Whamcloud - gitweb
phil [Wed, 2 Mar 2005 23:19:29 +0000 (23:19 +0000)]
b=5637,4952
r=adilger
This patch tries to support all of the common VFS-enforced options like
nosuid, nodev, and noexec. There are some others that will require server-side
code -- notably read-only -- that will have to come later.
phil [Wed, 2 Mar 2005 23:16:57 +0000 (23:16 +0000)]
b=5779
There's one error path in the DLM's enqueue code where, after a timeout and
user abort, the client will keep its local copy of the lock.
This has a few follow-on effects, all of which should be fixed by this patch:
- the reference on the lock is never dropped, so we never try to cancel it, so
we never find out that our view of the lock state differs from the server's.
This could perhaps cause some corruption.
- we try to match this lock on future enqueues; although the lock is marked as
failed, search_queue is only checking for destroyed (bug). I don't know
precisely why we need two flags for this, but that's a more subtle change than
I'm willing to make right now.
- once we have a handle on that lock, the completion AST does check that flag,
so it returns an error right away -- but we don't check its return code in the
match path (bug) and plow on
- the lock enqueue was originally aborted before it got to the part that updates
the KMS and sets the LDLM_FL_CAN_MATCH flag. So each match attempt will wait
100 seconds for that flag to get set, which of course never happens. We should
print a pretty serious warning if that timeout happens, but fixes for the
previous two bugs should prevent us from getting here in the first place.
This has been running at NERSC for the last week, so I think it's ready for
more exposure.
phil [Wed, 2 Mar 2005 23:12:23 +0000 (23:12 +0000)]
r=adilger
HP submitted a patch to dynamically calculate the page size in lfs, but we
decided that it made more sense to hard code it to at least 64kB. That's
the largest page size that we could commonly come into contact with (ia64)
nic [Wed, 2 Mar 2005 22:15:52 +0000 (22:15 +0000)]
b=5800
extend the same affection shown to lconf to lmc. we have no favorite child
adilger [Wed, 2 Mar 2005 20:21:51 +0000 (20:21 +0000)]
Branch: b1_4
Fix build error if _LARGEFILE64_SOURCE not defined when lustre_user.h is
included. lov_user_mds_data isn't widely used, just silently drop it.
Quiet compiler warning.
r=jacob
adilger [Wed, 2 Mar 2005 14:05:52 +0000 (14:05 +0000)]
Land b1_4_smallfix onto b1_4 (20050302_0257)
- new delete_thread patch for 81chaos kernel (b=5669)
- use KIOBUF_GET_BLOCKS() for PPC kernel (from b1_4_bgl)
- add "instantaneous" rates to llstat.pl in addition to overall average rate
adilger [Wed, 2 Mar 2005 08:18:49 +0000 (08:18 +0000)]
Branch: b1_4
Add debugging for bug 5793 in case it happens again.
b=5793
phil [Tue, 1 Mar 2005 21:24:56 +0000 (21:24 +0000)]
b=5800
Info:
HP SFS 2250/CFS 5800
On em64t/x86_64 systems the %{_libdir} rpm path resolves to
/usr/lib64 rather than /usr/lib. This patch updates lconf to
look in both locations for the Lustre extension modules.
adilger [Tue, 1 Mar 2005 09:52:12 +0000 (09:52 +0000)]
Branch: b1_4
Disable extents/mballoc with old patch, but allow Lustre to build.
adilger [Tue, 1 Mar 2005 09:17:46 +0000 (09:17 +0000)]
Branch: b1_4
Move the MDC EA/cookie max size initialization later in the mount process,
until we have established our permanent connections. This avoids needless
a bunch of client->OST connect+disconnect RPCs, and also fixes a bug where
the client didn't get the reply for the first disconnect, and then hit a
sanity check (target handle mismatch) on the server.
This will tie in directly to the bug 5707 fix, as that will additionally
store a "default" EA size to avoid allocating and sending a very large
(and otherwise useless) request buffer, just to have a place to store
the LOV EA for a MDS open replay. Includes regression test.
b=5317
r=nathan (with minor fix to avoid the storage of lovdesc at all)
phil [Tue, 1 Mar 2005 00:20:14 +0000 (00:20 +0000)]
Omnibus lconf update, courtesy of HP.
lconf_write_conf_failover-cfs5111.patch
b=5111
Info:
HP SFS 1575/CFS 5111
Allow lconf --write_conf to handle stopping when the MDS device
started in recovery.
lconf_pidfile_fix-cfs4903.patch:
b=4903
Info:
HP SFS 1515, 1597/CFS 4903, 5091, 5452
Fix the my_int() routine to handle arguments that are already
integers correctly, and also handle octal values in the same way
as hexadecimal ones are handled, and handle non-numeric strings,
which will trigger TypeError exception, appropriately.
Partially address issues identified with starting up and shutting
down the acceptor daemon. For startup we attempt to deal with the
possibility of another lconf instance racing with us that is also
attempting to start the daemon. For shutdown we wait for up to 15
seconds for the daemon to shut down properly before continuing.
lconf_validate_upcall-sfs1487.patch:
Info:
HP SFS 1487
Fixes erroneous comment in LCTLInterface class.
Adds validation check to make sure that the specified upcall
exists and is executable if appropriate.
lconf_whitespace_fixup.patch:
Info:
Fix up white space on all lines to be consistent as part of the
lustre 1.4.0 merge.
phil [Tue, 1 Mar 2005 00:12:35 +0000 (00:12 +0000)]
b=2811
HP's utils_pagesize_issues.patch:
Info:
HP SFS 1678
Enhance the lfs command to warn if stripe size is not an exact
multiple of page size.
HP SFS 211/CFS 2811
Fix the lctl command to use dynamically calculated page size
rather than compiled in page size setting. This means the command
doesn't need to be rebuilt/re-installed if we reboot on a kernel
with a different page size.
nic [Tue, 1 Mar 2005 00:02:25 +0000 (00:02 +0000)]
b=4481
revert patch, it is causing tests to fail
adilger [Mon, 28 Feb 2005 23:12:47 +0000 (23:12 +0000)]
Branch: b1_4
Add optional path for DBENCH client scripts.
Change test 72 to only run for kernel 43 or higher.
nic [Mon, 28 Feb 2005 18:35:18 +0000 (18:35 +0000)]
fix typo
nic [Mon, 28 Feb 2005 18:04:25 +0000 (18:04 +0000)]
b=5585
backout both patches for this bug as it is breaking conf-sanity
nkj [Mon, 28 Feb 2005 14:27:20 +0000 (14:27 +0000)]
fix to problem introduced by patch for bug #5585
adilger [Mon, 28 Feb 2005 13:24:45 +0000 (13:24 +0000)]
Branch: b1_4
Add export-zap-page-range.patch for HP/PNNL kernel.
b=5780
nkj [Sun, 27 Feb 2005 16:45:38 +0000 (16:45 +0000)]
fix for bug #5585 - and syntax error fix (missing newline)
nkj [Sun, 27 Feb 2005 16:22:52 +0000 (16:22 +0000)]
fix for bug #4481
adilger [Sun, 27 Feb 2005 00:20:49 +0000 (00:20 +0000)]
Branch: b1_4
Need to also allow building against an updated kernel version.
phil [Sat, 26 Feb 2005 19:44:24 +0000 (19:44 +0000)]
A think-o that Andreas didn't catch; really change version to 43
green [Sat, 26 Feb 2005 14:18:41 +0000 (14:18 +0000)]
b=5773
added to rhel4 serie: export-show_task-2.6-vanilla.patch and remove-suid-2.6-suse
vfs_intent-2.6-rhel4.patch: got rid of IT_CHDIR, moved LOOKUP_LAST and LOOKUP_NOT_LAST to not clash with LOOKUP_OPEN, incorporated fix for bug 4134
export_symbols-2.6-rhel4: provided declaration of filemap_populate.
phil [Sat, 26 Feb 2005 05:57:33 +0000 (05:57 +0000)]
b=5654
Landing the purely mechanical parts of bug 5654, to make the patch smaller for
inspection.
Change obd_increase_kms to obd_adjust_kms to allow growing or shrinking the
kms.
phil [Sat, 26 Feb 2005 05:49:00 +0000 (05:49 +0000)]
remove the last vestiges of LLI_F_PREFER_EXTENDED_SIZE
phil [Sat, 26 Feb 2005 04:19:04 +0000 (04:19 +0000)]
b=5695
Oleg's patch fixes the kernels to call setattr_raw inside remove_suid.
In 2.4, remove_suid just did nothing.
In 2.6, it would LBUG when notify_change was called instead.
phil [Sat, 26 Feb 2005 03:04:22 +0000 (03:04 +0000)]
b=5733,5638
Alex's patches to fix small bugs in extents and mballoc:
1) extents-related fixes:
1) callback's API used in ext3_ext_walk_space() changes a bit to
reflect that callback can be given >2^16 extent len (hole)
2) fsfilt_ext3 has changed to use updated callback API
3) minor race in ext3_ext_new_extent_cb() fixed
2) mballoc-related fixes:
1) free space searching has changed to be more smart
2) three possible races have been fixed
3) lots of minor fixes
4) mballoc doesn't regenerate buddies in clean umount case
phil [Sat, 26 Feb 2005 03:00:07 +0000 (03:00 +0000)]
b=5687,5210
A workaround for the crash in ll_unhash_aliases on 2.6.
The 2.6 lookup code is now lockless (thanks to the RCU machinations), which
allows it to race with the unhashing in ll_unhash_aliases. The 2.6 __d_drop
doesn't leave the list head ->pprev in a safe state, so the next pass through
this loop (which tries to re-hash it after we lose the dput race), would crash.
For the record, it looks like the kernel's own d_prune_aliases is broken in
the very same way, except that instead of crashing immediately, it just leaves
the d_hash pointers poisoned for someone else to crash on.
phil [Sat, 26 Feb 2005 02:53:54 +0000 (02:53 +0000)]
After a backend disk failure, we somehow got called in dio_complete_routine,
with what looked like a NULL bio->bi_private.
Add some debugging, in case we ever see this again.
phil [Sat, 26 Feb 2005 02:50:01 +0000 (02:50 +0000)]
Don't hold the pinger_sem in ptlrpc_pinger_sending_on_import.
It's an inherently racy operation in the first place, and until the pinger
thread is fixed not to hold the pinger_sem while it's waiting for an upcall
to complete, it prevents ptlrpcd from needlessly blocking.
phil [Sat, 26 Feb 2005 02:21:43 +0000 (02:21 +0000)]
b=5702
r=jacob
Scali's patch to fix a few problems with the init scripts on SLES9:
- Sources "/etc/init.d/functions" (doesn't exist on SLES9)
- Uses a function called "action" (doesn't exist on SLES9)
- The scripts don't have LSB-compliant comments to define what runlevels to
run on, dependencies on external subsystems and so on.
nic [Fri, 25 Feb 2005 21:38:23 +0000 (21:38 +0000)]
abort with -1 to get a non zero exit code
green [Fri, 25 Feb 2005 13:18:45 +0000 (13:18 +0000)]
Updates kernel patcheds for rhel4 kernel (2.6-rhel4 serie). Kernel compiles ok.
Starts updating ldiskfs serie for it too, extents and mballoc2 are remain to be
done for now
adilger [Thu, 24 Feb 2005 23:58:12 +0000 (23:58 +0000)]
Branch: b1_4
Be clearer when printing out the default directory striping pattern.
adilger [Thu, 24 Feb 2005 23:55:36 +0000 (23:55 +0000)]
Branch: b1_4
Quiet verbose and mostly useless message (printed loc_llcd == NULL always).
adilger [Thu, 24 Feb 2005 23:53:56 +0000 (23:53 +0000)]
Branch: b1_4
Allow mtime/atime to be set backward.
b=4829, b=5669
adilger [Thu, 24 Feb 2005 23:27:17 +0000 (23:27 +0000)]
Branch: b1_4
Allow mtime/atime to be set backward.
b=4829, b=5669
adilger [Thu, 24 Feb 2005 22:36:51 +0000 (22:36 +0000)]
Branch: b1_4
Add newlines to VMA_DEBUG messages in ll_vm_{open,close} to avoid console spew.
adilger [Thu, 24 Feb 2005 19:31:08 +0000 (19:31 +0000)]
Branch: b1_4
Hold ns lock when calling ldlm_del_waiting_lock() from ldlm_handle_ast_error().
This was detected by Phil's recent addition of "l_check_ns_lock()" therein.
b=5746
adilger [Thu, 24 Feb 2005 19:21:10 +0000 (19:21 +0000)]
Branch: b1_4
Fix bogus LASSERT when NFS exporting a filesystem.
This is for a structure we don't even use currently.
b=4655, 5760
r=green (original analysis + fix)
nic [Thu, 24 Feb 2005 16:35:41 +0000 (16:35 +0000)]
init the mpi layer earlier to prevent unknown options when run with a real mpi
layer. mpi init taken from write_append_truncate
eeb [Thu, 24 Feb 2005 10:27:10 +0000 (10:27 +0000)]
* Added network type 'ra' to comments/help strings in lmc
eeb [Thu, 24 Feb 2005 10:19:56 +0000 (10:19 +0000)]
* Added missing NALs to failure handling case of zeroconf ADD_PEER
adilger [Wed, 23 Feb 2005 23:33:03 +0000 (23:33 +0000)]
Branch: b1_4
Fix grammar.
adilger [Wed, 23 Feb 2005 23:25:37 +0000 (23:25 +0000)]
Branch: b1_4
Something I noticed when looking at the OST AMD patch. We should check
the "ost" index and not the "stripe" index.
eeb [Wed, 23 Feb 2005 21:43:14 +0000 (21:43 +0000)]
* First cut working vibnal
eeb [Wed, 23 Feb 2005 21:38:00 +0000 (21:38 +0000)]
* Added zeroconf ADD_PEER support for RANAL and VIBNAL
nic [Fri, 18 Feb 2005 23:52:39 +0000 (23:52 +0000)]
land changes from b_release_1_4_0_10
adilger [Fri, 18 Feb 2005 23:40:03 +0000 (23:40 +0000)]
Branch: b1_4
Minor fix to the MDS default stripe count->inode size calculations.
adilger [Fri, 18 Feb 2005 23:39:05 +0000 (23:39 +0000)]
Branch: b1_4
Ignore generated files.
adilger [Thu, 17 Feb 2005 18:01:09 +0000 (18:01 +0000)]
Land b1_4_smallfix onto b1_4 (20050210_1537)
Fix build problem for older kernels which don't have i_alloc_sem.
Use MDS_{UP,DOWN}_{READ,WRITE}_ORPHAN_SEM() to lock the MDS orphan
information to avoid a confusing name conflict.
b=5700
adilger [Wed, 16 Feb 2005 04:32:06 +0000 (04:32 +0000)]
Branch: b1_4
Add missing ChangeLog entry (bug not known).
r=eeb (original fix)
jacob [Tue, 15 Feb 2005 21:31:13 +0000 (21:31 +0000)]
b=5684
r=adilger
First stab at improving some user visible error messages.
jacob [Tue, 15 Feb 2005 21:25:04 +0000 (21:25 +0000)]
Add some comments so other people don't get fooled into thinking there was a bug here as i did.
jacob [Mon, 14 Feb 2005 19:32:22 +0000 (19:32 +0000)]
Add patch to fix running with 2.6 host systems.
eeb [Mon, 14 Feb 2005 15:30:14 +0000 (15:30 +0000)]
* Added loopback optimisation to lib-move.c
lctl --net ??? loopback # show current state
lctl --net ??? loopback on # enable loopback optimisation
lctl --net ??? loopback off # disable loopback optimisation
By default loopback optimisation is turned OFF
phil [Mon, 14 Feb 2005 05:52:52 +0000 (05:52 +0000)]
Andreas caught a think-o in the liblustre half of this checkin
phil [Sat, 12 Feb 2005 23:15:20 +0000 (23:15 +0000)]
brown paper bag bug in the lock cleanups; I'm amazed this passed as many tests
as it did, but here we are.
phil [Sat, 12 Feb 2005 23:03:41 +0000 (23:03 +0000)]
b1_4_mmap landing didn't update the ChangeLog (tsk tsk)
phil [Sat, 12 Feb 2005 22:59:54 +0000 (22:59 +0000)]
b=5628
r=jacob
- add a per-mdc/osc /proc file that lets you do a Lustre ping
- the old way, used by lfs check, would block when the MDS is down, because it
opened the mountpoint
- also exported it via liblustreapi as llapi_ping(obd_type, obd_name)
phil [Sat, 12 Feb 2005 21:32:17 +0000 (21:32 +0000)]
b=5656
Clean up a lot of calls to LDLM_DEBUG that weren't protected by an ns_lock.
This is probably not a comprehensive fix, but addresses all of the violations
that I saw.
phil [Sat, 12 Feb 2005 21:03:22 +0000 (21:03 +0000)]
Add an l_check_ns_lock to go with l_check_no_ns_lock.
phil [Sat, 12 Feb 2005 21:00:20 +0000 (21:00 +0000)]
remove the last vestiges of test_ldlm from {l,obd}ctl
phil [Sat, 12 Feb 2005 20:53:59 +0000 (20:53 +0000)]
Updated config files from 2.4.21-37EL, which enable a handful of new modules
(at least on x86_64)
phil [Sat, 12 Feb 2005 19:23:03 +0000 (19:23 +0000)]
bump the version to 1.4.0.10 for 1.4.1 rc1
phil [Sat, 12 Feb 2005 06:25:35 +0000 (06:25 +0000)]
A customer mentioned that we didn't have a get_hostaddr method for the LDAP
backend. LDAP is on its way out of favour, so I'm hardly shocked.
nic [Fri, 11 Feb 2005 23:42:08 +0000 (23:42 +0000)]
land b1_4_mmap on b1_4 (20050211_1829)
adilger [Thu, 10 Feb 2005 23:43:50 +0000 (23:43 +0000)]
Land b1_4_smallfix onto b1_4 (20050210_1537)
- fix incorrect LASSERT in mds_getattr_name (5635)
- fix configure check for grab_cache_page_nowait_gfp
- fix lsattr -v for directories so we can get inode generation from userspace
- fix mtime return from glimpse AST reply again (4829)
phil [Thu, 10 Feb 2005 23:33:09 +0000 (23:33 +0000)]
- BUGS should just point at bugzilla
- nodist doesn't need to reference obdext2 anymore
- rename COPYING to LICENSE and make it somewhat clearer
- remove the unused mgmt/ module; we know how to get it back.
phil [Thu, 10 Feb 2005 23:31:40 +0000 (23:31 +0000)]
remove the lustre book-building bits from Makefile.am, in preparation for
removal from the tree.
r=jacob
phil [Thu, 10 Feb 2005 05:55:19 +0000 (05:55 +0000)]
b=5277
r=adilger
Andreas's patch to avoid starting duplicate acceptors on servers with multiple
IP hostaddrs.
phil [Wed, 9 Feb 2005 19:50:22 +0000 (19:50 +0000)]
b=5544
A checkin was made on 17 Jan to move lov_get_info from llite to obdclass.
But here it is, still in llite, which is causing depmod to cry foul about
recursive dependencies with the lov.
phil [Sat, 5 Feb 2005 06:34:04 +0000 (06:34 +0000)]
b=5492
Back out the KMS locking change; I saw a problem with it yesterday, and the
new truncate buffalo failures confirmed my suspicion.
phil [Fri, 4 Feb 2005 19:55:46 +0000 (19:55 +0000)]
At some point, somebody added a 5th upcall argument, but didn't print it
with the rest of them.
phil [Fri, 4 Feb 2005 19:30:22 +0000 (19:30 +0000)]
b=5654
The lock ordering of i_sem and i_alloc_sem changed between 2.4 and 2.6,
which causes deadlocks if multiple threads try to truncate a single file.
I don't really like this solution, but it's the best I can come up with
right now.
eeb [Fri, 4 Feb 2005 18:49:21 +0000 (18:49 +0000)]
* Landed portals:b_port_step as follows...
- removed CFS_DECL_SPIN*
just use 'spinlock_t' and initialise with spin_lock_init()
- removed CFS_DECL_MUTEX*
just use 'struct semaphore' and initialise with init_mutex()
- removed CFS_DECL_RWSEM*
just use 'struct rw_semaphore' and initialise with init_rwsem()
- renamed cfs_sleep_chan -> cfs_waitq
cfs_sleep_link -> cfs_waitlink
- fixed race in linux version of arch-independent socknal
(the ENOMEM/EAGAIN decision).
- Didn't fix problems in Darwin version of arch-independent socknal
(resetting socket callbacks, eager ack hack, ENOMEM/EAGAIN decision)
- removed libcfs types from non-socknal header files (only some types
in the header files had been changed; the .c files hadn't been
updated at all).
- Updated lustre b1_4 to match
jacob [Thu, 3 Feb 2005 19:49:20 +0000 (19:49 +0000)]
ignore _netdev as well
jacob [Thu, 3 Feb 2005 17:09:23 +0000 (17:09 +0000)]
this is upstream now
jacob [Thu, 3 Feb 2005 16:58:13 +0000 (16:58 +0000)]
fix automake warning
jacob [Wed, 2 Feb 2005 23:29:07 +0000 (23:29 +0000)]
b=5383
r=phil
allow readahead to fail when low on memory.
nic [Wed, 2 Feb 2005 23:13:21 +0000 (23:13 +0000)]
land b1_4_smallfix on b1_4(20050202_1817)
- readahead could read extra page into cache that wasn't ejected (5388)
- prevent races in class_attach/setup/cleanup/detach (5260)
- don't dereference de->d_inode after l_dput of de (5458)
- use "int" for stripe value returned from lock_to_stripe (5544)
- mballoc allocation and error-checking fixes in 2.6 (5504)
- block device patches to fix I/O request sizes in 2.6 (5482)
- protect KMS changes with its own lock, not i_sem (5492, 5624, 3453)
- look up hostnames for IB nals (5602)
- fix mount usage message, return errors per mount(8) (5168)
- remove ancient LOV_MAGIC_V0 EA support (5047)
- add "disk I/Os in flight" and "I/O req time" stats in obdfilter
jacob [Wed, 2 Feb 2005 04:17:16 +0000 (04:17 +0000)]
update to SP1 kernel: 2.6.5-7.141
eeb [Tue, 1 Feb 2005 11:58:00 +0000 (11:58 +0000)]
* ranal debugging and lconf support
eeb [Mon, 31 Jan 2005 12:30:33 +0000 (12:30 +0000)]
* 5602 fix improves checks that NID is set correctly and causes incorrect
configurations to fail rather than trip a server assertion.
* 5606 fix ensures both peers agree on maximum "immediate" get.
* ranal updates
nkj [Fri, 28 Jan 2005 20:22:15 +0000 (20:22 +0000)]
Fix for bug 5511
nkj [Fri, 28 Jan 2005 15:30:09 +0000 (15:30 +0000)]
corrected filename
nic [Thu, 27 Jan 2005 00:38:22 +0000 (00:38 +0000)]
b=5594
remove hardcoded path to llmount that breaks rpm based testing
nkj [Wed, 26 Jan 2005 12:44:13 +0000 (12:44 +0000)]
landing patch for bug 5272 - making copy of list instead of using reference
nic [Wed, 19 Jan 2005 22:39:00 +0000 (22:39 +0000)]
b=5570
shorten name of suse drop to something we can use with lbuild and buffalo testing
eeb [Tue, 11 Jan 2005 03:37:38 +0000 (03:37 +0000)]
* Made openib not use the subnet manager to discover connection parameters
+ Openib listens on a TCP/IP port for service queries and responds with the
service id, port GID and pkey.
+ Openib peer table entries have become (NID, IP, port) tuples, where IP
and port connect to the peer's service query server.
+ /proc interfaces for port, listener timeout and backlog (restarts kernel
listener on update).
+ lmc/lconf support for new openib peers, including only running acceptor
if net == 'tcp'
+ Changed connection daemon to conduct both sides of the service query as
well as IB connection establishment. Spawning several since tcp/ip reads
can block for a timeout.
+ Added a reaper thread to do connection cleanup and timeout checks (the
single connection daemon used to do that too).
* Removed some unused lconf default constants
* Better openib automagic compilation check.
* Consistent usage (flipping + optional checksum) of all openib messages
(service query, connection requests and "normal" message flow).
* Exhaustive openib "are-you-still-the-same-person" checks
* Fixed bug which stopped idle persistent peers from getting removed from the
peer table.
* Fixed some ranal bugs which became obvious when similar problems were
debugged in openibnal
phil [Mon, 10 Jan 2005 21:51:53 +0000 (21:51 +0000)]
b=5445
Fix the x86-64 large-stack RHEL3 patch. The old patch would allocate 1 page
in alloc_task_struct, then free THREAD_ORDER pages in free_task_struct.
eeb [Fri, 7 Jan 2005 12:41:05 +0000 (12:41 +0000)]
* Included changes from *old* b1_4/b_cray
* OBD_FL_DEBUG_CHECK for consistent obdecho end-to-end checking
* struct echo_obd cleanup
* obdecho support for "difficult" reply testing (test_setattr)
* replaced cowardly eap_from_cookie() usage with EAP_FROM_COOKIE()
which asserts it is being called correctly
* bracketed echo_client_iocontrol() guts with unlock/lock_kernel()
for increased concurrency during echo ioctl execution
* better thread control and stats in obd.c
* n-threads-over-m-objects test_brw support (required for
obdfilter-survey)
* openibnal changes
* re-enabled openib on 2.5/2.6 builds
configure --with-openib=no can be used to disable openib explicitly if
required.
* fixed some 64 bit casts
* just use kernel's page_to_phys()
* fixed CERROR format problem in libcfs/module.c
buffalo-pull [Wed, 29 Dec 2004 23:12:57 +0000 (23:12 +0000)]
- keep version in line with latest tarball available on the FTP site;
- keep ia64 configs up-to-date with same tarball
jacob [Tue, 21 Dec 2004 21:39:32 +0000 (21:39 +0000)]
Update series to U4 kernel (2.4.21-27.EL)
adilger [Tue, 21 Dec 2004 09:43:56 +0000 (09:43 +0000)]
Land b1_4_smallfix onto b1_4 (20041221_0116)
- fix 2.6 build problems
- fix mount usage message, return errors per mount(8)
- change grep [] to grep "[]" in tests so they work in more UMLs
jacob [Mon, 20 Dec 2004 23:55:00 +0000 (23:55 +0000)]
Fix non-module build
phil [Mon, 20 Dec 2004 23:08:38 +0000 (23:08 +0000)]
Remove obd_ptlbd.h, and fix "make dist"
jacob [Mon, 20 Dec 2004 22:54:27 +0000 (22:54 +0000)]
add bigstack support to rhel 2.4.21
phil [Mon, 20 Dec 2004 22:52:15 +0000 (22:52 +0000)]
Remove one missed smfs/snapfs bit from lustre-core.m4
phil [Mon, 20 Dec 2004 22:49:36 +0000 (22:49 +0000)]
removing unused b1_4 code modules: cobd, ptlbd, smfs, snapfs
I'll send a separate mail to eng, but BE CAREFUL WHEN MERGING.
Merging into b1_2 or b_cray is fine, but be careful when you merge
into HEAD! If you want my help, please ask.