These macros are no longer used, so they can
be removed.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This l_wait_event_exclusive_head() will wait indefinitely
if the timeout is zero. If it does wait with a timeout
and times out, the timeout for next time is set to zero.
The can be mapped to a call to either
wait_event_idle_exclusive()
or
wait_event_idle_exclusive_timeout()
depending in the timeout setting.
The current code arranges for LIFO queuing of waiters,
but include/event.h doesn't support that yet.
Until it does, fall back on FIFO with
wait_event_idle_exclusive{,_timeout}().
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This is the last remaining use of l_wait_event().
It is the only use of LWI_TIMEOUT_INTR_ALL() which
has a meaning that timeouts can be interrupted.
Only interrupts by "fatal" signals are allowed, so
introduce l_wait_event_abortable_timeout() to
support this.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
replace l_wait_event() with wait_event_idle_timeout() and explicit
loop. This approach is easier to understand.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Rather an using l_wait_event(), use wait_event_idle_timeout()
with an explicit loop so it is easier to see what is happening.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Replace l_wait_event with wait_event_idle_timeout() and call the
handler function explicitly. This makes it more clear
what is happening.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We can replace l_wait_event() with
wait_event_idle_timeout() here providing we call the
timeout function when wait_event_idle_timeout() returns zero.
As ptlrpc_expired_set() returns 1, the l_wait_event() aborts of the
first timeout.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This use of l_wait_event() is a polling loop that re-checks
every second. Make this more obvious with a while loop
and wait_event_idle_timeout().
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When 'back_to_sleep()' is passed as the 'timeout' function,
the effect is to wait indefinitely for the event, polling
once after the timeout.
If LWI_ON_SIGNAL_NOOP is given, then after the timeout
we allow fatal signals to interrupt the wait.
Make this more obvious in both places "back_to_sleep()" is
used but using two explicit sleeps.
The code in ptlrpcd_add_req() looks odd - why not just have one
wait_event_idle()? However I believe this is a faithful
transformation of the existing code.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This waiter currently wakes up every second to re-test if
imp_flight is zero. If we ensure wakeup is called whenever
imp_flight is decremented to zero, we can just have a simple
wait_event_idle_timeout().
So add a wake_up_all to the one place it is missing, and simplify
the wait_event.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Two places that LWI_TIMEOUT_INTERVAL() is used, the outcome is a
simple polling loop that polls every second for some event (with a
limit).
So write a simple loop to make this more apparent.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If a signal-callback (lwi_on_signal) is set without lwi_allow_intr, as
is the case in ldlm_completion_ast(), the behavior depends on the
timeout set.
If a timeout is set, then signals are ignored. If the timeout is
reached, the timeout handler is called. If the timeout handler
return 0, which ldlm_expired_completion_wait() always does, the
l_wait_event() switches to exactly the behavior if no timeout was set.
If no timeout is set, then "fatal" signals are not ignored. If one
arrives the callback is run, but as the callback is empty in this
case, that is not relevant.
This can be simplified to:
if a timeout is wanted
wait_event_idle_timeout()
if that timed out, call the timeout handler
l_wait_event_abortable()
i.e. the code always waits indefinitely. Sometimes it performs a
non-abortable wait first. Sometimes it doesn't. But it only
aborts before the condition is true if it is signaled.
This doesn't quite agree with the comments and debug messages.
Now that we call the timeout handler (ldlm_expired_completion_wait())
wait directly, we can pass the two args directly rather then
using a special-purpose struct.
Reviewed-by: Patrick Farrell <paf@cray.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If l_wait_event() is given a function to be called on a signal,
but no timeout or timeout handler, then the intr function is simply
called at the end if the wait was aborted by a signal.
So a simpler way to write the code (in the one place this case is
used) it to open-code the body of the function after the
wait_event, if -ERESTARTSYS was returned.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
lustre sometimes wants to wait for an event, but abort if
one of a specific list of signals arrives. This is a little
bit like wait_event_killable(), except that the signals are
identified a different way.
So introduce l_wait_event_abortable() which provides this
functionality.
Having separate functions for separate needs is more in line
with the pattern set by include/linux/wait.h, than having a
single function which tries to include all possible needs.
Also introduce l_wait_event_abortable_exclusive().
Note that l_wait_event() return -EINTR on a signal, while
Linux wait_event functions return -ERESTARTSYS.
l_wait_event_{abortable_,}exclusive follow the Linux pattern.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the lwi arg has a timeout, but no timeout
callback function, l_wait_event() acts much the same as
wait_event_idle_timeout() - the wait is not interruptible and
simply waits for the event or the timeouts.
The most noticable difference is that the return value is
-ETIMEDOUT or 0, rather than 0 or non-zero.
Another difference is that if the timeout is zero, l_wait_event()
will not time out at all. In the one case where that is possible
we need to conditionally use wait_event_idle().
So replace all such calls with wait_event_idle_timeout(), being
careful of the return value.
In one case, there is no event expected, only the timeout
is needed. So use schedule_timeout_uninterruptible().
Note that the presence or absence of LWI_ON_SIGNAL_NOOP
has no effect in these cases. It only has effect if the timeout
callback is non-NULL, or the timeout is zero, or
LWI_TIMEOUT_INTR_ALL() is used.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
cfs_time_seconds() converts a number of seconds to the
matching number of jiffies.
The standard way to do this in Linux is "* HZ".
So discard cfs_time_seconds() and use "* HZ" instead.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the lwi arg is full of zeros, l_wait_event() behaves almost
identically to the standard wait_event_idle() interface, so use that
instead.
l_wait_event() uses TASK_INTERRUPTIBLE, but blocks all signals.
wait_event_idle() uses the new TASK_IDLE and so avoids adding
to the load average without needing to block signals.
In one case, wait_event_idle_exclusive() is needed.
Also remove all l_wait_condition*() macros which were short-cuts
for setting lwi to {0}.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This flag is never set, so remove checks and remove
the flag.
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Patrick Farrell <paf@cray.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The task of ll_find_alias() is now very similar to d_exact_alias().
We cannot use that function directly, but we can copy much of
the structure so that the similarities and differences are more
obvious.
Examining d_exact_alias() shows that the d_lock spinlock does not
need to be held in ll_find_alias as much as it currently is.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that ll_find_alias is only searching for one type
of dentry, we can return as soon as we find it.
This allows substantial simplification, and brings the
bonus that we don't need to take the d_lock again just
to increment the ref-count. We can increment it immediately
that the dentry is found.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that ll_find_alias() is never called for directories,
we can remove code that only applies to directories.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the Linux dcache a directory only ever has one dentry,
so d_splice_alias() can be used by ll_splice_alias() for directories.
It will find the one dentry whether it is DCACHE_DISCONNECTED or
IS_ROOT() or d_lustre_invalid().
Separating out the directories from non-directories will allow us
to simplify the non-directory code.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ll_dcompare is used in two slightly different contexts.
It is called (from __d_lookup, __d_lookup_rcu, and d_exact_alias)
to compare a name against a dentry that is already in the dcache.
It is also called (from d_alloc_parallel) to compare a name against
a dentry that is not in the dcache yet, but is part of an active
"lookup" or "atomic_open" call.
In the first case we need to avoid matching against "invalid" dentries
as a match implies something about ldlm locks which is not accurate.
In the second case we need to allow matching against "invalid" dentries
as the dentry will always be invalid (set by ll_d_init()) but we still
want to guard against multiple concurrent lookups of the same name.
d_alloc_parallel() will repeat the call to ll_dcompare() after
the lookup has finished, and if the dentry is still invalid, the whole
d_alloc_parallel() process is repeated. This assures us that it is safe
to report success whenever d_in_lookup().
With this patch, there will never be two threads concurrently in
ll_lookup_nd(), looking up the same name in the same directory.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 020ecc6f32 ("staging: lustre: llite: Remove IS_ERR tests")
changed ll_prep_inode to assume any error from ll_iget() meant
-ENOMEM because at that time it only returned NULL for errors.
Commit c3397e7e67 ("staging: lustre: llite: add error handler in
inode prepare phase") changed ll_iget() to once again return
meaningful codes, but nobody told ll_prep_inode().
So change ll_prep_inode() back to using PTR_ERR(*inode).
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This code tests various fields to see if they are different, except
for one where there test is if they are the same.
This is clearly wrong for a function that is tesding for equality.
So change "!strcmp()" which I always find hard to read, to
"strcmp() != 0" which obviously means that the strings are not equal.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As more people become involved with the progression of the lustre
client it needs to more clear what needs to be done to leave
staging. Update the TODO list with the various bugs and changes
to accomplish this. Some are simple bugs and others are far more
complex task that will change many lines of code. Some even cover
updating the user land utilities to meet the kernel requirements.
Several bugs have already been addressed and just need to be
pushed to the staging tree.
Signed-off-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Return value of error codes should typically be negative.
Issue reported by checkpatch.pl
Signed-off-by: Sumit Pundir <pundirsumit11@gmail.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Here is the big Staging and IIO driver patches for 4.16-rc1.
There is the normal amount of new IIO drivers added, like all releases.
The networking IPX and the ncpfs filesystem are moved into the staging
tree, as they are on their way out of the kernel due to lack of use
anymore.
The visorbus subsystem finall has started moving out of the staging tree
to the "real" part of the kernel, and the most and fsl-mc codebases are
almost ready to move out, that will probably happen for 4.17-rc1 if all
goes well.
Other than that, there is a bunch of license header cleanups in the
tree, along with the normal amount of coding style churn that we all
know and love for this codebase. I also got frustrated at the
Meltdown/Spectre mess and took it out on the dgnc tty driver, deleting
huge chunks of it that were never even being used.
Full details of everything is in the shortlog.
All of these patches have been in linux-next for a while with no
reported issues.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-----BEGIN PGP SIGNATURE-----
iG0EABECAC0WIQT0tgzFv3jCIUoxPcsxR9QN2y37KQUCWnLxoA8cZ3JlZ0Brcm9h
aC5jb20ACgkQMUfUDdst+yk4vgCgjeMlwhtar65DIticIRj626EFxiQAnjGmH8Kd
d9Xz2Piq8X47uSsC/6AE
=xxMT
-----END PGP SIGNATURE-----
Merge tag 'staging-4.16-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging
Pull staging/IIO updates from Greg KH:
"Here is the big Staging and IIO driver patches for 4.16-rc1.
There is the normal amount of new IIO drivers added, like all
releases.
The networking IPX and the ncpfs filesystem are moved into the staging
tree, as they are on their way out of the kernel due to lack of use
anymore.
The visorbus subsystem finall has started moving out of the staging
tree to the "real" part of the kernel, and the most and fsl-mc
codebases are almost ready to move out, that will probably happen for
4.17-rc1 if all goes well.
Other than that, there is a bunch of license header cleanups in the
tree, along with the normal amount of coding style churn that we all
know and love for this codebase. I also got frustrated at the
Meltdown/Spectre mess and took it out on the dgnc tty driver, deleting
huge chunks of it that were never even being used.
Full details of everything is in the shortlog.
All of these patches have been in linux-next for a while with no
reported issues"
* tag 'staging-4.16-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging: (627 commits)
staging: rtlwifi: remove redundant initialization of 'cfg_cmd'
staging: rtl8723bs: remove a couple of redundant initializations
staging: comedi: reformat lines to 80 chars or less
staging: lustre: separate a connection destroy from free struct kib_conn
Staging: rtl8723bs: Use !x instead of NULL comparison
Staging: rtl8723bs: Remove dead code
Staging: rtl8723bs: Change names to conform to the kernel code
staging: ccree: Fix missing blank line after declaration
staging: rtl8188eu: remove redundant initialization of 'pwrcfgcmd'
staging: rtlwifi: remove unused RTLHALMAC_ST and RTLPHYDM_ST
staging: fbtft: remove unused FB_TFT_SSD1325 kconfig
staging: comedi: dt2811: remove redundant initialization of 'ns'
staging: wilc1000: fix alignments to match open parenthesis
staging: wilc1000: removed unnecessary defined enums typedef
staging: wilc1000: remove unnecessary use of parentheses
staging: rtl8192u: remove redundant initialization of 'timeout'
staging: sm750fb: fix CamelCase for dispSet var
staging: lustre: lnet/selftest: fix compile error on UP build
staging: rtl8723bs: hal_com_phycfg: Remove unneeded semicolons
staging: rts5208: Fix "seg_no" calculation in reset_ms_card()
...
Pull crypto updates from Herbert Xu:
"API:
- Enforce the setting of keys for keyed aead/hash/skcipher
algorithms.
- Add multibuf speed tests in tcrypt.
Algorithms:
- Improve performance of sha3-generic.
- Add native sha512 support on arm64.
- Add v8.2 Crypto Extentions version of sha3/sm3 on arm64.
- Avoid hmac nesting by requiring underlying algorithm to be unkeyed.
- Add cryptd_max_cpu_qlen module parameter to cryptd.
Drivers:
- Add support for EIP97 engine in inside-secure.
- Add inline IPsec support to chelsio.
- Add RevB core support to crypto4xx.
- Fix AEAD ICV check in crypto4xx.
- Add stm32 crypto driver.
- Add support for BCM63xx platforms in bcm2835 and remove bcm63xx.
- Add Derived Key Protocol (DKP) support in caam.
- Add Samsung Exynos True RNG driver.
- Add support for Exynos5250+ SoCs in exynos PRNG driver"
* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (166 commits)
crypto: picoxcell - Fix error handling in spacc_probe()
crypto: arm64/sha512 - fix/improve new v8.2 Crypto Extensions code
crypto: arm64/sm3 - new v8.2 Crypto Extensions implementation
crypto: arm64/sha3 - new v8.2 Crypto Extensions implementation
crypto: testmgr - add new testcases for sha3
crypto: sha3-generic - export init/update/final routines
crypto: sha3-generic - simplify code
crypto: sha3-generic - rewrite KECCAK transform to help the compiler optimize
crypto: sha3-generic - fixes for alignment and big endian operation
crypto: aesni - handle zero length dst buffer
crypto: artpec6 - remove select on non-existing CRYPTO_SHA384
hwrng: bcm2835 - Remove redundant dev_err call in bcm2835_rng_probe()
crypto: stm32 - remove redundant dev_err call in stm32_cryp_probe()
crypto: axis - remove unnecessary platform_get_resource() error check
crypto: testmgr - test misuse of result in ahash
crypto: inside-secure - make function safexcel_try_push_requests static
crypto: aes-generic - fix aes-generic regression on powerpc
crypto: chelsio - Fix indentation warning
crypto: arm64/sha1-ce - get rid of literal pool
crypto: arm64/sha2-ce - move the round constant table to .rodata section
...
Pull misc vfs updates from Al Viro:
"All kinds of misc stuff, without any unifying topic, from various
people.
Neil's d_anon patch, several bugfixes, introduction of kvmalloc
analogue of kmemdup_user(), extending bitfield.h to deal with
fixed-endians, assorted cleanups all over the place..."
* 'work.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (28 commits)
alpha: osf_sys.c: use timespec64 where appropriate
alpha: osf_sys.c: fix put_tv32 regression
jffs2: Fix use-after-free bug in jffs2_iget()'s error handling path
dcache: delete unused d_hash_mask
dcache: subtract d_hash_shift from 32 in advance
fs/buffer.c: fold init_buffer() into init_page_buffers()
fs: fold __inode_permission() into inode_permission()
fs: add RWF_APPEND
sctp: use vmemdup_user() rather than badly open-coding memdup_user()
snd_ctl_elem_init_enum_names(): switch to vmemdup_user()
replace_user_tlv(): switch to vmemdup_user()
new primitive: vmemdup_user()
memdup_user(): switch to GFP_USER
eventfd: fold eventfd_ctx_get() into eventfd_ctx_fileget()
eventfd: fold eventfd_ctx_read() into eventfd_read()
eventfd: convert to use anon_inode_getfd()
nfs4file: get rid of pointless include of btrfs.h
uvc_v4l2: clean copyin/copyout up
vme_user: don't use __copy_..._user()
usx2y: don't bother with memdup_user() for 16-byte structure
...
Pull kern_recvmsg reduction from Al Viro:
"kernel_recvmsg() is a set_fs()-using wrapper for sock_recvmsg(). In
all but one case that is not needed - use of ITER_KVEC for ->msg_iter
takes care of the data and does not care about set_fs(). The only
exception is svc_udp_recvfrom() where we want cmsg to be store into
kernel object; everything else can just use sock_recvmsg() and be done
with that.
A followup converting svc_udp_recvfrom() away from set_fs() (and
killing kernel_recvmsg() off) is *NOT* in here - I'd like to hear what
netdev folks think of the approach proposed in that followup)"
* 'work.sock_recvmsg' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
tipc: switch to sock_recvmsg()
smc: switch to sock_recvmsg()
ipvs: switch to sock_recvmsg()
mISDN: switch to sock_recvmsg()
drbd: switch to sock_recvmsg()
lustre lnet_sock_read(): switch to sock_recvmsg()
cfs2: switch to sock_recvmsg()
ncpfs: switch to sock_recvmsg()
dlm: switch to sock_recvmsg()
svc_recvfrom(): switch to sock_recvmsg()
-----BEGIN PGP SIGNATURE-----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=iihN
-----END PGP SIGNATURE-----
Merge tag 'iversion-v4.16-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux
Pull inode->i_version rework from Jeff Layton:
"This pile of patches is a rework of the inode->i_version field. We
have traditionally incremented that field on every inode data or
metadata change. Typically this increment needs to be logged on disk
even when nothing else has changed, which is rather expensive.
It turns out though that none of the consumers of that field actually
require this behavior. The only real requirement for all of them is
that it be different iff the inode has changed since the last time the
field was checked.
Given that, we can optimize away most of the i_version increments and
avoid dirtying inode metadata when the only change is to the i_version
and no one is querying it. Queries of the i_version field are rather
rare, so we can help write performance under many common workloads.
This patch series converts existing accesses of the i_version field to
a new API, and then converts all of the in-kernel filesystems to use
it. The last patch in the series then converts the backend
implementation to a scheme that optimizes away a large portion of the
metadata updates when no one is looking at it.
In my own testing this series significantly helps performance with
small I/O sizes. I also got this email for Christmas this year from
the kernel test robot (a 244% r/w bandwidth improvement with XFS over
DAX, with 4k writes):
https://lkml.org/lkml/2017/12/25/8
A few of the earlier patches in this pile are also flowing to you via
other trees (mm, integrity, and nfsd trees in particular)".
* tag 'iversion-v4.16-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux: (22 commits)
fs: handle inode->i_version more efficiently
btrfs: only dirty the inode in btrfs_update_time if something was changed
xfs: avoid setting XFS_ILOG_CORE if i_version doesn't need incrementing
fs: only set S_VERSION when updating times if necessary
IMA: switch IMA over to new i_version API
xfs: convert to new i_version API
ufs: use new i_version API
ocfs2: convert to new i_version API
nfsd: convert to new i_version API
nfs: convert to new i_version API
ext4: convert to new i_version API
ext2: convert to new i_version API
exofs: switch to new i_version API
btrfs: convert to new i_version API
afs: convert to new i_version API
affs: convert to new i_version API
fat: convert to new i_version API
fs: don't take the i_lock in inode_inc_iversion
fs: new API for handling inode->i_version
ntfs: remove i_version handling
...
The logic of the original commit 4d99b2581e ("staging: lustre: avoid
intensive reconnecting for ko2iblnd") was assumed conditional free of
struct kib_conn if the second argument free_conn in function
kiblnd_destroy_conn(struct kib_conn *conn, bool free_conn) is true.
But this hunk of code was dropped from original commit. As result the logic
works wrong and current code use struct kib_conn after free.
> drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_cb.c
> 3317 kiblnd_destroy_conn(conn, !peer);
> ^^^^ Freed always (but should be conditionally)
> 3318
> 3319 spin_lock_irqsave(lock, flags);
> 3320 if (!peer)
> 3321 continue;
> 3322
> 3323 conn->ibc_peer = peer;
> ^^^^^^^^^^^^^^ Use after free
> 3324 if (peer->ibp_reconnected < KIB_RECONN_HIGH_RACE)
> 3325 list_add_tail(&conn->ibc_list,
> ^^^^^^^^^^^^^^ Use after free
> 3326 &kiblnd_data.kib_reconn_list);
> 3327 else
> 3328 list_add_tail(&conn->ibc_list,
> ^^^^^^^^^^^^^^ Use after free
> 3329 &kiblnd_data.kib_reconn_wait);
To avoid confusion this fix moved the freeing a struct kib_conn outside of
the function kiblnd_destroy_conn() and free as it was intended in original
commit.
Cc: <stable@vger.kernel.org> # v4.6
Fixes: 4d99b2581e ("staging: lustre: avoid intensive reconnecting for ko2iblnd")
Signed-off-by: Dmitry Eremin <Dmitry.Eremin@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When compiled without CONFIG_SMP, we get a compile error
as ->ctb_parts is not defined.
There is already a function, cfs_cpt_cpumask(), which will get the
cpumask we need, and which handles the UP case by returning a NULL pointer.
So use that and handle NULL.
Also avoid the #ifdef by allocating a cpumask_var and copying
into it, rather than sharing the mask.
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Fixes: 6106c0f824 ("staging: lustre: lnet: convert selftest to use workqueues")
Signed-off-by: NeilBrown <neilb@suse.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Because this is a post-op instead of a pre-op, then it means we check
if knl_buffer[-1] is a space. It doesn't really hurt anything, but
it causes a static checker warning so let's fix it.
Fixes: d7e09d0397 ("staging: add Lustre file system client support")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since kfree(NULL) is safe there is no need to place a check before it.
Issue reported by checkpatch.pl
Signed-off-by: Sumit Pundir <pundirsumit11@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use kcalloc for allocating an array instead of kzalloc with
multiply. kcalloc is the preferred API. Issue reported by
checkpatch.pl
Signed-off-by: Sumit Pundir <pundirsumit11@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch properly left aligns all member identifiers in every
struct defined in obd_class.h for better readability.
Signed-off-by: Fabian Huegel <fabian_huegel@web.de>
Signed-off-by: Christoph Volkert <linux@christoph-volkert.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Most multi-line comments started on the first line, but the preferred
linux kernel style is to start multi-line comments on the second line.
Some comments became less readable after the change, so we changed them
to single-line comments.
Signed-off-by: Fabian Huegel <fabian_huegel@web.de>
Signed-off-by: Christoph Volkert <linux@christoph-volkert.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
gcc warns that the latest workqueue change leads to returning an
uninitialized variable:
drivers/staging/lustre/lnet/selftest/module.c: In function 'lnet_selftest_init':
drivers/staging/lustre/lnet/selftest/module.c:98:10: error: 'rc' may be used uninitialized in this function [-Werror=maybe-uninitialized]
A failure from alloc_ordered_workqueue() tends to indicate an
out-of-memory condition, so return -ENOMEM in both cases.
The second error path was a preexisting bug, where we always
returned zero after a kvmalloc_array() failure.
Fixes: 6106c0f824 ("staging: lustre: lnet: convert selftest to use workqueues")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With the latest patches from Neil, we are now getting a build warning:
drivers/staging/lustre/lnet/selftest/module.c: In function ‘lnet_selftest_init’:
drivers/staging/lustre/lnet/selftest/module.c:98:10: warning: ‘rc’ may be used uninitialized in this function [-Wmaybe-uninitialized]
return rc;
^~
So fix this up by giving a default error value to rc.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Cc: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch right aligns all backslashes in multi-line macros
in obd_class.h for better readability.
Signed-off-by: Fabian Huegel <fabian_huegel@web.de>
Signed-off-by: Christoph Volkert <linux@christoph-volkert.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fixed four lines that went over the 80 character limit
to reduce checkpatch warnings.
Signed-off-by: Fabian Huegel <fabian_huegel@web.de>
Signed-off-by: Christoph Volkert <linux@christoph-volkert.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Checkpatch was complaining about missing identifier names in function
declarations. So we added the missing names according to the names in
the respective function implementation. *obd_import* was sometimes named
*import* and sometimes *imp* (in genops.c), so to avoid confusion we just
named it imp everywhere.
Signed-off-by: Fabian Huegel <fabian_huegel@web.de>
Signed-off-by: Christoph Volkert <linux@christoph-volkert.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This macro was only used in four places to declare two variables.
It saved one line of code, but in our opinion hurt readability.
So we removed the macro, substituting every occurrence with the
declaration of the two variables (like the preprocessor would have done).
Signed-off-by: Fabian Huegel <fabian_huegel@web.de>
Signed-off-by: Christoph Volkert <linux@christoph-volkert.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some complex multi-line macros were not enclosed by a do-while(0),
so we fixed that.
Signed-off-by: Fabian Huegel <fabian_huegel@web.de>
Signed-off-by: Christoph Volkert <linux@christoph-volkert.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are now no users. workqueues are doing the job
that this used to do.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of the cfs workitem library, use workqueues.
As lnet wants to provide a cpu mask of allowed cpus, it
needs to be a WQ_UNBOUND work queue so that tasks can
run on cpus other than where they were submitted.
This patch also exported apply_workqueue_attrs() which is
a documented part of the workqueue API, that isn't currently
exported. lustre needs it to allow workqueue thread to be limited
to a subset of CPUs.
Acked-by: Tejun Heo <tj@kernel.org> (for export of apply_workqueue_attrs)
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We need to consistently enforce that keyed hashes cannot be used without
setting the key. To do this we need a reliable way to determine whether
a given hash algorithm is keyed or not. AF_ALG currently does this by
checking for the presence of a ->setkey() method. However, this is
actually slightly broken because the CRC-32 algorithms implement
->setkey() but can also be used without a key. (The CRC-32 "key" is not
actually a cryptographic key but rather represents the initial state.
If not overridden, then a default initial state is used.)
Prepare to fix this by introducing a flag CRYPTO_ALG_OPTIONAL_KEY which
indicates that the algorithm has a ->setkey() method, but it is not
required to be called. Then set it on all the CRC-32 algorithms.
The same also applies to the Adler-32 implementation in Lustre.
Also, the cryptd and mcryptd templates have to pass through the flag
from their underlying algorithm.
Cc: stable@vger.kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
LIBCFS_ALLOC
LIBCFS_ALLOC_ATOMIC
LIBCFS_ALLOC_POST
LIBCFS_CPT_ALLOC
LIBCFS_FREE
are no longer used, and so are removed.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
NeilBrown