LU-8791 osd-zfs: hold oo_guard read lock for object write

In order to avoid the deadlock of changing object block size and
writing the object at the same time.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: Id1c3c7e66e74d4f61e2136311a0723b8da2da3bb
Reviewed-on: http://review.whamcloud.com/23550
Tested-by: Jenkins
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-8763 ldlm: do not dump update recovery list

Do not dump the update recovery list when recovery is aborted
or when checking whether recovery is complete.  The output
is not useful and is high volume on production systems.

Change-Id: I7f3cd71165475570353cb264b5587749ec252855
Signed-off-by: Olaf Faaland <faaland1@llnl.gov>
Reviewed-on: http://review.whamcloud.com/23444
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: wangdi <di.wang@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-8700 build: dkms do not install llite_lloop

Do not build/install llite_lloop.ko module in DKMS.

Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Change-Id: I61b3455ae35477a83193ef2afca5815135db21cd
Reviewed-on: http://review.whamcloud.com/23228
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Jian Yu <jian.yu@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>

LU-8634 quota: fix return code of intent quota lock

Intent quota operation should return error code in lock_policy_res2
like other intent operations, otherwise, it'll be confused with the
error code returned by intent locking.

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Change-Id: I99fd04f72eeb6d10380ebd30da928b5749e74443
Reviewed-on: http://review.whamcloud.com/22751
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Jenkins
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

New tag 2.8.60

Change-Id: Id8cdb79c3377385f1ae53f18e0969c99e52be59a
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8246 ldlm: Do not grant a lock twice in a race

This leads to wrong ldlm pool accounting of granted locks.
Also handle the case of a destroyed lock.

Change-Id: Ied262d6688766e37f71304e6ee6659b48124e7ad
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: http://review.whamcloud.com/20839
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>

LU-8573 lnet: Revert LU-7650 patches

These patches are causing LU-8573

Revert "LU-7650 o2iblnd: Put back work queue check previously removed"

This reverts commit bde1da1ec098450f40887587b0a46c9eb86a4f6c.

Revert "LU-7650 o2iblnd: handle mixed page size configurations."

This reverts commit 399a5ac1fc73343c69e0fd737032adf5329df1b2.

Change-Id: I4517fab64ac5b1023e615874f58b1bb4902e8c43
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/23439
Tested-by: Jenkins
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Tested-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8707 build: fix lbuild-sles for kernel_module_package

kernel_module_package macro also checking for
/boot/symsets-$kver-$flavor.tag.gz
in case of lbuild, we need to point it to lbuild
kernel-source directory

Change-Id: I3cf9c1f43fe9ea543f67967773fc8715325a47e9
Signed-off-by: Minh Diep <minh.diep@intel.com>
Reviewed-on: http://review.whamcloud.com/23166
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8731 utils: propagate errors in lfs df

Add llapi_obd_fstatfs() which does the same thing as
llapi_obd_statfs() but takes an open file descriptor instead of a
path. Refector the handler for 'lfs df' to use llapi_obd_fstatfs(),
thereby avoiding opening the mount point for each target and making
the error conditions easier to understand. Propagate errors from
llapi_obd_fstatfs() as the exit status of 'lfs df'.

In conf-sanity.sh test_64() allow 'lfs df' to fail when a target is
offline.

Signed-off-by: John L. Hammond <john.hammond@intel.com>
Change-Id: Iabfc92a65571b1a277de7fd42431f5b7e45ad440
Reviewed-on: http://review.whamcloud.com/23286
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Hongchao Zhang <hongchao.zhang@intel.com>

LU-8723 llapi: correct open() handling in llapi_obd_statfs()

In llapi_obd_statfs() remove a spurious errno test and retry after
open().

Signed-off-by: John L. Hammond <john.hammond@intel.com>
Change-Id: I1667f1f0acf0e1f0d700049bc39a5e6c462b9df6
Reviewed-on: http://review.whamcloud.com/23285
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>

LU-8748 osd-zfs: set block size of echo object

Set block size for zfs echo object.

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Change-Id: I6efab645181ab3de6686bf82f4ecbf9ea3384b1b
Reviewed-on: http://review.whamcloud.com/23323
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8311 doc: add NIDs examples to mkfs.lustre and mount.lustre

This patch adds examples of how NIDs can be specified on
mkfs.lustre and mount command lines.

Test-Parameters: trivial

Signed-off-by: Jian Yu <jian.yu@intel.com>
Change-Id: I93a8f843755582f94844504f604409dd43b617f9
Reviewed-on: http://review.whamcloud.com/23355
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Zhiqi Tao <zhiqi.tao@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8775 osd: do not report special writes in brw stats

do not report special writes (e.g. last_rcvd, etc) made with
copying write in brw_stats.

Change-Id: Id92ed13a8d241e6489731d51a546c3583a2156b8
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/23363
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Zhiqi Tao <zhiqi.tao@intel.com>
Tested-by: Zhiqi Tao <zhiqi.tao@intel.com>
Tested-by: Jenkins
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8755 kernel: kernel update [SLES12 SP1 3.12.62-60.64.8]

Update target and kernel_config files for new version

Test-Parameters: clientdistro=sles12 testgroup=review-ldiskfs \
  mdsdistro=sles12 ossdistro=sles12 mdsfilesystemtype=ldiskfs \
  mdtfilesystemtype=ldiskfs ostfilesystemtype=ldiskfs

Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Change-Id: I52ab5a1ecff81e470b965af536fd0f638a120546
Reviewed-on: http://review.whamcloud.com/23364
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8751 kernel: kernel update RHEL7.2 [3.10.0-327.36.3.el7]

update RHEL7.2 kernel to 3.10.0-327.36.3.el7

Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Change-Id: I0c6c457a6e48c4166508572d91e7b98f9ed4ad86
Reviewed-on: http://review.whamcloud.com/23335
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8749 osd-ldiskfs: inherit S_ISGID correctly

For remote directory S_ISGID is inherited on agent, not where file
resides, and also the group inherited from parent.

Update sanity 6g to test this.

Signed-off-by: Lai Siyao <lai.siyao@intel.com>
Change-Id: I159b2687ad00fdc7c35f60a18668a015240a1953
Reviewed-on: http://review.whamcloud.com/23329
Tested-by: Jenkins
Reviewed-by: Niu Yawei <yawei.niu@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-8733 gnilnd: Remove read capability of cksum_test procfile

When the old create proc interface was deprecated cksum_test was
updated  to use the new file operations table. Inadvertantly read
was left as a capability without actually defining a function
that the file would use when someone tried to read the file.
This causes a kernel crash when cksum_test is read, though it can
only be done by the root user.

The fix is to remove the .read op from the fops table for the
cksum_test proc entry

Test-Parameters: trivial
Signed-off-by: Chris Horn <hornc@cray.com>
Change-Id: I406b076f1b66b6d991694c69a9b748ed42c09f39
Reviewed-on: http://review.whamcloud.com/23255
Tested-by: Jenkins
Reviewed-by: James Simmons <uja.ornl@yahoo.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-7206 mdd: stop orphan cleanup before finish FLD

Stop orphan cleanup thread in PRE_CLEANUP phase.
Because orphan cleanup threads might need lookup
FLD, (__mdd_orphan_cleanup()->mdd_object_init()->
lod_object_alloc() ->lod_fld_lookup()), so let's
stop orphan cleanup threads before FLD cleanup.

Signed-off-by: Di Wang <di.wang@intel.com>
Change-Id: I8df9832c633017e2fca866579b497f8215054d31
Reviewed-on: http://review.whamcloud.com/23029
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-8347 ldlm: granting conflicting locks

Postpone lock reprocess during lock replay stage.
Reprocess is needed during request replay stage
beacause local locks are still in use until
client ACK.

Change-Id: I250d22fee471db643f12a900fdfc51eacfa94aa2
Seagate-bug-id: MRP-3516
Signed-off-by: Andriy Skulysh <andriy.skulysh@seagate.com>
Reviewed-on: http://review.whamcloud.com/21059
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8633 llite: do not clear uptodate bit in page delete

Otherwise, if the race between page fault and truncate occurs, it
will cause the page fault routine to return an EIO error.

In filemap_fault() {
    page_not_uptodate:
...
        ClearPageError(page);
        error = mapping->a_ops->readpage(file, page);
        if (!error) {
                wait_on_page_locked(page);
                if (!PageUptodate(page))
                        error = -EIO;
}
...
}

However, I tend to think this is a defect in kernel implementation,
because it assumes PageUptodate shouldn't be cleared but file read
routine doesn't make the same assumption.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: Ic4a919607a6121098e41eaf56b8ce3200f778ecf
Reviewed-on: http://review.whamcloud.com/22827
Tested-by: Maloo <hpdd-maloo@intel.com>
Tested-by: Jenkins
Reviewed-by: Li Dongyang <dongyang.li@anu.edu.au>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8680 osc: soft lock - osc_makes_rpc()

It is possible that an osc_extent contains more than 256 chunks, and
the IO engine won't add this extent in one RPC
(try_to_add_extent_for_io) so that osc_check_rpcs() run into a loop
upon this extent and never break.

This patch changes osc_max_write_chunks() to make sure the value
can cover all possible osc_extent, so that all osc_extent will be
added into one RPC.

This patch also add another field erd_max_extents in extent_rpc_data
to make sure not to add too many fragments in a single RPC.

Signed-off-by: Bobi Jam <bobijam.xu@intel.com>
Change-Id: Icf58a6bd04655bb9aa5589dd002e118c21ed932d
Reviewed-on: http://review.whamcloud.com/23326
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-930 doc: move DLC doc to Documentation dir

Move DLC documentation to top-level Documentation/ directory instead
of in the lustre/doc directory where man pages are kept.

Test-Parameters: trivial
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I4e5c01a9ee796e099dbfcfb73a315c8187931cf0
Reviewed-on: http://review.whamcloud.com/22931
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Joseph Gmitter <joseph.gmitter@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8590 gss: Move DH parameter generation out of upcall

This change adds the Diffie-Hellman parameter generation to the
lgss_sk utility prior to key loading.  The parameters are now
persistent to prevent long DH parameter generation times which
can cause mount command and connection timeouts.

This is based on recommendations from Matt Wood at Intel's
security review.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: Iba840168da533662ed8ec78004be9e4dc5369c68
Reviewed-on: http://review.whamcloud.com/23322
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: Oleg Drokin <oleg.drokin@intel.com>

LU-8186 llite: Typo in ll_rw_extents_stats_pp_seq_show

Add a missing quote character to ll_rw_extents_stats_pp_seq_show. Also
correct leading whitespace to match coding guidelines.

This corrects the text displayed on clients in
/proc/fs/lustre/llite/.../extents_stats_per_process

Test-Parameters: trivial
Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: I3046836372182925ea0f3b0f5909ae7f8dc5efd1
Reviewed-on: http://review.whamcloud.com/23248
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8124 osd-zfs: fix statfs small blocksize inode estimate

When a small recordsize is specified for the MDT dataset (e.g. 4KB)
the current statfs estimate for the total number of dnodes available
is constrained to assume one dnode per 4KB block.  However, if the
ZFS sector size is 4KB (ashift=12) then the SA (xattr) spill block
will also be allocated in 4KB units and ditto'd, consuming 8.5KB per
dnode plus extra overhead (OI, directory ZAP, etc).  If lots of
directories are created, there will be up to 64KB of space consumed
per dnode.  This throws off the dnode estimations significantly.

Instead, do not constrain the statfs dnode calculation by the small
recordsize and use the actual average space per dnode when estimating
the total number of dnodes the filesystem can hold.

Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I5403a855a0bd3d9077ef0e661d2f262ffa2cab07
Reviewed-on: http://review.whamcloud.com/20123
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8250 mdd: move linkea prepare out of transaction.

Move linkea prepare out of transaction to avoid reading
linkea remotely inside the transaciton.

Signed-off-by: Di Wang <di.wang@intel.com>
Change-Id: I10f0979c0c496fdcc5349f36ac5cca123d42c8a5
Reviewed-on: http://review.whamcloud.com/23096
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8378 all: remove set but unused variables

Remove set but unused variables as report list.

Signed-off-by: Yang Sheng <yang.sheng@intel.com>
Change-Id: If9120ec088a2dd0b65564330bc295c08a1e579b7
Reviewed-on: http://review.whamcloud.com/23221
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Emoly Liu <emoly.liu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8189 osc: osc_match_base prototype differs from declaration

The patch updates the prototype in osc_internal.h to match the
enums used in the declaration.

The osc_match_base declaration in lustre/osc/osc_request.c uses
enums for stricter checking on the type and mode parameters:

int osc_match_base(struct obd_export *exp,
   ...
-->    enum ldlm_type type,
   union ldlm_policy_data *policy,
-->    enum ldlm_mode mode,
   ... int unref)

The prototype in lustre/osc/osc_internal.h instead used unsigned ints:

int osc_match_base(struct obd_export *exp,
   ...
-->    __u32 type,
   union ldlm_policy_data *policy,
-->    __u32 mode,
   ... int unref);

Test-Parameters: trivial
Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: I7ccc6383e0e12bf4fe5b5c3bad822f3322aaa1ff
Reviewed-on: http://review.whamcloud.com/23167
Tested-by: Jenkins
Reviewed-by: Frank Zago <fzago@cray.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6303 osc: remove handling cl_avail_grant less than zero

Earlier cl_avail_grant was changed to an unsigned int. Juila
Lawall reported for the upstream client the following which
affects the Intel branch as well:

drivers/staging/lustre/lustre/osc/osc_request.c:1045:5-24: WARNING: Unsigned
     expression compared with zero: cli -> cl_avail_grant < 0

Since cl_avail_grant can never be negative we can remove the
code handling the negative value case.

Change-Id: I10f7ac3aaab7ebf03a7f7ac0717b60134f09cddf
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: http://review.whamcloud.com/23155
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Jenkins
Reviewed-by: Frank Zago <fzago@cray.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8697 llite: remove IS_ERR(master_inode) check

The kernel function ilookup5_nowait never returns
IS_ERR so we can remove the IS_ERR check in the
ll_md_blocking_ast() function.

Change-Id: I5e72a8f70857f178a2377e9db80b2e2139c56ec3
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: http://review.whamcloud.com/23151
Tested-by: Jenkins
Reviewed-by: Frank Zago <fzago@cray.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8705 tests: do not skip lnet-selftest for DNE

Patch removes the skip added by LU-4181.

Test-Parameters: mdtfilesystemtype=ldiskfs mdsfilesystemtype=ldiskfs ostfilesystemtype=ldiskfs mdscount=2 mdtcount=4 testlist=lnet-selftest
Seagate-bug-id: MRP-3912
Signed-off-by: Elena Gryaznova <elena.gryaznova@seagate.com>
Reviewed-by: Vladimir Saveliev <vladimir.saveliev@seagate.com>
Reviewed-by: Alexander Lezhoev <alexander.lezhoev@seagate.com>
Tested-by: Alexander Lezhoev <alexander.lezhoev@seagate.com>
Change-Id: Id78ef3909896325d55569ea948f75906cf0b7c87
Reviewed-on: http://review.whamcloud.com/23138
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-7593 target: take ted_lcd_lock after transaction started

Otherwise the thread1 may be blocked during the transaction start
in tgt_client_data_update() with 'ted_lcd_lock' held because another
thread2 may be blocked by such lock in tgt_txn_stop_cb() but with
transaction handle started. That is deadlock.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: Id623d171e43beaa54ae4a9718fb4dc52c474df01
Reviewed-on: http://review.whamcloud.com/23129
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Tested-by: Jenkins
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8692 kernel: kernel update RHEL7.2 [3.10.0-327.36.2.el7]

Update RHEL7.2 kernel to 3.10.0-327.36.2.el7

Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Change-Id: I936a4d6f73913d64af76fdbaf51964d7ad2c7e8f
Reviewed-on: http://review.whamcloud.com/23102
Tested-by: Jenkins
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8695 target: use -1 as an offset to declare write

at the end of recovery or filesystem setup the number of clients
may increase significantly. this can lead to underestimated space
or credits reserved.

Change-Id: Id4f3755dc481f8a29a1a2a673c26d64d12f7dbf0
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/23082
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: Oleg Drokin <oleg.drokin@intel.com>

LU-8685 kernel: jbd2: fix incorrect unlock on j_list_lock

This patch has been back-ported to avoid kernel Oopses/BUG()s
due to j_list_lock found unlocked when expected to be locked!

In jbd2_journal_get_create_access(),
when 'jh->b_transaction == transaction' (asserted by below)

  J_ASSERT_JH(jh, (jh->b_transaction == transaction || ...

'journal->j_list_lock' will be incorrectly unlocked, since
the the lock is aquired only at the end of if / else-if
statements (missing the else case).

This bug has been introduced by an earlier change named
"jbd2: minimize region locked by j_list_lock in
journal_get_create_access()".

Signed-off-by: Taesoo Kim <tsgatesv@gmail.com>
Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Change-Id: Ifb8b038333e523caa1b274f53f49317182895de5
Reviewed-on: http://review.whamcloud.com/23050
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Wang Shilong <wshilong@ddn.com>

LU-8683 readahead: update ras window correctly

When stride-RA hit case miss, we only reset normal sequential
read-ahead window, but not reset the stride IO to avoid the overhead
of re-detecting stride IO. While when the normal RA window is set
to not insect with the stride-RA window, when we try to increase
the stride-RA window length later, the presumption does not hold.

This patch resets the stride IO as well in this case.

Lustre-change: http://review.whamcloud.com/17343
Lustre-commit: 88ef5af0bed93c88984c226db755d07601aef60f

Signed-off-by: Bobi Jam <bobijam.xu@intel.com>
Change-Id: Iba6e51f12ac4d00548cc99b7bd423502b754db13
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: http://review.whamcloud.com/23032
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: wangdi <di.wang@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>

LU-8631 quota: better error message for 'lfs quota'

'lfs quota' should return useful error message when it's being
issued on a non-Lustre fs.

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Change-Id: I9dd08982077132756c8684b44430251da53cbb90
Reviewed-on: http://review.whamcloud.com/23021
Tested-by: Jenkins
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8492 ptlrpc: Correctly calculate hrp->hrp_nthrs

cpu_pattern can specify exactly 1 cpu in a partition:
"0[0]". That means CPT0 will have CPU 0. CPU 0 can have
hyperthreading enabled. This combination would result in

weight = cfs_cpu_ht_nsiblings(0);
hrp->hrp_nthrs = cfs_cpt_weight(ptlrpc_hr.hr_cpt_table, i);
hrp->hrp_nthrs /= weight;

evaluating to 0. Where
cfs_cpt_weight(ptlrpc_hr.hr_cpt_table, i) == 1
weight == 2

Therefore, if hrp_nthrs becomes zero, just set it to 1.

Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Change-Id: Id89d381436b2c61354d925420f2efce8d9a54864
Reviewed-on: http://review.whamcloud.com/19106
Tested-by: Jenkins
Reviewed-by: Liang Zhen <liang.zhen@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8667 osp: validate FID before initializing precreate seq

A few fixes for FID on OST...

Check if the OSP FID is valid before initializing
precreate sequence, in case the last_seq/oid file
is corrupted. Only MDT0 can use IDIF, and non-MDT0
can only use normal FID, which will also make sure
the following orphan cleanup will use the valid
sequence.

OFD will validate the object sequence to make sure
IDIF request will only operate on MDT0 group.

If MDT can not get new sequence from OST, it will
sleep 2 seconds before retry, to offer OSTs some
extra time to setup FID service with sequence
controller (MDT0).

Signed-off-by: Di Wang <di.wang@intel.com>
Change-Id: Id9bbfaf9170ff1a9240719eaee73ddcb4fd804e5
Reviewed-on: http://review.whamcloud.com/22934
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: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4931 ladvise: add code for ladvise_hdr into wirecheck.c

Add code into wirecheck.c to generate the ladvise_hdr checks
in wiretest.c.

Test-Parameters: trivial

Signed-off-by: Gu Zheng <gzheng@ddn.com>
Change-Id: Ic4488b2d6004d284a4fbf123ab7a0688da227212
Reviewed-on: http://review.whamcloud.com/21940
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: Oleg Drokin <oleg.drokin@intel.com>

LU-8089 lwp: change lwp export only at first connect

Fix lwp connection logic, change export only at first
connection.

Signed-off-by: Mikhail Pershin <mike.pershin@intel.com>
Change-Id: I5b528c2f6d142b8cfdd8d84f3f540289e45d557f
Reviewed-on: http://review.whamcloud.com/13726
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8694 docs: ZFS hostid mkfs.lustre(8) man page update

Add note about needing to reload spl module after creating /etc/hostid
file.

Test-Parameters: trivial
Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: Ie4c3697567ed2722742324fd70f382bb46a886d6
Reviewed-on: http://review.whamcloud.com/23118
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Steve Guminski <stephenx.guminski@intel.com>
Reviewed-by: Cliff White <cliff.white@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8713 utils: Try loading zfs.ko during zfs_init

Newer version of zfs (0.6.5.8 and 0.7.0) do not autoload zfs module at
boot nor do they load it during libzfs_init(), so try loading it
during initialization.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: I44765b0c2b7ea8b1c8c6d45a9107842b17623dbc
Reviewed-on: http://review.whamcloud.com/23210
Tested-by: Jenkins
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@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-8574 osd-ldiskfs: fix FID-in-dirent properly

Sometimes, the directory entry may be corrupted and contains bad
ino# information as to point to the inode that belong to other.
Two cases for that:

1) If such inode is unless, then the up layer namespace LFSCK may
   handle it as dangling name entry (by guess), and as the LFSCK
   progressing, its original inode (orphan) may be found then the
   LFSCK will fix the such bad name entry to reference the orphan.

2) If such inode is used by other, then osd_dirent_check_repair()
   will find that the FID-in-dirent does not match the FID-in-LMA,
   and fix the FID-in-dirent. So the up layer namespace LFSCK will
   get the wrong FID (not the original FID-in-dirent). Under such
   case, the up layer LFSCK will fix it as unmatched pairs, then
   the orphan cannot be recovered.

In fact, when injecting failure stub for LFSCK sanity test, we
hit similar issues as the 2) case. To resolve such trouble, we
can enhance the osd_dirent_check_repair() logic as following:
If the FID-in-dirent does not match the inode's FID-in-LMA, then
check the inode's linkEA, if it recognizes the dirent entry (with
parent dir's FID + child name), then trouble the inode, and fix
the FID-in-dirent with the FID-in-LMA. Otherwise, the dirent may
be corrupted, as the LFSCK processing, the right orphan may be
found, then fix it later. The worst case is that the inconsistence
is detected but kept there, but it will make wrong fixing.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I3516a6ad9cc1766453612c440df7db02bc2f09a4
Reviewed-on: http://review.whamcloud.com/22310
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6808 ptlrpc: no need to reassign mbits for replay

It's not necessary reassgin & re-adjust rq_mbits for replay
request in ptlrpc_set_bulk_mbits(), they all must have already
been correctly assigned before.

Such unecessary reassign could make the first matchbit not
PTLRPC_BULK_OPS_MASK aligned, that'll trigger LASSERT in
ptlrpc_register_bulk():

- ptlrpc_set_bulk_mbits() is called when first time sending
  request, rq_mbits is set as xid, which is BULK_OPS aligned;

- ptlrpc_set_bulk_mbits() continue to adjust the mbits for
  multi-bulk RPC, rq_mbits is not aligned anymore, then rq_xid
  is changed accordingly if client is connecting to an old
  server, so rq_xid became unaligned too;

- The request is replayed, ptlrpc_set_bulk_mbits() reassign
  the rq_mbits as rq_xid, which isn't aligned already, but
  ptlrpc_register_bulk() still assumes this value as the
  first matchbits and LASSERT it's BULK_OPS aligned.

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Change-Id: Ib5d5a969702d3b621fb44643586cc19bf931c365
Reviewed-on: http://review.whamcloud.com/23048
Tested-by: Jenkins
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8450 nodemap: modify ldlm_revoke_export_locks

Modify ldlm_revoke_export_locks to use cfs_hash_for_each_nolock
instead of cfs_hash_for_each_empty, to avoid looping multiple times
over the export hash.

Signed-off-by: Kit Westneat <kit.westneat@gmail.com>
Change-Id: I8ab475b8117493ed55961e94e747b4955141f003
Reviewed-on: http://review.whamcloud.com/23120
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-8413 test: limit # of processes for sanity test_101f

It tries to test mmap readahead does not miss too much pages,
so we need limit the number of processes of the iozone read.

Test-Parameters: trivial
Signed-off-by: Bobi Jam <bobijam.xu@intel.com>
Change-Id: Ic5c753549234fe2bffcaa42f75208c2295f84ac0
Reviewed-on: http://review.whamcloud.com/23039
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8682 llite: protect from accessing NULL lli_clob

Need to check file's lli_clob object before calling
lov_read_and_clear_async_rc().

lustre-change: http://review.whamcloud.com/16954
lustre-commit: 4e416d205b0d04e291e0f279741af85fa73845d2

Signed-off-by: Bobi Jam <bobijam.xu@intel.com>
Intel-bug-id: LDEV-231
Change-Id: I210b22d4e17dd4a407378aa987434d1940799f1f
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: http://review.whamcloud.com/23031
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-8681 osd: ingore ENODATA during unlink agent parent

If the directory is created before 2.0, there are no LMA
for the directory, then osd_delete_from_remote_parent()
will return ENODATA. So we should ignore ENODATA in this
case otherwise it will cause unlink fails.

Lustre-commit: 860283cb433dff4246a5c255ed89325323ee8e7c
Lustre-change: http://review.whamcloud.com/16760

Signed-off-by: Wang Di <di.wang@intel.com>
Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Intel-bug-id: LDEV-64
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Change-Id: I728c081f324253080c5d2cb740b3a11b26f9d570
Reviewed-on: http://review.whamcloud.com/23030
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>

LU-8669 kernel: kernel update RHEL6.8 [2.6.32-642.6.1.el6]

Update RHEL6.8 kernel to 2.6.32-642.6.1.el6

Test-Parameters: clientdistro=el6.8 mdsdistro=el6.8 ossdistro=el6.8 \
  mdsfilesystemtype=ldiskfs mdtfilesystemtype=ldiskfs \
  ostfilesystemtype=ldiskfs testgroup=review-ldiskfs

Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Change-Id: I84e92c261aec3c438112504bdf51555214fcc845
Reviewed-on: http://review.whamcloud.com/22960
Tested-by: Jenkins
Reviewed-by: Minh Diep <minh.diep@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8636 build: prevent src download at the same time

If the lbuild kernelsrc/kernelrpm directory is shared
among the builders, we need to keep them from using
it while it's being downloaded

Test-Parameters: trivial

Change-Id: I240996c84ea541f5985b8d6ec73e3c6a56d2d805
Signed-off-by: Minh Diep <minh.diep@intel.com>
Reviewed-on: http://review.whamcloud.com/22939
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8660 mgs: handle return code of server_make_name()

Make sure server_make_name() actually succeeded when
called in mkfs_lustre utility and mgs_set_index().

Change-Id: I218351f0f3dd98e1b928664f872c1702a419a7cc
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: http://review.whamcloud.com/22867
Tested-by: Jenkins
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Jian Yu <jian.yu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7501 utils: keep lfs arguments consistent

lfs getstripe use -m and deprecate -M to match lfs setstripe

lfs setstripe use --ost to match lfs find and deprecate --ost-list

lfs find add --mdt-index to match lfs setstripe

lfs getstripe and migrate add --mdt to match lfs find

lfs setdirstripe add --mdt-count|-c, --mdt-index|-m, --mdt-hash as
    aliases for the existing --count|-c and --index|-i and
    --hash-type|-t options

Signed-off-by: Yang Sheng <yang.sheng@intel.com>
Change-Id: I8983b7438e7046657a087e97d9fbc467242c9dd8
Reviewed-on: http://review.whamcloud.com/22581
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
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: Oleg Drokin <oleg.drokin@intel.com>

LU-8239 utils: llanalyze drops useful logs

llanalyze should not execute entering_rpc and
leaving_rpc if no rpctrace option set, otherwise wrong
pid is set and usefull logs are skipped.

This patch executes entering_rpc and leaving_rpc
only if rpctrace is set.

Test-Parameters: trivial
Signed-off-by: Artem Blagodarenko <artem.blagodarenko@seagate.com>
Change-Id: I4676b0ca80f8ed314716e9368558489b64825ffa
Reviewed-on: http://review.whamcloud.com/20641
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-8239 utils: llanalyze logs parser fix

llanalyze looks to be broken:
- thread PID parsed not correctly;
- competition and blocking upcalls start and finish message is parsed
wrong way.

This patch fixes problems with logs parsing.

Test-Parameters: trivial
Signed-off-by: Artem Blagodarenko <artem.blagodarenko@seagate.com>
Change-Id: I412120a2c7e877ba5374178a6a246984e2dcca08
Reviewed-on: http://review.whamcloud.com/20632
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-8289 utils: add ll_decode_linkea tool

A MDT recovered by fsck might contain some files under lost+found
directory of ldiskfs. And in order to get the right path to move
them to, the xattr of trusted.link could be used to extract the
parent FIDs.

This path adds an new tool ll_decode_linkea to dump the parent FIDs
of a file.

Test-Parameters: trivial
Signed-off-by: Li Xi <lixi@ddn.com>
Signed-off-by: Qian Yingjin <qian@ddn.com>
Change-Id: I1da6ea78ab2f9e2db8fbdfdcfb8690c57e9eb2b0
Reviewed-on: http://review.whamcloud.com/20444
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6635 lfsck: block replacing the OST-object for test

For sanity-lfsck test_18e, sometimes, before the client wirte
happened, the layout LFSCK has already gone into the phase2
and replaced the new created OST-object with the old orphan
OST-object, then cause the subsequent check failure.

To resolve such trouble, we will hold the layout LFSCK when
replacing the new created OST-object until the client write
happened.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: Ia5be2cd2ea920fa02bb281ca5c59d8d252f3fd7b
Reviewed-on: http://review.whamcloud.com/18146
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: Oleg Drokin <oleg.drokin@intel.com>

LU-8645 ptlrpc: update imp_known_replied_xid on resend-replay

The imp_known_replied_xid should be updated when try to resend
an already replied replay request, because the xid of this replay
request could be less than current imp_known_replied_xid.

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Change-Id: Iaafcb187efb24cc88dc96b30a63083fac83a9078
Reviewed-on: http://review.whamcloud.com/22776
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Tested-by: Jenkins
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8569 lfsck: cleanup lfsck requests list before exit

When the lfsck assistant thread hits some failure and exit
at the second stage scanning, it does not cleanup the requests
list 'lfsck_assistant_data::lad_req_list', that may cause the
lfsck main engine hit "LASSERT(list_empty(&lad->lad_req_list))"
when handles double scan.

This patch unifies the assistant thread exit process: before
cleanup the list 'lfsck_assistant_data::lad_req_list' set the
thread as "stopping" to prevent more lfsck requests being added,
then cleanup the list 'lfsck_assistant_data::lad_req_list'.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I7facd20e2742c7ad5d4fbee1c975dacd8f6ea363
Reviewed-on: http://review.whamcloud.com/22723
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 ssk: fix SK_IV_REV_START on 32-bit systems

Use a 64-bit constant for SK_IV_REV_START so that it doesn't
overflow on 32-bit systems.

Test-Parameters: trivial testlist=sanity-sec
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: Ie790f7ea438847b18e7b83689e3816ceffe2ddd7
Reviewed-on: http://review.whamcloud.com/23089
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 gss: Fix issues with SK privacy and integrity mode

This patch has several fixes for skpi:

1. The original SK patches failed to account for out of order
handling of RPCs and bulk pages during encryption.  As a result
clients would be out of sync with the IV used for decryption.
This patches moves the encryption to a format similar to RFC3686
to handle these RPCs and bulk pages.

2. A header was added to the SK mode RPCs to allow versioning and
send the unencrypted IV used for an RPC.  The versioning will allow
for future protocol changes.

3. Several changes to fix or impove security of the implementation
based on a security review from Matthew Wood at Intel:
- Derive a unique key for integrity modes instead of using the
  shared secret key (ska, ski, and skpi modes).  This helps prevent
  replays.
- Use PBKDF2 instead of HMAC to derive keys for integrity and
  encryption.
- Have the server side pass a random value (like the client) and
  incorporate this value into the key binding information.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I247187ecbd8cb23c602cec6a92eca938f135e564
Reviewed-on: http://review.whamcloud.com/21922
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>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8429 gnilnd: Option to not reconnect after conn timeout

When routers time out a client connection during a catastrophic
network disturbance like a cabinet EPO, there still may be
traffic from the file system that is using the router for the
return path to the client. This will cause a new connection to try
to be formed before the network has quiesced causing multiple failed
connection attempts which need to be put in purgatory since they could
possibly connect in the future. This can cause the gart space to be
consumed with registrations.

To avoid this, add a module parameter to_reconn_disable which when set
will change the state of the peer that has timed out to PEER_TIMED_OUT
which will act just like PEER_DOWN so that no traffic will be
attempted to a peer in this state.

When the network recovers, the client will form a new connection and
the state will change back to PEER_UP.

Changed gnp_down to gnp_state and GNILND_RCA_NODE_* to GNILND_PEER_*.

To add this option to routers, update /etc/modprobe.conf.local with:
options kgnilnd to_reconn_disable=1

To dynamically add this parameter to a booted node:
echo 1 > /sys/module/kgnilnd/parameters/to_reconn_disable

Tested functionality with both timing out a connection and bringing
down nodes to check the proper states are entered.

Test-Parameters: trivial
Signed-off-by: Chris Horn <hornc@cray.com>
Change-Id: I19cebab401208133d94e29c603eb340f77354684
Reviewed-on: http://review.whamcloud.com/21459
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-6387 tests: fix lp_utils build issues on Power8

With the latest RHEL7 image for power8 some compile
issues have surfaced. Currently the issue is with
begin() and end() inline functions for the lp_util
code. The solution appears to not make the functions
inline.

Change-Id: I8f11ed488890407c117f13ebc741d7140702c647
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: http://review.whamcloud.com/22941
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Frank Zago <fzago@cray.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8575 lod: clear ost usable flag to avoid striping.

clear ost->ltd_qos.ltq_usable before checking
OBD_FAIL_MDS_OSC_PRECREATE param and some other flags in
lod_alloc_qos() while finding all ost's which are valid stripe
candidates. Call to lod_alloc_qos before setting this flag/param
sets ost->ltd_qos.ltq_usable to 1 for all valid ost but after
setting this flag/param ost->ltd_qos.ltq_usable not getting
cleared so there is need to clear it so that it will not get added
into stripe array.

Also in test_27u corrects problem where log file gets created in
lustre mount directory instead of /tmp. This make sures TLOG file
will not get created on ost0.

Seagate-bug-id: MRP-2847

Signed-off-by: Jadhav Vikram <jadhav.vikram@seagate.com>
Change-Id: I6dd16c44b331b09a0ffee530b3eb8508bde64294
Reviewed-on: http://es-gerrit.xyus.xyratex.com:8080/9757
Tested-by: Jenkins
Reviewed-by: Ujjwal Lanjewar <ujjwal.lanjewar@seagate.com>
Reviewed-by: Vladimir Saveliev <vladimir.saveliev@seagate.com>
Tested-by: Elena V. Gryaznova <elena.gryaznova@seagate.com>
Reviewed-on: http://review.whamcloud.com/22367
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7903 ptlrpc: leaked rs on difficult reply

reply_out_callback() should call ptlrpc_schedule_difficult_reply()
to finalize the rs if it's already not on uncommitted list, otherwise,
the rs and the export held by rs could be leaked:

- target_send_reply() sends a difficult reply before the transaction
  committed, the reply is linked to scp_rep_active;

- export gets disconnected by umount or whatever reason,
  server_disconnect_export() is called to complete all outstanding
  replies, which will calls into ptlrpc_handle_rs() to dispose of
  the rs, so the rs is removed from the uncommitted list and
  LNetMDUnlink() is called to unlink the reply buffer and generate
  an unlink event;

- reply_out_callback() is called to process above unlink event,
  ptlrpc_schedule_difficult_reply() is supposed to be called to
  dispose of the rs finally. However, it could be skipped because of
  following flawed code snippet:

  if (!rs->rs_no_ack ||
      rs->rs_transno <= rs->rs_export->exp_obd->obd_last_committed)
          ptlrpc_schedule_difficult_reply(rs);

  The intention of above code is: if rs_no_ack is true (COS enabled),
  and transaction is not committed, we should rely on commit callback
  to release the rs. However, it overlooked the situation that rs
  could have been removed from the uncommitted list by disconnecting
  export.

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
Change-Id: I76d1a9e08c94340520fb731e8671b7b9205e0eb1
Reviewed-on: http://review.whamcloud.com/22696
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8592 mdt: hold mdt_device::mdt_md_root until service stop

Otherwise, if someone is using such object, it may trigger
object reference ASSERTION(atomic_read(&o->lo_header->loh_ref) > 0).

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: Ifd7adbce9a68da537f592c64117f4ecafb0a9ec4
Reviewed-on: http://review.whamcloud.com/22438
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8668 tests: print more information about dd failure

Don't drop the dd output so that we can see why this is failing.

Test-Parameters: trivial testlist=sanityn,sanityn,sanityn,sanityn
Test-Parameters: trivial testlist=sanityn,sanityn,sanityn,sanityn
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I3364bca75817ee4d0128079744e4b01a3ac938a0
Reviewed-on: http://review.whamcloud.com/22990
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 gss: Change the handling of keys for SK

Servers were automatically loading keys of the client type to allow
server to server communication to work by only including a path
to the --skpath option of mount.lustre.  However, this has multiple
issues due to ordering with multiple keys and can be unpredictable.
Instead keys that will be used for server to server communication
must be loaded manually or by a pre-mount script using lgss_sk
and specifiying the client type.

In addition client's should only load a single key with --skpath so
a check is added to not allow directories on the client.

Signed-off-by: Jeremy Filizetti <jeremy.filizetti@gmail.com>
Change-Id: I239753fa1a2bff19bed598e6d2a073e8567d1002
Reviewed-on: http://review.whamcloud.com/22626
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>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8658 ptlrpc: Suppress error for flock requests

-EAGAIN is a normal return when requesting POSIX flocks.
We can't recognize exactly that case here, but it's the
only case that should result in -EAGAIN on LDLM_ENQUEUE, so
don't print to console in that case.

Signed-off-by: Patrick Farrell <paf@cray.com>
Change-Id: Idfbaf671023ac2c3dc84ddd62d2e547427b1f50b
Reviewed-on: http://review.whamcloud.com/22856
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8648 all: remove all Sun license and URL references

The link to the GPL license at Sun is no longer there.
Also Sun Inc is gone so lets remove all references to
it in our source code.

Test-Parameters: trivial

Change-Id: Ib47d07dcef49a1ef624d0d8a4a91b9731f486278
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: http://review.whamcloud.com/22800
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Frank Zago <fzago@cray.com>

LU-7111 revert: b=22386 disallow wrong conf_param options

Don't fail mount if there are unrecognized parameters.  They may
change or be retired over time.

This reverts commit 12b6a5593a1d876ca1be91303da85a93ef6a92ca.

Change-Id: I806fd2345b854c8464632125210f84c7a029b78d
Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-on: http://review.whamcloud.com/21112
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Parinay Kondekar <parinay.kondekar@seagate.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8226 tests: Change check_catastrophe() to check_node_health()

Change check_catastrophe() to check_node_health() and take care of
following,
    - whether the remote node was running,
    - whether there was an LBUG/LASSERT

and print a proper error message in both case for particular
node.

Seagate-bug-id: MRP-3380
Change-Id: Ia42600ab6e6ee3925d3bda9ac9561149d60d8d72
Signed-off-by: Parinay Kondekar <parinay.kondekar@seagate.com>
Reviewed-on: http://review.whamcloud.com/20539
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Jian Yu <jian.yu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7988 hsm: make mdt_hsm_cdt_start a static function

It's only used in the file it's declared in.

Test-Parameters: trivial
Signed-off-by: Frank Zago <fzago@cray.com>
Change-Id: Ic0b3af0c18cfb09f2f94efbac814429f0c43cb4b
Reviewed-on: http://review.whamcloud.com/22704
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Vinayak <vinayakswami.hariharmath@seagate.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6245 libcfs: remove cfs_fs_time handling

Only the mdd layer uses cfs_fs_time handling. We can
easily convert it to the standard linux kernel time
keeping api in the latest kernels. Besides this change
the handling of time in 64 bits to avoid the 2038
limit is introduced.

Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Change-Id: I0bac2020d078c2d91347f5a3a68e99d25ba97575
Reviewed-on: http://review.whamcloud.com/22857
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Frank Zago <fzago@cray.com>
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6245 libcfs: remove last of types abstractions

With the complete removal of the types abstraction
in the lustre kernel code we can now remove the
types.h header in libcfs.

Change-Id: I59a4dfbf546db72154d2cb461ee25517c6bd6015
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: http://review.whamcloud.com/22732
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: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4931 ofd: use thread buffer for ladvise

A buffer is allocated everytime when handling willread advice.
This is not efficient, so replace it by using thread buffer
instead.

Signed-off-by: Li Xi <lixi@ddn.com>
Change-Id: I34015ea0658cb116b3348f8cab67128645718cac
Reviewed-on: http://review.whamcloud.com/22489
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>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8209 ldlm: engage ELC for all ldlm enqueue req

If there is no request passed into ldlm_cli_enqueue, the enqueue
request will not engage ELC to drop unneeded locks. currently,
this kind of request is mainly related to EXTENT locks enqueue
requests (except for glimpse EXTENT lock for it has an intent).

Change-Id: Ibb34117cbec6764f643457d9b0af142393263608
Signed-off-by: Hongchao Zhang <hongchao.zhang@intel.com>
Reviewed-on: http://review.whamcloud.com/21739
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Vitaly Fertman <vitaly.fertman@seagate.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7919 mount: Buffer overflow issue while parsing mount

check is added in parse_options, parse_ldd, add_mgsnids,
parse_ldd and main function to avoid buffer overflow
while parsing mount options.

Seagate-bug-id: MRP-2384
Change-Id: I1af9db4221ccdf1fd64a9a565063e5948cd5ba16
Signed-off-by: Lokesh Nagappa Jaliminche <lokesh.jaliminche@seagate.com>
Reviewed-on: http://review.whamcloud.com/19158
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Frank Zago <fzago@cray.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8630 build: Fix definition of KMP_MODDIR

Without KMP_MODDIR defined, it won't be correctly replaced in
dkms.mkconf, thus causing dkms install of lustre to fail

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: Ie671ede26b6cdf28aebeb80e131ff8506b9c05df
Reviewed-on: http://review.whamcloud.com/22663
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8519 build: make SLES use lbuild kernel-devel

SLES %kernel_module_package assume kernel-devel installed
in standard location /usr/src/. However, lbuild extracts
kernel-devel into its location; so we need to change
%kernel_module_package macros to use lbuild's kernel obj

Create a cleanup_rpmmacros to remove the change afterward

Test-parameters: trivial

Change-Id: I8f1d8ac50436455be23e15b0a277afb6c7def7c3
Signed-off-by: Minh Diep <minh.diep@intel.com>
Reviewed-on: http://review.whamcloud.com/22484
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-1882 llite: Adding timed wait in ll_umount_begin

There exists timing race between umount and other
thread which will increment the reference count on
mnt e.g. getattr. If umount thread lose the race
then umount fails with EBUSY error. To avoid this
timed wait is added so that umount thread will wait
for user to decrement the mnt reference count.

This patch also fixes below two problems
1.the fail_loc code in conf-sanity test 45 to
OR with the OBD_FAIL_ONCE (0x80000000) to prevent
the fail_loc from looping forever.

2.fixes the fail_loc code name from
OBD_FAIL_PTLRPC_LONG_UNLINK to
OBD_FAIL_PTLRPC_LONG_REPL_UNLINK.

Signed-off-by: Rahul Deshmukh <rahul.deshmukh@seagate.com>
Signed-off-by: Lokesh Nagappa Jaliminche <lokesh.jaliminche@seagate.com>
Signed-off-by: Jian Yu <jian.yu@intel.com>
Change-Id: Icab9b560cadcb8623c8592bfc4c5a842277ad266
Seagate-bug-id: MRP-1192
Reviewed-on: http://review.whamcloud.com/20061
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: Oleg Drokin <oleg.drokin@intel.com>

LU-8654 obd: access ocd_connect_flags2 only when present

Before 2.7.50 clients will send a struct obd_connect_data_v1 rather
than a full struct obd_connect_data. So check for OBD_CONNECT_FLAGS2
before accessing the ocd_connect_flags2 member of struct
obd_connect_data.

Signed-off-by: John L. Hammond <john.hammond@intel.com>
Change-Id: Ic1c2b54be9b3cbaec9af894f204663c74fa41213
Reviewed-on: http://review.whamcloud.com/22837
Tested-by: Jenkins
Reviewed-by: Fan Yong <fan.yong@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-8527 osd: ot_credits must be 32bit

ot_credits used to calculate journal credits must be 32bit,
otherwise it can overflow by huge transactions (e.g. 1K striped dirs)

Change-Id: I53545989fd3653e72b5227281fc2ec6b2ccd309d
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/22782
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

New tag 2.8.59

Change-Id: Id6d5c22072d86addabc6a968f1c8d986970ba81e
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4931 doc: update ladvise man page

Add documentation for the willread and dontneed advice types.

Test-Parameters: trivial
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I624aea124322e3cd105e9b8aa7d40106a1301e56
Reviewed-on: http://review.whamcloud.com/22910
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3289 gss: don't build SSK if libssl-1.0+ not available

The SSK functionality depends on libssl 1.0 or newer to have the
proper HMAC support.  If that is not available (e.g. SLES11) then
don't try to build this feature at all.

Rename the configure check to be OPENSSL_SSK since this is used in
several places, and is more clear than HAVE_INT_OPENSSL_HMAC_FUNCS.

Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I3b15f819bba421539664e629a4017599e23ebbe5
Reviewed-on: http://review.whamcloud.com/22806
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-8327 ldiskfs: release bh in make_indexed_dir

RHEL7.2 ext4 code leak to release bh in make_indexed_dir.
It was cause by upstream 6050d47adcadbb53582434d919ed7f038d936712

Signed-off-by: Yang Sheng <yang.sheng@intel.com>
Change-Id: Ie27225cfbfb2e43f1ee48799a9836f873fefacc9
Reviewed-on: http://review.whamcloud.com/22738
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8561 osd-zfs: make smatch happy

By removing the check of null pointer and call dmu_objset_disown()
directly.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: I3e6e684968946a923ec4b1a59ea881145f3bdede
Reviewed-on: http://review.whamcloud.com/22245
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: Oleg Drokin <oleg.drokin@intel.com>

LU-8509 llite: drop_caches hangs in cl_inode_fini()

This patch releases cl_pages on error in ll_write_begin()
to avoid memory and object reference leaks. Also, it
reuses per-cpu lu_env in ll_invalidatepage() in the same
way as done in ll_releasepage().

Change-Id: I11a658941aff4dacd3cc83685ae3df67f28093a2
Signed-off-by: Andrew Perepechko <andrew.perepechko@seagate.com>
Seagate-bug-id: MRP-3504
Reviewed-on: http://review.whamcloud.com/22745
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8637 obdclass: LWP callback hold export reference

The lwp_notify_main thread itself needs to hold the 'exp'
reference to prevent 'exp' being freed during LWP callback.

On the other hand, there is race condition between the
lustre_register_lwp_item() and lustre_notify_lwp_list():

When lustre_register_lwp_item() (call it as thread1) adds
'lwp_register_item' into the 'lwp_register_list', the
"*lri_exp" is NULL. After that, the thread1 releases the
'lwp_register_list_lock', then the LWP notify thread (call
it as thread2) assigns the "lri::lri_exp" via the
'lustre_notify_lwp_list()' by race. That will cause thread1
to trigger the 'lri::lri_cb_func'. At the same time, thread2
will also trigger the 'lri::lri_cb_func'. Then the callback
are called repeatedly.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I9eb5407931149a818b620a393fdc70fec0d0d099
Reviewed-on: http://review.whamcloud.com/22724
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: wangdi <di.wang@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8612 mdt: maintain Sync-on-Lock-Cancel locks per target

Sync-on-Lock-Cancel locks should be maintained per target, so that
if there are more than one target(MDT) on one server, transaction
commit on different target won't interfere with each other.

Signed-off-by: Lai Siyao <lai.siyao@intel.com>
Change-Id: Ib2834c64089f18fb9d2e06d64f770ddaf9c2e2d3
Reviewed-on: http://review.whamcloud.com/22490
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-4185 llite: Revise create with no open optimization

Currently ll_lookup_nd just returns a negative when we are trying
to create something with no open (read mkdir). This is all fine most
of the cases, except if the directory where we are trying to do is
not writeable by us. In that case vfs_create would return EPERM
seeing as how a negative dentry means the create cannot proceed.
But in reality if there is an existing name there that we just did
not have cached, the proper return is EEXIST that could only happen
if we did do the lookup.

So amend the optimization to only take place if the directory is
writeable by us, otherwise do the full lookup.

Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Bobi Jam <bobijam.xu@intel.com>
Change-Id: I1480320bd56431a2e47ae686450678324ca17633
Reviewed-on: http://review.whamcloud.com/8257
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>

LU-8340 tests: in sanity-sec 25 client nodemap must be admin

If the ENABLE_QUOTA is set, then cleanup_and_setup_lustre will use
the client node to setup quota, which requires admin privileges on
the nodemap. This patch enables admin privileges on all clients
before calling cleanup_and_setup_lustre.

Test-Parameters: envdefinitions=ENABLE_QUOTA=yes
Signed-off-by: Kit Westneat <kit.westneat@gmail.com>
Change-Id: If8d92d20aaa8cf75e3d443d1a81ff2f201d118cc
Reviewed-on: http://review.whamcloud.com/22750
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Jian Yu <jian.yu@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-5718 lnet: add offset for selftest brw

In current lnet selftest, both client and server side bulk have
no offset and we can only test page aligned IO, this patch changed
this:

- user can set brw offset by lst add_test ... brw off=OFFSET ...
- offset is only effective on client side so far
- to simply implementation, offset needs to be eight bytes aligned

Signed-off-by: Liang Zhen <liang.zhen@intel.com>
Change-Id: I34690d2c8fbb627b827291afcd46d7aa59831a3f
Reviewed-on: http://review.whamcloud.com/12496
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8413 llite: specify READA debug mask for sanity_101f

So that debug log only contains relevant messages for debugging
purpose.

Test-Parameters: trivial
Signed-off-by: Bobi Jam <bobijam.xu@intel.com>
Change-Id: Ia02e6ed41a3a2166dcf6b04441027517594bb35d
Reviewed-on: http://review.whamcloud.com/22753
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8632 osc: remove of usage removed 'page_count' local variable

Address of function 'page_count' is used instead of a local variable
which was removed in commit 7f2aae8d80a73de7408668bbe569d5f4d8553efe.

Test-Parameters: trivial

Change-Id: I92626b27b21c291101299b52b46287262d27a4bd
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-on: http://review.whamcloud.com/22654
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>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>

LU-8579 osd-ldiskfs: code cleanup for osd_ldiskfs_add_entry

Commit 07660ad33a7d1 can give more warning message when directory
size exceeds limitation. But some variables that are allocated
on stack are unnecessary and may cause overflow.

Further more, the "osd_thread_info::oti_dev" is obsolete, should
not be used any more. The fid_seq_is_mdt0() is used against zero
FID, will be replaced by checking whether the device is for MDT0.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: Idfe2a4546535b3863e84ce5753b8a86c369dadae
Reviewed-on: http://review.whamcloud.com/22307
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6401 header: remove assert from interval_set()

Currently interval_tree.h is used by user land and
kernel space. To unwind libcfs.h from user space
we need to remove LASSERT from the uapi headers.
In the case of interval_tree.h only interval_set()
uses LASSERT which is removed in this patch and
interval_set() instead reports a real error.

Change-Id: Iac716f61542eb6e1a6e060e2810e6b7a175b25d9
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: http://review.whamcloud.com/22522
Tested-by: Jenkins
Reviewed-by: Frank Zago <fzago@cray.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-3888 utils: print lmm_fid as part of getstripe

Print the lmm_fid in the standard DFID format in addition to the older
lmm_object_id and lmm_seq fields when using "lfs getstripe -v".  This
is still a bit of a hack, since lmm_mds_md_{v1,v3} do not store a
proper FID in the header, so the fields need to be parsed separately
instead of using lmm_oi_seq() and lmm_oi_id(), but at least userspace
can start using this output in scripts instead of needing to call
path2fid separately.

In the future, new layout formats should use proper FIDs instead of
ost_id fields.

Add "lfs getstripe -F" to print the lmm_fid of a file.  This should
match the inode FID from "lfs fid2path" for 2.x-created files.

Document "lfs getstripe -g" option in lfs.1 and usage message.

Fix lfs_fid2path() to not complain on the MDS if an invalid FID is
requested, since an error is returned to the application anyway.
Don't print errors in llapi_fid2path(), there is already one being
printed in lfs_fid2path().

Rename VERBOSE_ALL to VERBOSE_DEFAULT to make it more clear what this
is for, since it didn't actually print all of the components.

Add PFL VERBOSE_COMP_* flags to lustreapi.h to avoid future conflicts.

Test-Parameters: trivial
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Lokesh Nagappa Jaliminche <lokesh.jaliminche@seagate.com>
Change-Id: I4e27797ec3905327ea0a88911b1c75193d500c1e
Reviewed-on: http://review.whamcloud.com/10465
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>

LU-4474 osd: Add nodelalloc to ldiskfs mount options

If vfs_mount failed mount, process hangs during osd_mount cleanup.
Mutex sb->s_umount cause of this hang. During cleanup
ldiskfs_quota_off() is executed and if delalloc option is set,
then sync_filesystem executed, but only after taking of s_umount
mutex. We can't exlude this mutex here or move to another place,
because there are some code pathes there ldiskfs_quota_off
executed without s_umount mutex.

This patch disable the delalloc option.

Signed-off-by: Artem Blagodarenko <artem.blagodarenko@seagate.com>
Seagate-bug-id: MRP-3225
Change-Id: Ib68d1c424f4a99eb7df5c373e9775aeafe8505b4
Reviewed-on: http://review.whamcloud.com/21181
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Faccini Bruno <bruno.faccini@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>