.Zarochentev [Wed, 20 Jan 2010 18:41:39 +0000 (10:41 -0800)]
b=21716 Reduce memory consumptions in directio utility

use one memory mapped file buffer in directio.c instead of two.

i=tappro
i=andrew.perepechko

14 years agob=21709 do not trigger quota processing for SOM operation
Fan Yong [Wed, 20 Jan 2010 18:41:39 +0000 (10:41 -0800)]
b=21709 do not trigger quota processing for SOM operation

Do not trigger quota processing for SOM operation.

i=robert.read
i=vitaly.fertman

14 years agob=18790 Add config check to avoid type conflict.
yangsheng [Wed, 20 Jan 2010 18:41:39 +0000 (10:41 -0800)]
b=18790 Add config check to avoid type conflict.

i=panda
i=yangsheng

14 years agob=20758 Update SLES10 kernel to 2.6.16.60-0.42.8(only patchless).
yangsheng [Wed, 20 Jan 2010 18:41:39 +0000 (10:41 -0800)]
b=20758 Update SLES10 kernel to 2.6.16.60-0.42.8(only patchless).

i=johann

14 years agob=19673 Add ldiskfs maxdirsize mount option
Cliff White [Tue, 19 Jan 2010 19:54:20 +0000 (11:54 -0800)]
b=19673 Add ldiskfs maxdirsize mount option

i=girish.shilamkar
i=johann

14 years agob=19720 use min_t() to force comparison to unsigned
Brian J. Murrell [Tue, 19 Jan 2010 19:14:49 +0000 (14:14 -0500)]
b=19720 use min_t() to force comparison to unsigned

In older kernels num_online_cpus() is an int, and in newer
kernels it is an unsigned so force the comparison to unsigned
so that it's portable to both new and old kernels.

i=panda
i=whitebear

14 years agob=21671 Eliminate a warning for the ldiskfs module
Andrew Perepechko [Tue, 19 Jan 2010 18:34:21 +0000 (21:34 +0300)]
b=21671 Eliminate a warning for the ldiskfs module

o=Yang Sheng
i=Andrew Perepechko
i=Rahul Deshmukh

14 years agob=21489 fix several write+utimes race conditions
Andrew Perepechko [Tue, 19 Jan 2010 18:10:50 +0000 (21:10 +0300)]
b=21489 fix several write+utimes race conditions

Disable mtime updates on write and serialize fmd_mactime_xid checks in order to
avoid certain write(2)+utimes(2) race conditions on OSS

i=Johann Lombardi
i=Vitaly Fertman

14 years agob=15864 insanity test_6 : use stat -f instead of df for determining the recovery...
Elena Gryaznova [Tue, 12 Jan 2010 20:36:09 +0000 (23:36 +0300)]
b=15864 insanity test_6 : use stat -f instead of df for determining the recovery status

i=Jian.Yu

14 years agob=19702 fix COUNT to work properly
Elena Gryaznova [Tue, 12 Jan 2010 20:26:35 +0000 (23:26 +0300)]
b=19702 fix COUNT to work properly

i=Andrew.Perepechko

14 years agob=17167 libcfs: ensure all libcfs exported symbols to have cfs_ prefix
Lisa Week [Thu, 14 Jan 2010 05:44:06 +0000 (22:44 -0700)]
b=17167 libcfs: ensure all libcfs exported symbols to have cfs_ prefix

i=Andreas Dilger

14 years agob=21670 use int instead of char to hold getopt rc in accordance with its prototype
Andrew Perepechko [Tue, 12 Jan 2010 19:13:42 +0000 (22:13 +0300)]
b=21670 use int instead of char to hold getopt rc in accordance with its prototype

i=Brian Murrell

14 years agob=21670 use int instead of char to hold getopt rc in accordance with its prototype
Andrew Perepechko [Mon, 11 Jan 2010 12:55:42 +0000 (15:55 +0300)]
b=21670 use int instead of char to hold getopt rc in accordance with its prototype

i=Brian Murrell

14 years agob=21670 use the correct fmt specifier for *printf in check_cur_qunit
Andrew Perepechko [Mon, 11 Jan 2010 14:45:48 +0000 (17:45 +0300)]
b=21670 use the correct fmt specifier for *printf in check_cur_qunit

i=Dmitry Zogin
i=Sheng Yang

14 years agob=21586 More stderr/stdout redirections
Brian J. Murrell [Fri, 8 Jan 2010 16:25:44 +0000 (11:25 -0500)]
b=21586 More stderr/stdout redirections

Just a few more redirections to get some commands' output into the
appropriate log files.

We should actually return the 255, not just assign it to a unused variable.

i=yangsheng
i=wangyb

14 years agob=21670 revert new version of find_linux_devel_paths
Andrew Perepechko [Fri, 8 Jan 2010 16:25:43 +0000 (11:25 -0500)]
b=21670 revert new version of find_linux_devel_paths

i=Andrew Perepechko
o=Brian Murrell

14 years agob=21586 further build message cleanups
Andrew Perepechko [Fri, 8 Jan 2010 16:25:42 +0000 (11:25 -0500)]
b=21586 further build message cleanups

i=Yibin Wang
i=Sheng Yang

14 years agob=21759 Miscellaneous build fixes
Brian J. Murrell [Tue, 12 Jan 2010 17:06:04 +0000 (12:06 -0500)]
b=21759 Miscellaneous build fixes

The message string given to fatal() cannot be slit with line continuations
as you would strings elsewhere -- for whatever reason.  So let's just un-
split them for now.
Coding style fixups.
Adds a "--set-var" option to lbuild to set/override an environment variable.
This is mainly meant for lbuild testers.
Fix missing - from tar so that the --exclude parameters will be honoured.
Some stderr->stdout redirections to get output into the correct log.

14 years agob=20645 fix potential stale-read.
Eric Mei [Fri, 8 Jan 2010 23:30:39 +0000 (16:30 -0700)]
b=20645 fix potential stale-read.

Due to race between read and truncate, read may populate cache pages outside
of the file size range, then next truncate-expand will see the stale page.
To fix this, we do a second vmtruncate at the end of truncation.

i=jay
i=robert

14 years agob=21656 Quiet dirty_max CERROR
Dmitry Zogin [Fri, 8 Jan 2010 16:52:31 +0000 (11:52 -0500)]
b=21656 Quiet dirty_max CERROR

The atomic_read() allowing the atomic_inc() are not covered
 by a lock. Thus they may safely race and trip this CERROR()
 unless we add in a small fudge factor (+1).

 i=adilger
 i=dmitry.zogin

14 years agob=19556 Ignore broken cancel_dirty_page() in OFED 1.4.1
Brian J. Murrell [Fri, 8 Jan 2010 14:42:59 +0000 (09:42 -0500)]
b=19556 Ignore broken cancel_dirty_page() in OFED 1.4.1

OFED 1.4.1 had a broken implementation of cancel_dirty_page for SLES10.
This patch detects that and ignores the function if found.

i=johann
i=yangsheng

14 years agob=20301 Fix mkfs.lustre for 16TB+ LUNs
Girish Shilamkar [Fri, 8 Jan 2010 11:19:55 +0000 (16:49 +0530)]
b=20301 Fix mkfs.lustre for 16TB+ LUNs

Mounting 16TB LUNs failed due to three bugs in mkfs.lustre. This patch
fixes this.

Patch by James Simmons.

a=simmonsja@ornl.gov
i=adilger
i=girish

14 years ago Prepare for Build 34 1.10.0.34 v1_10_0_34
Terry Rutledge [Fri, 8 Jan 2010 00:31:21 +0000 (17:31 -0700)]
Prepare for Build 34

      - Increment build number.

14 years agoRevert "b=19808 2.6.29-fc11 patchless client support"
Robert Read [Wed, 6 Jan 2010 02:05:45 +0000 (18:05 -0800)]
Revert "b=19808 2.6.29-fc11 patchless client support"

Causing kernel BUG in my testing, see bug 19808.

This reverts commit 95dbd166988cf0a2912c51305572c78a9aa2d324.

14 years agob=21519 fix race between ldlm_pools_thread and ldlm_space_free
Eric Mei [Tue, 5 Jan 2010 17:28:57 +0000 (10:28 -0700)]
b=21519 fix race between ldlm_pools_thread and ldlm_space_free

After ldlm_space_free dropped the refcount of a namespace to 0, it
may start to cleanup data structures in the namespace, while the
ldlm_pools_thread can still pick up the freeing namespace. This patch
is to instruct ldlm_pools_thread to skip namespace which is being
freed.

i=oleg
i=wangdi

14 years agob=21551 Ensure visible pages are sane.
Jay Xiong [Tue, 5 Jan 2010 07:47:53 +0000 (15:47 +0800)]
b=21551 Ensure visible pages are sane.

Revise the page initialization path so that if the page is able to be seen by
osc, it must be sane.

i=wangdi
i=eric.mei

14 years agob=19808 2.6.29-fc11 patchless client support
Rahul Deshmukh [Tue, 5 Jan 2010 07:17:35 +0000 (12:47 +0530)]
b=19808 2.6.29-fc11 patchless client support

Changes required to support fc11 patchless client on HEAD.

i=oleg.drokin
i=tappro

14 years agob=21433 Add debug information for quota_setdq/quota_chkdq.
Fan Yong [Tue, 5 Jan 2010 02:55:08 +0000 (10:55 +0800)]
b=21433 Add debug information for quota_setdq/quota_chkdq.

i=vitaly.fertman
i=tianzy

14 years agob=20591 fix race on obd_device::md_stats
LiuYing [Tue, 5 Jan 2010 00:42:18 +0000 (08:42 +0800)]
b=20591 fix race on obd_device::md_stats

call lprocfs_free_per_client_stats() before mdt_procfs_fini() frees
all the lprocfs data

i=liuying
i=wangdi
o=vitaly

14 years agob=21547 Move cascading_rw.c to correct location.
Robert Read [Mon, 4 Jan 2010 22:37:39 +0000 (14:37 -0800)]
b=21547 Move cascading_rw.c to correct location.

14 years agob=21547 add cascading_rw into lustre mpi tests
Elena Gryaznova [Thu, 24 Dec 2009 10:07:27 +0000 (13:07 +0300)]
b=21547 add cascading_rw into lustre mpi tests

o=James Simmons
i=Johann
i=grev
i=Robert Read

14 years agob=21670 output format specifiers fixes in LQS_DEBUG
Andrew Perepechko [Thu, 24 Dec 2009 14:10:10 +0000 (17:10 +0300)]
b=21670 output format specifiers fixes in LQS_DEBUG

i=Andreas Dilger

14 years agob=19808 : 2.6.29-fc11 patchless client support
Rahul Deshmukh [Thu, 24 Dec 2009 09:04:11 +0000 (12:04 +0300)]
b=19808 : 2.6.29-fc11 patchless client support

Changes need to fix to compile on gcc 4.4 (fc11)

i=tappro
i=vladimir.saveliev

14 years agob=20583 Do not create more than pre-created objects for sanity_27y
Fan Yong [Wed, 23 Dec 2009 01:41:29 +0000 (09:41 +0800)]
b=20583 Do not create more than pre-created objects for sanity_27y

i=robert.read
i=tom.wang

14 years agob=21454 Disable cl_page_in_io check for append write.
Jay Xiong [Wed, 23 Dec 2009 03:18:42 +0000 (06:18 +0300)]
b=21454 Disable cl_page_in_io check for append write.

This is because for append write, it's difficult to pass the uptodate file size to subio.

i=wangdi
i=eric.mei

14 years agob=18630 avoid the race of acquiring/releasing quota
Landen [Wed, 23 Dec 2009 07:52:25 +0000 (10:52 +0300)]
b=18630 avoid the race of acquiring/releasing quota

Before this patch, checking and computing the info of lqs
are located in two different functions(check_cur_qunit()
and schedule_dqacq()). It is racy. With this patch, they
will all happen in one function(check_cur_qunit()).

i=johann
i=panda

14 years agoRevert "b=21547 add cascading_rw into lustre mpi tests" 1.10.0.33a v1_10_0_33a
Robert Read [Tue, 22 Dec 2009 21:46:21 +0000 (13:46 -0800)]
Revert "b=21547 add cascading_rw into lustre mpi tests"

Build failure in cascade_rw.c.

This reverts commit f91373a85ceb65cb1ea8558b1da242a1770fc2a3.

14 years agoPrepare for Build 33 1.10.0.33 v1_10_0_33
Robert Read [Tue, 22 Dec 2009 19:35:20 +0000 (11:35 -0800)]
Prepare for Build 33

  - Update lustre/ChangeLog.
  - Update which_patch with correct kernels.
  - Increment build number.

14 years agob=21600 Fix unitialized pointer
Brian J. Murrell [Tue, 22 Dec 2009 18:29:12 +0000 (13:29 -0500)]
b=21600 Fix unitialized pointer

gcc warns that ctx may be used prior to assignment, so initialize it
at declaration

14 years agob=21547 add cascading_rw into lustre mpi tests
Elena Gryaznova [Tue, 22 Dec 2009 18:40:48 +0000 (10:40 -0800)]
b=21547 add cascading_rw into lustre mpi tests

o=James Simmons
 i=Johann
 i=grev

14 years agob=21544 t-f: support the dynamic switching GSS flavor
Elena Gryaznova [Tue, 22 Dec 2009 18:40:48 +0000 (10:40 -0800)]
b=21544 t-f: support the dynamic switching GSS flavor

i=Eric.Mei

14 years agob=19964 server-side extent lock for getattr
Vitaly Fertman [Tue, 22 Dec 2009 01:28:52 +0000 (04:28 +0300)]
b=19964 server-side extent lock for getattr

extend server-side locking used for punch only to other operations, getattr here

i=green
i=zam

14 years agob=19964 SOM EA
Vitaly Fertman [Tue, 22 Dec 2009 01:28:51 +0000 (04:28 +0300)]
b=19964 SOM EA

SOM stores proper size, blocks, IOEpoch and mountid in EA

14 years agob=19964 SOM cleanups, part2
Vitaly Fertman [Tue, 22 Dec 2009 01:28:50 +0000 (04:28 +0300)]
b=19964 SOM cleanups, part2

SOM cleanups, comments and some bugfixes

i=zam
i=vs

14 years agob=19964 SOM cleanups, part1
Vitaly Fertman [Tue, 22 Dec 2009 01:28:49 +0000 (04:28 +0300)]
b=19964 SOM cleanups, part1

SOM cleanups, comments and some bugfixes

i=tappro
i=vs

14 years agob=20098 mod use after free
Vitaly Fertman [Mon, 21 Dec 2009 20:39:26 +0000 (23:39 +0300)]
b=20098 mod use after free

add a refcount to md_open_data to control its usage in open-close and setattr-done-writing

i=green
i=tappro

14 years agob=20607 Use refcount on lov_request_set to prevent memory corruption.
Tom.Wang [Mon, 21 Dec 2009 13:05:46 +0000 (08:05 -0500)]
b=20607 Use refcount on lov_request_set to prevent memory corruption.

Correct the refcount of lov_request_set to make sure the set will only
be freed after all the references have been released.

i=adilger
i=ericm

14 years agob=15692 statahead should hold parent dir's i_mutex to synchronize with other operatio...
Fan Yong [Sat, 19 Dec 2009 16:08:46 +0000 (00:08 +0800)]
b=15692 statahead should hold parent dir's i_mutex to synchronize with other operations from VFS layer when inserting dentry to dcache

1) statahead should hold parent dir's i_mutex to synchronize with other operations from VFS layer when inserting dentry to dcache
2) drop unused ll_lookup_lock
3) drop unused DCACHE_LUSTRE_INVALID definition checking
4) more debug information
5) other code cleanup

i=robert.read
i=eric.mei

14 years agob=15977 Disable test 39 until it is fixed.
Robert Read [Sat, 19 Dec 2009 00:00:16 +0000 (16:00 -0800)]
b=15977 Disable test 39 until it is fixed.

This test produces are large number false positives because it is possible
for debug logs on some cpus being filled faster than other cpus. A alloc
message might appear on one cpu, but the freed message might be lost on the
second.

14 years agob=20722 Fix watchdog timer messages to be more clear.
dzogin [Fri, 18 Dec 2009 22:32:41 +0000 (17:32 -0500)]
b=20722 Fix watchdog timer messages to be more clear.

i=adilger

14 years agoRevert "b=19325 Adjust waiting extent locks on first enqeue."
Robert Read [Fri, 18 Dec 2009 23:15:49 +0000 (15:15 -0800)]
Revert "b=19325 Adjust waiting extent locks on first enqeue."

This reverts commit 6ded7c96c486ebcb069e9125d4fe94c1efddbf60.

Reverting because of issues with lock expansion.

14 years agoRevert "b=20645 Call vmtruncate again after truncate."
Robert Read [Fri, 18 Dec 2009 22:11:11 +0000 (14:11 -0800)]
Revert "b=20645 Call vmtruncate again after truncate."

This reverts commit 3bbe90f2827dd972dfb9bb5ca52bd3b60b553069.

Just hit the new assertion in racer, so backing this out for now.
http://buffalo.lustre.org:8080/display_report.pl?report_id=450093

14 years agob=17682 limit performance impact of rpctrace, dlmtrace & quota
dzogin [Fri, 18 Dec 2009 21:41:11 +0000 (16:41 -0500)]
b=17682 limit performance impact of rpctrace, dlmtrace & quota

Performance improvements for debug messages with
    D_RPCTRACE, D_LDLM, D_QUOTA

    i=alexey.lyashkov
    i=andrew.perepechko
    i=johann

14 years agob=21571 stacksize and locking fixes for loadgen patch from umka
Nathan Rutman [Fri, 18 Dec 2009 21:38:56 +0000 (13:38 -0800)]
b=21571 stacksize and locking fixes for loadgen patch from umka

i=adilger
i=nathan

14 years agob=16890 CERROR cleanups
Andrew Perepechko [Fri, 18 Dec 2009 20:38:59 +0000 (23:38 +0300)]
b=16890 CERROR cleanups

a whole lot of CERROR/CDEBUG cleanups

14 years agob=16186 Enable peer health detect for ksocklnd and ko2iblnd. i=isaac
Isaac Huang [Thu, 17 Dec 2009 23:12:19 +0000 (18:12 -0500)]
b=16186 Enable peer health detect for ksocklnd and ko2iblnd. i=isaac

14 years agob=21586 clarify lbuild error reporting
Brian J. Murrell [Thu, 17 Dec 2009 15:36:31 +0000 (10:36 -0500)]
b=21586 clarify lbuild error reporting

clarify the reporting of build errors by:
- disabling xtrace during the stack trace printing
- moving the ccache statistics printing exit trap so that it's only
  installed if ccache has been enabled
- remove some useless $() constructs around function calls
- only send the untrapped error debuggery to me

i=wangyb
i=yangsheng

14 years agob=17545 control DCACHE_LUSTRE_INVALID flag with MDS_INODELOCK_LOOKUP lock
Fan Yong [Thu, 17 Dec 2009 07:37:33 +0000 (15:37 +0800)]
b=17545 control DCACHE_LUSTRE_INVALID flag with MDS_INODELOCK_LOOKUP lock

1) "DCACHE_LUSTRE_INVALID" is controlled by "MDS_INODELOCK_LOOKUP" which is corresponding to "IT_LOOKUP", do not skip invalidate for other intent.
2) cleanup locks from former test cases for replay-single test_53d.

i=robert.read
i=jinshan.xiong

14 years agob=21593 sanityN test_33 print jbd stat fix [ext3|ext4]
Elena Gryaznova [Tue, 15 Dec 2009 13:57:39 +0000 (16:57 +0300)]
b=21593 sanityN test_33 print jbd stat fix [ext3|ext4]

i=Alex.Zhuravlev

14 years agob=20591 fix race on obd_device::md_stats
LiuYing [Thu, 17 Dec 2009 02:30:32 +0000 (10:30 +0800)]
b=20591 fix race on obd_device::md_stats

move md_stats out of export_stats_init() and cleanup mdt/mgs
procfs init/fini.

i=wangdi
i=liangzhen

14 years agoRevert "b=21527 Use %configure in the RPM SPEC file"
Robert Read [Thu, 17 Dec 2009 01:38:45 +0000 (17:38 -0800)]
Revert "b=21527 Use %configure in the RPM SPEC file"

Reverting due to build failure.

This reverts commit 98fd9942abbc7503fc9f1885e58b88887daa6e73.

14 years agob=21527 Use %configure in the RPM SPEC file
Brian J. Murrell [Wed, 16 Dec 2009 20:03:07 +0000 (15:03 -0500)]
b=21527 Use %configure in the RPM SPEC file

use the %configure macro defined by the RPM installation so that the
vendor's filesystem hierarchy preferences are observered

i=wangyb
i=yangsheng

14 years agob=16890 Reorder tests 29 and 30
Andrew Perepechko [Tue, 15 Dec 2009 20:22:55 +0000 (23:22 +0300)]
b=16890 Reorder tests 29 and 30

Put tests 29 and 30 in natural order

14 years agob=20645 Call vmtruncate again after truncate.
Eric Mei [Tue, 15 Dec 2009 18:13:12 +0000 (11:13 -0700)]
b=20645 Call vmtruncate again after truncate.

Due to race between read and truncate, read may populate cache pages outside of the file size range, then next truncate-expand will see the stale page. To fix this, we do a second vmtruncate at the end of truncation.

r=jay
r=rread

14 years agob=16466 add the support for journaled admin quota file
Landen [Tue, 15 Dec 2009 07:02:26 +0000 (15:02 +0800)]
b=16466 add the support for journaled admin quota file

i=tianzy
i=johann

14 years agob=21600 Fix unitialized pointer
Brian J. Murrell [Mon, 14 Dec 2009 21:03:16 +0000 (16:03 -0500)]
b=21600 Fix unitialized pointer

gcc warns that ctx may be used prior to assignment, so initialize it
at declaration

i=rread

14 years agoRemove lustrecvs. Not needed for git
Brian Reitz [Mon, 14 Dec 2009 06:49:09 +0000 (00:49 -0600)]
Remove lustrecvs.  Not needed for git

14 years agoIntroduce .gitignore files. GIT_EPOCH_MASTER
Brian Reitz [Sat, 12 Dec 2009 14:10:21 +0000 (08:10 -0600)]
Introduce .gitignore files.

The top level .gitignore file is new and is an attempt at
pulling in some of the common items that you might get for
free (by default)with CVS.  The other subdir/.gitignore files
are translated versions of their corresponding .cvsignore
file.  Because CVS does not descend into a subdir when
applying a ingore rule we have to prepend a "/" to
the pathname to get git to behave the same way.

14 years agoAdd ldiskfs/build symlink
Brian Reitz [Sat, 12 Dec 2009 14:06:45 +0000 (08:06 -0600)]
Add ldiskfs/build symlink

14 years agoBranch HEAD
robert.read [Fri, 11 Dec 2009 04:40:33 +0000 (04:40 +0000)]
Branch HEAD

Another test commit to ensure cvs is back to normal.

14 years agoBranch HEAD
robert.read [Fri, 11 Dec 2009 02:35:38 +0000 (02:35 +0000)]
Branch HEAD

Test commit to ensure cvs  is back to normal.

14 years agoBranch HEAD
robert.read [Fri, 11 Dec 2009 01:40:25 +0000 (01:40 +0000)]
Branch HEAD

Updating ChangeLog supported kernel section.

14 years agoUpdated for 2.0 build 32. v1_10_0_32
stinson1947 [Thu, 10 Dec 2009 17:30:44 +0000 (17:30 +0000)]
Updated for 2.0 build 32.

14 years ago Branch HEAD
dzogin [Thu, 10 Dec 2009 13:52:52 +0000 (13:52 +0000)]
 Branch HEAD
 b=21259
 i=andrew.perepechko
 i=alexey.lyashkov
----------------------------------------------------------------------
 Modified Files:
  lustre/ChangeLog lustre/obdclass/linux/linux-module.c
----------------------------------------------------------------------
Description: Allow non-root access for "lfs check".
Details    : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.

14 years agob=19689
brian [Thu, 10 Dec 2009 12:59:39 +0000 (12:59 +0000)]
b=19689
i=adilger
i=bobijam
i=nathan
o=Jim Garlick

Change tunefs.lustre and mkfs.lustre --mountfsoptions so that
exactly the specified mount options are used.
Leaving off any "mandatory" mount options is an error.
Leaving off any default mount options causes a warning, but is allowed.
Change errors=remount-ro from mandatory to default.
Sanitize the mount string before storing it.
Update man pages accordingly.

This hopefully makes two awkward situations less so:
- making errors=panic the default (before we had to append errors=remount-ro
  and hope ldiskfs parsing caused the last option to override the first)
- setting other mount options on the OST's dropped mballoc,extents
  without warning.

14 years agob=20237
grev [Thu, 10 Dec 2009 12:40:19 +0000 (12:40 +0000)]
b=20237
i=Robert.Read
more tests gather logs changes

14 years agob=21115
grev [Thu, 10 Dec 2009 12:21:09 +0000 (12:21 +0000)]
b=21115
i=Robert.Read
mount the same config on all clients

14 years agob=21115
grev [Thu, 10 Dec 2009 12:06:26 +0000 (12:06 +0000)]
b=21115
i=Robert.Read
mount the same config on all clients

14 years agoBranch HEAD
tianzy [Thu, 10 Dec 2009 06:07:03 +0000 (06:07 +0000)]
Branch HEAD
When a reply is wrong, lquota will try to remove the qunit in order
to avoid other threads hanging on it.
b=21372
i=yong.fan

14 years agob=20853,i=isaac:
isaac [Thu, 10 Dec 2009 05:15:35 +0000 (05:15 +0000)]
b=20853,i=isaac:
- Mark portal 63 as reserved for Cray DVS.

14 years agob=20580
grev [Wed, 9 Dec 2009 21:05:40 +0000 (21:05 +0000)]
b=20580
i=Johann
skip test 78 for local OST