LU-8066 libcfs: test for both __kernel_param_[un]lock and kernel_param_[un]lock

In earlier kernels like RHEL6 no locking is available. Later the
function __kernel_param_[un]lock() we introduced. In most recent
kernels per module locking was introduced with the functions
kernel_param_[un]lock() and __kernel_param_[un]lock() is no longer
visible to modules. Since this is the case we need to make sure
both HAVE_MODULE_PARAM_LOCKING and HAVE_KERNEL_PARAM_LOCK are not
set in the case of RHEL6.

Change-Id: I0957a16352c4fb49fb5d96c0ff4d331a8be9703a
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/28498
Tested-by: Jenkins
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9860 tests: Add conf-sanity tests to ALWAYS_EXCEPT list

The following tests fail when run with a separate MDS and MGS:
conf-sanity tests 33a, 43b, 53b, 54b, 70e, 80, 84, 87, 100,
102, 103, 104, 105 and 107.
We need to add these tests to the ALWAYS_EXCEPT list
when running with a separate MDS and MGS.

Test-Parameters: trivial combinedmdsmgs=false testlist=conf-sanity envdefinitions=SLOW=yes
Signed-off-by: James Nunez <james.a.nunez@intel.com>
Change-Id: I1b17714216e14ad04eb9a492cb5f1aa4ed82bd1a
Reviewed-on: https://review.whamcloud.com/28497
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Wei Liu <wei3.liu@intel.com>
Reviewed-by: Dilip Krishnagiri <dilipx.krishnagiri@intel.com>
Reviewed-by: Saurabh Tandan <saurabh.tandan@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9869 lnet: fix incorrect arguments order calling lstcon_session_new

The arguments args->lstio_ses_force and args->lstio_ses_timeout are
in the incorrect order. Fix this by swapping them around.

Detected by CoverityScan, CID#1226833 ("Arguments in wrong order")

Test-Parameters: trivial testlist=lnet-selftest

Change-Id: If11c574655425db5bbf21ba2264be8d83a7e8bf8
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/28487
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Sonia Sharma <sonia.sharma@intel.com>
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6210 ptlrpc: Use C99 initializer in ptlrpc_register_rqbd()

This patch makes no functional changes.  The struct initializer in
ptlrpc_register_rqbd() is updated to C99 syntax.

C89 positional initializers require values to be placed in the
correct order. This will cause errors if the fields of the struct
definition are reordered or fields are added or removed. C99 named
initializers avoid this problem, and also automatically clear any
values that are not explicitly set.

Test-Parameters: trivial
Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: I7c24bac3ba6be6732b206406cd74b0d4f8a1f9c2
Reviewed-on: https://review.whamcloud.com/28479
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: Oleg Drokin <oleg.drokin@intel.com>

LU-9856 mdd: handle NULL buffer in mdd_xattr_list()

The upper layer may call mdd_xattr_list() with a NULL buffer to get
the length of the xattr name list. Handle this case safely by skipping
the removal of the link xattr for unlinked objects.

Signed-off-by: John L. Hammond <john.hammond@intel.com>
Change-Id: Iae87fba20325b228ef75ee762acfa49353932b1b
Reviewed-on: https://review.whamcloud.com/28469
Tested-by: Jenkins
Reviewed-by: Andrew Perepechko <andrew.perepechko@seagate.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6210 utils: Use C99 struct initializers in lfs_getdirstripe

This patch makes no functional changes.  The option struct
initializer in lfs_getdirstripe() is updated to C99 syntax.

C89 positional initializers require values to be placed in the
correct order. This will cause errors if the fields of the struct
definition are reordered or fields are added or removed. C99 named
initializers avoid this problem, and also automatically clear any
values that are not explicitly set.

Test-Parameters: trivial
Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: I6f2d4a82e5a9ef2c76946746d6c46b1202e8c278
Reviewed-on: https://review.whamcloud.com/28421
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: Andreas Dilger <andreas.dilger@intel.com>

LU-832 test: Add error check when running run-llog.sh

Add error status check in sanity test_60a when calling
run-llog.sh

Test-Parameters: trivial

Change-Id: I1296907c8892b7dd54dac37045d8a7c4e03b1f52
Signed-off-by: Wei Liu <wei3.liu@intel.com>
Reviewed-on: https://review.whamcloud.com/28412
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9803 tests: cast st_blksize for printf

Compilation with -Werror=format complains about this printf. Expects
unsigned long but st_blksize has type __blksize_t. Cast it to unsigned
long for printf

Test-Parameters: trivial
Signed-off-by: Chris Horn <hornc@cray.com>
Change-Id: I1eeb5613e485132de8f0bce08bd4d89887e52cf6
Reviewed-on: https://review.whamcloud.com/28262
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-9781 llog: Improve catalog full warning

When warning that a catalog file is full, provide the name
of the catalog file. If the name of catalog file isn't
defined, print its FID.

Signed-off-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Change-Id: I559e43d08febfd8a1512ceb58fd3030b06372e9f
Reviewed-on: https://review.whamcloud.com/28093
Reviewed-by: Faccini Bruno <bruno.faccini@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Olaf Faaland-LLNL <faaland1@llnl.gov>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6210 utils: Use C99 initializers in lfs_changelog()

This patch makes no functional changes.  Struct initializers that
use C89 or GCC-only syntax are updated to C99 syntax.  Variables of
type struct option are renamed to long_opts for consistency.

C89 positional initializers require values to be placed in the
correct order. This will cause errors if the fields of the struct
definition are reordered or fields are added or removed. C99 named
initializers avoid this problem, and also automatically clear any
values that are not explicitly set.

This patch updates lfs_changelog() to use the C99 syntax.

Test-Parameters: trivial
Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: I4f9d82974f68742d788f00d58c5e3d61449fc5bb
Reviewed-on: https://review.whamcloud.com/27522
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>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>

LU-9593 tests: remove sanity-sec tests from ALWAYS_EXCEPT

sanity-sec tests 2, 5 and 6 no longer exist. Test 2 was
removed by LU-6971 patch change ID I06f4348b. Tests
5 and 6 were removed by LU-3105 patch change I865a92b57.

Remove sanity-sec tests 2, 5 and 6 from the ALWAYS_EXCEPT
list.

Test-Parameters: trivial testlist=sanity-sec

Signed-off-by: dilip krishnagiri <dilipx.krishnagiri@intel.com>
Change-Id: Ia0377ff0da41c4ba9df6c90bc26f0469cb9de9a6
Reviewed-on: https://review.whamcloud.com/27411
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Chris Hanna <hannac@iu.edu>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9591 tests: remove replay-vbr tests 12a from ALWAYS_EXCEPT

Removing replay-vbr test 12a - lock replay with VBR from
the ALWAYS_EXCEPT list. It is associated with bugzilla
ticket 16356 which is in NEW state.
This test did not run for years.

Test-Parameters: trivial testlist=replay-vbr

Signed-off-by: dilip krishnagiri <dilipx.krishnagiri@intel.com>
Change-Id: I251bbaeea744a11fdf3e34870a00fc6b53fae3b1
Reviewed-on: https://review.whamcloud.com/27406
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Saurabh Tandan <saurabh.tandan@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8276 ldlm: Make lru clear always discard read lock pages

A significant amount of time is sometimes spent during
lru clearing (IE, echo 'clear' > lru_size) checking
pages to see if they are covered by another read lock.
Since all unused read locks will be destroyed by this
operation, the pages will be freed momentarily anyway,
and this is a waste of time.

This patch sets the LDLM_FL_DISCARD_DATA flag on all the PR
locks which are slated for cancellation by
ldlm_prepare_lru_list when it is called from
ldlm_ns_drop_cache.

The case where another lock covers those pages (and is in
use and so does not get cancelled by lru clear) is safe for
a few reasons:

1. When discarding pages, we wait (discard_cb->cl_page_own)
until they are in the cached state before invalidating.
So if they are actively in use, we'll wait until that use
is done.

2. Removal of pages under a read lock is something that can
happen due to memory pressure, since these are VFS cache
pages. If a client reads something which is then removed
from the cache and goes to read it again, this will simply
generate a new read request.

This has a performance cost for that reader, but if anyone
is clearing the ldlm lru while actively doing I/O in that
namespace, then they cannot expect good performance.

In the case of many read locks on a single resource, this
improves cleanup time dramatically.  In internal testing at
Cray with ~80,000 read locks on a single file, this improves
cleanup time from ~60 seconds to ~0.5 seconds.  This also
slightly improves cleanup speed in the case of 1 or a few
read locks on a file.

Signed-off-by: Patrick Farrell <paf@cray.com>
Change-Id: I0c076b31ea474bb5f012373ed2033de3e447b62d
Reviewed-on: https://review.whamcloud.com/20785
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-5541 lustreapi: only export the API symbols

By default, all kind of symbols are exported from the library (dump,
libcfs_ukuc_start, l_ioctl, set_ioctl_dump, ...), which may create
external conflicts. Use the linker version-script options to only
export the API symbols, and prevent the export of internal symbols.

Only the symbols declared in the global section of liblustreapi.map
will be seen by applications.

Fix lshowmount to use libcfs and not internal liblustreapi symbol.

Change-Id: Ica4226c1ea9b6b159a056ad22bacaa2ffcf4b171
Signed-off-by: frank zago <fzago@cray.com>
Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-on: https://review.whamcloud.com/11643
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>
Reviewed-by: Henri Doreau <henri.doreau@cea.fr>

LU-5541 build: build static and dynamic liblustreapi

libtool knows how to build both, so no need to hack the Makefile. As
two added benefits, the utilities will now use the dynamic version,
thus reducing their footprint, and calling make twice in a row won't
rebuild objects already built.

Test-Parameters: trivial

Change-Id: If4191e1ff1564793c476ffe03f5d4b6ad5295421
Signed-off-by: frank zago <fzago@cray.com>
Reviewed-on: https://review.whamcloud.com/11625
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Henri Doreau <henri.doreau@cea.fr>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9848 llog: check padding size for update reclen

Update log only checks padding size for split case, which should also
be done if it's less than chunk size.

Signed-off-by: Lai Siyao <lai.siyao@intel.com>
Change-Id: Ie7819f67dd9bcbfb060713bb208c9777420c5178
Reviewed-on: https://review.whamcloud.com/28554
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: wangdi <di.wang@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9828 ptlrpc: Do not assert when bd_nob_transferred != 0

There is a case in the routine ptlrpc_register_bulk() where we were
asserting if bd_nob_transferred != 0 when not resending.  There is
evidence that network errors can create a situation where
this does happen.  So we should not be asserting!

This patch changes that assert to an error return code of -EIO.

Signed-off-by: Doug Oucharek <doug.s.oucharek@intel.com>
Change-Id: I6a73ca1b04a86f187744d3b8b5d46df71d95e416
Reviewed-on: https://review.whamcloud.com/28491
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Sonia Sharma <sonia.sharma@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9863 lmv: Off by two in lmv_fid2path()

We want to concatonate join string one, a '/' character, string two and
then a NUL terminator. The destination buffer holds ori_gf->gf_pathlen
characters. The strlen() function returns the number of characters not
counting the NUL terminator. So we should be adding two extra spaces,
one for the foward slash and one for the NULL.

Change-Id: Ia96461a2d1b3331f44d3791ca0148f6e836caf0d
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/28477
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Frank Zago <fzago@cray.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9019 mdd: migrate from jiffies64 to ktime

The mdd layer uses cfs_time_xxx_64() for 64 bit time percision.
This was written before ktime_t came into existence and it uses
64 bit version of jiffies which can vary between nodes due to
HZ being configurable. This provides a consistent format with
nanosecond precision on any node.

Change-Id: Ibec17227fd70a148c83296e8d1c41668f67e9201
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/28407
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@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-9657 pfl: llapi_layout_comp_usei should handle non-pfl file

This patch improves llapi_layout_comp_use() to treat non-composite
file as single component file. When doing "is composite" check,
"1" is returned when LLAPI_LAYOUT_COMP_USE_NEXT/PREV is specified.

Signed-off-by: Emoly Liu <emoly.liu@intel.com>
Change-Id: I3ba4f07ec843d9b61273af331060d5f8827c2f8b
Reviewed-on: https://review.whamcloud.com/27865
Tested-by: Jenkins
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>

LU-8993 utils: Use absolute pathname for debug_daemon log file

The lctl debug_daemon command is changed to always provide an
absolute pathname to the kernel.  The kernel code will return EINVAL
if the pathname does not begin with '/', leading to the confusing
error "Invalid argument". This patch allows the user to provide a
relative pathname to the command without generating this error.

The absolute_path function has been moved to string.c and renamed to
cfs_abs_path, so that it may be used by all utilities.

Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: I35af242bfcfcb9a56135aeabe0423e28e9634bab
Reviewed-on: https://review.whamcloud.com/25485
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9410 ldiskfs: no check mb bitmap if flex_bg enabled

When initializes (reformat) the filesystem, the number of
free blocks in the group descriptor is calculated via the
ext2fs_reserve_super_and_bgd() (e2fsprogs). As commented
in such function: "This is not necessarily the case when
the flex_bg feature is enabled, so callers should take care!".

So it is normal that we may find the block group descriptor
that has LDISKFS_BG_BLOCK_UNINIT flag but with 0 free blocks.
The ldiskfs_mb_check_ondisk_bitmap() should NOT report error
for such block group, instead, skip the check directly.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: Iba0fb2bf0632a6e54222472bc724a8ea0478e9ae
Reviewed-on: https://review.whamcloud.com/28566
Tested-by: Jenkins
Reviewed-by: Andreas Dilger <andreas.dilger@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-9841 lov: do not split IO for single striped file

stripe size for single striped file is not reliable, it shouldn't
be used to split I/O.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: I47c31d59b46b07d4a6760b8985e1c19da4765a5c
Reviewed-on: https://review.whamcloud.com/28451
Tested-by: Jenkins
Reviewed-by: Bobi Jam <bobijam@hotmail.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-9842 osd: return ENODATA for XATTR_NAME_FID on MDT

The XATTR_NAME_FID xattr is OST side EA, if someone calls
getxattr() for XATTR_NAME_FID on MDT, then return -ENODATA.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I18b1466cf62d10fa28f7ed9731490e963b6274f4
Reviewed-on: https://review.whamcloud.com/28434
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@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-9767 utils: validate filesystem name for mkfs.lustre

The patch "LU-6401 uapi: turn lustre_param.h into a proper
UAPI header" removed various user land functions used to
validate poolnames and file system names were removed. The
checks instead were enforced on the kernel side to ensure
any possible user land software directly interfacing to the
kernel wouldn't be able to break things badly. For the case
of formating the backend file system no kernel interaction
doesn't happen until it tries to mount the MDT/OST/MGT which
is very late in the process. So for this case lets add back
the file system name verification to mkfs.lustre to warn
users long before they try to mount anything.

Secondly we remove the verify_poolname() in lfs.c since
it duplicates extract_fsname_poolname() in obd.c. Their
is no need to do the same test twice. The function
pool_cmd() calls the ioctl for pool handling which in
turn returns an error code. Use this error code to notify
the user what mistake they did for their pool command.
For the MGS kernel code mgs_extract_fs_pool() was checking
MTI_NAME_MAXLEN instead of LUSTRE_MAXFSNAME. Also use
LUSTRE_MAXFSNAME instead of the raw number in the function
server_name2fsname() located in obd_mount.c.

Change-Id: If094644e56a70b6dd8e6b0378adc8736911aeef1
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/28070
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9913 lnet: balance references in lnet_discover_peer_locked()

In lnet_discover_peer_locked() avoid a leaked reference to the peer in
the non-blocking discovery case.

Signed-off-by: John L. Hammond <john.hammond@intel.com>
Change-Id: Ic48414859c923af1ebb197b0b0f2f8d6752043ac
Reviewed-on: https://review.whamcloud.com/28695
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Quentin Bouget <quentin.bouget@cea.fr>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9480 lnet: Multi-Rail Dynamic Discovery feature

Merge remote-tracking branch 'origin/multi-rail'

Change-Id: I63d21d1085f4bf665480d29d5d14c065b6a22191
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9480 lnet: cleanup lnetctl and cyaml

lnetctl set commands results in segmentation fault
if no values are provided. This patch makes them
show help if no values are provided to with set commands.

Made general cleanups in the lnetctl code to consolidate
where the help is being printed. Created a function
check_cmd() which checks for the expected number of
arguments and for the -h/--help option and prints
the help string if either scenario is encountered.

fixed the fsm transition in cyaml to allow proper
parsing of empty cyaml documents

Change-Id: Ia081e9304ba2d6baa804e4c8890fb1988d860c1c
Test-Parameters: trivial
Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Signed-off-by: Sonia Sharma <sonia.sharma@intel.com>
Reviewed-on: https://review.whamcloud.com/27349
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>

LU-9480 lnet: show peer state

It is important to show the peer state when debugging.
This patch exports the peer state from the kernel to
user space, and is shown when the detail level requested
in the peer show command is >= 3

Test-Parameters: trivial
Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I1e169b2b7bf80671ea302f04c6fb948bbcbbb245
Reviewed-on: https://review.whamcloud.com/26130
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>

LU-9480 lnet: add enhanced statistics

Added statistics to track the different types of
LNet messages which are sent/received/dropped

Test-Parameters: trivial
Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I7e1fc991a56df20181f9e55a794765349a4d2cb9
Reviewed-on: https://review.whamcloud.com/25795

LU-9480 lnet: add "lnetctl discover"

Add a "discover" subcommand to lnetctl

jt_discover() in lnetctl.c calls lustre_lnet_discover_nid()
to implement "lnetctl discover". The output is similar to
"lnetctl ping" command.
This patch also does some clean up in linlnetconfig.c
For parameters under global settings, the common code
for them is pulled in funtions ioctl_set_value() and
ioctl_show_global_values().

Test-Parameters: trivial
Signed-off-by: Sonia Sharma <sonia.sharma@intel.com>
Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I98ebb0b27de4b32ea07421f7dd71a4a1c96f3e05
Reviewed-on: https://review.whamcloud.com/25793

LU-9077 lnet: fix for static analysis issues

fixes the 11 static analysis issues found in
v2_9_52_0-66-gec839d4.

1. lustre_lnet_show_numa_range - fixed
2. lnet_select_pathway - fixed
3. lustre_lnet_show_discovery - fixed
4. lnet_discover_peer_locked - false positive
5. lustre_lnet_ping_nid - fixed
6. lustre_lnet_ping_nid - false positive
7. lustre_lnet_show_discovery - duplicate of 3
8. lustre_lnet_show_max_intf - fixed
9. lustre_lnet_show_max_intf - duplicate of 8
10. lnet_peer_set_primary_data - false positive
11. lustre_lnet_show_numa_range - fixed

Test-Parameters: trivial
Signed-off-by: Sonia Sharma <sonia.sharma@intel.com>
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I4cb03e4f64cd0c743ee3646f4628d34533b2d4ba
Reviewed-on: https://review.whamcloud.com/25792
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: add "lnetctl ping" command

Adds function jt_ping() in lnetctl.c and
lustre_lnet_ping_nid() in liblnetconfig.c file.
The output of "lnetctl ping" is similar to
"lnetctl peer show".

Function jt_ping() in lnetctl.c calls lustre_lnet_ping_nid()
to implement "lnetctl ping". Adds a function infra_ping_nid()
to be later reused for the ping similar lnetctl commands.
Uses a new ioctl call, IOC_LIBCFS_PING_PEER for "lnetctl ping".
With "lnetctl ping", multiple nids can be pinged. Uses a new
struct(lnet_ioctl_ping_data in lib-dlc.h) to pass the data
from kernel to user space for ping. Also changes lnet_ping()
function and its input parameters in lnet/lnet/api-ni.c

Test-Parameters: trivial
Signed-off-by: Sonia Sharma <sonia.sharma@intel.com>
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I67024d87fa5cca6aa7ff7a8099d4400a795f3a83
Reviewed-on: https://review.whamcloud.com/25791
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: add "lnetctl peer list"

Add IOC_LIBCFS_GET_PEER_LIST to obtain a list of the primary
NIDs of all peers known to the system. The list is written
into a userspace buffer by the kernel. The typical usage is
to make a first call to determine the required buffer size,
then a second call to obtain the list.

Extend the "lnetctl peer" set of commands with a "list"
subcommand that uses this interface.

Modify the IOC_LIBCFS_GET_PEER_NI ioctl (which is new in the
Multi-Rail code) to use a NID to indicate the peer to look
up, and then pass out the data for all NIDs of that peer.

Re-implement "lnetctl peer show" to obtain the list of NIDs
using IOC_LIBCFS_GET_PEER_LIST followed by one or more
IOC_LIBCFS_GET_PEER_NI calls to get information for each
peer.

Make sure to copy the structure from kernel space to
user space even if the ioctl handler returns an error.
This is needed because if the buffer passed in by the
user space is not big enough to copy the data, we want
to pass the requested size to user space in the structure
passed in. The return code in this case is -E2BIG.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I522c11e6ec09bec46121496d526bb258e10295f1
Reviewed-on: https://review.whamcloud.com/25790
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: implement Peer Discovery

Implement Peer Discovery.

A peer is queued for discovery by lnet_peer_queue_for_discovery().
This set LNET_PEER_DISCOVERING, to indicate that discovery is in
progress.

The discovery thread lnet_peer_discovery() checks the peer and
updates its state as appropriate.

If LNET_PEER_DATA_PRESENT is set, then a valid Push message or
Ping reply has been received. The peer is updated in accordance
with the data, and LNET_PEER_NIDS_UPTODATE is set.

If LNET_PEER_PING_FAILED is set, then an attempt to send a Ping
message failed, and peer state is updated accordingly. The discovery
thread can do some cleanup like unlinking an MD that cannot be done
from the message event handler.

If LNET_PEER_PUSH_FAILED is set, then an attempt to send a Push
message failed, and peer state is updated accordingly. The discovery
thread can do some cleanup like unlinking an MD that cannot be done
from the message event handler.

If LNET_PEER_PING_REQUIRED is set, we must Ping the peer in order to
correctly update our knowledge of it. This is set, for example, if
we receive a Push message for a peer, but cannot handle it because
the Push target was too small. In such a case we know that the
state of the peer is incorrect, but need to do extra work to obtain
the required information.

If discovery is not enabled, then the discovery process stops here
and the peer is marked with LNET_PEER_UNDISCOVERED. This tells the
discovery process that it doesn't need to revisit the peer while
discovery remains disabled.

If LNET_PEER_NIDS_UPTODATE is not set, then we have reason to think
the lnet_peer is not up to date, and will Ping it.

The peer needs a Push if it is multi-rail and the ping buffer
sequence number for this node is newer than the sequence number it
has acknowledged receiving by sending an Ack of a Push.

If none of the above is true, then discovery has completed its work
on the peer.

Discovery signals that it is done with a peer by clearing the
LNET_PEER_DISCOVERING flag, and setting LNET_PEER_DISCOVERED or
LNET_PEER_UNDISCOVERED as appropriate. It then dequeues the peer
and clears the LNET_PEER_QUEUED flag.

When the local node is discovered via the loopback network, the
peer structure that is created will have an lnet_peer_ni for the
local loopback interface. Subsequent traffic from this node to
itself will use the loopback net.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I30acd1e046604013025b231b5806be25468a2286
Reviewed-on: https://review.whamcloud.com/25789
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: add the Push target

Peer Discovery will send a Push message (same format as an
LNet Ping) to Multi-Rail capable peers to give the peer the
list of local interfaces.

Set up a target buffer for these pushes in the_lnet. The
size of this buffer defaults to LNET_MIN_INTERFACES, but it
is resized if required.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I09b5ad8ae504ba8368d908539001fb8afc2c2778
Reviewed-on: https://review.whamcloud.com/25788
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: tune lnet_peer_discovery_disabled with lnetctl

A new tunable, lnet_peer_discovery_disabled, has been introduced.
Make it tunable with lnetctl. Note that the state of discovery is
reported as 1/enabled, or 0/disabled, which is the inverse of the
module parameter.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Change-Id: I67333d86520c5b6db8ff99c924054c4b487c8029
Reviewed-on: https://review.whamcloud.com/25787
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>

LU-9480 lnet: add discovery thread

Add the discovery thread, which will be used to handle peer
discovery. This change adds the thread and the infrastructure
that starts and stops it. The thread itself does trivial work.

Peer Discovery gets its own event queue (ln_dc_eqh), a queue
for peers that are to be discovered (ln_dc_request), a queue
for peers waiting for an event (ln_dc_working), a wait queue
head so the thread can sleep (ln_dc_waitq), and start/stop
state (ln_dc_state).

Peer discovery is started from lnet_select_pathway(), for
GET and PUT messages not sent to the LNET_RESERVED_PORTAL.
This criterion means that discovery will not be triggered by
the messages used in discovery, and neither will an LNet ping
trigger it.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Change-Id: I38a48ab7f61c8ef1b994cd17069729f243912bdf
Reviewed-on: https://review.whamcloud.com/25786
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>

LU-9480 lnet: add msg_type to lnet_event

Add a msg_type field to the lnet_event structure. This makes
it possible for an event handler to tell whether LNET_EVENT_SEND
corresponds to a GET or a PUT message.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: If9ecc42c26eb078c19697f399a17f80b2e225639
Reviewed-on: https://review.whamcloud.com/25785
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: reference counts on lnet_peer/lnet_peer_net

Peer discovery will be keeping track of lnet_peer structures,
so there will be references to an lnet_peer independent of
the references implied by lnet_peer_ni structures. Manage
this by adding explicit reference counts to lnet_peer_net and
lnet_peer.

Each lnet_peer_net has a hold on the lnet_peer it links to
with its lpn_peer pointer. This hold is only removed when that
pointer is assigned a new value or the lnet_peer_net is freed.
Just removing an lnet_peer_net from the lp_peer_nets list does
not release this hold, it just prevents new lookups of the
lnet_peer_net via the lnet_peer.

Each lnet_peer_ni has a hold on the lnet_peer_net it links to
with its lpni_peer_net pointer. This hold is only removed when
that pointer is assigned a new value or the lnet_peer_ni is
freed. Just removing an lnet_peer_ni from the lpn_peer_nis
list does not release this hold, it just prevents new lookups
of the lnet_peer_ni via the lnet_peer_net.

This ensures that given a lnet_peer_ni *lpni, we can rely on
lpni->lpni_peer_net->lpn_peer pointing to a valid lnet_peer.

Keep a count of the total number of lnet_peer_ni attached to
an lnet_peer in lp_nnis.

Split the global ln_peers list into per-lnet_peer_table lists.
The CPT of the peer table in which the lnet_peer is linked is
stored in lp_cpt.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I465f9b732964834dad327fbe5177fba0cfb6775f
Reviewed-on: https://review.whamcloud.com/25784
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: add LNET_PEER_CONFIGURED flag

Add the LNET_PEER_CONFIGURED flag, which indicates that a peer
has been configured by DLC. This is used to enforce that only
DLC can modify such a peer.

This includes some further refactoring of the code that creates
or modifies peers to ensure that the flag is properly passed
through, set, and cleared.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I647116ec19bc2f577732a02bf8efb52dad48a391
Reviewed-on: https://review.whamcloud.com/25783
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: preferred NIs for non-Multi-Rail peers

When a node sends a message to a peer NI, there may be
a preferred local NI that should be the source of the
message. This is in particular the case for non-Multi-
Rail (NMR) peers, as an NMR peer depends in some cases
on the source address of a message to correctly identify
its origin. (This as opposed to using a UUID provided by
a higher protocol layer.)

Implement this by keeping an array of preferred local
NIDs in the lnet_peer_ni structure. The case where only
a single NID needs to be stored is optimized so that this
can be done without needing to allocate any memory.

A flag in the lnet_peer_ni, LNET_PEER_NI_NON_MR_PREF,
indicates that the preferred NI was automatically added
for an NMR peer. Note that a peer which has not been
explicitly configured as Multi-Rail will be treated as
non-Multi-Rail until proven otherwise. These automatic
preferences will be cleared if the peer is changed to
Multi-Rail.

- lnet_peer_ni_set_non_mr_pref_nid()
  set NMR preferred NI for peer_ni
- lnet_peer_ni_clr_non_mr_pref_nid()
  clear NMR preferred NI for peer_ni
- lnet_peer_clr_non_mr_pref_nids()
  clear NMR preferred NIs for all peer_ni

- lnet_peer_add_pref_nid()
  add a preferred NID
- lnet_peer_del_pref_nid()
  delete a preferred NID

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: If98501b34e83f099652f3b19aab5bbbf158f8280
Reviewed-on: https://review.whamcloud.com/25782
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: introduce LNET_PEER_MULTI_RAIL flag bit

Add lp_state as a flag word to lnet_peer, and add lp_lock
to protect it. This lock needs to be taken whenever the
field is updated, because setting or clearing a bit is
a read-modify-write cycle.

The lp_multi_rail is removed, its function is replaced by
the new LNET_PEER_MULTI_RAIL flag bit.

The helper lnet_peer_is_multi_rail() tests the bit.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I15034be7670bcb18460dc709accf675711a48113
Reviewed-on: https://review.whamcloud.com/25781
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: refactor lnet_add_peer_ni()

Refactor lnet_add_peer_ni() and the functions called by it. In
particular, lnet_peer_add_nid() adds an lnet_peer_ni to an
existing lnet_peer, lnet_peer_add() adds a new lnet_peer.

lnet_find_or_create_peer_locked() is removed.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: Iffcbf9ffc26460afc544f102bd0e0a56e23a83f3
Reviewed-on: https://review.whamcloud.com/25780
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: refactor lnet_del_peer_ni()

Refactor lnet_del_peer_ni(). In particular break out the code
that removes an lnet_peer_ni from an lnet_peer and put it into
a separate function, lnet_peer_del_nid().

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: Id5988b308afb093f83cc2e7029d3f2961171c906
Reviewed-on: https://review.whamcloud.com/25779
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: rename lnet_add/del_peer_ni_to/from_peer()

Rename lnet_add_peer_ni_to_peer() to lnet_add_peer_ni(), and
lnet_del_peer_ni_from_peer() to lnet_del_peer_ni().  This brings
the function names closer to the ioctls they implement:
IOCTL_LIBCFS_ADD_PEER_NI and IOCTL_LIBCFS_DEL_PEER_NI. These
names are also a more accturate description their effect: adding
or deleting an lnet_peer_ni to LNet.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I0eefb60cbdedb998a659002b48d4c8ddd3b11fb2
Reviewed-on: https://review.whamcloud.com/25778
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: cleanup of lnet_peer_ni_addref/decref_locked()

Address style issues in lnet_peer_ni_addref_locked() and
lnet_peer_ni_decref_locked(). In the latter routine, replace
a sequence of atomic_dec()/atomic_read() with atomic_dec_and_test().

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I9b7030ac9850b035f8bd80487a7b69b66b1d5858
Reviewed-on: https://review.whamcloud.com/25777
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: add sanity checks on ping-related constants

Add sanity checks for LNet ping related data structures and
constants to wirecheck.c, and update the generated code in
lnet_assert_wire_constants().

In order for the structures and macros to be visible to
wirecheck.c, which is a userspace program, they were moved
from kernel-only lnet/lib-types.h to lnet/types.h

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I2949d27445b29ec69cf8c17b7769291f270a5923
Reviewed-on: https://review.whamcloud.com/25776
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: add Multi-Rail and Discovery ping feature bits

Claim ping features bit for Multi-Rail and Discovery.

Assert in lnet_ping_target_update() that no unknown bits will
be send over the wire.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: Ie84b155f1ae45e3c228a4e49dc898160b81efb94
Reviewed-on: https://review.whamcloud.com/25775
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: automatic sizing of router pinger buffers

The router pinger uses fixed-size buffers to receive the data
returned by a ping. When a router has more than 16 interfaces
(including loopback) this means the data for some interfaces
is dropped.

Detect this situation, and track the number of remote NIs in
the lnet_rc_data_t structure.  lnet_create_rc_data_locked()
becomes lnet_update_rc_data_locked(), and modified to replace
an existing ping buffer if one is present. It is now also
called by lnet_ping_router_locked() when the existing ping
buffer is too small.

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I7286702b8606e25a5c82291ea4138479c4bf010f
Reviewed-on: https://review.whamcloud.com/25774
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: add struct lnet_ping_buffer

The Multi-Rail code will use the ping target buffer also as the
source of data to push to other nodes. This means that there
will be multiple MDs referencing the same buffer, and care must
be taken to ensure that the buffer is not freed while any such
reference remains.

Encapsulate the struct lnet_ping_info (aka lnet_ping_info_t) in
a struct lnet_ping_buffer. This adds a reference count, and the
number of NIDs for the encapsulated lnet_ping_info has been
sized.

For sizing the buffer the constant LNET_PINGINFO_SIZE is replaced
with LNET_PING_INFO_SIZE(NNIS).

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: Iae255a7ebd6099c050bddbea84fb1923a586ac66
Reviewed-on: https://review.whamcloud.com/25773
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Amir Shehata <amir.shehata@intel.com>
Tested-by: Amir Shehata <amir.shehata@intel.com>

LU-9480 lnet: configure lnet_interfaces_max tunable from dlc

Added the ability to configure lnet_interfaces_max from DLC.
Combined the configure and show of numa range and max interfaces
under a "global" YAML element when configuring using YAML.

Test-Parameters: trivial
Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Signed-off-by: Olaf Weber <olaf@sgi.com>
Change-Id: I6f8babdf7900f963cd86acf92468175a49bbaeee
Reviewed-on: https://review.whamcloud.com/25771

LU-9480 lnet: add lnet_interfaces_max tunable

Add an lnet_interfaces_max tunable value, that describes the maximum
number of interfaces per node. This tunable is primarily useful for
sanity checks prior to allocating memory.

Allow lnet_interfaces_max to be set and get from the sysfs interface.

Add LNET_INTERFACES_MIN, value 16, as the minimum value.

Add LNET_INTERFACES_MAX_DEFAULT, value 200, as the default value. This
value was chosen to ensure that the size of an LNet ping message with
any associated LND overhead would fit in 4096 bytes.

(The LNET_INTERFACES_MAX name was not reused to allow for the early
detection of issues when merging code that uses it.)

Rename LNET_NUM_INTERFACES to LNET_INTERFACES_NUM

Test-Parameters: trivial
Signed-off-by: Olaf Weber <olaf@sgi.com>
Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Change-Id: I9bdc72cc688a414f7658fed93f84c9885c8342be
Reviewed-on: https://review.whamcloud.com/25770

New tag 2.10.52

Change-Id: I673949d64dd0067f1f220426ce3389806a886b5b
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9888 tests: Do not run conf-sanity 32b with ZFS

With recent changes to this test to support ZFS 0.7.1,
conf-sanity test 32b consistently fails in automated testing
with a ZFS file system. Add conf-sanity test 32b to the
ALWAYS_EXCEPT list for ZFS testing while the fialure is
investigated.

Test-Parameters: trivial testgroup=review-zfs-part-2
Signed-off-by: James Nunez <james.a.nunez@intel.com>
Change-Id: I1d5f7e5d02f0c318153eab0db01d8ae67ad93f13
Reviewed-on: https://review.whamcloud.com/28602
Tested-by: Jenkins
Reviewed-by: Patrick Farrell <paf@cray.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>

LU-9887 tests: ignore error sanity-lfsck test 9a,b

sanity-lfsck tests 9a and 9b are failing consistently on
checking that speed limiting LFSCK takes less time than the
user defined maximum speed. We should ignore these errors
for now and print the layout or namespace to help understand
this issue.

Test-Parameters: trivial testgroup=review-zfs-part-1
Test-Parameters: testgroup=review-dne-part-2

Signed-off-by: James Nunez <james.a.nunez@intel.com>
Change-Id: I64cac59edd456e6fd519961a4055130c8dbc8a4a
Reviewed-on: https://review.whamcloud.com/28588
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>

LU-9745 dkms: Fix included dkms.conf file

When lustre-dkms is installed with other dkms packages,
the PRE/POST scripts don't seem to function correctly.
This includes the correct dkms.conf by default without having
to recreate and reread it during build.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: Ic6f25480db40d784dfcb3b650f7c869716b903ee
Reviewed-on: https://review.whamcloud.com/28210
Tested-by: Jenkins
Tested-by: Brian J. Murrell <brian.murrell@intel.com>
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-9850 patchless client should conflict patched kernel

Due to how dependencies work in RPM (and a bug in how kmod RPMs generate
their dependency lists), on a node where the server and client repos are
both configured, YUM could allow the patched kernel to satisfy the
patchless-client RPM's requirements.

Add Conflicts: and Provides: to the kernel RPM and lustre-client RPM to
prevent this from happening.

This change also allows one to force the installation of the patched
kernel RPM (yum install kernel-lustre) if one desires.

Test-Parameters: trivial

Signed-off-by: Brian J. Murrell <brian.murrell@intel.com>
Change-Id: If9c44a93937cd7603b0246676ebc9c8260a43b11
Reviewed-on: https://review.whamcloud.com/28557
Tested-by: Jenkins
Reviewed-by: Minh Diep <minh.diep@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9826 test: Recreate disk2_7-zfs image to have a bigger file system

Increase 2.7 zfs file system size and move conf_sanity
test_32b out of the ALWAYS_EXCEPT list

Test-Parameters: ostfilesystemtype=zfs mdtfilesystemtype=zfs \
testlist=conf-sanity

Change-Id: I61e40bc495b06fcc73d4d6d5707433686984bcc7
Signed-off-by: Wei Liu <wei3.liu@intel.com>
Reviewed-on: https://review.whamcloud.com/28464
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6210 utils: Use C99 struct initializers in lsnapshot

This patch makes no functional changes.  The long_options struct
initializers in lsnapshot are updated to C99 syntax.  The long
options are renamed to long_opts for consistency.

C89 positional initializers require values to be placed in the
correct order. This will cause errors if the fields of the struct
definition are reordered or fields are added or removed. C99 named
initializers avoid this problem, and also automatically clear any
values that are not explicitly set.

Test-Parameters: trivial
Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: Ieccc018218597ade289378fc8666988bf969d924
Reviewed-on: https://review.whamcloud.com/28417
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9832 tests: Use sys/sysmacros.h for major/minor/makedev

This is apparently the standard for GNU libc and they
started to warn about it now.

Change-Id: I5fe5c7b3cd08b724d77f752987ce1e753cb48d4b
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: https://review.whamcloud.com/28360
Reviewed-by: Patrick Farrell <paf@cray.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>

LU-9816 kernel: kernel upgrade RHEL7.4 [3.10.0-693.el7]

With this mod we switch our supported el7 version to RHEL 7.4

Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Change-Id: Ib0bcb3547fcf220e1b4665229930c6ff28c6906c
Reviewed-on: https://review.whamcloud.com/28301
Tested-by: Jenkins
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8999 quota: fix quota iteration interface

Since zfs 0.7.0, object accounting is maintained by DMU, so that quota
iteration interface should retrieve the information over there.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: I3d0744dfb52b1a9088b828bc72d648872ec4d00b
Reviewed-on: https://review.whamcloud.com/28345
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9829 lnet: Properly dereference string pointer

A bug highlighted by newer gcc. We really need to be doing
double dereference of the next pointer looking for end of string,
since the pointer is double indirect.

Change-Id: Id80522f43606a843383787d71eec1ddf00af5269
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: https://review.whamcloud.com/28358
Tested-by: Jenkins
Reviewed-by: Patrick Farrell <paf@cray.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>

LU-9295 test: skip sanity-lfsck 9 less than 2.8

sanity-lfsck tests 9a and 9b check the version of the MDS
to see if the test should be run. Unfortunately, the
server_version variable is never initialized and the tests
are skipped.

For sanity-lfsck tests 9a and 9b, get the server version
using the lustre_version_code routine.

Test-Parameters: trivial testlist=sanity-lfsck
Signed-off-by: James Nunez <james.a.nunez@intel.com>
Change-Id: I6c461a6940a0cbd0439c8f29d979ef799ea30229
Reviewed-on: https://review.whamcloud.com/26403
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Saurabh Tandan <saurabh.tandan@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8619 lbuild: update ZFS to use 0.7.1

Update lbuild to build against ZFS 0.7.1

Changelog: https://github.com/zfsonlinux/zfs/releases/tag/zfs-0.7.1

Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Change-Id: I04082cd6cd43c98477100f9fc308666e1b981c0a
Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-on: https://review.whamcloud.com/22569
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9799 mount: Call read_ldd with initialized mount type

When re-reading the ldd when doing relabel, ensure the correct mount
type is used for the read. Otherwise ldiskfs complains:

   mount.lustre FATAL: unhandled/unloaded fs type 0 'ext3'

and ZFS complains:

   e2label: No such file or directory while trying to open MGS/MGT
   Couldn't find valid filesystem superblock.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Change-Id: Ife53cff948d545c306e99e4b023989245a1ac3f7
Reviewed-on: https://review.whamcloud.com/28456
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Hongchao Zhang <hongchao.zhang@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9499 lfsck: set target bitmap properly

If the notify from the peer server has LF_INCOMPLETE flags,
then record it in the target bitmap unconditionally to avoid
missing to update the bitmap for some corner cases.

This patch also addes more debug information when the LFSCK
updates the bitmap and handle double_scan_result.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I3a6195136d608aa47e59e61f95c92978503e3a4b
Reviewed-on: https://review.whamcloud.com/27632
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9494 test: Improve message for skipping tests

Modify skip messages for consistency and clarity.

Signed-off-by: Ruth Klundt <rklundt@sandia.gov>
Change-Id: I44ced56e67aa63ed84da6a15c88282bc3ff19332
Test-Parameters: trivial testlist=sanity
Reviewed-on: https://review.whamcloud.com/27350
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Wei Liu <wei3.liu@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8066 obd: fix LPROC_SEQ_FOPS macros for debugfs

While porting to debugfs I discovered a bug in that
for the debugfs case no data was being passed into
single_open(). This would cause a crash when reading
debugfs files since the seq_file private field was
NULL. Second change is the rename of the *WO_TYPE*
macro to LPROC_SEQ_FOPS_WR_ONLY to match upstream.

Change-Id: I9b6baebecaeb82795d7055148533862489473a1d
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/26650
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Emoly Liu <emoly.liu@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6401 uapi: migrate remaining uapi headers to uapi directory

Move all the remaining lustre headers shared between user land
and kernel space to the uapi directory. OBD_MAX_IOCTL_BUFFER
can be removed since the utilites don't even acknowledge it
value.

Change-Id: I6dcdf3f824dae6bc8a0786f0e61787ca68c10bd8
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/25246
Tested-by: Jenkins
Reviewed-by: Quentin Bouget <quentin.bouget@cea.fr>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Ben Evans <bevans@cray.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9235 libcfs: don't dump stack if just touched

If some lc_watchdog was touched before lcw_dump_stack dumped
the stack of the thread, it should not dump it anymore for
the thread is verified to be active and no need to dump.

Change-Id: I8e4acc1793bb8458ee3b6dc73f2953670ed22896
Signed-off-by: Hongchao Zhang <hongchao.zhang@intel.com>
Reviewed-on: https://review.whamcloud.com/23162
Tested-by: Jenkins
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8618 tests: ha.sh improvements

Patch adds the following ha.sh changes:
- Customise SIMUL and IOR paths.
- Add -p max failover period parameter.
- Add -r dry run parameter.
- Add "iozone" load.
- Add the possibilities to set the number of mpi threads per client.
- CRM is not always configured to fail target back when
  the primary node is back. Add the possibility to execute
  failback command if required.
- The logs from all clients are required if non mpi load fails on
  one client only. Dump logs from all clients.
- Add the possibilities to:
  run ha.sh with custom ior, simul parameters;
  start only the defined list of applications;
  start MPI loads instances on defined number of clients.

Test-Parameters: trivial
Seagate-bug-id: MRP-2150, MRP-2896, MRP-3431, MRP-3252, MRP-3495
Signed-off-by: Elena Gryaznova <elena.gryaznova@seagate.com>
Reviewed-by: Vladimir Saveliev <vladimir.saveliev@seagate.com>
Reviewed-by: Alexander Zarochentsev <alexander.zarochentsev@seagate.com>
Reviewed-by: Alexander Lezhoev <alexander.lezhoev@seagate.com>
Reviewed-by: Vitaly Fertman <vitaly.fertman@seagate.com>
Change-Id: I252aa0945286b30ffa6bad40aebf0c2cbc0c7261
Reviewed-on: https://review.whamcloud.com/22528
Tested-by: Jenkins
Reviewed-by: Jian Yu <jian.yu@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-8578 ldlm: ELC shouldn't wait on lock flush

commit ad87adbff5148a74ac616dbc9bfa02a4e3814034
LU-5727 ldlm: revert the changes for lock canceling policy

but it removes the fix for LU-4300 when lru_resize is disabled.

Introduce ldlm_cancel_aged_no_wait_policy to be used by ELC.

Change-Id: Ic74a6bce4a261dc96c8c0f6d8f7e20315def8c0e
Seagate-bug-id: MRP-3662
Signed-off-by: Andriy Skulysh <andriy.skulysh@seagate.com>
Reviewed-on: https://review.whamcloud.com/22286
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Vitaly Fertman <vitaly.fertman@seagate.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6245 tests: remove it_test

The test application it_test pulls in lustre kernel code which is
the wrong approach and this will not work in the future when the
lustre utilites will be build against the upstream kernel lustre
code. Looking at where it_test is actually run I discovered it is
never used. Since this is the case we can delete it.

Test-Parameters: trivial

Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Change-Id: I8fdcf279590b0edf1c8da6d9f4618964162ec762
Reviewed-on: https://review.whamcloud.com/28406
Tested-by: Jenkins
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-9725 quota: always deregister lwp

qsd should always deregiter lwp upon finish no matter qsd_exp was
set before, otherwise the item will stay on the list, but qsd has
been freed.

Signed-off-by: Lai Siyao <lai.siyao@intel.com>
Change-Id: I0d6206f2f2bc8177d0aa35b350f534d85eab1c03
Reviewed-on: https://review.whamcloud.com/28356
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9376 tests: remove sanity.sh test_103b

Remove mount option "noacl" related things from sanity.sh test_103b
and mount.lustre.8.

Test-Parameters: trivial

Signed-off-by: Emoly Liu <emoly.liu@intel.com>
Change-Id: I3da4500cc177894b10854c2ccd9ed884fec74789
Reviewed-on: https://review.whamcloud.com/27109
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Jian Yu <jian.yu@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9826 tests: Do not run conf-sanity 32b with ZFS

With ZFS 0.7.0, conf-sanity test 32b consistently fails
in automated testing. Since the issue may be the VM test
environment, add conf-sanity test 32b to the
ALWAYS_EXCEPT list for ZFS testing.

Test-Parameters: trivial testlist=conf-sanity
Signed-off-by: James Nunez <james.a.nunez@intel.com>
Change-Id: Ib8582b85760057045bc7cce66d470e81e0e43dde
Reviewed-on: https://review.whamcloud.com/28408
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9054 tests: disable test_312 due to zdb issue

zdb used to work for datasets of exported pool by '-e -p' options,
this has been changed in zfs-0.7.0.

This patch temporarily disables test_312 until zfs upstream ticket
https://github.com/zfsonlinux/zfs/issues/6464 is solved.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Change-Id: Ib0c9eeed4964ea4a0abfed70760cb8fbaeb44496
Reviewed-on: https://review.whamcloud.com/28343
Tested-by: Jenkins
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
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-9558 kernel: replace CURRENT_TIME macro

CURRENT_TIME macro is not y2038 safe on 32 bit systems.

The patch replaces all the uses of CURRENT_TIME by current_time() for
filesystem times, and ktime_get_* functions for others.

struct timespec is also not y2038 safe.  Retain timespec for timestamp
representation here as lustre uses it internally everywhere.  These
references will be changed to use struct timespec64 in a separate patch.

This is also in preparation for the patch that transitions vfs
timestamps to use 64 bit time and hence make them y2038 safe.
current_time() is also planned to be transitioned to y2038 safe behavior
along with this change.

CURRENT_TIME macro will be deleted before merging the aforementioned
change.

Linux-commit: 47f38c539e9a42344ff5a664942075bd4df93876

Change-Id: Iec67df8f48066801c227aba2bc9221d444dada65
Link: http://lkml.kernel.org/r/1491613030-11599-10-git-send-email-deepa.kernel@gmail.com
Signed-off-by: Deepa Dinamani <deepa.kernel@gmail.com>
Reviewed-on: https://review.whamcloud.com/28305
Tested-by: Jenkins
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-9364 test: wait rmultiop_start to start

In rmultiop_start, the remote command could be delayed a while,
and wait some time for the command to run.

Test-Parameters: trivial testlist=replay-vbr

Change-Id: Ic8beec725edc89a527c74e2033e59e1da0d444c9
Signed-off-by: Hongchao Zhang <hongchao.zhang@intel.com>
Reviewed-on: https://review.whamcloud.com/26729
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Wei Liu <wei3.liu@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: Saurabh Tandan <saurabh.tandan@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9012 contrib: to rename file ending with :rhel5.ed

Git clone of lustre-release master fails on windows
because of file contrib/patches/ofed/
01-dont-include-openib-initscript:rhel5.ed
On Windows ':' is not support in file name.
This file can be renamed as contrib/patches/ofed/
01-dont-include-openib-initscript-rhel5.ed
i.e. replace ':' with '-'.

submitting this as commented by Andreas for rhel5.ed file

Test-Parameters: trivial

Signed-off-by: dilip krishnagiri <dilipx.krishnagiri@intel.com>
Change-Id: I8170b01e79b5ea30c258b0b605c8b2f95917d60d
Reviewed-on: https://review.whamcloud.com/24842
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-7899 osd: batch EA updates

during file creation we set number of EAs: LMA, VBR, LinkEA, LOVEA, ACLs.
calling into SA to refill spill again and again is expensive. thus it
makes sense to postpone this to osd_trans_stop() where all changed EAs
has been already collected in a temporary buffer.

Change-Id: Ia2604ddafdf8b2ca4f6db4d70ead6d2d2761cd26
Signed-off-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-on: http://review.whamcloud.com/19143
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-on: https://review.whamcloud.com/21893

LU-9748 lod: safely access update log stat

'lod_child_got_update_log' and 'ltd_got_update_log' are not accessed
with lock, so there is race to check all got update log, use lod_lock
to serialize setting and checking these flags, as a side effect,
this lock can act as barrier before wakeup.

Add some debug messages, which can help understand long DNE recovery.

Signed-off-by: Lai Siyao <lai.siyao@intel.com>
Change-Id: Icf65837fe24dbfef35963dcc8502888271334ba5
Reviewed-on: https://review.whamcloud.com/28000
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: wangdi <di.wang@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9266 hsm: don't add request when cdt is stopped

Check cdt_state after getting layout lock in mdt_hsm_add_actions.
Fix protects against several RESTORE records addressed to the
same object in llog. Such records causes mount to hung when
starting hsm:
D: 15524  TASK: ffff880068b5b540  CPU: 4   COMMAND: "lctl"
 #0 [ffff8800bacd9728] schedule at ffffffff81525d30
 #1 [ffff8800bacd97f0] ldlm_completion_ast at ffffffffa08527f5 [ptlrpc]
 #2 [ffff8800bacd9890] ldlm_cli_enqueue_local at ffffffffa0851b8e
[ptlrpc]
 #3 [ffff8800bacd9910] mdt_object_lock0 at ffffffffa0e4ec4c [mdt]
 #4 [ffff8800bacd99c0] mdt_object_lock at ffffffffa0e4f694 [mdt]
 #5 [ffff8800bacd99d0] mdt_object_find_lock at ffffffffa0e4f9c1 [mdt]
 #6 [ffff8800bacd9a00] hsm_restore_cb at ffffffffa0e9b533 [mdt]
 #7 [ffff8800bacd9a50] llog_process_thread at ffffffffa05fd699
[obdclass]
 #8 [ffff8800bacd9b10] llog_process_or_fork at ffffffffa05fdbaf
[obdclass]
 #9 [ffff8800bacd9b60] llog_cat_process_cb at ffffffffa0601250
[obdclass]

Change-Id: Ib09139795d847cac2e5f079a192a3548d32db09c
Seagate-bug-id: MRP-4251
Signed-off-by: Sergey Cheremencev <sergey.cheremencev@seagate.com>
Signed-off-by: Hongchao Zhang <hongchao.zhang@intel.com>
Reviewed-on: https://review.whamcloud.com/26215
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Faccini Bruno <bruno.faccini@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9817 lnet: safe access to msg

When tx credits are returned if there are pending messages they
need to be sent. Messages could have different tx_cpts, so the
correct one needs to be locked. After lnet_post_send_locked(),
if we locked a different CPT then we need to relock the correct one
However, as part of lnet_post_send_locked(), lnet_finalze() can
be called which can free the message. Therefore, the cpt of the
message being passed must be cached in order to prevent access to
freed memory.

Signed-off-by: Amir Shehata <amir.shehata@intel.com>
Change-Id: I959fdc30daf87b5575d8371da20d5cf6f64e7d3c
Reviewed-on: https://review.whamcloud.com/28308
Tested-by: Jenkins
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Sonia Sharma <sonia.sharma@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9731 Limit work-around to EL7 only

Since the workaround previously landed for LU-9731 only applies to
EL7, only apply it for EL7 builds.

Signed-off-by: Brian J. Murrell <brian.murrell@intel.com>
Change-Id: Id74f03f3af74f324320e094e32f7b7480259145c
Reviewed-on: https://review.whamcloud.com/28202
Reviewed-by: Minh Diep <minh.diep@intel.com>
Tested-by: Jenkins
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9597 ofd: fix race for project setattr

sanity-quota 33 exposed an intresting project quota bug,
problem is we could hit following condition:

step 1: create an empty file
step 2: buffer write data to file, at this time project id 0 is packed.
step 3: chattr file's project id

If write RPC is generated on client but not reaching OST yet, and step 3
chattr is issued in this time window and chattr on OST is done before
write PRC being processed on OST:

That means we have changed file's project ID, but ofd_attr_handle_id()
did not clear S_ISUID and S_ISGID for this case. Write RPC arrived on
OST, first write RPC will call ofd_attr_handle_id() to set attribute.
Unfortunately it will think this is first write as S_ISUID and S_ISGID
is set, project id will be reset with packed id 0, thus we got wrong
project accounting.

We should use another file mode to indicate whether
project id has been initialized, this patch tries to
fix this problem.

Change-Id: Ie2f9ce5b662c3011aa1cd2f59e1fb20526a3e3d7
Signed-off-by: Wang Shilong <wshilong@ddn.com>
Reviewed-on: https://review.whamcloud.com/28274
Tested-by: Jenkins
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8066 libcfs: migrate to debugfs

move /proc/sys/lnet to debugfs

Parameters in lnet sysctl are of debug quality, so let's move them
to debugfs instead.

Linux-commit: 0871d551af37c72c308397c16c31bae945e6a79d

Remove redundant lnet debugfs variables

/proc/sys/lnet/console_ratelimit, debug_path and panic_on_lbug
are module parameters with no special magic accessible via
/sys/module/libcfs/parameters/libcfs_console_ratelimit,
/sys/module/libcfs/parameters/libcfs_debug_file_path and
/sys/module/libcfs/parameters/libcfs_panic_on_lbug respectively.

As such just replace them with corresponding symlinks

Linux-commit: 1b4d97b6c6ca05446ff33843a7b174cdd765df5a

get rid of debugfs/lnet/console_backoff

module parameter libcfs_console_backoff accessible through
/sys/module/libcfs/parameters/libcfs_console_backoff would
do the same thing, just add a special "uintpos" parameter
type to disallow 0 values too.

Also add a symlink to the module parameter variable for
backwards compatibility

Linux-commit: 8710427dd68f4dff8976d221e220317cea20ecec

get rid of debugfs/lnet/debug_mb

It's just a fancy libcfs_debug_mb module parameter wrapper,
so just add debug buffer size check and resizing and the same
functionality now would be accessible via
/sys/module/libcfs/parameters/libcfs_debug_mb

Also add a symlink for backwards compatibility.

Linux-commit: 8dc08446d0671709bdec9e037845b014e33663b1

get rid of debugfs/lnet/console_{min, max}_delay_centisecs

They are just fancy module parameters wrappers,
so just the same functionality now would be accessible via
/sys/module/libcfs/parameters/libcfs_console_{min,max}_delay

Also install compatibility symlinks

Linux-commit: 35ca907d245e8c218f39b6ed8bd6a1c656d056cc

Fix kstrtouint return value check fix

Apparently kstrtouint could return not just -EINVAL, but also -ERANGE,
so make sure we just check the return value for something negative.

Noticed by Dan Carpenter <dan.carpenter@oracle.com>

Linux-commit: aa66d6f87f300b15ac2efdcde9e198bf5a9020ce

Properly handle debugfs read- and write-only files

It turns out that unlike procfs, debugfs does not really enforce
permissions for root (similar to regular filesystems), so we need
to ensure we are not providing ->write() method to read-only files
and ->read() method for write-only files at registration.

This fixes a couple of crashes on unexpected access.

Linux-commit:49a76d7040b466dd92f198ded4aba3cd3569d48f

Change-Id: I69ff4d9766969f2ca5d5478d4e822879ee78b467
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/24688
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>

LU-6245 uapi: move libcfs/lnet UAPI header into own uapi directory

Move all the libcfs/lnet headers shared between user land and
kernel space to the new lnet uapi directory.

Test-Parameters: trivial

Change-Id: Id6142b0e890fd47d3318dc3c98e64f72b5df2594
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/28089
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9712 kernel: correct malformed ChangeLog

Somewhwre in the recent set of landings for kernel updates
the wrong lines in ChangeLog were updated.
This mod corrects that.

Test-Parameters: trivial

Change-Id: I95a66eb4108aebdb6883f445cfa183988554ad74
Signed-off-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-on: https://review.whamcloud.com/28273
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Yang Sheng <yang.sheng@intel.com>
Reviewed-by: Minh Diep <minh.diep@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9744 mdt: avoid wrong CLF_HSM_DIRTY report in ChangeLog

In hsm_cdt_request_completed() and upon error being returned
from mdt_hsm_get_md_hsm(), where "struct md_hsm mh" has not been
populated, HS_DIRTY can be wrongly detected and thus CLF_HSM_DIRTY
flag will be reported by error
This can be the cause of errors in associated sanity-hsm sub-tests
test_220a, test_222c, test_222d, tests_224a, when analyzing
ChangeLog flags.
!IS_ERR(obj) should also be tested before adding CLF_HSM_DIRTY
in cl_flags.

Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Change-Id: I4469a55b35ea5d35a9f0be152f085bd676f74240
Reviewed-on: https://review.whamcloud.com/27962
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Quentin Bouget <quentin.bouget@cea.fr>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6210 utils: Use C99 initializer in jt_nodemap_del_range()

This patch makes no functional changes.  The long_options struct
initializer in jt_nodemap_del_range() is updated to C99 syntax.
It is renamed to long_opts for consistency.

C89 positional initializers require values to be placed in the
correct order. This will cause errors if the fields of the struct
definition are reordered or fields are added or removed. C99 named
initializers avoid this problem, and also automatically clear any
values that are not explicitly set.

Test-Parameters: trivial
Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: I442d2cb65c1ffefc2229ead83e04b7ec2d8160e5
Reviewed-on: https://review.whamcloud.com/27883
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-6210 utils: Use C99 initializer in jt_nodemap_test_id()

This patch makes no functional changes.  The long_options struct
initializer in jt_nodemap_test_id() is updated to C99 syntax.
It is renamed to long_opts for consistency.

C89 positional initializers require values to be placed in the
correct order. This will cause errors if the fields of the struct
definition are reordered or fields are added or removed. C99 named
initializers avoid this problem, and also automatically clear any
values that are not explicitly set.

Test-Parameters: trivial
Signed-off-by: Steve Guminski <stephenx.guminski@intel.com>
Change-Id: Iddbee594f8ccd9cada11c6c792e31656c16f7170
Reviewed-on: https://review.whamcloud.com/27881
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-8958 llite: remove llite_loop left overs

With the removal of llite_loop several pieces of code are still
present in the llite layer that were only used by the lloop device.
We can remove these no longer used pieces.

Change-Id: I67f2761ae29b8ba7bb3bc9bcc3e3f8ece73a3ea3
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/26795
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-9758 build: allow disabling lustre test and iokit rpm creation

While attempting to create a basic set of rpms that didn't
include the lustre test and iokit rpm I encountered build
breakage. Test the lustre_test conditional in the spec file
so we don't attempt to build special lustre test rpms. The
lustre test rpm is actually dependant on the lustre iokit
rpm so if --disable-iokit is set we should disable lustre
test rpms generation as well.

Test-Parameters: trivial

Change-Id: I0adeb0eded1a3b99481fc4b7ad3cda2afd32b6d2
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/27985
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-6142 socklnd: remove typedefs from ksocklnd

Change the typedefs in lnd ksocklnd to proper structures.
Several other style changes to fix checkpatch issues with
code impacted by typedef change.

Test-Parameters: trivial

Change-Id: I78d69aea46721f4e97a1775c64ae0d59879aa4fd
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/28275
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-9019 mgs: migrate to 64 bit time

Port the remaining time code from 32 bits to 64 bits.
Replace cfs_time_current_sec() to avoid the overflow
issues in 2038 with ktime_get_real_seconds(). Moved
lprocfs_ir_timeout_seq_show from printing 32 times
to 64 bit times. Do a direction compare instead of
using the  cfs_time_before() wrapper. Change the
following to time64_t:

fsdb_barrier_expected
fsdb_barrier_latest_create_time
mgs_start_time
ir_timeout

Change-Id: I10bfd3f1479ad2823e325bfc464b802b210c238f
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-on: https://review.whamcloud.com/27910
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-7356 build: handle configure options for mount plugin

Currently if you build lustre with configure options
--disable-shared and/or --disable-static it will fail
to build in the case of the mount plugins. The mount
plugins are always built to be sharable code using
libtool but this is not compatiable with the confiure
--disable-share option. The solution was to create a
temporary static library and for the static case we
directly link into the lustre utilites. If shared
libraries are enabled then that temporary library
is turned into a dynamic plugin. The lustre utilites
was updated to handle both cases.

Test-Parameters: trivial

Change-Id: I0eec25c8bce71c0e5c533ff3dfbe752677f37a26
Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-on: https://review.whamcloud.com/17581
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

LU-8275 tests: add flag to enable secret shared key for tests

When the SHARED_KEY environment variable is set to true,
test-framework will set up a shared key between the nodes and start
Lustre with shared key enabled. Three tests (28,29,30) are also
added to sanity-sec in order to test shared key features.

Signed-off-by: Kit Westneat <kit.westneat@gmail.com>
Signed-off-by: Chris Hanna <hannac@iu.edu>
Change-Id: I2abff35fbb7dd4f227c6cd12a66cb0b77f610157
Reviewed-on: https://review.whamcloud.com/20780
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Nunez <james.a.nunez@intel.com>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>