profile/ivi/kernel-x86-ivi.git
11 years agotun: compare with 0 instead of total_len
Weiping Pan [Tue, 13 Aug 2013 13:46:56 +0000 (21:46 +0800)]
tun: compare with 0 instead of total_len

Since we set "len = total_len" in the beginning of tun_get_user(),
so we should compare the new len with 0, instead of total_len,
or the if statement always returns false.

Signed-off-by: Weiping Pan <wpan@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agortnetlink: rtnl_bridge_getlink: Call nlmsg_find_attr() with ifinfomsg header
Asbjoern Sloth Toennesen [Mon, 12 Aug 2013 16:30:09 +0000 (16:30 +0000)]
rtnetlink: rtnl_bridge_getlink: Call nlmsg_find_attr() with ifinfomsg header

Fix the iproute2 command `bridge vlan show`, after switching from
rtgenmsg to ifinfomsg.

Let's start with a little history:

Feb 20:   Vlad Yasevich got his VLAN-aware bridge patchset included in
          the 3.9 merge window.
          In the kernel commit 6cbdceeb, he added attribute support to
          bridge GETLINK requests sent with rtgenmsg.

Mar 6th:  Vlad got this iproute2 reference implementation of the bridge
          vlan netlink interface accepted (iproute2 9eff0e5c)

Apr 25th: iproute2 switched from using rtgenmsg to ifinfomsg (63338dca)
          http://patchwork.ozlabs.org/patch/239602/
          http://marc.info/?t=136680900700007

Apr 28th: Linus released 3.9

Apr 30th: Stephen released iproute2 3.9.0

The `bridge vlan show` command haven't been working since the switch to
ifinfomsg, or in a released version of iproute2. Since the kernel side
only supports rtgenmsg, which iproute2 switched away from just prior to
the iproute2 3.9.0 release.

I haven't been able to find any documentation, about neither rtgenmsg
nor ifinfomsg, and in which situation to use which, but kernel commit
88c5b5ce seams to suggest that ifinfomsg should be used.

Fixing this in kernel will break compatibility, but I doubt that anybody
have been using it due to this bug in the user space reference
implementation, at least not without noticing this bug. That said the
functionality is still fully functional in 3.9, when reversing iproute2
commit 63338dca.

This could also be fixed in iproute2, but thats an ugly patch that would
reintroduce rtgenmsg in iproute2, and from searching in netdev it seams
like rtgenmsg usage is discouraged. I'm assuming that the only reason
that Vlad implemented the kernel side to use rtgenmsg, was because
iproute2 was using it at the time.

Signed-off-by: Asbjoern Sloth Toennesen <ast@fiberby.net>
Reviewed-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoethernet/arc/arc_emac - fix NAPI "work > weight" warning
Alexey Brodkin [Tue, 13 Aug 2013 13:04:36 +0000 (17:04 +0400)]
ethernet/arc/arc_emac - fix NAPI "work > weight" warning

Initially I improperly set a boundary for maximum number of input
packets to process on NAPI poll ("work") so it might be more than
expected amount ("weight").

This was really harmless but seeing WARN_ON_ONCE on every device boot is
not nice. So trivial fix ("<" instead of "<=") is here.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Mischa Jonker <mjonker@synopsys.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Grant Likely <grant.likely@linaro.org>
Cc: Rob Herring <rob.herring@calxeda.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip_tunnel: Do not use inner ip-header-id for tunnel ip-header-id.
Pravin B Shelar [Tue, 13 Aug 2013 08:41:06 +0000 (01:41 -0700)]
ip_tunnel: Do not use inner ip-header-id for tunnel ip-header-id.

Using inner-id for tunnel id is not safe in some rare cases.
E.g. packets coming from multiple sources entering same tunnel
can have same id. Therefore on tunnel packet receive we
could have packets from two different stream but with same
source and dst IP with same ip-id which could confuse ip packet
reassembly.

Following patch reverts optimization from commit
490ab08127 (IP_GRE: Fix IP-Identification.)

CC: Jarno Rajahalme <jrajahalme@nicira.com>
CC: Ansis Atteka <aatteka@nicira.com>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'bnx2x'
David S. Miller [Tue, 13 Aug 2013 23:04:38 +0000 (16:04 -0700)]
Merge branch 'bnx2x'

Dmitry Kravkov says:

====================
Please consider applying the series of bnx2x fixes to net:
* statistics may cause FW assert
* missing fairness configuration in DCB flow
* memory leak in sriov related part
* Illegal PTE access
* Pagefault crash in shutdown flow with cnic
v1->v2
* fixed sparse error pointed by Joe Perches
* added missing signed-off from Sergei Shtylyov
v2->v3
* added missing signed-off from Sergei Shtylyov
* fixed formatting from Sergei Shtylyov
v3->v4
* patch 1/6: fixed declaration order
* patch 2/6 replaced with: protect flows using set_bit constraints
v4->v5
* patch 2/6: replace proprietary locking with semaphore
* droped 1/6: since adds redundant code from Benjamin Poirier
The following patchset contains four netfilter fixes, they are:

* Fix possible invalid access and mangling of the TCPMSS option in
  xt_TCPMSS. This was spotted by Julian Anastasov.

* Fix possible off by one access and mangling of the TCP packet in
  xt_TCPOPTSTRIP, also spotted by Julian Anastasov.

* Fix possible information leak due to missing initialization of one
  padding field of several structures that are included in nfqueue and
  nflog netlink messages, from Dan Carpenter.

* Fix TCP window tracking with Fast Open, from Yuchung Cheng.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: prevent crash in shutdown flow with CNIC
Yuval Mintz [Mon, 12 Aug 2013 23:25:03 +0000 (02:25 +0300)]
bnx2x: prevent crash in shutdown flow with CNIC

There might be a crash as during shutdown flow CNIC might try
to access resources already freed by bnx2x.
Change bnx2x_close() into dev_close() in __bnx2x_remove (shutdown flow)
to guarantee CNIC is notified of the device's change of status.

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@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: fix PTE write access error
Barak Witkowsky [Mon, 12 Aug 2013 23:25:02 +0000 (02:25 +0300)]
bnx2x: fix PTE write access error

PTE write access error  might occur in MF_ALLOWED mode when IOMMU
is active. The patch adds rmmod HSI indicating to MFW to stop
running queries which might trigger this failure.

Signed-off-by: Barak Witkowsky <barak@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@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: fix memory leak in VF
Ariel Elior [Mon, 12 Aug 2013 23:25:01 +0000 (02:25 +0300)]
bnx2x: fix memory leak in VF

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: update fairness parameters following DCB negotiation
Dmitry Kravkov [Mon, 12 Aug 2013 23:25:00 +0000 (02:25 +0300)]
bnx2x: update fairness parameters following DCB negotiation

ETS can be enabled as a result of DCB negotiation, then
fairness must be recalculated after each negotiation.

Signed-off-by: Dmitry Kravkov <dmitry@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: protect different statistics flows
Dmitry Kravkov [Mon, 12 Aug 2013 23:24:59 +0000 (02:24 +0300)]
bnx2x: protect different statistics flows

Add locking to protect different statistics flows from
running simultaneously.
This in order to serialize statistics requests sent to FW,
otherwise two outstanding queries may cause FW assert.

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agomyri10ge: Update MAINTAINERS
Hyong-Youb Kim [Mon, 12 Aug 2013 09:20:02 +0000 (02:20 -0700)]
myri10ge: Update MAINTAINERS

Remove Andrew Gallatin, as he is no longer with Myricom. Add
Hyong-Youb Kim as the new maintainer. Update the website URL.

Signed-off-by: Hyong-Youb Kim <hykim@myri.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoskge: dma_sync the whole receive buffer
stephen hemminger [Sat, 10 Aug 2013 22:02:07 +0000 (15:02 -0700)]
skge: dma_sync the whole receive buffer

The DMA sync should sync the whole receive buffer, not just
part of it. Fixes log messages dma_sync_check.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agogenetlink: fix family dump race
Johannes Berg [Tue, 13 Aug 2013 07:04:05 +0000 (09:04 +0200)]
genetlink: fix family dump race

When dumping generic netlink families, only the first dump call
is locked with genl_lock(), which protects the list of families,
and thus subsequent calls can access the data without locking,
racing against family addition/removal. This can cause a crash.
Fix it - the locking needs to be conditional because the first
time around it's already locked.

A similar bug was reported to me on an old kernel (3.4.47) but
the exact scenario that happened there is no longer possible,
on those kernels the first round wasn't locked either. Looking
at the current code I found the race described above, which had
also existed on the old kernel.

Cc: stable@vger.kernel.org
Reported-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: sctp: sctp_transport_destroy{, _rcu}: fix potential pointer corruption
Daniel Borkmann [Fri, 9 Aug 2013 14:25:21 +0000 (16:25 +0200)]
net: sctp: sctp_transport_destroy{, _rcu}: fix potential pointer corruption

Probably this one is quite unlikely to be triggered, but it's more safe
to do the call_rcu() at the end after we have dropped the reference on
the asoc and freed sctp packet chunks. The reason why is because in
sctp_transport_destroy_rcu() the transport is being kfree()'d, and if
we're unlucky enough we could run into corrupted pointers. Probably
that's more of theoretical nature, but it's safer to have this simple fix.

Introduced by commit 8c98653f ("sctp: sctp_close: fix release of bindings
for deferred call_rcu's"). I also did the 8c98653f regression test and
it's fine that way.

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: sctp: sctp_assoc_control_transport: fix MTU size in SCTP_PF state
Daniel Borkmann [Fri, 9 Aug 2013 13:09:08 +0000 (15:09 +0200)]
net: sctp: sctp_assoc_control_transport: fix MTU size in SCTP_PF state

The SCTP Quick failover draft [1] section 5.1, point 5 says that the cwnd
should be 1 MTU. So, instead of 1, set it to 1 MTU.

  [1] https://tools.ietf.org/html/draft-nishida-tsvwg-sctp-failover-05

Reported-by: Karl Heiss <kheiss@gmail.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agostmmac: fix init_dma_desc_rings() to handle errors
Bartlomiej Zolnierkiewicz [Fri, 9 Aug 2013 12:02:08 +0000 (14:02 +0200)]
stmmac: fix init_dma_desc_rings() to handle errors

In stmmac_init_rx_buffers():
* add missing handling of dma_map_single() error
* remove superfluous unlikely() optimization while at it

Add stmmac_free_rx_buffers() helper and use it in dma_free_rx_skbufs().

In init_dma_desc_rings():
* add missing handling of kmalloc_array() errors
* fix handling of dma_alloc_coherent() and stmmac_init_rx_buffers() errors
* make function return an error value on error and 0 on success

In stmmac_open():
* add handling of init_dma_desc_rings() return value

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 agotipc: avoid possible deadlock while enable and disable bearer
dingtianhong [Fri, 9 Aug 2013 09:12:58 +0000 (17:12 +0800)]
tipc: avoid possible deadlock while enable and disable bearer

We met lockdep warning when enable and disable the bearer for commands such as:

tipc-config -netid=1234 -addr=1.1.3 -be=eth:eth0
tipc-config -netid=1234 -addr=1.1.3 -bd=eth:eth0

---------------------------------------------------

[  327.693595] ======================================================
[  327.693994] [ INFO: possible circular locking dependency detected ]
[  327.694519] 3.11.0-rc3-wwd-default #4 Tainted: G           O
[  327.694882] -------------------------------------------------------
[  327.695385] tipc-config/5825 is trying to acquire lock:
[  327.695754]  (((timer))#2){+.-...}, at: [<ffffffff8105be80>] del_timer_sync+0x0/0xd0
[  327.696018]
[  327.696018] but task is already holding lock:
[  327.696018]  (&(&b_ptr->lock)->rlock){+.-...}, at: [<ffffffffa02be58d>] bearer_disable+  0xdd/0x120 [tipc]
[  327.696018]
[  327.696018] which lock already depends on the new lock.
[  327.696018]
[  327.696018]
[  327.696018] the existing dependency chain (in reverse order) is:
[  327.696018]
[  327.696018] -> #1 (&(&b_ptr->lock)->rlock){+.-...}:
[  327.696018]        [<ffffffff810b3b4d>] validate_chain+0x6dd/0x870
[  327.696018]        [<ffffffff810b40bb>] __lock_acquire+0x3db/0x670
[  327.696018]        [<ffffffff810b4453>] lock_acquire+0x103/0x130
[  327.696018]        [<ffffffff814d65b1>] _raw_spin_lock_bh+0x41/0x80
[  327.696018]        [<ffffffffa02c5d48>] disc_timeout+0x18/0xd0 [tipc]
[  327.696018]        [<ffffffff8105b92a>] call_timer_fn+0xda/0x1e0
[  327.696018]        [<ffffffff8105bcd7>] run_timer_softirq+0x2a7/0x2d0
[  327.696018]        [<ffffffff8105379a>] __do_softirq+0x16a/0x2e0
[  327.696018]        [<ffffffff81053a35>] irq_exit+0xd5/0xe0
[  327.696018]        [<ffffffff81033005>] smp_apic_timer_interrupt+0x45/0x60
[  327.696018]        [<ffffffff814df4af>] apic_timer_interrupt+0x6f/0x80
[  327.696018]        [<ffffffff8100b70e>] arch_cpu_idle+0x1e/0x30
[  327.696018]        [<ffffffff810a039d>] cpu_idle_loop+0x1fd/0x280
[  327.696018]        [<ffffffff810a043e>] cpu_startup_entry+0x1e/0x20
[  327.696018]        [<ffffffff81031589>] start_secondary+0x89/0x90
[  327.696018]
[  327.696018] -> #0 (((timer))#2){+.-...}:
[  327.696018]        [<ffffffff810b33fe>] check_prev_add+0x43e/0x4b0
[  327.696018]        [<ffffffff810b3b4d>] validate_chain+0x6dd/0x870
[  327.696018]        [<ffffffff810b40bb>] __lock_acquire+0x3db/0x670
[  327.696018]        [<ffffffff810b4453>] lock_acquire+0x103/0x130
[  327.696018]        [<ffffffff8105bebd>] del_timer_sync+0x3d/0xd0
[  327.696018]        [<ffffffffa02c5855>] tipc_disc_delete+0x15/0x30 [tipc]
[  327.696018]        [<ffffffffa02be59f>] bearer_disable+0xef/0x120 [tipc]
[  327.696018]        [<ffffffffa02be74f>] tipc_disable_bearer+0x2f/0x60 [tipc]
[  327.696018]        [<ffffffffa02bfb32>] tipc_cfg_do_cmd+0x2e2/0x550 [tipc]
[  327.696018]        [<ffffffffa02c8c79>] handle_cmd+0x49/0xe0 [tipc]
[  327.696018]        [<ffffffff8143e898>] genl_family_rcv_msg+0x268/0x340
[  327.696018]        [<ffffffff8143ed30>] genl_rcv_msg+0x70/0xd0
[  327.696018]        [<ffffffff8143d4c9>] netlink_rcv_skb+0x89/0xb0
[  327.696018]        [<ffffffff8143e617>] genl_rcv+0x27/0x40
[  327.696018]        [<ffffffff8143d21e>] netlink_unicast+0x15e/0x1b0
[  327.696018]        [<ffffffff8143ddcf>] netlink_sendmsg+0x22f/0x400
[  327.696018]        [<ffffffff813f7836>] __sock_sendmsg+0x66/0x80
[  327.696018]        [<ffffffff813f7957>] sock_aio_write+0x107/0x120
[  327.696018]        [<ffffffff8117f76d>] do_sync_write+0x7d/0xc0
[  327.696018]        [<ffffffff8117fc56>] vfs_write+0x186/0x190
[  327.696018]        [<ffffffff811803e0>] SyS_write+0x60/0xb0
[  327.696018]        [<ffffffff814de852>] system_call_fastpath+0x16/0x1b
[  327.696018]
[  327.696018] other info that might help us debug this:
[  327.696018]
[  327.696018]  Possible unsafe locking scenario:
[  327.696018]
[  327.696018]        CPU0                    CPU1
[  327.696018]        ----                    ----
[  327.696018]   lock(&(&b_ptr->lock)->rlock);
[  327.696018]                                lock(((timer))#2);
[  327.696018]                                lock(&(&b_ptr->lock)->rlock);
[  327.696018]   lock(((timer))#2);
[  327.696018]
[  327.696018]  *** DEADLOCK ***
[  327.696018]
[  327.696018] 5 locks held by tipc-config/5825:
[  327.696018]  #0:  (cb_lock){++++++}, at: [<ffffffff8143e608>] genl_rcv+0x18/0x40
[  327.696018]  #1:  (genl_mutex){+.+.+.}, at: [<ffffffff8143ed66>] genl_rcv_msg+0xa6/0xd0
[  327.696018]  #2:  (config_mutex){+.+.+.}, at: [<ffffffffa02bf889>] tipc_cfg_do_cmd+0x39/ 0x550 [tipc]
[  327.696018]  #3:  (tipc_net_lock){++.-..}, at: [<ffffffffa02be738>] tipc_disable_bearer+ 0x18/0x60 [tipc]
[  327.696018]  #4:  (&(&b_ptr->lock)->rlock){+.-...}, at: [<ffffffffa02be58d>]             bearer_disable+0xdd/0x120 [tipc]
[  327.696018]
[  327.696018] stack backtrace:
[  327.696018] CPU: 2 PID: 5825 Comm: tipc-config Tainted: G           O 3.11.0-rc3-wwd-    default #4
[  327.696018] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
[  327.696018]  00000000ffffffff ffff880037fa77a8 ffffffff814d03dd 0000000000000000
[  327.696018]  ffff880037fa7808 ffff880037fa77e8 ffffffff810b1c4f 0000000037fa77e8
[  327.696018]  ffff880037fa7808 ffff880037e4db40 0000000000000000 ffff880037e4e318
[  327.696018] Call Trace:
[  327.696018]  [<ffffffff814d03dd>] dump_stack+0x4d/0xa0
[  327.696018]  [<ffffffff810b1c4f>] print_circular_bug+0x10f/0x120
[  327.696018]  [<ffffffff810b33fe>] check_prev_add+0x43e/0x4b0
[  327.696018]  [<ffffffff810b3b4d>] validate_chain+0x6dd/0x870
[  327.696018]  [<ffffffff81087a28>] ? sched_clock_cpu+0xd8/0x110
[  327.696018]  [<ffffffff810b40bb>] __lock_acquire+0x3db/0x670
[  327.696018]  [<ffffffff810b4453>] lock_acquire+0x103/0x130
[  327.696018]  [<ffffffff8105be80>] ? try_to_del_timer_sync+0x70/0x70
[  327.696018]  [<ffffffff8105bebd>] del_timer_sync+0x3d/0xd0
[  327.696018]  [<ffffffff8105be80>] ? try_to_del_timer_sync+0x70/0x70
[  327.696018]  [<ffffffffa02c5855>] tipc_disc_delete+0x15/0x30 [tipc]
[  327.696018]  [<ffffffffa02be59f>] bearer_disable+0xef/0x120 [tipc]
[  327.696018]  [<ffffffffa02be74f>] tipc_disable_bearer+0x2f/0x60 [tipc]
[  327.696018]  [<ffffffffa02bfb32>] tipc_cfg_do_cmd+0x2e2/0x550 [tipc]
[  327.696018]  [<ffffffff81218783>] ? security_capable+0x13/0x20
[  327.696018]  [<ffffffffa02c8c79>] handle_cmd+0x49/0xe0 [tipc]
[  327.696018]  [<ffffffff8143e898>] genl_family_rcv_msg+0x268/0x340
[  327.696018]  [<ffffffff8143ed30>] genl_rcv_msg+0x70/0xd0
[  327.696018]  [<ffffffff8143ecc0>] ? genl_lock+0x20/0x20
[  327.696018]  [<ffffffff8143d4c9>] netlink_rcv_skb+0x89/0xb0
[  327.696018]  [<ffffffff8143e608>] ? genl_rcv+0x18/0x40
[  327.696018]  [<ffffffff8143e617>] genl_rcv+0x27/0x40
[  327.696018]  [<ffffffff8143d21e>] netlink_unicast+0x15e/0x1b0
[  327.696018]  [<ffffffff81289d7c>] ? memcpy_fromiovec+0x6c/0x90
[  327.696018]  [<ffffffff8143ddcf>] netlink_sendmsg+0x22f/0x400
[  327.696018]  [<ffffffff813f7836>] __sock_sendmsg+0x66/0x80
[  327.696018]  [<ffffffff813f7957>] sock_aio_write+0x107/0x120
[  327.696018]  [<ffffffff813fe29c>] ? release_sock+0x8c/0xa0
[  327.696018]  [<ffffffff8117f76d>] do_sync_write+0x7d/0xc0
[  327.696018]  [<ffffffff8117fa24>] ? rw_verify_area+0x54/0x100
[  327.696018]  [<ffffffff8117fc56>] vfs_write+0x186/0x190
[  327.696018]  [<ffffffff811803e0>] SyS_write+0x60/0xb0
[  327.696018]  [<ffffffff814de852>] system_call_fastpath+0x16/0x1b

-----------------------------------------------------------------------

The problem is that the tipc_link_delete() will cancel the timer disc_timeout() when
the b_ptr->lock is hold, but the disc_timeout() still call b_ptr->lock to finish the
work, so the dead lock occurs.

We should unlock the b_ptr->lock when del the disc_timeout().

Remove link_timeout() still met the same problem, the patch:

http://article.gmane.org/gmane.network.tipc.general/4380

fix the problem, so no need to send patch for fix link_timeout() deadlock warming.

Signed-off-by: Wang Weidong <wangweidong1@huawei.com>
Signed-off-by: Ding Tianhong <dingtianhong@huawei.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agocan: pcan_usb: fix wrong memcpy() bytes length
Stephane Grosjean [Fri, 9 Aug 2013 09:44:06 +0000 (11:44 +0200)]
can: pcan_usb: fix wrong memcpy() bytes length

Fix possibly wrong memcpy() bytes length since some CAN records received from
PCAN-USB could define a DLC field in range [9..15].
In that case, the real DLC value MUST be used to move forward the record pointer
but, only 8 bytes max. MUST be copied into the data field of the struct
can_frame object of the skb given to the network core.

Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agomacvtap: fix two races
Eric Dumazet [Thu, 8 Aug 2013 15:06:14 +0000 (08:06 -0700)]
macvtap: fix two races

Since commit ac4e4af1e59e1 ("macvtap: Consistently use rcu functions"),
Thomas gets two different warnings :

BUG: using smp_processor_id() in preemptible [00000000] code: vhost-45891/45892
caller is macvtap_do_read+0x45c/0x600 [macvtap]
CPU: 1 PID: 45892 Comm: vhost-45891 Not tainted 3.11.0-bisecttest #13
Call Trace:
([<00000000001126ee>] show_trace+0x126/0x144)
 [<00000000001127d2>] show_stack+0xc6/0xd4
 [<000000000068bcec>] dump_stack+0x74/0xd8
 [<0000000000481066>] debug_smp_processor_id+0xf6/0x114
 [<000003ff802e9a18>] macvtap_do_read+0x45c/0x600 [macvtap]
 [<000003ff802e9c1c>] macvtap_recvmsg+0x60/0x88 [macvtap]
 [<000003ff80318c5e>] handle_rx+0x5b2/0x800 [vhost_net]
 [<000003ff8028f77c>] vhost_worker+0x15c/0x1c4 [vhost]
 [<000000000015f3ac>] kthread+0xd8/0xe4
 [<00000000006934a6>] kernel_thread_starter+0x6/0xc
 [<00000000006934a0>] kernel_thread_starter+0x0/0xc

And

BUG: using smp_processor_id() in preemptible [00000000] code: vhost-45897/45898
caller is macvlan_start_xmit+0x10a/0x1b4 [macvlan]
CPU: 1 PID: 45898 Comm: vhost-45897 Not tainted 3.11.0-bisecttest #16
Call Trace:
([<00000000001126ee>] show_trace+0x126/0x144)
 [<00000000001127d2>] show_stack+0xc6/0xd4
 [<000000000068bdb8>] dump_stack+0x74/0xd4
 [<0000000000481132>] debug_smp_processor_id+0xf6/0x114
 [<000003ff802b72ca>] macvlan_start_xmit+0x10a/0x1b4 [macvlan]
 [<000003ff802ea69a>] macvtap_get_user+0x982/0xbc4 [macvtap]
 [<000003ff802ea92a>] macvtap_sendmsg+0x4e/0x60 [macvtap]
 [<000003ff8031947c>] handle_tx+0x494/0x5ec [vhost_net]
 [<000003ff8028f77c>] vhost_worker+0x15c/0x1c4 [vhost]
 [<000000000015f3ac>] kthread+0xd8/0xe4
 [<000000000069356e>] kernel_thread_starter+0x6/0xc
 [<0000000000693568>] kernel_thread_starter+0x0/0xc
2 locks held by vhost-45897/45898:
 #0:  (&vq->mutex){+.+.+.}, at: [<000003ff8031903c>] handle_tx+0x54/0x5ec [vhost_net]
 #1:  (rcu_read_lock){.+.+..}, at: [<000003ff802ea53c>] macvtap_get_user+0x824/0xbc4 [macvtap]

In the first case, macvtap_put_user() calls macvlan_count_rx()
in a preempt-able context, and this is not allowed.

In the second case, macvtap_get_user() calls
macvlan_start_xmit() with BH enabled, and this is not allowed.

Reported-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Bisected-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Tested-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Cc: Vlad Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge tag 'batman-adv-fix-for-davem' of git://git.open-mesh.org/linux-merge
David S. Miller [Sat, 10 Aug 2013 22:38:59 +0000 (15:38 -0700)]
Merge tag 'batman-adv-fix-for-davem' of git://git.open-mesh.org/linux-merge

Included change:
- reassign pointers to data after skb reallocation to avoid kernel paging errors

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobatman-adv: fix potential kernel paging errors for unicast transmissions
Linus Lüssing [Tue, 6 Aug 2013 18:21:15 +0000 (20:21 +0200)]
batman-adv: fix potential kernel paging errors for unicast transmissions

There are several functions which might reallocate skb data. Currently
some places keep reusing their old ethhdr pointer regardless of whether
they became invalid after such a reallocation or not. This potentially
leads to kernel paging errors.

This patch fixes these by refetching the ethdr pointer after the
potential reallocations.

Signed-off-by: Linus Lüssing <linus.luessing@web.de>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf
David S. Miller [Sat, 10 Aug 2013 20:44:22 +0000 (13:44 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/pablo/nf

Pablo Neira Ayuso says:

====================
The following patchset contains four netfilter fixes, they are:

* Fix possible invalid access and mangling of the TCPMSS option in
  xt_TCPMSS. This was spotted by Julian Anastasov.

* Fix possible off by one access and mangling of the TCP packet in
  xt_TCPOPTSTRIP, also spotted by Julian Anastasov.

* Fix possible information leak due to missing initialization of one
  padding field of several structures that are included in nfqueue and
  nflog netlink messages, from Dan Carpenter.

* Fix TCP window tracking with Fast Open, from Yuchung Cheng.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonetfilter: nf_conntrack: fix tcp_in_window for Fast Open
Yuchung Cheng [Sat, 10 Aug 2013 00:21:27 +0000 (17:21 -0700)]
netfilter: nf_conntrack: fix tcp_in_window for Fast Open

Currently the conntrack checks if the ending sequence of a packet
falls within the observed receive window. However it does so even
if it has not observe any packet from the remote yet and uses an
uninitialized receive window (td_maxwin).

If a connection uses Fast Open to send a SYN-data packet which is
dropped afterward in the network. The subsequent SYNs retransmits
will all fail this check and be discarded, leading to a connection
timeout. This is because the SYN retransmit does not contain data
payload so

end == initial sequence number (isn) + 1
sender->td_end == isn + syn_data_len
receiver->td_maxwin == 0

The fix is to only apply this check after td_maxwin is initialized.

Reported-by: Michael Chan <mcfchan@stanford.edu>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Acked-by: Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
11 years agortnetlink: Fix inverted check in ndo_dflt_fdb_del()
Sridhar Samudrala [Thu, 8 Aug 2013 22:19:48 +0000 (15:19 -0700)]
rtnetlink: Fix inverted check in ndo_dflt_fdb_del()

Fix inverted check when deleting an fdb entry.

Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: stmmac: Fixed the condition of extend_desc for jumbo frame
Byungho An [Thu, 8 Aug 2013 06:30:26 +0000 (15:30 +0900)]
net: stmmac: Fixed the condition of extend_desc for jumbo frame

This patch fixed the condition of extend_desc for jumbo frame.
There is no check routine for extend_desc in the stmmac_jumbo_frm function.
Even though extend_desc is set if dma_tx is used instead of dma_etx.
It causes kernel panic.

Signed-off-by: Byungho An <bh74.an@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agovxlan: fix a soft lockup in vxlan module removal
Cong Wang [Wed, 7 Aug 2013 08:43:22 +0000 (16:43 +0800)]
vxlan: fix a soft lockup in vxlan module removal

This is a regression introduced by:

commit fe5c3561e6f0ac7c9546209f01351113c1b77ec8
Author: stephen hemminger <stephen@networkplumber.org>
Date:   Sat Jul 13 10:18:18 2013 -0700

    vxlan: add necessary locking on device removal

The problem is that vxlan_dellink(), which is called with RTNL lock
held, tries to flush the workqueue synchronously, but apparently
igmp_join and igmp_leave work need to hold RTNL lock too, therefore we
have a soft lockup!

As suggested by Stephen, probably the flush_workqueue can just be
removed and let the normal refcounting work. The workqueue has a
reference to device and socket, therefore the cleanups should work
correctly.

Suggested-by: Stephen Hemminger <stephen@networkplumber.org>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: David S. Miller <davem@davemloft.net>
Tested-by: Cong Wang <amwang@redhat.com>
Signed-off-by: Cong Wang <amwang@redhat.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agovxlan: fix a regression of igmp join
Cong Wang [Wed, 7 Aug 2013 08:35:45 +0000 (16:35 +0800)]
vxlan: fix a regression of igmp join

This is a regression introduced by:

commit 3fc2de2faba387218bdf9dbc6b13f513ac3b060a
Author: stephen hemminger <stephen@networkplumber.org>
Date:   Thu Jul 18 08:40:15 2013 -0700

    vxlan: fix igmp races

Before this commit, the old code was:

       if (vxlan_group_used(vn, vxlan->default_dst.remote_ip))
               ip_mc_join_group(sk, &mreq);
       else
               ip_mc_leave_group(sk, &mreq);

therefore we shoud check vxlan_group_used(), not its opposite,
for igmp_join.

Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: rename busy poll MIB counter
Eliezer Tamir [Wed, 7 Aug 2013 08:33:25 +0000 (11:33 +0300)]
net: rename busy poll MIB counter

Rename mib counter from "low latency" to "busy poll"

v1 also moved the counter to the ip MIB (suggested by Shawn Bohrer)
Eric Dumazet suggested that the current location is better.

So v2 just renames the counter to fit the new naming convention.

Signed-off-by: Eliezer Tamir <eliezer.tamir@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years ago8139cp: Fix skb leak in rx_status_loop failure path.
Dave Jones [Fri, 9 Aug 2013 18:16:34 +0000 (11:16 -0700)]
8139cp: Fix skb leak in rx_status_loop failure path.

Introduced in cf3c4c03060b688cbc389ebc5065ebcce5653e96
("8139cp: Add dma_mapping_error checking")

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: flow_dissector: add 802.1ad support
Eric Dumazet [Tue, 6 Aug 2013 11:35:06 +0000 (04:35 -0700)]
net: flow_dissector: add 802.1ad support

Same behavior than 802.1q : finds the encapsulated protocol and
skip 32bit header.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip_gre: fix ipgre_header to return correct offset
Timo Teräs [Tue, 6 Aug 2013 10:45:43 +0000 (13:45 +0300)]
ip_gre: fix ipgre_header to return correct offset

Fix ipgre_header() (header_ops->create) to return the correct
amount of bytes pushed. Most callers of dev_hard_header() seem
to care only if it was success, but af_packet.c uses it as
offset to the skb to copy from userspace only once. In practice
this fixes packet socket sendto()/sendmsg() to gre tunnels.

Regression introduced in c54419321455631079c7d6e60bc732dd0c5914c5
("GRE: Refactor GRE tunneling code.")

Cc: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Acked-by: Eric Dumazet <edumazet@google.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 [Thu, 8 Aug 2013 21:12:10 +0000 (14:12 -0700)]
Merge branch 'for-davem' of git://git./linux/kernel/git/linville/wireless

John W. Linville says:

====================
This is a batch of fixes intended for the 3.11 queue...

Regarding the mac80211 (and related) bits, Johannes says:

"I have a fix from Chris for an infinite loop along with fixes from
myself to prevent it entering the loop to start with (continue using
disabled channels, many thanks to Chris for his debug/test help) and a
workaround for broken APs that advertise a bad HT primary channel in
their beacons. Additionally, a fix for another attrbuf race in mac80211
and a fix to clean up properly while P2P GO interfaces go down."

Along with that...

Solomon Peachy corrects a range check in cw1200 that would lead to
a BUG_ON when starting AP mode.

Stanislaw Gruszka provides an iwl4965 patch to power-up the device
earlier (avoiding microcode errors), and another iwl4965 fix that
resets the firmware after turning rfkill off (resolving a bug in the
Red Hat Bugzilla).

Please let me know if there are problems!
====================

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 [Thu, 8 Aug 2013 17:12:42 +0000 (13:12 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless into for-davem

11 years agoipv6: don't stop backtracking in fib6_lookup_1 if subtree does not match
Hannes Frederic Sowa [Wed, 7 Aug 2013 00:34:31 +0000 (02:34 +0200)]
ipv6: don't stop backtracking in fib6_lookup_1 if subtree does not match

In case a subtree did not match we currently stop backtracking and return
NULL (root table from fib_lookup). This could yield in invalid routing
table lookups when using subtrees.

Instead continue to backtrack until a valid subtree or node is found
and return this match.

Also remove unneeded NULL check.

Reported-by: Teco Boot <teco@inf-net.nl>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Cc: David Lamparter <equinox@diac24.net>
Cc: <boutier@pps.univ-paris-diderot.fr>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agotcp: cubic: fix bug in bictcp_acked()
Eric Dumazet [Tue, 6 Aug 2013 03:05:12 +0000 (20:05 -0700)]
tcp: cubic: fix bug in bictcp_acked()

While investigating about strange increase of retransmit rates
on hosts ~24 days after boot, Van found hystart was disabled
if ca->epoch_start was 0, as following condition is true
when tcp_time_stamp high order bit is set.

(s32)(tcp_time_stamp - ca->epoch_start) < HZ

Quoting Van :

 At initialization & after every loss ca->epoch_start is set to zero so
 I believe that the above line will turn off hystart as soon as the 2^31
 bit is set in tcp_time_stamp & hystart will stay off for 24 days.
 I think we've observed that cubic's restart is too aggressive without
 hystart so this might account for the higher drop rate we observe.

Diagnosed-by: Van Jacobson <vanj@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobridge: correct the comment for file br_sysfs_br.c
Wang Sheng-Hui [Tue, 6 Aug 2013 00:44:46 +0000 (08:44 +0800)]
bridge: correct the comment for file br_sysfs_br.c

br_sysfs_if.c is for sysfs attributes of bridge ports, while br_sysfs_br.c
is for sysfs attributes of bridge itself. Correct the comment here.

Signed-off-by: Wang Sheng-Hui <shhuiw@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agotcp: cubic: fix overflow error in bictcp_update()
Eric Dumazet [Tue, 6 Aug 2013 00:10:15 +0000 (17:10 -0700)]
tcp: cubic: fix overflow error in bictcp_update()

commit 17a6e9f1aa9 ("tcp_cubic: fix clock dependency") added an
overflow error in bictcp_update() in following code :

/* change the unit from HZ to bictcp_HZ */
t = ((tcp_time_stamp + msecs_to_jiffies(ca->delay_min>>3) -
      ca->epoch_start) << BICTCP_HZ) / HZ;

Because msecs_to_jiffies() being unsigned long, compiler does
implicit type promotion.

We really want to constrain (tcp_time_stamp - ca->epoch_start)
to a signed 32bit value, or else 't' has unexpected high values.

This bugs triggers an increase of retransmit rates ~24 days after
boot [1], as the high order bit of tcp_time_stamp flips.

[1] for hosts with HZ=1000

Big thanks to Van Jacobson for spotting this problem.

Diagnosed-by: Van Jacobson <vanj@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobridge: don't try to update timers in case of broken MLD queries
Linus Lüssing [Mon, 5 Aug 2013 22:32:05 +0000 (00:32 +0200)]
bridge: don't try to update timers in case of broken MLD queries

Currently we are reading an uninitialized value for the max_delay
variable when snooping an MLD query message of invalid length and would
update our timers with that.

Fixing this by simply ignoring such broken MLD queries (just like we do
for IGMP already).

This is a regression introduced by:
"bridge: disable snooping if there is no querier" (b00589af3b04)

Reported-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Linus Lüssing <linus.luessing@web.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agofib_trie: remove potential out of bound access
Eric Dumazet [Mon, 5 Aug 2013 18:18:49 +0000 (11:18 -0700)]
fib_trie: remove potential out of bound access

AddressSanitizer [1] dynamic checker pointed a potential
out of bound access in leaf_walk_rcu()

We could allocate one more slot in tnode_new() to leave the prefetch()
in-place but it looks not worth the pain.

Bug added in commit 82cfbb008572b ("[IPV4] fib_trie: iterator recode")

[1] :
https://code.google.com/p/address-sanitizer/wiki/AddressSanitizerForKernel

Reported-by: Andrey Konovalov <andreyknvl@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobonding: modify only neigh_parms owned by us
Veaceslav Falico [Fri, 2 Aug 2013 17:07:39 +0000 (19:07 +0200)]
bonding: modify only neigh_parms owned by us

Otherwise, on neighbour creation, bond_neigh_init() will be called with a
foreign netdev.

Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoneighbour: populate neigh_parms on alloc before calling ndo_neigh_setup
Veaceslav Falico [Fri, 2 Aug 2013 17:07:38 +0000 (19:07 +0200)]
neighbour: populate neigh_parms on alloc before calling ndo_neigh_setup

dev->ndo_neigh_setup() might need some of the values of neigh_parms, so
populate them before calling it.

Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: esp{4,6}: fix potential MTU calculation overflows
Daniel Borkmann [Mon, 5 Aug 2013 10:49:35 +0000 (12:49 +0200)]
net: esp{4,6}: fix potential MTU calculation overflows

Commit 91657eafb ("xfrm: take net hdr len into account for esp payload
size calculation") introduced a possible interger overflow in
esp{4,6}_get_mtu() handlers in case of x->props.mode equals
XFRM_MODE_TUNNEL. Thus, the following expression will overflow

  unsigned int net_adj;
  ...
  <case ipv{4,6} XFRM_MODE_TUNNEL>
         net_adj = 0;
  ...
  return ((mtu - x->props.header_len - crypto_aead_authsize(esp->aead) -
           net_adj) & ~(align - 1)) + (net_adj - 2);

where (net_adj - 2) would be evaluated as <foo> + (0 - 2) in an unsigned
context. Fix it by simply removing brackets as those operations here
do not need to have special precedence.

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Benjamin Poirier <bpoirier@suse.de>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Acked-by: Benjamin Poirier <bpoirier@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet_sched: make dev_trans_start return vlan's real dev trans_start
nikolay@redhat.com [Sat, 3 Aug 2013 20:07:47 +0000 (22:07 +0200)]
net_sched: make dev_trans_start return vlan's real dev trans_start

Vlan devices are LLTX and don't update their own trans_start, so if
dev_trans_start has to be called with a vlan device then 0 or a stale
value will be returned. Currently the bonding is the only such user, and
it's needed for proper arp monitoring when the slaves are vlans.
Fix this by extracting the vlan's real device trans_start.

Suggested-by: David Miller <davem@davemloft.net>
Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Acked-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agovlan: make vlan_dev_real_dev work over stacked vlans
nikolay@redhat.com [Sat, 3 Aug 2013 20:07:46 +0000 (22:07 +0200)]
vlan: make vlan_dev_real_dev work over stacked vlans

Sometimes we might have stacked vlans on top of each other, and we're
interested in the first non-vlan real device on the path, so transform
vlan_dev_real_dev to go over the stacked vlans and extract the first
non-vlan device.

Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agocw1200: Fix spurious BUG_ON() trigger when starting AP mode.
Solomon Peachy [Fri, 2 Aug 2013 23:57:40 +0000 (19:57 -0400)]
cw1200: Fix spurious BUG_ON() trigger when starting AP mode.

There's an underlying race condition with the unjoin_work() call that is
sometimes triggered depending on scheduling order and the phase of the
moon.  This doesn't fix the race condition, but it does remove the
ill-advised BUG_ON() call in an easily-recoverable situation.

Signed-off-by: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agonet/vmw_vsock/af_vsock.c: drop unneeded semicolon
Julia Lawall [Mon, 5 Aug 2013 14:47:38 +0000 (16:47 +0200)]
net/vmw_vsock/af_vsock.c: drop unneeded semicolon

Drop the semicolon at the end of the list_for_each_entry loop header.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agomlx5: remove health handler plugin
Eli Cohen [Mon, 5 Aug 2013 13:05:32 +0000 (16:05 +0300)]
mlx5: remove health handler plugin

Remove this code, per Dave Miller's request, since it is not being used
anywhere in the kernel.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agomacvlan: validate flags
Michael S. Tsirkin [Mon, 5 Aug 2013 15:25:54 +0000 (18:25 +0300)]
macvlan: validate flags

commit df8ef8f3aaa6692970a436204c4429210addb23a
    macvlan: add FDB bridge ops and macvlan flags
added a flags field to macvlan, which can be
controlled from userspace.
The idea is to make the interface future-proof
so we can add flags and not new fields.

However, flags value isn't validated, as a result,
userspace can't detect which flags are supported.

Cc: "David S. Miller" <davem@davemloft.net>
Cc: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonetfilter: nfnetlink_{log,queue}: fix information leaks in netlink message
Dan Carpenter [Thu, 1 Aug 2013 09:36:57 +0000 (12:36 +0300)]
netfilter: nfnetlink_{log,queue}: fix information leaks in netlink message

These structs have a "_pad" member.  Also the "phw" structs have an 8
byte "hw_addr[]" array but sometimes only the first 6 bytes are
initialized.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
11 years agoskge: fix build on 32 bit
Stephen Hemminger [Mon, 5 Aug 2013 03:40:34 +0000 (20:40 -0700)]
skge: fix build on 32 bit

The following is needed as well to fix warning/error about shifting a 32 bit
value 32 bits which occurs if building on 32 bit platform caused by conversion
to using dma_addr_t

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobusy_poll: cleanup do-nothing placeholders
Eliezer Tamir [Sun, 4 Aug 2013 09:55:48 +0000 (12:55 +0300)]
busy_poll: cleanup do-nothing placeholders

When renaming ll_poll to busy poll, I introduced a typo
in the name of the do-nothing placeholder for sk_busy_loop
and called it sk_busy_poll.
This broke compile when busy poll was not configured.
Cong Wang submitted a patch to fixed that.
This patch removes the now redundant, misspelled placeholder.

Signed-off-by: Eliezer Tamir <eliezer.tamir@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoskge: add dma_mapping check
stephen hemminger [Mon, 5 Aug 2013 00:22:34 +0000 (17:22 -0700)]
skge: add dma_mapping check

This old driver never checked for DMA mapping errors.
Causing splats with the new DMA mapping checks:
WARNING: at lib/dma-debug.c:937 check_unmap+0x47b/0x930()
skge 0000:01:09.0: DMA-API: device driver failed to check map

Add checks and unwind code.

Reported-by: poma <pomidorabelisima@gmail.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Sat, 3 Aug 2013 22:00:23 +0000 (15:00 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Don't ignore user initiated wireless regulatory settings on cards
    with custom regulatory domains, from Arik Nemtsov.

 2) Fix length check of bluetooth information responses, from Jaganath
    Kanakkassery.

 3) Fix misuse of PTR_ERR in btusb, from Adam Lee.

 4) Handle rfkill properly while iwlwifi devices are offline, from
    Emmanuel Grumbach.

 5) Fix r815x devices DMA'ing to stack buffers, from Hayes Wang.

 6) Kernel info leak in ATM packet scheduler, from Dan Carpenter.

 7) 8139cp doesn't check for DMA mapping errors, from Neil Horman.

 8) Fix bridge multicast code to not snoop when no querier exists,
    otherwise mutlicast traffic is lost.  From Linus Lüssing.

 9) Avoid soft lockups in fib6_run_gc(), from Michal Kubecek.

10) Fix races in automatic address asignment on ipv6, which can result
    in incorrect lifetime assignments.  From Jiri Benc.

11) Cure build bustage when CONFIG_NET_LL_RX_POLL is not set and rename
    it CONFIG_NET_RX_BUSY_POLL to eliminate the last reference to the
    original naming of this feature.  From Cong Wang.

12) Fix crash in TIPC when server socket creation fails, from Ying Xue.

13) macvlan_changelink() silently succeeds when it shouldn't, from
    Michael S Tsirkin.

14) HTB packet scheduler can crash due to sign extension, fix from
    Stephen Hemminger.

15) With the cable unplugged, r8169 prints out a message every 10
    seconds, make it netif_dbg() instead of netif_warn().  From Peter
    Wu.

16) Fix memory leak in rtm_to_ifaddr(), from Daniel Borkmann.

17) sis900 gets spurious TX queue timeouts due to mismanagement of link
    carrier state, from Denis Kirjanov.

18) Validate somaxconn sysctl to make sure it fits inside of a u16.
    From Roman Gushchin.

19) Fix MAC address filtering on qlcnic, from Shahed Shaikh.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (68 commits)
  qlcnic: Fix for flash update failure on 83xx adapter
  qlcnic: Fix link speed and duplex display for 83xx adapter
  qlcnic: Fix link speed display for 82xx adapter
  qlcnic: Fix external loopback test.
  qlcnic: Removed adapter series name from warning messages.
  qlcnic: Free up memory in error path.
  qlcnic: Fix ingress MAC learning
  qlcnic: Fix MAC address filter issue on 82xx adapter
  net: ethernet: davinci_emac: drop IRQF_DISABLED
  netlabel: use domain based selectors when address based selectors are not available
  net: check net.core.somaxconn sysctl values
  sis900: Fix the tx queue timeout issue
  net: rtm_to_ifaddr: free ifa if ifa_cacheinfo processing fails
  r8169: remove "PHY reset until link up" log spam
  net: ethernet: cpsw: drop IRQF_DISABLED
  htb: fix sign extension bug
  macvlan: handle set_promiscuity failures
  macvlan: better mode validation
  tipc: fix oops when creating server socket fails
  net: rename CONFIG_NET_LL_RX_POLL to CONFIG_NET_RX_BUSY_POLL
  ...

11 years agoqlcnic: Fix for flash update failure on 83xx adapter
Himanshu Madhani [Sat, 3 Aug 2013 03:16:01 +0000 (23:16 -0400)]
qlcnic: Fix for flash update failure on 83xx adapter

Flash update routine was improperly checking register read API return value.
Modify register read API and perform proper error check.

Signed-off-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Fix link speed and duplex display for 83xx adapter
Rajesh Borundia [Sat, 3 Aug 2013 03:16:00 +0000 (23:16 -0400)]
qlcnic: Fix link speed and duplex display for 83xx adapter

o Set link speed and duplex to unknown when link is not up.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Fix link speed display for 82xx adapter
Rajesh Borundia [Sat, 3 Aug 2013 03:15:59 +0000 (23:15 -0400)]
qlcnic: Fix link speed display for 82xx adapter

o Do not obtain link speed from register when adapter
  link is down.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Fix external loopback test.
Shahed Shaikh [Sat, 3 Aug 2013 03:15:58 +0000 (23:15 -0400)]
qlcnic: Fix external loopback test.

Driver was not handling external loopback diagnostic
test request.

Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Removed adapter series name from warning messages.
Pratik Pujar [Sat, 3 Aug 2013 03:15:57 +0000 (23:15 -0400)]
qlcnic: Removed adapter series name from warning messages.

Signed-off-by: Pratik Pujar <pratik.pujar@qlogic.com>
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Free up memory in error path.
Himanshu Madhani [Sat, 3 Aug 2013 03:15:56 +0000 (23:15 -0400)]
qlcnic: Free up memory in error path.

Signed-off-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Fix ingress MAC learning
Shahed Shaikh [Sat, 3 Aug 2013 03:15:55 +0000 (23:15 -0400)]
qlcnic: Fix ingress MAC learning

o Delete MAC address from the adapter's filter table
  if the source MAC address of ingress packet matches.

Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Fix MAC address filter issue on 82xx adapter
Shahed Shaikh [Sat, 3 Aug 2013 03:15:54 +0000 (23:15 -0400)]
qlcnic: Fix MAC address filter issue on 82xx adapter

Driver was passing the address of a pointer instead of
the pointer itself.

Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: ethernet: davinci_emac: drop IRQF_DISABLED
Mugunthan V N [Sat, 3 Aug 2013 11:09:45 +0000 (16:39 +0530)]
net: ethernet: davinci_emac: drop IRQF_DISABLED

IRQF_DISABLED is a no-op by now and should be removed.

Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'for-3.11' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Sat, 3 Aug 2013 18:15:03 +0000 (11:15 -0700)]
Merge branch 'for-3.11' of git://linux-nfs.org/~bfields/linux

Pull nfsd bugfixes from Bruce Fields:
 "Most of this is due to a screwup on my part -- some gss-proxy crashes
  got fixed before the merge window but somehow never made it out of a
  temporary git repo on my laptop...."

* 'for-3.11' of git://linux-nfs.org/~bfields/linux:
  svcrpc: set cr_gss_mech from gss-proxy as well as legacy upcall
  svcrpc: fix kfree oops in gss-proxy code
  svcrpc: fix gss-proxy xdr decoding oops
  svcrpc: fix gss_rpc_upcall create error
  NFSD/sunrpc: avoid deadlock on TCP connection due to memory pressure.

11 years agoMerge tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck...
Linus Torvalds [Sat, 3 Aug 2013 18:14:25 +0000 (11:14 -0700)]
Merge tag 'hwmon-for-linus' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fix from Guenter Roeck:
 "Fix chip initialization/configuration in MAX6697 driver"

* tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (max6697) fix MAX6581 ideality

11 years agoMerge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm
Linus Torvalds [Sat, 3 Aug 2013 18:12:09 +0000 (11:12 -0700)]
Merge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm

Pull arm fixes fixes from Russell King:
 "This fixes a couple of problems with commit 48be69a026b2 ("ARM: move
  signal handlers into a vdso-like page"), one of which was originally
  discovered via my testing originally, but the fix for it was never
  actually committed.

  The other shows up on noMMU builds, and such platforms are extremely
  rare and as such are not part of my nightly testing"

* 'fixes' of git://git.linaro.org/people/rmk/linux-arm:
  ARM: fix nommu builds with 48be69a02 (ARM: move signal handlers into a vdso-like page)
  ARM: fix a cockup in 48be69a02 (ARM: move signal handlers into a vdso-like page)

11 years agohwmon: (max6697) fix MAX6581 ideality
Vivien Didelot [Tue, 30 Jul 2013 21:14:34 +0000 (17:14 -0400)]
hwmon: (max6697) fix MAX6581 ideality

Without this patch, the values for ideality (register 0x4b) and ideality
selection mask (register 0x4c) are inverted.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Cc: stable@vger.kernel.org # 3.9+
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
11 years agoMerge branch 'security-fixes' into fixes
Russell King [Sat, 3 Aug 2013 09:49:38 +0000 (10:49 +0100)]
Merge branch 'security-fixes' into fixes

11 years agoARM: fix nommu builds with 48be69a02 (ARM: move signal handlers into a vdso-like...
Russell King [Sat, 3 Aug 2013 09:39:51 +0000 (10:39 +0100)]
ARM: fix nommu builds with 48be69a02 (ARM: move signal handlers into a vdso-like page)

Olof reports that noMMU builds error out with:

arch/arm/kernel/signal.c: In function 'setup_return':
arch/arm/kernel/signal.c:413:25: error: 'mm_context_t' has no member named 'sigpage'

This shows one of the evilnesses of IS_ENABLED().  Get rid of it here
and replace it with #ifdef's - and as no noMMU platform can make use
of sigpage, depend on CONIFG_MMU not CONFIG_ARM_MPU.

Reported-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
11 years agoARM: fix a cockup in 48be69a02 (ARM: move signal handlers into a vdso-like page)
Russell King [Sat, 3 Aug 2013 09:30:05 +0000 (10:30 +0100)]
ARM: fix a cockup in 48be69a02 (ARM: move signal handlers into a vdso-like page)

Unfortunately, I never committed the fix to a nasty oops which can
occur as a result of that commit:

------------[ cut here ]------------
kernel BUG at /home/olof/work/batch/include/linux/mm.h:414!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM
Modules linked in:
CPU: 0 PID: 490 Comm: killall5 Not tainted 3.11.0-rc3-00288-gabe0308 #53
task: e90acac0 ti: e9be8000 task.ti: e9be8000
PC is at special_mapping_fault+0xa4/0xc4
LR is at __do_fault+0x68/0x48c

This doesn't show up unless you do quite a bit of testing; a simple
boot test does not do this, so all my nightly tests were passing fine.

The reason for this is that install_special_mapping() expects the
page array to stick around, and as this was only inserting one page
which was stored on the kernel stack, that's why this was blowing up.

Reported-by: Olof Johansson <olof@lixom.net>
Tested-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
11 years agonetlabel: use domain based selectors when address based selectors are not available
Paul Moore [Fri, 2 Aug 2013 18:45:08 +0000 (14:45 -0400)]
netlabel: use domain based selectors when address based selectors are not available

NetLabel has the ability to selectively assign network security labels
to outbound traffic based on either the LSM's "domain" (different for
each LSM), the network destination, or a combination of both.  Depending
on the type of traffic, local or forwarded, and the type of traffic
selector, domain or address based, different hooks are used to label the
traffic; the goal being minimal overhead.

Unfortunately, there is a bug such that a system using NetLabel domain
based traffic selectors does not correctly label outbound local traffic
that is not assigned to a socket.  The issue is that in these cases
the associated NetLabel hook only looks at the address based selectors
and not the domain based selectors.  This patch corrects this by
checking both the domain and address based selectors so that the correct
labeling is applied, regardless of the configuration type.

In order to acomplish this fix, this patch also simplifies some of the
NetLabel domainhash structures to use a more common outbound traffic
mapping type: struct netlbl_dommap_def.  This simplifies some of the code
in this patch and paves the way for further simplifications in the
future.

Signed-off-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: check net.core.somaxconn sysctl values
Roman Gushchin [Fri, 2 Aug 2013 14:36:40 +0000 (18:36 +0400)]
net: check net.core.somaxconn sysctl values

It's possible to assign an invalid value to the net.core.somaxconn
sysctl variable, because there is no checks at all.

The sk_max_ack_backlog field of the sock structure is defined as
unsigned short. Therefore, the backlog argument in inet_listen()
shouldn't exceed USHRT_MAX. The backlog argument in the listen() syscall
is truncated to the somaxconn value. So, the somaxconn value shouldn't
exceed 65535 (USHRT_MAX).
Also, negative values of somaxconn are meaningless.

before:
$ sysctl -w net.core.somaxconn=256
net.core.somaxconn = 256
$ sysctl -w net.core.somaxconn=65536
net.core.somaxconn = 65536
$ sysctl -w net.core.somaxconn=-100
net.core.somaxconn = -100

after:
$ sysctl -w net.core.somaxconn=256
net.core.somaxconn = 256
$ sysctl -w net.core.somaxconn=65536
error: "Invalid argument" setting key "net.core.somaxconn"
$ sysctl -w net.core.somaxconn=-100
error: "Invalid argument" setting key "net.core.somaxconn"

Based on a prior patch from Changli Gao.

Signed-off-by: Roman Gushchin <klamm@yandex-team.ru>
Reported-by: Changli Gao <xiaosuo@gmail.com>
Suggested-by: Eric Dumazet <edumazet@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agosis900: Fix the tx queue timeout issue
Denis Kirjanov [Fri, 2 Aug 2013 09:50:54 +0000 (13:50 +0400)]
sis900: Fix the tx queue timeout issue

[  198.720048] ------------[ cut here ]------------
[  198.720108] WARNING: CPU: 0 PID: 0 at net/sched/sch_generic.c:255 dev_watchdog+0x229/0x240()
[  198.720118] NETDEV WATCHDOG: eth0 (sis900): transmit queue 0 timed out
[  198.720125] Modules linked in: bridge stp llc dmfe sundance 3c59x sis900 mii
[  198.720159] CPU: 0 PID: 0 Comm: swapper Not tainted 3.11.0-rc3+ #12
[  198.720167] Hardware name: System Manufacturer System Name/TUSI-M, BIOS ASUS TUSI-M ACPI BIOS
Revision 1013 Beta 001 12/14/2001
[  198.720175]  000000ff c13fa6b9 c169ddcc c12208d6 c169ddf8 c1031e4d c1664a84 c169de24
[  198.720197]  00000000 c165f5ea 000000ff c13fa6b9 00000001 000000ff c1664a84 c169de10
[  198.720217]  c1031f13 00000009 c169de08 c1664a84 c169de24 c169de50 c13fa6b9 c165f5ea
[  198.720240] Call Trace:
[  198.720257]  [<c13fa6b9>] ? dev_watchdog+0x229/0x240
[  198.720274]  [<c12208d6>] dump_stack+0x16/0x20
[  198.720306]  [<c1031e4d>] warn_slowpath_common+0x7d/0xa0
[  198.720318]  [<c13fa6b9>] ? dev_watchdog+0x229/0x240
[  198.720330]  [<c1031f13>] warn_slowpath_fmt+0x33/0x40
[  198.720342]  [<c13fa6b9>] dev_watchdog+0x229/0x240
[  198.720357]  [<c103f158>] call_timer_fn+0x78/0x150
[  198.720369]  [<c103f0e0>] ? internal_add_timer+0x40/0x40
[  198.720381]  [<c13fa490>] ? dev_init_scheduler+0xa0/0xa0
[  198.720392]  [<c103f33f>] run_timer_softirq+0x10f/0x200
[  198.720412]  [<c103954f>] ? __do_softirq+0x6f/0x210
[  198.720424]  [<c13fa490>] ? dev_init_scheduler+0xa0/0xa0
[  198.720435]  [<c1039598>] __do_softirq+0xb8/0x210
[  198.720467]  [<c14b54d2>] ? _raw_spin_unlock+0x22/0x30
[  198.720484]  [<c1003245>] ? handle_irq+0x25/0xd0
[  198.720496]  [<c1039c0c>] irq_exit+0x9c/0xb0
[  198.720508]  [<c14bc9d7>] do_IRQ+0x47/0x94
[  198.720534]  [<c1056078>] ? hrtimer_start+0x28/0x30
[  198.720564]  [<c14bc8b1>] common_interrupt+0x31/0x38
[  198.720589]  [<c1008692>] ? default_idle+0x22/0xa0
[  198.720600]  [<c10083c7>] arch_cpu_idle+0x17/0x30
[  198.720631]  [<c106d23d>] cpu_startup_entry+0xcd/0x180
[  198.720643]  [<c14ae30a>] rest_init+0xaa/0xb0
[  198.720654]  [<c14ae260>] ? reciprocal_value+0x50/0x50
[  198.720668]  [<c17044e0>] ? repair_env_string+0x60/0x60
[  198.720679]  [<c1704bda>] start_kernel+0x29a/0x350
[  198.720690]  [<c17044e0>] ? repair_env_string+0x60/0x60
[  198.720721]  [<c1704269>] i386_start_kernel+0x39/0xa0
[  198.720729] ---[ end trace 81e0a6266f5c73a8 ]---
[  198.720740] eth0: Transmit timeout, status 00000204 00000000

timer routine checks the link status and if it's up calls
netif_carrier_on() allowing upper layer to start the tx queue
even if the auto-negotiation process is not finished.

Also remove ugly auto-negotiation check from the sis900_start_xmit()

CC: Duan Fugang <B38611@freescale.com>
CC: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Denis Kirjanov <kda@linux-powerpc.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Fri, 2 Aug 2013 21:58:30 +0000 (14:58 -0700)]
Merge tag 'rdma-for-linus' of git://git./linux/kernel/git/roland/infiniband

Pull infiniband/rdma fixes from Roland Dreier:
 - Fixes for the newly merged mlx5 hardware driver
 - Stack info leak fixes from Dan Carpenter
 - Fixes for pkey table handling with SR-IOV
 - A few other small things

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IPoIB: Fix pkey change flow for virtualization environments
  IPoIB: Make sure child devices use valid/proper pkeys
  IB/core: Create QP1 using the pkey index which contains the default pkey
  mlx5_core: Variable may be used uninitialized
  mlx5_core: Implement new initialization sequence
  mlx5_core: Fix use after free in mlx5_cmd_comp_handler()
  IB/mlx5: Fix stack info leak in mlx5_ib_alloc_ucontext()
  IB/mlx5: Fix error return code in init_one()
  IB/mlx4: Use default pkey when creating tunnel QPs
  RDMA/cma: Only call cma_save_ib_info() for CM REQs
  RDMA/cma: Fix accessing invalid private data for UD
  RDMA/cma: Fix gcc warning
  Revert "RDMA/nes: Fix compilation error when nes_debug is enabled"
  IB/qib: Add err_decode() call for ring dump
  RDMA/cxgb3: Fix stack info leak in iwch_create_cq()
  RDMA/nes: Fix info leaks in nes_create_qp() and nes_create_cq()
  RDMA/ocrdma: Fix several stack info leaks
  RDMA/cxgb4: Fix stack info leak in c4iw_create_qp()
  RDMA/ocrdma: Remove unused include

11 years agoMerge tag 'gpio-for-v3.11-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
Linus Torvalds [Fri, 2 Aug 2013 21:57:24 +0000 (14:57 -0700)]
Merge tag 'gpio-for-v3.11-3' of git://git./linux/kernel/git/linusw/linux-gpio

Pull GPIO fixes from Linus Walleij:
 "Yet another GPIO pull request, fixing the fix from the last one.  It
  turns out that fixing the boot path for device tree boots on OMAP
  breaks out antique systems (such as OMAP1) and we need to find a
  better way.  So we're reverting that "fix" for the moment and thinking
  about something better.

  Also fixing a build issue on the MSM driver"

* tag 'gpio-for-v3.11-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio:
  gpio_msm: Fix build error due to missing err.h
  Revert "gpio/omap: don't create an IRQ mapping for every GPIO on DT"
  Revert "gpio/omap: auto request GPIO as input if used as IRQ via DT"
  Revert "gpio/omap: fix build error when OF_GPIO is not defined."

11 years agonet: rtm_to_ifaddr: free ifa if ifa_cacheinfo processing fails
Daniel Borkmann [Fri, 2 Aug 2013 09:32:43 +0000 (11:32 +0200)]
net: rtm_to_ifaddr: free ifa if ifa_cacheinfo processing fails

Commit 5c766d642 ("ipv4: introduce address lifetime") leaves the ifa
resource that was allocated via inet_alloc_ifa() unfreed when returning
the function with -EINVAL. Thus, free it first via inet_free_ifa().

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Reviewed-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agor8169: remove "PHY reset until link up" log spam
Lekensteyn [Fri, 2 Aug 2013 08:36:55 +0000 (10:36 +0200)]
r8169: remove "PHY reset until link up" log spam

This message was added in commit a7154cb8 (June 2004, [PATCH] r8169:
link handling and phy reset rework) and is printed every ten seconds
when no cable is connected and runtime power management is disabled.
(Before that commit, "Reset RTL8169s PHY" would be printed instead.)

Signed-off-by: Peter Wu <lekensteyn@gmail.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: ethernet: cpsw: drop IRQF_DISABLED
Felipe Balbi [Fri, 2 Aug 2013 07:44:10 +0000 (10:44 +0300)]
net: ethernet: cpsw: drop IRQF_DISABLED

IRQF_DISABLED is a no-op by now and should be
removed.

Signed-off-by: Felipe Balbi <balbi@ti.com>
Acked-by: Mugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agohtb: fix sign extension bug
stephen hemminger [Fri, 2 Aug 2013 05:32:07 +0000 (22:32 -0700)]
htb: fix sign extension bug

When userspace passes a large priority value
the assignment of the unsigned value hopt->prio
to  signed int cl->prio causes cl->prio to become negative and the
comparison is with TC_HTB_NUMPRIO is always false.

The result is that HTB crashes by referencing outside
the array when processing packets. With this patch the large value
wraps around like other values outside the normal range.

See: https://bugzilla.kernel.org/show_bug.cgi?id=60669

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Linus Torvalds [Fri, 2 Aug 2013 21:39:49 +0000 (14:39 -0700)]
Merge branch 'merge' of git://git./linux/kernel/git/benh/powerpc

Pull powerpc fixes from Ben Herrenschmidt:
 "Here is not quite a handful of powerpc fixes for rc3.

  The windfarm fix is a regression fix (though not a new one), the PMU
  interrupt rename is not a fix per-se but has been submitted a long
  time ago and I kept forgetting to put it in (it puts us back in sync
  with x86), the other perf bit is just about putting an API/ABI bit
  definition in the right place for userspace to consume, and finally,
  we have a fix for the VPHN (Virtual Partition Home Node) feature
  (notification that the hypervisor is moving nodes around) which could
  cause lockups so we may as well fix it now"

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
  powerpc/windfarm: Fix noisy slots-fan on Xserve (rm31)
  powerpc: VPHN topology change updates all siblings
  powerpc/perf: Export PERF_EVENT_CONFIG_EBB_SHIFT to userspace
  powerpc: Rename PMU interrupts from CNT to PMI

11 years agoMerge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm
Linus Torvalds [Fri, 2 Aug 2013 21:37:45 +0000 (14:37 -0700)]
Merge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm

Pull ARM fixes from Russell King:
 "I've thought long and hard about what to say for this pull request,
  and I really can't work out anything sane to say to summarise much of
  these commits.  The problem is, for most of these are, yet again, lots
  of small bits scattered around the place without any real overall
  theme to them"

Most notable is probably the kuser page helper improvements.

* 'fixes' of git://git.linaro.org/people/rmk/linux-arm: (22 commits)
  ARM: Add .text annotations where required after __CPUINIT removal
  ARM: 7803/1: Fix deadlock scenario with smp_send_stop()
  ARM: make vectors page inaccessible from userspace
  ARM: move signal handlers into a vdso-like page
  ARM: allow kuser helpers to be removed from the vector page
  ARM: update FIQ support for relocation of vectors
  ARM: use linker magic for vectors and vector stubs
  ARM: move vector stubs
  ARM: poison memory between kuser helpers
  ARM: poison the vectors page
  ARM: 7801/1: v6: prevent gcc 4.5 from reordering extended CP15 reads above is_smp() test
  ARM: 7800/1: ARMv7-M: Fix name of NVIC handler function
  ARM: Fix sorting of machine- initializers
  ARM: 7791/1: a.out: remove partial a.out support
  ARM: 7790/1: Fix deferred mm switch on VIVT processors
  ARM: 7789/1: Do not run dummy_flush_tlb_a15_erratum() on non-Cortex-A15
  ARM: 7787/1: virt: ensure visibility of __boot_cpu_mode
  ARM: 7788/1: elf: fix lpae hwcap feature reporting in proc/cpuinfo
  ARM: 7786/1: hyp: fix macro parameterisation
  ARM: 7785/1: mm: restrict early_alloc to section-aligned memory
  ...

11 years agoMerge branch 'parisc-3.11-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/delle...
Linus Torvalds [Fri, 2 Aug 2013 21:36:32 +0000 (14:36 -0700)]
Merge branch 'parisc-3.11-rc4' of git://git./linux/kernel/git/deller/parisc-linux

Pull parisc updates from Helge Deller:
 "The majority of lines changed are due the addition of a defconfig for
  the C8000 machine.  Even the fix in parisc/kernel/cache.c file is
  actually ony a 10-line fix, but the change became bigger (and much
  nicer) to avoid errors of the checkpatch script.

  Here is the short-changelog:

  This round of parisc updates includes mostly fixes for the C8000
  workstation.  We have a new defconfig file for this machine, as well
  as fixes for it's serial port, the AGP driver and the cache routines
  to cope with the vmas of the FireGL card in a C8000.  The sys32.h
  header file was not used and as such it's now gone"

* 'parisc-3.11-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Fix interrupt routing for C8000 serial ports
  parisc: Remove arch/parisc/kernel/sys32.h header
  parisc: add defconfig for c8000 machine
  parisc: agp/parisc-agp: allow binding of user memory to the AGP GART
  parisc: Fix cache routines to ignore vma's with an invalid pfn

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Fri, 2 Aug 2013 21:22:15 +0000 (14:22 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

Pull HID fixes from Jiri Kosina:
 - fix hid-sony PS3 sixaxxis breakage from Benjamin Tissories
 - fix hidraw race condition from Yonghua Zheng
 - fix/bandaid for rare device enumeration problems of Logitech Unifying
   receivers from Nestor Lopez Casado

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: hidraw: fix improper mutex release
  HID: sony: fix HID mapping for PS3 sixaxis controller
  HID: hid-logitech-dj: querying_devices was never set
  HID: Revert "Revert "HID: Fix logitech-dj: missing Unifying device issue""

11 years agoMerge tag 'please-pull-fix-mce-regression' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Fri, 2 Aug 2013 21:21:44 +0000 (14:21 -0700)]
Merge tag 'please-pull-fix-mce-regression' of git://git./linux/kernel/git/ras/ras

Pull MCE fix from Tony Luck:
 "Fix a regression in mce-severity.c"

* tag 'please-pull-fix-mce-regression' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
  x86/mce: Fix mce regression from recent cleanup

11 years agoMerge tag 'pci-v3.11-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaa...
Linus Torvalds [Fri, 2 Aug 2013 20:12:52 +0000 (13:12 -0700)]
Merge tag 'pci-v3.11-fixes-1' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fixes from Bjorn Helgaas:
 "Yinghai fixed a couple regressions: one resource assignment problem
  introduced in v3.10 that showed up with SR-IOV on powerpc, and another
  SR-IOV hot-remove issue related to refcounting changes we merged for
  v3.11.

  Yinghai is still working on another SR-IOV-related fix or two, which
  will be simpler if pciehp is non-modular, so I included the Kconfig
  changes now to get them in earlier.

  Finally, a minor fix for the ARM Marvell EBU host bridge driver that
  was merged for v3.11

  Hotplug:
      PCI: pciehp: Fix null pointer deref when hot-removing SR-IOV device
      PCI: hotplug: Convert to be builtin only, not modular
      PCI: pciehp: Convert pciehp to be builtin only, not modular

  Resource allocation:
      PCI: Retry allocation of only the resource type that failed

  ARM:
      PCI: mvebu: Disable prefetchable memory support in PCI-to-PCI bridge"

* tag 'pci-v3.11-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI: mvebu: Disable prefetchable memory support in PCI-to-PCI bridge
  PCI: Retry allocation of only the resource type that failed
  PCI: pciehp: Convert pciehp to be builtin only, not modular
  PCI: hotplug: Convert to be builtin only, not modular
  PCI: pciehp: Fix null pointer deref when hot-removing SR-IOV device

11 years agoMerge tag 'pm+acpi-3.11-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Fri, 2 Aug 2013 19:21:32 +0000 (12:21 -0700)]
Merge tag 'pm+acpi-3.11-rc4' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI and power management fixes from Rafael Wysocki:

 - Revert two cpuidle commits added during the 3.8 development cycle
   that turn out to have introduced a significant performance regression
   as requested by Jeremy Eder.

 - The recent patches that made the freezer less heavy-weight introduced
   a regression causing user-space-driven hibernation using the ioctl()
   interface to block indefinitely when the hibernate process executes
   try_to_freeze().  Fix from Colin Cross addresses this by adding a
   process flag to mark the hibernate/suspend process to inform the
   freezer that that process should be ignored.

 - One of the recent cpufreq reverts uncovered a problem in the core
   causing the cpufreq driver module refcount to become negative after a
   system suspend-resume cycle.  Fix from Rafael J Wysocki.

 - The evaluation of the ACPI battery _BIX method has never worked
   correctly, because the commit that added support for it forgot to
   take the "Revision" field in the return package into account.  As a
   result, the reading of battery info doesn't work at all on some
   systems, which is addressed by a fix from Lan Tianyu.

* tag 'pm+acpi-3.11-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  freezer: set PF_SUSPEND_TASK flag on tasks that call freeze_processes
  ACPI / battery: Fix parsing _BIX return value
  cpufreq: Fix cpufreq driver module refcount balance after suspend/resume
  Revert "cpuidle: Quickly notice prediction failure for repeat mode"
  Revert "cpuidle: Quickly notice prediction failure in general case"

11 years agoMerge branch 'for-john' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211
John W. Linville [Fri, 2 Aug 2013 18:54:19 +0000 (14:54 -0400)]
Merge branch 'for-john' of git://git./linux/kernel/git/jberg/mac80211

11 years agoiwl4965: reset firmware after rfkill off
Stanislaw Gruszka [Thu, 1 Aug 2013 10:07:55 +0000 (12:07 +0200)]
iwl4965: reset firmware after rfkill off

Using rfkill switch can make firmware unstable, what cause various
Microcode errors and kernel warnings. Reseting firmware just after
rfkill off (radio on) helped with that.

Resolve:
https://bugzilla.redhat.com/show_bug.cgi?id=977053

Reported-and-tested-by: Justin Pearce <whitefox@guardianfox.net>
Cc: stable@vger.kernel.org
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agoiwl4965: set power mode early
Stanislaw Gruszka [Thu, 1 Aug 2013 10:07:13 +0000 (12:07 +0200)]
iwl4965: set power mode early

If device was put into a sleep and system was restarted or module
reloaded, we have to wake device up before sending other commands.
Otherwise it will fail to start with Microcode error.

Cc: stable@vger.kernel.org
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
11 years agomacvlan: handle set_promiscuity failures
Michael S. Tsirkin [Thu, 1 Aug 2013 10:50:10 +0000 (13:50 +0300)]
macvlan: handle set_promiscuity failures

It's quite unlikely that dev_set_promiscuity will fail,
but worth checking just in case.

Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agomacvlan: better mode validation
Michael S. Tsirkin [Thu, 1 Aug 2013 10:43:19 +0000 (13:43 +0300)]
macvlan: better mode validation

macvlan passthrough mode is special: it's not possible to switch to or
from it through a netlink command.

But if you try, the command will succeed, which is
confusing.

Validate input and return error to user.

Cc: Sridhar Samudrala <sri@us.ibm.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agotipc: fix oops when creating server socket fails
Ying Xue [Thu, 1 Aug 2013 12:29:18 +0000 (08:29 -0400)]
tipc: fix oops when creating server socket fails

When creation of TIPC internal server socket fails,
we get an oops with the following dump:

BUG: unable to handle kernel NULL pointer dereference at 0000000000000020
IP: [<ffffffffa0011f49>] tipc_close_conn+0x59/0xb0 [tipc]
PGD 13719067 PUD 12008067 PMD 0
Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
Modules linked in: tipc(+)
CPU: 4 PID: 4340 Comm: insmod Not tainted 3.10.0+ #1
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
task: ffff880014360000 ti: ffff88001374c000 task.ti: ffff88001374c000
RIP: 0010:[<ffffffffa0011f49>]  [<ffffffffa0011f49>] tipc_close_conn+0x59/0xb0 [tipc]
RSP: 0018:ffff88001374dc98  EFLAGS: 00010292
RAX: 0000000000000000 RBX: ffff880012ac09d8 RCX: 0000000000000000
RDX: 0000000000000046 RSI: 0000000000000001 RDI: ffff880014360000
RBP: ffff88001374dcb8 R08: 0000000000000001 R09: 0000000000000001
R10: 0000000000000000 R11: 0000000000000000 R12: ffffffffa0016fa0
R13: ffffffffa0017010 R14: ffffffffa0017010 R15: ffff880012ac09d8
FS:  0000000000000000(0000) GS:ffff880016600000(0063) knlGS:00000000f76668d0
CS:  0010 DS: 002b ES: 002b CR0: 000000008005003b
CR2: 0000000000000020 CR3: 0000000012227000 CR4: 00000000000006e0
Stack:
ffff88001374dcb8 ffffffffa0016fa0 0000000000000000 0000000000000001
ffff88001374dcf8 ffffffffa0012922 ffff88001374dce8 00000000ffffffea
ffffffffa0017100 0000000000000000 ffff8800134241a8 ffffffffa0017150
Call Trace:
[<ffffffffa0012922>] tipc_server_stop+0xa2/0x1b0 [tipc]
[<ffffffffa0009995>] tipc_subscr_stop+0x15/0x20 [tipc]
[<ffffffffa00130f5>] tipc_core_stop+0x1d/0x33 [tipc]
[<ffffffffa001f0d4>] tipc_init+0xd4/0xf8 [tipc]
[<ffffffffa001f000>] ? 0xffffffffa001efff
[<ffffffff8100023f>] do_one_initcall+0x3f/0x150
[<ffffffff81082f4d>] ? __blocking_notifier_call_chain+0x7d/0xd0
[<ffffffff810cc58a>] load_module+0x11aa/0x19c0
[<ffffffff810c8d60>] ? show_initstate+0x50/0x50
[<ffffffff8190311c>] ? retint_restore_args+0xe/0xe
[<ffffffff810cce79>] SyS_init_module+0xd9/0x110
[<ffffffff8190dc65>] sysenter_dispatch+0x7/0x1f
Code: 6c 24 70 4c 89 ef e8 b7 04 8f e1 8b 73 04 4c 89 e7 e8 7c 9e 32 e1 41 83 ac 24
b8 00 00 00 01 4c 89 ef e8 eb 0a 8f e1 48 8b 43 08 <4c> 8b 68 20 4d 8d a5 48 03 00
00 4c 89 e7 e8 04 05 8f e1 4c 89
RIP  [<ffffffffa0011f49>] tipc_close_conn+0x59/0xb0 [tipc]
RSP <ffff88001374dc98>
CR2: 0000000000000020
---[ end trace b02321f40e4269a3 ]---

We have the following call chain:

tipc_core_start()
    ret = tipc_subscr_start()
        ret = tipc_server_start(){
                  server->enabled = 1;
                  ret = tipc_open_listening_sock()
              }

I.e., the server->enabled flag is unconditionally set to 1, whatever
the return value of tipc_open_listening_sock().

This causes a crash when tipc_core_start() tries to clean up
resources after a failed initialization:

    if (ret == failed)
        tipc_subscr_stop()
            tipc_server_stop(){
                if (server->enabled)
                    tipc_close_conn(){
                        NULL reference of con->sock-sk
                        OOPS!
                }
            }

To avoid this, tipc_server_start() should only set server->enabled
to 1 in case of a succesful socket creation. In case of failure, it
should release all allocated resources before returning.

Problem introduced in commit c5fa7b3cf3cb22e4ac60485fc2dc187fe012910f
("tipc: introduce new TIPC server infrastructure") in v3.11-rc1.
Note that it won't be seen often; it takes a module load under memory
constrained conditions in order to trigger the failure condition.

Signed-off-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: rename CONFIG_NET_LL_RX_POLL to CONFIG_NET_RX_BUSY_POLL
Cong Wang [Thu, 1 Aug 2013 03:10:25 +0000 (11:10 +0800)]
net: rename CONFIG_NET_LL_RX_POLL to CONFIG_NET_RX_BUSY_POLL

Eliezer renames several *ll_poll to *busy_poll, but forgets
CONFIG_NET_LL_RX_POLL, so in case of confusion, rename it too.

Cc: Eliezer Tamir <eliezer.tamir@linux.intel.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: fix a compile error when CONFIG_NET_LL_RX_POLL is not set
Cong Wang [Thu, 1 Aug 2013 03:10:24 +0000 (11:10 +0800)]
net: fix a compile error when CONFIG_NET_LL_RX_POLL is not set

When CONFIG_NET_LL_RX_POLL is not set, I got:

net/socket.c: In function ‘sock_poll’:
net/socket.c:1165:4: error: implicit declaration of function ‘sk_busy_loop’ [-Werror=implicit-function-declaration]

Fix this by adding a nop when !CONFIG_NET_LL_RX_POLL.

Cc: Eliezer Tamir <eliezer.tamir@linux.intel.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet/mlx4_core: VFs must ignore the enable_64b_cqe_eqe module param
Jack Morgenstein [Thu, 1 Aug 2013 16:55:01 +0000 (19:55 +0300)]
net/mlx4_core: VFs must ignore the enable_64b_cqe_eqe module param

Slaves get the 64B CQE/EQE state from QUERY_HCA, not from the module parameter.

If the parameter is set to zero, the slave outputs an incorrect/irrelevant
warning message that 64B CQEs/EQEs are supported but not enabled (even if the
hypervisor has enabled 64B CQEs/EQEs).

Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet/mlx4_core: Don't give VFs MAC addresses which are derived from the PF MAC
Or Gerlitz [Thu, 1 Aug 2013 16:55:00 +0000 (19:55 +0300)]
net/mlx4_core: Don't give VFs MAC addresses which are derived from the PF MAC

If the user has not assigned a MAC address to a VM, then don't give it MAC which
is based on the PF one. The current derivation scheme is wrong and leads to VM
MAC collisions when the number of cards/hypervisors becomes big enough.

Instead, just give it zeros and let them figure out what to do with that.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv6: prevent race between address creation and removal
Jiri Benc [Thu, 1 Aug 2013 08:41:28 +0000 (10:41 +0200)]
ipv6: prevent race between address creation and removal

There's a race in IPv6 automatic addess assignment. The address is created
with zero lifetime when it's added to various address lists. Before it gets
assigned the correct lifetime, there's a window where a new address may be
configured. This causes the semi-initiated address to be deleted in
addrconf_verify.

This was discovered as a reference leak caused by concurrent run of
__ipv6_ifa_notify for both RTM_NEWADDR and RTM_DELADDR with the same
address.

Fix this by setting the lifetime before the address is added to
inet6_addr_lst.

A few notes:

1. In addrconf_prefix_rcv, by setting update_lft to zero, the
   if (update_lft) { ... } condition is no longer executed for newly
   created addresses. This is okay, as the ifp fields are set in
   ipv6_add_addr now and ipv6_ifa_notify is called (and has been called)
   through addrconf_dad_start.

2. The removal of the whole block under ifp->lock in inet6_addr_add is okay,
   too, as tstamp is initialized to jiffies in ipv6_add_addr.

Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv6: move peer_addr init into ipv6_add_addr()
Jiri Pirko [Thu, 1 Aug 2013 08:41:27 +0000 (10:41 +0200)]
ipv6: move peer_addr init into ipv6_add_addr()

Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv6: update ip6_rt_last_gc every time GC is run
Michal Kubeček [Thu, 1 Aug 2013 08:04:24 +0000 (10:04 +0200)]
ipv6: update ip6_rt_last_gc every time GC is run

As pointed out by Eric Dumazet, net->ipv6.ip6_rt_last_gc should
hold the last time garbage collector was run so that we should
update it whenever fib6_run_gc() calls fib6_clean_all(), not only
if we got there from ip6_dst_gc().

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv6: prevent fib6_run_gc() contention
Michal Kubeček [Thu, 1 Aug 2013 08:04:14 +0000 (10:04 +0200)]
ipv6: prevent fib6_run_gc() contention

On a high-traffic router with many processors and many IPv6 dst
entries, soft lockup in fib6_run_gc() can occur when number of
entries reaches gc_thresh.

This happens because fib6_run_gc() uses fib6_gc_lock to allow
only one thread to run the garbage collector but ip6_dst_gc()
doesn't update net->ipv6.ip6_rt_last_gc until fib6_run_gc()
returns. On a system with many entries, this can take some time
so that in the meantime, other threads pass the tests in
ip6_dst_gc() (ip6_rt_last_gc is still not updated) and wait for
the lock. They then have to run the garbage collector one after
another which blocks them for quite long.

Resolve this by replacing special value ~0UL of expire parameter
to fib6_run_gc() by explicit "force" parameter to choose between
spin_lock_bh() and spin_trylock_bh() and call fib6_run_gc() with
force=false if gc_thresh is reached but not max_size.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoPCI: mvebu: Disable prefetchable memory support in PCI-to-PCI bridge
Thomas Petazzoni [Thu, 1 Aug 2013 13:44:19 +0000 (15:44 +0200)]
PCI: mvebu: Disable prefetchable memory support in PCI-to-PCI bridge

The Marvell PCIe driver uses an emulated PCI-to-PCI bridge to be able
to dynamically set up MBus address decoding windows for PCI I/O and
memory regions depending on the PCI devices enumerated by Linux.

However, this emulated PCI-to-PCI bridge logic makes the Linux PCI
core believe that prefetchable memory regions are supported (because
the registers are read/write), while in fact no adress decoding window
is ever created for such regions. Since the Marvell MBus address
decoding windows do not distinguish memory regions and prefetchable
memory regions, this patch takes a simple approach: change the
PCI-to-PCI bridge emulation to let the Linux PCI core know that we
don't support prefetchable memory regions.

To achieve this, we simply make the prefetchable memory base a
read-only register that always returns 0. Reading/writing all the
other prefetchable memory related registers has no effect.

This problem was originally reported by Finn Hoffmann
<finn@uni-bremen.de>, who couldn't get a RTL8111/8168B PCI NIC working
on the NSA310 Kirkwood platform after updating to 3.11-rc. The problem
was that the PCI-to-PCI bridge emulation was making the Linux PCI core
believe that we support prefetchable memory, so the Linux PCI core was
only filling the prefetchable memory base and limit registers, which
does not lead to a MBus window being created. The below patch has been
confirmed by Finn Hoffmann to fix his problem on Kirkwood, and has
otherwise been successfully tested on the Armada XP GP platform with a
e1000e PCIe NIC and a Marvell SATA PCIe card.

Reported-by: Finn Hoffmann <finn@uni-bremen.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>