]> git.proxmox.com Git - mirror_ubuntu-eoan-kernel.git/log
mirror_ubuntu-eoan-kernel.git
12 years agomwifiex: clear previous security setting during association
Amitkumar Karwar [Fri, 10 Feb 2012 02:32:22 +0000 (18:32 -0800)]
mwifiex: clear previous security setting during association

Driver maintains different flags for WEP, WPA, WPA2 security modes.
Appropriate flag is set using security information provided in
connect request. mwifiex_is_network_compatible() routine uses them
to check if driver's setting is compatible with AP. Association is
aborted if the routine fails.

For some corner cases, it is observed that association is failed
even for valid security information based on association history.
This patch fixes the problem by clearing previous security setting
during each association.

We should set WEP key provided in connect request as default tx key.
This missing change is also added here.

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>
12 years agomac80211: do not call rate control .tx_status before .rate_init
Felix Fietkau [Wed, 8 Feb 2012 18:17:11 +0000 (19:17 +0100)]
mac80211: do not call rate control .tx_status before .rate_init

Most rate control implementations assume .get_rate and .tx_status are only
called once the per-station data has been fully initialized.
minstrel_ht crashes if this assumption is violated.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Tested-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agomac80211: call rate control only after init
Johannes Berg [Fri, 20 Jan 2012 12:55:23 +0000 (13:55 +0100)]
mac80211: call rate control only after init

There are situations where we don't have the
necessary rate control information yet for
station entries, e.g. when associating. This
currently doesn't really happen due to the
dummy station handling; explicitly disabling
rate control when it's not initialised will
allow us to remove dummy stations.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoBluetooth: Fix possible use after free in delete path
Ulisses Furquim [Mon, 30 Jan 2012 20:26:29 +0000 (18:26 -0200)]
Bluetooth: Fix possible use after free in delete path

We need to use the _sync() version for cancelling the info and security
timer in the L2CAP connection delete path. Otherwise the delayed work
handler might run after the connection object is freed.

Signed-off-by: Ulisses Furquim <ulisses@profusion.mobi>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: Remove usage of __cancel_delayed_work()
Ulisses Furquim [Mon, 30 Jan 2012 20:26:28 +0000 (18:26 -0200)]
Bluetooth: Remove usage of __cancel_delayed_work()

__cancel_delayed_work() is being used in some paths where we cannot
sleep waiting for the delayed work to finish. However, that function
might return while the timer is running and the work will be queued
again. Replace the calls with safer cancel_delayed_work() version
which spins until the timer handler finishes on other CPUs and
cancels the delayed work.

Signed-off-by: Ulisses Furquim <ulisses@profusion.mobi>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: btusb: Add vendor specific ID (0a5c 21f3) for BCM20702A0
Manoj Iyer [Thu, 2 Feb 2012 15:32:36 +0000 (09:32 -0600)]
Bluetooth: btusb: Add vendor specific ID (0a5c 21f3) for BCM20702A0

T: Bus=01 Lev=02 Prnt=02 Port=03 Cnt=03 Dev#= 5 Spd=12 MxCh= 0
D: Ver= 2.00 Cls=ff(vend.) Sub=01 Prot=01 MxPS=64 #Cfgs= 1
P: Vendor=0a5c ProdID=21f3 Rev=01.12
S: Manufacturer=Broadcom Corp
S: Product=BCM20702A0
S: SerialNumber=74DE2B344A7B
C: #Ifs= 4 Cfg#= 1 Atr=e0 MxPwr=0mA
I: If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=01 Prot=01 Driver=(none)
I: If#= 1 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=01 Prot=01 Driver=(none)
I: If#= 2 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
I: If#= 3 Alt= 0 #EPs= 0 Cls=fe(app. ) Sub=01 Prot=01 Driver=(none)

Signed-off-by: Manoj Iyer <manoj.iyer@canonical.com>
Tested-by: Dennis Chua <dennis.chua@canonical.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: Add missing QUIRK_NO_RESET test to hci_dev_do_close
Johan Hedberg [Fri, 3 Feb 2012 19:29:40 +0000 (21:29 +0200)]
Bluetooth: Add missing QUIRK_NO_RESET test to hci_dev_do_close

We should only perform a reset in hci_dev_do_close if the
HCI_QUIRK_NO_RESET flag is set (since in such a case a reset will not be
performed when initializing the device).

Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
12 years agoBluetooth: Fix RFCOMM session reference counting issue
Octavian Purdila [Fri, 27 Jan 2012 17:32:39 +0000 (19:32 +0200)]
Bluetooth: Fix RFCOMM session reference counting issue

There is an imbalance in the rfcomm_session_hold / rfcomm_session_put
operations which causes the following crash:

[  685.010159] BUG: unable to handle kernel paging request at 6b6b6b6b
[  685.010169] IP: [<c149d76d>] rfcomm_process_dlcs+0x1b/0x15e
[  685.010181] *pdpt = 000000002d665001 *pde = 0000000000000000
[  685.010191] Oops: 0000 [#1] PREEMPT SMP
[  685.010247]
[  685.010255] Pid: 947, comm: krfcommd Tainted: G         C  3.0.16-mid8-dirty #44
[  685.010266] EIP: 0060:[<c149d76d>] EFLAGS: 00010246 CPU: 1
[  685.010274] EIP is at rfcomm_process_dlcs+0x1b/0x15e
[  685.010281] EAX: e79f551c EBX: 6b6b6b6b ECX: 00000007 EDX: e79f40b4
[  685.010288] ESI: e79f4060 EDI: ed4e1f70 EBP: ed4e1f68 ESP: ed4e1f50
[  685.010295]  DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
[  685.010303] Process krfcommd (pid: 947, ti=ed4e0000 task=ed43e5e0 task.ti=ed4e0000)
[  685.010308] Stack:
[  685.010312]  ed4e1f68 c149eb53 e5925150 e79f4060 ed500000 ed4e1f70 ed4e1f80 c149ec10
[  685.010331]  00000000 ed43e5e0 00000000 ed4e1f90 ed4e1f9c c149ec87 0000bf54 00000000
[  685.010348]  00000000 ee03bf54 c149ec37 ed4e1fe4 c104fe01 00000000 00000000 00000000
[  685.010367] Call Trace:
[  685.010376]  [<c149eb53>] ? rfcomm_process_rx+0x6e/0x74
[  685.010387]  [<c149ec10>] rfcomm_process_sessions+0xb7/0xde
[  685.010398]  [<c149ec87>] rfcomm_run+0x50/0x6d
[  685.010409]  [<c149ec37>] ? rfcomm_process_sessions+0xde/0xde
[  685.010419]  [<c104fe01>] kthread+0x63/0x68
[  685.010431]  [<c104fd9e>] ? __init_kthread_worker+0x42/0x42
[  685.010442]  [<c14dae82>] kernel_thread_helper+0x6/0xd

This issue has been brought up earlier here:

https://lkml.org/lkml/2011/5/21/127

The issue appears to be the rfcomm_session_put in rfcomm_recv_ua. This
operation doesn't seem be to required as for the non-initiator case we
have the rfcomm_process_rx doing an explicit put and in the initiator
case the last dlc_unlink will drive the reference counter to 0.

There have been several attempts to fix these issue:

6c2718d Bluetooth: Do not call rfcomm_session_put() for RFCOMM UA on closed socket
683d949 Bluetooth: Never deallocate a session when some DLC points to it

but AFAICS they do not fix the issue just make it harder to reproduce.

Signed-off-by: Octavian Purdila <octavian.purdila@intel.com>
Signed-off-by: Gopala Krishna Murala <gopala.krishna.murala@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: Fix potential deadlock
Andre Guedes [Fri, 27 Jan 2012 22:42:02 +0000 (19:42 -0300)]
Bluetooth: Fix potential deadlock

We don't need to use the _sync variant in hci_conn_hold and
hci_conn_put to cancel conn->disc_work delayed work. This way
we avoid potential deadlocks like this one reported by lockdep.

======================================================
[ INFO: possible circular locking dependency detected ]
3.2.0+ #1 Not tainted
-------------------------------------------------------
kworker/u:1/17 is trying to acquire lock:
 (&hdev->lock){+.+.+.}, at: [<ffffffffa0041155>] hci_conn_timeout+0x62/0x158 [bluetooth]

but task is already holding lock:
 ((&(&conn->disc_work)->work)){+.+...}, at: [<ffffffff81035751>] process_one_work+0x11a/0x2bf

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #2 ((&(&conn->disc_work)->work)){+.+...}:
       [<ffffffff81057444>] lock_acquire+0x8a/0xa7
       [<ffffffff81034ed1>] wait_on_work+0x3d/0xaa
       [<ffffffff81035b54>] __cancel_work_timer+0xac/0xef
       [<ffffffff81035ba4>] cancel_delayed_work_sync+0xd/0xf
       [<ffffffffa00554b0>] smp_chan_create+0xde/0xe6 [bluetooth]
       [<ffffffffa0056160>] smp_conn_security+0xa3/0x12d [bluetooth]
       [<ffffffffa0053640>] l2cap_connect_cfm+0x237/0x2e8 [bluetooth]
       [<ffffffffa004239c>] hci_proto_connect_cfm+0x2d/0x6f [bluetooth]
       [<ffffffffa0046ea5>] hci_event_packet+0x29d1/0x2d60 [bluetooth]
       [<ffffffffa003dde3>] hci_rx_work+0xd0/0x2e1 [bluetooth]
       [<ffffffff810357af>] process_one_work+0x178/0x2bf
       [<ffffffff81036178>] worker_thread+0xce/0x152
       [<ffffffff81039a03>] kthread+0x95/0x9d
       [<ffffffff812e7754>] kernel_thread_helper+0x4/0x10

-> #1 (slock-AF_BLUETOOTH-BTPROTO_L2CAP){+.+...}:
       [<ffffffff81057444>] lock_acquire+0x8a/0xa7
       [<ffffffff812e553a>] _raw_spin_lock_bh+0x36/0x6a
       [<ffffffff81244d56>] lock_sock_nested+0x24/0x7f
       [<ffffffffa004d96f>] lock_sock+0xb/0xd [bluetooth]
       [<ffffffffa0052906>] l2cap_chan_connect+0xa9/0x26f [bluetooth]
       [<ffffffffa00545f8>] l2cap_sock_connect+0xb3/0xff [bluetooth]
       [<ffffffff81243b48>] sys_connect+0x69/0x8a
       [<ffffffff812e6579>] system_call_fastpath+0x16/0x1b

-> #0 (&hdev->lock){+.+.+.}:
       [<ffffffff81056d06>] __lock_acquire+0xa80/0xd74
       [<ffffffff81057444>] lock_acquire+0x8a/0xa7
       [<ffffffff812e3870>] __mutex_lock_common+0x48/0x38e
       [<ffffffff812e3c75>] mutex_lock_nested+0x2a/0x31
       [<ffffffffa0041155>] hci_conn_timeout+0x62/0x158 [bluetooth]
       [<ffffffff810357af>] process_one_work+0x178/0x2bf
       [<ffffffff81036178>] worker_thread+0xce/0x152
       [<ffffffff81039a03>] kthread+0x95/0x9d
       [<ffffffff812e7754>] kernel_thread_helper+0x4/0x10

other info that might help us debug this:

Chain exists of:
  &hdev->lock --> slock-AF_BLUETOOTH-BTPROTO_L2CAP --> (&(&conn->disc_work)->work)

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock((&(&conn->disc_work)->work));
                               lock(slock-AF_BLUETOOTH-BTPROTO_L2CAP);
                               lock((&(&conn->disc_work)->work));
  lock(&hdev->lock);

 *** DEADLOCK ***

2 locks held by kworker/u:1/17:
 #0:  (hdev->name){.+.+.+}, at: [<ffffffff81035751>] process_one_work+0x11a/0x2bf
 #1:  ((&(&conn->disc_work)->work)){+.+...}, at: [<ffffffff81035751>] process_one_work+0x11a/0x2bf

stack backtrace:
Pid: 17, comm: kworker/u:1 Not tainted 3.2.0+ #1
Call Trace:
 [<ffffffff812e06c6>] print_circular_bug+0x1f8/0x209
 [<ffffffff81056d06>] __lock_acquire+0xa80/0xd74
 [<ffffffff81021ef2>] ? arch_local_irq_restore+0x6/0xd
 [<ffffffff81022bc7>] ? vprintk+0x3f9/0x41e
 [<ffffffff81057444>] lock_acquire+0x8a/0xa7
 [<ffffffffa0041155>] ? hci_conn_timeout+0x62/0x158 [bluetooth]
 [<ffffffff812e3870>] __mutex_lock_common+0x48/0x38e
 [<ffffffffa0041155>] ? hci_conn_timeout+0x62/0x158 [bluetooth]
 [<ffffffff81190fd6>] ? __dynamic_pr_debug+0x6d/0x6f
 [<ffffffffa0041155>] ? hci_conn_timeout+0x62/0x158 [bluetooth]
 [<ffffffff8105320f>] ? trace_hardirqs_off+0xd/0xf
 [<ffffffff812e3c75>] mutex_lock_nested+0x2a/0x31
 [<ffffffffa0041155>] hci_conn_timeout+0x62/0x158 [bluetooth]
 [<ffffffff810357af>] process_one_work+0x178/0x2bf
 [<ffffffff81035751>] ? process_one_work+0x11a/0x2bf
 [<ffffffff81055af3>] ? lock_acquired+0x1d0/0x1df
 [<ffffffffa00410f3>] ? hci_acl_disconn+0x65/0x65 [bluetooth]
 [<ffffffff81036178>] worker_thread+0xce/0x152
 [<ffffffff810407ed>] ? finish_task_switch+0x45/0xc5
 [<ffffffff810360aa>] ? manage_workers.isra.25+0x16a/0x16a
 [<ffffffff81039a03>] kthread+0x95/0x9d
 [<ffffffff812e7754>] kernel_thread_helper+0x4/0x10
 [<ffffffff812e5db4>] ? retint_restore_args+0x13/0x13
 [<ffffffff8103996e>] ? __init_kthread_worker+0x55/0x55
 [<ffffffff812e7750>] ? gs_change+0x13/0x13

Signed-off-by: Andre Guedes <andre.guedes@openbossa.org>
Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@openbossa.org>
Reviewed-by: Ulisses Furquim <ulisses@profusion.mobi>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: silence lockdep warning
Octavian Purdila [Sat, 21 Jan 2012 22:28:34 +0000 (00:28 +0200)]
Bluetooth: silence lockdep warning

Since bluetooth uses multiple protocols types, to avoid lockdep
warnings, we need to use different lockdep classes (one for each
protocol type).

This is already done in bt_sock_create but it misses a couple of cases
when new connections are created. This patch corrects that to fix the
following warning:

<4>[ 1864.732366] =======================================================
<4>[ 1864.733030] [ INFO: possible circular locking dependency detected ]
<4>[ 1864.733544] 3.0.16-mid3-00007-gc9a0f62 #3
<4>[ 1864.733883] -------------------------------------------------------
<4>[ 1864.734408] t.android.btclc/4204 is trying to acquire lock:
<4>[ 1864.734869]  (rfcomm_mutex){+.+.+.}, at: [<c14970ea>] rfcomm_dlc_close+0x15/0x30
<4>[ 1864.735541]
<4>[ 1864.735549] but task is already holding lock:
<4>[ 1864.736045]  (sk_lock-AF_BLUETOOTH){+.+.+.}, at: [<c1498bf7>] lock_sock+0xa/0xc
<4>[ 1864.736732]
<4>[ 1864.736740] which lock already depends on the new lock.
<4>[ 1864.736750]
<4>[ 1864.737428]
<4>[ 1864.737437] the existing dependency chain (in reverse order) is:
<4>[ 1864.738016]
<4>[ 1864.738023] -> #1 (sk_lock-AF_BLUETOOTH){+.+.+.}:
<4>[ 1864.738549]        [<c1062273>] lock_acquire+0x104/0x140
<4>[ 1864.738977]        [<c13d35c1>] lock_sock_nested+0x58/0x68
<4>[ 1864.739411]        [<c1493c33>] l2cap_sock_sendmsg+0x3e/0x76
<4>[ 1864.739858]        [<c13d06c3>] __sock_sendmsg+0x50/0x59
<4>[ 1864.740279]        [<c13d0ea2>] sock_sendmsg+0x94/0xa8
<4>[ 1864.740687]        [<c13d0ede>] kernel_sendmsg+0x28/0x37
<4>[ 1864.741106]        [<c14969ca>] rfcomm_send_frame+0x30/0x38
<4>[ 1864.741542]        [<c1496a2a>] rfcomm_send_ua+0x58/0x5a
<4>[ 1864.741959]        [<c1498447>] rfcomm_run+0x441/0xb52
<4>[ 1864.742365]        [<c104f095>] kthread+0x63/0x68
<4>[ 1864.742742]        [<c14d5182>] kernel_thread_helper+0x6/0xd
<4>[ 1864.743187]
<4>[ 1864.743193] -> #0 (rfcomm_mutex){+.+.+.}:
<4>[ 1864.743667]        [<c1061ada>] __lock_acquire+0x988/0xc00
<4>[ 1864.744100]        [<c1062273>] lock_acquire+0x104/0x140
<4>[ 1864.744519]        [<c14d2c70>] __mutex_lock_common+0x3b/0x33f
<4>[ 1864.744975]        [<c14d303e>] mutex_lock_nested+0x2d/0x36
<4>[ 1864.745412]        [<c14970ea>] rfcomm_dlc_close+0x15/0x30
<4>[ 1864.745842]        [<c14990d9>] __rfcomm_sock_close+0x5f/0x6b
<4>[ 1864.746288]        [<c1499114>] rfcomm_sock_shutdown+0x2f/0x62
<4>[ 1864.746737]        [<c13d275d>] sys_socketcall+0x1db/0x422
<4>[ 1864.747165]        [<c14d42f0>] syscall_call+0x7/0xb

Signed-off-by: Octavian Purdila <octavian.purdila@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: Fix using an absolute timeout on hci_conn_put()
Vinicius Costa Gomes [Wed, 4 Jan 2012 14:57:17 +0000 (11:57 -0300)]
Bluetooth: Fix using an absolute timeout on hci_conn_put()

queue_delayed_work() expects a relative time for when that work
should be scheduled.

Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@openbossa.org>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: l2cap_set_timer needs jiffies as timeout value
Andrzej Kaczmarek [Wed, 4 Jan 2012 11:10:42 +0000 (12:10 +0100)]
Bluetooth: l2cap_set_timer needs jiffies as timeout value

After moving L2CAP timers to workqueues l2cap_set_timer expects timeout
value to be specified in jiffies but constants defined in miliseconds
are used. This makes timeouts unreliable when CONFIG_HZ is not set to
1000.

__set_chan_timer macro still uses jiffies as input to avoid multiple
conversions from/to jiffies for sk_sndtimeo value which is already
specified in jiffies.

Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@tieto.com>
Ackec-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: Fix sk_sndtimeo initialization for L2CAP socket
Andrzej Kaczmarek [Wed, 4 Jan 2012 11:10:41 +0000 (12:10 +0100)]
Bluetooth: Fix sk_sndtimeo initialization for L2CAP socket

sk_sndtime value should be specified in jiffies thus initial value
needs to be converted from miliseconds. Otherwise this timeout is
unreliable when CONFIG_HZ is not set to 1000.

Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@tieto.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: Remove bogus inline declaration from l2cap_chan_connect
Johan Hedberg [Sun, 8 Jan 2012 20:51:16 +0000 (22:51 +0200)]
Bluetooth: Remove bogus inline declaration from l2cap_chan_connect

As reported by Dan Carpenter this function causes a Sparse warning and
shouldn't be declared inline:

include/net/bluetooth/l2cap.h:837:30 error: marked inline, but without a
definition"

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
12 years agoBluetooth: Don't mark non xfer isoc endpoint URBs with URB_ISO_ASAP
Daniel Wagner [Tue, 3 Jan 2012 10:53:53 +0000 (11:53 +0100)]
Bluetooth: Don't mark non xfer isoc endpoint URBs with URB_ISO_ASAP

[ 2096.384084] btusb_send_frame:684: hci0
[ 2096.384087] usb 3-1: BOGUS urb flags, 2 --> 0
[ 2096.384091] Bluetooth: hci0 urb ffff8801b61d3a80 submission failed (22)

According the documentation in usb_submit_urb() URB_ISO_ASAP
flag is only allowed for endpoints of type USB_ENDPOINT_XFER_ISOC.

This reverts commit b8aabfc92249b239c425da7e4ca85b7e4855e984.

Signed-off-by: Daniel Wagner <daniel.wagner@bmw-carit.de>
Acked-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agoBluetooth: Fix l2cap conn failures for ssp devices
Peter Hurley [Fri, 13 Jan 2012 14:11:30 +0000 (15:11 +0100)]
Bluetooth: Fix l2cap conn failures for ssp devices

Commit 330605423c fixed l2cap conn establishment for non-ssp remote
devices by not setting HCI_CONN_ENCRYPT_PEND every time conn security
is tested (which was always returning failure on any subsequent
security checks).

However, this broke l2cap conn establishment for ssp remote devices
when an ACL link was already established at SDP-level security. This
fix ensures that encryption must be pending whenever authentication
is also pending.

Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Tested-by: Daniel Wagner <daniel.wagner@bmw-carit.de>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
12 years agortlwifi: Modify rtl_pci_init to return 0 on success
Simon Graham [Thu, 9 Feb 2012 14:55:13 +0000 (09:55 -0500)]
rtlwifi: Modify rtl_pci_init to return 0 on success

Fixes problem where caller would think routine succeeded when it failed
leading to divide by zero panic.

Signed-off-by: Simon Graham <simon.graham@virtualcomputer.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agomac80211: Fix a rwlock bad magic bug
Mohammed Shafi Shajakhan [Thu, 9 Feb 2012 14:29:43 +0000 (19:59 +0530)]
mac80211: Fix a rwlock bad magic bug

read_lock(&tpt_trig->trig.leddev_list_lock) is accessed via the path
ieee80211_open (->) ieee80211_do_open (->) ieee80211_mod_tpt_led_trig
(->) ieee80211_start_tpt_led_trig (->) tpt_trig_timer before initializing
it.
the intilization of this read/write lock happens via the path
ieee80211_led_init (->) led_trigger_register, but we are doing
'ieee80211_led_init'  after 'ieeee80211_if_add' where we
register netdev_ops.
so we access leddev_list_lock before initializing it and causes the
following bug in chrome laptops with AR928X cards with the following
script

while true
do
sudo modprobe -v ath9k
sleep 3
sudo modprobe -r ath9k
sleep 3
done

BUG: rwlock bad magic on CPU#1, wpa_supplicant/358, f5b9eccc
Pid: 358, comm: wpa_supplicant Not tainted 3.0.13 #1
Call Trace:

[<8137b9df>] rwlock_bug+0x3d/0x47
[<81179830>] do_raw_read_lock+0x19/0x29
[<8137f063>] _raw_read_lock+0xd/0xf
[<f9081957>] tpt_trig_timer+0xc3/0x145 [mac80211]
[<f9081f3a>] ieee80211_mod_tpt_led_trig+0x152/0x174 [mac80211]
[<f9076a3f>] ieee80211_do_open+0x11e/0x42e [mac80211]
[<f9075390>] ? ieee80211_check_concurrent_iface+0x26/0x13c [mac80211]
[<f9076d97>] ieee80211_open+0x48/0x4c [mac80211]
[<812dbed8>] __dev_open+0x82/0xab
[<812dc0c9>] __dev_change_flags+0x9c/0x113
[<812dc1ae>] dev_change_flags+0x18/0x44
[<8132144f>] devinet_ioctl+0x243/0x51a
[<81321ba9>] inet_ioctl+0x93/0xac
[<812cc951>] sock_ioctl+0x1c6/0x1ea
[<812cc78b>] ? might_fault+0x20/0x20
[<810b1ebb>] do_vfs_ioctl+0x46e/0x4a2
[<810a6ebb>] ? fget_light+0x2f/0x70
[<812ce549>] ? sys_recvmsg+0x3e/0x48
[<810b1f35>] sys_ioctl+0x46/0x69
[<8137fa77>] sysenter_do_call+0x12/0x2

Cc: <stable@vger.kernel.org>
Cc: Gary Morain <gmorain@google.com>
Cc: Paul Stewart <pstew@google.com>
Cc: Abhijit Pradhan <abhijit@qca.qualcomm.com>
Cc: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Cc: Rajkumar Manoharan <rmanohar@qca.qualcomm.com>
Acked-by: Johannes Berg <johannes.berg@intel.com>
Tested-by: Mohammed Shafi Shajakhan <mohammed@qca.qualcomm.com>
Signed-off-by: Mohammed Shafi Shajakhan <mohammed@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agozd1211rw: firmware needs duration_id set to zero for non-pspoll frames
Tomas Vanek [Sun, 5 Feb 2012 13:51:53 +0000 (15:51 +0200)]
zd1211rw: firmware needs duration_id set to zero for non-pspoll frames

Some devices (iwl5100) cannot connect to zd1211rw based AP. It appears that
zd1211 firmware messes up duration_id field if it is not set to zero by driver.

Sniffing traffic shows that zd1211 is transmitting frames with duration_id bits
14 and 15 set and other bits appearing random. Setting duration_id at driver to
zero results zd1211 outputting sane duration_id. This means that firmware is
setting correct values itself and expects duration_id to be zero in first
place.

Looking at vendor driver shows that only PSPoll frames have duration_id set by
driver, for other frames duration_id left zero.

Original bug-report and attached patch at:
  http://sourceforge.net/mailarchive/message.php?msg_id=28759111

Reported-by: Tomas Vanek <Tomas.Vanek@fbl.cz>
[modified original patch from bug-report, added check for pspoll frame]
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agobcma: don't fail for bad SPROM CRC
Henrik Rydberg [Tue, 31 Jan 2012 19:22:15 +0000 (14:22 -0500)]
bcma: don't fail for bad SPROM CRC

The brcmsmac driver is now using the bcma SPROM CRC check, which does
not recognize all chipsets that were functional prior to the switch. In
particular, the current code bails out on odd CRC errors in recent
Macbooks. This patch ignores those errors, with the argument that an
unrecognized SPROM should be treated similarly to a non-existing one.

Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoath9k_hw: fix a RTS/CTS timeout regression
Felix Fietkau [Sun, 5 Feb 2012 20:15:18 +0000 (21:15 +0100)]
ath9k_hw: fix a RTS/CTS timeout regression

commit adb5066 "ath9k_hw: do not apply the 2.4 ghz ack timeout
workaround to cts" reduced the hardware CTS timeout to the normal
values specified by the standard, but it turns out while it doesn't
need the same extra time that it needs for the ACK timeout, it
does need more than the value specified in the standard, but only
for 2.4 GHz.

This patch brings the CTS timeout value in sync with the initialization
values, while still allowing adjustment for bigger distances.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Cc: stable@vger.kernel.org
Reported-by: Seth Forshee <seth.forshee@canonical.com>
Reported-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoath9k: fix a WEP crypto related regression
Felix Fietkau [Sun, 5 Feb 2012 20:15:17 +0000 (21:15 +0100)]
ath9k: fix a WEP crypto related regression

commit b4a82a0 "ath9k_hw: fix interpretation of the rx KeyMiss flag"
fixed the interpretation of the KeyMiss flag for keycache based lookups,
however WEP encryption uses a static index, so KeyMiss is always asserted
for it, even though frames are decrypted properly.
Fix this by clearing the ATH9K_RXERR_KEYMISS flag if no keycache based
lookup was performed.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Cc: stable@vger.kernel.org
Reported-by: Laurent Bonnans <bonnans.l@gmail.com>
Reported-by: Jurica Vukadin <u.ra604@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agomwifiex: add NULL checks in driver unload path
Amitkumar Karwar [Sat, 4 Feb 2012 04:34:02 +0000 (20:34 -0800)]
mwifiex: add NULL checks in driver unload path

If driver load is failed, sometimes few pointers may remain
uninitialized ex. priv->wdev, priv->netdev, adapter->sleep_cfm
This will cause NULL pointer dereferance while unloading the
driver.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Kiran Divekar <dkiran@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoath9k: Fix kernel panic during driver initilization
Mohammed Shafi Shajakhan [Thu, 2 Feb 2012 10:59:05 +0000 (16:29 +0530)]
ath9k: Fix kernel panic during driver initilization

all works need to be initialized before ieee80211_register_hw
to prevent mac80211 call backs such as drv_start, drv_config
getting started. otherwise we would queue/cancel works before
initializing them and it leads to kernel panic.
this issue can be recreated with the following script
in Chrome laptops with AR928X cards, with background scan
running (or) Network manager is running

while true
do
sudo modprobe -v ath9k
sleep 3
sudo modprobe -r ath9k
sleep 3
done

 EIP: [<81040a47>] __cancel_work_timer+0xb8/0xe1 SS:ESP 0068:f6be9d70
 ---[ end trace 4f86d6139a9900ef ]---
 Registered led device: ath9k-phy0
 ieee80211 phy0: Atheros AR9280 Rev:2 mem=0xf88a0000,
 irq=16
 Kernel panic - not syncing: Fatal exception
 Pid: 456, comm: wpa_supplicant Tainted: G      D
 3.0.13 #1
Call Trace:
 [<81379e21>] panic+0x53/0x14a
 [<81004a30>] oops_end+0x73/0x81
 [<81004b53>] die+0x4c/0x55
 [<81002710>] do_trap+0x7c/0x83
 [<81002855>] ? do_bounds+0x58/0x58
 [<810028cc>] do_invalid_op+0x77/0x81
 [<81040a47>] ? __cancel_work_timer+0xb8/0xe1
 [<810489ec>] ? sched_clock_cpu+0x81/0x11f
 [<8103f809>] ? wait_on_work+0xe2/0xf7
 [<8137f807>] error_code+0x67/0x6c
 [<810300d8>] ? wait_consider_task+0x4ba/0x84c
 [<81040a47>] ? __cancel_work_timer+0xb8/0xe1
 [<810380c9>] ? try_to_del_timer_sync+0x5f/0x67
 [<81040a91>] cancel_work_sync+0xf/0x11
 [<f88d7b7c>] ath_set_channel+0x62/0x25c [ath9k]
 [<f88d67d1>] ? ath9k_tx_last_beacon+0x26a/0x85c [ath9k]
 [<f88d8899>] ath_radio_disable+0x3f1/0x68e [ath9k]
 [<f90d0edb>] ieee80211_hw_config+0x111/0x116 [mac80211]
 [<f90dd95c>] __ieee80211_recalc_idle+0x919/0xa37 [mac80211]
 [<f90dda76>] __ieee80211_recalc_idle+0xa33/0xa37 [mac80211]
 [<812dbed8>] __dev_open+0x82/0xab

Cc: <stable@vger.kernel.org>
Cc: Gary Morain <gmorain@google.com>
Cc: Paul Stewart <pstew@google.com>
Cc: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Tested-by: Mohammed Shafi Shajakhan <mohammed@qca.qualcomm.com>
Signed-off-by: Rajkumar Manoharan <rmanohar@qca.qualcomm.com>
Signed-off-by: Mohammed Shafi Shajakhan <mohammed@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agomwifiex: handle association failure case correctly
Amitkumar Karwar [Thu, 2 Feb 2012 04:41:43 +0000 (20:41 -0800)]
mwifiex: handle association failure case correctly

Currently even if association is failed "iw link" shows some
information about connected BSS and "Tx timeout" error is seen in
dmesg log.

This patch fixes below issues in the code to handle assoc failure
case correctly.
1) "status" variable in mwifiex_wait_queue_complete() is not
correctly updated. Hence driver doesn't inform cfg80211 stack
about association failure.
2) During association network queues are stopped but carrier is
not cleared, which gives Tx timeout error in failure case

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>
12 years agoath9k: use WARN_ON_ONCE in ath_rc_get_highest_rix
John W. Linville [Mon, 30 Jan 2012 15:55:13 +0000 (10:55 -0500)]
ath9k: use WARN_ON_ONCE in ath_rc_get_highest_rix

The device seems to survive the issue, so no need to flood the logs
about it...

Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agomac80211: timeout a single frame in the rx reorder buffer
Eliad Peller [Wed, 1 Feb 2012 16:48:09 +0000 (18:48 +0200)]
mac80211: timeout a single frame in the rx reorder buffer

The current code checks for stored_mpdu_num > 1, causing
the reorder_timer to be triggered indefinitely, but the
frame is never timed-out (until the next packet is received)

Signed-off-by: Eliad Peller <eliad@wizery.com>
Cc: <stable@vger.kernel.org>
Acked-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoiwlwifi: don't mess up QoS counters with non-QoS frames
Emmanuel Grumbach [Wed, 1 Feb 2012 15:01:32 +0000 (07:01 -0800)]
iwlwifi: don't mess up QoS counters with non-QoS frames

In my AMPDU rework, I rely on the sequence numbers of frames. But
I didn't check that the frame has a valid tid before updating the
tracking counters. As a result, the Tx queues were stalled. People
who hit this bug saw that we simply didn't let any data out.

This bug was introduced in 3.3.

This patch fixes that and checks that the frame is a QoS frame before
looking at its tid and changing the counters.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agossb: fix cardbus slot in hostmode
Hauke Mehrtens [Tue, 31 Jan 2012 23:13:53 +0000 (00:13 +0100)]
ssb: fix cardbus slot in hostmode

ssb supports one extra device on the cardbus. This results in two
devices in total, one beeing the host controller itself and the other
the external device connected to the hostcontroller over cardbus.

This makes the cardbus slot work on the Linksys WRT150N.

Reported-by: Aaron Z <aaronz@pls-net.org>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoiwlwifi: make "Tx aggregation enabled on ra =" be at DEBUG level
John W. Linville [Tue, 31 Jan 2012 19:40:49 +0000 (14:40 -0500)]
iwlwifi: make "Tx aggregation enabled on ra =" be at DEBUG level

Average users either don't care or they think that message indicates a
problem...

Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agort2800lib: fix wrong -128dBm when signal is stronger than -12dBm
Luigi Tarenga [Tue, 31 Jan 2012 17:51:23 +0000 (18:51 +0100)]
rt2800lib: fix wrong -128dBm when signal is stronger than -12dBm

This patch correct the type of variables containing the rssi
values read from the rxwi.

In function rt2800_agc_to_rssi() 3 variables (rssi0, rssi1, rss2)
defined as int was assigned a 16bit signed values as unsigned.
From a test with a hi-gain antenna I verified that the rxwi
contains signed rssi values in the  range -13/+81 (inclusive)
with 0 as an error condition. In case of negative values a
condition is triggered and the function return -128dBm while
the signal is at its maximum. This patch correct the cast so
negative values are not treated as very high positive values
(ex. -13 does not become 243).

Signed-off-by: Luigi Tarenga <luigi.tarenga@gmail.com>
Reviewed-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agobcma: Fix mem leak in bcma_bus_scan()
Jesper Juhl [Sun, 29 Jan 2012 20:34:04 +0000 (21:34 +0100)]
bcma: Fix mem leak in bcma_bus_scan()

bcma_bus_scan() leaks 'struct bcma_device' bytes if
bcma_get_next_core() returns error.

Restructure the code so we always kfree() the memory we allocate to
the variable 'core' before it goes out of scope.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoaf_unix: fix EPOLLET regression for stream sockets
Eric Dumazet [Sat, 28 Jan 2012 16:11:03 +0000 (16:11 +0000)]
af_unix: fix EPOLLET regression for stream sockets

Commit 0884d7aa24 (AF_UNIX: Fix poll blocking problem when reading from
a stream socket) added a regression for epoll() in Edge Triggered mode
(EPOLLET)

Appropriate fix is to use skb_peek()/skb_unlink() instead of
skb_dequeue(), and only call skb_unlink() when skb is fully consumed.

This remove the need to requeue a partial skb into sk_receive_queue head
and the extra sk->sk_data_ready() calls that added the regression.

This is safe because once skb is given to sk_receive_queue, it is not
modified by a writer, and readers are serialized by u->readlock mutex.

This also reduce number of spinlock acquisition for small reads or
MSG_PEEK users so should improve overall performance.

Reported-by: Nick Mathewson <nickm@freehaven.net>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Alexey Moiseytsev <himeraster@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: fix tcp_trim_head() to adjust segment count with skb MSS
Neal Cardwell [Sat, 28 Jan 2012 17:29:46 +0000 (17:29 +0000)]
tcp: fix tcp_trim_head() to adjust segment count with skb MSS

This commit fixes tcp_trim_head() to recalculate the number of
segments in the skb with the skb's existing MSS, so trimming the head
causes the skb segment count to be monotonically non-increasing - it
should stay the same or go down, but not increase.

Previously tcp_trim_head() used the current MSS of the connection. But
if there was a decrease in MSS between original transmission and ACK
(e.g. due to PMTUD), this could cause tcp_trim_head() to
counter-intuitively increase the segment count when trimming bytes off
the head of an skb. This violated assumptions in tcp_tso_acked() that
tcp_trim_head() only decreases the packet count, so that packets_acked
in tcp_tso_acked() could underflow, leading tcp_clean_rtx_queue() to
pass u32 pkts_acked values as large as 0xffffffff to
ca_ops->pkts_acked().

As an aside, if tcp_trim_head() had really wanted the skb to reflect
the current MSS, it should have called tcp_set_skb_tso_segs()
unconditionally, since a decrease in MSS would mean that a
single-packet skb should now be sliced into multiple segments.

Signed-off-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/tcp: Fix tcp memory limits initialization when !CONFIG_SYSCTL
Glauber Costa [Mon, 30 Jan 2012 01:20:17 +0000 (01:20 +0000)]
net/tcp: Fix tcp memory limits initialization when !CONFIG_SYSCTL

sysctl_tcp_mem() initialization was moved to sysctl_tcp_ipv4.c
in commit 3dc43e3e4d0b52197d3205214fe8f162f9e0c334, since it
became a per-ns value.

That code, however, will never run when CONFIG_SYSCTL is
disabled, leading to bogus values on those fields - causing hung
TCP sockets.

This patch fixes it by keeping an initialization code in
tcp_init(). It will be overwritten by the first net namespace
init if CONFIG_SYSCTL is compiled in, and do the right thing if
it is compiled out.

It is also named properly as tcp_init_mem(), to properly signal
its non-sysctl side effect on TCP limits.

Reported-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: David S. Miller <davem@davemloft.net>
Link: http://lkml.kernel.org/r/4F22D05A.8030604@parallels.com
[ renamed the function, tidied up the changelog a bit ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet caif: Register properly as a pernet subsystem.
Eric W. Biederman [Thu, 26 Jan 2012 14:04:53 +0000 (14:04 +0000)]
net caif: Register properly as a pernet subsystem.

caif is a subsystem and as such it needs to register with
register_pernet_subsys instead of register_pernet_device.

Among other problems using register_pernet_device was resulting in
net_generic being called before the caif_net structure was allocated.
Which has been causing net_generic to fail with either BUG_ON's or by
return NULL pointers.

A more ugly problem that could be caused is packets in flight why the
subsystem is shutting down.

To remove confusion also remove the cruft cause by inappropriately
trying to fix this bug.

With the aid of the previous patch I have tested this patch and
confirmed that using register_pernet_subsys makes the failure go away as
it should.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Tested-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetns: Fail conspicously if someone uses net_generic at an inappropriate time.
Eric W. Biederman [Thu, 26 Jan 2012 14:02:55 +0000 (14:02 +0000)]
netns: Fail conspicously if someone uses net_generic at an inappropriate time.

By definition net_generic should never be called when it can return
NULL.  Fail conspicously with a BUG_ON to make it clear when people mess
up that a NULL return should never happen.

Recently there was a bug in the CAIF subsystem where it was registered
with register_pernet_device instead of register_pernet_subsys.  It was
erroneously concluded that net_generic could validly return NULL and
that net_assign_generic was buggy (when it was just inefficient).
Hopefully this BUG_ON will prevent people to coming to similar erroneous
conclusions in the futrue.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Tested-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless
David S. Miller [Sat, 28 Jan 2012 01:40:18 +0000 (20:40 -0500)]
Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless

12 years agonet: explicitly add jump_label.h header to sock.h
Glauber Costa [Thu, 26 Jan 2012 12:09:28 +0000 (12:09 +0000)]
net: explicitly add jump_label.h header to sock.h

Commit 36a1211970193ce215de50ed1e4e1272bc814df1 removed linux/module.h
include statement from one of the headers that end up in net/sock.h.
It was providing us with static_branch() definition implicitly, so
after its removal the build got broken.

To fix this, and avoid having this happening in the future,
let me do the right thing and include linux/jump_label.h
explicitly in sock.h.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Reported-by: Randy Dunlap <rdunlap@xenotime.net>
CC: David S. Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: RTNETLINK adjusting values of min_ifinfo_dump_size
Stefan Gula [Thu, 26 Jan 2012 11:01:06 +0000 (11:01 +0000)]
net: RTNETLINK adjusting values of min_ifinfo_dump_size

Setting link parameters on a netdevice changes the value
of if_nlmsg_size(), therefore it is necessary to recalculate
min_ifinfo_dump_size.

Signed-off-by: Stefan Gula <steweg@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv6: Fix ip_gre lockless xmits.
Willem de Bruijn [Thu, 26 Jan 2012 10:34:35 +0000 (10:34 +0000)]
ipv6: Fix ip_gre lockless xmits.

Tunnel devices set NETIF_F_LLTX to bypass HARD_TX_LOCK.  Sit and
ipip set this unconditionally in ops->setup, but gre enables it
conditionally after parameter passing in ops->newlink. This is
not called during tunnel setup as below, however, so GRE tunnels are
still taking the lock.

modprobe ip_gre
ip tunnel add test0 mode gre remote 10.5.1.1 dev lo
ip link set test0 up
ip addr add 10.6.0.1 dev test0
 # cat /sys/class/net/test0/features
 # $DIR/test_tunnel_xmit 10 10.5.2.1
ip route add 10.5.2.0/24 dev test0
ip tunnel del test0

The newlink callback is only called in rtnl_netlink, and only if
the device is new, as it calls register_netdevice internally. Gre
tunnels are created at 'ip tunnel add' with ioctl SIOCADDTUNNEL,
which calls ipgre_tunnel_locate, which calls register_netdev.
rtnl_newlink is called at 'ip link set', but skips ops->newlink
and the device is up with locking still enabled. The equivalent
ipip tunnel works fine, btw (just substitute 'method gre' for
'method ipip').

On kernels before /sys/class/net/*/features was removed [1],
the first commented out line returns 0x6000 with method gre,
which indicates that NETIF_F_LLTX (0x1000) is not set. With ipip,
it reports 0x7000. This test cannot be used on recent kernels where
the sysfs file is removed (and ETHTOOL_GFEATURES does not currently
work for tunnel devices, because they lack dev->ethtool_ops).

The second commented out line calls a simple transmission test [2]
that sends on 24 cores at maximum rate. Results of a single run:

ipip: 19,372,306
gre before patch:  4,839,753
gre after patch: 19,133,873

This patch replicates the condition check in ipgre_newlink to
ipgre_tunnel_locate. It works for me, both with oseq on and off.
This is the first time I looked at rtnetlink and iproute2 code,
though, so someone more knowledgeable should probably check the
patch. Thanks.

The tail of both functions is now identical, by the way. To avoid
code duplication, I'll be happy to rework this and merge the two.

[1] http://patchwork.ozlabs.org/patch/104610/
[2] http://kernel.googlecode.com/files/xmit_udp_parallel.c

Signed-off-by: Willem de Bruijn <willemb@google.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoxen-netfront: correct MAX_TX_TARGET calculation.
Wei Liu [Thu, 26 Jan 2012 07:23:23 +0000 (07:23 +0000)]
xen-netfront: correct MAX_TX_TARGET calculation.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetns: fix net_alloc_generic()
Eric Dumazet [Thu, 26 Jan 2012 00:41:38 +0000 (00:41 +0000)]
netns: fix net_alloc_generic()

When a new net namespace is created, we should attach to it a "struct
net_generic" with enough slots (even empty), or we can hit the following
BUG_ON() :

[  200.752016] kernel BUG at include/net/netns/generic.h:40!
...
[  200.752016]  [<ffffffff825c3cea>] ? get_cfcnfg+0x3a/0x180
[  200.752016]  [<ffffffff821cf0b0>] ? lockdep_rtnl_is_held+0x10/0x20
[  200.752016]  [<ffffffff825c41be>] caif_device_notify+0x2e/0x530
[  200.752016]  [<ffffffff810d61b7>] notifier_call_chain+0x67/0x110
[  200.752016]  [<ffffffff810d67c1>] raw_notifier_call_chain+0x11/0x20
[  200.752016]  [<ffffffff821bae82>] call_netdevice_notifiers+0x32/0x60
[  200.752016]  [<ffffffff821c2b26>] register_netdevice+0x196/0x300
[  200.752016]  [<ffffffff821c2ca9>] register_netdev+0x19/0x30
[  200.752016]  [<ffffffff81c1c67a>] loopback_net_init+0x4a/0xa0
[  200.752016]  [<ffffffff821b5e62>] ops_init+0x42/0x180
[  200.752016]  [<ffffffff821b600b>] setup_net+0x6b/0x100
[  200.752016]  [<ffffffff821b6466>] copy_net_ns+0x86/0x110
[  200.752016]  [<ffffffff810d5789>] create_new_namespaces+0xd9/0x190

net_alloc_generic() should take into account the maximum index into the
ptr array, as a subsystem might use net_generic() anytime.

This also reduces number of reallocations in net_assign_generic()

Reported-by: Sasha Levin <levinsasha928@gmail.com>
Tested-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Sjur Brændeland <sjur.brandeland@stericsson.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: bind() optimize port allocation
Flavio Leitner [Wed, 25 Jan 2012 08:34:52 +0000 (08:34 +0000)]
tcp: bind() optimize port allocation

Port autoselection finds a port and then drop the lock,
then right after that, gets the hash bucket again and lock it.

Fix it to go direct.

Signed-off-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: Marcelo Ricardo Leitner <mleitner@redhat.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: bind() fix autoselection to share ports
Flavio Leitner [Wed, 25 Jan 2012 08:34:51 +0000 (08:34 +0000)]
tcp: bind() fix autoselection to share ports

The current code checks for conflicts when the application
requests a specific port.  If there is no conflict, then
the request is granted.

On the other hand, the port autoselection done by the kernel
fails when all ports are bound even when there is a port
with no conflict available.

The fix changes port autoselection to check if there is a
conflict and use it if not.

Signed-off-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: Marcelo Ricardo Leitner <mleitner@redhat.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agol2tp: l2tp_ip - fix possible oops on packet receive
James Chapman [Wed, 25 Jan 2012 02:39:05 +0000 (02:39 +0000)]
l2tp: l2tp_ip - fix possible oops on packet receive

When a packet is received on an L2TP IP socket (L2TPv3 IP link
encapsulation), the l2tpip socket's backlog_rcv function calls
xfrm4_policy_check(). This is not necessary, since it was called
before the skb was added to the backlog. With CONFIG_NET_NS enabled,
xfrm4_policy_check() will oops if skb->dev is null, so this trivial
patch removes the call.

This bug has always been present, but only when CONFIG_NET_NS is
enabled does it cause problems. Most users are probably using UDP
encapsulation for L2TP, hence the problem has only recently
surfaced.

EIP: 0060:[<c12bb62b>] EFLAGS: 00210246 CPU: 0
EIP is at l2tp_ip_recvmsg+0xd4/0x2a7
EAX: 00000001 EBX: d77b5180 ECX: 00000000 EDX: 00200246
ESI: 00000000 EDI: d63cbd30 EBP: d63cbd18 ESP: d63cbcf4
 DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
Call Trace:
 [<c1218568>] sock_common_recvmsg+0x31/0x46
 [<c1215c92>] __sock_recvmsg_nosec+0x45/0x4d
 [<c12163a1>] __sock_recvmsg+0x31/0x3b
 [<c1216828>] sock_recvmsg+0x96/0xab
 [<c10b2693>] ? might_fault+0x47/0x81
 [<c10b2693>] ? might_fault+0x47/0x81
 [<c1167fd0>] ? _copy_from_user+0x31/0x115
 [<c121e8c8>] ? copy_from_user+0x8/0xa
 [<c121ebd6>] ? verify_iovec+0x3e/0x78
 [<c1216604>] __sys_recvmsg+0x10a/0x1aa
 [<c1216792>] ? sock_recvmsg+0x0/0xab
 [<c105a99b>] ? __lock_acquire+0xbdf/0xbee
 [<c12d5a99>] ? do_page_fault+0x193/0x375
 [<c10d1200>] ? fcheck_files+0x9b/0xca
 [<c10d1259>] ? fget_light+0x2a/0x9c
 [<c1216bbb>] sys_recvmsg+0x2b/0x43
 [<c1218145>] sys_socketcall+0x16d/0x1a5
 [<c11679f0>] ? trace_hardirqs_on_thunk+0xc/0x10
 [<c100305f>] sysenter_do_call+0x12/0x38
Code: c6 05 8c ea a8 c1 01 e8 0c d4 d9 ff 85 f6 74 07 3e ff 86 80 00 00 00 b9 17 b6 2b c1 ba 01 00 00 00 b8 78 ed 48 c1 e8 23 f6 d9 ff <ff> 76 0c 68 28 e3 30 c1 68 2d 44 41 c1 e8 89 57 01 00 83 c4 0c

Signed-off-by: James Chapman <jchapman@katalix.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Tue, 24 Jan 2012 23:51:40 +0000 (15:51 -0800)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

Davem says:

1) Fix JIT code generation on x86-64 for divide by zero, from Eric Dumazet.

2) tg3 header length computation correction from Eric Dumazet.

3) More build and reference counting fixes for socket memory cgroup
   code from Glauber Costa.

4) module.h snuck back into a core header after all the hard work we
   did to remove that, from Paul Gortmaker and Jesper Dangaard Brouer.

5) Fix PHY naming regression and add some new PCI IDs in stmmac, from
   Alessandro Rubini.

6) Netlink message generation fix in new team driver, should only advertise
   the entries that changed during events, from Jiri Pirko.

7) SRIOV VF registration and unregistration fixes, and also add a
   missing PCI ID, from Roopa Prabhu.

8) Fix infinite loop in tx queue flush code of brcmsmac, from Stanislaw Gruszka.

9) ftgmac100/ftmac100 build fix, missing interrupt.h include.

10) Memory leak fix in net/hyperv do_set_mutlicast() handling, from Wei Yongjun.

11) Off by one fix in netem packet scheduler, from Vijay Subramanian.

12) TCP loss detection fix from Yuchung Cheng.

13) TCP reset packet MD5 calculation uses wrong address, fix from Shawn Lu.

14) skge carrier assertion and DMA mapping fixes from Stephen Hemminger.

15) Congestion recovery undo performed at the wrong spot in BIC and CUBIC
    congestion control modules, fix from Neal Cardwell.

16) Ethtool ETHTOOL_GSSET_INFO is unnecessarily restrictive, from MichaÅ‚ MirosÅ‚aw.

17) Fix triggerable race in ipv6 sysctl handling, from Francesco Ruggeri.

18) Statistics bug fixes in mlx4 from Eugenia Emantayev.

19) rds locking bug fix during info dumps, from your's truly.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (67 commits)
  rds: Make rds_sock_lock BH rather than IRQ safe.
  netprio_cgroup.h: dont include module.h from other includes
  net: flow_dissector.c missing include linux/export.h
  team: send only changed options/ports via netlink
  net/hyperv: fix possible memory leak in do_set_multicast()
  drivers/net: dsa/mv88e6xxx.c files need linux/module.h
  stmmac: added PCI identifiers
  llc: Fix race condition in llc_ui_recvmsg
  stmmac: fix phy naming inconsistency
  dsa: Add reporting of silicon revision for Marvell 88E6123/88E6161/88E6165 switches.
  tg3: fix ipv6 header length computation
  skge: add byte queue limit support
  mv643xx_eth: Add Rx Discard and Rx Overrun statistics
  bnx2x: fix compilation error with SOE in fw_dump
  bnx2x: handle CHIP_REVISION during init_one
  bnx2x: allow user to change ring size in ISCSI SD mode
  bnx2x: fix Big-Endianess in ethtool -t
  bnx2x: fixed ethtool statistics for MF modes
  bnx2x: credit-leakage fixup on vlan_mac_del_all
  macvlan: fix a possible use after free
  ...

12 years agords: Make rds_sock_lock BH rather than IRQ safe.
David S. Miller [Tue, 24 Jan 2012 22:03:44 +0000 (17:03 -0500)]
rds: Make rds_sock_lock BH rather than IRQ safe.

rds_sock_info() triggers locking warnings because we try to perform a
local_bh_enable() (via sock_i_ino()) while hardware interrupts are
disabled (via taking rds_sock_lock).

There is no reason for rds_sock_lock to be a hardware IRQ disabling
lock, none of these access paths run in hardware interrupt context.

Therefore making it a BH disabling lock is safe and sufficient to
fix this bug.

Reported-by: Kumar Sanghvi <kumaras@chelsio.com>
Reported-by: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetprio_cgroup.h: dont include module.h from other includes
Paul Gortmaker [Tue, 24 Jan 2012 11:33:19 +0000 (11:33 +0000)]
netprio_cgroup.h: dont include module.h from other includes

A considerable effort was invested in wiping out module.h
from being present in all the other standard includes.  This
one leaked back in, but once again isn't strictly necessary,
so remove it.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: flow_dissector.c missing include linux/export.h
Jesper Dangaard Brouer [Tue, 24 Jan 2012 21:03:33 +0000 (16:03 -0500)]
net: flow_dissector.c missing include linux/export.h

The file net/core/flow_dissector.c seems to be missing
including linux/export.h.

Signed-off-by: Jesper Dangaard Brouer <hawk@comx.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoteam: send only changed options/ports via netlink
Jiri Pirko [Tue, 24 Jan 2012 05:16:00 +0000 (05:16 +0000)]
team: send only changed options/ports via netlink

This patch changes event message behaviour to send only updated records
instead of whole list. This fixes bug on which userspace receives non-actual
data in case multiple events occur in row.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/hyperv: fix possible memory leak in do_set_multicast()
Wei Yongjun [Tue, 24 Jan 2012 10:21:28 +0000 (10:21 +0000)]
net/hyperv: fix possible memory leak in do_set_multicast()

do_set_multicast() may not free the memory malloc in
netvsc_set_multicast_list().

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agodrivers/net: dsa/mv88e6xxx.c files need linux/module.h
Paul Gortmaker [Tue, 24 Jan 2012 10:41:40 +0000 (10:41 +0000)]
drivers/net: dsa/mv88e6xxx.c files need linux/module.h

An implicit instance of module.h leaked back into existence
and was masking the fact that these drivers weren't calling
out the include for itself.  Fix the drivers before we remove
the implicit include path via net/netprio_cgroup.h file.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agostmmac: added PCI identifiers
Alessandro Rubini [Mon, 23 Jan 2012 23:08:56 +0000 (23:08 +0000)]
stmmac: added PCI identifiers

STM has a device ID within its own VENDOR space, and it is being
used in the STA2X11 I/O Hub.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Giancarlo Asnaghi <giancarlo.asnaghi@st.com>
Acked-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agollc: Fix race condition in llc_ui_recvmsg
Radu Iliescu [Thu, 19 Jan 2012 03:57:57 +0000 (03:57 +0000)]
llc: Fix race condition in llc_ui_recvmsg

There is a race on sk_receive_queue between llc_ui_recvmsg and
sock_queue_rcv_skb.

Our current solution is to protect skb_eat in llc_ui_recvmsg
with the queue spinlock.

Signed-off-by: Radu Iliescu <riliescu@ixiacom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agostmmac: fix phy naming inconsistency
Alessandro Rubini [Mon, 23 Jan 2012 23:26:48 +0000 (23:26 +0000)]
stmmac: fix phy naming inconsistency

After commit "db8857b stmmac: use an unique MDIO bus name" my
device stopped being probed because two different names were being
used in different places. This fixes the inconsistency.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Giancarlo Asnaghi <giancarlo.asnaghi@st.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Florian Fainelli <florian@openwrt.org>
Acked-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs
Linus Torvalds [Tue, 24 Jan 2012 20:12:40 +0000 (12:12 -0800)]
Merge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  quota: Pass information that quota is stored in system file to userspace
  ext2: protect inode changes in the SETVERSION and SETFLAGS ioctls
  jbd: Issue cache flush after checkpointing

12 years agoiwlwifi: fix PCI-E transport "inta" race
Johannes Berg [Thu, 19 Jan 2012 16:20:57 +0000 (08:20 -0800)]
iwlwifi: fix PCI-E transport "inta" race

When an interrupt comes in, we read the reason
bits and collect them into "trans_pcie->inta".
This happens with the spinlock held. However,
there's a bug resetting this variable -- that
happens after the spinlock has been released.
This means that it is possible for interrupts
to be missed if the reset happens after some
other interrupt reasons were already added to
the variable.

I found this by code inspection, looking for a
reason that we sometimes see random commands
time out. It seems possible that this causes
such behaviour, but I can't say for sure right
now since it happens extremely infrequently on
my test systems.

Cc: stable@vger.kernel.org [3.2]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agomac80211: set bss_conf.idle when vif is connected
Eliad Peller [Wed, 11 Jan 2012 11:11:50 +0000 (13:11 +0200)]
mac80211: set bss_conf.idle when vif is connected

__ieee80211_recalc_idle() iterates through the vifs,
sets bss_conf.idle = true if they are disconnected,
and increases "count" if they are not (which later
gets evaluated in order to determine whether the
device is idle).

However, the loop doesn't set bss_conf.idle = false
(along with increasing "count"), causing the device
idle state and the vif idle state to get out of sync
in some cases.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agomac80211: update oper_channel on ibss join
Eliad Peller [Tue, 10 Jan 2012 13:19:54 +0000 (15:19 +0200)]
mac80211: update oper_channel on ibss join

Commit 13c40c5 ("mac80211: Add HT operation modes for IBSS") broke
ibss operation by mistakenly removing the local->oper_channel
update (causing ibss to start on the wrong channel). fix it.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Acked-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoregulator: Fix documentation for of_node parameter of regulator_register()
Mark Brown [Tue, 24 Jan 2012 11:17:26 +0000 (11:17 +0000)]
regulator: Fix documentation for of_node parameter of regulator_register()

Commit 5bc75a886353 ("kernel-doc: fix new warning in regulator core")
added documentation for of_node to address a warning but the
documentation didn't explain what the parameter is for so would be
likely to be unhelpful for users.  Clarify that.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Linus Torvalds [Tue, 24 Jan 2012 18:25:29 +0000 (10:25 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda - Fix silent outputs from docking-station jacks of Dell laptops
  ALSA: HDA: Use model=auto for Thinkpad T510
  ALSA: hda - Fix buffer-alignment regression with Nvidia HDMI
  ALSA: hda - Fix a unused variable warning
  snd-hda-intel: better Alienware M17x R3 quirk
  ALSA: hda/realtek - Remove use_jack_tbl field
  ALSA: hda/realtek - Avoid conflict of unsol-events with static quirks
  ALSA: hda/realtek - Avoid multi-ios conflicting with multi-speakers

12 years agogma500: Fix shmem mapping
Alan Cox [Tue, 24 Jan 2012 16:57:42 +0000 (16:57 +0000)]
gma500: Fix shmem mapping

GMA500 did it the old way and it's been on the TODO list to fix.
Current kernels now blow up if we use the old way so we'd better
do the work!

Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomigrate_mode.h is not exported to user mode
Stephen Rothwell [Tue, 24 Jan 2012 00:41:32 +0000 (11:41 +1100)]
migrate_mode.h is not exported to user mode

so move its include into fs.h inside the __KERNEL__ protection.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge tag 'pm-fixes-for-3.3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Mon, 23 Jan 2012 23:11:27 +0000 (15:11 -0800)]
Merge tag 'pm-fixes-for-3.3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Power management fixes for 3.3

Two fixes for regressions introduced during the merge window, one fix for
a long-standing obscure issue in the computation of hibernate image size
and two small PM documentation fixes.

* tag 'pm-fixes-for-3.3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  PM / Sleep: Fix read_unlock_usermodehelper() call.
  PM / Hibernate: Rewrite unlock_system_sleep() to fix s2disk regression
  PM / Hibernate: Correct additional pages number calculation
  PM / Documentation: Fix minor issue in freezing_of_tasks.txt
  PM / Documentation: Fix spelling mistake in basic-pm-debugging.txt

12 years agoMerge tag 'arm-soc-imx-move' of git://git.kernel.org/pub/scm/linux/kernel/git/arm...
Linus Torvalds [Mon, 23 Jan 2012 22:50:30 +0000 (14:50 -0800)]
Merge tag 'arm-soc-imx-move' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc

Consolidate i.MX 5 platforms to be under the new shared i.MX 3/5/6 tree.

* tag 'arm-soc-imx-move' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ARM i.MX: Update defconfig
  ARM i.MX: Merge i.MX5 support into mach-imx
  ARM i.MX5: remove unnecessary includes from board files

Fix up fairly trivial conflicts due to various changes nearby in
arch/arm/{mach,plat}-imx/{Kconfig,Makefile}

Pull request had been sent to the wrong email address, but happened
before the merge window closed.  I'm merging the MX 5 consolidation,
since it apparently will help the next development window and will avoid
conflicts later as per Arnd.

12 years agoPM / Sleep: Fix read_unlock_usermodehelper() call.
Tetsuo Handa [Mon, 23 Jan 2012 20:59:08 +0000 (21:59 +0100)]
PM / Sleep: Fix read_unlock_usermodehelper() call.

Commit b298d289
 "PM / Sleep: Fix freezer failures due to racy usermodehelper_is_disabled()"
added read_unlock_usermodehelper() but read_unlock_usermodehelper() is called
without read_lock_usermodehelper() when kmalloc() failed.

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Acked-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
12 years agodsa: Add reporting of silicon revision for Marvell 88E6123/88E6161/88E6165 switches.
Chris Healy [Sun, 22 Jan 2012 21:20:54 +0000 (21:20 +0000)]
dsa: Add reporting of silicon revision for Marvell 88E6123/88E6161/88E6165 switches.

Add reporting of silicon revision during the probe function for Marvell 88E6123/88E6161/88E6165 switches.

Signed-off-by: Chris Healy <cphealy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotg3: fix ipv6 header length computation
Eric Dumazet [Mon, 23 Jan 2012 01:22:09 +0000 (01:22 +0000)]
tg3: fix ipv6 header length computation

tg3_start_xmit() makes the wrong assumption for TSOV6 that skb->head
doesnt include any payload data.

if (skb_is_gso_v6(skb))
hdr_len = skb_headlen(skb) - ETH_HLEN;

This is not true anymore after commit f07d960df3 (tcp: avoid frag
allocation for small frames)

We should instead use : skb_transport_offset(skb) + tcp_hdrlen(skb)

Its also true for IPv4

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Matt Carlson <mcarlson@broadcom.com>
CC: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoskge: add byte queue limit support
stephen hemminger [Sun, 22 Jan 2012 09:40:40 +0000 (09:40 +0000)]
skge: add byte queue limit support

This also changes the cleanup logic slightly to aggregate
completed notifications for multiple packets.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agomv643xx_eth: Add Rx Discard and Rx Overrun statistics
Paulius Zaleckas [Mon, 23 Jan 2012 01:16:35 +0000 (01:16 +0000)]
mv643xx_eth: Add Rx Discard and Rx Overrun statistics

These statistics helped me a lot while searching who is losing
packets in my setup.
I added these stats to MIB group since they are very similar,
but just in other registers.
I have tested this patch on 88F6281 SoC.

Signed-off-by: Paulius Zaleckas <paulius.zaleckas@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: fix compilation error with SOE in fw_dump
Yuval Mintz [Mon, 23 Jan 2012 07:31:56 +0000 (07:31 +0000)]
bnx2x: fix compilation error with SOE in fw_dump

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>
12 years agobnx2x: handle CHIP_REVISION during init_one
Ariel Elior [Mon, 23 Jan 2012 07:31:55 +0000 (07:31 +0000)]
bnx2x: handle CHIP_REVISION during init_one

The macro `CHIP_IS_E1x' requires `bp' to be initialized.
As `bp' is not yet initialized during this phase of `bnx2x_init_dev',
it accessed uninitialized fields in the struct.

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>
12 years agobnx2x: allow user to change ring size in ISCSI SD mode
Dmitry Kravkov [Mon, 23 Jan 2012 07:31:54 +0000 (07:31 +0000)]
bnx2x: allow user to change ring size in ISCSI SD mode

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>
12 years agobnx2x: fix Big-Endianess in ethtool -t
Dmitry Kravkov [Mon, 23 Jan 2012 07:31:53 +0000 (07:31 +0000)]
bnx2x: fix Big-Endianess in ethtool -t

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>
12 years agobnx2x: fixed ethtool statistics for MF modes
Yuval Mintz [Mon, 23 Jan 2012 07:31:52 +0000 (07:31 +0000)]
bnx2x: fixed ethtool statistics for MF modes

Previosuly, in MF modes `ethtool -S' lacked some of the statistics
which appeared in non-MF modes. This has been fixed.

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>
12 years agobnx2x: credit-leakage fixup on vlan_mac_del_all
Yuval Mintz [Mon, 23 Jan 2012 07:31:51 +0000 (07:31 +0000)]
bnx2x: credit-leakage fixup on vlan_mac_del_all

Upon insertion of elements into the execution queue, it is validated
that there are enough credits to support additional vlan-macs,
and the credits are consumed. However, when removing a pending
command in `bnx2x_vland_mac_del_all' the consumed credits are not
released, which might cause leakage and eventually the inability to
add new vlan-macs in certain scenarios.

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>
12 years agomacvlan: fix a possible use after free
Eric Dumazet [Mon, 23 Jan 2012 05:38:59 +0000 (05:38 +0000)]
macvlan: fix a possible use after free

Commit bc416d9768 (macvlan: handle fragmented multicast frames) added a
possible use after free in macvlan_handle_frame(), since
ip_check_defrag() uses pskb_may_pull() : skb header can be reallocated.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Ben Greear <greearb@candelatech.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'kernel-doc' from Randy Dunlap
Linus Torvalds [Mon, 23 Jan 2012 18:08:08 +0000 (10:08 -0800)]
Merge branch 'kernel-doc' from Randy Dunlap

The usual kernel-doc fixups from Randy.  Some of them David acked as
merged in his tree, this is the random left-overs.

* kernel-doc:
  docbook: fix sched source file names in device-drivers book
  docbook: change iomap source filename in deviceiobook
  docbook: don't use serial_core.h in device-drivers book
  kernel-doc: fix kernel-doc warnings in sched
  kernel-doc: fix new warnings in cfg80211.h
  kernel-doc: fix new warning in usb.h
  kernel-doc: fix new warnings in device.h
  kernel-doc: fix new warnings in debugfs
  kernel-doc: fix new warning in regulator core
  kernel-doc: fix new warnings in pci
  kernel-doc: fix new warnings in driver-core
  kernel-doc: fix new warnings in auditsc.c
  scripts/kernel-doc: fix fatal error caused by cfg80211.h

12 years agoMerge branch 'akpm'
Linus Torvalds [Mon, 23 Jan 2012 17:27:54 +0000 (09:27 -0800)]
Merge branch 'akpm'

Quoth Andrew:
  "Random fixes.  And a simple new LED driver which I'm trying to sneak
   in while you're not looking."

Sneaking successful.

* akpm:
  score: fix off-by-one index into syscall table
  mm: fix rss count leakage during migration
  SHM_UNLOCK: fix Unevictable pages stranded after swap
  SHM_UNLOCK: fix long unpreemptible section
  kdump: define KEXEC_NOTE_BYTES arch specific for s390x
  mm/hugetlb.c: undo change to page mapcount in fault handler
  mm: memcg: update the correct soft limit tree during migration
  proc: clear_refs: do not clear reserved pages
  drivers/video/backlight/l4f00242t03.c: return proper error in l4f00242t03_probe if regulator_get() fails
  drivers/video/backlight/adp88x0_bl.c: fix bit testing logic
  kprobes: initialize before using a hlist
  ipc/mqueue: simplify reading msgqueue limit
  leds: add led driver for Bachmann's ot200
  mm: __count_immobile_pages(): make sure the node is online
  mm: fix NULL ptr dereference in __count_immobile_pages
  mm: fix warnings regarding enum migrate_mode

12 years agoMerge branch 'for-linus-3.3' of git://git.linaro.org/people/sumitsemwal/linux-dma-buf
Linus Torvalds [Mon, 23 Jan 2012 17:26:37 +0000 (09:26 -0800)]
Merge branch 'for-linus-3.3' of git://git.linaro.org/people/sumitsemwal/linux-dma-buf

* 'for-linus-3.3' of git://git.linaro.org/people/sumitsemwal/linux-dma-buf:
  MAINTAINERS: Add dma-buf sharing framework maintainer

12 years agoALSA: hda - Fix silent outputs from docking-station jacks of Dell laptops
Takashi Iwai [Mon, 23 Jan 2012 17:23:36 +0000 (18:23 +0100)]
ALSA: hda - Fix silent outputs from docking-station jacks of Dell laptops

The recent change of the power-widget handling for IDT codecs caused
the silent output from the docking-station line-out jack.  This was
partially fixed by the commit f2cbba7602383cd9cdd21f0a5d0b8bd1aad47b33
"ALSA: hda - Fix the lost power-setup of seconary pins after PM resume".
But the line-out on the docking-station is still silent when booted
with the jack plugged even by this fix.

The remainig bug is that the power-widget is set off in stac92xx_init()
because the pins in cfg->line_out_pins[] aren't checked there properly
but only hp_pins[] are checked in is_nid_hp_pin().

This patch fixes the problem by checking both HP and line-out pins
and leaving the power-map correctly.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=42637

Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
12 years agoMerge git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Mon, 23 Jan 2012 16:59:49 +0000 (08:59 -0800)]
Merge git://git.samba.org/sfrench/cifs-2.6

* git://git.samba.org/sfrench/cifs-2.6:
  CIFS: Rename *UCS* functions to *UTF16*
  [CIFS] ACL and FSCACHE support no longer EXPERIMENTAL
  [CIFS] Fix build break with multiuser patch when LANMAN disabled
  cifs: warn about impending deprecation of legacy MultiuserMount code
  cifs: fetch credentials out of keyring for non-krb5 auth multiuser mounts
  cifs: sanitize username handling
  keys: add a "logon" key type
  cifs: lower default wsize when unix extensions are not used
  cifs: better instrumentation for coalesce_t2
  cifs: integer overflow in parse_dacl()
  cifs: Fix sparse warning when calling cifs_strtoUCS
  CIFS: Add descriptions to the brlock cache functions

12 years agodocbook: fix sched source file names in device-drivers book
Randy Dunlap [Sat, 21 Jan 2012 19:03:24 +0000 (11:03 -0800)]
docbook: fix sched source file names in device-drivers book

Fix warning since kernel scheduler functions and kernel-doc
notation were moved to other files.

docproc: lnx-33-rc1/kernel/sched.c: No such file or directory

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agodocbook: change iomap source filename in deviceiobook
Randy Dunlap [Sat, 21 Jan 2012 19:03:20 +0000 (11:03 -0800)]
docbook: change iomap source filename in deviceiobook

Fix warning since kernel-doc comments were moved to another
source file.

Warning(lib/iomap.c): no structured comments found

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agodocbook: don't use serial_core.h in device-drivers book
Randy Dunlap [Sat, 21 Jan 2012 19:03:16 +0000 (11:03 -0800)]
docbook: don't use serial_core.h in device-drivers book

Fix new kernel-doc warning.  This file no longer contains
kernel-doc comments.

Warning(include/linux/serial_core.h): no structured comments found

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix kernel-doc warnings in sched
Randy Dunlap [Sat, 21 Jan 2012 19:03:13 +0000 (11:03 -0800)]
kernel-doc: fix kernel-doc warnings in sched

Fix new kernel-doc notation warnings:

Warning(include/linux/sched.h:2094): No description found for parameter 'p'
Warning(include/linux/sched.h:2094): Excess function parameter 'tsk' description in 'is_idle_task'
Warning(kernel/sched/cpupri.c:139): No description found for parameter 'newpri'
Warning(kernel/sched/cpupri.c:139): Excess function parameter 'pri' description in 'cpupri_set'
Warning(kernel/sched/cpupri.c:208): Excess function parameter 'bootmem' description in 'cpupri_init'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix new warnings in cfg80211.h
Randy Dunlap [Sat, 21 Jan 2012 19:03:00 +0000 (11:03 -0800)]
kernel-doc: fix new warnings in cfg80211.h

Fix new kernel-doc warnings:

Warning(include/net/cfg80211.h:1165): No description found for parameter 'channel_type'
Warning(include/net/cfg80211.h:2090): No description found for parameter 'probe_resp_offload'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: linux-wireless@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix new warning in usb.h
Randy Dunlap [Sat, 21 Jan 2012 19:02:56 +0000 (11:02 -0800)]
kernel-doc: fix new warning in usb.h

Fix new kernel-doc warning:

Warning(include/linux/usb.h:1251): No description found for parameter 'num_mapped_sgs'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix new warnings in device.h
Randy Dunlap [Sat, 21 Jan 2012 19:02:51 +0000 (11:02 -0800)]
kernel-doc: fix new warnings in device.h

Fix new kernel-doc warnings:

Warning(include/linux/device.h:299): No description found for parameter 'name'
Warning(include/linux/device.h:299): No description found for parameter 'subsys'
Warning(include/linux/device.h:299): No description found for parameter 'node'
Warning(include/linux/device.h:299): No description found for parameter 'add_dev'
Warning(include/linux/device.h:299): No description found for parameter 'remove_dev'
Warning(include/linux/device.h:685): No description found for parameter 'id'
Warning(include/linux/device.h:1009): No description found for parameter '__driver'
Warning(include/linux/device.h:1009): No description found for parameter '__register'
Warning(include/linux/device.h:1009): No description found for parameter '__unregister'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Lars-Peter Clausen <lars@metafoo.de>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix new warnings in debugfs
Randy Dunlap [Sat, 21 Jan 2012 19:02:42 +0000 (11:02 -0800)]
kernel-doc: fix new warnings in debugfs

Fix new kernel-doc warnings:

Warning(fs/debugfs/file.c:556): No description found for parameter 'nregs'
Warning(fs/debugfs/file.c:556): Excess function parameter 'mregs' description in 'debugfs_print_regs32'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix new warning in regulator core
Randy Dunlap [Sat, 21 Jan 2012 19:02:38 +0000 (11:02 -0800)]
kernel-doc: fix new warning in regulator core

Fix new kernel-doc warning:

Warning(drivers/regulator/core.c:2741): No description found for parameter 'of_node'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Liam Girdwood <lrg@ti.com>
Cc: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix new warnings in pci
Randy Dunlap [Sat, 21 Jan 2012 19:02:35 +0000 (11:02 -0800)]
kernel-doc: fix new warnings in pci

Fix new kernel-doc warnings:

Warning(drivers/pci/pci.c:2811): No description found for parameter 'dev'
Warning(drivers/pci/pci.c:2811): Excess function parameter 'pdev' description in 'pci_intx_mask_supported'
Warning(drivers/pci/pci.c:2894): No description found for parameter 'dev'
Warning(drivers/pci/pci.c:2894): Excess function parameter 'pdev' description in 'pci_check_and_mask_intx'
Warning(drivers/pci/pci.c:2908): No description found for parameter 'dev'
Warning(drivers/pci/pci.c:2908): Excess function parameter 'pdev' description in 'pci_check_and_unmask_intx'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix new warnings in driver-core
Randy Dunlap [Sat, 21 Jan 2012 19:02:28 +0000 (11:02 -0800)]
kernel-doc: fix new warnings in driver-core

Fix new kernel-doc warnings:

Warning(drivers/base/bus.c:925): No description found for parameter 'key'
Warning(drivers/base/bus.c:1241): No description found for parameter 'subsys'
Warning(drivers/base/bus.c:1241): No description found for parameter 'groups'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel-doc: fix new warnings in auditsc.c
Randy Dunlap [Sat, 21 Jan 2012 19:02:24 +0000 (11:02 -0800)]
kernel-doc: fix new warnings in auditsc.c

Fix new kernel-doc warnings in auditsc.c:

Warning(kernel/auditsc.c:1875): No description found for parameter 'success'
Warning(kernel/auditsc.c:1875): No description found for parameter 'return_code'
Warning(kernel/auditsc.c:1875): Excess function parameter 'pt_regs' description in '__audit_syscall_exit'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Eric Paris <eparis@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoscripts/kernel-doc: fix fatal error caused by cfg80211.h
Randy Dunlap [Sat, 21 Jan 2012 18:31:54 +0000 (10:31 -0800)]
scripts/kernel-doc: fix fatal error caused by cfg80211.h

include/net/cfg80211.h uses __must_check in functions that
have kernel-doc notation.  This was confusing scripts/kernel-doc,
so have scripts/kernel-doc ignore "__must_check".

Error(include/net/cfg80211.h:2702): cannot understand prototype: 'struct cfg80211_bss * __must_check cfg80211_inform_bss(...)

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoscore: fix off-by-one index into syscall table
Dan Rosenberg [Fri, 20 Jan 2012 22:34:27 +0000 (14:34 -0800)]
score: fix off-by-one index into syscall table

If the provided system call number is equal to __NR_syscalls, the
current check will pass and a function pointer just after the system
call table may be called, since sys_call_table is an array with total
size __NR_syscalls.

Whether or not this is a security bug depends on what the compiler puts
immediately after the system call table.  It's likely that this won't do
anything bad because there is an additional NULL check on the syscall
entry, but if there happens to be a non-NULL value immediately after the
system call table, this may result in local privilege escalation.

Signed-off-by: Dan Rosenberg <drosenberg@vsecurity.com>
Cc: <stable@vger.kernel.org>
Cc: Chen Liqin <liqin.chen@sunplusct.com>
Cc: Lennox Wu <lennox.wu@gmail.com>
Cc: Eugene Teo <eugeneteo@kernel.sg>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: fix rss count leakage during migration
Konstantin Khlebnikov [Fri, 20 Jan 2012 22:34:24 +0000 (14:34 -0800)]
mm: fix rss count leakage during migration

Memory migration fills a pte with a migration entry and it doesn't
update the rss counters.  Then it replaces the migration entry with the
new page (or the old one if migration failed).  But between these two
passes this pte can be unmaped, or a task can fork a child and it will
get a copy of this migration entry.  Nobody accounts for this in the rss
counters.

This patch properly adjust rss counters for migration entries in
zap_pte_range() and copy_one_pte().  Thus we avoid extra atomic
operations on the migration fast-path.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoSHM_UNLOCK: fix Unevictable pages stranded after swap
Hugh Dickins [Fri, 20 Jan 2012 22:34:21 +0000 (14:34 -0800)]
SHM_UNLOCK: fix Unevictable pages stranded after swap

Commit cc39c6a9bbde ("mm: account skipped entries to avoid looping in
find_get_pages") correctly fixed an infinite loop; but left a problem
that find_get_pages() on shmem would return 0 (appearing to callers to
mean end of tree) when it meets a run of nr_pages swap entries.

The only uses of find_get_pages() on shmem are via pagevec_lookup(),
called from invalidate_mapping_pages(), and from shmctl SHM_UNLOCK's
scan_mapping_unevictable_pages().  The first is already commented, and
not worth worrying about; but the second can leave pages on the
Unevictable list after an unusual sequence of swapping and locking.

Fix that by using shmem_find_get_pages_and_swap() (then ignoring the
swap) instead of pagevec_lookup().

But I don't want to contaminate vmscan.c with shmem internals, nor
shmem.c with LRU locking.  So move scan_mapping_unevictable_pages() into
shmem.c, renaming it shmem_unlock_mapping(); and rename
check_move_unevictable_page() to check_move_unevictable_pages(), looping
down an array of pages, oftentimes under the same lock.

Leave out the "rotate unevictable list" block: that's a leftover from
when this was used for /proc/sys/vm/scan_unevictable_pages, whose flawed
handling involved looking at pages at tail of LRU.

Was there significance to the sequence first ClearPageUnevictable, then
test page_evictable, then SetPageUnevictable here? I think not, we're
under LRU lock, and have no barriers between those.

Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michel Lespinasse <walken@google.com>
Cc: <stable@vger.kernel.org> [back to 3.1 but will need respins]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoSHM_UNLOCK: fix long unpreemptible section
Hugh Dickins [Fri, 20 Jan 2012 22:34:19 +0000 (14:34 -0800)]
SHM_UNLOCK: fix long unpreemptible section

scan_mapping_unevictable_pages() is used to make SysV SHM_LOCKed pages
evictable again once the shared memory is unlocked.  It does this with
pagevec_lookup()s across the whole object (which might occupy most of
memory), and takes 300ms to unlock 7GB here.  A cond_resched() every
PAGEVEC_SIZE pages would be good.

However, KOSAKI-san points out that this is called under shmem.c's
info->lock, and it's also under shm.c's shm_lock(), both spinlocks.
There is no strong reason for that: we need to take these pages off the
unevictable list soonish, but those locks are not required for it.

So move the call to scan_mapping_unevictable_pages() from shmem.c's
unlock handling up to shm.c's unlock handling.  Remove the recently
added barrier, not needed now we have spin_unlock() before the scan.

Use get_file(), with subsequent fput(), to make sure we have a reference
to mapping throughout scan_mapping_unevictable_pages(): that's something
that was previously guaranteed by the shm_lock().

Remove shmctl's lru_add_drain_all(): we don't fault in pages at SHM_LOCK
time, and we lazily discover them to be Unevictable later, so it serves
no purpose for SHM_LOCK; and serves no purpose for SHM_UNLOCK, since
pages still on pagevec are not marked Unevictable.

The original code avoided redundant rescans by checking VM_LOCKED flag
at its level: now avoid them by checking shp's SHM_LOCKED.

The original code called scan_mapping_unevictable_pages() on a locked
area at shm_destroy() time: perhaps we once had accounting cross-checks
which required that, but not now, so skip the overhead and just let
inode eviction deal with them.

Put check_move_unevictable_page() and scan_mapping_unevictable_pages()
under CONFIG_SHMEM (with stub for the TINY case when ramfs is used),
more as comment than to save space; comment them used for SHM_UNLOCK.

Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michel Lespinasse <walken@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>