Whamcloud - gitweb
fs/lustre-release.git
22 years agofix portals/archdep.m4
wangdi [Mon, 26 May 2003 11:04:05 +0000 (11:04 +0000)]
fix portals/archdep.m4

22 years agob=1173
pschwan [Mon, 26 May 2003 08:18:18 +0000 (08:18 +0000)]
b=1173
Remove all files from lib/, put the functions into various other
modules as appropriate, and remove the bits to create the symlinks.
Tested in UML and lightly on dev.  Builds, inserts, mounts.

22 years agoMostly working full cookie create-send-cancel cycle for MDS unlinks.
adilger [Sun, 25 May 2003 18:17:54 +0000 (18:17 +0000)]
Mostly working full cookie create-send-cancel cycle for MDS unlinks.
MDS now establishes a connection (LOV only, getting UUID from LOVDESC pending
  bug 1280 resolution).
We cannot clean up as a result of this (MDS->LOV connection).  Might be
  possible to fix by changing cleanup order so we do llite/mdc/mds cleanup
  before lov/osc/ost cleanup.

22 years agofix for build portals in user space
meijia [Sat, 24 May 2003 08:49:12 +0000 (08:49 +0000)]
fix for build portals in user space

22 years ago* Added portals + lustre wire constant and struct size/offset assertions
eeb [Fri, 23 May 2003 22:04:03 +0000 (22:04 +0000)]
*   Added portals + lustre wire constant and struct size/offset assertions

*   Added portals + lustre wirecheck programs to generate these assertions
    on the reference platform.

22 years agomerge b_devel to b_eq since there are problems with old b_devel
meijia [Fri, 23 May 2003 10:01:12 +0000 (10:01 +0000)]
merge b_devel to b_eq since there are problems with old b_devel
  DEVEL_EQ_UPDATE_PARENT_20030523 -> DEVEL_EQ_UPDATE_CHILD_20030523

22 years agoMerge b_devel -> b_orphan.
adilger [Fri, 23 May 2003 07:03:13 +0000 (07:03 +0000)]
Merge b_devel -> b_orphan.
  DEVEL_ORPHAN_UPDATE_PARENT_20030522 -> DEVEL_ORPHAN_UPDATE_PARENT_20030522

22 years agomerge b_devel to b_eq:
meijia [Wed, 21 May 2003 15:43:09 +0000 (15:43 +0000)]
merge b_devel to b_eq:
  DEVEL_EQ_UPDATE_PARENT_20030521 -> DEVEL_EQ_UPDATE_CHILD_20030521

22 years agoRemove bogus local function prototypes that were causing me grief.
adilger [Wed, 21 May 2003 06:25:00 +0000 (06:25 +0000)]
Remove bogus local function prototypes that were causing me grief.
Fix header mess as necessary to be able to compile.
Most of MDS unlink cookie generation complete.
Remove some local variable declarations that are only used in debug messages
  that appeared when compiling with CDEBUG() as a no-op (reduce stack usage).

22 years ago- Update kernel patchs to work on hp4_pnnl6. Note that a new branch of
mmex [Mon, 19 May 2003 17:07:44 +0000 (17:07 +0000)]
- Update kernel patchs to work on hp4_pnnl6.  Note that a new branch of
dev_read_only was made.  dev_read_only_hp can not longer be used
because multiple other series (not the hp ones) now use it and regularly
update it in conflicting ways.

22 years agomerge b_devel into b_eq:
meijia [Mon, 19 May 2003 17:04:54 +0000 (17:04 +0000)]
merge b_devel into b_eq:
  DEVEL_EQ_UPDATE_PARENT_20030519 -> DEVEL_EQ_UPDATE_CHILD_20030519

only resolve confliction.

22 years agomerge b_ad (cleaned up a bit) into b_devel; entirely 2.5 compat fixes
pschwan [Mon, 19 May 2003 06:29:23 +0000 (06:29 +0000)]
merge b_ad (cleaned up a bit) into b_devel; entirely 2.5 compat fixes

22 years agothe chaos kernel can use the iopen-2.4.18.patch
pschwan [Mon, 19 May 2003 05:19:54 +0000 (05:19 +0000)]
the chaos kernel can use the iopen-2.4.18.patch

22 years agoland b_proto on b_devel:
pschwan [Mon, 19 May 2003 03:15:41 +0000 (03:15 +0000)]
land b_proto on b_devel:
 - major rework of the rpc and bulk APIs, lots of new assertions and cleanups
 - many rpc and bulk bugfixes, including bugs 937, 1057, 1072, 1162
 - perform brw and getattr operations to multiple OSTs in parallel
 - all bulks are tied to exactly one request; multiple requests can be
   tied to a single request_set

22 years agolanding b_open onto b_devel:
pschwan [Sun, 18 May 2003 16:39:34 +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

22 years ago- remainder of the b_devel into b_open merge.
braam [Sat, 17 May 2003 21:18:06 +0000 (21:18 +0000)]
- remainder of the b_devel into b_open merge.

22 years agoAdded lprocfs counters for obd_ops stats, obdfilter and ptlrpc services. Bugzilla...
thantry [Fri, 16 May 2003 05:55:52 +0000 (05:55 +0000)]
Added lprocfs counters for obd_ops stats, obdfilter and ptlrpc services. Bugzilla#1107.

22 years agoMerge b_devel (ORPHAN_BASE6 -> ORPHAN_BASE7) to
adilger [Thu, 15 May 2003 17:01:48 +0000 (17:01 +0000)]
Merge b_devel (ORPHAN_BASE6 -> ORPHAN_BASE7) to
b_orphan (DEVEL_ORPHAN_UPDATE_CHILD_20030514).

22 years agoUpdate the .pc to match the renamed patch
pschwan [Wed, 14 May 2003 17:48:31 +0000 (17:48 +0000)]
Update the .pc to match the renamed patch

22 years agob=1046
rread [Wed, 14 May 2003 17:44:08 +0000 (17:44 +0000)]
b=1046

update the config version.

22 years agob=1046
rread [Wed, 14 May 2003 17:42:55 +0000 (17:42 +0000)]
b=1046

- add a new argument to upcalls. The first argument identifies the reason
  of upcall.  The upcall script path is now /proc/sys/lustre/upcall, and the
  config option has changed from --recovery_upcall to --lustre_upcall.

  Current lustre upcall types and arguments:

    FAILED_IMPORT tgt_uuid dev_uuid net_uuid
    RECOVERY_OVER dev_uuid

- The config tools also support the new --portals_upcall, and both of the
  upcalls can be set to the same script with the --upcall option.

  There is only one portals upcall type:

    LBUG filename function line

22 years ago- applying uml-2.4.20-4 to vanilla 2.4.20 makes kernel non-compilable for i386
alex_tomas [Tue, 13 May 2003 19:07:38 +0000 (19:07 +0000)]
- applying uml-2.4.20-4 to vanilla 2.4.20 makes kernel non-compilable for i386

22 years ago- core UML patch
alex_tomas [Tue, 13 May 2003 19:04:36 +0000 (19:04 +0000)]
- core UML patch

22 years ago- add ptlrpc_dump_connections() for debugging ldlm cleanup problems
rread [Tue, 13 May 2003 00:46:08 +0000 (00:46 +0000)]
- add ptlrpc_dump_connections() for debugging ldlm cleanup problems
- cleanup the import's export ptlrpc_import_connect if the connect fails.
  this prevents an obd_refcount clean on the client during a failed mount.

22 years agob=1215
rread [Wed, 7 May 2003 23:39:57 +0000 (23:39 +0000)]
b=1215
r=shaver@clusterfs.com

- add obd_refcount, and use it to ensure the device not being used
  during cleanup. If --force is used, then cleanup will set a stopping
  flag on the device, disconnect the exports, and then wait until
  the device is unused.

- removed class_disconnect_all and moved/renamed forcibly_detatch_exports
  to class_disconnect_exports.

- ptlrpc_import_disconnect now disconnects imp_export

- if --failover, then filter cleanup will not delete files marked
  for delete on close.

- manage imp_active flag in ptlrpc_set_import_active

22 years agob=1026
rread [Wed, 7 May 2003 19:12:10 +0000 (19:12 +0000)]
b=1026
r=adilger@clusterfs.com

- add initial support for creating client connections on the MDS,
  configs will need to be recreated. (The version has been updated
  to force this.)

- create a new config object, "filesystem" which represents a
  specific filesystem.  In the zero-config future, the MDS will be
  need to be aware of what filesystems it "owns," so this is perhaps one
  tiny step in that direction.

22 years agoadd config-linux-2.4.20-uml
wangdi [Wed, 7 May 2003 11:14:29 +0000 (11:14 +0000)]
add config-linux-2.4.20-uml

22 years agoadd uml.spec
wangdi [Wed, 7 May 2003 11:06:35 +0000 (11:06 +0000)]
add uml.spec

22 years agopartial merge of b_open into b_devel:
pschwan [Mon, 5 May 2003 04:29:49 +0000 (04:29 +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

22 years agofixed -ESTALE bug in vfs_intent-2.4.18-18.patch (already fixed in other
pschwan [Mon, 5 May 2003 03:32:49 +0000 (03:32 +0000)]
fixed -ESTALE bug in vfs_intent-2.4.18-18.patch (already fixed in other
vfs_intent patches)

22 years agoremove exports_2.4.20-rh.pc, which can be replaced by exports_hp_2.4.20.pc
wangdi [Sun, 4 May 2003 11:04:14 +0000 (11:04 +0000)]
remove exports_2.4.20-rh.pc, which can be replaced by exports_hp_2.4.20.pc

22 years agoremove exports_2.4.20-rh.patch, which can be replaced by exports_hp_2.4.20.patch
wangdi [Sun, 4 May 2003 11:01:41 +0000 (11:01 +0000)]
remove exports_2.4.20-rh.patch, which can be replaced by exports_hp_2.4.20.patch

22 years agoadd ext3-noread-2.4.20.patch
wangdi [Sun, 4 May 2003 10:22:16 +0000 (10:22 +0000)]
add ext3-noread-2.4.20.patch

22 years ago- fixes to b_open
braam [Sun, 4 May 2003 05:04:40 +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.

22 years agoMerge from b_devel to b_orphan (ORPHAN_BASE5 -> ORPHAN_BASE6).
adilger [Sun, 4 May 2003 00:33:30 +0000 (00:33 +0000)]
Merge from b_devel to b_orphan (ORPHAN_BASE5 -> ORPHAN_BASE6).

22 years ago- this is a reconstruction of b_kernel and a merge from current
braam [Sat, 3 May 2003 07:48:44 +0000 (07:48 +0000)]
- this is a reconstruction of b_kernel and a merge from current
  b_devel (as of OPEN_BASE).  Due to tag confusion we had to do this
  by hand.

- DO NOT USE B_KERNEL, MAIL DIFF's TO PETER & PHIL

22 years agomerge b_devel into b_proto
pschwan [Sat, 3 May 2003 06:00:59 +0000 (06:00 +0000)]
merge b_devel into b_proto
from DEVEL_PROTO_UPDATE_PARENT_20030502 tag on b_devel
  to DEVEL_PROTO_UPDATE_CHILD_20030502 tag on b_proto

22 years agoAdding it to b_devel too.
adilger [Fri, 2 May 2003 23:24:54 +0000 (23:24 +0000)]
Adding it to b_devel too.

22 years agob=1178
pschwan [Sun, 27 Apr 2003 17:49:18 +0000 (17:49 +0000)]
b=1178
Preparations for removing unused "addr" from lustre_handle

22 years ago- merge b_devel into b_proto
pschwan [Fri, 25 Apr 2003 06:06:04 +0000 (06:06 +0000)]
- merge b_devel into b_proto
- simplify the check_set/set_wait logic a bit

22 years agoMerge from b_devel -> b_orphan (up to ORPHAN_BASE5 tag on b_devel)
adilger [Thu, 24 Apr 2003 08:28:30 +0000 (08:28 +0000)]
Merge from b_devel -> b_orphan (up to ORPHAN_BASE5 tag on b_devel)

22 years agob=1075
rread [Thu, 24 Apr 2003 01:36:34 +0000 (01:36 +0000)]
b=1075
* added --failover option to lconf and lctl cleanup
* failover flag added to obd_disconnect and obd_cleanup
* if failover ==1, then MDS and OST will not update the client
  state in last_rcvd.
* Also changes to lconf to add the --group support to MDS.

22 years agob=1075
rread [Mon, 21 Apr 2003 20:00:03 +0000 (20:00 +0000)]
b=1075
  If a reply comes back with -ENOTCONN, then this means that we've
  either been evicted, or the device has been removed from the
  server. First attempt to reconnect, then call the upcall and attempt
  recovery.

22 years agoFix build: ldlm_hooks_referenced wasn't defined or set.
shaver [Sun, 20 Apr 2003 16:55:04 +0000 (16:55 +0000)]
Fix build: ldlm_hooks_referenced wasn't defined or set.

22 years agob=1111
pschwan [Sun, 20 Apr 2003 15:43:29 +0000 (15:43 +0000)]
b=1111
- add ptlrpc_ldlm_hooks_referenced(), so we can determine in
  ldlm_cleanup if "force" is needed
- add a little bit more debug information to hunt connection leaks

22 years agoMerge b_devel changes into b_proto
pschwan [Sat, 19 Apr 2003 15:53:46 +0000 (15:53 +0000)]
Merge b_devel changes into b_proto

22 years agob=1075
rread [Sat, 19 Apr 2003 01:07:24 +0000 (01:07 +0000)]
b=1075
  The config part of failback.  Lconf now supports --group for setup and
  cleanup.  If --group is used,  lconf will only setup/cleanup the
  devices that are in the group AND are currently active on this node. Lconf
  will still setup the modules needed by the devices (networking, routing),
  but cleanup will leave these untouched if there are still other devices
  left after cleaning up the group.

* small fix enable config version support in ldap

22 years ago- add verison to config, so lconf can make sure the xml/ldap was created
rread [Tue, 15 Apr 2003 21:46:42 +0000 (21:46 +0000)]
- add verison to config, so lconf can make sure the xml/ldap was created
  by a compatible lmc.

- recreate clients if remounting

22 years ago[liblustre]:
meijia [Sun, 13 Apr 2003 14:52:58 +0000 (14:52 +0000)]
[liblustre]:
 - initial support symlink

22 years ago[liblustre]:
meijia [Sun, 13 Apr 2003 13:56:51 +0000 (13:56 +0000)]
[liblustre]:
 - add support to mkdir

22 years ago[liblustre]:
meijia [Sun, 13 Apr 2003 01:45:05 +0000 (01:45 +0000)]
[liblustre]:
 - support setattr. but currently we don't involved any locking, thus
   contains ugly workaround, maybe totally wrong.

22 years ago- merge b_devel into b_proto (incl b_swab changes)
pschwan [Sat, 12 Apr 2003 20:07:45 +0000 (20:07 +0000)]
- merge b_devel into b_proto (incl b_swab changes)
- fix leak-all-bulk-pages bug
- Mike fixed a bug where a failed ptlrpc_queue_wait send buf would
  fail, but the user would still have to wait for the timeout to pass
  before being able to interrupt

22 years agob=1111
pschwan [Sat, 12 Apr 2003 17:42:08 +0000 (17:42 +0000)]
b=1111
- Pass 'force' down to all of the cleanup methods, instead of just
  having special meaning for exports
- in ldlm_cleanup, if forcing, remove the ldlm hooks from ptlrpc so we
  can unload the ldlm module

22 years agob=1084
rread [Fri, 11 Apr 2003 19:27:59 +0000 (19:27 +0000)]
b=1084
* add a --group=tag option to mds and ost devices, so related devices
  can be failed over together.
* rewrite lactve to failover a specific device group, instead of all
  devices on a node.

22 years ago[liblustre]:
meijia [Fri, 11 Apr 2003 07:48:34 +0000 (07:48 +0000)]
[liblustre]:
 - do proper cleanup with an inode is about to be destroyed

22 years ago[liblustre]:
meijia [Fri, 11 Apr 2003 07:12:39 +0000 (07:12 +0000)]
[liblustre]:
 - add inode operation close()

22 years ago[liblustre]:
meijia [Fri, 11 Apr 2003 06:12:54 +0000 (06:12 +0000)]
[liblustre]:
 - actually retrive stripe ea data from server
 - fix logic in name resolution, save fid for each file

22 years ago[liblustre]:
nfshp [Thu, 10 Apr 2003 02:07:12 +0000 (02:07 +0000)]
[liblustre]:
 - add a fake ioctl which could dump mount options to file
 - sysio_mount use UUIDs from the mount options
 - minor fix for pnode count

22 years ago[liblustre]: now we don't need the syscall workaround in libsysio anymore
nfshp [Wed, 9 Apr 2003 09:43:19 +0000 (09:43 +0000)]
[liblustre]: now we don't need the syscall workaround in libsysio anymore

22 years ago[liblustre]: following the recent libsysio changes.
nfshp [Wed, 9 Apr 2003 08:43:21 +0000 (08:43 +0000)]
[liblustre]: following the recent libsysio changes.

22 years agofix cut and paste braino
rread [Wed, 9 Apr 2003 00:20:49 +0000 (00:20 +0000)]
fix cut and paste braino

22 years ago[liblustre]: must feed NULL-terminated string to mdc_getattr_name()
nfshp [Tue, 8 Apr 2003 11:33:54 +0000 (11:33 +0000)]
[liblustre]: must feed NULL-terminated string to mdc_getattr_name()

22 years agoland b_cray on b_devel
nfshp [Tue, 8 Apr 2003 10:22:09 +0000 (10:22 +0000)]
land b_cray on b_devel

22 years ago* Landing b_swab
eeb [Tue, 8 Apr 2003 00:32:50 +0000 (00:32 +0000)]
*   Landing b_swab

22 years agomerge b_devel into b_cray
nfshp [Mon, 7 Apr 2003 07:21:35 +0000 (07:21 +0000)]
merge b_devel into b_cray

22 years agomerge b_devel into b_cray
nfshp [Fri, 4 Apr 2003 16:16:39 +0000 (16:16 +0000)]
merge b_devel into b_cray

22 years ago- set_osc_active becomes ptlrpc_set_import_active (though it still only works on
shaver [Thu, 3 Apr 2003 21:20:53 +0000 (21:20 +0000)]
- set_osc_active becomes ptlrpc_set_import_active (though it still only works on
OSCs) and is automatically called for non-recoverable OSCs before we invoke the
upcall. (part of b=1038)
- When aborting recovery, shut down the recovery timer, abort the recovery
queue's requests, and don't do it twice.
- Make sure we call obd_disconnect correctly for all non-lctl imports when
aborting recovery or forcing shutdown, so that we don't leave the last_rcvd bits
around for the next time.
- Fix leak of last_rcvd-spawn exports (every one!).
- Get and put exports for recovery-queued requests and replies, so that
  we don't race badly with teardown.

22 years agob=1089
rread [Thu, 3 Apr 2003 18:02:21 +0000 (18:02 +0000)]
b=1089
 Python does not automatically convert strings to integers when it might
 seem appropriate.

22 years agob=957
pschwan [Wed, 2 Apr 2003 03:56:25 +0000 (03:56 +0000)]
b=957
- Infrastructure for the pinger thread; needs multi-rpc and some refinements
- Add last_request_time to obd_export

- Fixed export leak in lov_brw success case
- test22 rears its head again: tar gets pissy if it can't access $PWD,
  which is common for homedirs at LLNL.  cd to /tmp before running tar.

Merged many of the changes from b_proto:
- multi-rpc infrastructure, but not the key bits
- removed some bitfield flags in favour of individual struct members
- merged obd_brw_write and obd_brw_read into a single obd_brw_internal
- move the Elan-related page wrangling from the IOD into the OSC

22 years agoadding cmdline.py to pymod_SCRIPTS
amrutjoshi [Tue, 1 Apr 2003 08:08:17 +0000 (08:08 +0000)]
adding cmdline.py to pymod_SCRIPTS

22 years agoadding scripts to EXTRA_DIST.
amrutjoshi [Tue, 1 Apr 2003 07:38:03 +0000 (07:38 +0000)]
adding scripts to EXTRA_DIST.

22 years agolanding b_recovery on b_devel
pschwan [Mon, 31 Mar 2003 18:33:36 +0000 (18:33 +0000)]
landing b_recovery on b_devel

* bug fixes
 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
 - per-import/export recovery handling (958, 931, 959)
 - multiple last-rcvd slots, for serving multiple FSes (949)
 - connections are again shared between multiple imp/exports (963, 964)
 - "umount -f" would hang if any requests needed to be sent (393, 978)
 - avoid pinning large req buffer by copying for queued messages (989)
 - add "uuid" to "lctl device" command to help upcalls (991)
 - "open" RPCs with transnos would confuse recovery counters (1037)
* major user-visible changes
 - fail out/fail over policy now controlled by the upcall (993)
* protocol changes
 - add OBD_PING to check server availability and failure (954)

22 years agolast merge of b_devel into b_recovery
pschwan [Mon, 31 Mar 2003 18:05:30 +0000 (18:05 +0000)]
last merge of b_devel into b_recovery

22 years agoMerge from b_devel.
shaver [Wed, 26 Mar 2003 07:09:48 +0000 (07:09 +0000)]
Merge from b_devel.

22 years agoMerge b_devel into b_recovery.
shaver [Wed, 19 Mar 2003 18:32:17 +0000 (18:32 +0000)]
Merge b_devel into b_recovery.

22 years agoBug828 fix. Approved by Andreas.
pengzhao [Tue, 11 Mar 2003 18:10:12 +0000 (18:10 +0000)]
Bug828 fix. Approved by Andreas.

22 years ago- purge the write cache before truncate stomps on the local i_size so that we
zab [Tue, 11 Mar 2003 02:34:52 +0000 (02:34 +0000)]
- purge the write cache before truncate stomps on the local i_size so that we
  don't zero-extend the file when truncate's lock acquisition purges the
  writeback after having updated the i_size.  I'm not sure how 2.5 will do this,
  but it probably needs to.

22 years agoMerge from b_devel, mainly to get the fixes for 977/978/722.
shaver [Sun, 9 Mar 2003 21:29:15 +0000 (21:29 +0000)]
Merge from b_devel, mainly to get the fixes for 977/978/722.

22 years agomerge b_devel into b_recovery
pschwan [Wed, 5 Mar 2003 18:40:29 +0000 (18:40 +0000)]
merge b_devel into b_recovery

22 years agoWhile it is true that ALL_PAGES is a good indicator of rmap, we don't get
adilger [Tue, 4 Mar 2003 09:51:52 +0000 (09:51 +0000)]
While it is true that ALL_PAGES is a good indicator of rmap, we don't get
this define without already including an rmap-specific header.  Instead
use PG_inactive_dirty.

22 years agoIt seems PG_lru also exists on the PNNL kernel, so use ALL_ZONES instead?
adilger [Tue, 4 Mar 2003 09:45:18 +0000 (09:45 +0000)]
It seems PG_lru also exists on the PNNL kernel, so use ALL_ZONES instead?

22 years agoThis makes very large IORs pass on a 400 node cluster..
zab [Tue, 4 Mar 2003 08:22:44 +0000 (08:22 +0000)]
This makes very large IORs pass on a 400 node cluster..

- ifdef PG_lru to check for rmap instead of pte_chain, which didn't work (!)
- put the vm specific memory pressure decision in should_writeback()
- stop prepare_write throttling/writeback when vm pressure lets up
- if we're still under pressure after writing back lots of dirty data then
  we also wait on any inodes to complete their writeback until the VM thinks
  we're happy or there's nothing to wait on..

22 years agoZach and I found a nasty race condition with the sets: we were
pschwan [Mon, 3 Mar 2003 21:05:47 +0000 (21:05 +0000)]
Zach and I found a nasty race condition with the sets: we were
overloading the refcount to mean both "number of outstanding I/Os" and
also "number of threads using this structure".  It was possible in
extreme circumstances to free the set in the reading thread before the
event handling thread called wake_up, which would thus crash.

To address this, there is a new count (brw_desc_count), which counts
the number of remaining I/Os.  The other count, brw_refcount, now
properly tracks users of the structure to prevent early freeing.

22 years agouse obd_brw_sets like the OSC, instead of putting it on the stack and
pschwan [Sun, 2 Mar 2003 09:07:22 +0000 (09:07 +0000)]
use obd_brw_sets like the OSC, instead of putting it on the stack and
playing silly buggers with the refcount

22 years ago03:26 < zab> oh, jesus.
pschwan [Sun, 2 Mar 2003 08:30:12 +0000 (08:30 +0000)]
03:26 < zab> oh, jesus.
03:27 < zab> the do { } while (llwp->num_pages) is totally hosed
03:27 < zab> its spinning on the same llwp for ever.
03:28 < zab> memset(llwp just before the ll_get_dirty_pages

also, I removed some tabs

22 years agoI took zab's patch, broke it, and the committed it to cvs.
pschwan [Sun, 2 Mar 2003 08:14:23 +0000 (08:14 +0000)]
I took zab's patch, broke it, and the committed it to cvs.

Then he read my checkin message and corrected me.

22 years agozab's patch to correctly track the number of fragments and full pages,
pschwan [Sun, 2 Mar 2003 08:10:17 +0000 (08:10 +0000)]
zab's patch to correctly track the number of fragments and full pages,
so that we put as many pages (up to 16) as possible on the wire at once

22 years agoland b_io on b_devel (b_io will be deleted as soon as it's confirmed
pschwan [Sun, 2 Mar 2003 05:58:07 +0000 (05:58 +0000)]
land b_io on b_devel (b_io will be deleted as soon as it's confirmed
that the merge happend without incident)

22 years agomerge 0.5.20.3 b_devel into b_io (ready to land)
pschwan [Sun, 2 Mar 2003 05:19:27 +0000 (05:19 +0000)]
merge 0.5.20.3 b_devel into b_io (ready to land)

22 years agocatch up with b_devel
shaver [Fri, 28 Feb 2003 21:42:20 +0000 (21:42 +0000)]
catch up with b_devel

22 years ago- bring b_io up to b_devel
zab [Thu, 27 Feb 2003 21:28:34 +0000 (21:28 +0000)]
- bring b_io up to b_devel

22 years agoFix uninitialized variable warning.
adilger [Tue, 25 Feb 2003 23:26:52 +0000 (23:26 +0000)]
Fix uninitialized variable warning.

22 years agoMake uuid_nid_data.nid a ptl_nid_t (u64) instead of a u32.
adilger [Tue, 25 Feb 2003 11:17:56 +0000 (11:17 +0000)]
Make uuid_nid_data.nid a ptl_nid_t (u64) instead of a u32.

22 years ago- bring b_devel changes into b_io in preparation for file size fixes
zab [Tue, 25 Feb 2003 01:09:36 +0000 (01:09 +0000)]
- bring b_devel changes into b_io in preparation for file size fixes

22 years agoLanding b_malt onto b_devel
rread [Thu, 20 Feb 2003 20:42:27 +0000 (20:42 +0000)]
Landing b_malt onto b_devel
- move uuid and handles to lustre, with config changes
- b=204,667: fix router config so network interfaces can be
  created in any order.
- Make sure all the ioctl calls  in obd.c are packed.

22 years agoUpdate b_multinet
rread [Thu, 20 Feb 2003 05:31:27 +0000 (05:31 +0000)]
Update b_multinet
  minor conflicts found in mds/handler.c and niobuf.c

22 years agomerge b_setattr into b_devel. The highlights:
pschwan [Tue, 18 Feb 2003 19:39:46 +0000 (19:39 +0000)]
merge b_setattr into b_devel.  The highlights:

- proper LDLM_DEBUG macro fix, for gcc 3.2 (850)
- failed open()s could cause deadlock; fixed (867, 869)
- stop cancelling OST locks when files are closed (481)
- overlapping XID spaces caused network corruption (851, 853)
- fix unsafe fsfilt counter arithmetic; change to atomic_t
- setattr_raw added, to do single-RPC, server-side setattrs
- fix crashy race condition between ptlrpc_free_req and osc_close

22 years agomerge b_devel into b_setattr. highlights:
pschwan [Mon, 17 Feb 2003 18:42:56 +0000 (18:42 +0000)]
merge b_devel into b_setattr.  highlights:
- kernel 2.5 and liblustre build changes
- fixes to Makefile.am for 7.3-era versions of automake, which is quite possibly
  the worst piece of software with which I come in contact on a daily basis
- small unlink fix

22 years ago- rebase b_io against HEAD in preparation for the latest write cache code
zab [Sat, 15 Feb 2003 20:21:37 +0000 (20:21 +0000)]
- rebase b_io against HEAD in preparation for the latest write cache code

22 years agoinitial check-in for SAN support. There's still not clean
nfshp [Wed, 12 Feb 2003 08:57:21 +0000 (08:57 +0000)]
initial check-in for SAN support. There's still not clean
stuff like mount option(data=writeback) of ost.