LU-8182 mdt: check connection data properly

For mdt_obd_connect() case, the parameter @data should never be
NULL, so drop the redundant check of "data != NULL", replace it
with "LASSERT(data != NULL);"

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: Ia6754752527e69c89e7a40d3d012232eaba94d3e
Reviewed-on: http://review.whamcloud.com/20391
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>

LU-8178 lproc: fix negative recovery_duration

There is a window in target_recovery_thread where
obd_recovery_end is not set despite of
obd_recovering becomes 0. In such case
recovery_duration shows difference between 0 and
obd_recovery_start like below:
recovery_duration: 18446744072323902150.

Change-Id: Id385156585c99d12726e89980b55c588e0bf3ceb
Signed-off-by: Sergey Cheremencev <sergey.cheremencev@seagate.com>
Seagate-bug-id: MRP-1519
Reviewed-by: Alexander Nikolaevich Boyko <alexander.boyko@seagate.com>
Reviewed-by: Artem Blagodarenko <artem.blagodarenko@seagate.com>
Tested-by: Parinay Vijayprakash Kondekar <parinay.kondekar@seagate.com>
Reviewed-on: http://review.whamcloud.com/20381
Tested-by: Jenkins
Reviewed-by: Alexander Boyko <alexander.boyko@seagate.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7334 lprocfs: Allow a new line before null terminator

Kernel string parsing functions support a single new line
before the null terminator of a string. Our numeric string
parsing should do the same. This allows strings to be
echo'ed directly into proc files.

Signed-off-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Change-Id: Ia9d36b4f4f6849547391115d31be34a14215fc2a
Reviewed-on: http://review.whamcloud.com/20353
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Reviewed-by: Frank Zago <fzago@cray.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-8171 hsm: stack overrun in hai_dump_data_field

With the stock 3.10 kernel, hai_dump_data_field will do a stack buffer
overrun when cat'ing /proc/fs/lustre/.../hsm/actions if an action has
some data in it.

hai_dump_data_field uses snprintf. But there is no check for
truncation, and the value returned by snprintf is used as-is.  The
coordinator code calls hai_dump_data_field with 12 bytes in the
buffer. The 6th byte of data is printed incompletely to make room for
the terminating NUL. However snprintf still returns 2, so when
hai_dump_data_field writes the final NUL, it does it outside the
reserved buffer, in the 13th byte of the buffer. This stack buffer
overrun hangs my VM.

This doesn't happen on CentOS 6.

Fix by checking that there is enough room for the next 2 characters
plus the NUL terminator. Don't print half bytes. Change the format to
02X instead of .2X, which makes more sense.

Signed-off-by: frank zago <fzago@cray.com>
Change-Id: Icb6e658fc63f441b7a698f595ac71a1236ff3588
Reviewed-on: http://review.whamcloud.com/20338
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Jean-Baptiste Riaux <riaux.jb@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8163 hsm: user_request_mask can't be unset in test 24c

The following lines in sanity-hsm.sh test_24c

  set_hsm_param user_request_mask ""
  set_hsm_param group_request_mask ""
  set_hsm_param other_request_mask ""

generate the following errors in the test, which isn't caught:

  error: set_param: setting mdt.lustre-MDT0000.hsm.user_request_mask:
                    no value

This is because "lct set_param" cannot accept an empty value:

  $ lctl set_param -n mdt.lustre-MDT0000.hsm.user_request_mask=""
  error: set_param: setting mdt.lustre-MDT0000.hsm.user_request_mask:
                    no value

Since the test checked earlier that [user|group|other]_request_mask
was RESTORE, we don't actually need to change their values to
nothing. They just have to be not ARCHIVE.

Test-Parameters: trivial testlist=sanity-hsm,sanity-hsm

Signed-off-by: frank zago <fzago@cray.com>
Change-Id: I00622d93157e6577651996bf482ec8f710f86c9e
Reviewed-on: http://review.whamcloud.com/20320
Reviewed-by: Henri Doreau <henri.doreau@cea.fr>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8138 utils: misc fixes in wirecheck

Include required headers that were missing.
Fix structure field name that has been embedded into an union.

Signed-off-by: Henri Doreau <henri.doreau@cea.fr>
Change-Id: If87d673ffbe9ccdd965c77161dfc69ed9e983fc3
Reviewed-on: http://review.whamcloud.com/20189
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>

LU-8056 libcfs: Support for linux 4.2 kernels

In kernel version 4.2 commit df6b35f409af0a8ff1ef62f552b8402f3fef8665
header file i387.h was renamed to fpu/api.h

Change-Id: Id4f5f6e73f3c2aeac67b5c87b9c1d310a0a50a50
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-on: http://review.whamcloud.com/20009
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8102 ldlm: Correlate ptlrpc req w/AST error

Enhance AST error and debug messages with the pointer and xid of the
corresponding ptlrpc request.

Signed-off-by: Chris Horn <hornc@cray.com>
Change-Id: I47834d092df0ac8638b03aa6447b3bd772ee5e0b
Reviewed-on: http://review.whamcloud.com/19983
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-5560 obd: reserve connection flag OBD_CONNECT2_FILE_SECCTX

The connection flag OBD_CONNECT2_FILE_SECCTX will be set (in
ocd_connect_flags2) if an MDT supports setting the file security
context at create time.

This patch is ported from the following one:
Lustre-commit: f0f79ede2105dd40665bd8b60c2c5f274290ab57
Lustre-change: http://review.whamcloud.com/18420

Signed-off-by: John L. Hammond <john.hammond@intel.com>
Signed-off-by: Sebastien Buisson <sbuisson@ddn.com>
Intel-bug-id: LDEV-236
Change-Id: Ib79ef50bbcd68baa624ece109fa204dfd7f8b128
Reviewed-on: http://review.whamcloud.com/19970
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-8036 utils: remove ARRAY_SIZE() define from lustreapi.h

Remove the definition of ARRAY_SIZE() from lustreapi.h.

Signed-off-by: John L. Hammond <john.hammond@intel.com>
Change-Id: I57b1893e7ecd57d841f5b9348a8d5285cbc02124
Reviewed-on: http://review.whamcloud.com/19818
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>

LU-7920 hsm: Account for decreasing max request count

If HSM is active and the max request count is decreased, account for
that when checking to see if more requests should be started.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: I4dcf9ea192f9b354a9933e08d2ced9ced7b3bc7c
Reviewed-on: http://review.whamcloud.com/19382
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Frank Zago <fzago@cray.com>
Reviewed-by: Aurelien Degremont <aurelien.degremont@cea.fr>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7993 test: Fix division by 0 in info prints

Some of the information only prints in sanity/116a don't account for a
0 in the minimum number and thus divide by zero, but only print if MIN
is above zero, so this just doesn't do the calculation if it isn't
needed.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: I7003739f6b518f981a1058d41b19bca258e1d11e
Reviewed-on: http://review.whamcloud.com/19359
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>

LU-7776 tests: lnet-selftest.sh local_mode failure

Update lnet-selftest to call stopall for local_mode
and cleanupall otherwise.

Signed-off-by: Abrarahmed Momin <abrar.habib@seagate.com>
Change-Id: I5dff3df8b45afc4a582556f7dd6bc92409efd380
Reviewed-on: http://review.whamcloud.com/19308
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>

LU-7955 gnilnd: Add ability to set bte_get/put_dlvr_mode

Currently the bte_dlvr_mode applies to all BTE transactions
initiated on the node. Based on testing we would like to be able to
set the mode for BTE_GET and BTE_PUT independently.

Changes
1. Add bte_get_dlvr_mode
2. Add bte_put_dlvr_mode
3. Remove old bte_dlvr_mode parameter
4. Set default for both new modes to be performance.

Signed-off-by: Chris Horn <hornc@cray.com>
Change-Id: Ifa7388b2a5b3d2f17b96f68890d3d98f082e3de6
Reviewed-on: http://review.whamcloud.com/19202
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Chuck Fossen <chuckf@cray.com>
Reviewed-by: James Shimek <jshimek@cray.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7259 osd-ldiskfs: fix signed one bit field

Bit field 'ot_remove_agents' has one bit and is signed.

Change-Id: Ia844a41552f8ca496ca39774296ecfa32b5966b3
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-on: http://review.whamcloud.com/19198
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Frank Zago <fzago@cray.com>

LU-7258 osc: fix signed one bit field

Bit field 'oi_lockless' and 'oi_is_active' has one bit and is signed.

Change-Id: Ib2fb92faf29ced057438a090dad382258eb9fc40
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-on: http://review.whamcloud.com/19196
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Frank Zago <fzago@cray.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>

LU-7901 osd: put sa_bulk_attr_t in osd_thread_info

and don't allocate it for each attr_set/attr_init

Change-Id: I02e00ae72702728e97056ecd63862ff217134231
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/19087
Tested-by: Jenkins
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7908 osp: fake precreate support

to be able to benchmark file creations locally like we
have very fast OSTs.

Change-Id: Ic5bb0f1c9306fee018da74a38f4ce702bf28d293
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/19083
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>

LU-7890 lov: Ensure correct operation for large object sizes

If a backing filesystem (ZFS) returns that it supports very large
(LLONG_MAX) object sizes, that should be correctly supported.  This
fixes the check for unitialized stripe_maxbytes in
lsm_unpackmd_common(), so that ZFS can return LLONG_MAX and it will be
okay.  This issue is excersized by writing to or past the 2TB boundry
of a singly stripped file.  A regression test is added to check for
this.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: I669096281bd826d69b5e86ada50077740412d268
Reviewed-on: http://review.whamcloud.com/19066
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7830 ost: do not evict during truncate

OST punch request can last for a long time, so
it is needed to prolong taken lock
until it completes.

Seagate-bug-id: MRP-3332
Change-Id: Icad7d122bc559f9e0bd693c7f88cbac14235bd38
Signed-off-by: Andriy Skulysh <andriy.skulysh@seagate.com>
Reviewed-on: http://review.whamcloud.com/18739
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Chris Horn <hornc@cray.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7805 llite: Remove unused members of struct ll_sb_info

ll_orphan_dentry_list and ll_conn_chain are not used
and could be removed.

Change-Id: Ibc306b501acb9d541582af82558f6bcfc4582638
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: http://review.whamcloud.com/18583
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>

LU-7302 scrub: join the running OI scrub properly

Sometimes, the OI scrub can be triggered as partial mode by
randomly found inconsistent OI mapping(s). Under such case,
if the sysadmin triggers LFSCK/OI scrub (full mode) manually,
the partial mode OI scrub needs to be converted as full mode
via osd_scrub_join(). Before that, it will check whether the
current OI scrub is full mode or not. Originally, we checked
the SF_AUTO flags (scrub::os_file.sf_flags) only, but ignored
the partial mode (scrub::os_partial_scan). The partial mode
OI scrub will NOT change the scrub iteration cursor (srub::
os_pos_current). So if miss to convert the partial OI scrub,
the subsequent inode table based iteration will be blocked
inside osd_otable_it_next() as following:

if (it->ooi_cache.ooc_pos_preload >= scrub->os_pos_current)
        l_wait_event(thread->t_ctl_waitq,
                     osd_otable_it_wakeup(scrub, it),
                     &lwi);

This patch fix such issue by checking the conditions properly.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I41cf519e717f356f21a1eebebefda1e44a3bd5ae
Reviewed-on: http://review.whamcloud.com/18175
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7486 obdclass: health_check to report unhealthy upon LBUG

When a LBUG has occured, without panic_on_lbug being set, health_check
/proc file must return an unhealthy state.

Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Change-Id: I3fb89fe825cb3e85f2be99609141131aef6b6af7
Reviewed-on: http://review.whamcloud.com/17981
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7626 ldlm: refcount nonzero (1) after lock cleanup

It is a regression after LU-2177.
LCK_NL is used by obdecho also.

Perform flock sanity checks only for LDLM_FLOCK.

Change-Id: I83f87b41a13c4d666af36af389ccf1a127197294
Xyratex-bug-id: MRP-2336
Signed-off-by: Andriy Skulysh <andriy.skulysh@seagate.com>
Reviewed-on: http://es-gerrit.xyus.xyratex.com:8080/9406
Tested-by: Jenkins
Reviewed-by: Mirza Arshad Mirza Hussain <arshad.hussain@seagate.com>
Reviewed-by: Alexey Leonidovich Lyashkov <alexey.lyashkov@seagate.com>
Tested-by: Parinay Vijayprakash Kondekar <parinay.kondekar@seagate.com>
Reviewed-on: http://review.whamcloud.com/17791
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-5954 IOC_MDC_GETFILEINFO returns the wrong ino

req_capsule_server_get() through  __req_capsule_get in ll_dir_ioctl()
returns a pointer to a PTLRPC request or reply buffer, which is assigned
to struct mdt_body.

If the command is IOC_MDS_GETFILEINFO then the inode "st.st_ino" should
be assigned from one extracted from mdt_body through cl_fid_build_ino().

Change-Id: I8429d726ff5f6c2c2d342e2c59f163608c794307
Signed-off-by: John Hammond <john.hammond@intel.com>
Signed-off-by: akam kumar bharathi <azurelustre@gmail.com>
Reviewed-on: http://review.whamcloud.com/17618
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-5092 nodemap: handle config changes while mid-flight

This adds the ability to detect when the nodemap configuration has
changed between messages for very large configurations. If the
configuration is over 1MB (about 25k entries), it needs to be sent
in multiple messages. If the configuration changes in-between these
messages, then the client needs to abort the previous get config
operation and restart.

Test-Parameters: envdefinitions=SLOW=yes testlist=sanity-sec
Signed-off-by: Kit Westneat <kit.westneat@gmail.com>
Change-Id: I2bf6e4bade947fea8331c2922d9c33ab49100577
Reviewed-on: http://review.whamcloud.com/16941
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6523 lmv: Error not handled for lmv_find_target

This issue is found by smatch; has been reported as-
Unchecked usage of potential ERR_PTR result in lmv_hsm_req_count
and lmv_hsm_req_build.
Added ERR_PTR in both functions and also return value check added.

Signed-off-by: Ulka Vaze <ulka.vaze@yahoo.in>
Change-Id: I9651bb70724a1b2f2a1901a3bf5093f84b22e6ea
Signed-off-by: Aditya Pandit <panditadityashreesh@yahoo.com>
Reviewed-on: http://review.whamcloud.com/14918
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8157 tests: functional testing for swap layout

Add a few tests for the swap layout operations.

Test-Parameters: trivial

Signed-off-by: frank zago <fzago@cray.com>
Signed-off-by: Henri Doreau <henri.doreau@cea.fr>
Change-Id: I30210c634a83c01c8c15de06b23581beee6a8eb9
Reviewed-on: http://review.whamcloud.com/13441
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3434 misc: style check whole patch when rebasing

When rebasing a patch, style check the whole patch instead of just
the part that has changed.  Otherwise, if the patch was previously
developed on a system without the prepare-commit-msg hook it will
not be properly checked before submission, possibly incurring the
ire of HPDD Checkpatch.

Test-Parameters: trivial

Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I603577450d3524ab7bb3d4a4d0bdf8ce463ebbe5
Reviewed-on: http://review.whamcloud.com/8644
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-1595 build: improve prepare-commit-msg git hook

Improve the Git prepare-commit-msg hook to still add the default
comment for Signed-off-by: and the Vim modeline, even if checkpatch.pl
isn't available.

Add a few more comments to the script.

Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I253931c34ee64afa2d9cf160e675c779b32e9359
Reviewed-on: http://review.whamcloud.com/13427
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8241 ofed: fix compilation with MOFED 3.3

Add OFED compat rdma header file during autoconf checks also.

Test-Parameters: trivial

Change-Id: Ib280311fa1565cda3c7322308f33575159218e42
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-on: http://review.whamcloud.com/20644
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7428 osd: set device read-only correctly

Freeze the journal before setting device read-only to prevent
starting a new handle or commit a transaction in that time.

Seagate-bug-id: MRP-2135

Some of the patch is from http://review.whamcloud.com/20535

Change-Id: Iadcb0a7ccf9189e8bb564aa4783bae9a2ef86a76
Signed-off-by: Alexey Lyashkov <alexey.lyashkov@seagate.com>
Signed-off-by: vinayakswami hariharmath <vinayakswami.hariharmath@seagate.com>
Signed-off-by: Gu Zheng <gzheng@ddn.com>
Signed-off-by: Hongchao Zhang <hongchao.zhang@intel.com>
Reviewed-on: http://review.whamcloud.com/20586
Reviewed-by: Wang Shilong <wshilong@ddn.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-8216 ldiskfs: fix journal quota files

When we hit Lustre crash, e2fsck will always
complains Quota accounting mismatch, and sometimes
differences are huge.

Problems are we don't journal quota updates properly,
and with current codes quota files will be only
journaled when syncing filesystem,
that only happen by sync call or unmounting.

Signed-off-by: Wang Shilong <wshilong@ddn.com>
Change-Id: I94c148a6b48f2464e96320bbadc88a3c234eacb0
Reviewed-on: http://review.whamcloud.com/20503
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>

LU-8179 libcfs: Do not include libcfs/util/ioctl.h in kernel code

lustre/include/lustre_ioctl.h included libcfs/include/libcfs/util/ioctl.h
by mistake during kernel compiles, but it's strictly a user-space header
so move it into appropriate ifdef.

Test-Parameters: trivial
Change-Id: I598ef54ad2cf8bab9788eefe487adddeb73f90c1
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: http://review.whamcloud.com/20375
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Frank Zago <fzago@cray.com>

LU-8019 llite: Restore proper opencache operations

Just returning 0 from revalidate for all opens broke open cache
because when we enter ll_lookup_nd, we have no inode and need to
do an RPC to get there which defeats the purpose.

Instead allow returning 1 for non-creating opens,
mark all dentries passing via ll_iget_for_nfs as such
so that ll_file_open would know where did we
come from and request the open lock with the handle.

Change-Id: Id8a0ade892bc26cee96a9ba2b1498cc3199255f0
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: http://review.whamcloud.com/20354
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Li Xi <lixi@ddn.com>

LU-8146 osd-zfs: Build against SPL/ZFS 0.6.5.7

Build against SPL/ZFS 0.6.5.7.

ZFS Changelogs since 0.6.5.6:
https://github.com/zfsonlinux/zfs/releases/tag/zfs-0.6.5.7

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: Icdde350e4f7fee6e72f2eab2548ef07b44f0dacb
Reviewed-on: http://review.whamcloud.com/20219
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>

LU-6245 libcfs: remove types abstraction from libcfs/LNet code

Originally when lustre code was built for userland we needed
a proper way to handle 32 bit and 64 bit platforms when
reporting unsigned longs. Now that this code is only built
for kernel space and the kernel has it own special string
handling functions we don't need this abstraction anymore.
The libcfs types.h is not removed yet so not to break
any other patches ready to land. Had checkpatch.pl updated
to report to developers to not use the LP* abstraction.
This patch only updates the LNet/libcfs layers.

Change-Id: I0a338b4eb181d86b9c7d0f535d8c12c9f7045fb3
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: http://review.whamcloud.com/20204
Reviewed-by: Frank Zago <fzago@cray.com>
Tested-by: Jenkins
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8044 mgs: Only add OSP for registered MDT

Only add OSP for those really registered MDT,
by checking if its llog is empty, otherwise
it will incorrectly add "setup OSP" record in
the begining of config log, which will cause
setup failure.

Signed-off-by: Di Wang <di.wang@intel.com>
Change-Id: I9df26a5e6653b4ac0413dad01ea2abfda6f1c1fb
Reviewed-on: http://review.whamcloud.com/19658
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7997 obd: RCU stalls in lu_cache_shrink_count()

The algorithm for counting freeable objects in the lu_cache shrinker
does not scale with the number of cpus. The LU_SS_LRU_LEN counter
for each cpu is read and summed at shrink time while holding the
lu_sites_guard mutex. With a large number of cpus and low memory
conditions, processes bottleneck on the mutex.

This mod reduces the time spent counting by using the kernel's percpu
counter functions to maintain the length of a site's lru. The summing
occurs when a percpu value is incremented or decremented and a
threshhold is exceeded. lu_cache_shrink_count() simply returns the
last such computed sum.

This mod also replaces the lu_sites_guard mutex with a rw semaphore.
The lock protects the lu_site list, which is modified when a file
system is mounted/umounted or when the lu_site is purged.
lu_cache_shrink_count simply reads data so it does not need to wait
for other readers. lu_cache_shrink_scan, which actually frees the
unused objects, is still serialized.

Signed-off-by: Ann Koehler <amk@cray.com>
Change-Id: I62d006d8047db99d8bbe3ee4edd9d4489bd3739f
Reviewed-on: http://review.whamcloud.com/19390
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-5969 lustreapi: replace llapi_get_version()

The existing llapi_get_version() helper has an awkward calling
style, requiring a temporary input buffer, as well as an output
pointer.

Replace this with llapi_get_version_string() which returns the
the simple version string.  Mark llapi_get_version() deprecated.
Update "lctl lustre_build_version" to use llapi_version_string().

These versions are extracted from {/proc,/sys}/fs/lustre/version
instead of an ioctl, to avoid the messiness that is Lustre ioctls.

Update the "version" file to only print the "lustre:" version,
since the "build:" version is now identical, and the "kernel:"
version has been "patchless_client" for many years already.

Print out the build version at the start of each test so it is
logged in the test results for convenience.

Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I681a353e6f65fd1ec2d1434fe38dfa55ed3ebbe5
Reviewed-on: http://review.whamcloud.com/16721
Tested-by: Jenkins
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8006 ptlrpc: cleanup codes of TBF command

The codes of parsing TBF command are improved so as to make it
easier to add new commands. And command format has been changed
to:
For NRS "tbf nid" policy:
lctl set_param ost.OSS.ost_io.nrs_tbf_rule=
"start $RULE_NAME nid={$NID} rate=$RATE"
For NRS "tbf jobid" policy:
lctl set_param ost.OSS.ost_io.nrs_tbf_rule=
"start $RULE_NAME jobid={$JOBID} rate=$RATE"

lctl set_param ost.OSS.ost_io.nrs_tbf_rule=
"change $RULE_NAME rate=$RATE"

Signed-off-by: Li Xi <lixi@ddn.com>
Signed-off-by: Qian Yingjin <qian@ddn.com>
Change-Id: Id9861f4c3d894bea614288089622c11124e7f4de
Reviewed-on: http://review.whamcloud.com/19499
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Wang Shilong <wshilong@ddn.com>
Reviewed-by: Gu Zheng <gzheng@ddn.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6861 scrub: only trigger full OI scrub when necessary

Generally, scanning the whole device for OI scrub routine check may
takes some long time. If the whole system only contains several bad
OI mappings, then it is not worth to trigger OI scrub automatically
with full speed when some bad OI mapping is auto-detected. Instead,
we can make the OI scrub to fix the found bad OI mappings only, and
if more and more bad OI mappings are found as to exceeds some given
threshold that can be adjusted via some proc interface, then the OI
scrub will run with full speed to scan whole device.

Our current OI scrub auto-trigger control mechanism can avoid most
of unnecessary whole-device based scanning. But it missed one case
that the FID lookup for remote MDT-object and may cause unexpected
whole-device based OI scrub. This patch is for handle such corner.

Fix the sanity-scrub.sh test_4a, test_4b, and test_4c to search for
"prior_updated" and not "updated_prior" in the scrub status file.

Signed-off-by: James Nunez <james.a.nunez@intel.com>
Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I776c44b1dbc7357817d42d9b1ff6011226fe292b
Reviewed-on: http://review.whamcloud.com/16951
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8057 ko2iblnd: Replace sg++ with sg = sg_next(sg)

With scatterlist chaining, simply incrementing the array does not
work. sg_next macro was thus introduced to follow the chain links
when necessary. So replace sg++ with sg_next.

This change was made with the help of the following Coccinelle
semantic patch:
//<smpl>
@@
 struct scatterlist *sg;
@@
 -sg++
 +sg = sg_next(sg)
//</smpl>

In my test of the upstream client this change exposed a long
standing issues where we have a offset that is not page aligned
would cause us to access memory beyond the scatter gather list
which was causing memory corruption when all 256 fragments were
in use.

Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Change-Id: I5bc140645b841aa2a92da434ffa65795112c6cec
Signed-off-by: Amitoj Kaur Chawla <amitoj1606@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-on: http://review.whamcloud.com/19342
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7418 tests: cancel all locks in sanity test_29

Cancel both MDC and OSC locks in sanity.sh test_29, since any
background writeout of dirty pages under OSC locks will request
a layout lock from the MDS, causing the test to fail.

Test-Parameters: trivial testlist=sanity,sanity,sanity,sanity
Test-Parameters: testlist=sanity,sanity,sanity,sanity
Test-Parameters: testlist=sanity,sanity,sanity,sanity
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: Ie2cc1b241d5fac5b439509a3d01384f6a531e654
Reviewed-on: http://review.whamcloud.com/20741
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 gss: Add two additional security flavors for sk

This patch adds shared key null (skn) and shared key auth (ska)
flavors to make shared key consistent with the kerberos
implementation.  Shared key null requires a key to establish the
security context but does not use integrity or privacy outside of
the SEC_CTX_INIT RPC.  Shared key auth enables integrity for normal
service but not bulk.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I55fa52dfe1089f3dc9a40ffad28997a0b08aadec
Reviewed-on: http://review.whamcloud.com/18773
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 gss: Cleanup gss print statements and comments

Some minor cleanup of GSS messages and comments.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I9cf4e7d7189142b47c2fb9ab13428cb65bc08754
Reviewed-on: http://review.whamcloud.com/17602
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 gss: Add option for loading keys during mount

This adds functionality for loading GSS shared keys using
mount.lustre (mount) command.  The --skpath option has been
added which supports a directory or file of keys to load.
The context from which the mount occurs is used to determine
which type of key to load (MGS, MDT/OST, or client).

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I4ffbfb9ea76d32ed69f94da93f82e90e161cb99d
Reviewed-on: http://review.whamcloud.com/17601
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-3289 gss: Add userspace support for GSS null and sk

Add the support to lgss_keyring (client) and lsvcgssd (server) for GSS
null and shared key.  This includes the common functionality for the
creation of the session key in privacy mode and shared key for
integrity and privacy mode.

A new utility lgss_sk has been added for key management.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: Id2574b940d344d18f793113f61e69a393b6df4f0
Reviewed-on: http://review.whamcloud.com/17600
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-3289 gss: Return GSS major and minor status

Return the major and minor status codes back from the server side
userspace component (lsvcgssd) instead of defaulting to
GSS_S_FAILURE for the major and 0 for the minor.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I14d1a4b4ab6fb982e3c6c9c062d585865aa77ca0
Reviewed-on: http://review.whamcloud.com/17599
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-3289 osp: osp_precreate_thread does not check for errors

If servers do not have a valid key for GSS the call to
osp_statfs_update() can fail.  This patch breaks the loop in the
event of a failure from osp_statfs_update().

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I576cff86d285c4263e8d98cccb9b3c6943af6f75
Reviewed-on: http://review.whamcloud.com/17598
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 gss: Add Shared key and GSS Null functionality

Update the GSS null and SK mechanisms framework with the actual
implementation for wrapping and signing of RPCs.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I050944ccc2413d398442351d8615c08ef48e3500
Reviewed-on: http://review.whamcloud.com/17597
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 gss: Interface and code changes for shared key

This patch moves and modifies some of the kerberos code to
be more general so it can be utilized by the sk and null
gss implementations.  This includes the userspace portion
for lsvcgssd and the kernel pipefs interface.  Since the gss
interface has changed the version has also been incremented.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: Ib46b74c9b0e01c4f71f16d288eae83c5c1567bfb
Reviewed-on: http://review.whamcloud.com/16728
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7372 tests: disable replay-dual test_26

This test is continually failing, preventing replay-dual from being
re-enabled during normal review testing.  Disable it for now so that
we can return replay-dual to normal review test coverage.

Test-Parameters: trivial
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: Ib7129ccbf48197a31bedd8f37ea63fd9038cab07
Reviewed-on: http://review.whamcloud.com/20669
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Jian Yu <jian.yu@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>

LU-4257 test: Correct error_ignore message

The error_ignore routine requires a bug number to indicate why an
error should be ignored. In sanity test 248, there is no ticket
indicating why the test results should be ignored, but, due to running
the test in a VM, the performance numbers may not be what is
expected and, thus, the results can be ignored.

Add a reason to ignore the error to the call to error_ignore.

Test-Parameters: trivial testlist=sanity,sanity,sanity,sanity

Signed-off-by: James Nunez <james.a.nunez@intel.com>
Change-Id: Ieda9d06facd3e1519ac6a84e15505bc4c74aa60a
Reviewed-on: http://review.whamcloud.com/20647
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>

New tag 2.8.54

Change-Id: Idb98285ba26f15117a1b569064e3491826b194f5
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7352 tests: conf-sanity 78 don't check missing files

Commit b666d02e13e stopped conf-sanity test_78 failing if all of
the required files could not be created.  However, the test also
checks later whether those files are still available after the
filesystem resize.  Reduce $num_files to match the number of
files actually created.

Test-Parameters: trivial envdefinitions=ONLY=78 testlist=conf-sanity,conf-sanity,conf-sanity,conf-sanity
Test-Parameters: trivial envdefinitions=ONLY=78 testlist=conf-sanity,conf-sanity,conf-sanity,conf-sanity
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I0ccfbd3f6597167c937e75cf9b5d0892cb5cab07
Reviewed-on: http://review.whamcloud.com/20558
Reviewed-by: Wang Shilong <wshilong@ddn.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4257 llite: fast read implementation

For read operation, if a page is already in cache, it must be covered
by a DLM lock. We can take advantage of this by reading cached page
without interacting with Lustre. Traditional read will go on if fast
read fails.

This patch can improve small read performance significantly.
These are the performance data I collected:

+------------+----------------+-----------------+
|            | read bs=4k     | read bs=1M      |
+------------+----------------+-----------------+
| w/o patch  | 257 MB/s       | 1.1 GB/s        |
+------------+----------------+-----------------+
| w/ patch   | 1.2 GB/s       | 1.4 GB/s        |
+------------+----------------+-----------------+

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: I2ebac609a6c80b135ee64ba001f75fa2cc80faf2
Reviewed-on: http://review.whamcloud.com/20255
Tested-by: Jenkins
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4257 llite: fix up iov_iter implementation

Clean up and fix up the usage of iov_iter in client.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: I331b48a0978dd3258946ee7bae8362e60026b037
Reviewed-on: http://review.whamcloud.com/20256
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4257 obdclass: Get rid of cl_env hash table

cl_env hash table is under heavy contention when there are lots of
processes doing IO at the same time;
reduce lock contention by replacing cl_env cache with percpu array;
remove cl_env_nested_get() and cl_env_nested_put();
remove cl_env_reenter() and cl_env_reexit();

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: I04137c93e2d48f21298fe954a1c2afe1006f44d5
Reviewed-on: http://review.whamcloud.com/20254
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8068 osd-zfs: large dnode support

In ZFS 0.7.x support for large dnode has been added.  This allows
Lustre to optionally specify the size of a dnode which ZFS will
then use to store metadata such as xattrs.

The default dnode size specified by the 'dnodesize' property on the
dataset will be used unless a specific value is provided.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ned Bass <bass6@llnl.gov>
Change-Id: Iaa7211ca50e7089af4b1fabcccdaa6e16477e641
Reviewed-on: http://review.whamcloud.com/20367
Tested-by: Jenkins
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8136 tests: allow CT to registers with all MDTs

With DNE, CT must register with all MDTs and thus need
additional time to do so.

Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Change-Id: Id2342f6d2060649d08d423703173a70b3407ee8d
Reviewed-on: http://review.whamcloud.com/20258
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Sebastien Buisson <sebastien.buisson@bull.net>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8132 tests: do not truncate fsname from paths to ost/mdt

Previously the default path for ost/mdt, while defined in cfg/local.sh
to contain the filesystem name ("/mnt/<fsname>"), was truncated
(to "/mnt"). This lead the paths of two different filesystems to
overlapse.
This patch fixes this by keeping the fsname in the path to ost/mdt
(default path now look like "/mnt/<fsname>-<component><id>").

Signed-off-by: Quentin Bouget <quentin.bouget.ocre@cea.fr>
Change-Id: Icf0baab4a9ceee42116bb12f64c0a07999b10551
Reviewed-on: http://review.whamcloud.com/20188
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Henri Doreau <henri.doreau@cea.fr>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8116 build: Cleanup GSS configure script messages

Some of the GSS configure warning are unnecessarily multi-line,
and also redundant.  For instance:

  checking for keyctl_search in -lkeyutils... no
  configure: WARNING:

  libkeyutils is not found, which is required by gss keyring backend

  configure: WARNING:

  Cannot enable gss keyring. See above for details.

Why so many lines?  Why the double warning?

Also, there is a bug in the "for Kerberos v5" check where it
fails to use AC_MSG_RESULT in the negative case.  A completely
different function winds up appending an AC_MSG_WARN:

  checking for Kerberos v5... configure: WARNING: not found!

This patch addresses those minor issues.

Change-Id: Ic9fb8b5687849688d965cc67b394e4eb569204be
Signed-off-by: Christopher J. Morrone <morrone2@llnl.gov>
Reviewed-on: http://review.whamcloud.com/20056
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7918 mdd: no need to declare attr_set at create

as dt_create() will set the attributes we need. also, we can save
couple dt_record_write() in tgt_txn_start_cb() as one is not needed
with multislot enabled while another is not needed at all.
these changes decrease disk/memory reservation from ~33MB to 6MB.

Change-Id: I65e537dd053eb7b4e4d2cccab41973096072b1db
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/19132
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-7846 mdt: mount with fileset info from nodemap

Enforce subdir mount if client is in a nodemap for
which fileset information is set.

Signed-off-by: Sebastien Buisson <sbuisson@ddn.com>
Change-Id: Ibf1077389dac66fce7c2e88ab64573619545119b
Reviewed-on: http://review.whamcloud.com/18784
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Wang Shilong <wshilong@ddn.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7846 nodemap: add fileset info to nodemap

Give the ability to set a fileset information on a nodemap.
When set, a client pertaining to this nodemap will be allowed to
mount only the subdirectory represented by this fileset.

Fileset info of nodemap is stored permanently by using
'lctl set_param -P' commands. It means MDS and OSS will be able to
retrieve fileset info of nodemap after a restart (when they read
params llog).
MGS will not see fileset info after restart, but this does not prevent
the feature from working.

Signed-off-by: Sebastien Buisson <sbuisson@ddn.com>
Signed-off-by: Kit Westneat <kit.westneat@gmail.com>
Change-Id: Ic16abaeb0b2fe583a9a4745258a82bbd23fc259e
Reviewed-on: http://review.whamcloud.com/18783
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7064 obd: detect errors from llog_declare_destroy()

The return value from llog_declare_destroy() in
llog_cancel_rec() is not used. This patch adds the same check.

Change-Id: I31c09d426e9e4ae5418531acd73a6428927e9bd3
Signed-off-by: Parinay Kondekar <parinay.kondekar@seagate.com>
Reviewed-on: http://review.whamcloud.com/17999
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4017 quota: redefine LL_MAXQUOTAS for Lustre

Lustre code should similarly be patched to use LL_MAXQUOTAS
like EXT4_MAXQUOTAS so that it is independent of the kernel
MAXQUOTAS value.

Signed-off-by: Wang Shilong <wshilong@ddn.com>
Change-Id: I95c894600e6b9fa4932b367a1ef10635a9aa222d
Reviewed-on: http://review.whamcloud.com/19843
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>

LU-6518 obd: Unhandled possible allocation failure in lustre_start_mgc

Patch to handle un-handled possible allocation failure
of struct obd_uuid.

Change-Id: Ib19d4c53a6dae5e2f6a80e0fdf755fa7aac81fd1
Signed-off-by: Parinay Kondekar <parinay.kondekar@seagate.com>
Reviewed-on: http://review.whamcloud.com/17493
Tested-by: Jenkins
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Frank Zago <fzago@cray.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8210 osd-ldiskfs: fix setting pages PageUptodate state

In our VM autotest
sanity:156 test failed in rhel7 plataform for writecache
disable and readcache enabled.

After debugging this is because of we miss to set PageUptodate
after reading bio finish, this will make read cache total
unusable for rhel7, because osd_read_prep() will think pages
always not uptodate and triger bio submitting for every reading
request.

Problems come from pages iterating, since bio_for_each_segment_all
introduced, bio_for_each_segment will no longer work right
for pages interating anymore, we should use new interface
for rhel7.

So this problems only exist for some specific Drivers:
From Documentation/block/biodoc.txt

-------------------------------
Drivers which can't process a large bio in one shot can
use the bi_idx field to keep track of the next bio_vec entry
to process.(e.g a 1MB bio_vec needs to be handled in max 128kB
chunks for IDE) [TBD: Should preferably also have a bi_voffset
and bi_vlen to avoid modifying bi_offset an len fields]
-------------------------------

This might explain why intel autotest could not reproduce it
but our clustered can always reproduce the problem.

Signed-off-by: Wang Shilong <wshilong@ddn.com>
Change-Id: I5d3de5bf5facbee8d9450c15b8ab7189d7f2fbfe
Reviewed-on: http://review.whamcloud.com/20478
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>

LU-8100 lmv: Correctly generate target_obd

The target_obd proc file was not being generated correctly
in cases where nonconsecutive MDT indices were used when
generating a filesystem.

Signed-off-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Change-Id: I4bd3aaffd56707ea29b9b2d4366de5e20b3d3760
Reviewed-on: http://review.whamcloud.com/20336
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8126 kernel: new kernel RHEL 6.8 [2.6.32-642.el6]

Changes to support new kernel version in new el6.8 release distro
New kernel-config and target files.
Revised base kernel and ldiskfs patches.
Small adjustments to lbuild and autoconf.

Test-Parameters: clientdistro=el6.8 mdsdistro=el6.8 ossdistro=el6.8

Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Change-Id: I94c0c06f43ca2973133a1dabb0184079290dbc6a
Reviewed-on: http://review.whamcloud.com/20106
Tested-by: Jenkins
Reviewed-by: Minh Diep <minh.diep@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8071 ldiskfs: handle system freeze protection

For RHEL6 2.6.32, or kernel 3.6 or newer, it supports wrapped
functions for filesystem freeze protection, such as sb_{start,
stop}_write(). Usually, they are called by VFS. Since Lustre
uses ldiskfs backend bypass VFS layer, it is the OSD's duty to
handle filesystem freeze protection by calling related wrapped
filesystem freeze protection functions. As for other kernel,
such as 3.0.101, related logic has been handled inside ldiskfs
via vfs_check_frozen().

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: Idb2b4ca02d569ec818a06b534e857b6ee44f89a2
Reviewed-on: http://review.whamcloud.com/20062
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Wang Shilong <wshilong@ddn.com>

LU-7185 ldlm: Restore connect flags on failure

Restore connect flags on failure of ptlrpc_connect_import()
to prevent an LBUG due to flags mismatch.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I9149c8d2f512c1f920b3bdfa8f3435b03dc72ab2
Reviewed-on: http://review.whamcloud.com/16950
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Reviewed-by: Sebastien Buisson <sebastien.buisson@bull.net>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-8147 osd-zfs: fix osd_mount error path

The osd_oi_fini will grab osl_seq_list_lock for cleanup.
But this lock is not initilized if osd_mount return and
failed before osd_oi_init.

Signed-off-by: Yang Sheng <yang.sheng@intel.com>
Change-Id: I8f76a87e22d131e6ab9034d8ecc42b47e301ff25
Reviewed-on: http://review.whamcloud.com/20309
Tested-by: Jenkins
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-5092 nodemap: transfer nodemaps between MGS, MDTs, and OSTs

This creates a new config lock on the MGS for transferring the
nodemaps.  The target MGCs can enqueue this lock, which is then
revoked by the MGS when the nodemaps are modified. The MGCs then
issue a read config RPC on the nodemap config lock, which causes
the MGS to transfer the nodemap index file to the MGCs. The MGCs
then process the index file, similarly to the way the MGS
processes it on start.

Signed-off-by: Kit Westneat <kit.westneat@gmail.com>
Change-Id: I99a34773298484f70a912761f5831b75196d41d8
Reviewed-on: http://review.whamcloud.com/11830
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7906 osd: no blocksize on non-OST objects

osd-zfs should not set blocksize on non-OST objects.

Change-Id: Idafe2236a49ffcb86bf06971d3687c8fceb516a2
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/19286
Tested-by: Jenkins
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-7728 osp: soft lockup in osp_precreate_reserve()

LU-4313 commit 3a2073d3bf9fab33fa26c5c5f0da2b0821185131
treats -ENOTCONN as recoverable error, so it is needed
to add it to osp_precreate_ready_condition() to wait
for an event instead of busy-looping.

Seagate-bug-id: MRP-3247
Change-Id: I8a0a77a7369be0325fbeb9d0a2c321950905afbb
Signed-off-by: Andriy Skulysh <andriy.skulysh@seagate.com>
Reviewed-on: http://review.whamcloud.com/18244
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4330 lnet: Allocate MEs and small MDs in own kmem_caches

As part of LU-3848 and LU-4330, it has been discovered that LNET
MEs and small MDs (<=128 Bytes) are allocated in <size-128> kmem_cache
and thus can suffer quite frequent corruptions, from other modules or
Kernel parts, that occur there. To avoid this, MEs and small-MDs
specific kmem_cache have been created.

Also, lnet_libmd struct fields have been re-ordered to optimize its
memory foot-print.

Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Change-Id: I6cfbed0c3f0a4c2447fe22b2d0d30dba8e3f3173
Reviewed-on: http://review.whamcloud.com/18586
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7975 lod: fix delayed stripe error path & Client resend

This patch is a follow-on to LU-4260/LU-4971/LU-5346 to fix
cleanup in error path during layout/stripe creation.
It is intended to fix special case when layout creation has been
delayed and will be with a non-0 size (ie truncate).

Problem has been unveiled due to the current limitation where a
delayed layout creation with a non-0 size/truncate will not be
resent if connection with OST is down, but is rather returning
-EAGAIN/-EWOULDBLOCK upon next open, and finally MDS will crash for
"(lod_object.c:700:lod_ah_init()) ASSERTION( lc->ldo_stripenr == 0 )"
LBUG.

Thus, patch also fixes this limitation by instead returning
-EINPROGRESS,this to avoid using costly resend operation from MDS
(will keep a ptlrpc thread busy) but offload this responsibility to
Client side (in after_reply()) which will handle this process in a
much less agressive way.

Specific sanity/test_27F sub-test has been created to test both
layout creation error path correct cleanup and Client resend.

Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Change-Id: Ibc1734f52c071c66cb5974e3fe21d8819d725c2a
Reviewed-on: http://review.whamcloud.com/19302
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8141 tests: fix for acl test

Replace earlier fix for LU-3227 with a better version.
The old fix still allows some failures on sanity, test 103a

Lustre-commit: 30ee7bf159e53fc5b19a4a4605387d09191436e6
Lustre-change: http://review.whamcloud.com/14036

Signed-off-by: Yang Sheng <yang.sheng@intel.com>
Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: James Simmons <uja.ornl@gmail.com>
Reviewed-by: Jian Yu <jian.yu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Change-Id: I4d240eefc4f5d5364edd776aef76397202c36f51
Reviewed-on: http://review.whamcloud.com/20198
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>

LU-8030 hsm: prevent duplicated HSM requests

It is possible to insert identical HSM requests in the HSM action
list. The following command

  lfs hsm-archive foo foo foo foo

results in 4 new archive requests in the "actions" list for the same
file.

Identify such requests and suppress the duplicates.

Signed-off-by: frank zago <fzago@cray.com>
Change-Id: If83d2855948578b94babca96dc72c33f96ae4f71
Reviewed-on: http://review.whamcloud.com/19635
Tested-by: Jenkins
Reviewed-by: jacques-Charles Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7134 utils: Ensure hostid set for ZFS during mkfs

Ensure import protection in ZFS.  spl_hostid must be set to a
non-zero value for it to work correctly.  Error out during mkfs and
tunefs unless override is passed.  Also print error during mount.

Ensure hostid is set in conf-sanity.sh

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: If37318c20e4502a643992e6bc21ac4bc4ad20ed7
Reviewed-on: http://review.whamcloud.com/16611
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-5092 nodemap: users of ted_nodemap should take ref

In order to quickly map clients, the client's assigned nodemap is
saved on the export data struct. When mapping occurs, a reference to
that nodemap is used for mapping. If no reference is taken during
mapping and the nodemap is put elsewhere, then the nodemap could be
destroyed before the mapping is finished. This patch ensures that a
reference is taken before mapping.

Signed-off-by: Kit Westneat <kit.westneat@gmail.com>
Change-Id: I5c4d707970b1a86fb061a501240182ff3dc7be51
Reviewed-on: http://review.whamcloud.com/19674
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8193 ptlrpc: set proper mbits for EINPROGRESS resend

Set mbits for EINPROGRESS resend in ptl_send_rpc().

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Change-Id: If9db4a67530371ce5c2504c60ea79f9c9db6ce15
Reviewed-on: http://review.whamcloud.com/20377
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Liang Zhen <liang.zhen@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7422 llite: don't panic when fid is insane

LASSERT should never be done on data that is
received to over the network. Return EINVAL
when server returns invalid fid despite of
it_status == 0.

Change-Id: I4341d1002761ab4c86bd7af4d0ca5718fe2a3623
Signed-off-by: Sergey Cheremencev <sergey.cheremencev@seagate.com>
Seagate-bug-id: MRP-3073
Reviewed-on: http://review.whamcloud.com/17985
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8022 lnet: Don't access NULL NI on failure path

In kiblnd_passive_connect(), if we are failing the connection
attempt because we cannot find a valid NI (we have a NULL NI),
we were coring after the "goto fail" because the failure
path was assuming non-NULL NI.

This patch ensures we don't dereference a NULL NI on that
failure path.

Signed-off-by: Doug Oucharek <doug.s.oucharek@intel.com>
Change-Id: I4f3deb78b4d3e9d81551ba915ceee852ec3b0792
Reviewed-on: http://review.whamcloud.com/19614
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Matt Ezell <ezellma@ornl.gov>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7848 target: Do not fail MDT-MDT connection

When one MDT restarts, and try to connect to another
MDT with the new cookie, let's do not fail it, otherwise
it might cause eviction between MDT.

Signed-off-by: Di Wang <di.wang@intel.com>
Change-Id: I3d2e0712ee0ed05bdf46c7781e190206a6dcfb1c
Reviewed-on: http://review.whamcloud.com/18800
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4781 test: wait for dbench to stop

In lustre-rsync-test.sh after sending a SIGSTOP to dbench, wait for
all dbench processes to show a stopped state.

Test-Parameters: testlist=lustre-rsync-test,lustre-rsync-test,lustre-rsync-test,lustre-rsync-test,lustre-rsync-test,lustre-rsync-test
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Change-Id: I22767b5002ca4e1b5b1d596380591dc5130ee11d
Reviewed-on: http://review.whamcloud.com/20471
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>

LU-8155 tests: Fix timing of sanity-hsm/16

sanity-hsm/16 occationally fails due to a timing issue.
The measured duration is 19s when the expected duration
is 20s, this is most likely due to rounding of half seconds.

Test-Parameters: trivial testlist=sanity-hsm

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: Iff8b4dec6d9d562b1560cc0101f251bf4b039508
Reviewed-on: http://review.whamcloud.com/20271
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-7352 tests: don't fail conf-sanity test_78 on ENOSPC

Add debugging to conf-sanity.sh test_78 to see why this test is
regularly reporting ENOSPC creating 100x 1MB test files, when
there should be many GB of space available during the test.

Don't fail the test if the OST runs out of space, or if the MDS
refuses to create more files when OST has no available blocks,
since the main goal of this test is to verify ldiskfs resizing.

Test-Parameters: trivial testlist=conf-sanity,conf-sanity,conf-sanity
Test-Parameters: testlist=conf-sanity,conf-sanity,conf-sanity,conf-sanity

Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: Id5a7049aa434c794a5a8f8aa5398bf7ca9bcab07
Reviewed-on: http://review.whamcloud.com/20166
Tested-by: Jenkins
Reviewed-by: Wang Shilong <wshilong@ddn.com>
Reviewed-by: Emoly Liu <emoly.liu@intel.com>

LU-8049 obdclass: use __BIG_ENDIAN to detect local endianness

Originally, we used LUSTRE_MSG_MAGIC to detect the local endianness
as "if (LUSTRE_MSG_MAGIC != cpu_to_le32(LUSTRE_MSG_MAGIC))". But it
is not a good way. It is gratuitously confusing for no reason not
only because it has nothing specific to do with LUSTRE_MSG_MAGIC,
but also because this is a compile-time check but looks like runtime
check. This patch fixes it by checking "__BIG_ENDIAN".

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I7fc15890119b15c178da22b6dc6eebe102403734
Reviewed-on: http://review.whamcloud.com/19812
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>

LU-8134 kernel: kernel update RHEL7.2 [3.10.0-327.18.2.el7]

Update RHEL7.2 kernel to 3.10.0-327.18.2.el7

Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Change-Id: I3b587d11e9ad8051665f38c879c45f1ec5727213
Reviewed-on: http://review.whamcloud.com/20206
Reviewed-by: Minh Diep <minh.diep@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8014 hsm: remove invalid kuc_free in coordinator

Commit 9091a934fc ("LU-8014 hsm: remove kuc_ispayload") moved the KUC
allocation in mdt_hsm_agent_send() and I left a kuc_free in an error
path in mdt_coordinator(). The HAL is no longer a KUC and it is freed
outside the for loop. That kuc_free is incorrect.

Signed-off-by: frank zago <fzago@cray.com>
Change-Id: If0894d1a8bcbc7960b728e33a34bddb95806acf8
Reviewed-on: http://review.whamcloud.com/19869
Tested-by: Jenkins
Reviewed-by: Henri Doreau <henri.doreau@cea.fr>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8080 utils: Replace calls to signal with sigaction

The man page for signal (2) states that sigaction should replace
signal for portability. It also states:
"The effects of signal() in a multithreaded process are unspecified."

This patch affects utils/lhsmtool_posix.c for the second reason and
utils/liblustreapi_hsm.c for completeness even though: "The only
portable use of signal() is to set a signal’s disposition to
SIG_DFL or SIG_IGN." (which is what is done in liblustreapi_hsm.c)

Signed-off-by: Quentin Bouget <quentin.bouget.ocre@cea.fr>
Change-Id: Ib247fab937e3aac514f418c7260cdd67fc9e442c
Reviewed-on: http://review.whamcloud.com/19855
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Henri Doreau <henri.doreau@cea.fr>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8078 iokit: correct obdfilter-survey output data format

A space is needed between min and max status numbers in the
obdfilter-survey script, otherwise plot-obdfilter can't parse
the summary file correctly, so that the related .dat files and
graphs can't be generated correctly.

Test-Parameters: trivial testlist=obdfilter-survey
Signed-off-by: Emoly Liu <emoly.liu@intel.com>
Change-Id: I5bf4790f2db1ee11cb6a7cc95d5527cb0d68e35a
Reviewed-on: http://review.whamcloud.com/19841
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Jian Yu <jian.yu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7905 osd: pin OI objects

so that corresponing dnodes are kept in memory.

Change-Id: Ic312bfc57e3adc1032cefdd97efcf8f819418dc8
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/19103
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7795 quota: tuneable soft least qunit

Introduce a tuneable least qunit for soft block quota, which is
used when soft limit is exceeded to retain an acceptable write
performance.

Remove the stale declarations of quota proc function.

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Change-Id: I172d1d08494418a6f7bc261f022c587da34b784d
Reviewed-on: http://review.whamcloud.com/18896
Tested-by: Jenkins
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Wang Shilong <wshilong@ddn.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-5810 tests: add client hostname to lctl mark

Improve debug messages to include the originating hostname.

Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I441bf8294c38135276a5a0f0853dbebf4358c563
Reviewed-on: http://review.whamcloud.com/13113
Tested-by: Jenkins
Reviewed-by: Emoly Liu <emoly.liu@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6215 lnet: split struct ib_send_wr

In v4.4 Linux kernel split up struct ib_send_wr so that all non-trivial
verbs use their own structure which embeds struct ib_send_wr.

Linux-commit: e622f2f4ad2142d2a613a57fb85f8cf737935ef5
Change-Id: Iea2599ea57a576a4c061841be47a989aba097cc6
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-on: http://review.whamcloud.com/19168
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Li Dongyang <dongyang.li@anu.edu.au>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>