Whamcloud - gitweb
cvs2svn [Fri, 5 Mar 2004 03:52:39 +0000 (03:52 +0000)]
This commit was manufactured by cvs2svn to create branch 'unlabeled-1.2.4'.
adilger [Fri, 5 Mar 2004 03:52:30 +0000 (03:52 +0000)]
Land b1_2 onto HEAD (20040304_171022)
b=2733, b=2773, b=2529, b=2730, b=2819, b=2814, b=2822, b=1450, b=2676, b=2681
b=2817, b=2706, b=2816, b=1987, b=2884, b=1191, b=2809, b=2765, b=2805, b=1972
morrone [Fri, 5 Mar 2004 01:30:50 +0000 (01:30 +0000)]
Merge of b1_2.
ericm [Thu, 4 Mar 2004 12:29:44 +0000 (12:29 +0000)]
file vanilla-2.6.3-nfs4 was initially added on branch b_v26.
ericm [Thu, 4 Mar 2004 11:07:43 +0000 (11:07 +0000)]
liblustre:
- don't build mpi programs by default, as Bob's suggestion
buffalo-pull [Thu, 4 Mar 2004 00:04:24 +0000 (00:04 +0000)]
- tagging RC_CURRENT
ericm [Wed, 3 Mar 2004 16:52:32 +0000 (16:52 +0000)]
liblustre:
- another inode drop fix
- fix a stupid error of last checkin
ericm [Wed, 3 Mar 2004 14:37:03 +0000 (14:37 +0000)]
liblustre:
- supress a warning of glimpse callback
- remove the debug code accidently checked in sometime before
ericm [Wed, 3 Mar 2004 14:33:55 +0000 (14:33 +0000)]
liblustre:
- don't blindly drop inode
- fix iget
- root inode checking
zab [Tue, 2 Mar 2004 23:12:35 +0000 (23:12 +0000)]
file build_check.h was initially added on branch b_cray_portals_merge.
zab [Tue, 2 Mar 2004 23:12:28 +0000 (23:12 +0000)]
file lustre_list.h was initially added on branch b_cray_portals_merge.
phil [Tue, 2 Mar 2004 17:22:56 +0000 (17:22 +0000)]
file llite_gns.c was initially added on branch b_gns.
ericm [Tue, 2 Mar 2004 15:00:53 +0000 (15:00 +0000)]
liblustre:
- b#1179: fix byteorder definition
- minor code cleanup in liblustre.h
ericm [Tue, 2 Mar 2004 14:56:14 +0000 (14:56 +0000)]
liblustre:
- fix broken readdir test
ericm [Tue, 2 Mar 2004 13:38:08 +0000 (13:38 +0000)]
liblustre:
- must update kms after write
yury [Tue, 2 Mar 2004 13:13:16 +0000 (13:13 +0000)]
Changes in vanilla-2.4.20 patch set. Added tmpfs related patches.
ericm [Tue, 2 Mar 2004 08:57:01 +0000 (08:57 +0000)]
liblustre build fix:
- follow the file size glimpse/probe change
buffalo-pull [Tue, 2 Mar 2004 03:03:30 +0000 (03:03 +0000)]
- tagging RC_CURRENT
zab [Tue, 2 Mar 2004 01:34:26 +0000 (01:34 +0000)]
file module.c was initially added on branch b_cray_portals_merge.
zab [Tue, 2 Mar 2004 01:34:21 +0000 (01:34 +0000)]
file list.h was initially added on branch b_cray_portals_merge.
zab [Tue, 2 Mar 2004 01:34:20 +0000 (01:34 +0000)]
file libcfs.h was initially added on branch b_cray_portals_merge.
phil [Mon, 1 Mar 2004 18:16:43 +0000 (18:16 +0000)]
Don't dereference NULL lock in error path
cvs2svn [Mon, 1 Mar 2004 15:35:42 +0000 (15:35 +0000)]
This commit was manufactured by cvs2svn to create branch 'unlabeled-1.2.4'.
yury [Mon, 1 Mar 2004 15:35:32 +0000 (15:35 +0000)]
Updated iopen patch for tmpfs.
hitao [Mon, 1 Mar 2004 15:20:48 +0000 (15:20 +0000)]
to incorporate change of lib_finalize function in portals/portals/lib-msg.c
hitao [Mon, 1 Mar 2004 15:15:19 +0000 (15:15 +0000)]
merge gmnal code in b_llp2 to HEAD
ericm [Mon, 1 Mar 2004 12:14:38 +0000 (12:14 +0000)]
file vallina-2.6.3-nfs4 was initially added on branch b_v26.
ericm [Mon, 1 Mar 2004 12:14:37 +0000 (12:14 +0000)]
file uml-patch-2.6.3-rc2-1.patch was initially added on branch b_v26.
ericm [Mon, 1 Mar 2004 12:14:36 +0000 (12:14 +0000)]
file uml-fix-2.6.3.patch was initially added on branch b_v26.
ericm [Mon, 1 Mar 2004 12:14:35 +0000 (12:14 +0000)]
file linux-2.6.3-nfs-intent.patch was initially added on branch b_v26.
ericm [Mon, 1 Mar 2004 12:14:34 +0000 (12:14 +0000)]
file linux-2.6.3-CITI_NFS4_ALL.patch was initially added on branch b_v26.
ericm [Mon, 1 Mar 2004 12:14:33 +0000 (12:14 +0000)]
file export-2.6.3.patch was initially added on branch b_v26.
buffalo-pull [Sat, 28 Feb 2004 00:10:13 +0000 (00:10 +0000)]
- tagging RC_CURRENT
adilger [Fri, 27 Feb 2004 22:18:21 +0000 (22:18 +0000)]
file branch.sh was initially added on branch b_smallfix.
eeb [Fri, 27 Feb 2004 19:07:14 +0000 (19:07 +0000)]
* Changed obdecho persistent object id to 1; object id 0 is now illegal.
* Fixed obdecho persistent object size at 1 MByte.
* Changed test_brw data verification tests to work on a fixed 4k block
rather than on page size to allow different page sized servers and
clients.
eeb [Fri, 27 Feb 2004 16:43:23 +0000 (16:43 +0000)]
* Applied the last patch in Bug 2306, which changes the portals router/NAL
forwarding interface from a single struct iovec for the message header +
payload, to passing the message header by pointer, and the message body
by ptl_kiov_t (== page,offset,size). This restores portals router
performance by allowing it to use socknal zero copy again.
* Applied the socknal connection load balancing fix from Bug 2817
buffalo-pull [Fri, 27 Feb 2004 01:11:08 +0000 (01:11 +0000)]
- tagging RC_CURRENT
phil [Thu, 26 Feb 2004 19:50:39 +0000 (19:50 +0000)]
b=2819
If the completion AST arrives before the enqueue reply, don't clobber
the newer LVB
b=2806
Remove unnecessary CERROR when we lose a glimpse vs. cancel race.
buffalo-pull [Thu, 26 Feb 2004 00:05:42 +0000 (00:05 +0000)]
- tagging RC_CURRENT
wangdi [Wed, 25 Feb 2004 16:15:00 +0000 (16:15 +0000)]
Update smfs: minor fixs in smfs_unlink
wangdi [Wed, 25 Feb 2004 09:18:29 +0000 (09:18 +0000)]
Update smfs:some minor fix in smfs
shaocheng [Tue, 24 Feb 2004 23:38:51 +0000 (23:38 +0000)]
update with current HEAD
wangdi [Tue, 24 Feb 2004 18:44:49 +0000 (18:44 +0000)]
Update smfs: 1)unlock cache inode after we call bottom fs methods 2)minor fix in symlink ops
phil [Tue, 24 Feb 2004 07:51:55 +0000 (07:51 +0000)]
some scripts were mysteriously deleted when b_size landed
phil [Tue, 24 Feb 2004 07:25:06 +0000 (07:25 +0000)]
Fix a fencepost I was pretty sure I had already fixed (I checked; the
other places were still fixed). NB: a lock on [x,y] protects a KMS of
up to y+1 bytes.
youfeng [Tue, 24 Feb 2004 06:56:16 +0000 (06:56 +0000)]
b 2295
r Adilger
move defination IOC_MDC_GETSTRIPE and declaration op_create_file() to lustre_user.h cause it is needed for user-space apps.
add a funcation int get_file_stripe(char *path, struct lov_user_md *lum) in liblustreapi.c, which is used by user apps want to get Lustre files' stripe info
adilger [Tue, 24 Feb 2004 02:35:15 +0000 (02:35 +0000)]
Fix RPM tag
adilger [Tue, 24 Feb 2004 01:47:59 +0000 (01:47 +0000)]
Land b_smallfix onto HEAD (20040223_1817)
adilger [Tue, 24 Feb 2004 01:11:59 +0000 (01:11 +0000)]
Update b_smallfix from HEAD (20040223_1758) (mostly b_size landing).
jacob [Tue, 24 Feb 2004 00:00:52 +0000 (00:00 +0000)]
r=phil
if mount.lustre is not installed, don't oops but print a message
suggesting it be checked.
cvs2svn [Mon, 23 Feb 2004 23:37:49 +0000 (23:37 +0000)]
This commit was manufactured by cvs2svn to create branch 'unlabeled-1.2.4'.
phil [Mon, 23 Feb 2004 23:37:30 +0000 (23:37 +0000)]
b=1021,2720
land b_size on HEAD for 1.2
phil [Mon, 23 Feb 2004 22:57:39 +0000 (22:57 +0000)]
merge HEAD into b_size, in preparation for landing
adilger [Mon, 23 Feb 2004 17:52:03 +0000 (17:52 +0000)]
Create ll_truncate_complete_page() compat macro.
wangdi [Mon, 23 Feb 2004 17:39:07 +0000 (17:39 +0000)]
update smfs 1)Add kml in smfs, but still not finish it, will change later according to llog. 2)some minor fixs
ericm [Mon, 23 Feb 2004 14:45:38 +0000 (14:45 +0000)]
port several patches to 2.6.3
ericm [Mon, 23 Feb 2004 14:14:50 +0000 (14:14 +0000)]
2.6 compile fix under obdfilter.
ericm [Mon, 23 Feb 2004 13:58:56 +0000 (13:58 +0000)]
module's init function will not get called when loading in 2.6.
wangdi [Mon, 23 Feb 2004 04:24:35 +0000 (04:24 +0000)]
get rid of some .h file we do not need in options.c
phil [Sun, 22 Feb 2004 04:19:28 +0000 (04:19 +0000)]
1. When one thread enqueues and another thread matches, the matching
thread can get the lock before the enqueueing thread updates the lock
value block. The correct fix is to do the granting inside of the
completion AST -- which can then be wrapped as appropriate to handle
the LVB mechanics -- but that's higher risk than I want right now.
Instead, I added a temporary LDLM_FL_CAN_MATCH flag, which gets set
only after the enqueueing thread is finished. Threads trying to match
will wait for this flag to be set after the lock is granted.
2. Fixes a fencepost error in the KMS calculation; a lock on [x, y]
actually protects a file of y + 1 bytes, not y bytes.
3. I spotted that ldlm_completion_ast might not call wake_up in
certain cases, but I think that's a theoretical concern. Fixed by
checking a different set of flags.
eeb [Sat, 21 Feb 2004 09:54:05 +0000 (09:54 +0000)]
* Applied fix for 1888
wangdi [Fri, 20 Feb 2004 13:59:39 +0000 (13:59 +0000)]
copy inode attribute to bottom inode in smfs_dirty_inode
jacob [Fri, 20 Feb 2004 08:44:33 +0000 (08:44 +0000)]
b=2774
r=phil
if portals_nid2str() is used twice in a single printf-type thing it
will end up printing the same ip both times.
this patch adds a second buffer so that it prints the correct things.
youfeng [Fri, 20 Feb 2004 06:03:58 +0000 (06:03 +0000)]
b_2074
r = Adilger
b 2074 comment 33 -- cleanup code using fstat()
adilger [Thu, 19 Feb 2004 21:51:06 +0000 (21:51 +0000)]
Update b_smallfix from HEAD (20040219_1143) (landing b_bug974, b_eq)
phil [Thu, 19 Feb 2004 15:55:41 +0000 (15:55 +0000)]
Some last b_size work:
- use the KMS in prepare_write
- remove now-useless ll_extent_lock_no_validate
- I noticed that lookup_it_finish still took whole-file locks when a getattr
comes in through lookup instead of revalidate; fixed.
- record the last time a lock was used: lock->l_last_used
- if we see a glimpse callback for a lock that hasn't been used in 10 seconds,
call both the glimpse and blocking AST callbacks
Repeated "ls -l /mnt/lustre/foo" then leads to:
[Write a file]
First ls, inode instantiation and glimpsing: 118 debug-log lines
[10 seconds pass]
Second ls, glimpse runs and blocking ast runs: 75 debug-log lines
Third ls, glimpse runs and returns a lock: 40 debug-log lines
Fourth ls, glimpse matches local lock: 19 debug-log lines
This can be condensed somewhat when "instant cancellation" is possible
for locks with no outstanding dirty pages to writeback.
wangdi [Thu, 19 Feb 2004 14:33:30 +0000 (14:33 +0000)]
Oops, Just forget to commit smfs_internal.h in the last commit
wangdi [Thu, 19 Feb 2004 14:24:03 +0000 (14:24 +0000)]
Update smfs 1. Use d_alloc to alloc the cache dentry 2. some minor fixs
buffalo-pull [Thu, 19 Feb 2004 03:46:05 +0000 (03:46 +0000)]
- tagging RC_CURRENT
rread [Wed, 18 Feb 2004 21:37:24 +0000 (21:37 +0000)]
b=2751
Don't do precreate orphan recovery on an OST if the MDS is in recovery
mode. The OSC's last_id info is not initialized until after MDS
recovery is complete, so clearing the orphans here would result in
removing all the objects on the OST.
adilger [Wed, 18 Feb 2004 20:07:54 +0000 (20:07 +0000)]
Fix endian conversion of socknal major/minor version. They are __u16 not __u32.
b=2464
wangdi [Wed, 18 Feb 2004 15:55:12 +0000 (15:55 +0000)]
1. Replace i_mapping with the bottom inode imapping in mmap method according to peter's advice. 2. some minor fix
adilger [Tue, 17 Feb 2004 23:31:25 +0000 (23:31 +0000)]
Increase the number of request buffers to 4096 from 1024 for a 4GB machine
so we can remove max_rpcs_in_flight workaround on ALC.
b=2729
wangdi [Tue, 17 Feb 2004 09:01:27 +0000 (09:01 +0000)]
add address method of smfs for mmap
jacob [Tue, 17 Feb 2004 04:03:18 +0000 (04:03 +0000)]
r=shaver
Fix configure option indenting.
wangdi [Mon, 16 Feb 2004 13:42:44 +0000 (13:42 +0000)]
copy bottom inode attribute out to the vfs inode
wangdi [Mon, 16 Feb 2004 12:33:48 +0000 (12:33 +0000)]
Update smfs: 1.fix bugs in smfs_link 2) decrease i_count in smfs_clear_inode
green [Mon, 16 Feb 2004 10:18:51 +0000 (10:18 +0000)]
Update to HEAD. Should now contain fix for 2710/2743
adilger [Mon, 16 Feb 2004 07:15:58 +0000 (07:15 +0000)]
Shouldn't have been merged.
adilger [Mon, 16 Feb 2004 06:56:45 +0000 (06:56 +0000)]
Add newline to ll_readpage() debug message.
adilger [Mon, 16 Feb 2004 00:33:18 +0000 (00:33 +0000)]
Re-add CVS helper scripts. Not sure how they were lost.
wangdi [Sun, 15 Feb 2004 15:50:02 +0000 (15:50 +0000)]
1. Adding iput(cache_inode) in smfs_clear_inode for clearing cache inode according to peter's advice. 2. other minor fixes in smfs_clear_inode
ericm [Sun, 15 Feb 2004 14:31:50 +0000 (14:31 +0000)]
remove the ancient workaround for liblustre in osc_announce_cached().
Andreas: I tested that OST will never grant more space to liblustre than
the initial 4M.
shaver [Sun, 15 Feb 2004 05:03:03 +0000 (05:03 +0000)]
b=2726: make sure that we properly balance page locking in the case of a failed
writepage.
patch=zab
nic [Sat, 14 Feb 2004 07:24:26 +0000 (07:24 +0000)]
update b_size from HEAD (20040213_2304)
buffalo-pull [Sat, 14 Feb 2004 03:35:41 +0000 (03:35 +0000)]
- tagging RC_CURRENT
cvs2svn [Sat, 14 Feb 2004 03:16:39 +0000 (03:16 +0000)]
This commit was manufactured by cvs2svn to create branch 'unlabeled-1.2.4'.
rread [Sat, 14 Feb 2004 03:16:22 +0000 (03:16 +0000)]
land b_eq on HEAD
adilger [Sat, 14 Feb 2004 02:34:34 +0000 (02:34 +0000)]
Update b_refinelocks from HEAD (20040213_1918) (b_bug974 mostly).
adilger [Sat, 14 Feb 2004 01:31:10 +0000 (01:31 +0000)]
Update b_eq from HEAD (20040213_1644) (b_bug974 landing).
Eric Mei, per your comment in osc_announce_cached(), we should only be sending
OBD_MD_FLGRANT if the client is caching data. If (AFAIK) liblustre isn't
doing any client-side write cache then it shouldn't be requesting any grant
from the OST. This isn't critical, since the OST _should_ stop giving out
grant when the client isn't consuming it, but it would be good to verify.
adilger [Fri, 13 Feb 2004 23:42:52 +0000 (23:42 +0000)]
Oops.
cvs2svn [Fri, 13 Feb 2004 23:18:00 +0000 (23:18 +0000)]
This commit was manufactured by cvs2svn to create branch 'unlabeled-1.2.4'.
adilger [Fri, 13 Feb 2004 23:17:53 +0000 (23:17 +0000)]
Landing b_bug974 onto HEAD (20040213_1538).
Adds support for client-side write cache accounting via OST write credits
(grants). There is not yet a "correction" algorithm for grants if they
get out of sync between the client and OST (OST can handle clients who think
they have grant but do not). Peter had previously suggested that grants
should decay over time (faster on the clients than on the OSTs) so idle
nodes do not consume space they aren't using.
Use /proc/fs/lustre/osc/OSC*/cur_grant_bytes (and existing cur_dirty_bytes)
to check client-side cache values.
Includes lustre-side changes for lfsck support (create files with
specific objids, create specific objids on OSTs).
Also includes some changes to the testing scripts (local.sh, lov.sh, acc-sm)
so that they include a "client nid '*'" client and can be used for sanityN.sh
testing. This has been put into the main acceptance-small.sh loop so that
sanityN.sh will be run on both single and multi-OST setups.
b=974 b=2349
r=shaver
adilger [Fri, 13 Feb 2004 21:32:19 +0000 (21:32 +0000)]
Land b_smallfix onto HEAD (20040213_1402)
adilger [Fri, 13 Feb 2004 18:00:39 +0000 (18:00 +0000)]
Or rather just old ones, since this bites me sometimes.
adilger [Fri, 13 Feb 2004 17:59:21 +0000 (17:59 +0000)]
Ignore all .mergeinfo files.
wangdi [Fri, 13 Feb 2004 14:28:46 +0000 (14:28 +0000)]
Update smfs: fix bugs of smfs_create
wangdi [Fri, 13 Feb 2004 07:23:00 +0000 (07:23 +0000)]
Update smfs: some fix about clear_inode in smfs
nic [Thu, 12 Feb 2004 20:13:10 +0000 (20:13 +0000)]
b=2618
r=shaver
get lconf to panic() when no host entry is found. using panic() will prevent
lconf from exiting cleanly when this happens.
adilger [Thu, 12 Feb 2004 19:55:25 +0000 (19:55 +0000)]
Update b_smallfix from HEAD (20040212_1234)