kernel/kernel-generic.git
11 years agopkt_sched: fq: rate limiting improvements
Eric Dumazet [Tue, 1 Oct 2013 16:10:16 +0000 (09:10 -0700)]
pkt_sched: fq: rate limiting improvements

FQ rate limiting suffers from two problems, reported
by Steinar :

1) FQ enforces a delay when flow quantum is exhausted in order
to reduce cpu overhead. But if packets are small, current
delay computation is slightly wrong, and observed rates can
be too high.

Steinar had this problem because he disabled TSO and GSO,
and default FQ quantum is 2*1514.

(Of course, I wish recent TSO auto sizing changes will help
to not having to disable TSO in the first place)

2) maxrate was not used for forwarded flows (skbs not attached
to a socket)

Tested:

tc qdisc add dev eth0 root est 1sec 4sec fq maxrate 8Mbit
netperf -H lpq84 -l 1000 &
sleep 10 ; tc -s qdisc show dev eth0
qdisc fq 8003: root refcnt 32 limit 10000p flow_limit 100p buckets 1024
 quantum 3028 initial_quantum 15140 maxrate 8000Kbit
 Sent 16819357 bytes 11258 pkt (dropped 0, overlimits 0 requeues 0)
 rate 7831Kbit 653pps backlog 7570b 5p requeues 0
  44 flows (43 inactive, 1 throttled), next packet delay 2977352 ns
  0 gc, 0 highprio, 5545 throttled

lpq83:~# tcpdump -p -i eth0 host lpq84 -c 12
09:02:52.079484 IP lpq83 > lpq84: . 1389536928:1389538376(1448) ack 3808678021 win 457 <nop,nop,timestamp 961812 572609068>
09:02:52.079499 IP lpq83 > lpq84: . 1448:2896(1448) ack 1 win 457 <nop,nop,timestamp 961812 572609068>
09:02:52.079906 IP lpq84 > lpq83: . ack 2896 win 16384 <nop,nop,timestamp 572609080 961812>
09:02:52.082568 IP lpq83 > lpq84: . 2896:4344(1448) ack 1 win 457 <nop,nop,timestamp 961815 572609071>
09:02:52.082581 IP lpq83 > lpq84: . 4344:5792(1448) ack 1 win 457 <nop,nop,timestamp 961815 572609071>
09:02:52.083017 IP lpq84 > lpq83: . ack 5792 win 16384 <nop,nop,timestamp 572609083 961815>
09:02:52.085678 IP lpq83 > lpq84: . 5792:7240(1448) ack 1 win 457 <nop,nop,timestamp 961818 572609074>
09:02:52.085693 IP lpq83 > lpq84: . 7240:8688(1448) ack 1 win 457 <nop,nop,timestamp 961818 572609074>
09:02:52.086117 IP lpq84 > lpq83: . ack 8688 win 16384 <nop,nop,timestamp 572609086 961818>
09:02:52.088792 IP lpq83 > lpq84: . 8688:10136(1448) ack 1 win 457 <nop,nop,timestamp 961821 572609077>
09:02:52.088806 IP lpq83 > lpq84: . 10136:11584(1448) ack 1 win 457 <nop,nop,timestamp 961821 572609077>
09:02:52.089217 IP lpq84 > lpq83: . ack 11584 win 16384 <nop,nop,timestamp 572609090 961821>

Reported-by: Steinar H. Gunderson <sesse@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip6tnl: allow to use rtnl ops on fb tunnel
Nicolas Dichtel [Tue, 1 Oct 2013 16:05:00 +0000 (18:05 +0200)]
ip6tnl: allow to use rtnl ops on fb tunnel

rtnl ops where introduced by c075b13098b3 ("ip6tnl: advertise tunnel param via
rtnl"), but I forget to assign rtnl ops to fb tunnels.

Now that it is done, we must remove the explicit call to
unregister_netdevice_queue(), because  the fallback tunnel is added to the queue
in ip6_tnl_destroy_tunnels() when checking rtnl_link_ops of all netdevices (this
is valid since commit 0bd8762824e7 ("ip6tnl: add x-netns support")).

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agosit: allow to use rtnl ops on fb tunnel
Nicolas Dichtel [Tue, 1 Oct 2013 16:04:59 +0000 (18:04 +0200)]
sit: allow to use rtnl ops on fb tunnel

rtnl ops where introduced by ba3e3f50a0e5 ("sit: advertise tunnel param via
rtnl"), but I forget to assign rtnl ops to fb tunnels.

Now that it is done, we must remove the explicit call to
unregister_netdevice_queue(), because  the fallback tunnel is added to the queue
in sit_destroy_tunnels() when checking rtnl_link_ops of all netdevices (this
is valid since commit 5e6700b3bf98 ("sit: add support of x-netns")).

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'ip_tunnel'
David S. Miller [Tue, 1 Oct 2013 16:42:28 +0000 (12:42 -0400)]
Merge branch 'ip_tunnel'

ip_tunnel bug fixes from Steffen Klassert.

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip_tunnel: Remove double unregister of the fallback device
Steffen Klassert [Tue, 1 Oct 2013 09:37:37 +0000 (11:37 +0200)]
ip_tunnel: Remove double unregister of the fallback device

When queueing the netdevices for removal, we queue the
fallback device twice in ip_tunnel_destroy(). The first
time when we queue all netdevices in the namespace and
then again explicitly. Fix this by removing the explicit
queueing of the fallback device.

Bug was introduced when network namespace support was added
with commit 6c742e714d8 ("ipip: add x-netns support").

Cc: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip_tunnel_core: Change __skb_push back to skb_push
Steffen Klassert [Tue, 1 Oct 2013 09:35:51 +0000 (11:35 +0200)]
ip_tunnel_core: Change __skb_push back to skb_push

Git commit 0e6fbc5b ("ip_tunnels: extend iptunnel_xmit()")
moved the IP header installation to iptunnel_xmit() and
changed skb_push() to __skb_push(). This makes possible
bugs hard to track down, so change it back to skb_push().

Cc: Pravin Shelar <pshelar@nicira.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip_tunnel: Add fallback tunnels to the hash lists
Steffen Klassert [Tue, 1 Oct 2013 09:34:48 +0000 (11:34 +0200)]
ip_tunnel: Add fallback tunnels to the hash lists

Currently we can not update the tunnel parameters of
the fallback tunnels because we don't find them in the
hash lists. Fix this by adding them on initialization.

Bug was introduced with commit c544193214
("GRE: Refactor GRE tunneling code.")

Cc: Pravin Shelar <pshelar@nicira.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip_tunnel: Fix a memory corruption in ip_tunnel_xmit
Steffen Klassert [Tue, 1 Oct 2013 09:33:59 +0000 (11:33 +0200)]
ip_tunnel: Fix a memory corruption in ip_tunnel_xmit

We might extend the used aera of a skb beyond the total
headroom when we install the ipip header. Fix this by
calling skb_cow_head() unconditionally.

Bug was introduced with commit c544193214
("GRE: Refactor GRE tunneling code.")

Cc: Pravin Shelar <pshelar@nicira.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf
David S. Miller [Tue, 1 Oct 2013 16:39:35 +0000 (12:39 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/pablo/nf

Pablo Neira Ayuso says:

====================
The following patchset contains Netfilter/IPVS fixes for your net
tree, they are:

* Fix BUG_ON splat due to malformed TCP packets seen by synproxy, from
  Patrick McHardy.

* Fix possible weight overflow in lblc and lblcr schedulers due to
  32-bits arithmetics, from Simon Kirby.

* Fix possible memory access race in the lblc and lblcr schedulers,
  introduced when it was converted to use RCU, two patches from
  Julian Anastasov.

* Fix hard dependency on CPU 0 when reading per-cpu stats in the
  rate estimator, from Julian Anastasov.

* Fix race that may lead to object use after release, when invoking
  ipvsadm -C && ipvsadm -R, introduced when adding RCU, from Julian
  Anastasov.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Fix SR-IOV configuration
Manish Chopra [Tue, 1 Oct 2013 06:23:48 +0000 (02:23 -0400)]
qlcnic: Fix SR-IOV configuration

o Interface needs to be brought down and up while configuring SR-IOV.
  Protect interface up/down using rtnl_lock()/rtnl_unlock()

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoll_temac: Reset dma descriptors indexes on ndo_open
Ricardo Ribalda [Tue, 1 Oct 2013 06:17:10 +0000 (08:17 +0200)]
ll_temac: Reset dma descriptors indexes on ndo_open

The dma descriptors indexes are only initialized on the probe function.

If a packet is on the buffer when temac_stop is called, the dma
descriptors indexes can be left on a incorrect state where no other
package can be sent.

So an interface could be left in an usable state after ifdow/ifup.

This patch makes sure that the descriptors indexes are in a proper
status when the device is open.

Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoskbuff: size of hole is wrong in a comment
Nicolas Dichtel [Mon, 30 Sep 2013 12:16:41 +0000 (14:16 +0200)]
skbuff: size of hole is wrong in a comment

Since commit c93bdd0e03e8 ("netvm: allow skb allocation to use PFMEMALLOC
reserves"), hole size is one bit less than what is written in the comment.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'fixes-for-3.12' of git://gitorious.org/linux-can/linux-can
David S. Miller [Tue, 1 Oct 2013 05:31:05 +0000 (22:31 -0700)]
Merge branch 'fixes-for-3.12' of git://gitorious.org/linux-can/linux-can

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv6 mcast: use in6_dev_put in timer handlers instead of __in6_dev_put
Salam Noureddine [Sun, 29 Sep 2013 20:41:34 +0000 (13:41 -0700)]
ipv6 mcast: use in6_dev_put in timer handlers instead of __in6_dev_put

It is possible for the timer handlers to run after the call to
ipv6_mc_down so use in6_dev_put instead of __in6_dev_put in the
handler function in order to do proper cleanup when the refcnt
reaches 0. Otherwise, the refcnt can reach zero without the
inet6_dev being destroyed and we end up leaking a reference to
the net_device and see messages like the following,

unregister_netdevice: waiting for eth0 to become free. Usage count = 1

Tested on linux-3.4.43.

Signed-off-by: Salam Noureddine <noureddine@aristanetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv4 igmp: use in_dev_put in timer handlers instead of __in_dev_put
Salam Noureddine [Sun, 29 Sep 2013 20:39:42 +0000 (13:39 -0700)]
ipv4 igmp: use in_dev_put in timer handlers instead of __in_dev_put

It is possible for the timer handlers to run after the call to
ip_mc_down so use in_dev_put instead of __in_dev_put in the handler
function in order to do proper cleanup when the refcnt reaches 0.
Otherwise, the refcnt can reach zero without the in_device being
destroyed and we end up leaking a reference to the net_device and
see messages like the following,

unregister_netdevice: waiting for eth0 to become free. Usage count = 1

Tested on linux-3.4.43.

Signed-off-by: Salam Noureddine <noureddine@aristanetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoethernet: moxa: fix incorrect placement of __initdata tag
Bartlomiej Zolnierkiewicz [Mon, 30 Sep 2013 13:18:27 +0000 (15:18 +0200)]
ethernet: moxa: fix incorrect placement of __initdata tag

__initdata tag should be placed between the variable name and equal
sign for the variable to be placed in the intended .init.data section.

In this particular case __initdata is incorrect as moxart_mac_driver
can be used after the driver gets initialized.

Also while at it static-ize moxart_mac_driver.

Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv6: gre: correct calculation of max_headroom
Hannes Frederic Sowa [Sun, 29 Sep 2013 03:40:50 +0000 (05:40 +0200)]
ipv6: gre: correct calculation of max_headroom

gre_hlen already accounts for sizeof(struct ipv6_hdr) + gre header,
so initialize max_headroom to zero. Otherwise the

if (encap_limit >= 0) {
max_headroom += 8;
mtu -= 8;
}

increments an uninitialized variable before max_headroom was reset.

Found with coverity: 728539

Cc: Dmitry Kozlov <xeb@mail.ru>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agopowerpc/83xx: gianfar_ptp: select 1588 clock source through dts file
Aida Mynzhasova [Fri, 27 Sep 2013 13:40:27 +0000 (17:40 +0400)]
powerpc/83xx: gianfar_ptp: select 1588 clock source through dts file

Currently IEEE 1588 timer reference clock source is determined through
hard-coded value in gianfar_ptp driver. This patch allows to select ptp
clock source by means of device tree file node.

For instance:

fsl,cksel = <0>;

for using external (TSEC_TMR_CLK input) high precision timer
reference clock.

Other acceptable values:

<1> : eTSEC system clock
<2> : eTSEC1 transmit clock
<3> : RTC clock input

When this attribute isn't used, eTSEC system clock will serve as
IEEE 1588 timer reference clock.

Signed-off-by: Aida Mynzhasova <aida.mynzhasova@skitlab.ru>
Acked-by: Kumar Gala <galak@codeaurora.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoRevert "powerpc/83xx: gianfar_ptp: select 1588 clock source through dts file"
David S. Miller [Tue, 1 Oct 2013 04:16:17 +0000 (21:16 -0700)]
Revert "powerpc/83xx: gianfar_ptp: select 1588 clock source through dts file"

This reverts commit 894116bd0e9b7749a0c4b6c62dec13c2a0ccef68.

I applied the wrong version of this patch, correct
version coming up.

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobonding: Fix broken promiscuity reference counting issue
Neil Horman [Fri, 27 Sep 2013 16:22:15 +0000 (12:22 -0400)]
bonding: Fix broken promiscuity reference counting issue

Recently grabbed this report:
https://bugzilla.redhat.com/show_bug.cgi?id=1005567

Of an issue in which the bonding driver, with an attached vlan encountered the
following errors when bond0 was taken down and back up:

dummy1: promiscuity touches roof, set promiscuity failed. promiscuity feature of
device might be broken.

The error occurs because, during __bond_release_one, if we release our last
slave, we take on a random mac address and issue a NETDEV_CHANGEADDR
notification.  With an attached vlan, the vlan may see that the vlan and bond
mac address were in sync, but no longer are.  This triggers a call to dev_uc_add
and dev_set_rx_mode, which enables IFF_PROMISC on the bond device.  Then, when
we complete __bond_release_one, we use the current state of the bond flags to
determine if we should decrement the promiscuity of the releasing slave.  But
since the bond changed promiscuity state during the release operation, we
incorrectly decrement the slave promisc count when it wasn't in promiscuous mode
to begin with, causing the above error

Fix is pretty simple, just cache the bonding flags at the start of the function
and use those when determining the need to set promiscuity.

This is also needed for the ALLMULTI flag

CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
CC: Mark Wu <wudxw@linux.vnet.ibm.com>
CC: "David S. Miller" <davem@davemloft.net>
Reported-by: Mark Wu <wudxw@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agotcp: TSQ can use a dynamic limit
Eric Dumazet [Fri, 27 Sep 2013 10:28:54 +0000 (03:28 -0700)]
tcp: TSQ can use a dynamic limit

When TCP Small Queues was added, we used a sysctl to limit amount of
packets queues on Qdisc/device queues for a given TCP flow.

Problem is this limit is either too big for low rates, or too small
for high rates.

Now TCP stack has rate estimation in sk->sk_pacing_rate, and TSO
auto sizing, it can better control number of packets in Qdisc/device
queues.

New limit is two packets or at least 1 to 2 ms worth of packets.

Low rates flows benefit from this patch by having even smaller
number of packets in queues, allowing for faster recovery,
better RTT estimations.

High rates flows benefit from this patch by allowing more than 2 packets
in flight as we had reports this was a limiting factor to reach line
rate. [ In particular if TX completion is delayed because of coalescing
parameters ]

Example for a single flow on 10Gbp link controlled by FQ/pacing

14 packets in flight instead of 2

$ tc -s -d qd
qdisc fq 8001: dev eth0 root refcnt 32 limit 10000p flow_limit 100p
buckets 1024 quantum 3028 initial_quantum 15140
 Sent 1168459366606 bytes 771822841 pkt (dropped 0, overlimits 0
requeues 6822476)
 rate 9346Mbit 771713pps backlog 953820b 14p requeues 6822476
  2047 flow, 2046 inactive, 1 throttled, delay 15673 ns
  2372 gc, 0 highprio, 0 retrans, 9739249 throttled, 0 flows_plimit

Note that sk_pacing_rate is currently set to twice the actual rate, but
this might be refined in the future when a flow is in congestion
avoidance.

Additional change : skb->destructor should be set to tcp_wfree().

A future patch (for linux 3.13+) might remove tcp_limit_output_bytes

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Wei Liu <wei.liu2@citrix.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agodm9601: fix IFF_ALLMULTI handling
Peter Korsgaard [Mon, 30 Sep 2013 21:28:20 +0000 (23:28 +0200)]
dm9601: fix IFF_ALLMULTI handling

Pass-all-multicast is controlled by bit 3 in RX control, not bit 2
(pass undersized frames).

Reported-by: Joseph Chang <joseph_chang@davicom.com.tw>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agopkt_sched: fq: qdisc dismantle fixes
Eric Dumazet [Fri, 27 Sep 2013 21:20:01 +0000 (14:20 -0700)]
pkt_sched: fq: qdisc dismantle fixes

fq_reset() should drops all packets in queue, including
throttled flows.

This patch moves code from fq_destroy() to fq_reset()
to do the cleaning.

fq_change() must stop calling fq_dequeue() if all remaining
packets are from throttled flows.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: flow_dissector: fix thoff for IPPROTO_AH
Eric Dumazet [Thu, 26 Sep 2013 15:44:06 +0000 (08:44 -0700)]
net: flow_dissector: fix thoff for IPPROTO_AH

In commit 8ed781668dd49 ("flow_keys: include thoff into flow_keys for
later usage"), we missed that existing code was using nhoff as a
temporary variable that could not always contain transport header
offset.

This is not a problem for TCP/UDP because port offset (@poff)
is 0 for these protocols.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Daniel Borkmann <dborkman@redhat.com>
Cc: Nikolay Aleksandrov <nikolay@redhat.com>
Acked-by: Nikolay Aleksandrov <nikolay@redhat.com>
Acked-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMAINTAINERS: add myself as maintainer of xen-netback
Wei Liu [Thu, 26 Sep 2013 11:41:53 +0000 (12:41 +0100)]
MAINTAINERS: add myself as maintainer of xen-netback

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoxen-netback: Handle backend state transitions in a more robust way
Paul Durrant [Thu, 26 Sep 2013 11:09:52 +0000 (12:09 +0100)]
xen-netback: Handle backend state transitions in a more robust way

When the frontend state changes netback now specifies its desired state to
a new function, set_backend_state(), which transitions through any
necessary intermediate states.
This fixes an issue observed with some old Windows frontend drivers where
they failed to transition through the Closing state and netback would not
behave correctly.

Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv6: Fix preferred_lft not updating in some cases
Paul Marks [Wed, 25 Sep 2013 22:12:55 +0000 (15:12 -0700)]
ipv6: Fix preferred_lft not updating in some cases

Consider the scenario where an IPv6 router is advertising a fixed
preferred_lft of 1800 seconds, while the valid_lft begins at 3600
seconds and counts down in realtime.

A client should reset its preferred_lft to 1800 every time the RA is
received, but a bug is causing Linux to ignore the update.

The core problem is here:
  if (prefered_lft != ifp->prefered_lft) {

Note that ifp->prefered_lft is an offset, so it doesn't decrease over
time.  Thus, the comparison is always (1800 != 1800), which fails to
trigger an update.

The most direct solution would be to compute a "stored_prefered_lft",
and use that value in the comparison.  But I think that trying to filter
out unnecessary updates here is a premature optimization.  In order for
the filter to apply, both of these would need to hold:

  - The advertised valid_lft and preferred_lft are both declining in
    real time.
  - No clock skew exists between the router & client.

So in this patch, I've set "update_lft = 1" unconditionally, which
allows the surrounding code to be greatly simplified.

Signed-off-by: Paul Marks <pmarks@google.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip_tunnel: Do not use stale inner_iph pointer.
Pravin B Shelar [Wed, 25 Sep 2013 16:57:47 +0000 (09:57 -0700)]
ip_tunnel: Do not use stale inner_iph pointer.

While sending packet skb_cow_head() can change skb header which
invalidates inner_iph pointer to skb header. Following patch
avoid using it. Found by code inspection.

This bug was introduced by commit 0e6fbc5b6c6218 (ip_tunnels: extend
iptunnel_xmit()).

Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: qmi_wwan: fix Cinterion PLXX product ID
Aleksander Morgado [Wed, 25 Sep 2013 15:02:36 +0000 (17:02 +0200)]
net: qmi_wwan: fix Cinterion PLXX product ID

Cinterion PLXX LTE devices have a 0x0060 product ID, not 0x12d1.

The blacklisting in the serial/option driver does actually use the correct PID,
as per commit 8ff10bdb14a52e3f25d4ce09e0582a8684c1a6db ('USB: Blacklisted
Cinterion's PLxx WWAN Interface').

CC: Hans-Christoph Schemmel <hans-christoph.schemmel@gemalto.com>
CC: Christian Schmiedl <christian.schmiedl@gemalto.com>
CC: Nicolaus Colberg <nicolaus.colberg@gemalto.com>
Signed-off-by: Aleksander Morgado <aleksander@lanedo.com>
Acked-by: Bjørn Mork <bjorn@mork.no>
Acked-by: Christian Schmiedl <christian.schmiedl@gemalto.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agopowerpc/83xx: gianfar_ptp: select 1588 clock source through dts file
Aida Mynzhasova [Wed, 25 Sep 2013 07:24:23 +0000 (11:24 +0400)]
powerpc/83xx: gianfar_ptp: select 1588 clock source through dts file

Currently IEEE 1588 timer reference clock source is determined through
hard-coded value in gianfar_ptp driver. This patch allows to select ptp
clock source by means of device tree file node.

For instance:

fsl,cksel = <0>;

for using external (TSEC_TMR_CLK input) high precision timer
reference clock.

Other acceptable values:

<1> : eTSEC system clock
<2> : eTSEC1 transmit clock
<3> : RTC clock input

When this attribute isn't used, eTSEC system clock will serve as
IEEE 1588 timer reference clock.

Signed-off-by: Aida Mynzhasova <aida.mynzhasova@skitlab.ru>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agovxlan: Use RCU apis to access sk_user_data.
Pravin B Shelar [Tue, 24 Sep 2013 17:25:40 +0000 (10:25 -0700)]
vxlan: Use RCU apis to access sk_user_data.

Use of RCU api makes vxlan code easier to understand.  It also
fixes bug due to missing ACCESS_ONCE() on sk_user_data dereference.
In rare case without ACCESS_ONCE() compiler might omit vs on
sk_user_data dereference.
Compiler can use vs as alias for sk->sk_user_data, resulting in
multiple sk_user_data dereference in rcu read context which
could change.

CC: Jesse Gross <jesse@nicira.com>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agocan: flexcan: fix flexcan_chip_start() on imx6
Marc Kleine-Budde [Fri, 27 Sep 2013 10:15:05 +0000 (12:15 +0200)]
can: flexcan: fix flexcan_chip_start() on imx6

In the flexcan_chip_start() function first the flexcan core is going through
the soft reset sequence, then the RX FIFO is enabled.

With the hardware is put into FIFO mode, message buffers 1...7 are reserved by
the FIFO engine. The remaining message buffers are in reset default values.
This patch removes the bogus initialization of the message buffers, as it
causes an imprecise external abort on imx6.

Cc: linux-stable <stable@vger.kernel.org>
Reported-by: Lothar Waßmann <LW@KARO-electronics.de>
Tested-by: Lothar Waßmann <LW@KARO-electronics.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agonetfilter: synproxy: fix BUG_ON triggered by corrupt TCP packets
Patrick McHardy [Mon, 30 Sep 2013 07:51:46 +0000 (08:51 +0100)]
netfilter: synproxy: fix BUG_ON triggered by corrupt TCP packets

TCP packets hitting the SYN proxy through the SYNPROXY target are not
validated by TCP conntrack. When th->doff is below 5, an underflow happens
when calculating the options length, causing skb_header_pointer() to
return NULL and triggering the BUG_ON().

Handle this case gracefully by checking for NULL instead of using BUG_ON().

Reported-by: Martin Topholm <mph@one.com>
Tested-by: Martin Topholm <mph@one.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
11 years agonet: net_secret should not depend on TCP
Eric Dumazet [Tue, 24 Sep 2013 13:19:57 +0000 (06:19 -0700)]
net: net_secret should not depend on TCP

A host might need net_secret[] and never open a single socket.

Problem added in commit aebda156a570782
("net: defer net_secret[] initialization")

Based on prior patch from Hannes Frederic Sowa.

Reported-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Hannes Frederic Sowa <hannes@strressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: Delay default_device_exit_batch until no devices are unregistering v2
Eric W. Biederman [Tue, 24 Sep 2013 04:19:49 +0000 (21:19 -0700)]
net: Delay default_device_exit_batch until no devices are unregistering v2

There is currently serialization network namespaces exiting and
network devices exiting as the final part of netdev_run_todo does not
happen under the rtnl_lock.  This is compounded by the fact that the
only list of devices unregistering in netdev_run_todo is local to the
netdev_run_todo.

This lack of serialization in extreme cases results in network devices
unregistering in netdev_run_todo after the loopback device of their
network namespace has been freed (making dst_ifdown unsafe), and after
the their network namespace has exited (making the NETDEV_UNREGISTER,
and NETDEV_UNREGISTER_FINAL callbacks unsafe).

Add the missing serialization by a per network namespace count of how
many network devices are unregistering and having a wait queue that is
woken up whenever the count is decreased.  The count and wait queue
allow default_device_exit_batch to wait until all of the unregistration
activity for a network namespace has finished before proceeding to
unregister the loopback device and then allowing the network namespace
to exit.

Only a single global wait queue is used because there is a single global
lock, and there is a single waiter, per network namespace wait queues
would be a waste of resources.

The per network namespace count of unregistering devices gives a
progress guarantee because the number of network devices unregistering
in an exiting network namespace must ultimately drop to zero (assuming
network device unregistration completes).

The basic logic remains the same as in v1.  This patch is now half
comment and half rtnl_lock_unregistering an expanded version of
wait_event performs no extra work in the common case where no network
devices are unregistering when we get to default_device_exit_batch.

Reported-by: Francesco Ruggeri <fruggeri@aristanetworks.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoIPv6 NAT: Do not drop DNATed 6to4/6rd packets
Catalin\(ux\) M. BOIE [Mon, 23 Sep 2013 20:04:19 +0000 (23:04 +0300)]
IPv6 NAT: Do not drop DNATed 6to4/6rd packets

When a router is doing DNAT for 6to4/6rd packets the latest
anti-spoofing commit 218774dc ("ipv6: add anti-spoofing checks for
6to4 and 6rd") will drop them because the IPv6 address embedded does
not match the IPv4 destination. This patch will allow them to pass by
testing if we have an address that matches on 6to4/6rd interface.  I
have been hit by this problem using Fedora and IPV6TO4_IPV4ADDR.
Also, log the dropped packets (with rate limit).

Signed-off-by: Catalin(ux) M. BOIE <catab@embedromix.ro>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoUSBNET: fix handling padding packet
Ming Lei [Mon, 23 Sep 2013 12:59:35 +0000 (20:59 +0800)]
USBNET: fix handling padding packet

Commit 638c5115a7949(USBNET: support DMA SG) introduces DMA SG
if the usb host controller is capable of building packet from
discontinuous buffers, but missed handling padding packet when
building DMA SG.

This patch attachs the pre-allocated padding packet at the
end of the sg list, so padding packet can be sent to device
if drivers require that.

Reported-by: David Laight <David.Laight@aculab.com>
Acked-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Ming Lei <ming.lei@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Fix register device in FAILED state for 82xx.
Sucheta Chakraborty [Fri, 27 Sep 2013 06:12:36 +0000 (02:12 -0400)]
qlcnic: Fix register device in FAILED state for 82xx.

o Commit 7e2cf4feba058476324dc545e3d1b316998c91e6
  ("qlcnic: change driver hardware interface mechanism")
  has overwritten
  commit b43e5ee76a4320c070cf0fe65cf4927198fbb4d1
  ("qlcnic: Register device in FAILED state")

Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobe2net: Fix to display the VLAN priority for a VF
Ajit Khaparde [Fri, 27 Sep 2013 20:18:56 +0000 (15:18 -0500)]
be2net: Fix to display the VLAN priority for a VF

VLAN priority is not being displayed for a VF currently when user executes
"ip link show" command. This patch fixes it.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobe2net: Fix to configure VLAN priority for a VF interface.
Ajit Khaparde [Fri, 27 Sep 2013 20:18:46 +0000 (15:18 -0500)]
be2net: Fix to configure VLAN priority for a VF interface.

Thix fix allows the VLAN priority to be configured for a VF interface
via the "ip link set DEVICE vf NUM" path.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobe2net: Fix to allow VLAN configuration on VF interfaces.
Ajit Khaparde [Fri, 27 Sep 2013 20:18:28 +0000 (15:18 -0500)]
be2net: Fix to allow VLAN configuration on VF interfaces.

Now the VF interfaces have privilege to add VLANs.
Allow VLANs to be configured on these interfaces.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobe2net: Fix number of VLANs supported in UMC mode for BE3-R.
Ajit Khaparde [Fri, 27 Sep 2013 20:18:16 +0000 (15:18 -0500)]
be2net: Fix number of VLANs supported in UMC mode for BE3-R.

In BE3-R, when UMC is enabled, the number of VLANs that can be added
to the interface is reduced to 15.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobe2net: Fix VLAN promiscuous mode programming
Ajit Khaparde [Fri, 27 Sep 2013 20:17:58 +0000 (15:17 -0500)]
be2net: Fix VLAN promiscuous mode programming

When the interface runs out of the allocated entries in VLAN table,
we program the interface in VLAN promiscuous mode.
Use OPCODE_COMMON_NTWK_RX_FILTER to set VLAN Promiscuous mode
instead of OPCODE_COMMON_NTWK_VLAN_CONFIG.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobe2net: Fix the size of be_nic_res_desc structure
Ajit Khaparde [Fri, 27 Sep 2013 20:17:31 +0000 (15:17 -0500)]
be2net: Fix the size of be_nic_res_desc structure

Size of be_nic_res_desc structure is incorrect. Fix it.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobe2net: Fix to prevent Tx stall on SH-R when packet size < 32
Ajit Khaparde [Fri, 27 Sep 2013 20:17:04 +0000 (15:17 -0500)]
be2net: Fix to prevent Tx stall on SH-R when packet size < 32

Tx on SH-R can lockup if the packet size is less than 32 bytes.
Pad such packets to a safer 36-byte size.
Patch uses the Lancer-R workaround - which checks for packet <= 32-bytes

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville...
David S. Miller [Fri, 27 Sep 2013 20:05:34 +0000 (16:05 -0400)]
Merge branch 'for-davem' of git://git./linux/kernel/git/linville/wireless

John W. Linville says:

====================
Regarding the Bluetooth bits, Gustavo says:

"First Bluetooth fixes to 3.12, it includes:

* 3 patches to add device id for 3 new hardwares.

* 2 patches from Johan to fix the rfkill behaviour during setup stage

* a small clean up in the rfcomm TTY code that fixes a potential racy
condition (by Gianluca Anzolin)

* 2 fixes to proper set encryption key size and security level in the
peripheral role of Bluetooth LE devices. (by Andre Guedes)

* a fix for dealing devices where pairing is not necessary, we were keeping
the Bluetooth ACL connection alive for too much time. (by Syam Sidhardhan)"

Also, I fixed-up an curly-brace indentation problem in the Bluetooth
code while merging.  On top of that...

Alexey Khoroshilov brings a p54usb fix to avoid a resource leak when
request_firmware_nowait fails.

Amitkumar Karwar fixes a firmware hang caused by too much header data
being appended for USB devices using the mwifiex driver.

Arend van Spriel provides three fixes: a brcmfmac fix to relocate some
driver code outside of an .init section; a scheduling while atomic
fix for bcma; and, another scheduling while atomic fix for brcmsmac.

Bing Zhao offers a pair of mwifiex fixes: a code change to avoid
firmware timeouts on USB; and a fix for a NULL pointer dereference.

Christian Lamparter adds a device ID to p54usb.

Felix Fietkau implements a quartet of small ath9k fixes, including
a locking fix, a list management fix, a fix to properly mark a stale
buffer, and an aggregate buffering fix.

Larry Finger champions a data alignment fix to make rtlwifi work
better with ARM builds.

Solomon Peachy reverts an earlier interrupt handling fix for cw1200
and replaces it with a new threaded oneshot irq handler implementation.

Sujith Manoharan fixes an ath9k regression by reverting an earlier
patch.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net
David S. Miller [Fri, 27 Sep 2013 17:27:49 +0000 (13:27 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/jkirsher/net

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates

This series contains updates to igb and i40e.

Todd provides a fix for 82580 devices in igb, where the ethtool
loopback test was missing 82580 copper devices.

Jesse provides five fixes/cleanups to i40e based on feedback from
Joe Perches and the community.

v2: fixed up patch 5 in the series based on feedback from Joe Perches
    and David Miller
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
John W. Linville [Fri, 27 Sep 2013 17:09:17 +0000 (13:09 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless into for-davem

Also fixed-up a badly indented closing brace...

Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoi40e: clean up coccicheck reported errors
Jesse Brandeburg [Fri, 13 Sep 2013 08:23:22 +0000 (08:23 +0000)]
i40e: clean up coccicheck reported errors

coccicheck shows:

drivers/net/ethernet/intel/i40e/i40e_adminq.c:704:2-8: Replace memcpy
with struct assignment
drivers/net/ethernet/intel/i40e/i40e_adminq.c:763:1-7: Replace memcpy
with struct assignment
drivers/net/ethernet/intel/i40e/i40e_adminq.c:810:2-8: Replace memcpy
with struct assignment
drivers/net/ethernet/intel/i40e/i40e_common.c:510:2-8: Replace memcpy
with struct assignment

Fix each of them with a *a = *b;

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
11 years agoi40e: better return values
Jesse Brandeburg [Wed, 25 Sep 2013 23:41:14 +0000 (23:41 +0000)]
i40e: better return values

As mentioned by Joe Perches, clean up return values in some functions
making sure to have consistent return types, not mixing types.

A couple of Joe's comments suggested returning void, but since
the functions in question are ndo defined, the return values are fixed.
So make a comment in the header that notes this is a function called by
net_device_ops.

v2: fix post increment bug in return

CC: Joe Perches <joe@perches.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
11 years agoi40e: convert ret to aq_ret
Jesse Brandeburg [Fri, 13 Sep 2013 08:23:20 +0000 (08:23 +0000)]
i40e: convert ret to aq_ret

When calling admin queue functions the driver should use aq_ret
variable to help make clear that the return value is not a regular
return variable.

This allows for clean up of the return types that were previously
converted to int.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
11 years agoi40e: small clean ups from review
Jesse Brandeburg [Fri, 13 Sep 2013 08:23:19 +0000 (08:23 +0000)]
i40e: small clean ups from review

As mentioned by Joe Perches clean up a loop flow.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
CC: Joe Perches <joe@perches.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
11 years agoi40e: use common failure flow
Jesse Brandeburg [Fri, 13 Sep 2013 08:23:18 +0000 (08:23 +0000)]
i40e: use common failure flow

As mentioned by Joe Perches, we should be using
foo = alloc(...)
if (!foo)
return -ENOMEM;

return 0;

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
CC: Joe Perches <joe@perches.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
11 years agoigb: Fix ethtool loopback test for 82580 copper
Todd Fujinaka [Tue, 17 Sep 2013 05:08:48 +0000 (05:08 +0000)]
igb: Fix ethtool loopback test for 82580 copper

Add back 82580 loopback tests to ethtool.

Signed-off-by: Todd Fujinaka <todd.fujinaka@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
11 years agoMerge branch 'fixes-for-3.12' of git://gitorious.org/linux-can/linux-can
David S. Miller [Fri, 27 Sep 2013 06:08:03 +0000 (02:08 -0400)]
Merge branch 'fixes-for-3.12' of git://gitorious.org/linux-can/linux-can

11 years agovia-rhine: fix VLAN priority field (PCP, IEEE 802.1p)
Roger Luethi [Sat, 21 Sep 2013 12:24:11 +0000 (14:24 +0200)]
via-rhine: fix VLAN priority field (PCP, IEEE 802.1p)

Outgoing packets sent by via-rhine have their VLAN PCP field off by one
(when hardware acceleration is enabled). The TX descriptor expects only VID
and PCP (without a CFI/DEI bit).

Peter Boström noticed and reported the bug.

Signed-off-by: Roger Luethi <rl@hellgate.ch>
Cc: Peter Boström <peter.bostrom@netrounds.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobrcmsmac: call bcma_core_pci_power_save() from non-atomic context
Arend van Spriel [Wed, 25 Sep 2013 10:11:03 +0000 (12:11 +0200)]
brcmsmac: call bcma_core_pci_power_save() from non-atomic context

This patch adds explicit call to bcma_core_pci_power_save() from
a non-atomic context resolving 'scheduling while atomic' issue.

[   13.224317] BUG: scheduling while atomic: dhcpcd/1800/0x00000202
[   13.224322] Modules linked in: brcmsmac nouveau coretemp kvm_intel kvm cordic brcmutil bcma dell_wmi atl1c ttm mxm_wmi wmi
[   13.224354] CPU: 0 PID: 1800 Comm: dhcpcd Tainted: G        W    3.11.0-wl #1
[   13.224359] Hardware name: Alienware M11x R2/M11x R2, BIOS A04 11/23/2010
[   13.224363]  ffff880177c12c40 ffff880170fd1968 ffffffff8169af5b 0000000000000007
[   13.224374]  ffff880170fd1ad0 ffff880170fd1978 ffffffff81697ee2 ffff880170fd19f8
[   13.224383]  ffffffff816a19f5 00000000000f4240 000000000000d080 ffff880170fd1fd8
[   13.224391] Call Trace:
[   13.224399]  [<ffffffff8169af5b>] dump_stack+0x4f/0x84
[   13.224403]  [<ffffffff81697ee2>] __schedule_bug+0x43/0x51
[   13.224409]  [<ffffffff816a19f5>] __schedule+0x6e5/0x810
[   13.224412]  [<ffffffff816a1c34>] schedule+0x24/0x70
[   13.224416]  [<ffffffff816a04fc>] schedule_hrtimeout_range_clock+0x10c/0x150
[   13.224420]  [<ffffffff810684e0>] ? update_rmtp+0x60/0x60
[   13.224424]  [<ffffffff8106915f>] ? hrtimer_start_range_ns+0xf/0x20
[   13.224429]  [<ffffffff816a054e>] schedule_hrtimeout_range+0xe/0x10
[   13.224432]  [<ffffffff8104f6fb>] usleep_range+0x3b/0x40
[   13.224437]  [<ffffffffa003733a>] bcma_pcie_mdio_read.isra.5+0x8a/0x100 [bcma]
[   13.224442]  [<ffffffffa00374a5>] bcma_pcie_mdio_writeread.isra.6.constprop.13+0x25/0x30 [bcma]
[   13.224448]  [<ffffffffa00374f9>] bcma_core_pci_power_save+0x49/0x80 [bcma]
[   13.224452]  [<ffffffffa003765d>] bcma_core_pci_up+0x2d/0x60 [bcma]
[   13.224460]  [<ffffffffa03dc17c>] brcms_c_up+0xfc/0x430 [brcmsmac]
[   13.224467]  [<ffffffffa03d1a7d>] brcms_up+0x1d/0x20 [brcmsmac]
[   13.224473]  [<ffffffffa03d2498>] brcms_ops_start+0x298/0x340 [brcmsmac]
[   13.224478]  [<ffffffff81600a12>] ? cfg80211_netdev_notifier_call+0xd2/0x5f0
[   13.224483]  [<ffffffff815fa53d>] ? packet_notifier+0xad/0x1d0
[   13.224487]  [<ffffffff81656e75>] ieee80211_do_open+0x325/0xf80
[   13.224491]  [<ffffffff8106ac09>] ? __raw_notifier_call_chain+0x9/0x10
[   13.224495]  [<ffffffff81657b41>] ieee80211_open+0x71/0x80
[   13.224498]  [<ffffffff81526267>] __dev_open+0x87/0xe0
[   13.224502]  [<ffffffff8152650c>] __dev_change_flags+0x9c/0x180
[   13.224505]  [<ffffffff815266a3>] dev_change_flags+0x23/0x70
[   13.224509]  [<ffffffff8158cd68>] devinet_ioctl+0x5b8/0x6a0
[   13.224512]  [<ffffffff8158d5c5>] inet_ioctl+0x75/0x90
[   13.224516]  [<ffffffff8150b38b>] sock_do_ioctl+0x2b/0x70
[   13.224519]  [<ffffffff8150b681>] sock_ioctl+0x71/0x2a0
[   13.224523]  [<ffffffff8114ed47>] do_vfs_ioctl+0x87/0x520
[   13.224528]  [<ffffffff8113f159>] ? ____fput+0x9/0x10
[   13.224533]  [<ffffffff8106228c>] ? task_work_run+0x9c/0xd0
[   13.224537]  [<ffffffff8114f271>] SyS_ioctl+0x91/0xb0
[   13.224541]  [<ffffffff816aa252>] system_call_fastpath+0x16/0x1b

Cc: <stable@vger.kernel.org> # 3.11.x
Cc: Tod Jackson <tod.jackson@gmail.com>
Cc: Joe Perches <joe@perches.com>
Cc: Rafal Milecki <zajec5@gmail.com>
Cc: Hauke Mehrtens <hauke@hauke-m.de>
Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agobcma: make bcma_core_pci_{up,down}() callable from atomic context
Arend van Spriel [Wed, 25 Sep 2013 10:11:02 +0000 (12:11 +0200)]
bcma: make bcma_core_pci_{up,down}() callable from atomic context

This patch removes the bcma_core_pci_power_save() call from
the bcma_core_pci_{up,down}() functions as it tries to schedule
thus requiring to call them from non-atomic context. The function
bcma_core_pci_power_save() is now exported so the calling module
can explicitly use it in non-atomic context. This fixes the
'scheduling while atomic' issue reported by Tod Jackson and
Joe Perches.

[   13.210710] BUG: scheduling while atomic: dhcpcd/1800/0x00000202
[   13.210718] Modules linked in: brcmsmac nouveau coretemp kvm_intel kvm cordic brcmutil bcma dell_wmi atl1c ttm mxm_wmi wmi
[   13.210756] CPU: 2 PID: 1800 Comm: dhcpcd Not tainted 3.11.0-wl #1
[   13.210762] Hardware name: Alienware M11x R2/M11x R2, BIOS A04 11/23/2010
[   13.210767]  ffff880177c92c40 ffff880170fd1948 ffffffff8169af5b 0000000000000007
[   13.210777]  ffff880170fd1ab0 ffff880170fd1958 ffffffff81697ee2 ffff880170fd19d8
[   13.210785]  ffffffff816a19f5 00000000000f4240 000000000000d080 ffff880170fd1fd8
[   13.210794] Call Trace:
[   13.210813]  [<ffffffff8169af5b>] dump_stack+0x4f/0x84
[   13.210826]  [<ffffffff81697ee2>] __schedule_bug+0x43/0x51
[   13.210837]  [<ffffffff816a19f5>] __schedule+0x6e5/0x810
[   13.210845]  [<ffffffff816a1c34>] schedule+0x24/0x70
[   13.210855]  [<ffffffff816a04fc>] schedule_hrtimeout_range_clock+0x10c/0x150
[   13.210867]  [<ffffffff810684e0>] ? update_rmtp+0x60/0x60
[   13.210877]  [<ffffffff8106915f>] ? hrtimer_start_range_ns+0xf/0x20
[   13.210887]  [<ffffffff816a054e>] schedule_hrtimeout_range+0xe/0x10
[   13.210897]  [<ffffffff8104f6fb>] usleep_range+0x3b/0x40
[   13.210910]  [<ffffffffa00371af>] bcma_pcie_mdio_set_phy.isra.3+0x4f/0x80 [bcma]
[   13.210921]  [<ffffffffa003729f>] bcma_pcie_mdio_write.isra.4+0xbf/0xd0 [bcma]
[   13.210932]  [<ffffffffa0037498>] bcma_pcie_mdio_writeread.isra.6.constprop.13+0x18/0x30 [bcma]
[   13.210942]  [<ffffffffa00374ee>] bcma_core_pci_power_save+0x3e/0x80 [bcma]
[   13.210953]  [<ffffffffa003765d>] bcma_core_pci_up+0x2d/0x60 [bcma]
[   13.210975]  [<ffffffffa03dc17c>] brcms_c_up+0xfc/0x430 [brcmsmac]
[   13.210989]  [<ffffffffa03d1a7d>] brcms_up+0x1d/0x20 [brcmsmac]
[   13.211003]  [<ffffffffa03d2498>] brcms_ops_start+0x298/0x340 [brcmsmac]
[   13.211020]  [<ffffffff81600a12>] ? cfg80211_netdev_notifier_call+0xd2/0x5f0
[   13.211030]  [<ffffffff815fa53d>] ? packet_notifier+0xad/0x1d0
[   13.211064]  [<ffffffff81656e75>] ieee80211_do_open+0x325/0xf80
[   13.211076]  [<ffffffff8106ac09>] ? __raw_notifier_call_chain+0x9/0x10
[   13.211086]  [<ffffffff81657b41>] ieee80211_open+0x71/0x80
[   13.211101]  [<ffffffff81526267>] __dev_open+0x87/0xe0
[   13.211109]  [<ffffffff8152650c>] __dev_change_flags+0x9c/0x180
[   13.211117]  [<ffffffff815266a3>] dev_change_flags+0x23/0x70
[   13.211127]  [<ffffffff8158cd68>] devinet_ioctl+0x5b8/0x6a0
[   13.211136]  [<ffffffff8158d5c5>] inet_ioctl+0x75/0x90
[   13.211147]  [<ffffffff8150b38b>] sock_do_ioctl+0x2b/0x70
[   13.211155]  [<ffffffff8150b681>] sock_ioctl+0x71/0x2a0
[   13.211169]  [<ffffffff8114ed47>] do_vfs_ioctl+0x87/0x520
[   13.211180]  [<ffffffff8113f159>] ? ____fput+0x9/0x10
[   13.211198]  [<ffffffff8106228c>] ? task_work_run+0x9c/0xd0
[   13.211202]  [<ffffffff8114f271>] SyS_ioctl+0x91/0xb0
[   13.211208]  [<ffffffff816aa252>] system_call_fastpath+0x16/0x1b
[   13.211217] NOHZ: local_softirq_pending 202

The issue was introduced in v3.11 kernel by following commit:

commit aa51e598d04c6acf5477934cd6383f5a17ce9029
Author: Hauke Mehrtens <hauke@hauke-m.de>
Date:   Sat Aug 24 00:32:31 2013 +0200

    brcmsmac: use bcma PCIe up and down functions

    replace the calls to bcma_core_pci_extend_L1timer() by calls to the
    newly introduced bcma_core_pci_ip() and bcma_core_pci_down()

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Cc: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This fix has been discussed with Hauke Mehrtens [1] selection
option 3) and is intended for v3.12.

Ref:
[1] http://mid.gmane.org/5239B12D.3040206@hauke-m.de

Cc: <stable@vger.kernel.org> # 3.11.x
Cc: Tod Jackson <tod.jackson@gmail.com>
Cc: Joe Perches <joe@perches.com>
Cc: Rafal Milecki <zajec5@gmail.com>
Cc: Hauke Mehrtens <hauke@hauke-m.de>
Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agobrcmfmac: obtain platform data upon module initialization
Arend van Spriel [Wed, 25 Sep 2013 10:11:01 +0000 (12:11 +0200)]
brcmfmac: obtain platform data upon module initialization

The driver uses platform_driver_probe() to obtain platform data
if any. However, that function is placed in the .init section so
it must be called upon driver module initialization.

The problem was reported by Fenguang Wu resulting in a kernel
oops because the .init section was already freed.

[   48.966342] Switched to clocksource tsc
[   48.970002] kernel tried to execute NX-protected page - exploit attempt? (uid: 0)
[   48.970851] BUG: unable to handle kernel paging request at ffffffff82196446
[   48.970957] IP: [<ffffffff82196446>] classes_init+0x26/0x26
[   48.970957] PGD 1e76067 PUD 1e77063 PMD f388063 PTE 8000000002196163
[   48.970957] Oops: 0011 [#1]
[   48.970957] CPU: 0 PID: 17 Comm: kworker/0:1 Not tainted 3.11.0-rc7-00444-gc52dd7f #23
[   48.970957] Workqueue: events brcmf_driver_init
[   48.970957] task: ffff8800001d2000 ti: ffff8800001d4000 task.ti: ffff8800001d4000
[   48.970957] RIP: 0010:[<ffffffff82196446>]  [<ffffffff82196446>] classes_init+0x26/0x26
[   48.970957] RSP: 0000:ffff8800001d5d40  EFLAGS: 00000286
[   48.970957] RAX: 0000000000000001 RBX: ffffffff820c5620 RCX: 0000000000000000
[   48.970957] RDX: 0000000000000001 RSI: ffffffff816f7380 RDI: ffffffff820c56c0
[   48.970957] RBP: ffff8800001d5d50 R08: ffff8800001d2508 R09: 0000000000000002
[   48.970957] R10: 0000000000000000 R11: 0001f7ce298c5620 R12: ffff8800001c76b0
[   48.970957] R13: ffffffff81e91d40 R14: 0000000000000000 R15: ffff88000e0ce300
[   48.970957] FS:  0000000000000000(0000) GS:ffffffff81e84000(0000) knlGS:0000000000000000
[   48.970957] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[   48.970957] CR2: ffffffff82196446 CR3: 0000000001e75000 CR4: 00000000000006b0
[   48.970957] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   48.970957] DR3: 0000000000000000 DR6: 0000000000000000 DR7: 0000000000000000
[   48.970957] Stack:
[   48.970957]  ffffffff816f7df8 ffffffff820c5620 ffff8800001d5d60 ffffffff816eeec9
[   48.970957]  ffff8800001d5de0 ffffffff81073dc5 ffffffff81073d68 ffff8800001d5db8
[   48.970957]  0000000000000086 ffffffff820c5620 ffffffff824f7fd0 0000000000000000
[   48.970957] Call Trace:
[   48.970957]  [<ffffffff816f7df8>] ? brcmf_sdio_init+0x18/0x70
[   48.970957]  [<ffffffff816eeec9>] brcmf_driver_init+0x9/0x10
[   48.970957]  [<ffffffff81073dc5>] process_one_work+0x1d5/0x480
[   48.970957]  [<ffffffff81073d68>] ? process_one_work+0x178/0x480
[   48.970957]  [<ffffffff81074188>] worker_thread+0x118/0x3a0
[   48.970957]  [<ffffffff81074070>] ? process_one_work+0x480/0x480
[   48.970957]  [<ffffffff8107aa17>] kthread+0xe7/0xf0
[   48.970957]  [<ffffffff810829f7>] ? finish_task_switch.constprop.57+0x37/0xd0
[   48.970957]  [<ffffffff8107a930>] ? __kthread_parkme+0x80/0x80
[   48.970957]  [<ffffffff81a6923a>] ret_from_fork+0x7a/0xb0
[   48.970957]  [<ffffffff8107a930>] ? __kthread_parkme+0x80/0x80
[   48.970957] Code: cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc
cc cc cc cc cc cc <cc> cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc
[   48.970957] RIP  [<ffffffff82196446>] classes_init+0x26/0x26
[   48.970957]  RSP <ffff8800001d5d40>
[   48.970957] CR2: ffffffff82196446
[   48.970957] ---[ end trace 62980817cd525f14 ]---

Cc: <stable@vger.kernel.org> # 3.10.x, 3.11.x
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Tested-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agomwifiex: fix NULL pointer dereference in usb suspend handler
Bing Zhao [Wed, 25 Sep 2013 02:31:25 +0000 (19:31 -0700)]
mwifiex: fix NULL pointer dereference in usb suspend handler

Bug 60815 - Interface hangs in mwifiex_usb
https://bugzilla.kernel.org/show_bug.cgi?id=60815

[ 2.883807] BUG: unable to handle kernel NULL pointer dereference
            at 0000000000000048
[ 2.883813] IP: [<ffffffff815a65e0>] pfifo_fast_enqueue+0x90/0x90

[ 2.883834] CPU: 1 PID: 3220 Comm: kworker/u8:90 Not tainted
            3.11.1-monotone-l0 #6
[ 2.883834] Hardware name: Microsoft Corporation Surface with
            Windows 8 Pro/Surface with Windows 8 Pro,
            BIOS 1.03.0450 03/29/2013

On Surface Pro, suspend to ram gives a NULL pointer dereference in
pfifo_fast_enqueue(). The stack trace reveals that the offending
call is clearing carrier in mwifiex_usb suspend handler.

Since commit 1499d9f "mwifiex: don't drop carrier flag over suspend"
has removed the carrier flag handling over suspend/resume in SDIO
and PCIe drivers, I'm removing it in USB driver too. This also fixes
the bug for Surface Pro.

Cc: <stable@vger.kernel.org> # 3.5+
Tested-by: Dmitry Khromov <icechrome@gmail.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agomwifiex: fix hang issue for USB chipsets
Amitkumar Karwar [Wed, 25 Sep 2013 02:31:24 +0000 (19:31 -0700)]
mwifiex: fix hang issue for USB chipsets

Bug 60815 - Interface hangs in mwifiex_usb
https://bugzilla.kernel.org/show_bug.cgi?id=60815

We have 4 bytes of interface header for packets delivered to SDIO
and PCIe, but not for USB interface.

In Tx AMSDU case, currently 4 bytes of garbage data is unnecessarily
appended for USB packets. This sometimes leads to a firmware hang,
because it may not interpret the data packet correctly.

Problem is fixed by removing this redundant headroom for USB.

Cc: <stable@vger.kernel.org> # 3.5+
Tested-by: Dmitry Khromov <icechrome@gmail.com>
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>
11 years agop54usb: add USB ID for Corega WLUSB2GTST USB adapter
Christian Lamparter [Tue, 24 Sep 2013 19:56:46 +0000 (21:56 +0200)]
p54usb: add USB ID for Corega WLUSB2GTST USB adapter

Added USB ID for Corega WLUSB2GTST USB adapter.

Cc: <stable@vger.kernel.org>
Reported-by: Joerg Kalisch <the_force@gmx.de>
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agocw1200: Use a threaded oneshot irq handler for cw1200_spi
Solomon Peachy [Mon, 23 Sep 2013 20:00:04 +0000 (16:00 -0400)]
cw1200: Use a threaded oneshot irq handler for cw1200_spi

This supercedes the older patch ("cw1200: Don't perform SPI transfers in
interrupt context") that badly attempted to fix this problem.

This is a far simpler solution, which has the added benefit of
actually working.

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoRevert "cw1200: Don't perform SPI transfers in interrupt context"
Solomon Peachy [Mon, 23 Sep 2013 20:00:03 +0000 (16:00 -0400)]
Revert "cw1200: Don't perform SPI transfers in interrupt context"

This reverts commit aec8e88c947b7017e2b4bbcb68a4bfc4a1f8ad35.

This solution turned out to cause interrupt delivery problems, and
rather than trying to fix this approach, it has been scrapped in favor
of an alternative (and far simpler) implementation.

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agomwifiex: fix PCIe hs_cfg cancel cmd timeout
Bing Zhao [Sat, 21 Sep 2013 02:56:45 +0000 (19:56 -0700)]
mwifiex: fix PCIe hs_cfg cancel cmd timeout

For pcie8897, the hs_cfg cancel command (0xe5) times out when host
comes out of suspend. This is caused by an incompleted host sleep
handshake between driver and firmware.

Like SDIO interface, PCIe also needs to go through firmware power
save events to complete the handshake for host sleep configuration.
Only USB interface doesn't require power save events for hs_cfg.

Cc: <stable@vger.kernel.org> # 3.10+
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agortlwifi: Align private space in rtl_priv struct
Larry Finger [Thu, 19 Sep 2013 02:21:35 +0000 (21:21 -0500)]
rtlwifi: Align private space in rtl_priv struct

The private array at the end of the rtl_priv struct is not aligned.
On ARM architecture, this causes an alignment trap and is fixed by aligning
that array with __align(sizeof(void *)). That should properly align that
space according to the requirements of all architectures.

Reported-by: Jason Andrews <jasona@cadence.com>
Tested-by: Jason Andrews <jasona@cadence.com>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: Stable <stable@vger.kernel.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoath9k: add txq locking for ath_tx_aggr_start
Felix Fietkau [Wed, 18 Sep 2013 13:23:41 +0000 (15:23 +0200)]
ath9k: add txq locking for ath_tx_aggr_start

Prevents race conditions when un-aggregated frames are pending in the
driver.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agop54usb: fix leak at failure path in p54u_load_firmware()
Alexey Khoroshilov [Tue, 17 Sep 2013 20:57:59 +0000 (00:57 +0400)]
p54usb: fix leak at failure path in p54u_load_firmware()

If request_firmware_nowait() fails in p54u_load_firmware(),
p54u_load_firmware_cb is not called and no one decrements usb_dev refcnt.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoath9k: don't use BAW tracking on PS responses for non-AMPDU packets
Felix Fietkau [Tue, 17 Sep 2013 10:05:15 +0000 (12:05 +0200)]
ath9k: don't use BAW tracking on PS responses for non-AMPDU packets

When .release_buffered_frames was implemented, only A-MPDU packets were
buffered internally. Now that this has changed, the BUF_AMPDU flag needs
to be checked before calling ath_tx_addto_baw

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoath9k: Fix regression in LNA diversity
Sujith Manoharan [Mon, 16 Sep 2013 04:54:51 +0000 (10:24 +0530)]
ath9k: Fix regression in LNA diversity

The commit "ath9k: Optimize LNA check" tried
to use the "rs_firstaggr" flag to optimize the LNA
combining algorithm when processing subframes in
an A-MPDU. This doesn't appear to work well in practice,
so revert it and use the old method of relying on
"rs_moreaggr".

Cc: stable@vger.kernel.org # 3.11
Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoath9k: do not link bf_next across multiple A-MPDUs
Felix Fietkau [Sat, 14 Sep 2013 15:02:29 +0000 (17:02 +0200)]
ath9k: do not link bf_next across multiple A-MPDUs

This might trip up tx completion processing, although the condition that
triggers this should not (yet) occur in practice.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoath9k: fix stale flag handling on buffer clone
Felix Fietkau [Sat, 14 Sep 2013 13:46:33 +0000 (15:46 +0200)]
ath9k: fix stale flag handling on buffer clone

Fixes a regression from commit
"ath9k: shrink a few data structures by reordering fields"

When cloning a buffer, the stale flag (part of bf_state now) needs to be
reset after copying the state to prevent tx processing hangs.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth...
John W. Linville [Thu, 26 Sep 2013 17:47:05 +0000 (13:47 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/bluetooth/bluetooth

11 years agoipv6: udp packets following an UFO enqueued packet need also be handled by UFO
Hannes Frederic Sowa [Sat, 21 Sep 2013 04:27:00 +0000 (06:27 +0200)]
ipv6: udp packets following an UFO enqueued packet need also be handled by UFO

In the following scenario the socket is corked:
If the first UDP packet is larger then the mtu we try to append it to the
write queue via ip6_ufo_append_data. A following packet, which is smaller
than the mtu would be appended to the already queued up gso-skb via
plain ip6_append_data. This causes random memory corruptions.

In ip6_ufo_append_data we also have to be careful to not queue up the
same skb multiple times. So setup the gso frame only when no first skb
is available.

This also fixes a shortcoming where we add the current packet's length to
cork->length but return early because of a packet > mtu with dontfrag set
(instead of sutracting it again).

Found with trinity.

Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlge: call ql_core_dump() only if dump memory was allocated.
malahal@us.ibm.com [Fri, 20 Sep 2013 21:21:17 +0000 (16:21 -0500)]
qlge: call ql_core_dump() only if dump memory was allocated.

Also changed a log message to indicate that memory was not allocated
instead of memory not available!

Signed-off-by: Malahal Naineni <malahal@us.ibm.com>
Acked-by: Jitendra Kalsaria <jitendra.kalsaria@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoskge: fix invalid value passed to pci_unmap_sigle
Mikulas Patocka [Fri, 20 Sep 2013 17:53:22 +0000 (13:53 -0400)]
skge: fix invalid value passed to pci_unmap_sigle

In my patch c194992cbe71c20bb3623a566af8d11b0bfaa721 ("skge: fix
broken driver") I didn't fix the skge bug correctly. The value of the
new mapping (not old) was passed to pci_unmap_single.

If we enable CONFIG_DMA_API_DEBUG, it results in this warning:
WARNING: CPU: 0 PID: 0 at lib/dma-debug.c:986 check_sync+0x4c4/0x580()
skge 0000:02:07.0: DMA-API: device driver tries to sync DMA memory it has
not allocated [device address=0x000000023a0096c0] [size=1536 bytes]

This patch makes the skge driver pass the correct value to
pci_unmap_single and fixes the warning. It copies the old descriptor to
on-stack variable "ee" and unmaps it if mapping of the new descriptor
succeeded.

This patch should be backported to 3.11-stable.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reported-by: Francois Romieu <romieu@fr.zoreil.com>
Tested-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: raw: do not report ICMP redirects to user space
Duan Jiong [Fri, 20 Sep 2013 10:21:25 +0000 (18:21 +0800)]
net: raw: do not report ICMP redirects to user space

Redirect isn't an error condition, it should leave
the error handler without touching the socket.

Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: udp: do not report ICMP redirects to user space
Duan Jiong [Fri, 20 Sep 2013 10:20:28 +0000 (18:20 +0800)]
net: udp: do not report ICMP redirects to user space

Redirect isn't an error condition, it should leave
the error handler without touching the socket.

Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agomrp: add periodictimer to allow retries when packets get lost
Noel Burton-Krahn [Wed, 18 Sep 2013 19:24:40 +0000 (12:24 -0700)]
mrp: add periodictimer to allow retries when packets get lost

MRP doesn't implement the periodictimer in 802.1Q, so it never retries
if packets get lost.  I ran into this problem when MRP sent a MVRP
JoinIn before the interface was fully up.  The JoinIn was lost, MRP
didn't retry, and MVRP registration failed.

Tested against Juniper QFabric switches

Signed-off-by: Noel Burton-Krahn <noel@burton-krahn.com>
Acked-by: David Ward <david.ward@ll.mit.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet/lapb: re-send packets on timeout
josselin.costanzi@mobile-devices.fr [Wed, 18 Sep 2013 10:00:35 +0000 (12:00 +0200)]
net/lapb: re-send packets on timeout

Actually re-send packets when the T1 timer runs out. This fixes a bug
where packets are waiting on the write queue until disconnection when
no other traffic is outstanding.

Signed-off-by: Josselin Costanzi <josselin.costanzi@mobile-devices.fr>
Signed-off-by: Maxime Jayat <maxime.jayat@mobile-devices.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoBluetooth: btusb: Add support for Belkin F8065bf
Ken O'Brien [Sat, 21 Sep 2013 18:14:43 +0000 (19:14 +0100)]
Bluetooth: btusb: Add support for Belkin F8065bf

Add generic rule on encountering Belkin bluetooth usb device F8065bf.

Relevant section from /sys/kernel/debug/usb/devices:

T:  Bus=03 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  2 Spd=12   MxCh= 0
D:  Ver= 2.00 Cls=ff(vend.) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=050d ProdID=065a Rev= 1.12
S:  Manufacturer=Broadcom Corp
S:  Product=BCM20702A0
S:  SerialNumber=0002723E2D29
C:* #Ifs= 4 Cfg#= 1 Atr=a0 MxPwr=100mA
I:* If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=01 Prot=01 Driver=btusb
E:  Ad=81(I) Atr=03(Int.) MxPS=  16 Ivl=1ms
E:  Ad=82(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=   0 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=   0 Ivl=1ms
I:  If#= 1 Alt= 1 #EPs= 2 Cls=ff(vend.) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=   9 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=   9 Ivl=1ms

Signed-off-by: Ken O'Brien <kernel@kenobrien.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
11 years agoMerge branch 'bnx2x'
David S. Miller [Mon, 23 Sep 2013 20:17:16 +0000 (16:17 -0400)]
Merge branch 'bnx2x'

Yuval Mintz says:

====================
This patch contains various bug fixes, half of which are SR-IOV related
(some fixing issues in the recently added VF RSS support), while the other fix
a wide assortments of issues in the driver.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: handle known but unsupported VF messages
Ariel Elior [Mon, 23 Sep 2013 07:12:55 +0000 (10:12 +0300)]
bnx2x: handle known but unsupported VF messages

Commit b9871bcf "bnx2x: VF RSS support - PF side" has deprecated one of
the previous existing messages. If an old VF driver were to send this message
to the PF then the PF will not reply and leave the mailbox in an unsteady
state (and cause a timeout on the VF side).

Wait until firmware ack is written before unlocking channel

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: prevent masked MCP parities from appearing
Yuval Mintz [Mon, 23 Sep 2013 07:12:54 +0000 (10:12 +0300)]
bnx2x: prevent masked MCP parities from appearing

During flows which mask block attentions (e.g., register dump) all parities
are masked. However, unlike other blocks the MCP's attention is not masked
inside the block but rather the indication to the driver. If another attention
(e.g., link change) will occour while there's an MCP parity, the driver will
ignore the fact that the parity is masked and erroneously report a parity.

This patch forces the driver to read the MCP masking while checking for
parities.

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: prevent masking error from cnic
Yuval Mintz [Mon, 23 Sep 2013 07:12:53 +0000 (10:12 +0300)]
bnx2x: prevent masking error from cnic

During error flows while loading cnic the return value was incorrectly replaced
by that of bnx2x_set_real_num_queues(); If that function was to finish
successfully then the cnic would have mistakenly thought the load ended
successfully, causing issues (& panics) later on.

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: add missing VF resource allocation during init
Ariel Elior [Mon, 23 Sep 2013 07:12:52 +0000 (10:12 +0300)]
bnx2x: add missing VF resource allocation during init

bnx2x_iov_static_resc() should be called after IGU was read for information on
the number of available VFs, so that resources will be correctly set.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Fix support for VFs on some PFs
Ariel Elior [Mon, 23 Sep 2013 07:12:51 +0000 (10:12 +0300)]
bnx2x: Fix support for VFs on some PFs

Due to incorrect usage of PF macros when reading information relating to
interrupts, some PFs were erroneously unable to support VFs.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Prevent mistaken hangup between driver & FW
Eilon Greenstein [Mon, 23 Sep 2013 07:12:50 +0000 (10:12 +0300)]
bnx2x: Prevent mistaken hangup between driver & FW

When system CPU is stressed it's possible that the driver will not be able
to pulse the FW every second, which will cause the log to be filled with
error messages.

Increasing the threshold to 5 seconds seems to be enough to eliminate the
issue.

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Fix 848xx duplex settings
Yaniv Rosner [Sun, 22 Sep 2013 11:59:26 +0000 (14:59 +0300)]
bnx2x: Fix 848xx duplex settings

On 848xx PHY (10G-baseT), half-duplex was always advertised regardless of the
actual configuration. Change the 848xx duplex settings to advertise half-duplex
only if configured.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Specific Active-DAC is not detected on 57810
Yaniv Rosner [Sun, 22 Sep 2013 11:59:25 +0000 (14:59 +0300)]
bnx2x: Specific Active-DAC is not detected on 57810

Fix Warpcore mode setting when active DAC (Direct Attached Cable) is detected.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: 57840 non-external loopback test fail on 1G
Yaniv Rosner [Sun, 22 Sep 2013 11:59:24 +0000 (14:59 +0300)]
bnx2x: 57840 non-external loopback test fail on 1G

when 1G-optic module was plugged in, internal loopback test failed because the
driver used to check the optic module (with no need), and for 1G optic module,
the link speed was forced down to 1G, while the XMAC (10G MAC) was enabled.
This patch avoid accessing optic module in case internal loopback was selected,
and update the link speed in case 1G optic module was detected during init
stage.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: KR2 disablement fix
Yaniv Rosner [Sun, 22 Sep 2013 11:59:23 +0000 (14:59 +0300)]
bnx2x: KR2 disablement fix

Relocate bnx2x_disable_kr2 function, and use it to disable KR2 in case it is not
configured in order to clear it's configuration, otherwise the link may come up
at 20G instead of the requested 10G-KR. In addition, restart AN after
disabling KR2 as part of the KR2 work-around.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Generalize KR work-around
Yaniv Rosner [Sun, 22 Sep 2013 11:59:22 +0000 (14:59 +0300)]
bnx2x: Generalize KR work-around

Previously, in case of KR link down, the driver would reset the PHY and restart
auto negotiation only when old Warpcore microcode was used (below D108).
This patch comes to generalize this by keep trying to restart KR link,
regardless of Warpcore microcode, since it was found that it solves another link
issue which source is a link-partner. As part of this change, the signal
detect is no longer a condition to apply the work-around to cover this new case.
Like before, as long as the link is down, AN will be restarted every 2 seconds.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'sfc-3.12' of git://git.kernel.org/pub/scm/linux/kernel/git/bwh/sfc
David S. Miller [Sat, 21 Sep 2013 18:20:08 +0000 (14:20 -0400)]
Merge branch 'sfc-3.12' of git://git./linux/kernel/git/bwh/sfc

Ben Hutchings says:

====================
Daniel Pieczko fixed two bugs in reset handling that particularly
affected the new SFC9120 controller.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoslcan: rewrite of slc_bump and slc_encaps
Andre Naujoks [Fri, 13 Sep 2013 17:37:13 +0000 (19:37 +0200)]
slcan: rewrite of slc_bump and slc_encaps

The old implementation was heavy on str* functions and sprintf calls.
This version is more manual, but faster.

Profiling just the printing of a 3 char CAN-id resulted in 60 instructions
for the manual method and over 2000 for the sprintf method. Bear in
mind the profiling was done against libc and not the kernel sprintf.

Together with this rewrite an issue with sending and receiving of RTR frames
has been fixed by Oliver for the cases that the DLC is not zero.

Signed-off-by: Andre Naujoks <nautsch2@gmail.com>
Tested-by: Oliver Hartkopp <socketcan@hartkopp.net>
Acked-by: Oliver Hartkopp <socketcan@hartkopp.net>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agolib: introduce upper case hex ascii helpers
Andre Naujoks [Fri, 13 Sep 2013 17:37:12 +0000 (19:37 +0200)]
lib: introduce upper case hex ascii helpers

To be able to use the hex ascii functions in case sensitive environments
the array hex_asc_upper[] and the needed functions for hex_byte_pack_upper()
are introduced.

Signed-off-by: Andre Naujoks <nautsch2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoslip/slcan: added locking in wakeup function
Andre Naujoks [Fri, 13 Sep 2013 17:37:11 +0000 (19:37 +0200)]
slip/slcan: added locking in wakeup function

The locking is needed, since the the internal buffer for the CAN frames is
changed during the wakeup call. This could cause buffer inconsistencies
under high loads, especially for the outgoing short CAN packet skbuffs.

The needed locks led to deadlocks before commit
"5ede52538ee2b2202d9dff5b06c33bfde421e6e4 tty: Remove extra wakeup from pty
write() path", which removed the direct callback to the wakeup function from the
tty layer.

As slcan.c is based on slip.c the issue in the original code is fixed, too.

Signed-off-by: Andre Naujoks <nautsch2@gmail.com>
Acked-by: Oliver Hartkopp <socketcan@hartkopp.net>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoBluetooth: don't release the port in rfcomm_dev_state_change()
Gianluca Anzolin [Tue, 27 Aug 2013 16:28:46 +0000 (18:28 +0200)]
Bluetooth: don't release the port in rfcomm_dev_state_change()

When the dlc is closed, rfcomm_dev_state_change() tries to release the
port in the case it cannot get a reference to the tty. However this is
racy and not even needed.

Infact as Peter Hurley points out:

1. Only consider dlcs that are 'stolen' from a connected socket, ie.
   reused. Allocated dlcs cannot have been closed prior to port
   activate and so for these dlcs a tty reference will always be avail
   in rfcomm_dev_state_change() -- except for the conditions covered by
   #2b below.
2. If a tty was at some point previously created for this rfcomm, then
   either
   (a) the tty reference is still avail, so rfcomm_dev_state_change()
       will perform a hangup. So nothing to do, or,
   (b) the tty reference is no longer avail, and the tty_port will be
       destroyed by the last tty_port_put() in rfcomm_tty_cleanup.
       Again, no action required.
3. Prior to obtaining the dlc lock in rfcomm_dev_add(),
   rfcomm_dev_state_change() will not 'see' a rfcomm_dev so nothing to
   do here.
4. After releasing the dlc lock in rfcomm_dev_add(),
   rfcomm_dev_state_change() will 'see' an incomplete rfcomm_dev if a
   tty reference could not be obtained. Again, the best thing to do here
   is nothing. Any future attempted open() will block on
   rfcomm_dev_carrier_raised(). The unconnected device will exist until
   released by ioctl(RFCOMMRELEASEDEV).

The patch removes the aforementioned code and uses the
tty_port_tty_hangup() helper to hangup the tty.

Signed-off-by: Gianluca Anzolin <gianluca@sottospazio.it>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
11 years agocan: pcan_usb_core: fix memory leak on failure paths in peak_usb_start()
Alexey Khoroshilov [Wed, 4 Sep 2013 21:46:09 +0000 (01:46 +0400)]
can: pcan_usb_core: fix memory leak on failure paths in peak_usb_start()

Tx and rx urbs are not deallocated if something goes wrong in peak_usb_start().
The patch fixes error handling to deallocate all the resources.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Acked-by: Stephane Grosjean <s.grosjean@peak-system.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agosfc: Increase MCDI status timeout to 250ms
Daniel Pieczko [Fri, 20 Sep 2013 15:45:10 +0000 (16:45 +0100)]
sfc: Increase MCDI status timeout to 250ms

The SFC9120 MC firmware often takes longer than 20ms to reboot and
update the warm boot count in BIU_MC_SFT_STATUS_REG.  A timeout of
250ms is very generous for an MC reboot.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>