firefly-linux-kernel-4.4.55.git
9 years agoniu: fix error handling in niu_class_to_ethflow()
Dan Carpenter [Thu, 26 Feb 2015 16:56:56 +0000 (19:56 +0300)]
niu: fix error handling in niu_class_to_ethflow()

There is a discrepancy here because the niu_class_to_ethflow() returns
zero on failure and one on success but the caller expected zero on
success and negative on failure.

The problem means that we allow the user to pass classes and flow_types
which we don't want.  I've looked at it a bit and I don't see it as a
very serious bug.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: smc91x: use run-time configuration on all ARM machines
Arnd Bergmann [Wed, 25 Feb 2015 15:31:57 +0000 (16:31 +0100)]
net: smc91x: use run-time configuration on all ARM machines

The smc91x driver traditionally gets configured at compile-time
for whichever hardware it runs on. This no longer works on
ARM as we continue to move to building all-in-one kernels.

Most ARM configurations with this driver already use run-time
configuration through DT or through platform_data, but a
few have not been converted yet.

I've checked all ARM boards that use this driver in their
legacy board files, and converted the ones that were using
compile-time configuration in smc91x.h to behave like the
other ones and provide the interrupt polarity along with
the MMIO configuration (width, stride) at platform device
creation time.

In particular, these combinations were previously selectable
in Kconfig but in fact broken:

- sa1100 assabet plus pleb
- msm combined with any other armv6/v7 platform
- pxa-idp combined with any non-DMA pxa variant
- LogicPD PXA270 combined with any other pxa
- nomadik combined with any other armv4/v5 platform,
  e.g. versatile.

None of these seem critical enough to warrant a backport
to stable, but it would be nice to clean this up for good.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
----
I would like the patch to get merged through netdev, after
Robert and/or Linus have verified it on at least some hardware.

There are a few other non-ARM platforms using this driver,
I could do the same patch for those if we want to take
it further.

 arch/arm/mach-msm/board-halibut.c    |   8 ++++-
 arch/arm/mach-msm/board-qsd8x50.c    |   8 ++++-
 arch/arm/mach-pxa/idp.c              |   5 +++
 arch/arm/mach-pxa/lpd270.c           |   8 ++++-
 arch/arm/mach-realview/core.c        |   7 ++++
 arch/arm/mach-realview/realview_eb.c |   2 +-
 arch/arm/mach-sa1100/neponset.c      |   6 ++++
 arch/arm/mach-sa1100/pleb.c          |   7 ++++
 drivers/net/ethernet/smsc/smc91x.c   |   9 +++--
 drivers/net/ethernet/smsc/smc91x.h   | 114 ++----------------------------------------------------------
 10 files changed, 57 insertions(+), 117 deletions(-)
Tested-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: use cond_resched()
Eric Dumazet [Thu, 26 Feb 2015 15:20:34 +0000 (07:20 -0800)]
rhashtable: use cond_resched()

If a hash table has 128 slots and 16384 elems, expand to 256 slots
takes more than one second. For larger sets, a soft lockup is detected.

Holding cpu for that long, even in a work queue is a show stopper
for non preemptable kernels.

cond_resched() at strategic points to allow process scheduler
to reschedule us.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net
David S. Miller [Fri, 27 Feb 2015 22:48:17 +0000 (17:48 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/jkirsher/net

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates 2015-02-26

This series contains fixes for i40e and i40evf only.

Alexey Khoroshilov found a possible leak of 'cmd_buf' when copy_from_user()
failed in i40e_dbg_command_write(), so resolved by calling kfree().

Shannon provides a fix to ensure the shift and bitwise precedences do not
work backwards for us by adding parans.  Fixed the driver by preventing
the driver from allowing stray interrupts or causing system logs from
un-handled interrupts by combining the ICR0 shutdown with the standard
interrupt shutdown and add the interrupt clearing to the PCI shutdown
path.  Fixed an issue where a NVM write times out before a transaction
can complete, so Shannon added logic to make another attempt by
reacquiring the semaphore, then retry the write, if the one retry fails,
we will then give up.  Adds checks to pointers before their use to ensure
we do not try to dereference NULL pointers when returning values from the
AdminQ calls.

Akeem adds a check to bail out if the device is already down when checking
for Tx hang subtask.

Anjali fixes TSO with more than 8 frags per segment issue.  The hardware
has some limitations which the driver needs to adhere to:
  1) no more than 8 descriptors per packet on the wire
  2) no header can span more than 3 descriptors
If one of these events happens, the hardware will generate an internal
error and freeze the Tx queue, so Anjali fixes this by linearizes the skb
to avoid these situations.  Fixed an issue where the per Traffic Class
queue count was higher than queues enabled, which will fix a warning
with multiple function mode where systems regularly have more cores than
vectors.  Fixed TCP/IPv6 over VXLAN Tx checksum offload, where we were
checking the outer protocol flags and deciding the flow for the inner
header.

Jesse fixes a race condition in the transmit hang detection.  Before we
were having issues of false Tx hang detection, no the driver makes more
direct with the checks for progress forward by directly checking the head
write back address and tail register when determining progress.  This
avoids Tx hangs where the software gets behind, because we are directly
checking hardware state when determining a hang state.

Neerav fixes the transmit ring Qset handle when DCB reconfigures. The issue
was when DCB is reconfigured to a single traffic class (TC) and the driver
did not reset the Tx ring Qset handle to correct the mapping, which caused
the Tx queue to disable timeouts.  Also as part of DCB reconfiguration flow
if the Tx queue disable times out, then issue a PF reset to do some level
of recovery.

Mitch stops flow director on shutdown because, in some cases, the hardware
would continue to try to access the FDIR ring after entering D3Hot state,
which would cause either PCIe errors or NMIs, depending upon the system
configuration.

* NOTE * I have verified that this series of patches for net will not cause
any merge issues when you sync up your net tree with your net-next tree.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoamd-xgbe: Request IRQs only after driver is fully setup
Lendacky, Thomas [Wed, 25 Feb 2015 19:50:12 +0000 (13:50 -0600)]
amd-xgbe: Request IRQs only after driver is fully setup

It is possible that the hardware may not have been properly shutdown
before this driver gets control, through use by firmware, for example.
Until the driver is loaded, interrupts associated with the hardware
could go pending. When the IRQs are requested napi support has not
been initialized yet, but the ISR will get control and schedule napi
processing resulting in a kernel panic because the poll routine has not
been set.

Adjust the code so that the driver is fully ready to handle and process
interrupts as soon as the IRQs are requested. This involves requesting
and freeing IRQs during start and stop processing and ordering the napi
add and delete calls appropriately.

Also adjust the powerup and powerdown routines to match the start and
stop routines in regards to the ordering of tasks, including napi
related calls.

Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: asix: add support for the Sitecom LN-028 USB adapter
Luca Ceresoli [Wed, 25 Feb 2015 23:58:12 +0000 (00:58 +0100)]
net: asix: add support for the Sitecom LN-028 USB adapter

Just another AX88178-based 10/100/1000 USB-to-Ethernet dongle. This one
shows up in lsusb as: "Sitecom Europe B.V. LN-028 Network USB 2.0 Adapter".

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Francois Romieu <romieu@fr.zoreil.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: linux-usb@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'rhashtable'
David S. Miller [Fri, 27 Feb 2015 21:06:21 +0000 (16:06 -0500)]
Merge branch 'rhashtable'

Daniel Borkmann says:

====================
rhashtable updates

As discussed, I'm sending out rhashtable fixups for -net.

I have a couple of more patches I was working on last week pending,
i.e. to get rid of ht->nelems and ht->shift atomic operations which
speed-up pure insertions/deletions, e.g. on my laptop I have 2 threads,
inserting 7M entries each, that will reduce insertion time from ~1,450 ms
to 865 ms (performance should even be better after removing the
grow/shrink indirections). I guess that however is rather something
for net-next.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: remove indirection for grow/shrink decision functions
Daniel Borkmann [Wed, 25 Feb 2015 15:31:54 +0000 (16:31 +0100)]
rhashtable: remove indirection for grow/shrink decision functions

Currently, all real users of rhashtable default their grow and shrink
decision functions to rht_grow_above_75() and rht_shrink_below_30(),
so that there's currently no need to have this explicitly selectable.

It can/should be generic and private inside rhashtable until a real
use case pops up. Since we can make this private, we'll save us this
additional indirection layer and can improve insertion/deletion time
as well.

Reference: http://patchwork.ozlabs.org/patch/443040/
Suggested-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: unconditionally grow when max_shift is not specified
Daniel Borkmann [Wed, 25 Feb 2015 15:31:53 +0000 (16:31 +0100)]
rhashtable: unconditionally grow when max_shift is not specified

While commit c0c09bfdc415 ("rhashtable: avoid unnecessary wakeup for
worker queue") rightfully moved part of the decision making of
whether we should expand or shrink from the expand/shrink functions
themselves into insert/delete functions in order to avoid unnecessary
worker wake-ups, it however introduced a regression by doing so.

Before that change, if no max_shift was specified (= 0) on rhashtable
initialization, rhashtable_expand() would just grow unconditionally
and lets the available memory be the limiting factor. After that
change, if no max_shift was specified, there would be _no_ expansion
step at all.

Given that netlink and tipc have a max_shift specified, it was not
visible there, but Josh Hunt reported that if nft that starts out
with a default element hint of 3 if not otherwise provided, would
slow i.e. inserts down trememdously as it cannot grow larger to
relax table occupancy.

Given that the test case verifies shrinks/expands manually, we also
must remove pointer to the helper functions to explicitly avoid
parallel resizing on insertions/deletions. test_bucket_stats() and
test_rht_lookup() could also be wrapped around rhashtable mutex to
explicitly synchronize a walk from resizing, but I think that defeats
the actual test case which intended to have explicit test steps,
i.e. 1) inserts, 2) expands, 3) shrinks, 4) deletions, with object
verification after each stage.

Reported-by: Josh Hunt <johunt@akamai.com>
Fixes: c0c09bfdc415 ("rhashtable: avoid unnecessary wakeup for worker queue")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Ying Xue <ying.xue@windriver.com>
Cc: Josh Hunt <johunt@akamai.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agovhost: drop hard-coded num_buffers size
Michael S. Tsirkin [Wed, 25 Feb 2015 14:20:01 +0000 (15:20 +0100)]
vhost: drop hard-coded num_buffers size

The 2 that we use for copy_to_iter comes from sizeof(u16),
it used to be that way before the iov iter update.
Fix it up, making it obvious the size of stack access
is right.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agovhost: cleanup iterator update logic
Michael S. Tsirkin [Wed, 25 Feb 2015 14:19:28 +0000 (15:19 +0100)]
vhost: cleanup iterator update logic

Recent iterator-related changes in vhost made it
harder to follow the logic fixing up the header.
In fact, the fixup always happens at the same
offset: sizeof(virtio_net_hdr): sometimes the
fixup iterator is updated by copy_to_iter,
sometimes-by iov_iter_advance.

Rearrange code to make this obvious.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorocker: silence shift wrapping warning
Dan Carpenter [Wed, 25 Feb 2015 13:36:12 +0000 (16:36 +0300)]
rocker: silence shift wrapping warning

"val" is declared as a u64 so static checkers complain that this shift
can wrap.  I don't have the hardware but probably it's doesn't have over
31 ports.  Still we may as well silence the warning even if it's not a
real bug.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Scott Feldman <sfeldma@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorocker: add a check for NULL in rocker_probe_ports()
Dan Carpenter [Wed, 25 Feb 2015 13:35:32 +0000 (16:35 +0300)]
rocker: add a check for NULL in rocker_probe_ports()

Make sure kmalloc() succeeds.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Scott Feldman <sfeldma@gmail.com>
Acked-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agocxgb4: Fix PCI-E Memory window interface for big-endian systems
Hariprasad Shenai [Wed, 25 Feb 2015 11:20:04 +0000 (16:50 +0530)]
cxgb4: Fix PCI-E Memory window interface for big-endian systems

When doing reads and writes to adapter memory via the PCI-E Memory Window
interface, data gets swizzled on 4-byte boundaries on Big-Endian systems
because we need to account for the register read/write interface which
incorporates a swizzle onto the Little-Endian PCI-E Bus.

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 agoenic: do notify_check before returning credits
Sujith Sankar [Wed, 25 Feb 2015 09:56:55 +0000 (15:26 +0530)]
enic: do notify_check before returning credits

We should complete notify_check before returning the credits. Once we return the
credits, adaptor may access the notify data.

Signed-off-by: Sujith Sankar <ssujith@cisco.com>
Signed-off-by: Govindarajulu Varadarajan <_govind@gmx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoi40e: check pointers before use
Shannon Nelson [Sat, 21 Feb 2015 06:45:28 +0000 (06:45 +0000)]
i40e: check pointers before use

Make sure we don't try to dereference NULL pointers when returning values
from the AdminQ calls.

Change-ID: Ia6694f2f415d50acf0aba063c863568742799aff
Signed-off-by: Shannon Nelson <shannon.nelson@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: catch NVM write semaphore timeout and retry
Shannon Nelson [Sat, 21 Feb 2015 06:45:10 +0000 (06:45 +0000)]
i40e: catch NVM write semaphore timeout and retry

In some circumstances, a multi-write transaction takes longer than the
default 3 minute timeout on the write semaphore.  If the write failed with
an EBUSY status, this is likely the problem, so here we try to reacquire
the semaphore then retry the write.  We only do one retry, then give up.

Change-ID: I1c8be60688acc2f39573839579baf601207c4a36
Signed-off-by: Shannon Nelson <shannon.nelson@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: stop flow director on shutdown
Mitch A Williams [Sat, 21 Feb 2015 06:44:51 +0000 (06:44 +0000)]
i40e: stop flow director on shutdown

In some cases, the hardware would continue to try to access the FDIR
ring after entering D3Hot state, which would cause either PCIe errors or
NMIs, depending upon system configuration.

Explicitly stop FDIR in our shutdown routine to eliminate this
possibility.

Change-ID: I1bd9fc7fd8f151fe24cad132ac9adddab923e3af
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: disconnect irqs on shutdown
Shannon Nelson [Sat, 21 Feb 2015 06:44:33 +0000 (06:44 +0000)]
i40e: disconnect irqs on shutdown

Combine the ICR0 shutdown with the standard interrupt shutdown, and
add the interrupt clearing to the PCI shutdown path.

This prevents the driver from allowing stray interrupts or causing
system logs from un-handled interrupts.

Change-ID: I48f6ab95cad7f8ca77c1f26c92a51cc1034ced43
Signed-off-by: Shannon Nelson <shannon.nelson@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40evf: TCP/IPv6 over Vxlan Tx checksum offload fix
Anjali Singhai [Sat, 21 Feb 2015 06:44:16 +0000 (06:44 +0000)]
i40evf: TCP/IPv6 over Vxlan Tx checksum offload fix

We were checking the outer Protocol flags and deciding the flow for
inner header. This patch fixes that.
This fixes the Tx checksum offload for TCP/IPv6 over vxlan.

Change-ID: I837aaea921d34f71b24c2bc32aaadea5001ddf78
Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: Issue a PF reset if Tx queue disable timeout
Parikh, Neerav [Sat, 21 Feb 2015 06:43:55 +0000 (06:43 +0000)]
i40e: Issue a PF reset if Tx queue disable timeout

As part of DCB reconfiguration flow if the Tx queue disable times out
then issue a PF reset to do some level of recovery.

Change-ID: I7550021c55bff355351c0365e61e1f05fcaff46d
Signed-off-by: Neerav Parikh <neerav.parikh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: Fix the Tx ring qset handle when DCB reconfigures
Parikh, Neerav [Sat, 21 Feb 2015 06:43:37 +0000 (06:43 +0000)]
i40e: Fix the Tx ring qset handle when DCB reconfigures

When DCB is reconfigured to single TC the driver did not reset the
Tx ring Qset handle to the correct mapping; which caused Tx queue
disable timeouts.

Change-ID: I4da5915ec92a83c281b478d653fae6ef1b72edfe
Signed-off-by: Neerav Parikh <neerav.parikh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: Fix the case where per TC queue count was higher than queues enabled
Anjali Singhai [Sat, 21 Feb 2015 06:43:19 +0000 (06:43 +0000)]
i40e: Fix the case where per TC queue count was higher than queues enabled

When the driver or hardware gets less interrupt vectors than the actual
number of CPU cores, limit the queue count for the priority queue
traffic class (TC) queues.

This will fix a warning with multiple function mode where systems
regularly have more cores than vectors.

Also add extra comment for readability.

Change-ID: I4f02226263aa3995e1f5ee5503eac0cd6ee12fbd
Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: fix race in hang check
Jesse Brandeburg [Tue, 24 Feb 2015 05:26:03 +0000 (05:26 +0000)]
i40e: fix race in hang check

The driver was having some issues with false Tx hang detection. This
makes the driver a little more direct with the checks for progress
forward by directly checking the head write back address and tail register
when determining progress.  This avoids Tx hangs where the software
gets behind, because we are directly checking hardware state when
determining hang state.

Change-ID: I774f0e861c9e8ab5ccb213634100fe15440ae24a
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: Fix TSO with more than 8 frags per segment issue
Anjali Singhai [Sat, 21 Feb 2015 06:42:35 +0000 (06:42 +0000)]
i40e: Fix TSO with more than 8 frags per segment issue

The hardware has some limitations the driver needs to adhere to,
that we found in extended testing.
  1) no more than 8 descriptors per packet on the wire
  2) no header can span more than 3 descriptors

If one of these events occurs, the hardware will generate an internal
error and freeze the Tx queue.

This patch linearizes the skb to avoid these situations.

Change-ID: I37dab7d3966e14895a9663ec4d0aaa8eb0d9e115
Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: Don't check for Tx hang when PF down
Akeem G Abodunrin [Sat, 21 Feb 2015 06:42:15 +0000 (06:42 +0000)]
i40e: Don't check for Tx hang when PF down

This patch adds check to bail out if device is already down when checking
for Tx hang subtask.

Change-ID: I3853fb7a6d11cb9a4c349b687cb25c15b19977a0
Signed-off-by: Akeem G Abodunrin <akeem.g.abodunrin@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: fix shift precedence issue
Shannon Nelson [Sat, 21 Feb 2015 06:41:47 +0000 (06:41 +0000)]
i40e: fix shift precedence issue

Add parens to make sure the shift and bitwise precedences don't work backwards
for us.

Change-ID: I60c10ef4fad6bc654522b9d8a53da2e270a0f268
Reported-by: Joe Perches <joe@perches.com>
Signed-off-by: Shannon Nelson <shannon.nelson@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoi40e: Fix memory leak at failure path in i40e_dbg_command_write()
Alexey Khoroshilov [Sat, 21 Feb 2015 07:32:47 +0000 (07:32 +0000)]
i40e: Fix memory leak at failure path in i40e_dbg_command_write()

The patch fixes a leak of 'cmd_buf' when copy_from_user() failed
in i40e_dbg_command_write().

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

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
9 years agoMAINTAINERS: update my email address
Andy Gospodarek [Wed, 25 Feb 2015 22:00:16 +0000 (17:00 -0500)]
MAINTAINERS: update my email address

I have been signing off on patches with this address so I'll change it.

Signed-off-by: Andy Gospodarek <gospo@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoamd-xgbe-phy: PHY KX/KR mode differences
Tom Lendacky [Tue, 24 Feb 2015 16:47:49 +0000 (10:47 -0600)]
amd-xgbe-phy: PHY KX/KR mode differences

The PHY requires different settings for the Decision Feedback Analyzer
(DFE) when running in KX mode vs. KR mode. Update the code to change
these settings when changing modes in order to provide a more stable
link.

Additionally, adjust the 10GbE PQ skew default setting to a more sane
value.

Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agor8169: Fix trivial typo in rtl_check_firmware
Yannick Guerrini [Tue, 24 Feb 2015 12:03:51 +0000 (13:03 +0100)]
r8169: Fix trivial typo in rtl_check_firmware

Change 'firwmare' to 'firmware'

Signed-off-by: Yannick Guerrini <yguerrini@tomshardware.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoxen-netback: release pending index before pushing Tx responses
David Vrabel [Tue, 24 Feb 2015 11:17:59 +0000 (11:17 +0000)]
xen-netback: release pending index before pushing Tx responses

If the pending indexes are released /after/ pushing the Tx response
then a stale pending index may be used if a new Tx request is
immediately pushed by the frontend.  The may cause various WARNINGs or
BUGs if the stale pending index is actually still in use.

Fix this by releasing the pending index before pushing the Tx
response.

The full barrier for the pending ring update is not required since the
the Tx response push already has a suitable write barrier.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoaf_packet: don't pass empty blocks for PACKET_V3
Alexander Drozdov [Tue, 24 Feb 2015 05:18:28 +0000 (08:18 +0300)]
af_packet: don't pass empty blocks for PACKET_V3

Before da413eec729d ("packet: Fixed TPACKET V3 to signal poll when block is
closed rather than every packet") poll listening for an af_packet socket was
not signaled if there was no packets to process. After the patch poll is
signaled evety time when block retire timer expires. That happens because
af_packet closes the current block on timeout even if the block is empty.

Passing empty blocks to the user not only wastes CPU but also wastes ring
buffer space increasing probability of packets dropping on small timeouts.

Signed-off-by: Alexander Drozdov <al.drozdov@gmail.com>
Cc: Dan Collins <dan@dcollins.co.nz>
Cc: Willem de Bruijn <willemb@google.com>
Cc: Guy Harris <guy@alum.mit.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agortnetlink: avoid 0 sized arrays
Sasha Levin [Tue, 24 Feb 2015 19:14:35 +0000 (14:14 -0500)]
rtnetlink: avoid 0 sized arrays

Arrays (when not in a struct) "shall have a value greater than zero".

GCC complains when it's not the case here.

Fixes: ba7d49b1f0 ("rtnetlink: provide api for getting and setting slave info")
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoipv6: addrconf: validate new MTU before applying it
Marcelo Leitner [Mon, 23 Feb 2015 14:17:13 +0000 (11:17 -0300)]
ipv6: addrconf: validate new MTU before applying it

Currently we don't check if the new MTU is valid or not and this allows
one to configure a smaller than minimum allowed by RFCs or even bigger
than interface own MTU, which is a problem as it may lead to packet
drops.

If you have a daemon like NetworkManager running, this may be exploited
by remote attackers by forging RA packets with an invalid MTU, possibly
leading to a DoS. (NetworkManager currently only validates for values
too small, but not for too big ones.)

The fix is just to make sure the new value is valid. That is, between
IPV6_MIN_MTU and interface's MTU.

Note that similar check is already performed at
ndisc_router_discovery(), for when kernel itself parses the RA.

Signed-off-by: Marcelo Ricardo Leitner <mleitner@redhat.com>
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoaltera_tse: Fixes in NAPI and interrupt handling paths
Vlastimil Setka [Mon, 23 Feb 2015 17:30:29 +0000 (11:30 -0600)]
altera_tse: Fixes in NAPI and interrupt handling paths

Incorrect NAPI polling caused WARNING at net/core/dev.c net_rx_action.
Some stability issues were also seen at high throughput and system
load before this patch.

This patch contains several changes in altera_tse_main.c:

- tse_rx() is fixed to not process more than `limit` frames

- tse_poll() is refactored to match NAPI logic
  - only received frames are counted for return value
  - removed bogus condition `(rxcomplete >= budget || txcomplete > 0)`
  - replace by: if (rxcomplete < budget) -> call __napi_complete and enable irq

- altera_isr()
  - replace spin_lock_irqsave() by spin_lock() - we are in isr
  - use spinlocks just over irq manipulation, not over __napi_schedule
  - reset IRQ first, then disable and schedule napi

This is a cleaned up resubmission from Vlastimil's recent submission.

Signed-off-by: Vlastimil Setka <setka@vsis.cz>
Signed-off-by: Roman Pisl <rpisl@kky.zcu.cz>
Signed-off-by: Vince Bridgers <vbridger@opensource.altera.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoaltera_tse: Correct typo in obtaining tx_fifo_depth from devicetree
Vlastimil Setka [Mon, 23 Feb 2015 17:27:37 +0000 (11:27 -0600)]
altera_tse: Correct typo in obtaining tx_fifo_depth from devicetree

This patch corrects a typo in the way tx_fifo_depth is read from the
devicetree. This patch was submitted by Vlastimil about a week ago,
and is now cleaned up and resubmitted.

Signed-off-by: Vlastimil Setka <setka@vsis.cz>
Signed-off-by: Vince Bridgers <vbridger@opensource.altera.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: compat: Ignore MSG_CMSG_COMPAT in compat_sys_{send, recv}msg
Catalin Marinas [Mon, 23 Feb 2015 18:12:56 +0000 (18:12 +0000)]
net: compat: Ignore MSG_CMSG_COMPAT in compat_sys_{send, recv}msg

With commit a7526eb5d06b (net: Unbreak compat_sys_{send,recv}msg), the
MSG_CMSG_COMPAT flag is blocked at the compat syscall entry points,
changing the kernel compat behaviour from the one before the commit it
was trying to fix (1be374a0518a, net: Block MSG_CMSG_COMPAT in
send(m)msg and recv(m)msg).

On 32-bit kernels (!CONFIG_COMPAT), MSG_CMSG_COMPAT is 0 and the native
32-bit sys_sendmsg() allows flag 0x80000000 to be set (it is ignored by
the kernel). However, on a 64-bit kernel, the compat ABI is different
with commit a7526eb5d06b.

This patch changes the compat_sys_{send,recv}msg behaviour to the one
prior to commit 1be374a0518a.

The problem was found running 32-bit LTP (sendmsg01) binary on an arm64
kernel. Arguably, LTP should not pass 0xffffffff as flags to sendmsg()
but the general rule is not to break user ABI (even when the user
behaviour is not entirely sane).

Fixes: a7526eb5d06b (net: Unbreak compat_sys_{send,recv}msg)
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoirda: replace current->state by set_current_state()
Fabian Frederick [Mon, 23 Feb 2015 17:38:24 +0000 (18:38 +0100)]
irda: replace current->state by set_current_state()

Use helper functions to access current->state.
Direct assignments are prone to races and therefore buggy.

current->state = TASK_RUNNING can be replaced by __set_current_state()

Thanks to Peter Zijlstra for the exact definition of the problem.

Suggested-By: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Fabian Frederick <fabf@skynet.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: sched: export tc_connmark.h so it is uapi accessible
Jamal Hadi Salim [Mon, 23 Feb 2015 13:17:12 +0000 (08:17 -0500)]
net: sched: export tc_connmark.h so it is uapi accessible

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoteam: fix possible null pointer dereference in team_handle_frame
Jiri Pirko [Mon, 23 Feb 2015 13:02:54 +0000 (14:02 +0100)]
team: fix possible null pointer dereference in team_handle_frame

Currently following race is possible in team:

CPU0                                        CPU1
                                            team_port_del
                                              team_upper_dev_unlink
                                                priv_flags &= ~IFF_TEAM_PORT
team_handle_frame
  team_port_get_rcu
    team_port_exists
      priv_flags & IFF_TEAM_PORT == 0
    return NULL (instead of port got
                 from rx_handler_data)
                                              netdev_rx_handler_unregister

The thing is that the flag is removed before rx_handler is unregistered.
If team_handle_frame is called in between, team_port_exists returns 0
and team_port_get_rcu will return NULL.
So do not check the flag here. It is guaranteed by netdev_rx_handler_unregister
that team_handle_frame will always see valid rx_handler_data pointer.

Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Fixes: 3d249d4ca7d0 ("net: introduce ethernet teaming device")
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agodecnet: Fix obvious o/0 typo
Rasmus Villemoes [Mon, 23 Feb 2015 11:02:51 +0000 (12:02 +0100)]
decnet: Fix obvious o/0 typo

Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: initialize all rhashtable walker members
Sasha Levin [Mon, 23 Feb 2015 09:35:06 +0000 (04:35 -0500)]
rhashtable: initialize all rhashtable walker members

Commit f2dba9c6ff ("rhashtable: Introduce rhashtable_walk_*") forgot to
initialize the members of struct rhashtable_walker after allocating it, which
caused an undefined value for 'resize' which is used later on.

Fixes: f2dba9c6ff ("rhashtable: Introduce rhashtable_walk_*")
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetoot...
David S. Miller [Mon, 23 Feb 2015 20:05:39 +0000 (15:05 -0500)]
Merge branch 'for-upstream' of git://git./linux/kernel/git/bluetooth/bluetooth

Johan Hedberg says:

====================
pull request: bluetooth 2015-02-23

Here's one important fix for the 4.0-rc series. Refactoring of Intel
Bluetooth controller detection ended up disabling some older ones which
are based on CSR hardware. This patch re-introduces the necessary USB id
and fixes the breakage.

Please let me know if there are any issues pulling. Thanks.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoBluetooth: btusb: Fix issue with CSR based Intel Wireless controllers
Marcel Holtmann [Sun, 22 Feb 2015 23:41:18 +0000 (15:41 -0800)]
Bluetooth: btusb: Fix issue with CSR based Intel Wireless controllers

Older Wireless controllers from Intel used CSR chips to provide support
for Bluetooth.

The commit d0ac9eb72 (Bluetooth: btusb: Ignore unknown Intel devices
with generic descriptor) disabled these older controllers. To enable
them again, put them into the blacklist and mark them clearly as CSR
based controllers.

T:  Bus=02 Lev=02 Prnt=02 Port=05 Cnt=01 Dev#=  3 Spd=12   MxCh= 0
D:  Ver= 2.00 Cls=e0(wlcon) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=8087 ProdID=07da Rev=78.69
C:* #Ifs= 2 Cfg#= 1 Atr=e0 MxPwr=  0mA
I:* If#= 0 Alt= 0 #EPs= 3 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=81(I) Atr=03(Int.) MxPS=  16 Ivl=1ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=82(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=   0 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=   0 Ivl=1ms
I:  If#= 1 Alt= 1 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=   9 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=   9 Ivl=1ms
I:  If#= 1 Alt= 2 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  17 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  17 Ivl=1ms
I:  If#= 1 Alt= 3 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  25 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  25 Ivl=1ms
I:  If#= 1 Alt= 4 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  33 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  33 Ivl=1ms
I:  If#= 1 Alt= 5 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  49 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  49 Ivl=1ms

Reported-by: Kenneth R. Crudup <kenny@panix.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
9 years agotcp: fix tcp_should_expand_sndbuf() to use tcp_packets_in_flight()
Neal Cardwell [Fri, 20 Feb 2015 18:33:16 +0000 (13:33 -0500)]
tcp: fix tcp_should_expand_sndbuf() to use tcp_packets_in_flight()

tcp_should_expand_sndbuf() does not expand the send buffer if we have
filled the congestion window.

However, it should use tcp_packets_in_flight() instead of
tp->packets_out to make this check.

Testing has established that the difference matters a lot if there are
many SACKed packets, causing a needless performance shortfall.

Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: pktgen: disable xmit_clone on virtual devices
Eric Dumazet [Mon, 23 Feb 2015 01:03:41 +0000 (17:03 -0800)]
net: pktgen: disable xmit_clone on virtual devices

Trying to use burst capability (aka xmit_more) on a virtual device
like bonding is not supported.

For example, skb might be queued multiple times on a qdisc, with
various list corruptions.

Fixes: 38b2cf2982dc ("net: pktgen: packet bursting via skb->xmit_more")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Alexei Starovoitov <ast@plumgrid.com>
Acked-by: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agor8169: Revert BQL and xmit_more support.
David S. Miller [Sun, 22 Feb 2015 20:54:29 +0000 (15:54 -0500)]
r8169: Revert BQL and xmit_more support.

There are certain regressions which are pointing to
these two commits which we are having a hard time
resolving.  So revert them for now.

Specifically this reverts:

commit 0bec3b700d106a8b0a34227b2976d1a582f1aab7
Author: Florian Westphal <fw@strlen.de>
Date:   Wed Jan 7 10:49:49 2015 +0100

    r8169: add support for xmit_more

and

commit 1e918876853aa85435e0f17fd8b4a92dcfff53d6
Author: Florian Westphal <fw@strlen.de>
Date:   Wed Oct 1 13:38:03 2014 +0200

    r8169: add support for Byte Queue Limits

There were some attempts by Eric Dumazet to address some obvious
problems in the TX flow, to see if they would fix the problems,
but none of them seem to help for the regression reporters.

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agowan: cosa: replace current->state by set_current_state()
Fabian Frederick [Fri, 20 Feb 2015 18:12:56 +0000 (19:12 +0100)]
wan: cosa: replace current->state by set_current_state()

Use helper functions to access current->state.
Direct assignments are prone to races and therefore buggy.

current->state = TASK_RUNNING is replaced by __set_current_state()

Thanks to Peter Zijlstra for the exact definition of the problem.

Suggested-By: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Fabian Frederick <fabf@skynet.be>
Acked-By: Jan "Yenya" Kasprzak <kas@fi.muni.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agohso: replace current->state by __set_current_state()
Fabian Frederick [Fri, 20 Feb 2015 18:12:55 +0000 (19:12 +0100)]
hso: replace current->state by __set_current_state()

Use helper functions to access current->state.
Direct assignments are prone to races and therefore buggy.

Thanks to Peter Zijlstra for the exact definition of the problem.

Suggested-By: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Fabian Frederick <fabf@skynet.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agomISDN: replace current->state by set_current_state()
Fabian Frederick [Fri, 20 Feb 2015 18:12:52 +0000 (19:12 +0100)]
mISDN: replace current->state by set_current_state()

Use helper function to access current->state.
Direct assignments are prone to races and therefore buggy.

Thanks to Peter Zijlstra for the exact definition of the problem.

Signed-off-by: Fabian Frederick <fabf@skynet.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoaf_packet: allow packets defragmentation not only for hash fanout type
Alexander Drozdov [Fri, 20 Feb 2015 05:24:27 +0000 (08:24 +0300)]
af_packet: allow packets defragmentation not only for hash fanout type

Packets defragmentation was introduced for PACKET_FANOUT_HASH only,
see 7736d33f4262 ("packet: Add pre-defragmentation support for ipv4
fanouts")

It may be useful to have defragmentation enabled regardless of
fanout type. Without that, the AF_PACKET user may have to:
1. Collect fragments from different rings
2. Defragment by itself

Signed-off-by: Alexander Drozdov <al.drozdov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: ensure cache line alignment on bucket_table
Eric Dumazet [Fri, 20 Feb 2015 14:48:57 +0000 (06:48 -0800)]
rhashtable: ensure cache line alignment on bucket_table

struct bucket_table contains mostly read fields :

size, locks_mask, locks.

Make sure these are not sharing a cache line with buckets[]

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: reject creation of netdev names with colons
Matthew Thode [Wed, 18 Feb 2015 00:31:57 +0000 (18:31 -0600)]
net: reject creation of netdev names with colons

colons are used as a separator in netdev device lookup in dev_ioctl.c

Specific functions are SIOCGIFTXQLEN SIOCETHTOOL SIOCSIFNAME

Signed-off-by: Matthew Thode <mthode@mthode.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: dsa: bcm_sf2: fix 64-bits register reads
Florian Fainelli [Thu, 19 Feb 2015 19:09:27 +0000 (11:09 -0800)]
net: dsa: bcm_sf2: fix 64-bits register reads

Reading 64-bits register was not working because we inverted the steps
between reading the lower 32-bits of the register and reading the upper
32-bits. Swapping these operations is how the HW guarantees that 64-bits
reads are latched correctly. We only have a handful of 64-bits registers
for now, mostly MIB counters, so the imapct is low.

Fixes: 246d7f773c13 ("net: dsa: add Broadcom SF2 switch driver")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: allow to unload test module
Daniel Borkmann [Thu, 19 Feb 2015 23:53:39 +0000 (00:53 +0100)]
rhashtable: allow to unload test module

There's no good reason why to disallow unloading of the rhashtable
test case module.

Commit 9d6dbe1bbaf8 moved the code from a boot test into a stand-alone
module, but only converted the subsys_initcall() handler into a
module_init() function without a related exit handler, and thus
preventing the test module from unloading.

Fixes: 9d6dbe1bbaf8 ("rhashtable: Make selftest modular")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: better high order allocation attempts
Daniel Borkmann [Thu, 19 Feb 2015 23:53:38 +0000 (00:53 +0100)]
rhashtable: better high order allocation attempts

When trying to allocate future tables via bucket_table_alloc(), it seems
overkill on large table shifts that we probe for kzalloc() unconditionally
first, as it's likely to fail.

Only probe with kzalloc() for more reasonable table sizes and use vzalloc()
either as a fallback on failure or directly in case of large table sizes.

Fixes: 7e1e77636e36 ("lib: Resizable, Scalable, Concurrent Hash Table")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: don't test for shrink on insert, expansion on delete
Daniel Borkmann [Thu, 19 Feb 2015 23:53:37 +0000 (00:53 +0100)]
rhashtable: don't test for shrink on insert, expansion on delete

Restore pre 54c5b7d311c8 behaviour and only probe for expansions on inserts
and shrinks on deletes. Currently, it will happen that on initial inserts
into a sparse hash table, we may i.e. shrink it first simply because it's
not fully populated yet, only to later realize that we need to grow again.

This however is counter intuitive, e.g. an initial default size of 64
elements is already small enough, and in case an elements size hint is given
to the hash table by a user, we should avoid unnecessary expansion steps,
so a shrink is clearly unintended here.

Fixes: 54c5b7d311c8 ("rhashtable: introduce rhashtable_wakeup_worker helper function")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Ying Xue <ying.xue@windriver.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf
David S. Miller [Fri, 20 Feb 2015 22:36:20 +0000 (17:36 -0500)]
Merge git://git./pub/scm/linux/kernel/git/pablo/nf

Pablo Neira Ayuso says:

====================
Netfilter/IPVS fixes for net

The following patchset contains updates for your net tree, they are:

1) Fix removal of destination in IPVS when the new mixed family support
   is used, from Alexey Andriyanov via Simon Horman.

2) Fix module refcount undeflow in nft_compat when reusing a match /
   target.

3) Fix iptables-restore when the recent match is used with a new hitcount
   that exceeds threshold, from Florian Westphal.

4) Fix stack corruption in xt_socket due to using stack storage to save
   the inner IPv6 header, from Eric Dumazet.

I'll follow up soon with another batch with more fixes that are still
cooking.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agocaif: fix a signedness bug in cfpkt_iterate()
Dan Carpenter [Thu, 19 Feb 2015 09:13:13 +0000 (12:13 +0300)]
caif: fix a signedness bug in cfpkt_iterate()

The cfpkt_iterate() function can return -EPROTO on error, but the
function is a u16 so the negative value gets truncated to a positive
unsigned short.  This causes a static checker warning.

The only caller which might care is cffrml_receive(), when it's checking
the frame checksum.  I modified cffrml_receive() so that it never says
-EPROTO is a valid checksum.

Also this isn't ever going to be inlined so I removed the "inline".

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agocxgb4: Fix incorrect 'c' suffix to %pI4, use %pISc instead
Anish Bhatt [Wed, 18 Feb 2015 23:29:45 +0000 (15:29 -0800)]
cxgb4: Fix incorrect 'c' suffix to %pI4, use %pISc instead

Issue caught by 0-day kernel test infrastructure. Code changed to use sockaddr
members so that %pISc can be used instead.

Fixes: b5a02f503caa ('cxgb4 : Update ipv6 address handling api')
Signed-off-by: Anish Bhatt <anish@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoethtool: Add hw-switch-offload to netdev_features_strings.
Rami Rosen [Fri, 20 Feb 2015 19:34:58 +0000 (21:34 +0200)]
ethtool: Add hw-switch-offload to netdev_features_strings.

commit aafb3e98b279 (netdev: introduce new NETIF_F_HW_SWITCH_OFFLOAD feature
flag for switch device offloads) add a new feature without adding it to
netdev_features_strings array; this patch fixes this.

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoipvlan: Fix text that talks about ip util support
Mahesh Bandewar [Wed, 18 Feb 2015 23:15:57 +0000 (15:15 -0800)]
ipvlan: Fix text that talks about ip util support

ipvlan was added into 3.19 release and iproute2 added support
for the same in iproute2-3.19 package.

Signed-off-by: Mahesh Bandewar <maheshb@google.com>
CC: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agorhashtable: don't allocate ht structure on stack in test_rht_init
Daniel Borkmann [Fri, 20 Feb 2015 20:14:21 +0000 (21:14 +0100)]
rhashtable: don't allocate ht structure on stack in test_rht_init

With object runtime debugging enabled, the rhashtable test suite
will rightfully throw a warning "ODEBUG: object is on stack, but
not annotated" from rhashtable_init().

This is because run_work is (correctly) being initialized via
INIT_WORK(), and not annotated by INIT_WORK_ONSTACK(). Meaning,
rhashtable_init() is okay as is, we just need to move ht e.g.,
into global scope.

It never triggered anything, since test_rhashtable is rather a
controlled environment and effectively runs to completion, so
that stack memory is not vanishing underneath us, we shouldn't
confuse any testers with it though.

Fixes: 7e1e77636e36 ("lib: Resizable, Scalable, Concurrent Hash Table")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/appletalk: LTPC needs virt_to_bus
Arnd Bergmann [Wed, 18 Feb 2015 19:48:54 +0000 (20:48 +0100)]
net/appletalk: LTPC needs virt_to_bus

The ltpc driver is rather outdated and does not get built on most
platforms because it requires the ISA_DMA_API symbol. However
there are some ARM platforms that have ISA_DMA_API but no virt_to_bus,
and they get this build error when enabling the ltpc driver.

drivers/net/appletalk/ltpc.c: In function 'handlefc':
drivers/net/appletalk/ltpc.c:380:2: error: implicit declaration of function 'virt_to_bus' [-Werror=implicit-function-declaration]
  set_dma_addr(dma,virt_to_bus(ltdmacbuf));
  ^

This adds another dependency in Kconfig to avoid that configuration.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: smc91x: improve neponset hack
Arnd Bergmann [Wed, 18 Feb 2015 19:47:30 +0000 (20:47 +0100)]
net: smc91x: improve neponset hack

The smc91x driver tries to support multiple platforms at compile
time, but they are mutually exclusive at runtime, and not clearly
defined.

Trying to build for CONFIG_SA1100_ASSABET without CONFIG_ASSABET_NEPONSET
results in this link error:

drivers/built-in.o: In function `smc_drv_probe':
:(.text+0x33310c): undefined reference to `neponset_ncr_frob'

since the neponset_ncr_set function is not defined otherwise.

Similarly, building for both CONFIG_SA1100_ASSABET and CONFIG_SA1100_PLEB
results in a different build error:

smsc/smc91x.c: In function 'smc_drv_probe':
smsc/smc91x.c:2299:2: error: implicit declaration of function 'neponset_ncr_set' [-Werror=implicit-function-declaration]
  neponset_ncr_set(NCR_ENET_OSC_EN);
  ^
smsc/smc91x.c:2299:19: error: 'NCR_ENET_OSC_EN' undeclared (first use in this function)
  neponset_ncr_set(NCR_ENET_OSC_EN);
                   ^

This is an attempt to fix the call site responsible for both
errors, making sure we call the function exactly when the driver
is actually trying to run on the assabet/neponset machine. With
this patch, I no longer see randconfig build errors in this file.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agosock: sock_dequeue_err_skb() needs hard irq safety
Eric Dumazet [Wed, 18 Feb 2015 13:47:55 +0000 (05:47 -0800)]
sock: sock_dequeue_err_skb() needs hard irq safety

Non NAPI drivers can call skb_tstamp_tx() and then sock_queue_err_skb()
from hard IRQ context.

Therefore, sock_dequeue_err_skb() needs to block hard irq or
corruptions or hangs can happen.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Fixes: 364a9e93243d1 ("sock: deduplicate errqueue dequeue")
Fixes: cb820f8e4b7f7 ("net: Provide a generic socket error queue delivery method for Tx time stamps.")
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: Initialize all members in skb_gro_remcsum_init()
Geert Uytterhoeven [Wed, 18 Feb 2015 10:38:06 +0000 (11:38 +0100)]
net: Initialize all members in skb_gro_remcsum_init()

skb_gro_remcsum_init() initializes the gro_remcsum.delta member only,
leading to compiler warnings about a possibly uninitialized
gro_remcsum.offset member:

drivers/net/vxlan.c: In function â€˜vxlan_gro_receive’:
drivers/net/vxlan.c:602: warning: â€˜grc.offset’ may be used uninitialized in this function
net/ipv4/fou.c: In function â€˜gue_gro_receive’:
net/ipv4/fou.c:262: warning: â€˜grc.offset’ may be used uninitialized in this function

While these are harmless for now:
  - skb_gro_remcsum_process() sets offset before changing delta,
  - skb_gro_remcsum_cleanup() checks if delta is non-zero before
    accessing offset,
it's safer to let the initialization function initialize all members.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoethernet/ixp4xx: prevent allmulti from clobbering promisc
Derrick Pallas [Wed, 18 Feb 2015 08:50:25 +0000 (00:50 -0800)]
ethernet/ixp4xx: prevent allmulti from clobbering promisc

If both promisc and allmulti are set, promisc should trump allmulti and
disable the MAC filter; otherwise, the interface is not really promisc.

Previously, this code checked IFF_ALLMULTI prior to and without regard for
IFF_PROMISC; if both were set, only multicast and direct unicast traffic
would make it through the filter.

Signed-off-by: Derrick Pallas <pallas@meraki.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoopenvswitch: Fix net exit.
Pravin B Shelar [Tue, 17 Feb 2015 19:23:10 +0000 (11:23 -0800)]
openvswitch: Fix net exit.

Open vSwitch allows moving internal vport to different namespace
while still connected to the bridge. But when namespace deleted
OVS does not detach these vports, that results in dangling
pointer to netdevice which causes kernel panic as follows.
This issue is fixed by detaching all ovs ports from the deleted
namespace at net-exit.

BUG: unable to handle kernel NULL pointer dereference at 0000000000000028
IP: [<ffffffffa0aadaa5>] ovs_vport_locate+0x35/0x80 [openvswitch]
Oops: 0000 [#1] SMP
Call Trace:
 [<ffffffffa0aa6391>] lookup_vport+0x21/0xd0 [openvswitch]
 [<ffffffffa0aa65f9>] ovs_vport_cmd_get+0x59/0xf0 [openvswitch]
 [<ffffffff8167e07c>] genl_family_rcv_msg+0x1bc/0x3e0
 [<ffffffff8167e319>] genl_rcv_msg+0x79/0xc0
 [<ffffffff8167d919>] netlink_rcv_skb+0xb9/0xe0
 [<ffffffff8167deac>] genl_rcv+0x2c/0x40
 [<ffffffff8167cffd>] netlink_unicast+0x12d/0x1c0
 [<ffffffff8167d3da>] netlink_sendmsg+0x34a/0x6b0
 [<ffffffff8162e140>] sock_sendmsg+0xa0/0xe0
 [<ffffffff8162e5e8>] ___sys_sendmsg+0x408/0x420
 [<ffffffff8162f541>] __sys_sendmsg+0x51/0x90
 [<ffffffff8162f592>] SyS_sendmsg+0x12/0x20
 [<ffffffff81764ee9>] system_call_fastpath+0x12/0x17

Reported-by: Assaf Muller <amuller@redhat.com>
Fixes: 46df7b81454("openvswitch: Add support for network namespaces.")
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reviewed-by: Thomas Graf <tgraf@noironetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoematch: Fix auto-loading of ematch modules.
Ignacy Gawędzki [Tue, 17 Feb 2015 19:15:20 +0000 (20:15 +0100)]
ematch: Fix auto-loading of ematch modules.

In tcf_em_validate(), after calling request_module() to load the
kind-specific module, set em->ops to NULL before returning -EAGAIN, so
that module_put() is not called again by tcf_em_tree_destroy().

Signed-off-by: Ignacy Gawędzki <ignacy.gawedzki@green-communications.fr>
Acked-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: phy: Fix verification of EEE support in phy_init_eee
Guenter Roeck [Tue, 17 Feb 2015 17:36:22 +0000 (09:36 -0800)]
net: phy: Fix verification of EEE support in phy_init_eee

phy_init_eee uses phy_find_setting(phydev->speed, phydev->duplex)
to find a valid entry in the settings array for the given speed
and duplex value. For full duplex 1000baseT, this will return
the first matching entry, which is the entry for 1000baseKX_Full.

If the phy eee does not support 1000baseKX_Full, this entry will not
match, causing phy_init_eee to fail for no good reason.

Fixes: 9a9c56cb34e6 ("net: phy: fix a bug when verify the EEE support")
Fixes: 3e7077067e80c ("phy: Expand phy speed/duplex settings array")
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoipv4: ip_check_defrag should correctly check return value of skb_copy_bits
Alexander Drozdov [Tue, 17 Feb 2015 10:33:46 +0000 (13:33 +0300)]
ipv4: ip_check_defrag should correctly check return value of skb_copy_bits

skb_copy_bits() returns zero on success and negative value on error,
so it is needed to invert the condition in ip_check_defrag().

Fixes: 1bf3751ec90c ("ipv4: ip_check_defrag must not modify skb before unsharing")
Signed-off-by: Alexander Drozdov <al.drozdov@gmail.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agousb: plusb: Add support for National Instruments host-to-host cable
Ben Shelton [Mon, 16 Feb 2015 19:47:06 +0000 (13:47 -0600)]
usb: plusb: Add support for National Instruments host-to-host cable

The National Instruments USB Host-to-Host Cable is based on the Prolific
PL-25A1 chipset.  Add its VID/PID so the plusb driver will recognize it.

Signed-off-by: Ben Shelton <ben.shelton@ni.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoehea: Register memory hotplug, reboot and crash hooks on adapter probe
Anton Blanchard [Sun, 15 Feb 2015 19:44:20 +0000 (17:44 -0200)]
ehea: Register memory hotplug, reboot and crash hooks on adapter probe

ehea creates memory hotplug, reboot and crash hooks even if there
are no adapters in the box. Just create them when we probe our
first adapter.

[cascardo: use ehea_register_memory_hooks return code]
Signed-off-by: Anton Blanchard <anton@samba.org>
Tested-by: Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agogen_stats.c: Duplicate xstats buffer for later use
Ignacy Gawędzki [Fri, 13 Feb 2015 22:47:05 +0000 (14:47 -0800)]
gen_stats.c: Duplicate xstats buffer for later use

The gnet_stats_copy_app() function gets called, more often than not, with its
second argument a pointer to an automatic variable in the caller's stack.
Therefore, to avoid copying garbage afterwards when calling
gnet_stats_finish_copy(), this data is better copied to a dynamically allocated
memory that gets freed after use.

[xiyou.wangcong@gmail.com: remove a useless kfree()]

Signed-off-by: Ignacy Gawędzki <ignacy.gawedzki@green-communications.fr>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge tag 'mfd-for-linus-3.20' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd
Linus Torvalds [Wed, 18 Feb 2015 17:05:48 +0000 (09:05 -0800)]
Merge tag 'mfd-for-linus-3.20' of git://git./linux/kernel/git/lee/mfd

Pull MFD updates from Lee Jones:
 "Changes to existing drivers:
   - fixr platform device collision; da9052, wm8994-core
   - regmap configuration amendments; tps65218
   - fix runtime PM deadlock; rtsx_usb
   - remove unused/superfluous code; db8500-prcmu, omap-usb-host
   - enable watchdog timer; lpc_sch
   - add start/stop RX URBs helpers; dln2
   - remove platform device (DT only); max77686, max77802
   - support suspend and resume; dln2
   - add Device Tree support; da9063
   - extra error checking; intel_soc_pmic
   - const'ify all the things; 88pm860x, hi6421-pmic, intel_soc_pmic,
                               max77686, lm3533, retu, pcf50633,
                               davinci_voicecodec, smsc-ece1099,
                               tps65218, mc13xxx, tps65217, twl-core,
                               twl6040

  New drivers/supported devices:
   - new driver for Richtek RT5033
   - new driver for DA9150 Charger and FuelGauge
   - new driver for Qualcomm Resource Power Manager (RPM)
   - add support for the ir-clk into sun6i-prcm
   - add support for FuelGauge into axp20x"

* tag 'mfd-for-linus-3.20' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd: (32 commits)
  mfd: intel_soc_pmic: Add missing error check for devm_kzalloc
  mfd: rtsx_usb: Defer autosuspend while card exists
  mfd: devicetree: Add bindings for DA9063
  mfd: da9063: Add device tree support
  regulator: qcom-rpm: Add missing state flag in call to RPM
  mfd: qcom-rpm: Driver for the Qualcomm RPM
  mfd: devicetree: bindings: Add Qualcomm RPM DT binding
  mfd: max77686/802: Remove support for board files
  mfd: omap-usb-host: Remove some unused functions
  mfd: twl6040: Constify struct regmap_config and reg_default array
  mfd: twl-core: Constify struct regmap_config and reg_default array
  mfd: tps65217: Constify struct regmap_config
  mfd: mc13xxx: i2c/spi: Constify struct regmap_config
  mfd: tps65218: Constify struct regmap_config
  mfd: smsc-ece1099: Constify struct regmap_config
  mfd: davinci_voicecodec: Constify struct regmap_config
  mfd: pcf50633: Constify struct regmap_config
  mfd: retu: Constify struct regmap_config
  mfd: lm3533: Constify struct regmap_config
  mfd: max77686: Constify struct regmap_config
  ...

9 years agoMerge tag 'pwm/for-3.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry...
Linus Torvalds [Wed, 18 Feb 2015 17:00:53 +0000 (09:00 -0800)]
Merge tag 'pwm/for-3.20-rc1' of git://git./linux/kernel/git/thierry.reding/linux-pwm

Pull pwm updates from Thierry Reding:
 "This contains two new drivers, one for Allwinner SoCs and the other
  for Imagination Technologies' Pistachio SoC.

  Complementing this are a couple of fixes to the Atmel HLCDC PWM and
  STi PWM drivers as well as minor cleanups to the core and the Tegra
  driver"

* tag 'pwm/for-3.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm:
  pwm: tegra: Use NSEC_PER_SEC
  pwm: Remove unnecessary check before of_node_put()
  pwm: Add device tree binding document for IMG PWM DAC
  pwm: Imagination Technologies PWM DAC driver
  pwm: sti: Maintain a bitmap of configured devices
  pwm: sunxi: document OF bindings
  pwm: Add Allwinner SoC support
  pwm: atmel-hlcdc: Prevent division by zero
  pwm: atmel-hlcdc: Depend on HAVE_CLK

9 years agoMerge branch 'for-linus' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Wed, 18 Feb 2015 16:49:20 +0000 (08:49 -0800)]
Merge branch 'for-linus' of git://git.infradead.org/users/vkoul/slave-dma

Pull dmaengine updates from Vinod Koul:
 "This update brings:

   - the big cleanup up by Maxime for device control and slave
     capabilities.  This makes the API much cleaner.

   - new IMG MDC driver by Andrew

   - new Renesas R-Car Gen2 DMA Controller driver by Laurent along with
     bunch of fixes on rcar drivers

   - odd fixes and updates spread over driver"

* 'for-linus' of git://git.infradead.org/users/vkoul/slave-dma: (130 commits)
  dmaengine: pl330: add DMA_PAUSE feature
  dmaengine: pl330: improve pl330_tx_status() function
  dmaengine: rcar-dmac: Disable channel 0 when using IOMMU
  dmaengine: rcar-dmac: Work around descriptor mode IOMMU errata
  dmaengine: rcar-dmac: Allocate hardware descriptors with DMAC device
  dmaengine: rcar-dmac: Fix oops due to unintialized list in error ISR
  dmaengine: rcar-dmac: Fix spinlock issues in interrupt
  dmaenegine: edma: fix sparse warnings
  dmaengine: rcar-dmac: Fix uninitialized variable usage
  dmaengine: shdmac: extend PM methods
  dmaengine: shdmac: use SET_RUNTIME_PM_OPS()
  dmaengine: pl330: fix bug that cause start the same descs in cyclic
  dmaengine: at_xdmac: allow muliple dwidths when doing slave transfers
  dmaengine: at_xdmac: simplify channel configuration stuff
  dmaengine: at_xdmac: introduce save_cc field
  dmaengine: at_xdmac: wait for in-progress transaction to complete after pausing a channel
  ioat: fail self-test if wait_for_completion times out
  dmaengine: dw: define DW_DMA_MAX_NR_MASTERS
  dmaengine: dw: amend description of dma_dev field
  dmatest: move src_off, dst_off, len inside loop
  ...

9 years agoMerge git://www.linux-watchdog.org/linux-watchdog
Linus Torvalds [Wed, 18 Feb 2015 16:42:47 +0000 (08:42 -0800)]
Merge git://www.linux-watchdog.org/linux-watchdog

Pull watchdog updates from Wim Van Sebroeck:
 "This adds the following new drivers:

   - ImgTec PDC Watchdog Timer Driver,
   - Mediatek SoC integrated watchdog

  Add support for BCM5301X, IT8783, NCT6791 and NCT6792 WDT's

  Add bcm47xx_wdt and da9063 restart handlers and contains overall
  improvements and fixes"

* git://www.linux-watchdog.org/linux-watchdog:
  watchdog: bcm47xx_wdt.c: allow enabling on BCM5301X arch
  watchdog: jz4740: Add DT support
  dt: watchdog: Add DT binding documentation for jz4740 watchdog timer
  watchdog: dw_wdt: Try to get a 30 second watchdog by default
  watchdog: dw_wdt: pat the watchdog before enabling it
  watchdog: w83627hf_wdt: Add support for NCT6791 and NCT6792
  watchdog: bcm47xx_wdt.c: add restart handler support
  watchdog: gpio_wdt: Add "always_running" feature to GPIO watchdog
  watchdog: da9063: Add restart handler support
  ARM: mediatek: dts: Add bindings for watchdog
  watchdog: Add driver for Mediatek watchdog
  watchdog: Fix omap watchdogs to enable the magic close bit
  watchdog: rt2880_wdt: minor clean up
  watchdog: hpwdt: Fix initialization message in hpwdt.c
  watchdog: it87_wdt: add IT8783 ID
  watchdog: imx2: Constify struct regmap_config and watchdog_ops
  DT: watchdog: Add ImgTec PDC Watchdog Timer binding documentation
  watchdog: ImgTec PDC Watchdog Timer Driver

9 years agoMerge tag 'firewire-updates' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee139...
Linus Torvalds [Wed, 18 Feb 2015 16:22:38 +0000 (08:22 -0800)]
Merge tag 'firewire-updates' of git://git./linux/kernel/git/ieee1394/linux1394

Pull fireware updates from Stefan Richter:
 "IEEE 1394 subsystem updates:

   - Replace made-up, unallocated Vendor and Model values of
     firewire-core's Configuration ROM register root directory by
     properly registered IDs.  (These IDs are visible to peer nodes on
     the bus and locally via sysfs, but they are not involved in
     protocol matching or driver matching, nor are they used in stock
     udev rules)

   - Remove some unneccessary code"

* tag 'firewire-updates' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394:
  firewire: core: use correct vendor/model IDs
  firewire: sbp2: remove redundant check for bidi command
  firewire: ohci: Remove unused function

9 years agoMerge tag 'for-linus-20150216' of git://git.infradead.org/linux-mtd
Linus Torvalds [Wed, 18 Feb 2015 16:01:44 +0000 (08:01 -0800)]
Merge tag 'for-linus-20150216' of git://git.infradead.org/linux-mtd

Pull MTD updates from Brian Norris:
 "NAND:

   - Add new Hisilicon NAND driver for Hip04
   - Add default reboot handler, to ensure all outstanding erase
     transactions complete in time
   - jz4740: convert to use GPIO descriptor API
   - Atmel: add support for sama5d4
   - Change default bitflip threshold to 75% of correction strength
   - Miscellaneous cleanups and bugfixes

  SPI NOR:

   - Freescale QuadSPI:
   - Fix a few probe() and remove() issues
   - Add a MAINTAINERS entry for this driver
   - Tweak transfer size to increase read performance
   - Add suspend/resume support
   - Add Micron quad I/O support
   - ST FSM SPI: miscellaneous fixes

  JFFS2:

   - gracefully handle corrupted 'offset' field found on flash

  Other:

   - bcm47xxpart: add tweaks for a few new devices
   - mtdconcat: set return lengths properly for mtd_write_oob()
   - map_ram: enable use with mtdoops
   - maps: support fallback to ROM/UBI for write-protected NOR flash"

* tag 'for-linus-20150216' of git://git.infradead.org/linux-mtd: (46 commits)
  mtd: hisilicon: && vs & typo
  jffs2: fix handling of corrupted summary length
  mtd: hisilicon: add device tree binding documentation
  mtd: hisilicon: add a new NAND controller driver for hisilicon hip04 Soc
  mtd: avoid registering reboot notifier twice
  mtd: concat: set the return lengths properly
  mtd: kconfig: replace PPC_OF with PPC
  mtd: denali: remove unnecessary stubs
  mtd: nand: remove redundant local variable
  MAINTAINERS: add maintainer entry for FREESCALE QUAD SPI driver
  mtd: fsl-quadspi: improve read performance by increase AHB transfer size
  mtd: fsl-quadspi: Remove unnecessary 'map_failed' label
  mtd: fsl-quadspi: Remove unneeded success/error messages
  mtd: fsl-quadspi: Fix the error paths
  mtd: nand: omap: drop condition with no effect
  mtd: nand: jz4740: Convert to GPIO descriptor API
  mtd: nand: Request strength instead of bytes for soft BCH
  mtd: nand: default bitflip-reporting threshold to 75% of correction strength
  mtd: atmel_nand: introduce a new compatible string for sama5d4 chip
  mtd: atmel_nand: return max bitflips in all sectors in pmecc_correction()
  ...

9 years agopwm: tegra: Use NSEC_PER_SEC
Thierry Reding [Wed, 18 Feb 2015 07:40:29 +0000 (08:40 +0100)]
pwm: tegra: Use NSEC_PER_SEC

Instead of using the literal value for the number of nanoseconds per
second, use the macro instead to increase readability.

Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Wed, 18 Feb 2015 01:41:19 +0000 (17:41 -0800)]
Merge git://git./linux/kernel/git/davem/net

Pull networking updates from David Miller:

 1) Missing netlink attribute validation in nft_lookup, from Patrick
    McHardy.

 2) Restrict ipv6 partial checksum handling to UDP, since that's the
    only case it works for.  From Vlad Yasevich.

 3) Clear out silly device table sentinal macros used by SSB and BCMA
    drivers.  From Joe Perches.

 4) Make sure the remote checksum code never creates a situation where
    the remote checksum is applied yet the tunneling metadata describing
    the remote checksum transformation is still present.  Otherwise an
    external entity might see this and apply the checksum again.  From
    Tom Herbert.

 5) Use msecs_to_jiffies() where applicable, from Nicholas Mc Guire.

 6) Don't explicitly initialize timer struct fields, use setup_timer()
    and mod_timer() instead.  From Vaishali Thakkar.

 7) Don't invoke tg3_halt() without the tp->lock held, from Jun'ichi
    Nomura.

 8) Missing __percpu annotation in ipvlan driver, from Eric Dumazet.

 9) Don't potentially perform skb_get() on shared skbs, also from Eric
    Dumazet.

10) Fix COW'ing of metrics for non-DST_HOST routes in ipv6, from Martin
    KaFai Lau.

11) Fix merge resolution error between the iov_iter changes in vhost and
    some bug fixes that occurred at the same time.  From Jason Wang.

12) If rtnl_configure_link() fails we have to perform a call to
    ->dellink() before unregistering the device.  From WANG Cong.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (39 commits)
  net: dsa: Set valid phy interface type
  rtnetlink: call ->dellink on failure when ->newlink exists
  com20020-pci: add support for eae single card
  vhost_net: fix wrong iter offset when setting number of buffers
  net: spelling fixes
  net/core: Fix warning while make xmldocs caused by dev.c
  net: phy: micrel: disable NAND-tree for KSZ8021, KSZ8031, KSZ8051, KSZ8081
  ipv6: fix ipv6_cow_metrics for non DST_HOST case
  openvswitch: Fix key serialization.
  r8152: restore hw settings
  hso: fix rx parsing logic when skb allocation fails
  tcp: make sure skb is not shared before using skb_get()
  bridge: netfilter: Move sysctl-specific error code inside #ifdef
  ipv6: fix possible deadlock in ip6_fl_purge / ip6_fl_gc
  ipvlan: add a missing __percpu pcpu_stats
  tg3: Hold tp->lock before calling tg3_halt() from tg3_init_one()
  bgmac: fix device initialization on Northstar SoCs (condition typo)
  qlcnic: Delete existing multicast MAC list before adding new
  net/mlx5_core: Fix configuration of log_uar_page_sz
  sunvnet: don't change gso data on clones
  ...

9 years agoMerge tag 'md/3.20-fixes' of git://neil.brown.name/md
Linus Torvalds [Wed, 18 Feb 2015 01:34:21 +0000 (17:34 -0800)]
Merge tag 'md/3.20-fixes' of git://neil.brown.name/md

Pull md bugfixes from Neil Brown:
 "Three bug md fixes for 3.20

  yet-another-livelock in raid5, and a problem with write errors to
  4K-block devices"

* tag 'md/3.20-fixes' of git://neil.brown.name/md:
  md/raid5: Fix livelock when array is both resyncing and degraded.
  md/raid10: round up to bdev_logical_block_size in narrow_write_error.
  md/raid1: round up to bdev_logical_block_size in narrow_write_error

9 years agoMerge tag 'please-pull-fixmcelog' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 18 Feb 2015 01:03:07 +0000 (17:03 -0800)]
Merge tag 'please-pull-fixmcelog' of git://git./linux/kernel/git/ras/ras

Pull mcelog regression fix from Tony Luck:
 "Fix regression - functions on the mce notifier chain should not be
  able to decide that an event should not be logged"

* tag 'please-pull-fixmcelog' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
  x86/mce: Fix regression. All error records should report via /dev/mcelog

9 years agoMerge tag 'docs-fix' of git://git.lwn.net/linux-2.6
Linus Torvalds [Wed, 18 Feb 2015 01:02:04 +0000 (17:02 -0800)]
Merge tag 'docs-fix' of git://git.lwn.net/linux-2.6

Pull DocBook build fix from Jonathan Corbet:
 "Fix the DocBook build failure caused by the move of the i2o subsystem
  to the staging tree"

* tag 'docs-fix' of git://git.lwn.net/linux-2.6:
  Fix docs build failure caused by i2o removal

9 years agoMerge branch 'for-3.20' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Wed, 18 Feb 2015 01:00:54 +0000 (17:00 -0800)]
Merge branch 'for-3.20' of git://linux-nfs.org/~bfields/linux

Pull nfsd bugfixes from Bruce Fields:
 "These are fixes for two bugs introduced during the merge window"

* 'for-3.20' of git://linux-nfs.org/~bfields/linux:
  nfsd4: fix v3-less build
  nfsd: fix comparison in fh_fsid_match()

9 years agomd/raid5: Fix livelock when array is both resyncing and degraded.
NeilBrown [Wed, 18 Feb 2015 00:35:14 +0000 (11:35 +1100)]
md/raid5: Fix livelock when array is both resyncing and degraded.

Commit a7854487cd7128a30a7f4f5259de9f67d5efb95f:
  md: When RAID5 is dirty, force reconstruct-write instead of read-modify-write.

Causes an RCW cycle to be forced even when the array is degraded.
A degraded array cannot support RCW as that requires reading all data
blocks, and one may be missing.

Forcing an RCW when it is not possible causes a live-lock and the code
spins, repeatedly deciding to do something that cannot succeed.

So change the condition to only force RCW on non-degraded arrays.

Reported-by: Manibalan P <pmanibalan@amiindia.co.in>
Bisected-by: Jes Sorensen <Jes.Sorensen@redhat.com>
Tested-by: Jes Sorensen <Jes.Sorensen@redhat.com>
Signed-off-by: NeilBrown <neilb@suse.de>
Fixes: a7854487cd7128a30a7f4f5259de9f67d5efb95f
Cc: stable@vger.kernel.org (v3.7+)
9 years agoMerge branch 'lazytime' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Wed, 18 Feb 2015 00:12:34 +0000 (16:12 -0800)]
Merge branch 'lazytime' of git://git./linux/kernel/git/viro/vfs

Pull lazytime mount option support from Al Viro:
 "Lazytime stuff from tytso"

* 'lazytime' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  ext4: add optimization for the lazytime mount option
  vfs: add find_inode_nowait() function
  vfs: add support for a lazytime mount option

9 years agoMerge branch 'iov_iter' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Tue, 17 Feb 2015 23:48:33 +0000 (15:48 -0800)]
Merge branch 'iov_iter' of git://git./linux/kernel/git/viro/vfs

Pull iov_iter updates from Al Viro:
 "More iov_iter work - missing counterpart of iov_iter_init() for
  bvec-backed ones and vfs_read_iter()/vfs_write_iter() - wrappers for
  sync calls of ->read_iter()/->write_iter()"

* 'iov_iter' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  fs: add vfs_iter_{read,write} helpers
  new helper: iov_iter_bvec()

9 years agoMerge branch 'getname2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Tue, 17 Feb 2015 23:27:47 +0000 (15:27 -0800)]
Merge branch 'getname2' of git://git./linux/kernel/git/viro/vfs

Pull getname/putname updates from Al Viro:
 "Rework of getname/getname_kernel/etc., mostly from Paul Moore.  Gets
  rid of quite a pile of kludges between namei and audit..."

* 'getname2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  audit: replace getname()/putname() hacks with reference counters
  audit: fix filename matching in __audit_inode() and __audit_inode_child()
  audit: enable filename recording via getname_kernel()
  simpler calling conventions for filename_mountpoint()
  fs: create proper filename objects using getname_kernel()
  fs: rework getname_kernel to handle up to PATH_MAX sized filenames
  cut down the number of do_path_lookup() callers

9 years agoMerge branch 'debugfs_automount' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 17 Feb 2015 23:18:19 +0000 (15:18 -0800)]
Merge branch 'debugfs_automount' of git://git./linux/kernel/git/viro/vfs

Pull debugfs patches from Al Viro:
 "debugfs patches, mostly to make it possible for something like tracefs
  to be transparently automounted on given directory in debugfs.

  New primitive in there is debugfs_create_automount(name, parent, func,
  arg), which creates a directory and makes its ->d_automount() return
  func(arg).  Another missing primitive was debugfs_create_file_size() -
  open-coded in quite a few places.  Dave's patch adds it and converts
  the open-code instances to calling it"

* 'debugfs_automount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  debugfs: Provide a file creation function that also takes an initial size
  new primitive: debugfs_create_automount()
  debugfs: split end_creating() into success and failure cases
  debugfs: take mode-dependent parts of debugfs_get_inode() into callers
  fold debugfs_mknod() into callers
  fold debugfs_create() into caller
  fold debugfs_mkdir() into caller
  debugfs_mknod(): get rid useless arguments
  fold debugfs_link() into caller
  debugfs: kill __create_file()
  debugfs: split the beginning and the end of __create_file() off
  debugfs_{mkdir,create,link}(): get rid of redundant argument

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Tue, 17 Feb 2015 22:56:45 +0000 (14:56 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

Pull misc VFS updates from Al Viro:
 "This cycle a lot of stuff sits on topical branches, so I'll be sending
  more or less one pull request per branch.

  This is the first pile; more to follow in a few.  In this one are
  several misc commits from early in the cycle (before I went for
  separate branches), plus the rework of mntput/dput ordering on umount,
  switching to use of fs_pin instead of convoluted games in
  namespace_unlock()"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  switch the IO-triggering parts of umount to fs_pin
  new fs_pin killing logics
  allow attaching fs_pin to a group not associated with some superblock
  get rid of the second argument of acct_kill()
  take count and rcu_head out of fs_pin
  dcache: let the dentry count go down to zero without taking d_lock
  pull bumping refcount into ->kill()
  kill pin_put()
  mode_t whack-a-mole: chelsio
  file->f_path.dentry is pinned down for as long as the file is open...
  get rid of lustre_dump_dentry()
  gut proc_register() a bit
  kill d_validate()
  ncpfs: get rid of d_validate() nonsense
  selinuxfs: don't open-code d_genocide()

9 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Tue, 17 Feb 2015 22:35:02 +0000 (14:35 -0800)]
Merge branch 'akpm' (patches from Andrew)

Merge yet more updates from Andrew Morton:

 - a pile of minor fs fixes and cleanups

 - kexec updates

 - random misc fixes in various places: vmcore, rbtree, eventfd, ipc, seccomp.

 - a series of python-based kgdb helper scripts

* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (58 commits)
  seccomp: cap SECCOMP_RET_ERRNO data to MAX_ERRNO
  samples/seccomp: improve label helper
  ipc,sem: use current->state helpers
  scripts/gdb: disable pagination while printing from breakpoint handler
  scripts/gdb: define maintainer
  scripts/gdb: convert CpuList to generator function
  scripts/gdb: convert ModuleList to generator function
  scripts/gdb: use a generator instead of iterator for task list
  scripts/gdb: ignore byte-compiled python files
  scripts/gdb: port to python3 / gdb7.7
  scripts/gdb: add basic documentation
  scripts/gdb: add lx-lsmod command
  scripts/gdb: add class to iterate over CPU masks
  scripts/gdb: add lx_current convenience function
  scripts/gdb: add internal helper and convenience function for per-cpu lookup
  scripts/gdb: add get_gdbserver_type helper
  scripts/gdb: add internal helper and convenience function to retrieve thread_info
  scripts/gdb: add is_target_arch helper
  scripts/gdb: add helper and convenience function to look up tasks
  scripts/gdb: add task iteration class
  ...

9 years agoseccomp: cap SECCOMP_RET_ERRNO data to MAX_ERRNO
Kees Cook [Tue, 17 Feb 2015 21:48:00 +0000 (13:48 -0800)]
seccomp: cap SECCOMP_RET_ERRNO data to MAX_ERRNO

The value resulting from the SECCOMP_RET_DATA mask could exceed MAX_ERRNO
when setting errno during a SECCOMP_RET_ERRNO filter action.  This makes
sure we have a reliable value being set, so that an invalid errno will not
be ignored by userspace.

Signed-off-by: Kees Cook <keescook@chromium.org>
Reported-by: Dmitry V. Levin <ldv@altlinux.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Will Drewry <wad@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agosamples/seccomp: improve label helper
Kees Cook [Tue, 17 Feb 2015 21:47:58 +0000 (13:47 -0800)]
samples/seccomp: improve label helper

Fixes a potential corruption with uninitialized stack memory in the
seccomp BPF sample program.

[akpm@linux-foundation.org: coding-style fixlet]
Signed-off-by: Kees Cook <keescook@chromium.org>
Reported-by: Robert Swiecki <swiecki@google.com>
Tested-by: Robert Swiecki <swiecki@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoipc,sem: use current->state helpers
Davidlohr Bueso [Tue, 17 Feb 2015 21:47:55 +0000 (13:47 -0800)]
ipc,sem: use current->state helpers

Call __set_current_state() instead of assigning the new state directly.
These interfaces also aid CONFIG_DEBUG_ATOMIC_SLEEP environments, keeping
track of who changed the state.

Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoscripts/gdb: disable pagination while printing from breakpoint handler
Jan Kiszka [Tue, 17 Feb 2015 21:47:52 +0000 (13:47 -0800)]
scripts/gdb: disable pagination while printing from breakpoint handler

While reporting the (refreshed) list of modules on automatic updates we
may hit the page boundary of the output console and cause a stop if
pagination is enabled.  However, gdb does not accept user input while
running over the breakpoint handler.  So we get stuck, and the user is
forced to interrupt gdb.

Resolve this by disabling pagination during automatic symbol updates.  We
restore the user's configuration once done.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Borislav Petkov <bp@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoscripts/gdb: define maintainer
Jan Kiszka [Tue, 17 Feb 2015 21:47:49 +0000 (13:47 -0800)]
scripts/gdb: define maintainer

I'm proposing myself for keeping an eye on these scripts and integrating
contributions.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Borislav Petkov <bp@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>