Servers have a finite amount of memory to store NFSv4 open and lock
owners. Moreover, servers may have a difficult time determining when
they can reap their state owner table, thanks to gray areas in the
NFSv4 protocol specification. Thus clients should be careful to reuse
state owners when possible.
Currently Linux is not too careful. When a user has closed all her
files on one mount point, the state owner's reference count goes to
zero, and it is released. The next OPEN allocates a new one. A
workload that serially opens and closes files can run through a large
number of open owners this way.
When a state owner's reference count goes to zero, slap it onto a free
list for that nfs_server, with an expiry time. Garbage collect before
looking for a state owner. This makes state owners for active users
available for re-use.
Now that there can be unused state owners remaining at umount time,
purge the state owner free list when a server is destroyed. Also be
sure not to reclaim unused state owners during state recovery.
This change has benefits for the client as well. For some workloads,
this approach drops the number of OPEN_CONFIRM calls from the same as
the number of OPEN calls, down to just one. This reduces wire traffic
and thus open(2) latency. Before this patch, untarring a kernel
source tarball shows the OPEN_CONFIRM call counter steadily increasing
through the test. With the patch, the OPEN_CONFIRM count remains at 1
throughout the entire untar.
As long as the expiry time is kept short, I don't think garbage
collection should be terribly expensive, although it does bounce the
clp->cl_lock around a bit.
[ At some point we should rationalize the use of the nfs_server
->destroy method. ]
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
[Trond: Fixed a garbage collection race and a few efficiency issues]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
There's no longer a need to check the so_server field in the state
owner, because nowadays the RB tree we search for state owners
contains owners for that only server.
Make nfs4_find_state_owners_locked() use the same tree searching logic
as nfs4_insert_state_owner_locked().
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The NFSv4 bitmap size is unbounded: a server can return an arbitrary
sized bitmap in an FATTR4_WORD0_ACL request. Replace using the
nfs4_fattr_bitmap_maxsz as a guess to the maximum bitmask returned by a server
with the inclusion of the bitmap (xdr length plus bitmasks) and the acl data
xdr length to the (cached) acl page data.
This is a general solution to commit e5012d1f "NFSv4.1: update
nfs4_fattr_bitmap_maxsz" and fixes hitting a BUG_ON in xdr_shrink_bufhead
when getting ACLs.
Fix a bug in decode_getacl that returned -EINVAL on ACLs > page when getxattr
was called with a NULL buffer, preventing ACL > PAGE_SIZE from being retrieved.
Cc: stable@kernel.org
Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
This change modifies filelayout_get_dense_offset() to use the functions
in math64.h and thus avoid a 32-bit platform compile error trying to
use do_div() on an s64 type.
Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
Reviewed-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
We are either initializing or resetting a session. Initialize or reset
the session slot tables accordingly.
Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Setting the security context of a NFSv4 mount via the context= mount
option is currently broken. The NFSv4 codepath allocates a parsed
options struct, and then parses the mount options to fill it. It
eventually calls nfs4_remote_mount which calls security_init_mnt_opts.
That clobbers the lsm_opts struct that was populated earlier. This bug
also looks like it causes a small memory leak on each v4 mount where
context= is used.
Fix this by moving the initialization of the lsm_opts into
nfs_alloc_parsed_mount_data. Also, add a destructor for
nfs_parsed_mount_data to make it easier to free all of the allocations
hanging off of it, and to ensure that the security_free_mnt_opts is
called whenever security_init_mnt_opts is.
I believe this regression was introduced quite some time ago, probably
by commit c02d7adf.
Cc: stable@vger.kernel.org
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
From c6d615d2b97fe305cbf123a8751ced859dca1d5e Mon Sep 17 00:00:00 2001
From: NeilBrown <neilb@suse.de>
Date: Wed, 16 Nov 2011 09:39:05 +1100
Subject: [PATCH] NFS - fix recent breakage to NFS error handling.
commit 02c24a8218 made a small and
presumably unintended change to write error handling in NFS.
Previously an error from filemap_write_and_wait_range would only be of
interest if nfs_file_fsync did not return an error. After this commit,
an error from filemap_write_and_wait_range would mean that (the rest of)
nfs_file_fsync would not even be called.
This means that:
1/ you are more likely to see EIO than e.g. EDQUOT or ENOSPC.
2/ NFS_CONTEXT_ERROR_WRITE remains set for longer so more writes are
synchronous.
This patch restores previous behaviour.
Cc: stable@kernel.org
Cc: Josef Bacik <josef@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Lukas Razik <linux@razik.name> reports that on his SPARC system,
booting with an NFS root file system stopped working after commit
56463e50 "NFS: Use super.c for NFSROOT mount option parsing."
We found that the network switch to which Lukas' client was attached
was delaying access to the LAN after the client's NIC driver reported
that its link was up. The delay was longer than the timeouts used in
the NFS client during mounting.
NFSROOT worked for Lukas before commit 56463e50 because in those
kernels, the client's first operation was an rpcbind request to
determine which port the NFS server was listening on. When that
request failed after a long timeout, the client simply selected the
default NFS port (2049). By that time the switch was allowing access
to the LAN, and the mount succeeded.
Neither of these client behaviors is desirable, so reverting 56463e50
is really not a choice. Instead, introduce a mechanism that retries
the NFSROOT mount request several times. This is the same tactic that
normal user space NFS mounts employ to overcome server and network
delays.
Signed-off-by: Lukas Razik <linux@razik.name>
[ cel: match kernel coding style, add proper patch description ]
[ cel: add exponential back-off ]
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Lukas Razik <linux@razik.name>
Cc: stable@kernel.org # > 2.6.38
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Instead of hacking specific service names into gss_encode_v1_msg, we should
just allow the caller to specify the service name explicitly.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: J. Bruce Fields <bfields@redhat.com>
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
fix CAN MAINTAINERS SCM tree type
mwifiex: fix crash during simultaneous scan and connect
b43: fix regression in PIO case
ath9k: Fix kernel panic in AR2427 in AP mode
CAN MAINTAINERS update
net: fsl: fec: fix build for mx23-only kernel
sch_qfq: fix overflow in qfq_update_start()
Revert "Bluetooth: Increase HCI reset timeout in hci_dev_do_close"
bitmap size sanity checks should be done *before* allocating ->s_root;
there their cleanup on failure would be correct. As it is, we do iput()
on root inode, but leak the root dentry...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Josh Boyer <jwboyer@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is the temporary simple fix for 3.2, we need more changes in this
area.
1. do_signal_stop() assumes that the running untraced thread in the
stopped thread group is not possible. This was our goal but it is
not yet achieved: a stopped-but-resumed tracee can clone the running
thread which can initiate another group-stop.
Remove WARN_ON_ONCE(!current->ptrace).
2. A new thread always starts with ->jobctl = 0. If it is auto-attached
and this group is stopped, __ptrace_unlink() sets JOBCTL_STOP_PENDING
but JOBCTL_STOP_SIGMASK part is zero, this triggers WANR_ON(!signr)
in do_jobctl_trap() if another debugger attaches.
Change __ptrace_unlink() to set the artificial SIGSTOP for report.
Alternatively we could change ptrace_init_task() to copy signr from
current, but this means we can copy it for no reason and hide the
possible similar problems.
Acked-by: Tejun Heo <tj@kernel.org>
Cc: <stable@kernel.org> [3.1]
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Test-case:
int main(void)
{
int pid, status;
pid = fork();
if (!pid) {
for (;;) {
if (!fork())
return 0;
if (waitpid(-1, &status, 0) < 0) {
printf("ERR!! wait: %m\n");
return 0;
}
}
}
assert(ptrace(PTRACE_ATTACH, pid, 0,0) == 0);
assert(waitpid(-1, NULL, 0) == pid);
assert(ptrace(PTRACE_SETOPTIONS, pid, 0,
PTRACE_O_TRACEFORK) == 0);
do {
ptrace(PTRACE_CONT, pid, 0, 0);
pid = waitpid(-1, NULL, 0);
} while (pid > 0);
return 1;
}
It fails because ->real_parent sees its child in EXIT_DEAD state
while the tracer is going to change the state back to EXIT_ZOMBIE
in wait_task_zombie().
The offending commit is 823b018e which moved the EXIT_DEAD check,
but in fact we should not blame it. The original code was not
correct as well because it didn't take ptrace_reparented() into
account and because we can't really trust ->ptrace.
This patch adds the additional check to close this particular
race but it doesn't solve the whole problem. We simply can't
rely on ->ptrace in this case, it can be cleared if the tracer
is multithreaded by the exiting ->parent.
I think we should kill EXIT_DEAD altogether, we should always
remove the soon-to-be-reaped child from ->children or at least
we should never do the DEAD->ZOMBIE transition. But this is too
complex for 3.2.
Reported-and-tested-by: Denys Vlasenko <vda.linux@googlemail.com>
Tested-by: Lukasz Michalik <lmi@ift.uni.wroc.pl>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: <stable@kernel.org> [3.0+]
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This reverts commit 93b2ec0128.
The call to "schedule_work()" in rtc_initialize_alarm() happens too
early, and can cause oopses at bootup
Neil Brown explains why we do it:
"If you set an alarm in the future, then shutdown and boot again after
that time, then you will end up with a timer_queue node which is in
the past.
When this happens the queue gets stuck. That entry-in-the-past won't
get removed until and interrupt happens and an interrupt won't happen
because the RTC only triggers an interrupt when the alarm is "now".
So you'll find that e.g. "hwclock" will always tell you that
'select' timed out.
So we force the interrupt work to happen at the start just in case."
and has a patch that convert it to do things in-process rather than with
the worker thread, but right now it's too late to play around with this,
so we just revert the patch that caused problems for now.
Reported-by: Sander Eikelenboom <linux@eikelenboom.it>
Requested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Requested-by: John Stultz <john.stultz@linaro.org>
Cc: Neil Brown <neilb@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Turned out the ntlmv2 (default security authentication)
upgrade was harder to test than expected, and we ran
out of time to test against Apple and a few other servers
that we wanted to. Delay upgrade of default security
from ntlm to ntlmv2 (on mount) to 3.3. Still works
fine to specify it explicitly via "sec=ntlmv2" so this
should be fine.
Acked-by: Jeff Layton <jlayton@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
The current check looks to see if the RFC1002 length is larger than
CIFSMaxBufSize, and fails if it is. The buffer is actually larger than
that by MAX_CIFS_HDR_SIZE.
This bug has been around for a long time, but the fact that we used to
cap the clients MaxBufferSize at the same level as the server tended
to paper over it. Commit c974befa changed that however and caused this
bug to bite in more cases.
Reported-and-Tested-by: Konstantinos Skarlatos <k.skarlatos@gmail.com>
Tested-by: Shirish Pargaonkar <shirishpargaonkar@gmail.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
This reverts commit c0afabd3d5.
It causes failures on Toshiba laptops - instead of disabling the alarm,
it actually seems to enable it on the affected laptops, resulting in
(for example) the laptop powering on automatically five minutes after
shutdown.
There's a patch for it that appears to work for at least some people,
but it's too late to play around with this, so revert for now and try
again in the next merge window.
See for example
http://bugs.debian.org/652869
Reported-and-bisected-by: Andreas Friedrich <afrie@gmx.net> (Toshiba Tecra)
Reported-by: Antonio-M. Corbi Bellot <antonio.corbi@ua.es> (Toshiba Portege R500)
Reported-by: Marco Santos <marco.santos@waynext.com> (Toshiba Portege Z830)
Reported-by: Christophe Vu-Brugier <cvubrugier@yahoo.fr> (Toshiba Portege R830)
Cc: Jonathan Nieder <jrnieder@gmail.com>
Requested-by: John Stultz <john.stultz@linaro.org>
Cc: stable@kernel.org # for the versions that applied this
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
vfork parent uninterruptibly and unkillably waits for its child to
exec/exit. This wait is of unbounded length. Ignore such waits
in the hung_task detector.
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reported-by: Sasha Levin <levinsasha928@gmail.com>
LKML-Reference: <1325344394.28904.43.camel@lappy>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: John Kacur <jkacur@redhat.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 1e39f384bb ("evm: fix build problems") makes the stub version
of security_old_inode_init_security() return 0 when CONFIG_SECURITY is
not set.
But that makes callers such as reiserfs_security_init() assume that
security_old_inode_init_security() has set name, value, and len
arguments properly - but security_old_inode_init_security() left them
uninitialized which then results in interesting failures.
Revert security_old_inode_init_security() to the old behavior of
returning EOPNOTSUPP since both callers (reiserfs and ocfs2) handle this
just fine.
[ Also fixed the S_PRIVATE(inode) case of the actual non-stub
security_old_inode_init_security() function to return EOPNOTSUPP
for the same reason, as pointed out by Mimi Zohar.
It got incorrectly changed to match the new function in commit
fb88c2b6cb: "evm: fix security/security_old_init_security return
code". - Linus ]
Reported-by: Jorge Bastos <mysql.jorge@decimal.pt>
Acked-by: James Morris <jmorris@namei.org>
Acked-by: Mimi Zohar <zohar@us.ibm.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
As pointed out by Joe Perches the SCM tree type was missing in my patch.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
CC: Oliver Hartkopp <oliver.hartkopp@volkswagen.de>
CC: Urs Thuermann <urs.thuermann@volkswagen.de>
CC: Wolfgang Grandegger <wg@grandegger.com>
CC: Marc Kleine-Budde <mkl@pengutronix.de>
CC: linux-can@vger.kernel.org
If 'iw connect' command is fired when driver is already busy in
serving 'iw scan' command, ssid specific scan operation for connect
is skipped. In this case cmd wait queue handler gets called with no
command in queue (i.e. adapter->cmd_queued = NULL).
This patch adds a NULL check in mwifiex_wait_queue_complete()
routine to fix crash observed during simultaneous scan and assoc
operations.
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch fixes the regression, introduced by
commit 17030f48e3
From: Rafał Miłecki <zajec5@gmail.com>
Date: Thu, 11 Aug 2011 17:16:27 +0200
Subject: [PATCH] b43: support new RX header, noticed to be used in 598.314+ fw
in PIO case.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Update the CAN MAINTAINERS section:
- point out active maintainers
- pull the CAN driver discussion away from netdev ML
- point to the new CAN web site on gitorious.org
- add CAN development git repository URL to submit patches
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
CC: Oliver Hartkopp <oliver.hartkopp@volkswagen.de>
CC: Urs Thuermann <urs.thuermann@volkswagen.de>
CC: Wolfgang Grandegger <wg@grandegger.com>
CC: Marc Kleine-Budde <mkl@pengutronix.de>
CC: linux-can@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
If one only selects mx23-based boards, compile fails:
drivers/net/ethernet/freescale/fec.c:410:2: error: 'FEC_HASH_TABLE_HIGH' undeclared (first use in this function)
drivers/net/ethernet/freescale/fec.c:411:2: error: 'FEC_HASH_TABLE_LOW' undeclared (first use in this function)
This is because fec.h uses CONFIG_SOC_IMX28 to determine the register
layout of the core which makes sense since the MX23 does not have a fec.
However, Kconfig uses the broader ARCH_MXS symbol and this way even
makes the fec-driver default for MX23. Adapt Kconfig to use the more
precise SOC_IMX28 as well.
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: Fabio Estevam <fabio.estevam@freescale.com>
Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Cc: Shawn Guo <shawn.guo@linaro.org>
Cc: David S. Miller <davem@davemloft.net>
Acked-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
grp->slot_shift is between 22 and 41, so using 32bit wide variables is
probably a typo.
This could explain QFQ hangs Dave reported to me, after 2^23 packets ?
(23 = 64 - 41)
Reported-by: Dave Taht <dave.taht@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
CC: Dave Taht <dave.taht@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
netfilter: ctnetlink: fix timeout calculation
ipvs: try also real server with port 0 in backup server
skge: restore rx multicast filter on resume and after config changes
mlx4_en: nullify cq->vector field when closing completion queue
The current gspca core code has a regression where it no longer properly
falls back to lower alt settings when there is not enough bandwidth.
This causes many iso based usb-1 cameras to not work when plugged into a
usb2 hub or a sandybridge chipset motherboard!
This patch fixes this.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
It was found (by Sasha) that if you use a futex located in the gate
area we get stuck in an uninterruptible infinite loop, much like the
ZERO_PAGE issue.
While looking at this problem, PeterZ realized you'll get into similar
trouble when hitting any install_special_pages() mapping. And are there
still drivers setting up their own special mmaps without page->mapping,
and without special VM or pte flags to make get_user_pages fail?
In most cases, if page->mapping is NULL, we do not need to retry at all:
Linus points out that even /proc/sys/vm/drop_caches poses no problem,
because it ends up using remove_mapping(), which takes care not to
interfere when the page reference count is raised.
But there is still one case which does need a retry: if memory pressure
called shmem_writepage in between get_user_pages_fast dropping page
table lock and our acquiring page lock, then the page gets switched from
filecache to swapcache (and ->mapping set to NULL) whatever the refcount.
Fault it back in to get the page->mapping needed for key->shared.inode.
Reported-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The sanity check (timeout < 0) never works; the dividend is unsigned
and so is the division, which should have been a signed division.
long timeout = (ct->timeout.expires - jiffies) / HZ;
if (timeout < 0)
timeout = 0;
This patch converts the time values to signed for the division.
Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
We should not forget to try for real server with port 0
in the backup server when processing the sync message. We should
do it in all cases because the backup server can use different
forwarding method.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Restore skge hardware registers for multicast filtering to their
appropriate values after system resume and after hardware restarts
that are done when changing certain settings.
Signed-off-by: Florian Zumbiehl <florz@florz.de>
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Caused loss of connectivity when changing ring size.
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
* git://www.linux-watchdog.org/linux-watchdog:
watchdog: iTCO_wdt.c - problems with newer hardware due to SMI clearing (part 2)
watchdog: hpwdt: Changes to handle NX secure bit in 32bit path
watchdog: sp805: Fix section mismatch in ID table.
watchdog: move coh901327 state holders
