Whamcloud - gitweb
braam [Tue, 8 Jul 2003 22:41:21 +0000 (22:41 +0000)]
- we could call this "merely a merge from b_devel". Notes:
- lprocfs is not managable
- this oopses on mount, will be fixed soon.
zab [Fri, 27 Jun 2003 21:06:52 +0000 (21:06 +0000)]
- include version.h before using KERNEL_VERSION
- have ll_ts2u64 in 2.4 also take a pointer to the time in question
- ifdef out some 2.5 cache paths so 2.4 isn't sad
- fix up remenants of llwp's move to a properly allocated structure
- add some missing break;s to the end of default: blocks
zab [Fri, 27 Jun 2003 05:36:27 +0000 (05:36 +0000)]
- remove the empty parameters to CDEBUG_STACK to appease ancient gcc that
I was accidentally using.
- we're still #including kernel headers in userspace (sigh), 2.5 needs to
play along. It would be great to clean up these includes.
eeb [Mon, 23 Jun 2003 19:07:05 +0000 (19:07 +0000)]
* Changed socknal to cope with sk_xxxxx
* Removed spurious #define of smp_processor_id() in kp30.h
* Tried to get everything to compile under 2.4 as well as 2.5 but
retired hurt
eeb [Wed, 11 Jun 2003 18:05:10 +0000 (18:05 +0000)]
* Merged latest b_devel
* **** NOT TESTED ****
zab [Tue, 10 Jun 2003 18:53:55 +0000 (18:53 +0000)]
b=1349: always perform the final socket fput() in the reaper context.
in_interrupt() isn't the right test on 2.5, so we'll just
toss it instead of reserving the right to get it wrong again.
rread [Thu, 5 Jun 2003 04:12:57 +0000 (04:12 +0000)]
merge from b_devel -> b_ad
DEVEL_AD_PARENT_20030526 -> DEVEL_AD_PARENT_20030605
Large, squirrelly merge. It mounts, and acceptance_small.sh finishes
the test and then fails in cleanup, just like b_devel.
amrutjoshi [Tue, 3 Jun 2003 18:25:02 +0000 (18:25 +0000)]
1) changes to debug msg format (bug 1387)
2) parser changes for llanalyze
3) llanalyze -logfile=filename
amrutjoshi [Tue, 3 Jun 2003 13:08:16 +0000 (13:08 +0000)]
1) changed debug mesg format.
2) changed llanalyze parser to parse new format
3) changed llanalyze to show different colors for entry and exit.
4) give log filename as 1st arg to llanalyze e.g. "llanalyze FILENALE".
5) give console column width as 2nd arg to llanalyze so that the output doesnt wrap around.
alex_tomas [Sun, 1 Jun 2003 08:41:02 +0000 (08:41 +0000)]
- __exit modificater has been commented out. now b_ad compiles w/o UML
braam [Sun, 1 Jun 2003 02:24:43 +0000 (02:24 +0000)]
- add the tests directory to the compile
zab [Tue, 27 May 2003 01:25:08 +0000 (01:25 +0000)]
get b_ad compiling after the b_devel merge.
- move the possibly broken waitfor_one_page to one place
- o_mtime is seconds, that's what ll_ts2u64 returns. s/%Lu/LPU64/
- get rid of client_tgt2uuid2obd copies, export it from obdclass
- move client_obd_{setup,cleanup} into obdclass, export, kill dupes
- remove empty ptlrpc_lib
- get rid of random sanobd dupe in lov
- hoist ll_alloc_brw out of a kernel version ifdef
- no, seriously, unsigned long flags.
zab [Mon, 26 May 2003 22:50:00 +0000 (22:50 +0000)]
- bring recent b_devel changes into b_ad, only minor corrections needed. the
only significant of which was making sure changes in super.c made it to
b_ad's llite_lib.c
braam [Mon, 26 May 2003 08:55:36 +0000 (08:55 +0000)]
- add llite to 2.5 compile chain.
- start abstracting super.c commonality in llite_lib.c
- fix obvious problems
- start restoring the vfs_intent patch for 2.5 to one that hinges on
path_release not the sprinkle of intent_release stuff from calsoft.
braam [Mon, 26 May 2003 05:28:13 +0000 (05:28 +0000)]
- automatically set up a 2.5 bk repository with symlink and Makefile
fixes for builds
- fix up patches, series files etc. Use some new names.
zab [Fri, 23 May 2003 21:34:00 +0000 (21:34 +0000)]
- add mdc, obdfilter, and lov to the mix
- static out more duplicate lprocfs symbols, xprocfs too
- move lustre code to the 2.5 d_hash hlist_ api and have compat map it back to
2.4's list_heads
- ->bmap now enjoys sector_t
zab [Fri, 23 May 2003 00:22:05 +0000 (00:22 +0000)]
- this seems to pass runregression-net.sh..
- bring the osc and ost into the mix for the benefit of echo.sh
- get lprocfs building into the kernel by not defining the same 3 symbols in
every lprocfs object (sigh)
- compat out path_lookup, we may need to EXPORT it for 2.5 later
- s/uint flags/ulong flags/ in ptlrpc (eep)
zab [Thu, 22 May 2003 21:30:08 +0000 (21:30 +0000)]
- build ldlm and ptlrpc 'cause obdecho wants them
- don't mess about with 2.5's host app shared library mechanisms, just link
ptlctl in explicitly
- add a 2.5 ext3 san patch for ext3_get_block_handle and journal_handle_stop
- at least pretend to track the patches in series/jdike-2.5
- 'int flags', doh.
zab [Thu, 22 May 2003 18:26:20 +0000 (18:26 +0000)]
get utils/ building
- use 2.5's nice shared lib helpers to build libptlctl
zab [Thu, 22 May 2003 07:51:48 +0000 (07:51 +0000)]
- N instances of lib objs will collide in the final link, lets try building
it explicitly..
- get obdecho/lib compiling, make headway into obdclass
- I'm tired of adding Makefile.mk entries to configure.in
- lproc's awesome code duplication was saddening the final link, just call me
cap'n static
zab [Thu, 22 May 2003 05:02:12 +0000 (05:02 +0000)]
get mds/ building in 2.5
- fix the still-born KDEVT_VAL
- 2.5 uses a extra-awesome special list/head datatype for hashes. we introduce
some compatibility macros so we can drive both it and 2.4's list_head. this
isn't quite done..
- cook some 2.5 make fu to link into ../lib
zab [Wed, 21 May 2003 00:57:52 +0000 (00:57 +0000)]
- eeb points out that we don't actually need test_bit
- don't try and vmalloc _16 meg_ in the ping server, 16k is fine, I promise.
- nurr, really default to the socknal in the ping server.
zab [Tue, 20 May 2003 22:26:31 +0000 (22:26 +0000)]
- get pinger support utils building
- oops, s/obj/src/ .. the 2.5 build can have different source and target trees
- make all kp30's thread_size business conditional on __kernel__
- when compiled in SYMBOL_GET amounts to a weak reference which won't be
satisfied by a static symbol in the link, only kping_client wasn't right
zab [Tue, 20 May 2003 01:56:17 +0000 (01:56 +0000)]
- include the portals ping client/server in the 2.5 build
- default to the socknal in the ping server
zab [Tue, 20 May 2003 00:42:15 +0000 (00:42 +0000)]
uml+portals boots..
- change the makefile ordering so that socknal gets setup after the router
- while we're at it, init some router data at build time so its not as
susceptible to module_init() ordering
zab [Mon, 19 May 2003 23:22:18 +0000 (23:22 +0000)]
portals builds and the kernel tries to link..
- add 2.5 build spew to .cvsignore
- more relative makefile inclusion fixes
- more unsigned long flags
zab [Mon, 19 May 2003 23:05:22 +0000 (23:05 +0000)]
get libcfs/ and portals/ building
- add portals kernel_compat header
- get THREAD_SIZE from the kernel
- minor typos in makefiles, don't ask to build non-existant files
- continue to fix up the const propogation
zab [Mon, 19 May 2003 22:07:37 +0000 (22:07 +0000)]
I like 2.5's sanity checks already. This cleans up socknal's build..
- fix up portals Makefile.mk relative file references, 2.5 doesn't -C
- locks take 'unsigned long' flags arguments
- bitfields do too, and it could be a memory corrupting bug to use otherwise.
- call the kernel's sock_sendmsg wrapper instead of exposing ourselves to
the prot call's variations
zab [Mon, 19 May 2003 21:10:58 +0000 (21:10 +0000)]
- come up to modern 2.5's bdevname api
- sprinkle come consts around our consumers of LINE/FUNC to shut up gcc3
zab [Mon, 19 May 2003 18:57:54 +0000 (18:57 +0000)]
- merge recent b_devel build fixes into b_ad
cvs2svn [Mon, 19 May 2003 06:26:24 +0000 (06:26 +0000)]
This commit was manufactured by cvs2svn to create branch
'unlabeled-1.1.2.1.2'.
pschwan [Mon, 19 May 2003 06:26:23 +0000 (06:26 +0000)]
Fix some minor autoconf incompatibilities, replace portals cvsignore files
braam [Mon, 19 May 2003 06:17:37 +0000 (06:17 +0000)]
add missed file
braam [Mon, 19 May 2003 04:27:44 +0000 (04:27 +0000)]
- add portals to Lustre
- fix up makefiles
- add initial part of 2.5 in kernel make infrastructure
- cleanup tcpnal prototypes etc.
pschwan [Sun, 18 May 2003 17:50:08 +0000 (17:50 +0000)]
- fork iopen.patch into iopen-2.4.18 and iopen-2.4.20
- fix the EXPORT in extN-san
- lprocfs update for b_proto's new getattr_async obd function
pschwan [Sun, 18 May 2003 17:05:50 +0000 (17:05 +0000)]
merged b_devel into b_proto,
20030518
pschwan [Sun, 18 May 2003 16:39:25 +0000 (16:39 +0000)]
landing b_open onto b_devel:
- kernel patches for bug 1203
- fixes directory open()s
- re-enables metadata caching on the MDS
- fixes bug in ext3_noread kernel patch
- fixes some symbol export problems in other kernels
- fixes compilation problems when running on Red Hat 2.4.20 kernels
If you update past this merge, you will require kernel version 18
meijia [Sun, 18 May 2003 16:25:40 +0000 (16:25 +0000)]
[liblustre]:
- code reorganize
braam [Sun, 18 May 2003 04:34:44 +0000 (04:34 +0000)]
- fix a bug Ninja Schwan spotted: one too many igets...
braam [Sat, 17 May 2003 20:29:06 +0000 (20:29 +0000)]
- fix broken patches: kmem_cache_validate-2.4.20
- add iopen for chaos series
- fix broken exec.c fix in vanilla-2.4.20 and chaos
- update from b_devel
braam [Sat, 17 May 2003 17:55:55 +0000 (17:55 +0000)]
- a file handle cache in the MDS:
- Implemented within ext3 as a new file in the 2.4.20 series: iopen
- Originates from Ted T'so's iopen patch
- Many many changes to that patch
- simplifications in mds_fid2dentry
- The mds has a dentry mds->mds_fid_de for cached fid's
- The FH directory is gone.
- The CVS test in sanity is now actually working, except for commit
adilger [Fri, 16 May 2003 23:16:22 +0000 (23:16 +0000)]
Pass peer/target UUID to the log handling functions, so we can track which
logs/catalogs belong to each target. Targets should only be sent the logs
that pertain to themselves. We don't have these values in the kernel yet.
thantry [Fri, 16 May 2003 06:01:52 +0000 (06:01 +0000)]
User space utiltity to format /proc/fs/lustre/...*_stat files.
adilger [Thu, 15 May 2003 20:23:03 +0000 (20:23 +0000)]
Use S_UNDEFINED subsystem for now, as S_LOG doesn't seem to appear in logs...
Fix deadlock in log locking when doing cancels.
Allow osc_log_cancel() to be called with NULL params to force a cancel flush.
Force a cancel flush and free cancel params when disconnecting OSC from target.
Send an empty reply message from OST for cancel.
adilger [Wed, 14 May 2003 20:17:51 +0000 (20:17 +0000)]
Set up logging on the MDS.
Add fields to MDS last_rcvd to more closely match filter last_rcvd
- only last_objid is missing, as it would change the on-disk MDS layout
- that could be added if/when we need to do other MDS format changes
Add handling of unlink cookies on OSC, LOV and OST, but not full-circle yet.
Delete empty log files from catalog at close time.
Add locking for logs/catalog.
rread [Tue, 13 May 2003 00:10:14 +0000 (00:10 +0000)]
b_devel -> b_proto
from DEVEL_PROTO_UPDATE_PARENT_20030512
to DEVEL_PROTO_UPDATE_CHILD_20030512
wangdi [Sat, 10 May 2003 10:02:42 +0000 (10:02 +0000)]
fix the config file
adilger [Sat, 10 May 2003 05:25:52 +0000 (05:25 +0000)]
WIP checkin of b_orphan. Catalog is now an object, and the log headers
include a creation timestamp.
amrutjoshi [Fri, 9 May 2003 08:52:21 +0000 (08:52 +0000)]
macro to assign values to kdev_t
amrutjoshi [Thu, 8 May 2003 16:01:36 +0000 (16:01 +0000)]
pc files for patches
amrutjoshi [Thu, 8 May 2003 15:56:41 +0000 (15:56 +0000)]
patch with san functions for ext3.
amrutjoshi [Thu, 8 May 2003 15:55:04 +0000 (15:55 +0000)]
export truncate patch for 2.5.63
amrutjoshi [Thu, 8 May 2003 13:29:18 +0000 (13:29 +0000)]
missing 2.5 types. I am not sure this is the right place for this.
amrutjoshi [Thu, 8 May 2003 13:11:57 +0000 (13:11 +0000)]
compatibity macros
wangdi [Wed, 7 May 2003 12:05:33 +0000 (12:05 +0000)]
add asm/fixmap.h in dsp.patch
wangdi [Wed, 7 May 2003 11:38:41 +0000 (11:38 +0000)]
*** empty log message ***
wangdi [Wed, 7 May 2003 11:18:58 +0000 (11:18 +0000)]
add config-linux-2.4.20-i386-rh
wangdi [Wed, 7 May 2003 11:13:29 +0000 (11:13 +0000)]
add config-linux-2.4.18-p4smp-52chaos
wangdi [Wed, 7 May 2003 11:09:16 +0000 (11:09 +0000)]
add config-linux-2.4.18-uml
wangdi [Wed, 7 May 2003 10:01:30 +0000 (10:01 +0000)]
add dsp.pc
wangdi [Wed, 7 May 2003 10:00:31 +0000 (10:00 +0000)]
add mcore-2.4.20-8.pc
wangdi [Wed, 7 May 2003 09:58:45 +0000 (09:58 +0000)]
add dsp.patch
wangdi [Wed, 7 May 2003 09:57:54 +0000 (09:57 +0000)]
add mcore-2.4.20-8.patch
adilger [Tue, 6 May 2003 23:54:50 +0000 (23:54 +0000)]
Fix numerous tiny bugs in the logging code. The code now will generate logs
and cookies, send cookie back from filter_create() to client, and client
passes them on to MDS along with LOV EA (didn't observe this directly).
MDS now needs to take this cookie and hook it into commit callback and then
call obd_log_cancel on (not yet set up) connection to LOV/OSC.
wangdi [Mon, 5 May 2003 12:44:58 +0000 (12:44 +0000)]
add ext3_orphan_lock-2.4.20-rh.pc
wangdi [Mon, 5 May 2003 12:43:40 +0000 (12:43 +0000)]
add ext3_orphan_lock-2.4.20-rh.patch
wangdi [Mon, 5 May 2003 10:20:05 +0000 (10:20 +0000)]
add uml-patch-2.4.20-4.pc
wangdi [Mon, 5 May 2003 10:17:42 +0000 (10:17 +0000)]
add uml-patch-2.4.20-4.patch
pschwan [Mon, 5 May 2003 04:30:09 +0000 (04:30 +0000)]
partial merge of b_open into b_devel:
- Lustre kernel patch version 16
- extN no longer built; ext3 patched as part of the kernel patch
- related changes to lconf/lmc, to load and mkfs ext3 instead of extN now
adilger [Sun, 4 May 2003 19:31:03 +0000 (19:31 +0000)]
.pc file for delete thread patch.
adilger [Sun, 4 May 2003 19:28:40 +0000 (19:28 +0000)]
New version of delete_thread patch for 2.4.20 kernel.
wangdi [Sun, 4 May 2003 10:28:56 +0000 (10:28 +0000)]
add ext3-noread-2.4.20.pc
wangdi [Sun, 4 May 2003 10:25:09 +0000 (10:25 +0000)]
add ext3-noread-2.4.20 patch
braam [Sun, 4 May 2003 05:04:44 +0000 (05:04 +0000)]
- fixes to b_open
- new UML patch 2.4.20-4 which works with Red Hat 9
- minor fixes to handling handles in open/close to avoid oopsing
- fix erroneous mds_mfd_put in mds_open
- fix the CVS tests added to sanity
- the tests exposed more problem in this branch, probably related to
the lack of a file handle cache. Only good parts of the branch will
be merged.
adilger [Sat, 3 May 2003 18:06:30 +0000 (18:06 +0000)]
Make obdfilter logging actually work:
- don't dput parent directories (doh!)
- revert a bunch of changes no longer needed for logging
meijia [Sat, 3 May 2003 03:18:10 +0000 (03:18 +0000)]
merge b_devel -> b_eq
wangdi [Fri, 2 May 2003 11:20:31 +0000 (11:20 +0000)]
fix a flaw in linux-2.4.20-xattr-0.8.54-chaos.patch
wangdi [Fri, 2 May 2003 08:05:34 +0000 (08:05 +0000)]
add kmem_cache_validate_2.4.20-rh.pc
wangdi [Fri, 2 May 2003 08:03:38 +0000 (08:03 +0000)]
add exports_2.4.20-rh.pc
wangdi [Fri, 2 May 2003 08:02:30 +0000 (08:02 +0000)]
add exports_2.4.20-rh.patch
wangdi [Fri, 2 May 2003 08:00:41 +0000 (08:00 +0000)]
add kmem_cache_validate_2.4.20-rh.patch
adilger [Fri, 2 May 2003 00:00:54 +0000 (00:00 +0000)]
Another not-yet-working-but-checking-in-not-to-lose-data landing of b_orphan.
Will do another merge with b_devel, but really dreading when b_proto lands...
mmex [Wed, 30 Apr 2003 16:18:12 +0000 (16:18 +0000)]
Changes to hp-pnnl patch files to make them work on hp-pnnl kernels.
- drop hp-pnnl specific intent patch, the vanilla works
- redo xattr patch, exported symbols must be in uniquely named files if
modversions is turned on. These changes should likely also be ported
to the other series, since they can have the same problem if
modversions gets turned on. Be careful when adding new exports to
make sure the file including the export is uniquely named across the
entire kernel.
wangdi [Wed, 30 Apr 2003 10:57:33 +0000 (10:57 +0000)]
add the uml.spec
rread [Tue, 29 Apr 2003 21:46:35 +0000 (21:46 +0000)]
- use --failover on the kimberlite example script; this is required
for failback, but is broken for normal shutdown.
meijia [Mon, 28 Apr 2003 14:27:57 +0000 (14:27 +0000)]
[liblustre]:
- temorarily add script to generate shared lib of liblustre. will merge
into makefile later
meijia [Mon, 28 Apr 2003 14:24:47 +0000 (14:24 +0000)]
[liblustre]:
- remove the unneeded 'printf'
meijia [Mon, 28 Apr 2003 14:22:08 +0000 (14:22 +0000)]
[liblustre]:
- add liblustre startup wrapper script
meijia [Mon, 28 Apr 2003 14:20:26 +0000 (14:20 +0000)]
[liblustre]:
- add a init/fini function wraper for libluster
- debug message fix
pschwan [Sun, 27 Apr 2003 17:40:04 +0000 (17:40 +0000)]
b=1178
- Preparations for removing unused "addr" from lustre_handle
- Some mild internal API consistency fixups, before I copied this API
to three other places that need handles
wangdi [Sat, 26 Apr 2003 03:25:39 +0000 (03:25 +0000)]
check whether the system has installed lsdiff before check_pc_match
mmex [Fri, 25 Apr 2003 21:48:39 +0000 (21:48 +0000)]
Updates to hp-pnnl kernel to bring it up to date with hp6-pnnl2 kernel.
adilger [Fri, 25 Apr 2003 08:36:59 +0000 (08:36 +0000)]
More for logging:
- Store recovery status in export, and check that, so we can have a
non-recovery connection to the obdfilter and not do logging, transnos, etc.
- Split up filter_setup() so that it can be set up by the MDS on an already
mounted filesystem
- Rename obdfilter "last_rcvd" file to be "filter_data" to avoid clashing on
an MDS+obdfilter setup
- Pass struct obd_trans_info around, and put the journal handle in that,
instead of using desc_private (for journal nesting).
Replace "force" and "failover" with "flags" + OBD_OPT_FORCE, OBD_OPT_FAILOVER.
mmex [Thu, 24 Apr 2003 20:54:44 +0000 (20:54 +0000)]
Fix syntax error in script.
adilger [Thu, 24 Apr 2003 08:06:39 +0000 (08:06 +0000)]
More work on the logging code.
Move "desc_private" journal handle into obd_trans_info (already passed to
many of the OBD API methods) so that we can do nested transactions when
calling OBD API methods.
pschwan [Thu, 24 Apr 2003 05:19:26 +0000 (05:19 +0000)]
- remove double-CDEBUG in ptlrpc_get_connection path
- add some bug numbers to sanity-ldlm.sh
wangdi [Thu, 24 Apr 2003 05:03:09 +0000 (05:03 +0000)]
add error handler in check_pc_match
pschwan [Thu, 24 Apr 2003 05:02:46 +0000 (05:02 +0000)]
- convert all kmem_cache_alloc/PORTAL_SLAB_ALLOC to OBD_SLAB_ALLOC
- POISON is sometimes a no-op, so use memset() for unused handle.addr
- in target_handle_reconnect, only compare "cookie", not the whole handle
- a cleanup here, a cleanup there, while I hunted for the corruption
pschwan [Thu, 24 Apr 2003 04:56:25 +0000 (04:56 +0000)]
remove the mds_file_cache slab, in favour of new-style handles
wangdi [Wed, 23 Apr 2003 11:09:39 +0000 (11:09 +0000)]
add fs/exec patch