cascardo/linux.git
9 years agowil6210: fix access after free in wil_pcie_remove()
Vladimir Kondratiev [Wed, 6 Aug 2014 07:32:01 +0000 (10:32 +0300)]
wil6210: fix access after free in wil_pcie_remove()

'wil'  released in wil_if_free(); save iomapped address aside to
properly unmap it.

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: cfg80211_rx_mgmt to use GFP_ATOMIC
Vladimir Kondratiev [Wed, 6 Aug 2014 07:32:00 +0000 (10:32 +0300)]
wil6210: cfg80211_rx_mgmt to use GFP_ATOMIC

Internally, cfg80211_rx_mgmt() takes lock:
spin_lock_bh(&wdev->mgmt_registrations_lock);

and therefore one can't use it with GFP_KERNEL allocation flags.
When called with GFP_KERNEL, "sleep while atomic" warning triggered:

<3>[242645.446224] BUG: sleeping function called from invalid context at kernel/mm/slub.c:926
<3>[242645.455516] in_atomic(): 1, irqs_disabled(): 0, pid: 18881, name: kworker/u16:0
<6>[242645.462868] CPU: 0 PID: 18881 Comm: kworker/u16:0 Tainted: G           O 3.10.40-g619a884-00001-g357fa9c #2
<6>[242645.463044] Workqueue: wil6210_wmi wmi_event_worker [wil6210]
<6>[242645.463087] Call trace:
<6>[242645.463157] [<ffffffc000087390>] dump_backtrace+0x0/0x144
<6>[242645.463205] [<ffffffc0000874e4>] show_stack+0x10/0x1c
<6>[242645.463259] [<ffffffc0009abc04>] dump_stack+0x1c/0x28
<6>[242645.463313] [<ffffffc0000c10c0>] __might_sleep+0xe0/0xf0
<6>[242645.463365] [<ffffffc00015a514>] kmem_cache_alloc+0x48/0x1e4
<6>[242645.463414] [<ffffffc00080d468>] __alloc_skb+0x48/0x164
<6>[242645.463466] [<ffffffc00097f2c4>] nl80211_send_mgmt+0x4c/0x1dc
<6>[242645.463511] [<ffffffc000980268>] cfg80211_rx_mgmt+0x1bc/0x25c
<6>[242645.463647] [<ffffffbffc005d64>] wmi_evt_rx_mgmt+0x21c/0x23c [wil6210]
<6>[242645.463783] [<ffffffbffc0073b8>] wmi_event_worker+0x230/0x2f8 [wil6210]
<6>[242645.463838] [<ffffffc0000afffc>] process_one_work+0x258/0x3d0
<6>[242645.463887] [<ffffffc0000b0fe0>] worker_thread+0x1f0/0x310
<6>[242645.463946] [<ffffffc0000b642c>] kthread+0xac/0xb8

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: fix free'd memory access in wil_if_free()
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:59 +0000 (10:31 +0300)]
wil6210: fix free'd memory access in wil_if_free()

In the wil_priv_deinit(), wdev->netdev is accessed, so free_netdev()
should not be called before mentioned call.

Set wdev->netdev to NULL Make sure no more attempts to use it.
It is used for debug printk if not NULL.

This fix kernel panic on module unload and in case error on probe;
if memory allocation debugging enabled.

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: Limit max number of associated stations
Dedy Lansky [Wed, 6 Aug 2014 07:31:58 +0000 (10:31 +0300)]
wil6210: Limit max number of associated stations

Add a module parameter to support setting the maximum
number of stations that can be associated to an AP.

Signed-off-by: Dedy Lansky <qca_dlansky@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: fix false "scan timeout"
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:57 +0000 (10:31 +0300)]
wil6210: fix false "scan timeout"

When sending scan request, if hardware is not ready, scan timer was started
and scan timeout mis-reported.

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: fix beamforming data reporting
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:56 +0000 (10:31 +0300)]
wil6210: fix beamforming data reporting

When reading 'bf' file on debugfs, query beam forming status from firmware.
Ignore CID's that return error or return all zeros.

Remove obsolete code that used to maintain statistics on per-device basis,
as now it is reported be per-CID and current.

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: convert debugfs to the table mode
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:55 +0000 (10:31 +0300)]
wil6210: convert debugfs to the table mode

Define tables for similar file entries, i.e.
- table for debugfs_create_file
- table for static parameters
- table for parameters that are members of "wil" structure
- table for blobs
- table for interrupt controllers

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: Workaround for Sparrow with bad device id
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:54 +0000 (10:31 +0300)]
wil6210: Workaround for Sparrow with bad device id

If firmware broken on the boot time, Sparrow card enumerates as
device 0x302 instead of 0x310. Allow using this ID

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: wait longer for hardware reset completion
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:53 +0000 (10:31 +0300)]
wil6210: wait longer for hardware reset completion

Typical time for hardware reset, as measured, is about 200..250 msec.
Adjust sleeping to use more relaxed msleep, increase sleep interval
and increase total sleep time to be about 1sec

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: check error in wil_target_reset()
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:52 +0000 (10:31 +0300)]
wil6210: check error in wil_target_reset()

When resetting target in wil_target_reset(), error may occur.
Indicate error via error code. In case of error, don't attempt to further
interact with card, specifically don't attempt firmware download or
interrupts enabling.

Move wil_rx_fini() to be always executed to prevent memory leak.

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: update copyright year 2014
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:51 +0000 (10:31 +0300)]
wil6210: update copyright year 2014

Fix Copyright headers in all files changed in 2014, to mention 2014

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: fix race in reset
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:50 +0000 (10:31 +0300)]
wil6210: fix race in reset

It is important to halt USER CPU first, then MAC CPU
Otherwise, race happens in the firmware

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: map MAC timer for packet lifetime into debugfs
Vladimir Kondratiev [Wed, 6 Aug 2014 07:31:49 +0000 (10:31 +0300)]
wil6210: map MAC timer for packet lifetime into debugfs

expose reading RGF_MAC_MTRL_COUNTER_0 in debugfs

Signed-off-by: Dedy Lansky <qca_dlansky@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agomwifiex: fix decimal printf format specifiers prefixed with 0x
Hans Wennborg [Wed, 6 Aug 2014 04:42:55 +0000 (21:42 -0700)]
mwifiex: fix decimal printf format specifiers prefixed with 0x

The prefix suggests the number should be printed in hex, so use
the %x specifier to do that.

Found by using regex suggested by Joe Perches.

Signed-off-by: Hans Wennborg <hans@hanshq.net>
Acked-by: Avinash Patil <patila@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agocw1200: Introduce the use of devm_kzalloc
Himangi Saraogi [Sat, 2 Aug 2014 14:52:46 +0000 (20:22 +0530)]
cw1200: Introduce the use of devm_kzalloc

This patch introduces the use of devm_kzalloc and does away with the
kfrees in the probe and remove functions.

Signed-off-by: Himangi Saraogi <himangi774@gmail.com>
Acked-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agonet: wireless: wl1251: Remove unnecessary free_irq
Himangi Saraogi [Sat, 2 Aug 2014 11:21:30 +0000 (16:51 +0530)]
net: wireless: wl1251: Remove unnecessary free_irq

This patch does away with free_irq in the remove function as the irq is
allocated with a devm_request_irq.

Signed-off-by: Himangi Saraogi <himangi774@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agob43: N-PHY: support setting custom TX power
Rafał Miłecki [Thu, 31 Jul 2014 19:59:48 +0000 (21:59 +0200)]
b43: N-PHY: support setting custom TX power

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agob43: implement PPR (Power Per Rate) management/API
Rafał Miłecki [Thu, 31 Jul 2014 19:59:47 +0000 (21:59 +0200)]
b43: implement PPR (Power Per Rate) management/API

Broadcom hardware supports auto-adjustment of TX power depending on the
currently used rate. So far all calculations were handled without any
helpers (API) using big arrays and magic offsets.
It seems Broadcom recently decided to clean this up by developing PPR.
Their wlc_ppr.h can be found in open parts of the SDK.
As we plan to implement support for rate-based TX power it makes sense
to also implement our version of PPR as well.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agob43: N-PHY: add RF power tables for radio 0x2057 revs 9 & 14
Rafał Miłecki [Thu, 31 Jul 2014 19:59:46 +0000 (21:59 +0200)]
b43: N-PHY: add RF power tables for radio 0x2057 revs 9 & 14

Don't write them as we don't have gains configured correctly yet.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agob43: N-PHY: update rev3+ gain control workarounds
Rafał Miłecki [Thu, 31 Jul 2014 19:59:45 +0000 (21:59 +0200)]
b43: N-PHY: update rev3+ gain control workarounds

This fixes workarounds on rev 6+ and cleans code slightly.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agob43: flush some writes on Broadcom MIPS SoCs
Rafał Miłecki [Thu, 7 Aug 2014 05:45:37 +0000 (07:45 +0200)]
b43: flush some writes on Broadcom MIPS SoCs

Access to PHY and radio registers is indirect on Broadcom hardware and
it seems that addressing on some MIPS SoCs may require flushing. So far
this problem was noticed on 0x4716 SoC only (marketing names: BCM4717,
BCM4718).

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agob43: don't duplicate common PHY read/write ops
Rafał Miłecki [Thu, 31 Jul 2014 19:59:43 +0000 (21:59 +0200)]
b43: don't duplicate common PHY read/write ops

Most of the PHYs use the same way of accessing registers, so move that
code to the shared place. An exception is G-PHY which sometimes access
A-PHY regs and requires special handling.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agob43: update flushing many writes performed in a row
Rafał Miłecki [Thu, 31 Jul 2014 19:59:42 +0000 (21:59 +0200)]
b43: update flushing many writes performed in a row

Flush radio writes as well and add some tiny optimizations (e.g.
masksetting PHY reg involves reading it, so reset the counter).

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoMerge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetoot...
John W. Linville [Mon, 25 Aug 2014 19:58:02 +0000 (15:58 -0400)]
Merge branch 'for-upstream' of git://git./linux/kernel/git/bluetooth/bluetooth-next

9 years agoath10k: flush hif buffers before recovery
Michal Kazior [Fri, 22 Aug 2014 12:33:18 +0000 (14:33 +0200)]
ath10k: flush hif buffers before recovery

Transport buffers weren't flushed and processed
before queueing hw recovery request to mac80211.

This could in theory result in an unwanted htt/wmi
rx events being processed while mac80211 recovers
the device and possibly interfere or even crash
the system.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: remove ar_pci->started
Michal Kazior [Fri, 22 Aug 2014 12:33:17 +0000 (14:33 +0200)]
ath10k: remove ar_pci->started

There are basically no more uses for
ar_pci->started. It is also perfectly safe to call
hif_stop without hif_start now.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: ignore ar_pci->started in pipe cleanup
Michal Kazior [Fri, 22 Aug 2014 12:33:16 +0000 (14:33 +0200)]
ath10k: ignore ar_pci->started in pipe cleanup

Structures used by these functions are now
guaranteed to remain accessible until driver is
unregistered.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: update comment regarding warm reset
Michal Kazior [Fri, 22 Aug 2014 12:33:15 +0000 (14:33 +0200)]
ath10k: update comment regarding warm reset

The old comment was a little out of date. HTT Rx
ring is a more relevant problem when stopping
transport layer.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: rework posting pci rx buffers
Michal Kazior [Fri, 22 Aug 2014 12:33:14 +0000 (14:33 +0200)]
ath10k: rework posting pci rx buffers

It was possible on a host system running low on
memory to end up with no rx buffers on pci pipes.

This makes the driver more robust as it won't fail
to start if it can't allocate all rx buffers right
away. If it is fatal then upper layers will notice
trouble anyway.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: remove early irq handling
Michal Kazior [Fri, 22 Aug 2014 12:23:34 +0000 (14:23 +0200)]
ath10k: remove early irq handling

It's not really necessary to have a dedicated irq
handler just for the sake of catching early fw
crashes anymore. It is now safe to use one handler
even during early stages of device boot up.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: make sure to really disable irqs
Michal Kazior [Fri, 22 Aug 2014 12:23:33 +0000 (14:23 +0200)]
ath10k: make sure to really disable irqs

This fixes two corner cases.

One is a race between disabling copy engine
interrupts and unhandled pending interrupts on the
host. This could end up with a runaway tasklet and
consequently memory leak of a few copy engine
rx buffers.

The other one is an unexpected (and non-maskable
via device CSR) MSI fw indication interrupt during
teardown. This could trigger the same problem as
the first corner case.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: split ce irq/handler setup
Michal Kazior [Fri, 22 Aug 2014 12:23:32 +0000 (14:23 +0200)]
ath10k: split ce irq/handler setup

It doesn't make much sense to overwrite send_cb
and recv_cb callbacks over and over again whenever
transport starts. Just make sure to unmask copy
engine interrupts when starting.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: setup irq method in probe
Michal Kazior [Fri, 22 Aug 2014 12:23:31 +0000 (14:23 +0200)]
ath10k: setup irq method in probe

It doesn't make sense to re-init irqs completely
whenever transport is started/stopped. Do it just
once upon probing/removing.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: fix legacy irq workaround
Michal Kazior [Fri, 22 Aug 2014 12:23:30 +0000 (14:23 +0200)]
ath10k: fix legacy irq workaround

Wrong register was being set up. This could
prevent firmware from booting in some rare cases
when using legacy interrupts.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: move fw init print
Michal Kazior [Fri, 22 Aug 2014 12:23:29 +0000 (14:23 +0200)]
ath10k: move fw init print

Firmware probing is done only once when driver is
registered and firmware version is guaranteed to
remain the same until driver is unregistered.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: print more driver info when firmware crashes
Kalle Valo [Mon, 25 Aug 2014 05:37:45 +0000 (08:37 +0300)]
ath10k: print more driver info when firmware crashes

Sometimes users forget to include important info like firmware version,
so better to print all the info.

Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: rename ath10k_pci_hif_dump_area() to ath10k_pci_fw_crashed_dump()
Kalle Valo [Mon, 25 Aug 2014 05:37:37 +0000 (08:37 +0300)]
ath10k: rename ath10k_pci_hif_dump_area() to ath10k_pci_fw_crashed_dump()

Better to have a clear name for the function. While at it, clear up the title
for the register dump.

Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: provide firmware crash info via debugfs
Ben Greear [Mon, 25 Aug 2014 05:37:32 +0000 (08:37 +0300)]
ath10k: provide firmware crash info via debugfs

Store the firmware registers and other relevant data to a firmware crash dump
file and provide it to user-space via debugfs. Should help with figuring out
why the firmware crashed.

kvalo: remove dbglog support, rework and refactor the code to avoid ifdefs and
otherwise simplify it as well

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: add ath10k_pci_diag_* helpers
Kalle Valo [Mon, 25 Aug 2014 05:37:26 +0000 (08:37 +0300)]
ath10k: add ath10k_pci_diag_* helpers

ath10k_pci_diag_read32() is for reading u32 from a device and ath10k_pci_diag_read_hi()
is a helper for reading data using "host interest" table.

Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: fix a conflict bug in wmi service bitmap
Michal Kazior [Tue, 19 Aug 2014 11:22:32 +0000 (13:22 +0200)]
ath10k: fix a conflict bug in wmi service bitmap

Service mapping for main firmware branch was incorrectly used for 10.x firmware
and vice-versa.  This caused wmi_services in debugfs to print wrong values.

This fixes commit cff990ce7ddd6 ("ath10k: fix wmi service bitmap debug") where
for some reason there was either a conflict that wasn't resolved properly or
git had a bad day.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath6kl: fix %d confusingly prefixed with 0x in format strings
Hans Wennborg [Sun, 17 Aug 2014 21:57:11 +0000 (14:57 -0700)]
ath6kl: fix %d confusingly prefixed with 0x in format strings

Signed-off-by: Hans Wennborg <hans@hanshq.net>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agomac80211: fix channel switch for chanctx-based drivers
Michal Kazior [Mon, 18 Aug 2014 11:19:09 +0000 (13:19 +0200)]
mac80211: fix channel switch for chanctx-based drivers

The new_ctx pointer is set only for non-chanctx drivers.  This yielded a
crash for chanctx-based drivers during channel switch finalization:

  BUG: unable to handle kernel NULL pointer dereference at 0000000000000020
  IP: ieee80211_vif_use_reserved_switch+0x71c/0xb00 [mac80211]

Use an adequate chanctx pointer to fix this.

Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Fri, 22 Aug 2014 21:33:18 +0000 (14:33 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:
 "Here are some bug fixes that have piled up during ksummit/linuxcon.

   1) Fix endian problems in ibmveth, from Anton Blanchard.

   2) IPV6 routing code does GFP_KERNEL allocation in atomic, fix from
      Benjamin Block.

   3) SCTP association fixes from Daniel Borkmann.

   4) When multiple VLAN headers are present we have to make sure the
      second and subsequent ones are pullable in the SKB otherwise we
      blindly dereference garbage.  From Jiri Benc.

   5) The argument adjustment of the signature of hlist_add_after*()
      introduced a regression in the batman-adv code, fix from Sven
      Eckelmann.

   6) Fix TX hang handling to avoid a panic in i40e, from Anjali Singhai
      Jain.

   7) PTP flag test is inverted in i40e driver, from Jesse Brandeburg.

   8) ATM LEC driver needs to hold RTNL mutex over MTU changes, from
      Chas Williams.

   9) Truncate packets larger then the TPACKET_V3 format configured
      buffers, otherwise we overwrite past the end of said buffers.
      From Eric Dumazet.

  10) Fix endianness bugs in qlcnic firmware handling, from Rajesh
      Borundia and Shahed Shaikh.

  11) CXGB4 sometimes doesn't get all of the TX completion events it
      should resulting in SKBs getting stuck in the TX queue, from
      Hariprasad Shenai.

  12) When the FEC chip's PTP clock is disabled, you can't access the
      register.  Add necessary checks to avoid the resulting hang, from
      Fugang Duan"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (37 commits)
  drivers: isdn: eicon: xdi_msg.h: Fix typo in #ifndef
  net: sctp: fix suboptimal edge-case on non-active active/retrans path selection
  net: sctp: spare unnecessary comparison in sctp_trans_elect_best
  net: ethernet: broadcom: bnx2x: Remove redundant #ifdef
  ibmveth: Fix endian issues with rx_no_buffer statistic
  net: xgene: fix possible NULL dereference in xgene_enet_free_desc_rings()
  openvswitch: fix panic with multiple vlan headers
  net: ipv6: fib: don't sleep inside atomic lock
  net: fec: ptp: avoid register access when ipg clock is disabled
  cxgb4: Free completed tx skbs promptly
  cxgb4: Fix race condition in cleanup
  sctp: not send SCTP_PEER_ADDR_CHANGE notifications with failed probe
  bnx2x: Revert UNDI flushing mechanism
  qlcnic: Fix endianess issue in firmware load from file operation
  qlcnic: Fix endianess issue in FW dump template header
  qlcnic: Fix flash access interface to application
  MAINTAINERS: Add section for MRF24J40 IEEE 802.15.4 radio driver
  macvlan: Allow setting multicast filter on all macvlan types
  packet: handle too big packets for PACKET_V3
  MAINTAINERS: add entry for ec_bhf driver
  ...

9 years agodrivers: isdn: eicon: xdi_msg.h: Fix typo in #ifndef
Rasmus Villemoes [Fri, 22 Aug 2014 16:39:14 +0000 (18:39 +0200)]
drivers: isdn: eicon: xdi_msg.h: Fix typo in #ifndef

Test for definedness of the macro which is actually defined (the
change is hard to see: it is s/SSS/SSA/).

Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: sctp: fix suboptimal edge-case on non-active active/retrans path selection
Daniel Borkmann [Fri, 22 Aug 2014 11:03:30 +0000 (13:03 +0200)]
net: sctp: fix suboptimal edge-case on non-active active/retrans path selection

In SCTP, selection of active (T.ACT) and retransmission (T.RET)
transports is being done whenever transport control operations
(UP, DOWN, PF, ...) are engaged through sctp_assoc_control_transport().

Commits 4c47af4d5eb2 ("net: sctp: rework multihoming retransmission
path selection to rfc4960") and a7288c4dd509 ("net: sctp: improve
sctp_select_active_and_retran_path selection") have both improved
it towards a more fine-grained and optimal path selection.

Currently, the selection algorithm for T.ACT and T.RET is as follows:

1) Elect the two most recently used ACTIVE transports T1, T2 for
   T.ACT, T.RET, where T.ACT<-T1 and T1 is most recently used
2) In case primary path T.PRI not in {T1, T2} but ACTIVE, set
   T.ACT<-T.PRI and T.RET<-T1
3) If only T1 is ACTIVE from the set, set T.ACT<-T1 and T.RET<-T1
4) If none is ACTIVE, set T.ACT<-best(T.PRI, T.RET, T3) where
   T3 is the most recently used (if avail) in PF, set T.RET<-T.PRI

Prior to above commits, 4) was simply a camp on T.ACT<-T.PRI and
T.RET<-T.PRI, ignoring possible paths in PF. Camping on T.PRI is
still slightly suboptimal as it can lead to the following scenario:

Setup:
        <A>                                <B>
    T1: p1p1 (10.0.10.10) <==>  .'`)  <==> p1p1 (10.0.10.12)  <= T.PRI
    T2: p1p2 (10.0.10.20) <==> (_ . ) <==> p1p2 (10.0.10.22)

    net.sctp.rto_min = 1000
    net.sctp.path_max_retrans = 2
    net.sctp.pf_retrans = 0
    net.sctp.hb_interval = 1000

T.PRI is permanently down, T2 is put briefly into PF state (e.g. due to
link flapping). Here, the first time transmission is sent over PF path
T2 as it's the only non-INACTIVE path, but the retransmitted data-chunks
are sent over the INACTIVE path T1 (T.PRI), which is not good.

After the patch, it's choosing better transports in both cases by
modifying step 4):

4) If none is ACTIVE, set T.ACT_new<-best(T.ACT_old, T3) where T3 is
   the most recently used (if avail) in PF, set T.RET<-T.ACT_new

This will still select a best possible path in PF if available (which
can also include T.PRI/T.RET), and set both T.ACT/T.RET to it.

In case sctp_assoc_control_transport() *just* put T.ACT_old into INACTIVE
as it transitioned from ACTIVE->PF->INACTIVE and stays in INACTIVE just
for a very short while before going back ACTIVE, it will guarantee that
this path will be reselected for T.ACT/T.RET since T3 (PF) is not
available.

Previously, this was not possible, as we would only select between T.PRI
and T.RET, and a possible T3 would be NULL due to the fact that we have
just transitioned T3 in sctp_assoc_control_transport() from PF->INACTIVE
and would select a suboptimal path when T.PRI/T.RET have worse properties.

In the case that T.ACT_old permanently went to INACTIVE during this
transition and there's no PF path available, plus T.PRI and T.RET are
INACTIVE as well, we would now camp on T.ACT_old, but if everything is
being INACTIVE there's really not much we can do except hoping for a
successful HB to bring one of the transports back up again and, thus
cause a new selection through sctp_assoc_control_transport().

Now both tests work fine:

Case 1:

 1. T1 S(ACTIVE) T.ACT
    T2 S(ACTIVE) T.RET

 2. T1 S(ACTIVE) T.ACT, T.RET
    T2 S(PF)

 3. T1 S(ACTIVE) T.ACT, T.RET
    T2 S(INACTIVE)

 5. T1 S(PF) T.ACT, T.RET
    T2 S(INACTIVE)

[ 5.1 T1 S(INACTIVE) T.ACT, T.RET
      T2 S(INACTIVE) ]

 6. T1 S(ACTIVE) T.ACT, T.RET
    T2 S(INACTIVE)

 7. T1 S(ACTIVE) T.ACT
    T2 S(ACTIVE) T.RET

Case 2:

 1. T1 S(ACTIVE) T.ACT
    T2 S(ACTIVE) T.RET

 2. T1 S(PF)
    T2 S(ACTIVE) T.ACT, T.RET

 3. T1 S(INACTIVE)
    T2 S(ACTIVE) T.ACT, T.RET

 5. T1 S(INACTIVE)
    T2 S(PF) T.ACT, T.RET

[ 5.1 T1 S(INACTIVE)
      T2 S(INACTIVE) T.ACT, T.RET ]

 6. T1 S(INACTIVE)
    T2 S(ACTIVE) T.ACT, T.RET

 7. T1 S(ACTIVE) T.ACT
    T2 S(ACTIVE) T.RET

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: sctp: spare unnecessary comparison in sctp_trans_elect_best
Daniel Borkmann [Fri, 22 Aug 2014 11:03:29 +0000 (13:03 +0200)]
net: sctp: spare unnecessary comparison in sctp_trans_elect_best

When both transports are the same, we don't have to go down that
road only to realize that we will return the very same transport.
We are guaranteed that curr is always non-NULL. Therefore, just
short-circuit this special case.

Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: ethernet: broadcom: bnx2x: Remove redundant #ifdef
Rasmus Villemoes [Wed, 20 Aug 2014 13:14:49 +0000 (15:14 +0200)]
net: ethernet: broadcom: bnx2x: Remove redundant #ifdef

Nothing defines _ASM_GENERIC_INT_L64_H, it is a weird way to check for
64 bit longs, and u64 should be printed using %llx anyway.

Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoibmveth: Fix endian issues with rx_no_buffer statistic
Anton Blanchard [Fri, 22 Aug 2014 01:36:52 +0000 (11:36 +1000)]
ibmveth: Fix endian issues with rx_no_buffer statistic

Hidden away in the last 8 bytes of the buffer_list page is a solitary
statistic. It needs to be byte swapped or else ethtool -S will
produce numbers that terrify the user.

Since we do this in multiple places, create a helper function with a
comment explaining what is going on.

Signed-off-by: Anton Blanchard <anton@samba.org>
Cc: stable@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: xgene: fix possible NULL dereference in xgene_enet_free_desc_rings()
Iyappan Subramanian [Thu, 21 Aug 2014 20:44:48 +0000 (13:44 -0700)]
net: xgene: fix possible NULL dereference in xgene_enet_free_desc_rings()

A NULL pointer dereference is possible for the argument ring->buf_pool
which is passed to xgene_enet_free_desc_ring(), as ring could be NULL.

And now since NULL pointers are being checked for before the calls to
xgene_enet_free_desc_ring(), might as well take advantage of them and
not call the function if the argument would be NULL.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Iyappan Subramanian <isubramanian@apm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoopenvswitch: fix panic with multiple vlan headers
Jiri Benc [Thu, 21 Aug 2014 19:33:44 +0000 (21:33 +0200)]
openvswitch: fix panic with multiple vlan headers

When there are multiple vlan headers present in a received frame, the first
one is put into vlan_tci and protocol is set to ETH_P_8021Q. Anything in the
skb beyond the VLAN TPID may be still non-linear, including the inner TCI
and ethertype. While ovs_flow_extract takes care of IP and IPv6 headers, it
does nothing with ETH_P_8021Q. Later, if OVS_ACTION_ATTR_POP_VLAN is
executed, __pop_vlan_tci pulls the next vlan header into vlan_tci.

This leads to two things:

1. Part of the resulting ethernet header is in the non-linear part of the
   skb. When eth_type_trans is called later as the result of
   OVS_ACTION_ATTR_OUTPUT, kernel BUGs in __skb_pull. Also, __pop_vlan_tci
   is in fact accessing random data when it reads past the TPID.

2. network_header points into the ethernet header instead of behind it.
   mac_len is set to a wrong value (10), too.

Reported-by: Yulong Pei <ypei@redhat.com>
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: ipv6: fib: don't sleep inside atomic lock
Benjamin Block [Thu, 21 Aug 2014 17:37:48 +0000 (19:37 +0200)]
net: ipv6: fib: don't sleep inside atomic lock

The function fib6_commit_metrics() allocates a piece of memory in mode
GFP_KERNEL while holding an atomic lock from higher up in the stack, in
the function __ip6_ins_rt(). This produces the following BUG:

> BUG: sleeping function called from invalid context at mm/slub.c:1250
> in_atomic(): 1, irqs_disabled(): 0, pid: 2909, name: dhcpcd
> 2 locks held by dhcpcd/2909:
>  #0:  (rtnl_mutex){+.+.+.}, at: [<ffffffff81978e67>] rtnl_lock+0x17/0x20
>  #1:  (&tb->tb6_lock){++--+.}, at: [<ffffffff81a6951a>] ip6_route_add+0x65a/0x800
> CPU: 1 PID: 2909 Comm: dhcpcd Not tainted 3.17.0-rc1 #1
> Hardware name: ASUS All Series/Q87T, BIOS 0216 10/16/2013
>  0000000000000008 ffff8800c8f13858 ffffffff81af135a 0000000000000000
>  ffff880212202430 ffff8800c8f13878 ffffffff810f8d3a ffff880212202c98
>  0000000000000010 ffff8800c8f138c8 ffffffff8121ad0e 0000000000000001
> Call Trace:
>  [<ffffffff81af135a>] dump_stack+0x4e/0x68
>  [<ffffffff810f8d3a>] __might_sleep+0x10a/0x120
>  [<ffffffff8121ad0e>] kmem_cache_alloc_trace+0x4e/0x190
>  [<ffffffff81a6bcd6>] ? fib6_commit_metrics+0x66/0x110
>  [<ffffffff81a6bcd6>] fib6_commit_metrics+0x66/0x110
>  [<ffffffff81a6cbf3>] fib6_add+0x883/0xa80
>  [<ffffffff81a6951a>] ? ip6_route_add+0x65a/0x800
>  [<ffffffff81a69535>] ip6_route_add+0x675/0x800
>  [<ffffffff81a68f2a>] ? ip6_route_add+0x6a/0x800
>  [<ffffffff81a6990c>] inet6_rtm_newroute+0x5c/0x80
>  [<ffffffff8197cf01>] rtnetlink_rcv_msg+0x211/0x260
>  [<ffffffff81978e67>] ? rtnl_lock+0x17/0x20
>  [<ffffffff81119708>] ? lock_release_holdtime+0x28/0x180
>  [<ffffffff81978e67>] ? rtnl_lock+0x17/0x20
>  [<ffffffff8197ccf0>] ? __rtnl_unlock+0x20/0x20
>  [<ffffffff819a989e>] netlink_rcv_skb+0x6e/0xd0
>  [<ffffffff81978ee5>] rtnetlink_rcv+0x25/0x40
>  [<ffffffff819a8e59>] netlink_unicast+0xd9/0x180
>  [<ffffffff819a9600>] netlink_sendmsg+0x700/0x770
>  [<ffffffff81103735>] ? local_clock+0x25/0x30
>  [<ffffffff8194e83c>] sock_sendmsg+0x6c/0x90
>  [<ffffffff811f98e3>] ? might_fault+0xa3/0xb0
>  [<ffffffff8195ca6d>] ? verify_iovec+0x7d/0xf0
>  [<ffffffff8194ec3e>] ___sys_sendmsg+0x37e/0x3b0
>  [<ffffffff8111ef15>] ? trace_hardirqs_on_caller+0x185/0x220
>  [<ffffffff81af979e>] ? mutex_unlock+0xe/0x10
>  [<ffffffff819a55ec>] ? netlink_insert+0xbc/0xe0
>  [<ffffffff819a65e5>] ? netlink_autobind.isra.30+0x125/0x150
>  [<ffffffff819a6520>] ? netlink_autobind.isra.30+0x60/0x150
>  [<ffffffff819a84f9>] ? netlink_bind+0x159/0x230
>  [<ffffffff811f989a>] ? might_fault+0x5a/0xb0
>  [<ffffffff8194f25e>] ? SYSC_bind+0x7e/0xd0
>  [<ffffffff8194f8cd>] __sys_sendmsg+0x4d/0x80
>  [<ffffffff8194f912>] SyS_sendmsg+0x12/0x20
>  [<ffffffff81afc692>] system_call_fastpath+0x16/0x1b

Fixing this by replacing the mode GFP_KERNEL with GFP_ATOMIC.

Signed-off-by: Benjamin Block <bebl@mageta.org>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: fec: ptp: avoid register access when ipg clock is disabled
Nimrod Andy [Thu, 21 Aug 2014 09:09:38 +0000 (17:09 +0800)]
net: fec: ptp: avoid register access when ipg clock is disabled

The current kernel hang on i.MX6SX with rootfs mount from MMC.
The root cause is that ptp uses a periodic timer to access enet register
even if ipg clock is disabled.

FEC ptp driver start one period timer to read 1588 counter register in the
ptp init function that is called after FEC driver is probed.

To save power, after FEC probe finish, FEC driver disable all clocks including
ipg clock that is needed for register access.

i.MX5x, i.MX6q/dl/sl FEC register access don't cause system hang when ipg clock
is disabled, just return zero value. But for i.MX6sx SOC, it cause system hang.

To avoid the issue, we need to check ptp clock status before ptp timer count access.

Signed-off-by: Fugang Duan <B38611@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Fri, 22 Aug 2014 16:08:20 +0000 (09:08 -0700)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Will Deacon:
 "This small set of fixes addresses a few issues introduced during the
  merge window, including:

   - fix typo in I-cache detection that was causing us to treat all
     I-caches as aliasing
   - hook up memfd_create and getrandom syscalls for native and compat
   - revert a temporary hack for defconfig builds in -next (the audit
     tree changes didn't make it in this merge window)
   - a couple of UEFI fixes for TEXT_OFFSET fuzzing and /memreserve/
   - a simple sparsemem fix for 48-bit physical addressing
   - small defconfig updates to get autotesters working with X-gene"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  Revert "arm64: Do not invoke audit_syscall_* functions if !CONFIG_AUDIT_SYSCALL"
  arm64: mm: update max pa bits to 48
  arm64: ignore DT memreserve entries when booting in UEFI mode
  arm64: configs: Enable X-Gene SATA and ethernet in defconfig
  arm64: align randomized TEXT_OFFSET on 4 kB boundary
  asm-generic: add memfd_create system call to unistd.h
  arm64: compat: wire up memfd_create and getrandom syscalls for aarch32
  arm64: fix typo in I-cache policy detection

9 years agoMerge tag 'iommu-fixes-v3.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 22 Aug 2014 16:06:22 +0000 (09:06 -0700)]
Merge tag 'iommu-fixes-v3.17-rc1' of git://git./linux/kernel/git/joro/iommu

Pull IOMMU fixes from Joerg Roedel:
 "The fixes include:

   - fix a crash in the VT-d driver when devices with a driver attached
     are hot-unplugged

   - fix a AMD IOMMU driver crash with device assignment of 32 bit PCI
     devices to KVM guests

   - fix for a copy&paste error in generic IOMMU code.  Now the right
     function pointer is checked before calling"

* tag 'iommu-fixes-v3.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/core: Check for the right function pointer in iommu_map()
  iommu/amd: Fix cleanup_domain for mass device removal
  iommu/vt-d: Defer domain removal if device is assigned to a driver

9 years agocxgb4: Free completed tx skbs promptly
Hariprasad Shenai [Thu, 21 Aug 2014 11:34:46 +0000 (17:04 +0530)]
cxgb4: Free completed tx skbs promptly

Description of problem:
The NIC card is not reporting back to the driver the transmitted skbs,
so they get stuck in the TX ring causing issues with reference
counters in other kernel components.

Developed a new Automatic Egress Queue Update firmware facility to slowly tick
through Egress Queues and send back any outstanding CIDX Updates which are
laying around.

Based on original work by Casey Leedom <leedom@chelsio.com>

Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge tag 'linux-can-fixes-for-3.17-20140821' of git://gitorious.org/linux-can/linux-can
David S. Miller [Fri, 22 Aug 2014 04:53:15 +0000 (21:53 -0700)]
Merge tag 'linux-can-fixes-for-3.17-20140821' of git://gitorious.org/linux-can/linux-can

Marc Kleine-Budde says:

====================
pull-request: can 2014-08-21

The first patch is from Mirza Krak, it fixes the initialization of the hardware
in the sja1000 driver. The next patch is contributed by Dan Carpenter, it fixes
the error handling in the c_can's probe function. Then there are two patches
for the flexcan driver, one by Alexander Stein, which fixes the resetting of
the bus error interrupt mask, the other one by Sebastian Andrzej Siewior which
adds an additional error state transition message.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agocxgb4: Fix race condition in cleanup
Anish Bhatt [Wed, 20 Aug 2014 20:44:06 +0000 (13:44 -0700)]
cxgb4: Fix race condition in cleanup

There is a possible race condition when we unregister the PCI Driver and then
flush/destroy the global "workq". This could lead to situations where there
are tasks on the Work Queue with references to now deleted adapter data
structures. Instead, have per-adapter Work Queues which were instantiated and
torn down in init_one() and remove_one(), respectively.

v2: Remove unnecessary call to flush_workqueue() before destroy_workqueue()

Signed-off-by: Anish Bhatt <anish@chelsio.com>
Signed-off-by: Casey Leedom <leedom@chelsio.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agosctp: not send SCTP_PEER_ADDR_CHANGE notifications with failed probe
zhuyj [Wed, 20 Aug 2014 09:31:43 +0000 (17:31 +0800)]
sctp: not send SCTP_PEER_ADDR_CHANGE notifications with failed probe

Since the transport has always been in state SCTP_UNCONFIRMED, it
therefore wasn't active before and hasn't been used before, and it
always has been, so it is unnecessary to bug the user with a
notification.

Reported-by: Deepak Khandelwal <khandelwal.deepak.1987@gmail.com>
Suggested-by: Vlad Yasevich <vyasevich@gmail.com>
Suggested-by: Michael Tuexen <tuexen@fh-muenster.de>
Suggested-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: Zhu Yanjun <Yanjun.Zhu@windriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Acked-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agobnx2x: Revert UNDI flushing mechanism
Yuval Mintz [Mon, 18 Aug 2014 19:36:23 +0000 (22:36 +0300)]
bnx2x: Revert UNDI flushing mechanism

Commit 91ebb929b6f8 ("bnx2x: Add support for Multi-Function UNDI") [which was
later supposedly fixed by de682941eef3 ("bnx2x: Fix UNDI driver unload")]
introduced a bug in which in some [yet-to-be-determined] scenarios the
alternative flushing mechanism which was to guarantee the Rx buffers are
empty before resetting them during device probe will fail.
If this happens, when device will be loaded once more a fatal attention will
occur; Since this most likely happens in boot from SAN scenarios, the machine
will fail to load.

Notice this may occur not only in the 'Multi-Function' scenario but in the
regular scenario as well, i.e., this introduced a regression in the driver's
ability to perform boot from SAN.

The patch reverts the mechanism and applies the old scheme to multi-function
devices as well as to single-function devices.

Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'qlcnic'
David S. Miller [Fri, 22 Aug 2014 00:43:20 +0000 (17:43 -0700)]
Merge branch 'qlcnic'

Shahed Shaikh says:

====================
qlcnic: Bug fixes

This series fixes some bugs related to endianess.

Please apply this series to net.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoqlcnic: Fix endianess issue in firmware load from file operation
Shahed Shaikh [Mon, 18 Aug 2014 13:31:55 +0000 (09:31 -0400)]
qlcnic: Fix endianess issue in firmware load from file operation

Firmware binary file is in little endian. On big-endian architecture, while
writing this binary FW file to adapters memory, writel() swaps the data resulting into
corruption of FW image. So, swap the data before writing into adapters memory.

Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoqlcnic: Fix endianess issue in FW dump template header
Rajesh Borundia [Mon, 18 Aug 2014 13:31:54 +0000 (09:31 -0400)]
qlcnic: Fix endianess issue in FW dump template header

Firmware dump template header is read from adapter using
readl() which swaps the data. So, adjust structure
element on the boundary of 32bit dword.

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>
9 years agoqlcnic: Fix flash access interface to application
Jitendra Kalsaria [Mon, 18 Aug 2014 13:31:53 +0000 (09:31 -0400)]
qlcnic: Fix flash access interface to application

Application expects flash data in little endian, but driver reads/writes
flash data using readl()/writel() APIs which swaps data on big endian machine.
So, swap the data after reading from and before writing to flash memory.

Signed-off-by: Jitendra Kalsaria <jitendra.kalsaria@qlogic.com>
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMAINTAINERS: Add section for MRF24J40 IEEE 802.15.4 radio driver
Alan Ott [Sat, 16 Aug 2014 21:09:03 +0000 (17:09 -0400)]
MAINTAINERS: Add section for MRF24J40 IEEE 802.15.4 radio driver

Alan is the original author of the driver. This change was discussed
with the 802.15.4 subsystem maintainer, Alexander Aring.

Signed-off-by: Alan Ott <alan@signal11.us>
Acked-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agomacvlan: Allow setting multicast filter on all macvlan types
Vlad Yasevich [Fri, 15 Aug 2014 17:04:59 +0000 (13:04 -0400)]
macvlan: Allow setting multicast filter on all macvlan types

Currently, macvlan code restricts multicast and unicast
filter setting only to passthru devices.  As a result,
if a guest using macvtap wants to receive multicast
traffic, it has to set IFF_ALLMULTI or IFF_PROMISC.

This patch makes it possible to use the fdb interface
to add multicast addresses to the filter thus allowing
a guest to receive only targeted multicast traffic.

CC: John Fastabend <john.r.fastabend@intel.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Jason Wang <jasowang@redhat.com>
Signed-off-by: Vladislav Yasevich <vyasevic@redhat.com>
Acked-by: John Fastabend <john.r.fastabend@intel.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agopacket: handle too big packets for PACKET_V3
Eric Dumazet [Fri, 15 Aug 2014 16:16:04 +0000 (09:16 -0700)]
packet: handle too big packets for PACKET_V3

af_packet can currently overwrite kernel memory by out of bound
accesses, because it assumed a [new] block can always hold one frame.

This is not generally the case, even if most existing tools do it right.

This patch clamps too long frames as API permits, and issue a one time
error on syslog.

[  394.357639] tpacket_rcv: packet too big, clamped from 5042 to 3966. macoff=82

In this example, packet header tp_snaplen was set to 3966,
and tp_len was set to 5042 (skb->len)

Signed-off-by: Eric Dumazet <edumazet@google.com>
Fixes: f6fb8f100b80 ("af-packet: TPACKET_V3 flexible buffer implementation.")
Acked-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMAINTAINERS: add entry for ec_bhf driver
Dariusz Marcinkiewicz [Fri, 15 Aug 2014 15:50:47 +0000 (17:50 +0200)]
MAINTAINERS: add entry for ec_bhf driver

Added entry for ec_bhf driver.

Signed-off-by: Dariusz Marcinkiewicz <reksio@newterm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agolec: Use rtnl lock/unlock when updating MTU
chas williams - CONTRACTOR [Thu, 14 Aug 2014 13:19:47 +0000 (09:19 -0400)]
lec: Use rtnl lock/unlock when updating MTU

The LECS response contains the MTU that should be used.  Correctly
synchronize with other layers when updating.

Signed-off-by: Chas Williams - CONTRACTOR <chas@cmf.nrl.navy.mil>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'for-3.17-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj...
Linus Torvalds [Thu, 21 Aug 2014 21:26:27 +0000 (14:26 -0700)]
Merge branch 'for-3.17-fixes' of git://git./linux/kernel/git/tj/libata

Pull libata fixes from Tejun Heo:
 "Nothing drastic but pushing out early due to build breakage in the new
  tegra platform.

  Additionally:

   - M550 tagged trim blacklist pattern is widened so that it matches
     the new 1TB model

   - three controller specific fixes"

* 'for-3.17-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata:
  libata: widen Crucial M550 blacklist matching
  pata_scc: propagate return value of scc_wait_after_reset
  ata: ahci_tegra: Change include to fix compilation
  pata_samsung_cf: change ret type to signed
  ahci_xgene: Removing NCQ support from the APM X-Gene SoC AHCI SATA Host Controller driver.

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Thu, 21 Aug 2014 21:25:20 +0000 (14:25 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

Pull HID fixes from Jiri Kosina:

 - fixes for a couple potential memory corruption problems (the HW would
   have to be manufactured to be deliberately evil to trigger those)
   found by Ben Hawkes
 - fix for potential infinite loop when using sysfs interface of
   logitech driver, from Simon Wood
 - a couple more simple driver fixes

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: fix a couple of off-by-ones
  HID: logitech: perform bounds checking on device_id early enough
  HID: logitech: fix bounds checking on LED report size
  HID: logitech: Prevent possibility of infinite loop when using /sys interface
  HID: rmi: print an error if F11 is not found instead of stopping the device
  HID: hid-sensor-hub: use devm_ functions consistently
  HID: huion: Use allocated buffer for DMA
  HID: huion: Fail on parameter retrieval errors

9 years agoMerge tag 'sound-3.17-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Thu, 21 Aug 2014 21:24:40 +0000 (14:24 -0700)]
Merge tag 'sound-3.17-rc2' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "A bunch of ASoC fixes with a few HD-audio fixes in this pull request.

  All fairly small, boring and device-specific fixes, in addition to
  MAINTAINERS update for better reviewing"

* tag 'sound-3.17-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda/hdmi - apply Valleyview fix-ups to Cherryview display codec
  ALSA: hda/hdmi - set depop_delay for haswell plus
  ALSA: hda - restore the gpio led after resume
  ALSA: hda/realtek - Avoid setting wrong COEF on ALC269 & co
  ASoC: pxa-ssp: drop SNDRV_PCM_FMTBIT_S24_LE
  ASoC: fsl-esai: Revert .xlate_tdm_slot_mask() support
  ASoC: mcasp: Fix implicit BLCK divider setting
  ASoC: arizona: Fix TDM slot length handling in arizona_hw_params
  ASoC: pcm512x: Correct Digital Playback control names
  ASoC: dapm: Fix uninitialized variable in snd_soc_dapm_get_enum_double()
  ASoC: Intel: Restore Baytrail ADSP streams only when ADSP was in reset
  ASoC: Intel: Wait Baytrail ADSP boot at resume_early stage
  ASoC: Intel: Merge Baytrail ADSP suspend_noirq into suspend_late
  MAINTAINERS: Add i.MX maintainers and paths to Freescale ASoC entry
  ASoC: Intel: Update Baytrail ADSP firmware name

9 years agoMerge branch 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Thu, 21 Aug 2014 21:07:44 +0000 (14:07 -0700)]
Merge branch 'i2c/for-next' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "Here is the fixup for the 'lowlight' of my last pull request.  I2C is
  not selected anymore by I2C_ACPI.  Instead, the code in question now
  depends on I2C=y.

  Also, Mika has agreed to support me and be the maintainer for I2C-ACPI
  related patches.  Finally, a new-ID-patch came along last week"

* 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  MAINTAINERS: add maintainer for ACPI parts of I2C
  i2c: i801: Add PCI ID for Intel Braswell
  i2c: rework kernel config I2C_ACPI

9 years agoMerge tag 'please-pull-memfd_create' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 21 Aug 2014 21:06:56 +0000 (14:06 -0700)]
Merge tag 'please-pull-memfd_create' of git://git./linux/kernel/git/aegl/linux

Pull ia64 update from Tony Luck:
 "Add memfd_create syscall to ia64"

* tag 'please-pull-memfd_create' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux:
  [IA64] Wire up memfd_create() system call

9 years agoMerge tag 'microblaze-3.17-rc2' of git://git.monstr.eu/linux-2.6-microblaze
Linus Torvalds [Thu, 21 Aug 2014 21:06:18 +0000 (14:06 -0700)]
Merge tag 'microblaze-3.17-rc2' of git://git.monstr.eu/linux-2.6-microblaze

Pull microblaze update from Michal Simek:
 "Wire-up seccomp/getrandom/memfd_create syscalls"

* tag 'microblaze-3.17-rc2' of git://git.monstr.eu/linux-2.6-microblaze:
  microblaze: Wire-up memfd_create syscall
  microblaze: Wire-up getrandom syscall
  microblaze: Wire-up seccomp syscall

9 years agoHID: fix a couple of off-by-ones
Jiri Kosina [Thu, 21 Aug 2014 14:57:48 +0000 (09:57 -0500)]
HID: fix a couple of off-by-ones

There are a few very theoretical off-by-one bugs in report descriptor size
checking when performing a pre-parsing fixup. Fix those.

Cc: stable@vger.kernel.org
Reported-by: Ben Hawkes <hawkes@google.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agoHID: logitech: perform bounds checking on device_id early enough
Jiri Kosina [Thu, 21 Aug 2014 14:57:17 +0000 (09:57 -0500)]
HID: logitech: perform bounds checking on device_id early enough

device_index is a char type and the size of paired_dj_deivces is 7
elements, therefore proper bounds checking has to be applied to
device_index before it is used.

We are currently performing the bounds checking in
logi_dj_recv_add_djhid_device(), which is too late, as malicious device
could send REPORT_TYPE_NOTIF_DEVICE_UNPAIRED early enough and trigger the
problem in one of the report forwarding functions called from
logi_dj_raw_event().

Fix this by performing the check at the earliest possible ocasion in
logi_dj_raw_event().

Cc: stable@vger.kernel.org
Reported-by: Ben Hawkes <hawkes@google.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agoHID: logitech: fix bounds checking on LED report size
Jiri Kosina [Thu, 21 Aug 2014 14:56:47 +0000 (09:56 -0500)]
HID: logitech: fix bounds checking on LED report size

The check on report size for REPORT_TYPE_LEDS in logi_dj_ll_raw_request()
is wrong; the current check doesn't make any sense -- the report allocated
by HID core in hid_hw_raw_request() can be much larger than
DJREPORT_SHORT_LENGTH, and currently logi_dj_ll_raw_request() doesn't
handle this properly at all.

Fix the check by actually trimming down the report size properly if it is
too large.

Cc: stable@vger.kernel.org
Reported-by: Ben Hawkes <hawkes@google.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agocan: flexcan: handle state passive -> warning transition
Sebastian Andrzej Siewior [Fri, 25 Jul 2014 18:16:40 +0000 (20:16 +0200)]
can: flexcan: handle state passive -> warning transition

Once the CAN-bus is open and a packet is sent, the controller switches
into the PASSIVE state. Once the BUS is closed again it goes the back
err-warning. The TX error counter goes 0 -> 0x80 -> 0x7f.
This patch makes sure that the user learns about this state chang
(CAN_STATE_ERROR_WARNING => CAN_STATE_ERROR_PASSIVE)

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Matthias Klein <matthias.klein@optimeas.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
9 years agocan: flexcan: Disable error interrupt when bus error reporting is disabled
Alexander Stein [Tue, 12 Aug 2014 08:47:21 +0000 (10:47 +0200)]
can: flexcan: Disable error interrupt when bus error reporting is disabled

In case we don't have FLEXCAN_HAS_BROKEN_ERR_STATE and the user set
CAN_CTRLMODE_BERR_REPORTING once it can not be unset again until reboot.
So in case neither hardware nor user wants the error interrupt disable
the bit.

Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
9 years agocan: c_can: checking IS_ERR() instead of NULL
Dan Carpenter [Fri, 1 Aug 2014 08:53:44 +0000 (11:53 +0300)]
can: c_can: checking IS_ERR() instead of NULL

devm_ioremap() returns NULL on error, not an ERR_PTR().

Fixes: 33cf75656923 ('can: c_can_platform: Fix raminit, use devm_ioremap() instead of devm_ioremap_resource()')

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: linux-stable <stable@vger.kernel.org> # >= v3.11
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
9 years agocan: sja1000: Validate initialization state in start method
Mirza Krak [Fri, 8 Aug 2014 12:30:50 +0000 (14:30 +0200)]
can: sja1000: Validate initialization state in start method

When sja1000 is not compiled as module the SJA1000 chip is only
initialized during device registration on kernel boot. Should the chip
get a hardware reset there is no way to reinitialize it without re-
booting the Linux kernel.

This patch adds a check in sja1000_start if the chip is initialized, if
not we initialize it.

Signed-off-by: Mirza Krak <mirza.krak@hostmobility.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
9 years agomicroblaze: Wire-up memfd_create syscall
Michal Simek [Thu, 21 Aug 2014 08:19:28 +0000 (10:19 +0200)]
microblaze: Wire-up memfd_create syscall

Add new memfd_create syscall.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
9 years agomicroblaze: Wire-up getrandom syscall
Michal Simek [Thu, 7 Aug 2014 13:59:05 +0000 (15:59 +0200)]
microblaze: Wire-up getrandom syscall

Add new getrandom syscall.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
9 years agomicroblaze: Wire-up seccomp syscall
Michal Simek [Thu, 7 Aug 2014 13:57:41 +0000 (15:57 +0200)]
microblaze: Wire-up seccomp syscall

Add new seccomp syscall.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
9 years agoMerge branch 'for-linus' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Wed, 20 Aug 2014 23:33:21 +0000 (18:33 -0500)]
Merge branch 'for-linus' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fixes from Steve French:
 "Most important fixes in this set include three SMB3 fixes for stable
  (including fix for possible kernel oops), and a workaround to allow
  writes to Mac servers (only cifs dialect, not more current SMB2.1,
  worked to Mac servers).  Also fallocate support added, and lease fix
  from Jeff"

* 'for-linus' of git://git.samba.org/sfrench/cifs-2.6:
  [SMB3] Enable fallocate -z support for SMB3 mounts
  enable fallocate punch hole ("fallocate -p") for SMB3
  Incorrect error returned on setting file compressed on SMB2
  CIFS: Fix wrong directory attributes after rename
  CIFS: Fix SMB2 readdir error handling
  [CIFS] Possible null ptr deref in SMB2_tcon
  [CIFS] Workaround MacOS server problem with SMB2.1 write  response
  cifs: handle lease F_UNLCK requests properly
  Cleanup sparse file support by creating worker function for it
  Add sparse file support to SMB2/SMB3 mounts
  Add missing definitions for CIFS File System Attributes
  cifs: remove unused function cifs_oplock_break_wait

9 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs
Linus Torvalds [Wed, 20 Aug 2014 23:32:16 +0000 (18:32 -0500)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs

Pull filesystem fixes from Jan Kara:
 "udf, isofs, and ext3 bug fixes"

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  ext3: Count internal journal as bsddf overhead in ext3_statfs
  isofs: Fix unbounded recursion when processing relocated directories
  udf: avoid unneeded up_write when fail to add entry in ->symlink

9 years agoMerge branch 'for_linus' of git://cavan.codon.org.uk/platform-drivers-x86
Linus Torvalds [Wed, 20 Aug 2014 23:23:06 +0000 (18:23 -0500)]
Merge branch 'for_linus' of git://cavan.codon.org.uk/platform-drivers-x86

Pull x86 platform driver revert from Matthew Garrett:
 "This clearly shouldn't have been merged.  No excuse on my part"

* 'for_linus' of git://cavan.codon.org.uk/platform-drivers-x86:
  Revert "platform/x86/toshiba-apci.c possible bad if test?"

9 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Wed, 20 Aug 2014 23:22:10 +0000 (18:22 -0500)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM fixes from Paolo Bonzini:
 "Reverting a 3.16 patch, fixing two bugs in device assignment (one has
  a CVE), and fixing some problems introduced during the merge window
  (the CMA bug came in via Andrew, the x86 ones via yours truly)"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  virt/kvm/assigned-dev.c: Set 'dev->irq_source_id' to '-1' after free it
  Revert "KVM: x86: Increase the number of fixed MTRR regs to 10"
  KVM: x86: do not check CS.DPL against RPL during task switch
  KVM: x86: Avoid emulating instructions on #UD mistakenly
  PC, KVM, CMA: Fix regression caused by wrong get_order() use
  kvm: iommu: fix the third parameter of kvm_iommu_put_pages (CVE-2014-3601)

9 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Wed, 20 Aug 2014 23:20:50 +0000 (18:20 -0500)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "These are the two bug fixes I mentioned in the final merge window
  pull.  One is a reversed logic check in the device busy tests which
  can cause a nasty hang and another crash seen in the new SCSI pool
  support if the use count ever goes to zero"

[ The device busy test already got merged from a patch earlier, so is
  now duplicated.  ]

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  [SCSI] save command pool address of Scsi_Host
  [SCSI] fix qemu boot hang problem

9 years agoRevert "platform/x86/toshiba-apci.c possible bad if test?"
Matthew Garrett [Wed, 20 Aug 2014 15:18:18 +0000 (08:18 -0700)]
Revert "platform/x86/toshiba-apci.c possible bad if test?"

This reverts commit bdc3ae7221213963f438faeaa69c8b4a2195f491.

Signed-off-by: Matthew Garrett <matthew.garrett@nebula.com>
9 years agoext3: Count internal journal as bsddf overhead in ext3_statfs
Chin-Tsung Cheng [Fri, 15 Aug 2014 07:49:31 +0000 (15:49 +0800)]
ext3: Count internal journal as bsddf overhead in ext3_statfs

The journal blocks of external journal device should not
be counted as overhead.

Signed-off-by: Chin-Tsung Cheng <chintzung@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
9 years agoRevert "arm64: Do not invoke audit_syscall_* functions if !CONFIG_AUDIT_SYSCALL"
Will Deacon [Tue, 19 Aug 2014 21:05:45 +0000 (22:05 +0100)]
Revert "arm64: Do not invoke audit_syscall_* functions if !CONFIG_AUDIT_SYSCALL"

For some reason, the audit patches didn't make it out of -next this
merge window, so revert our temporary hack and let the audit guys deal
with fixing up -next.

This reverts commit 2a8f45b040bcb9b2ad2845f061499d1b6f41cc7b.

Signed-off-by: Will Deacon <will.deacon@arm.com>
9 years agoarm64: mm: update max pa bits to 48
Ganapatrao Kulkarni [Thu, 14 Aug 2014 15:19:46 +0000 (20:49 +0530)]
arm64: mm: update max pa bits to 48

Now that we support 48-bit physical addressing, update MAX_PHYSMEM_BITS
accordingly.

Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Ganapatrao Kulkarni <ganapatrao.kulkarni@caviumnetworks.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
9 years agoarm64: ignore DT memreserve entries when booting in UEFI mode
Leif Lindholm [Mon, 28 Jul 2014 18:03:03 +0000 (19:03 +0100)]
arm64: ignore DT memreserve entries when booting in UEFI mode

UEFI provides its own method for marking regions to reserve, via the
memory map which is also used to initialise memblock. So when using the
UEFI memory map, ignore any memreserve entries present in the DT.

Reported-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Leif Lindholm <leif.lindholm@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
9 years agoarm64: configs: Enable X-Gene SATA and ethernet in defconfig
Mark Brown [Thu, 14 Aug 2014 19:57:16 +0000 (20:57 +0100)]
arm64: configs: Enable X-Gene SATA and ethernet in defconfig

Currently when run on an APM platform the ARMv8 defconfig has no viable
options for rootfs other than ramdisk which is rather limiting. Since
we already have both SATA and the bits needed for NFS root enabled we just
need to enable the relevant drivers so do that, helping enable direct
testing of upstream.

If the configuration ends up becoming too big we can consider modularising
some of the drivers and asking people to use an initramfs but for now this
is not an issue.

Signed-off-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
9 years agoarm64: align randomized TEXT_OFFSET on 4 kB boundary
Ard Biesheuvel [Wed, 13 Aug 2014 17:53:03 +0000 (18:53 +0100)]
arm64: align randomized TEXT_OFFSET on 4 kB boundary

When booting via UEFI, the kernel Image is loaded at a 4 kB boundary and
the embedded EFI stub is executed in place. The EFI stub relocates the
Image to reside TEXT_OFFSET bytes above a 2 MB boundary, and jumps into
the kernel proper.

In AArch64, PC relative symbol references are emitted using adrp/add or
adrp/ldr pairs, where the offset into a 4 kB page is resolved using a
separate :lo12: relocation. This implicitly assumes that the code will
always be executed at the same relative offset with respect to a 4 kB
boundary, or the references will point to the wrong address.

This means we should link the kernel at a 4 kB aligned base address in
order to remain compatible with the base address the UEFI loader uses
when doing the initial load of Image. So update the code that generates
TEXT_OFFSET to choose a multiple of 4 kB.

At the same time, update the code so it chooses from the interval [0..2MB)
as the author originally intended.

Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
9 years agotipc: Fix build.
David S. Miller [Tue, 19 Aug 2014 18:14:02 +0000 (11:14 -0700)]
tipc: Fix build.

Missing semicolon in range check fix.

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'cbq-fixes'
David S. Miller [Tue, 19 Aug 2014 17:59:12 +0000 (10:59 -0700)]
Merge branch 'cbq-fixes'

Vasily Averin says:

====================
cbq: incorrectly low bandwidth blocks limited traffic

v2: patch description changes
Fixes: f0f6ee1f70c4 ("cbq: incorrect processing of high limits")

Mainstream commit f0f6ee1f70c4 ("cbq: incorrect processing of high limits")
have side effect: if cbq bandwidth setting is less than real interface
throughput non-limited traffic can delay limited traffic for a very long time.

This happen because of q->now changes incorrectly in cbq_dequeue():
in described scenario L2T is much greater than real time delay,
and q->now gets an extra boost for each transmitted packet.

Accumulated boost prevents update q->now, and blocked class can wait
very long time until (q->now >= cl->undertime) will be true again.

More detailed problem description can be found here:
http://www.spinics.net/lists/netdev/msg292493.html

Following patches should fix the problem.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agocbq: now_rt removal
Vasily Averin [Thu, 14 Aug 2014 08:27:59 +0000 (12:27 +0400)]
cbq: now_rt removal

Now q->now_rt is identical to q->now and is not required anymore.

Signed-off-by: Vasily Averin <vvs@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agocbq: incorrectly low bandwidth setting blocks limited traffic
Vasily Averin [Thu, 14 Aug 2014 08:27:47 +0000 (12:27 +0400)]
cbq: incorrectly low bandwidth setting blocks limited traffic

Mainstream commit f0f6ee1f70c4 ("cbq: incorrect processing of high limits")
have side effect: if cbq bandwidth setting is less than real interface
throughput non-limited traffic can delay limited traffic for a very long time.

This happen because of q->now changes incorrectly in cbq_dequeue():
in described scenario L2T is much greater than real time delay,
and q->now gets an extra boost for each transmitted packet.

Accumulated boost prevents update q->now, and blocked class can wait
very long time until (q->now >= cl->undertime) will be true again.

To fix the problem the patch updates q->now on each cbq_update() call.
L2T-related pre-modification q->now was moved to cbq_update().

My testing confirmed that it fixes the problem and did not discover
any side-effects

Fixes: f0f6ee1f70c4 ("cbq: incorrect processing of high limits")

Signed-off-by: Vasily Averin <vvs@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>