Instead of only considering SD_WAKE_AFFINE | SD_PREFER_SIBLING
domains also allow all SD_PREFER_SIBLING domains below a
SD_WAKE_AFFINE domain to change the affinity target.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
LKML-Reference: <20091112145610.909723612@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Clean up the new affine to idle sibling bits while trying to
grok them. Should not have any function differences.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
LKML-Reference: <20091112145610.832503781@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Originally task_s/utime() were designed to return clock_t but
later changed to return cputime_t by following commit:
commit efe567fc82
Author: Christian Borntraeger <borntraeger@de.ibm.com>
Date: Thu Aug 23 15:18:02 2007 +0200
It only changed the type of return value, but not the
implementation. As the result the granularity of task_s/utime()
is still that of clock_t, not that of cputime_t.
So using task_s/utime() in __exit_signal() makes values
accumulated to the signal struct to be rounded and coarse
grained.
This patch removes casts to clock_t in task_u/stime(), to keep
granularity of cputime_t over the calculation.
v2:
Use div_u64() to avoid error "undefined reference to `__udivdi3`"
on some 32bit systems.
Signed-off-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: xiyou.wangcong@gmail.com
Cc: Spencer Candland <spencer@bluehost.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Stanislaw Gruszka <sgruszka@redhat.com>
LKML-Reference: <4AFB9029.9000208@jp.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
From the code in rt_mutex_setprio(), it is evident that the
intention is that task's with a RT 'prio' value as a consequence
of receiving a PI boost also have their 'sched_class' field set
to '&rt_sched_class'.
However, Peter noticed that the code in __setscheduler() could
result in this intention being frustrated. Fix it.
Reported-by: Peter Williams <pwil3058@bigpond.net.au>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
LKML-Reference: <1257880321.4108.457.camel@laptop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Commit 1b9508f, "Rate-limit newidle" has been confirmed to fix
the netperf UDP loopback regression reported by Alex Shi.
This is a cleanup and a fix:
- moved to a more out of the way spot
- fix to ensure that balancing doesn't try to balance
runqueues which haven't gone online yet, which can
mess up CPU enumeration during boot.
Reported-by: Alex Shi <alex.shi@intel.com>
Reported-by: Zhang, Yanmin <yanmin_zhang@linux.intel.com>
Signed-off-by: Mike Galbraith <efault@gmx.de>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: <stable@kernel.org> # .32.x: a1f84a3: sched: Check for an idle shared cache
Cc: <stable@kernel.org> # .32.x: 1b9508f: sched: Rate-limit newidle
Cc: <stable@kernel.org> # .32.x: fd21073: sched: Fix affinity logic
Cc: <stable@kernel.org> # .32.x
LKML-Reference: <1257821402.5648.17.camel@marge.simson.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
In 15934a3732,
field last_tick_seen is added to struct rq.
But it is unused now.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Guillaume Chazarain <guichaz@yahoo.fr>
LKML-Reference: <4AE6A513.6010100@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Rate limit newidle to migration_cost. It's a win for all
stages of sysbench oltp tests.
Signed-off-by: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When waking affine, check for an idle shared cache, and if
found, wake to that CPU/sibling instead of the waker's CPU.
This improves pgsql+oltp ramp up by roughly 8%. Possibly more
for other loads, depending on overlap. The trade-off is a
roughly 1% peak downturn if tasks are truly synchronous.
Signed-off-by: Mike Galbraith <efault@gmx.de>
Cc: Arjan van de Ven <arjan@infradead.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: <stable@kernel.org>
LKML-Reference: <1256654138.17752.7.camel@marge.simson.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Currently partition_sched_domains() takes a 'struct cpumask
*doms_new' which is a kmalloc'ed array of cpumask_t. You can't
have such an array if 'struct cpumask' is undefined, as we plan
for CONFIG_CPUMASK_OFFSTACK=y.
So, we make this an array of cpumask_var_t instead: this is the
same for the CONFIG_CPUMASK_OFFSTACK=n case, but requires
multiple allocations for the CONFIG_CPUMASK_OFFSTACK=y case.
Hence we add alloc_sched_domains() and free_sched_domains()
functions.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <200911031453.40668.rusty@rustcorp.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
find_lowest_rq() wants to call pick_optimal_cpu() on the
intersection of sched_domain_span(sd) and lowest_mask. Rather
than doing a cpus_and into a temporary, we can open-code it.
This actually makes the code slightly clearer, IMHO.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Gregory Haskins <ghaskins@novell.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
LKML-Reference: <200911031453.15350.rusty@rustcorp.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Peter Zijlstra suggested that we remove USER_SCHED at:
http://lkml.org/lkml/2009/3/21/67
Removing USER_SCHED removes a lot of code from the scheduler
and simplifies the code.
We already have the ability to do user based classification
which is tightened using PAM in userspace.
Schedule USER_SCHED for removal in 2.6.34
Signed-off-by: Dhaval Giani <dhaval@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Bharata B Rao <bharata@linux.vnet.ibm.com>
Cc: Serge E. Hallyn <serue@us.ibm.com>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
LKML-Reference: <20091103214544.GI5495@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
cpu_nr_migrations() is not used, remove it.
Signed-off-by: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <4AF12A66.6020609@ct.jp.nec.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
CPU time of a guest is always accounted in 'user' time
without concern for the nice value of its counterpart
process although the guest is scheduled under the nice
value.
This patch fixes the defect and accounts cpu time of
a niced guest in 'nice' time as same as a niced process.
And also the patch adds 'guest_nice' to cpuacct. The
value provides niced guest cpu time which is like 'nice'
to 'user'.
The original discussions can be found here:
http://www.mail-archive.com/kvm@vger.kernel.org/msg23982.htmlhttp://www.mail-archive.com/kvm@vger.kernel.org/msg23860.html
Signed-off-by: Ryota Ozaki <ozaki.ryota@gmail.com>
Acked-by: Avi Kivity <avi@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1256314810-7897-1-git-send-email-ozaki.ryota@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus:
move virtrng_remove to .devexit.text
move virtballoon_remove to .devexit.text
virtio_blk: Revert serial number support
virtio: let header files include virtio_ids.h
virtio_blk: revert QUEUE_FLAG_VIRT addition
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (21 commits)
niu: VLAN_ETH_HLEN should be used to make sure that the whole MAC header was copied to the head buffer in the Vlan packets case
KS8851: Fix ks8851_set_rx_mode() for IFF_MULTICAST
KS8851: Fix MAC address write order
KS8851: Add soft reset at probe time
net: fix section mismatch in fec.c
net: Fix struct inet_timewait_sock bitfield annotation
tcp: Try to catch MSG_PEEK bug
net: Fix IP_MULTICAST_IF
bluetooth: static lock key fix
bluetooth: scheduling while atomic bug fix
tcp: fix TCP_DEFER_ACCEPT retrans calculation
tcp: reduce SYN-ACK retrans for TCP_DEFER_ACCEPT
tcp: accept socket after TCP_DEFER_ACCEPT period
Revert "tcp: fix tcp_defer_accept to consider the timeout"
AF_UNIX: Fix deadlock on connecting to shutdown socket
ethoc: clear only pending irqs
ethoc: inline regs access
vmxnet3: use dev_dbg, fix build for CONFIG_BLOCK=n
virtio_net: use dev_kfree_skb_any() in free_old_xmit_skbs()
be2net: fix support for PCI hot plug
...
The function virtrng_remove is used only wrapped by __devexit_p so define
it using __devexit.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The function virtballoon_remove is used only wrapped by __devexit_p so
define it using __devexit.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This reverts "Add serial number support for virtio_blk, V4a".
Turns out that virtio_pci, lguest and s/390 all have an 8 bit limit
on virtio config space, so noone could ever use this.
This is coming back later in a cleaner form.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: john cooper <john.cooper@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Rusty,
commit 3ca4f5ca73
virtio: add virtio IDs file
moved all device IDs into a single file. While the change itself is
a very good one, it can break userspace applications. For example
if a userspace tool wanted to get the ID of virtio_net it used to
include virtio_net.h. This does no longer work, since virtio_net.h
does not include virtio_ids.h.
This patch moves all "#include <linux/virtio_ids.h>" from the C
files into the header files, making the header files compatible with
the old ones.
In addition, this patch exports virtio_ids.h to userspace.
CC: Fernando Luis Vazquez Cao <fernando@oss.ntt.co.jp>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It seems like the addition of QUEUE_FLAG_VIRT caueses major performance
regressions for Fedora users:
https://bugzilla.redhat.com/show_bug.cgi?id=509383https://bugzilla.redhat.com/show_bug.cgi?id=505695
while I can't reproduce those extreme regressions myself I think the flag
is wrong.
Rationale:
QUEUE_FLAG_VIRT expands to QUEUE_FLAG_NONROT which casus the queue
unplugged immediately. This is not a good behaviour for at least
qemu and kvm where we do have significant overhead for every
I/O operations. Even with all the latested speeups (native AIO,
MSI support, zero copy) we can only get native speed for up to 128kb
I/O requests we already are down to 66% of native performance for 4kb
requests even on my laptop running the Intel X25-M SSD for which the
QUEUE_FLAG_NONROT was designed.
If we ever get virtio-blk overhead low enough that this flag makes
sense it should only be set based on a feature flag set by the host.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
* 'for-linus' of git://git.infradead.org/users/eparis/notify:
dnotify: ignore FS_EVENT_ON_CHILD
inotify: fix coalesce duplicate events into a single event in special case
inotify: deprecate the inotify kernel interface
fsnotify: do not set group for a mark before it is on the i_list
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
Input: hp_sdc_rtc - fix test in hp_sdc_rtc_read_rt()
Input: atkbd - consolidate force release quirks for volume keys
Input: logips2pp - model 73 is actually TrackMan FX
Input: i8042 - add Sony Vaio VGN-FZ240E to the nomux list
Input: fix locking issue in /proc/bus/input/ handlers
Input: atkbd - postpone restoring LED/repeat rate at resume
Input: atkbd - restore resetting LED state at startup
Input: i8042 - make pnp_data_busted variable boolean instead of int
Input: synaptics - add another Protege M300 to rate blacklist
* 'kvm-updates/2.6.32' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: Prevent kvm_init from corrupting debugfs structures
KVM: MMU: fix pointer cast
KVM: use proper hrtimer function to retrieve expiration time
* git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-2.6-dm:
dm snapshot: allow chunk size to be less than page size
dm snapshot: use unsigned integer chunk size
dm snapshot: lock snapshot while supplying status
dm exception store: fix failed set_chunk_size error path
dm snapshot: require non zero chunk size by end of ctr
dm: dec_pending needs locking to save error value
dm: add missing del_gendisk to alloc_dev error path
dm log: userspace fix incorrect luid cast in userspace_ctr
dm snapshot: free exception store on init failure
dm snapshot: sort by chunk size to fix race
Increase TEST_SUSPEND_SECONDS to 10 so the warning in
suspend_test_finish() doesn't annoy the users of slower systems so much.
Also, make the warning print the suspend-resume cycle time, so that we
know why the warning actually triggered.
Patch prepared during the hacking session at the Kernel Summit in Tokyo.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This fixes a compile bug introduced in
6ef297f (ARM: 5720/1: Move MMCI header to amba include dir)
That commit moved arch/arm/include/asm/mach/mmc.h to
include/linux/amba/mmci.h. Just removing the include was enough.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: Bill Gatliff <bgat@billgatliff.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Pierre Ossman <drzeus@drzeus.cx>
Cc: linux-arm-kernel@lists.infradead.org
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'sh/for-2.6.32' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
sh: Kill off stray HAVE_FTRACE_SYSCALLS reference.
sh: Remove BKL from landisk gio.
sh: disabled cache handling fix.
sh: Fix up single page flushing to use PAGE_SIZE.
Fix a (small) memory leak in one of the error paths of the NFS mount
options parsing code.
Regression introduced in 2.6.30 by commit a67d18f (NFS: load the
rpc/rdma transport module automatically).
Reported-by: Yinghai Lu <yinghai@kernel.org>
Reported-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch fixes a null pointer exception in pipe_rdwr_open() which
generates the stack trace:
> Unable to handle kernel NULL pointer dereference at 0000000000000028 RIP:
> [<ffffffff802899a5>] pipe_rdwr_open+0x35/0x70
> [<ffffffff8028125c>] __dentry_open+0x13c/0x230
> [<ffffffff8028143d>] do_filp_open+0x2d/0x40
> [<ffffffff802814aa>] do_sys_open+0x5a/0x100
> [<ffffffff8021faf3>] sysenter_do_call+0x1b/0x67
The failure mode is triggered by an attempt to open an anonymous
pipe via /proc/pid/fd/* as exemplified by this script:
=============================================================
while : ; do
{ echo y ; sleep 1 ; } | { while read ; do echo z$REPLY; done ; } &
PID=$!
OUT=$(ps -efl | grep 'sleep 1' | grep -v grep |
{ read PID REST ; echo $PID; } )
OUT="${OUT%% *}"
DELAY=$((RANDOM * 1000 / 32768))
usleep $((DELAY * 1000 + RANDOM % 1000 ))
echo n > /proc/$OUT/fd/1 # Trigger defect
done
=============================================================
Note that the failure window is quite small and I could only
reliably reproduce the defect by inserting a small delay
in pipe_rdwr_open(). For example:
static int
pipe_rdwr_open(struct inode *inode, struct file *filp)
{
msleep(100);
mutex_lock(&inode->i_mutex);
Although the defect was observed in pipe_rdwr_open(), I think it
makes sense to replicate the change through all the pipe_*_open()
functions.
The core of the change is to verify that inode->i_pipe has not
been released before attempting to manipulate it. If inode->i_pipe
is no longer present, return ENOENT to indicate so.
The comment about potentially using atomic_t for i_pipe->readers
and i_pipe->writers has also been removed because it is no longer
relevant in this context. The inode->i_mutex lock must be used so
that inode->i_pipe can be dealt with correctly.
Signed-off-by: Earl Chew <earl_chew@agilent.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In ks8851_set_rx_mode() the case handling IFF_MULTICAST was also setting
the RXCR1_AE bit by accident. This meant that all unicast frames where
being accepted by the device. Remove RXCR1_AE from this case.
Note, RXCR1_AE was also masking a problem with setting the MAC address
properly, so needs to be applied after fixing the MAC write order.
Fixes a bug reported by Doong, Ping of Micrel. This version of the
patch avoids setting RXCR1_ME for all cases.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
The MAC address register was being written in the wrong order, so add
a new address macro to convert mac-address byte to register address and
a ks8851_wrreg8() function to write each byte without having to worry
about any difficult byte swapping.
Fixes a bug reported by Doong, Ping of Micrel.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Issue a full soft reset at probe time.
This was reported by Doong Ping of Micrel, but no explanation of why this
is necessary or what bug it is fixing. Add it as it does not seem to hurt
the current driver and ensures that the device is in a known state when we
start setting it up.
Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
fec_enet_init is called by both fec_probe and fec_resume, so it
shouldn't be marked as __init.
Signed-off-by: Steven King <sfking@fdwdc.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
commit 9e337b0f (net: annotate inet_timewait_sock bitfields)
added 4/8 bytes in struct inet_timewait_sock.
Fix this by declaring tw_ipv6_offset in the 'flags' bitfield
The 14 bits hole is named tw_pad to make it cleary apparent.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch tries to print out more information when we hit the
MSG_PEEK bug in tcp_recvmsg. It's been around since at least
2005 and it's about time that we finally fix it.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
When renaming kernel_fpu_using to irq_fpu_usable, the semantics of the
function is changed too, from mesuring whether kernel is using FPU,
that is, the FPU is NOT available, to measuring whether FPU is usable,
that is, the FPU is available.
But the usage of irq_fpu_usable in aesni-intel_glue.c is not changed
accordingly. This patch fixes this.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
ipv4/ipv6 setsockopt(IP_MULTICAST_IF) have dubious __dev_get_by_index() calls.
This function should be called only with RTNL or dev_base_lock held, or reader
could see a corrupt hash chain and eventually enter an endless loop.
Fix is to call dev_get_by_index()/dev_put().
If this happens to be performance critical, we could define a new dev_exist_by_index()
function to avoid touching dev refcount.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix TCP_DEFER_ACCEPT conversion between seconds and
retransmission to match the TCP SYN-ACK retransmission periods
because the time is converted to such retransmissions. The old
algorithm selects one more retransmission in some cases. Allow
up to 255 retransmissions.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Change SYN-ACK retransmitting code for the TCP_DEFER_ACCEPT
users to not retransmit SYN-ACKs during the deferring period if
ACK from client was received. The goal is to reduce traffic
during the deferring period. When the period is finished
we continue with sending SYN-ACKs (at least one) but this time
any traffic from client will change the request to established
socket allowing application to terminate it properly.
Also, do not drop acked request if sending of SYN-ACK fails.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Willy Tarreau and many other folks in recent years
were concerned what happens when the TCP_DEFER_ACCEPT period
expires for clients which sent ACK packet. They prefer clients
that actively resend ACK on our SYN-ACK retransmissions to be
converted from open requests to sockets and queued to the
listener for accepting after the deferring period is finished.
Then application server can decide to wait longer for data
or to properly terminate the connection with FIN if read()
returns EAGAIN which is an indication for accepting after
the deferring period. This change still can have side effects
for applications that expect always to see data on the accepted
socket. Others can be prepared to work in both modes (with or
without TCP_DEFER_ACCEPT period) and their data processing can
ignore the read=EAGAIN notification and to allocate resources for
clients which proved to have no data to send during the deferring
period. OTOH, servers that use TCP_DEFER_ACCEPT=1 as flag (not
as a timeout) to wait for data will notice clients that didn't
send data for 3 seconds but that still resend ACKs.
Thanks to Willy Tarreau for the initial idea and to
Eric Dumazet for the review and testing the change.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jan Blunck