Whamcloud - gitweb
fs/lustre-release.git
21 years agoMove version back to its original location, and put flags at the end.
adilger [Tue, 29 Oct 2002 07:56:58 +0000 (07:56 +0000)]
Move version back to its original location, and put flags at the end.
It is better to have version close to the beginning of the message, so
that it is possible to change fields after it without breaking things.

21 years agoQuiet compiler warning.
adilger [Tue, 29 Oct 2002 07:52:43 +0000 (07:52 +0000)]
Quiet compiler warning.

21 years agoWhile we are breaking BA OST compatibility, swap the osc_punch packing of
adilger [Tue, 29 Oct 2002 07:51:51 +0000 (07:51 +0000)]
While we are breaking BA OST compatibility, swap the osc_punch packing of
"start" and "end" so that "start" is in o_size, which is more consistent
with setting i_size on truncate.

Also add some more FIXMEs for bug 288.

21 years agoRe-do change that Phil killed in his ll_lookup2()/ll_invalidate_inode() merge.
adilger [Tue, 29 Oct 2002 07:35:34 +0000 (07:35 +0000)]
Re-do change that Phil killed in his ll_lookup2()/ll_invalidate_inode() merge.

21 years agoDisable the code which sets the file size on the MDS:
adilger [Tue, 29 Oct 2002 06:29:18 +0000 (06:29 +0000)]
Disable the code which sets the file size on the MDS:
1) we never (or should never) currently use the file size from the MDS
2) only the client which holds the size lock would need to update the MDS
3) that client already knows the correct file size locally, so no obd_getattr

Removing this hunk of code also has the benefit of removing an extraneous
obd_getattr() x stripe_count + mdc_setattr() per file, so it should improve
performance for tar/bonnie/etc noticably.

Passes multi-client runtests.

21 years agob=256
pschwan [Tue, 29 Oct 2002 05:27:13 +0000 (05:27 +0000)]
b=256
abstract ll_lookup2; make ll_lookup2 and ll_revalidate2 both use this
function.  For the first time, we might almost be properly cache coherant.

21 years agosacrifice 80 the column margin on one line for the broader goal of compiling on
rread [Tue, 29 Oct 2002 00:27:07 +0000 (00:27 +0000)]
sacrifice 80 the column margin on one line for the broader goal of compiling on
older versions of gcc.

21 years agoPut back lustre_msg.version (in a different spot, though, so out-of-date clients
shaver [Mon, 28 Oct 2002 23:42:49 +0000 (23:42 +0000)]
Put back lustre_msg.version (in a different spot, though, so out-of-date clients
will still barf -- I think this is a feature).

Some ChangeLog updates.

21 years agoAdd some null-proofing in DEBUG_REQ, and supply arguments for all the format
shaver [Mon, 28 Oct 2002 20:23:09 +0000 (20:23 +0000)]
Add some null-proofing in DEBUG_REQ, and supply arguments for all the format
fields (!) while we're at it.

Bullet-proof expired_request to see if we can figure out why we sometimes see
NULL derefs during timeout on dev.

Make the cleanup-from-sending-list stuff use D_ERROR, so that we can see it even
if the client crashes afterwards.  (Just playing a hunch.)

21 years agoOnly initialize fdd in terms of object_dentry once object_dentry has itself been
shaver [Mon, 28 Oct 2002 20:18:09 +0000 (20:18 +0000)]
Only initialize fdd in terms of object_dentry once object_dentry has itself been
initialized.

Also remove shadowing redecl of fdd.

21 years agoBackout sravi's changes; please contact Robert with your patches.
pschwan [Mon, 28 Oct 2002 19:30:55 +0000 (19:30 +0000)]
Backout sravi's changes; please contact Robert with your patches.

21 years agoFix space leak after recent change of openunlink code logic (fdd was
adilger [Mon, 28 Oct 2002 19:25:27 +0000 (19:25 +0000)]
Fix space leak after recent change of openunlink code logic (fdd was
previously released at d_delete time and is now done at d_release time,
so the mere presence of d_fsdata isn't enough to signal an open file).

21 years agoLconf modified accordingy lmc, to handle mdc info in the xml file.
sravi [Mon, 28 Oct 2002 19:19:19 +0000 (19:19 +0000)]
Lconf modified accordingy lmc, to handle mdc info in the xml file.

21 years ago- Fixed uninitialized flags, which tricked ldlm_cli_enqueue into thinking
pschwan [Mon, 28 Oct 2002 19:10:32 +0000 (19:10 +0000)]
- Fixed uninitialized flags, which tricked ldlm_cli_enqueue into thinking
that it was in recovery

21 years agoModified the this to generate mdc configuration.
sravi [Mon, 28 Oct 2002 18:49:49 +0000 (18:49 +0000)]
Modified the this to generate mdc configuration.

21 years agoPer Mike's observation, we already set lic.lic_body before this point.
adilger [Mon, 28 Oct 2002 17:49:14 +0000 (17:49 +0000)]
Per Mike's observation, we already set lic.lic_body before this point.

21 years agoAvoid CVS permission problems.
adilger [Mon, 28 Oct 2002 17:40:28 +0000 (17:40 +0000)]
Avoid CVS permission problems.

21 years agoShort version: replaying create and rename works now, including all the fixups
shaver [Mon, 28 Oct 2002 02:49:56 +0000 (02:49 +0000)]
Short version: replaying create and rename works now, including all the fixups
               you could ever, ever want.  (OK, except lock replay, but that's
               easy now.)  Single-client recovery should now be much, much
               more robust.

- Added LDLM_FL_INTENT_ONLY to avoid acquiring locks during intent-lock replay.
- Fix leak of open requests that was due to a misunderstanding (mine) of the
  way that the ptlrpc_request refcounting code (mine) really worked.
- Store FID data after create, so that we can fix up the world afterwards.
- Poor man's lustre_spare_buf support through simple over-allocation of request
  buffers.
- Cancel unused locks when we lose contact with an MDS.
- When replaying create, fix up to-be-replayed and delayed-for-recovery requests
  to contain the new FID generation, in addition to the actual in-memory inode.
- Don't sweat a "failed" return code for replayed requests, but do squeak a bit
  if the old and new status doesn't match up (virtually guaranteed, in fact,
  but I'll tune that down later).
- Skip no-transno requests during replay, since they don't affect MDS state.

21 years ago- Pass flags into lock policy function
pschwan [Mon, 28 Oct 2002 01:18:30 +0000 (01:18 +0000)]
- Pass flags into lock policy function

21 years agomost of the work for lock replay:
pschwan [Sun, 27 Oct 2002 03:03:50 +0000 (03:03 +0000)]
most of the work for lock replay:
  - add a dlm replay flag that:
    - prevents the calling of policy functions
    - tells the server to trust the client's judgement
  - a function to call ldlm_cli_enqueue for replay

next up:
  - something to help replay _all_ locks in a given namespace, probably in
    the form of some generic ldlm_namespace_foreach thing

21 years agoOne lock too many.
adilger [Sat, 26 Oct 2002 23:58:10 +0000 (23:58 +0000)]
One lock too many.

21 years agoOops, one more.
pschwan [Sat, 26 Oct 2002 18:54:22 +0000 (18:54 +0000)]
Oops, one more.

21 years ago- Fixes build break and &&-instead-of-& logic error
pschwan [Sat, 26 Oct 2002 18:40:00 +0000 (18:40 +0000)]
- Fixes build break and &&-instead-of-& logic error
- Andreas, will you please verify that I guessed your intentions with respect
  to spinlocks correctly?  In particular, the fed_lock and the extra
  spin_lock_init

21 years agoSupport for delayed-destroy of objects on the OST. It isn't perfect,
adilger [Sat, 26 Oct 2002 08:29:39 +0000 (08:29 +0000)]
Support for delayed-destroy of objects on the OST.  It isn't perfect,
in the sense that there is still some work to be done for handling
multi-client unlink properly, but it passes all of the normal tests,
so it is no worse than what we have now.

21 years ago- Convert version to flags -- a few days earlier than I said, but I'll back it
shaver [Fri, 25 Oct 2002 21:17:07 +0000 (21:17 +0000)]
- Convert version to flags -- a few days earlier than I said, but I'll back it
  out if people complain.
- Add some helpers for accessing the lower (general) and upper (op-specific)
  portions of the flags.
- Use an op flag (MDS_OPEN_HAS_EA) rather than bufcount checking to determine
  whether or not an open has an EA for delayed-create.
- Remove basically-unused rq_reply_md_h.
- Pack open-replay state in extra message buffer, instead of dangling it off the
  request.
- Add DEBUG_REQ for consistent and clean request-dumping, and use it somewhat
  liberally.
- Teach reint_link to be tolerant of -EEXIST if replaying.
- Likewise for reint_unlink and -ENOENT.
- Remove connection locking from ptlrpc_check_reply.  We don't lock in
  queue_wait, so if it's needed I need a better story anyway, and in the interim
  this will fix a guaranteed deadlock in replay.
- Factor sending-queue request disposition checks for legibility and reuse.
- Dump retained-request state before starting replay, to aid debugging.

21 years agoProc header for SNMP. Counter functionality/macros will go into a different header...
thantry [Fri, 25 Oct 2002 06:05:32 +0000 (06:05 +0000)]
Proc header for SNMP. Counter functionality/macros will go into a different header file

21 years ago- Update ChangeLog
pschwan [Fri, 25 Oct 2002 03:43:08 +0000 (03:43 +0000)]
- Update ChangeLog
- only send cancels to the LDLM portal, since Peter pointed out that my last
  "fix" didn't really fix anything

21 years agob=611336
pschwan [Fri, 25 Oct 2002 01:15:43 +0000 (01:15 +0000)]
b=611336
Land Alok's patch with a couple small modifications (which were my fault,
for lying to him about the return values of osc_brw_{read,write})

21 years agoQuiet compiler warning.
adilger [Fri, 25 Oct 2002 00:05:38 +0000 (00:05 +0000)]
Quiet compiler warning.

21 years agoDon't oops if data->ioc_inlbuf1 is not set.
adilger [Thu, 24 Oct 2002 21:30:22 +0000 (21:30 +0000)]
Don't oops if data->ioc_inlbuf1 is not set.

21 years agoReturn a valid error code on normal exit.
adilger [Thu, 24 Oct 2002 21:08:35 +0000 (21:08 +0000)]
Return a valid error code on normal exit.

21 years agoNew lmc manpage that describes the new lmc command line interface.
radhikav [Thu, 24 Oct 2002 18:47:04 +0000 (18:47 +0000)]
New lmc manpage that describes the new lmc command line interface.

21 years agoPatch for tinderbox
amrutjoshi [Thu, 24 Oct 2002 15:51:18 +0000 (15:51 +0000)]
Patch for tinderbox

21 years agoImproved patch.. for the portalslib
amrutjoshi [Thu, 24 Oct 2002 14:43:00 +0000 (14:43 +0000)]
Improved patch.. for the portalslib

21 years agoPatched configure.in to get the correct portalslib. Usually $enable_portalsdir/utils...
amrutjoshi [Thu, 24 Oct 2002 13:43:21 +0000 (13:43 +0000)]
Patched configure.in to get the correct portalslib. Usually $enable_portalsdir/utils/linux. To override use --with-portalslib=<dir>

21 years agoChangeLog for 0.5.15
pschwan [Thu, 24 Oct 2002 04:14:32 +0000 (04:14 +0000)]
ChangeLog for 0.5.15

21 years ago- Fix recovery crash bug in prepare_ost()
pschwan [Thu, 24 Oct 2002 03:59:56 +0000 (03:59 +0000)]
- Fix recovery crash bug in prepare_ost()

21 years agob=625069
pschwan [Thu, 24 Oct 2002 00:54:44 +0000 (00:54 +0000)]
b=625069
- Workaround for 625069: send all LDLM requests, not just callbacks, to the
  LDLM portal
- change CERROR in lov_getattr into CDEBUG

21 years ago"force" should _cause_ forcing, not disable forcing. Thanks to robert for the
shaver [Thu, 24 Oct 2002 00:51:03 +0000 (00:51 +0000)]
"force" should _cause_ forcing, not disable forcing.  Thanks to robert for the
catch.

21 years agoA patch only a mother could love to notify LOVs when OSCs are disabled due to
shaver [Thu, 24 Oct 2002 00:32:56 +0000 (00:32 +0000)]
A patch only a mother could love to notify LOVs when OSCs are disabled due to
OST failure or partition.

Remember to re-enable it in your upcall, after reconnecting!

21 years agoFix LOV stripe allocation.
adilger [Wed, 23 Oct 2002 23:07:03 +0000 (23:07 +0000)]
Fix LOV stripe allocation.
Don't re-assign lsm_stripe_offset because it is 0 (which is a valid offset).

21 years agoMake the cleanup and remount scripts consistent in accepting "NAME" as
adilger [Wed, 23 Oct 2002 22:06:01 +0000 (22:06 +0000)]
Make the cleanup and remount scripts consistent in accepting "NAME" as
the basename of the config script, so you can do:

NAME=lov sh llmount.sh
NAME=lov sh llmountcleanup.sh

and it will work.

Also fix a bug in llmount.sh, which I think would have prevented the
XML config file from being regenerated properly.

21 years ago- Beginning of new REINT replay infrastructure.
shaver [Wed, 23 Oct 2002 18:57:33 +0000 (18:57 +0000)]
- Beginning of new REINT replay infrastructure.
- Poison imports for failed OSCs, so future requests get -EIO.
- Remove the epoch checking for mounts and inodes, because we don't want to
  kill a while filesystem because a single OSC died.
- Close open files on MDS-disconnect.
- If asked with a "force" parameter, disconnect all remaining exports during
  OBD disconnection.
- Teach lconf to send that parameter when given -f.

21 years ago* startlevel and endlevel patch supplied by Evan.Felix@pnl.gov
rread [Wed, 23 Oct 2002 10:50:05 +0000 (10:50 +0000)]
* startlevel and endlevel patch supplied by Evan.Felix@pnl.gov

21 years agoSome random double pointer error..
thantry [Wed, 23 Oct 2002 03:18:36 +0000 (03:18 +0000)]
Some random double pointer error..

21 years agoFile containing read/write fptr implementations for PTLRPC device. To be added to...
thantry [Wed, 23 Oct 2002 03:05:50 +0000 (03:05 +0000)]
File containing read/write fptr implementations for PTLRPC device. To be added to make path after review

21 years agoFile containing read/write fptr implementations for OST device. To be added to make...
thantry [Wed, 23 Oct 2002 03:05:21 +0000 (03:05 +0000)]
File containing read/write fptr implementations for OST device. To be added to make path after review

21 years agoFile containing read/write fptr implementations for OSC device. To be added to make...
thantry [Wed, 23 Oct 2002 03:04:57 +0000 (03:04 +0000)]
File containing read/write fptr implementations for OSC device. To be added to make path after review

21 years agoFile containing read/write fptr implementations for OBDFILTER device. To be added...
thantry [Wed, 23 Oct 2002 03:04:22 +0000 (03:04 +0000)]
File containing read/write fptr implementations for OBDFILTER device. To be added to make path after review

21 years agoFile containing read/write fptr implementations for echo device. To be added to make...
thantry [Wed, 23 Oct 2002 03:03:47 +0000 (03:03 +0000)]
File containing read/write fptr implementations for echo device. To be added to make path after review

21 years agoFile containing read/write fptr implementations for MDS. To be added to make path...
thantry [Wed, 23 Oct 2002 03:03:16 +0000 (03:03 +0000)]
File containing read/write fptr implementations for MDS. To be added to make path after review

21 years agoFile containing read/write fptr implementations for MDC. To be added to make path...
thantry [Wed, 23 Oct 2002 03:02:49 +0000 (03:02 +0000)]
File containing read/write fptr implementations for MDC. To be added to make path after review

21 years agoFile containing read/write fptr implementations for LOV. To be added to make path...
thantry [Wed, 23 Oct 2002 03:02:20 +0000 (03:02 +0000)]
File containing read/write fptr implementations for LOV. To be added to make path after review

21 years agoFile containing read/write fptr implementations for mount_point. To be added to make...
thantry [Wed, 23 Oct 2002 03:01:45 +0000 (03:01 +0000)]
File containing read/write fptr implementations for mount_point. To be added to make path after review

21 years agoAdding to the source repository. Will be linked in to the makefile after
thantry [Wed, 23 Oct 2002 02:59:02 +0000 (02:59 +0000)]
Adding to the source repository. Will be linked in to the makefile after
review.

21 years agoPass ioctls on LOV device through to all underlying OSCs if unknown.
adilger [Tue, 22 Oct 2002 22:36:40 +0000 (22:36 +0000)]
Pass ioctls on LOV device through to all underlying OSCs if unknown.

21 years agoFix LOV striping code:
adilger [Tue, 22 Oct 2002 21:22:58 +0000 (21:22 +0000)]
Fix LOV striping code:
- don't set default stripe count at lov_connect() time if it is zero, defer
  it to lov_create() time in case the number of OSTs has changed
- validate the "stripe width < 4GB" requirement at lov_create() time, in
  case bad user-specified striping patterns are given or #OSTs has increased
- don't override lsm_stripe_count if it is already set
- choose from all OSTs as the starting OST, instead of only [0-active]
- return -ENOTTY for unimplemented ioctls

21 years agoMinor tweak to match RH 2.4.18-14 function name in kernel.
adilger [Tue, 22 Oct 2002 20:41:47 +0000 (20:41 +0000)]
Minor tweak to match RH 2.4.18-14 function name in kernel.

21 years agoRemove old -12.5 kernel patch, the -14 patch is exactly the same, and uptodate.
adilger [Tue, 22 Oct 2002 20:33:28 +0000 (20:33 +0000)]
Remove old -12.5 kernel patch, the -14 patch is exactly the same, and uptodate.

21 years ago- teeny fix to ll_revalidate2; this will be trumped by my fix for 626244
pschwan [Tue, 22 Oct 2002 06:55:14 +0000 (06:55 +0000)]
- teeny fix to ll_revalidate2; this will be trumped by my fix for 626244
  in ways that can scarcely be described by mere words
- multiple CONNECTs to the LOV would do more CONNECTs to the OSCs, each
  clobbering the last connection and preventing cleanup; fixed.
- LOV targets now have an 'active' bit; inactive targets will be skipped at
  CREATE and DISCONNECT time
- added locking for the above, along with a function to toggle the bit given an
  LOV and a UUID
- added a LOV_SET_OSC_ACTIVE ioctl to the lov driver and lctl

21 years ago- fixes to osc_brw path to address fsx failures. We in fact uncovered
braam [Mon, 21 Oct 2002 23:37:39 +0000 (23:37 +0000)]
- fixes to osc_brw path to address fsx failures. We in fact uncovered
  serious problems with writes of less than one page.
- minor fixes for 2.5

21 years agoCommit new intent patch for 2.4.18.
adilger [Mon, 21 Oct 2002 22:05:33 +0000 (22:05 +0000)]
Commit new intent patch for 2.4.18.

21 years agoRemove sig from end of patch.
adilger [Mon, 21 Oct 2002 22:05:11 +0000 (22:05 +0000)]
Remove sig from end of patch.

21 years agoThe curse of copy-n-paste. Fix compile warnings and maybe code too.
adilger [Mon, 21 Oct 2002 21:45:15 +0000 (21:45 +0000)]
The curse of copy-n-paste.  Fix compile warnings and maybe code too.

21 years agoGetting ready to land OBD open file handles, and OSC/LOV local lock handle
adilger [Mon, 21 Oct 2002 21:32:53 +0000 (21:32 +0000)]
Getting ready to land OBD open file handles, and OSC/LOV local lock handle
allocations.

21 years agoA little debugging I've been running in my tree for a long time.
adilger [Mon, 21 Oct 2002 21:31:04 +0000 (21:31 +0000)]
A little debugging I've been running in my tree for a long time.
It caught a couple of minor bugs, and should prevent more in the future.

21 years agoAdd lov_getconfig to lctl docs.
adilger [Mon, 21 Oct 2002 21:25:05 +0000 (21:25 +0000)]
Add lov_getconfig to lctl docs.

21 years agoChange lctl lovconfig command into lov_getconfig, and add lov_setconfig.
adilger [Mon, 21 Oct 2002 21:16:46 +0000 (21:16 +0000)]
Change lctl lovconfig command into lov_getconfig, and add lov_setconfig.
Update all scripts/docs which refer to the lovconfig command.

lov_getconfig compiles OK, and does not affect normal operation, but it
is not working yet (user-space segfault) and I need to run it on a real
system (can't run GDB under UML).

21 years agoAdd permission bits to open(O_CREAT) (b=624321).
adilger [Mon, 21 Oct 2002 18:26:53 +0000 (18:26 +0000)]
Add permission bits to open(O_CREAT) (b=624321).
Add a verifiable data pattern so that we can confirm if data is going to the
correct stripe.

21 years agoremoved demos/Makefile from configure.in
eeb [Mon, 21 Oct 2002 17:29:53 +0000 (17:29 +0000)]
removed demos/Makefile from configure.in

21 years agoOops, missed a hunk.
pschwan [Mon, 21 Oct 2002 07:17:26 +0000 (07:17 +0000)]
Oops, missed a hunk.

21 years agoMove AM_CONDITIONAL into configure.in
pschwan [Mon, 21 Oct 2002 07:16:15 +0000 (07:16 +0000)]
Move AM_CONDITIONAL into configure.in

21 years agoInclude version.h so we can check version.
adilger [Mon, 21 Oct 2002 07:05:15 +0000 (07:05 +0000)]
Include version.h so we can check version.

21 years ago- fix to deal with other versions of automake.... :(
braam [Mon, 21 Oct 2002 07:03:25 +0000 (07:03 +0000)]
- fix to deal with other versions of automake.... :(

21 years ago- compile fixes for 2.5 -- somewhat tentative here and there.
braam [Mon, 21 Oct 2002 06:47:07 +0000 (06:47 +0000)]
- compile fixes for 2.5 -- somewhat tentative here and there.

21 years ago2.4 fixes
braam [Mon, 21 Oct 2002 06:35:22 +0000 (06:35 +0000)]
2.4 fixes

21 years ago- 2.5 fixes for MDS
braam [Mon, 21 Oct 2002 05:58:41 +0000 (05:58 +0000)]
- 2.5 fixes for MDS

21 years ago- compile fixes for 2.5.44
braam [Mon, 21 Oct 2002 04:25:20 +0000 (04:25 +0000)]
- compile fixes for 2.5.44
- echo works again

21 years ago- put obdfilter before extN + mds to facilitate collab between peter +
braam [Sun, 20 Oct 2002 22:40:11 +0000 (22:40 +0000)]
- put obdfilter before extN + mds to facilitate collab between peter +
  andreas on 2.5

21 years ago2.4 build fixes
pschwan [Sun, 20 Oct 2002 19:18:38 +0000 (19:18 +0000)]
2.4 build fixes

21 years ago- Lustre Lite at least superficially ready for 2.5
braam [Sun, 20 Oct 2002 19:09:24 +0000 (19:09 +0000)]
- Lustre Lite at least superficially ready for 2.5

21 years ago- Back out the ldlm_everything_lock patch, which causes a deadlock on cleanup
pschwan [Sun, 20 Oct 2002 00:16:14 +0000 (00:16 +0000)]
- Back out the ldlm_everything_lock patch, which causes a deadlock on cleanup
- Move backslashes in lustre_lite.h macros to fit in 80 columns
- Added a couple of structure fields that support my LOV work

21 years ago- buffer checks are different for non-intent
braam [Sat, 19 Oct 2002 22:43:14 +0000 (22:43 +0000)]
- buffer checks are different for non-intent
- remove a it = ->d_it assignment
- set the intent at the end of lookup2

21 years ago- fix rdev stuff
braam [Sat, 19 Oct 2002 19:53:09 +0000 (19:53 +0000)]
- fix rdev stuff
- assign d_fsdata in lookup2

21 years ago- keep extN before MDS
braam [Sat, 19 Oct 2002 16:43:42 +0000 (16:43 +0000)]
- keep extN before MDS

21 years agoRestore Phil's header fix that I whacked.
braam [Sat, 19 Oct 2002 16:29:00 +0000 (16:29 +0000)]
Restore Phil's header fix that I whacked.

21 years agoONLY UPDATE IF YOU NEED THIS (i.e. Andreas probably will)
braam [Sat, 19 Oct 2002 16:09:43 +0000 (16:09 +0000)]
ONLY UPDATE IF YOU NEED THIS (i.e. Andreas probably will)

- more fixes for 2.5 builds, much overlap with Andreas' check in from
  yesterday.

- add vfs_inode to Lustre inode info: for 2.5 this makes this the
  whole inode.

- two versions of intent_release: Robert's and the new one for 2.5

- Robert's patch - so this may break some stuff, we'll sort that out.

- convert device numbers

- small fixes to make thread handling compatible with 2.5

- always compile super25.c and super.c; the contents is conditional on
  the version

21 years ago- Replace per-namespace recursive lock with an ldlm-global lock, to close the
pschwan [Sat, 19 Oct 2002 11:33:33 +0000 (11:33 +0000)]
- Replace per-namespace recursive lock with an ldlm-global lock, to close the
  race between ldlm_handle2lock and the lock destroy portion of ldlm_lock_put
- Add a special interface for ldlm_handle2lock to resolve the following
  conflict between our invariants:

  1. ldlm_handle2lock should never give a reference to a "dying" lock
  2. ldlm_lock_decref needs to use ldlm_handle2lock to get a reference to a lock
     so that it can decrease the readers/writers refcount
  3. The lock can't finish dying until readers/writers hits zero.

  That should be fixed now.

21 years ago- Split import reconnection and replay (OSC only needs to reconnect).
shaver [Sat, 19 Oct 2002 11:19:14 +0000 (11:19 +0000)]
- Split import reconnection and replay (OSC only needs to reconnect).
- Restore recovd-management of server-side connections (why did I take that out
  in the first place?).
- Abort inflight, uncommitted and waiting requests for imports that are
  invalidated by reconnection to OST.  (Generates -EIO, or should!)
- Mete out harsh, harsh justice to locks held by recovery-invalidated imports.
- Remove LBUG()s now that some error returns from ldlm_cli_cancel are "normal".
- Cancel locks on disconnection from OST (really filter).
- Ignore replies that come in after we've started recovery on a given connection.
  The socknal's implicit retransmission can otherwise cause us worlds and worlds
  of hurt.
- Wake up when we get aborted, and return -EIO.
- Remove connections from the recovd's care when they go unused.  (Which might
  actually happen, once Phil lands his refcounting fixes!)

21 years ago- Rename ldlm_namespace_cleanup's "local" parameter to "local_only" for clarity.
shaver [Sat, 19 Oct 2002 11:09:16 +0000 (11:09 +0000)]
- Rename ldlm_namespace_cleanup's "local" parameter to "local_only" for clarity.
- Remove spurious LBUG() or two.
- Export the cleanup and dump symbols we use in recovery.

21 years agoFix the connection refcount leaks. There were actually 3 separate bugs.
pschwan [Sat, 19 Oct 2002 09:12:24 +0000 (09:12 +0000)]
Fix the connection refcount leaks.  There were actually 3 separate bugs.

21 years agoUnlockPage->unlock_page. The two are identical, except that the former
adilger [Sat, 19 Oct 2002 06:21:46 +0000 (06:21 +0000)]
UnlockPage->unlock_page.  The two are identical, except that the former
does not exist in 2.5.

21 years agoHiding will not save you.
adilger [Fri, 18 Oct 2002 23:43:17 +0000 (23:43 +0000)]
Hiding will not save you.

21 years agoDon't refer to junk which doesn't exist.
adilger [Fri, 18 Oct 2002 23:39:18 +0000 (23:39 +0000)]
Don't refer to junk which doesn't exist.

21 years agoYet more junk.
adilger [Fri, 18 Oct 2002 23:38:54 +0000 (23:38 +0000)]
Yet more junk.

21 years agoMore cruftectomy.
adilger [Fri, 18 Oct 2002 23:37:55 +0000 (23:37 +0000)]
More cruftectomy.

21 years agoRemove demos directory entirely.
adilger [Fri, 18 Oct 2002 23:35:27 +0000 (23:35 +0000)]
Remove demos directory entirely.

21 years agoRemove ancient and unworking snapshot demo scripts.
adilger [Fri, 18 Oct 2002 23:34:45 +0000 (23:34 +0000)]
Remove ancient and unworking snapshot demo scripts.

21 years agoOne more straggler in obdfs.
adilger [Fri, 18 Oct 2002 23:27:46 +0000 (23:27 +0000)]
One more straggler in obdfs.

21 years agoRemove ancient and entirely non-working obdfs code from the HEAD.
adilger [Fri, 18 Oct 2002 23:20:52 +0000 (23:20 +0000)]
Remove ancient and entirely non-working obdfs code from the HEAD.