Whamcloud - gitweb
scjody [Wed, 12 Apr 2006 12:57:40 +0000 (12:57 +0000)]
bonsai test; please ignore
scjody [Tue, 11 Apr 2006 20:38:51 +0000 (20:38 +0000)]
bonsai test; please ignore
scjody [Tue, 11 Apr 2006 20:19:51 +0000 (20:19 +0000)]
bonsai test; please ignore
mjmac [Fri, 7 Apr 2006 17:47:04 +0000 (17:47 +0000)]
* turns out we actually need to build against real sources for patchless
* massage the kernel Makefile's EXTRAVERSION to match the running kernel
brian [Thu, 6 Apr 2006 14:15:05 +0000 (14:15 +0000)]
The variable, not the literal. Thanks mjmac.
brian [Thu, 6 Apr 2006 14:14:14 +0000 (14:14 +0000)]
For the patchless client, LNET needs to come from it's own branch too.
brian [Thu, 6 Apr 2006 14:03:12 +0000 (14:03 +0000)]
Revert my use of the configure_flags macro in the lustre.spec file and just
pass the --disable-server argument to configure with the CONFIGURE_FLAGS
variable.
Prefer (by boolean switch) mjmac's use of kernel-devel for building the
patchless client.
brian [Wed, 5 Apr 2006 17:35:48 +0000 (17:35 +0000)]
~sigh~ rpm macros are not empty if undefined.
wangyb [Wed, 5 Apr 2006 07:46:05 +0000 (07:46 +0000)]
No RPM is built when NO_SRC=0, but we should have the rpms being built.
mjmac [Wed, 5 Apr 2006 02:02:36 +0000 (02:02 +0000)]
* fix a tyop
* add some sanity-checking
brian [Wed, 5 Apr 2006 00:30:27 +0000 (00:30 +0000)]
Need a bit more kernel prep fu. Thanx Cliff and bug 7108.
Need to be able to tell the lustre.spec which kernel sources to build against.
lmake "cheats" and re-substs the lustre.spec.in template file.
brian [Tue, 4 Apr 2006 22:33:32 +0000 (22:33 +0000)]
Of course, we need to work outside of $DO_SRC == 1.
brian [Tue, 4 Apr 2006 22:02:15 +0000 (22:02 +0000)]
Ahhh. For patchless client, we still need the kernel source unpacked and
configure needs to know where it is.
brian [Tue, 4 Apr 2006 21:04:12 +0000 (21:04 +0000)]
A lack of a SERIES file specified in a target file means that we don't have
patches for a kernel, so why build one. Specifically this is for building a
patchless client.
This should all work if the concept of:
if [ -z "$LINUX" ] ; then
...
else
build_lustre
fi
works. I don't know that we test the build_lustre() function of lbuild with
any regularity, if at all though.
brian [Tue, 4 Apr 2006 20:47:45 +0000 (20:47 +0000)]
b_release_1_4_6-patchless is now using lustre-build from HEAD. Happy hacking.
brian [Tue, 4 Apr 2006 20:45:54 +0000 (20:45 +0000)]
r=green
b_release_1_4_6-patchless is now using lustre-build from HEAD. Happy hacking.
cliffw [Mon, 3 Apr 2006 15:55:06 +0000 (15:55 +0000)]
Removed per Peter Bojanic. Will return if the code can be fixed and tested.
cliffw [Fri, 31 Mar 2006 16:53:37 +0000 (16:53 +0000)]
b=10371
r=scjody@clusterfs.com
Fixed error with multiple scsidevs
zhanghc [Wed, 29 Mar 2006 10:56:02 +0000 (10:56 +0000)]
b=10329
DESCRIPTION:
When doing fcntl() file locking calls in a liblustre client via libsysio,
successful lock calls return random non-zero values due to use uninitialized
value as return value.
zhanghc [Wed, 29 Mar 2006 06:24:50 +0000 (06:24 +0000)]
b=10163
DESCRIPTION:
keep CFS up to date with latest libsysio version
wangdi [Tue, 28 Mar 2006 02:57:14 +0000 (02:57 +0000)]
Branch: HEAD
add b_new_cmd in buildcvs
brian [Fri, 24 Mar 2006 21:05:36 +0000 (21:05 +0000)]
Trickery to get the exit status of strace's subject.
brian [Fri, 24 Mar 2006 18:06:20 +0000 (18:06 +0000)]
strace that update_oldconfig to see why make oldconfig is timing out the expect script.
wangyb [Fri, 24 Mar 2006 04:44:37 +0000 (04:44 +0000)]
fix shebang -d problem
cliffw [Thu, 23 Mar 2006 22:21:33 +0000 (22:21 +0000)]
Added support for raw devices
brian [Thu, 23 Mar 2006 22:10:29 +0000 (22:10 +0000)]
Get some diagnostic output from expect to see if I can see why this fails on
RHEL3 on only boulder. Attempts to reproduce in a debugging environment are
fruitless. ~sigh~
scjody [Thu, 23 Mar 2006 20:58:04 +0000 (20:58 +0000)]
Merge ... old CVS version; this should now contain the best of old CVS and
20051107 from the website.
brian [Tue, 21 Mar 2006 20:20:29 +0000 (20:20 +0000)]
r=adilger
Try to find the cvs-modified-files.pl in the same place as the cvsdiffclient
if it's not in $PATH.
brian [Tue, 21 Mar 2006 16:19:32 +0000 (16:19 +0000)]
Be a bit more verbose than just "timeout".
brian [Tue, 21 Mar 2006 15:55:51 +0000 (15:55 +0000)]
Yet another expression found in the RHEL3/x86_64 make oldconfig.
brian [Tue, 21 Mar 2006 06:56:44 +0000 (06:56 +0000)]
Another different make config pattern. Where are these all coming from all
of a sudden?
brian [Mon, 20 Mar 2006 21:11:25 +0000 (21:11 +0000)]
Put the return status in a variable, and use the PIPESTATUS array instead of
the easily broken $?.
brian [Mon, 20 Mar 2006 18:48:33 +0000 (18:48 +0000)]
Recognize and answer an additional expression.
Be a bit more liberal with the format of the questions.
brian [Mon, 20 Mar 2006 18:09:05 +0000 (18:09 +0000)]
Apparently it's lindex, not index.
brian [Mon, 20 Mar 2006 16:38:32 +0000 (16:38 +0000)]
Try to get some more information when update_oldconfig fails. Probably some
missing expect patterns, etc.
scjody [Fri, 17 Mar 2006 17:31:36 +0000 (17:31 +0000)]
Branch b1_4
b=10143
r=adilger
Add --disable-readline option, make it the default for --enable-bgl.
adilger [Wed, 15 Mar 2006 19:44:11 +0000 (19:44 +0000)]
Branch b1_4
Remove SubVersion files on "make dist".
brian [Sat, 11 Mar 2006 17:37:12 +0000 (17:37 +0000)]
Build fix for RHEL3 native compiler on x86_64.
Hopefully doesn't break any other builds.
scjody [Fri, 3 Mar 2006 17:54:10 +0000 (17:54 +0000)]
Comment out SCSI devices in sgpdd-survey per Peter Braam
scjody [Fri, 3 Mar 2006 17:53:03 +0000 (17:53 +0000)]
Update based on lustre-iokit-
20051107.tar.gz from the customer download area.
liangzhen [Fri, 3 Mar 2006 05:56:31 +0000 (05:56 +0000)]
Smallfix for branch using universal headers(b1_5*), no effect to other branches.
alex [Thu, 2 Mar 2006 19:49:50 +0000 (19:49 +0000)]
- minor tweaks for b_release_1_4_6-patchless
liangzhen [Thu, 2 Mar 2006 13:39:55 +0000 (13:39 +0000)]
. Small change to lustre.spec.in to resolve problem in rpm building,
it's tested in b1_4 and b1_5.
. Fix rpm building problem in b1_5
. Rename $(platform)/types.h to $(platform)/lustre_types.h to avoid
confusing while compile.
. portability fix for lov_obd.c
ericm [Sun, 26 Feb 2006 23:04:58 +0000 (23:04 +0000)]
branch: b1_5
update from b1_4.
adilger [Thu, 23 Feb 2006 23:16:18 +0000 (23:16 +0000)]
Branch: HEAD
Add snmp module to all future builds by default.
adilger [Thu, 23 Feb 2006 22:06:36 +0000 (22:06 +0000)]
Land b_release_1_4_6 onto HEAD (20060223_1455)
* the b1_5_* and b_iam* branches will to continue to use the old libsysio
* until they can be updated from b1_4 liblustre because of the API change.
- API changegetdirentries->filldirentries changed from upstream
- symlink following fixup
- fix getcwd for out-of-libsysio directory
- lseek fixups for large files
- flock/fcntl fixups for 64-bit systems
adilger [Thu, 23 Feb 2006 21:53:34 +0000 (21:53 +0000)]
Land b_release_1_4_6 onto HEAD (20060223_1429)
- for some short time (until libsysio is also merged) some branches will
check out the wrong libsysio. Should be fixed before it is noticed.
huanghua [Wed, 15 Feb 2006 07:35:03 +0000 (07:35 +0000)]
add a branch and its table named "b_iam" for "Index API Module" from #colibri
This branch will be used to develop & test iam.
liangzhen [Thu, 26 Jan 2006 06:38:49 +0000 (06:38 +0000)]
Correct gcc flags for Tiger.
zhanghc [Mon, 23 Jan 2006 06:13:26 +0000 (06:13 +0000)]
update Entries to check out right lnet module
Entries:
b1_4_next_recovery_transno
b1_4_atime_update
b1_4_lfs_df
b1_4_lov_lvb_cleanup
zhanghc [Mon, 23 Jan 2006 05:51:24 +0000 (05:51 +0000)]
add a new Entry(b1_4_atime_update) in order to checkout the right modules
zhanghc [Fri, 20 Jan 2006 06:01:14 +0000 (06:01 +0000)]
add one more tag entry(b1_4_lfs_df) in buildcvs,
so this script can check out right libsysio module
brian [Tue, 17 Jan 2006 22:09:41 +0000 (22:09 +0000)]
Use sendmail and set the sender address to ensure that mail is not sent from a
domain name that is neither an A or MX record (and thusly gets refused by
mail servers).
adilger [Tue, 17 Jan 2006 05:16:43 +0000 (05:16 +0000)]
Branch HEAD
Check for presence of kernel sources before we check for $LINUX/.config
to avoid confusing (and common) error message that $LINUX/.config could
not be found (normally because kernel is not configured and not because
incorrect path was given to configure).
zhanghc [Thu, 12 Jan 2006 07:06:52 +0000 (07:06 +0000)]
add portalstag in b1_4_next_recovery_transno and b1_4_lov_lvb_cleanup
zhanghc [Thu, 12 Jan 2006 06:10:25 +0000 (06:10 +0000)]
update the LNET tag of b1_4_next_recovery_transno and b1_4_lov_lvb_cleanup
liangzhen [Wed, 11 Jan 2006 06:22:54 +0000 (06:22 +0000)]
1. Eliminate -m64 option while building kextensions in OS X (Kernel
of OS X, both Panther and Tiger, are 32bits. Although Tiger supports
64bits applications).
2. A little change to Darwin building cflags.
zhanghc [Tue, 10 Jan 2006 10:08:03 +0000 (10:08 +0000)]
add one tag in buildcvs in order to checkout right libsysio
zhanghc [Fri, 6 Jan 2006 04:01:10 +0000 (04:01 +0000)]
in order to test branch b1_4_lov_lvb_cleanup, change libsysio tag as
b_release_1_4_6 from HEAD
FILE: buildcvs
adilger [Thu, 5 Jan 2006 18:44:32 +0000 (18:44 +0000)]
Branch b1_4
Add special case for regression test.
b=10023
r=brian
phil [Tue, 3 Jan 2006 23:08:09 +0000 (23:08 +0000)]
grammatical and line-break fixes to the OS X installer text
nic [Thu, 22 Dec 2005 16:58:50 +0000 (16:58 +0000)]
add vmstat running in parallel
fix how remote_shells for running tests are handled -- we have a long running child process now, so we have to wait on specific pids
nic [Fri, 16 Dec 2005 02:58:48 +0000 (02:58 +0000)]
- allow most everything to be overridden on the command line
- still needs help inthe ssh/pdsh decision
brian [Mon, 12 Dec 2005 18:45:39 +0000 (18:45 +0000)]
When landing, strip the trailing segments off of the dirname (given to
land1.sh) to ensure we use the ChangeLog file at the top of the module.
This is needed when trying to land only a subdirectory of a module.
adilger [Mon, 12 Dec 2005 09:26:23 +0000 (09:26 +0000)]
Branch b1_4
Make this more obvious.
brian [Fri, 9 Dec 2005 19:34:45 +0000 (19:34 +0000)]
Add ".#*" to the list of files removed from the make dist target dir.
Optimize the multiple find ... | xargs into a single find.
brian [Fri, 9 Dec 2005 19:13:24 +0000 (19:13 +0000)]
b=9729
r=adilger
Some implicit options when --enable-bgl is specified.
green [Tue, 6 Dec 2005 21:16:36 +0000 (21:16 +0000)]
reverting test commit
green [Tue, 6 Dec 2005 21:16:16 +0000 (21:16 +0000)]
test commit
brian [Sat, 3 Dec 2005 00:56:29 +0000 (00:56 +0000)]
b=9729
r=+adilger
Add an --enable-bgl flag for future BGL specific tunings and disable the -m64
compiler flag checking if set.
brian [Fri, 25 Nov 2005 20:56:36 +0000 (20:56 +0000)]
Protect the * from shell expansion producing:
brian [Fri, 25 Nov 2005 16:50:48 +0000 (16:50 +0000)]
Remove the debugging flag to greatly reduce output.
Add an alternate expression for 2.4's make oldconfig.
yury [Fri, 18 Nov 2005 11:17:40 +0000 (11:17 +0000)]
- added missed swab of body->mode
green [Wed, 16 Nov 2005 12:40:20 +0000 (12:40 +0000)]
file lustre_ver.h.in was initially added on branch b1_4.
green [Wed, 16 Nov 2005 12:39:39 +0000 (12:39 +0000)]
b=7417
r=adilger
Introduce version exchange between servers and clients and ability to issue
warnings on old clients and refusal to accept old liblustre clients.
wangdi [Wed, 16 Nov 2005 07:17:39 +0000 (07:17 +0000)]
Branch b1_4_join
update from b_release_1_4_6.
gord [Wed, 9 Nov 2005 20:47:25 +0000 (20:47 +0000)]
Add kabi script to CVS.
brian [Mon, 7 Nov 2005 21:50:32 +0000 (21:50 +0000)]
b1_4_rhel4_22 has been re-parented to b_release_1_4_6 so it should use the
b_release_1_4_6 lnet/portals.
s/^I/ /g
brian [Wed, 2 Nov 2005 21:21:37 +0000 (21:21 +0000)]
Fix timed_run() so that it returns success when it's child does not time out.
Bomb out of lmake if the update_oldconfig times out.
brian [Wed, 2 Nov 2005 21:20:08 +0000 (21:20 +0000)]
I thought this would make expect more efficient, reducing the size of the
buffer that it had to match in but perhaps this is causing the mysterious
hangs.
brian [Wed, 2 Nov 2005 17:53:51 +0000 (17:53 +0000)]
Make sure update_oldconfig does not hang the whole works.
brian [Wed, 2 Nov 2005 14:21:51 +0000 (14:21 +0000)]
Expect debugging to see what is hanging it up.
brian [Tue, 1 Nov 2005 22:17:05 +0000 (22:17 +0000)]
r=nic
Build in the RPM _tmpdir.
Get the .config from the one just used -- it might contain differences from
the one supplied by Lustre -- for new options.
brian [Tue, 1 Nov 2005 14:28:58 +0000 (14:28 +0000)]
Don't log the output of the make oldconfig. The expect script displays what
new values needed setting and what they were set to.
brian [Mon, 31 Oct 2005 16:04:06 +0000 (16:04 +0000)]
Need to include update_oldconfig in the dist tarball.
brian [Mon, 31 Oct 2005 04:20:07 +0000 (04:20 +0000)]
An expect script to interact with make oldconfig answering questions for new
config values conservatively (m for moduleable code, n for all else).
When a new item is added by make oldconfig, QA should get notified and the
kernel-meister is responsible for getting the kernel_configs updated with
appropriate answers for the new values.
brian [Fri, 28 Oct 2005 19:36:48 +0000 (19:36 +0000)]
r=nic
Look in the right place for *config targets for kernel make.
brian [Fri, 28 Oct 2005 17:49:49 +0000 (17:49 +0000)]
r=nic
Anchor the search for the right target to rebuild the .config file to avoid
using the wrong one just because it's mentioned (as a non-target) in the
Makefile.
If make $oldconfig produces a .config that is different than we have for that
kernel in the Lustre tree, tell the QA folks so they can update the tree to
have the same kernel_config.
nic [Fri, 28 Oct 2005 03:52:22 +0000 (03:52 +0000)]
b_release braches are tagged for build/ too
nathan [Thu, 27 Oct 2005 23:20:19 +0000 (23:20 +0000)]
b=9501
r=adilger
automatically create /dev/lnet, /dev/obd when needed.
brian [Wed, 26 Oct 2005 23:15:01 +0000 (23:15 +0000)]
r=nic
Allow lbuild to build a kernel from a downloaded RHEL4 SRPM. It should only
need to do this once as it caches the .tar.bz2 it builds.
This technique can/should be ported to other vendor kernels we deal with.
Just more "hands off" automation of the build process.
nathan [Tue, 25 Oct 2005 15:20:15 +0000 (15:20 +0000)]
b=8080
create /dev/lnet
brian [Fri, 21 Oct 2005 22:33:34 +0000 (22:33 +0000)]
Array element access in bash MUST be enclosed with {}.
lsy [Fri, 21 Oct 2005 03:28:06 +0000 (03:28 +0000)]
remove unused LC_CONFIG_QUOTA in lustre-build.m4.
remove check for if_dqblk/dqinfo.
change parameters type from int to __u32, since it's used
in ioctl. and make it align. (according to Andreas' advice)
nathan [Thu, 20 Oct 2005 19:45:29 +0000 (19:45 +0000)]
add configure warning if building without CONFIG_KMOD
niu [Thu, 20 Oct 2005 09:45:15 +0000 (09:45 +0000)]
all children of b1_4 should use same lnet
nathan [Wed, 19 Oct 2005 21:49:51 +0000 (21:49 +0000)]
b=8080
better errors for 2.4 module autoloading
ericm [Tue, 18 Oct 2005 21:37:30 +0000 (21:37 +0000)]
b1_4_acl has updated from b1_4, use same portals/lnet as b1_4
nathan [Tue, 18 Oct 2005 18:23:08 +0000 (18:23 +0000)]
b=8080
- modules aren't as automatic as they should be for linux 2.4
- for some reason python in linux 2.4 didn't understand this lmc construct
nathan [Tue, 18 Oct 2005 16:12:39 +0000 (16:12 +0000)]
b=8080
Landing LNET (b1_4_newconfig)