platform/kernel/linux-arm64.git
10 years agonet: macb: Remove 'unlikely' optimization
Soren Brinkmann [Sun, 4 May 2014 22:43:01 +0000 (15:43 -0700)]
net: macb: Remove 'unlikely' optimization

Coverage data suggests that the unlikely case of receiving data while
the receive handler is running may not be that unlikely.
Coverage data after running iperf for a while:
    91320:  891: work_done = bp->macbgem_ops.mog_rx(bp, budget);
    91320:  892: if (work_done < budget) {
     2362:  893: napi_complete(napi);
        -:  894:
        -:  895: /* Packets received while interrupts were disabled */
     4724:  896: status = macb_readl(bp, RSR);
     2362:  897: if (unlikely(status)) {
      762:  898: if (bp->caps & MACB_CAPS_ISR_CLEAR_ON_WRITE)
      762:  899: macb_writel(bp, ISR, MACB_BIT(RCOMP));
        -:  900: napi_reschedule(napi);
        -:  901: } else {
     1600:  902: macb_writel(bp, IER, MACB_RX_INT_FLAGS);
        -:  903: }
        -:  904: }

Signed-off-by: Soren Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: macb: Re-enable RX interrupt only when RX is done
Soren Brinkmann [Sun, 4 May 2014 22:43:00 +0000 (15:43 -0700)]
net: macb: Re-enable RX interrupt only when RX is done

When data is received during the driver processing received data the
NAPI is re-scheduled. In that case the RX interrupt should not be
re-enabled.

Signed-off-by: Soren Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: macb: Clear interrupt flags
Soren Brinkmann [Sun, 4 May 2014 22:42:59 +0000 (15:42 -0700)]
net: macb: Clear interrupt flags

A few interrupt flags were not cleared in the ISR, resulting in a sytem
trapped in the ISR in cases one of those interrupts occurred. Clear all
flags to avoid such situations.

Signed-off-by: Soren Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: macb: Pass same size to DMA_UNMAP as used for DMA_MAP
Soren Brinkmann [Sun, 4 May 2014 22:42:58 +0000 (15:42 -0700)]
net: macb: Pass same size to DMA_UNMAP as used for DMA_MAP

Just as commit "net: macb: DMA-unmap full rx-buffer"
(48330e08fa168395b9fd9f369f06cca1df204361), pass the size that
was used for mapping the memory also to the unmap routine to
avoid warnings from the DMA_API.

Signed-off-by: Soren Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoip_tunnel: Set network header properly for IP_ECN_decapsulate()
Ying Cai [Sun, 4 May 2014 22:20:04 +0000 (15:20 -0700)]
ip_tunnel: Set network header properly for IP_ECN_decapsulate()

In ip_tunnel_rcv(), set skb->network_header to inner IP header
before IP_ECN_decapsulate().

Without the fix, IP_ECN_decapsulate() takes outer IP header as
inner IP header, possibly causing error messages or packet drops.

Note that this skb_reset_network_header() call was in this spot when
the original feature for checking consistency of ECN bits through
tunnels was added in eccc1bb8d4b4 ("tunnel: drop packet if ECN present
with not-ECT"). It was only removed from this spot in 3d7b46cd20e3
("ip_tunnel: push generic protocol handling to ip_tunnel module.").

Fixes: 3d7b46cd20e3 ("ip_tunnel: push generic protocol handling to ip_tunnel module.")
Reported-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Ying Cai <ycai@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net
David S. Miller [Mon, 5 May 2014 20:30:03 +0000 (16:30 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/jkirsher/net

Jeff Kirsher says:

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

This series contains updates to e1000e only.

David provides four fixes for e1000e, first is a workaround for a hardware
erratum on 82579 devices which experienced packet loss in gigabit and 100
speeds when interconnect between the PHY and MAC is exiting K1 power saving
state.  Second expands the scope of a workaround to include i217 and i218
parts as well to address over aggressive transmit behavior when connecting
at 10Mbs half-duplex.  Next is to resolve a reported link flap issue on
82579 parts which was root caused as an interoperability problem between
82579 and at least some Broadcom PHYs in the Energy Efficient Ethernet wake
mechanism.  Lastly, restricts the workaround of putting the PHY into MDIO
slow mode to access the PHY id to relevant parts since this issue has been
fixed on the newer hardware.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoe1000e: Restrict MDIO Slow Mode workaround to relevant parts
David Ertman [Thu, 1 May 2014 02:19:03 +0000 (02:19 +0000)]
e1000e: Restrict MDIO Slow Mode workaround to relevant parts

It has been determined that the workaround of putting the PHY into MDIO
slow mode to access the PHY id is not necessary with Lynx Point and newer
parts.  The issue that necessitated the workaround has been fixed on the
newer hardware.

We will maintains, as a last ditch attempt, the conversion to MDIO Slow
Mode in the failure branch when attempting to access the PHY id so as to
cover all contingencies.

Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
10 years agoe1000e: Fix issue with link flap on 82579
David Ertman [Thu, 1 May 2014 01:22:26 +0000 (01:22 +0000)]
e1000e: Fix issue with link flap on 82579

Several customers have reported a link flap issue on 82579. The symptoms
are random and intermittent link losses when 82579 is connected to specific
link partners. Issue has been root caused as interoperability problem
between 82579 and at least some Broadcom PHYs in the Energy Efficient
Ethernet wake mechanism.

To fix the issue, we are disabling the Phase Locked Loop shutdown in 100M
Low Power Idle.  This solution will cause an increase of power in 100M EEE
link. It will cost additional 28mW in this specific mode.

Cc: Lukasz Adamczuk <lukasz.adamczuk@intel.com>
Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
10 years agoe1000e: Expand workaround for 10Mb HD throughput bug
David Ertman [Tue, 22 Apr 2014 05:48:54 +0000 (05:48 +0000)]
e1000e: Expand workaround for 10Mb HD throughput bug

In commit 772d05c51c4f4896c120ad418b1e91144a2ac813 "e1000e: slow performance
between two 82579 connected via 10Mbit hub", a workaround was put into place
to address the overaggressive transmit behavior of 82579 parts when connecting
at 10Mbs half-duplex.

This same behavior is seen on i217 and i218 parts as well.  This patch expands
the original workaround to encompass these parts.

Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
10 years agoe1000e: Workaround for dropped packets in Gig/100 speeds on 82579
David Ertman [Tue, 22 Apr 2014 05:25:53 +0000 (05:25 +0000)]
e1000e: Workaround for dropped packets in Gig/100 speeds on 82579

This is a workaround for a HW erratum on 82579 devices.
Erratum is #23 in Intel 6 Series Chipset and Intel C200 Series Chipset
specification Update June 2013.

Problem: 82579 parts experience packet loss in Gig and 100 speeds
when interconnect between PHY and MAC is exiting K1 power saving state.
This was previously believed to only affect 1Gig speed, but has been observed
at 100Mbs also.

Workaround: Disable K1 for 82579 devices at Gig and 100 speeds.

Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
10 years agoMerge branch 'mlx4'
David S. Miller [Mon, 5 May 2014 19:49:29 +0000 (15:49 -0400)]
Merge branch 'mlx4'

Or Gerlitz says:

====================
This series contains fixes for 3.15-rc, mostly around SRIOV. The patches by Jack,
Matan and myself fix few issues related to mlx4 SRIOV support for RoCE and single
port VFs, and the patch from Eyal eliminates checking PCI caps for VFs which is misleading.

Patches done against the net tree, commit 014f1b2 "net: bonding: Fix format string
mismatch in bond_sysfs.c"

We'd be happy to get Eyal's patch queued in your -stable list for 3.14.y
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet/mlx4_core: Don't issue PCIe speed/width checks for VFs
Eyal Perry [Sun, 4 May 2014 14:07:25 +0000 (17:07 +0300)]
net/mlx4_core: Don't issue PCIe speed/width checks for VFs

Carrying out PCI speed/width checks through pcie_get_minimum_link()
on VFs yield wrong results, so remove them.

Fixes: b912b2f ('net/mlx4_core: Warn if device doesn't have enough PCI bandwidth')
Signed-off-by: Eyal Perry <eyalpe@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet/mlx4_core: Load the Eth driver first
Or Gerlitz [Sun, 4 May 2014 14:07:24 +0000 (17:07 +0300)]
net/mlx4_core: Load the Eth driver first

When running in SRIOV mode, VM that is assigned with a non-provisioned
Ethernet VFs get themselves a random mac when the Eth driver starts. In
this case, if the IB driver startup code that deals with RoCE runs first,
it will use a zero mac as the source mac for the Para-Virtual CM MADs
which is buggy. To handle that, we change the order of loading.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet/mlx4_core: Fix slave id computation for single port VF
Matan Barak [Sun, 4 May 2014 14:07:23 +0000 (17:07 +0300)]
net/mlx4_core: Fix slave id computation for single port VF

The code that deals with computing the slave id based on a given GID
gave wrong results when the number of single port VFs wasn't the
same for port 1 vs. port 2 and the relevant VF is single ported on
port 2. As a result, incoming CM MADs were dispatched to the wrong VF.
Fixed that and added documentation to clarify the computation steps.

Fixes: 449fc48 ('net/mlx4: Adapt code for N-Port VF')
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet/mlx4_core: Adjust port number in qp_attach wrapper when detaching
Jack Morgenstein [Sun, 4 May 2014 14:07:22 +0000 (17:07 +0300)]
net/mlx4_core: Adjust port number in qp_attach wrapper when detaching

When using single ported VFs and the VF is using port 2, we need
to adjust the port accordingly (change it from 1 to 2).

Fixes: 449fc48 ('net/mlx4: Adapt code for N-Port VF')
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Jack Morgenstein <jackm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: cdc_ncm: fix buffer overflow
Bjørn Mork [Fri, 2 May 2014 21:27:00 +0000 (23:27 +0200)]
net: cdc_ncm: fix buffer overflow

Commit 4d619f625a60 ("net: cdc_ncm: no point in filling up the NTBs
if we send ZLPs") changed the padding logic for devices with the ZLP
flag set.  This meant that frames of any size will be sent without
additional padding, except for the single byte added if the size is
a multiple of the USB packet size. But if the unpadded size is
identical to the maximum frame size, and the maximum size is a
multiplum of the USB packet size, then this one-byte padding will
overflow the buffer.

Prevent padding if already at maximum frame size, letting usbnet
transmit a ZLP instead in this case.

Fixes: 4d619f625a60 ("net: cdc_ncm: no point in filling up the NTBs if we send ZLPs")
Reported by: Yu-an Shih <yshih@nvidia.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoAltera TSE: ALTERA_TSE should depend on HAS_DMA
Geert Uytterhoeven [Fri, 2 May 2014 04:29:21 +0000 (06:29 +0200)]
Altera TSE: ALTERA_TSE should depend on HAS_DMA

If NO_DMA=y:

drivers/built-in.o: In function `altera_tse_probe':
altera_tse_main.c:(.text+0x25ec2e): undefined reference to `dma_set_mask'
altera_tse_main.c:(.text+0x25ec78): undefined reference to `dma_supported'
altera_tse_main.c:(.text+0x25ecb6): undefined reference to `dma_supported'
drivers/built-in.o: In function `sgdma_async_read':
altera_sgdma.c:(.text+0x25f620): undefined reference to `dma_sync_single_for_cpu'
drivers/built-in.o: In function `sgdma_uninitialize':
(.text+0x25f678): undefined reference to `dma_unmap_single'
drivers/built-in.o: In function `sgdma_uninitialize':
(.text+0x25f696): undefined reference to `dma_unmap_single'
drivers/built-in.o: In function `sgdma_initialize':
(.text+0x25f6f0): undefined reference to `dma_map_single'
drivers/built-in.o: In function `sgdma_initialize':
(.text+0x25f702): undefined reference to `dma_mapping_error'
drivers/built-in.o: In function `sgdma_tx_buffer':
(.text+0x25f92a): undefined reference to `dma_sync_single_for_cpu'
drivers/built-in.o: In function `sgdma_rx_status':
(.text+0x25fa24): undefined reference to `dma_sync_single_for_cpu'
make[3]: *** [vmlinux] Error 1

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Vince Bridgers <vbridgers2013@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agovsock: Make transport the proto owner
Andy King [Thu, 1 May 2014 22:20:43 +0000 (15:20 -0700)]
vsock: Make transport the proto owner

Right now the core vsock module is the owner of the proto family. This
means there's nothing preventing the transport module from unloading if
there are open sockets, which results in a panic. Fix that by allowing
the transport to be the owner, which will refcount it properly.

Includes version bump to 1.0.1.0-k

Passes checkpatch this time, I swear...

Acked-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Andy King <acking@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoMerge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville...
David S. Miller [Mon, 5 May 2014 17:06:01 +0000 (13:06 -0400)]
Merge branch 'for-davem' of git://git./linux/kernel/git/linville/wireless

John W. Linville says:

====================
pull request: wireless 2014-05-01

Please pull the following batch of fixes intended for the 3.15 stream!

For the Bluetooth bits, Gustavo says:

"Some fixes for 3.15. There is a revert for the intel driver, a new
device id, and two important SSP fixes from Johan."

On top of that...

Ben Hutchings gives us a fix for an unbalanced irq enable in an
rtl8192cu error path.

Colin Ian King provides an rtlwifi fix for an uninitialized variable.

Felix Fietkau brings a pair of ath9k fixes, one that corrects a
hardware initialization value and another that removes an (unnecessary)
flag that was being used in a way that led to a software tx queue
hang in ath9k.

Gertjan van Wingerde pushes a MAINTAINERS change to remove himself
from the rt2x00 maintainer team.

Hans de Goede fixes a brcmfmac firmware load hang.

Larry Finger changes rtlwifi to use the correct queue for V0 traffic
on rtl8192se.

Rajkumar Manoharan corrects a race in ath9k driver initialization.

Stanislaw Gruszka fixes an rt2x00 bug in which disabling beaconing
once on USB devices led to permanently disabling beaconing for those
devices.

Tim Harvey provides fixes for a pair of ath9k issues that can lead
to soft lockups in that driver.

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

Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: sched: lock imbalance in hhf qdisc
John Fastabend [Thu, 1 May 2014 16:23:06 +0000 (09:23 -0700)]
net: sched: lock imbalance in hhf qdisc

hhf_change() takes the sch_tree_lock and releases it but misses the
error cases. Fix the missed case here.

To reproduce try a command like this,

# tc qdisc change dev p3p2 root hhf quantum 40960 non_hh_weight 300000

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: mvmdio: Check for a valid interrupt instead of an error
Ezequiel Garcia [Wed, 30 Apr 2014 16:28:51 +0000 (13:28 -0300)]
net: mvmdio: Check for a valid interrupt instead of an error

The following commit:

commit 9ec36cafe43bf835f8f29273597a5b0cbc8267ef
Author: Rob Herring <robh@kernel.org>
Date:   Wed Apr 23 17:57:41 2014 -0500

    of/irq: do irq resolution in platform_get_irq

changed platform_get_irq() which now returns EINVAL and EPROBE_DEFER,
in addition to ENXIO. If there's no interrupt for mvmdio, platform_get_irq()
returns EINVAL, but we currently check only for ENXIO.

Fix this by looking for a positive integer, which is the proper way of
validating a virtual interrupt number.

While at it, add a proper handling for the deferral probe case.

Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Reviewed-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet phy: Check for aneg completion before setting state to PHY_RUNNING
Balakumaran Kannan [Thu, 24 Apr 2014 02:52:47 +0000 (08:22 +0530)]
net phy: Check for aneg completion before setting state to PHY_RUNNING

phy_state_machine should check whether auto-negotiatin is completed
before changing phydev->state from PHY_NOLINK to PHY_RUNNING. If
auto-negotiation is not completed phydev->state should be set to
PHY_AN.

Signed-off-by: Balakumaran Kannan <kumaran.4353@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
John W. Linville [Thu, 1 May 2014 15:23:21 +0000 (11:23 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless into for-davem

10 years agohyperv: Properly handle checksum offload
KY Srinivasan [Wed, 30 Apr 2014 18:58:25 +0000 (11:58 -0700)]
hyperv: Properly handle checksum offload

Do checksum offload only if the client of the driver wants checksum to be
offloaded.

In V1 version of this patch, I  addressed comments from
Stephen Hemminger <stephen@networkplumber.org> and
Eric Dumazet <eric.dumazet@gmail.com>.

In this version of the patch I have addressed comments from
David Miller.

This patch fixes a bug that is exposed in gateway scenarios.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agogianfar: Avoid unnecessary reg accesses in adjust_link()
Claudiu Manoil [Wed, 30 Apr 2014 11:27:21 +0000 (14:27 +0300)]
gianfar: Avoid unnecessary reg accesses in adjust_link()

For phy devices that don't issue interrupts upon link
state changes, phylib polls the link state resulting in
repeated calls to adjust_link(), even if the link state
didn't change.  As a result, some mac registers are
repeatedly read and written with the same values, which
is not ok.

To fix this, adjust_link() has been refactored to check
first whether the link state has changed and to take action
only if needed, updating mac registers and local state
variables.  The 'new_state' local flag, set if one of the
link params changed (link, speed or duplex), has been
rendered useless and removed by this refactoring.

Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agotcp_cubic: fix the range of delayed_ack
Liu Yu [Wed, 30 Apr 2014 09:34:09 +0000 (17:34 +0800)]
tcp_cubic: fix the range of delayed_ack

commit b9f47a3aaeab (tcp_cubic: limit delayed_ack ratio to prevent
divide error) try to prevent divide error, but there is still a little
chance that delayed_ack can reach zero. In case the param cnt get
negative value, then ratio+cnt would overflow and may happen to be zero.
As a result, min(ratio, ACK_RATIO_LIMIT) will calculate to be zero.

In some old kernels, such as 2.6.32, there is a bug that would
pass negative param, which then ultimately leads to this divide error.

commit 5b35e1e6e9c (tcp: fix tcp_trim_head() to adjust segment count
with skb MSS) fixed the negative param issue. However,
it's safe that we fix the range of delayed_ack as well,
to make sure we do not hit a divide by zero.

CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Liu Yu <allanyuliu@tencent.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agobpf_dbg: fix wrong register usage
Brendan Hickey [Wed, 30 Apr 2014 08:24:38 +0000 (10:24 +0200)]
bpf_dbg: fix wrong register usage

The AND instruction is erroneously using the X register instead
of the K register.

Signed-off-by: Brendan Hickey <bhickey@google.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoRevert "macvlan : fix checksums error when we are in bridge mode"
Vlad Yasevich [Tue, 29 Apr 2014 14:09:51 +0000 (10:09 -0400)]
Revert "macvlan : fix checksums error when we are in bridge mode"

This reverts commit 12a2856b604476c27d85a5f9a57ae1661fc46019.
The commit above doesn't appear to be necessary any more as the
checksums appear to be correctly computed/validated.

Additionally the above commit breaks kvm configurations where
one VM is using a device that support checksum offload (virtio) and
the other VM does not.
In this case, packets leaving virtio device will have CHECKSUM_PARTIAL
set.  The packets is forwarded to a macvtap that has offload features
turned off.  Since we use CHECKSUM_UNNECESSARY, the host does does not
update the checksum and thus a bad checksum is passed up to
the guest.

CC: Daniel Lezcano <daniel.lezcano@free.fr>
CC: Patrick McHardy <kaber@trash.net>
CC: Andrian Nord <nightnord@gmail.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Jason Wang <jasowang@redhat.com>
Signed-off-by: Vlad Yasevich <vyasevic@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agomactap: Fix checksum errors for non-gso packets in bridge mode
Vlad Yasevich [Tue, 29 Apr 2014 14:09:50 +0000 (10:09 -0400)]
mactap: Fix checksum errors for non-gso packets in bridge mode

The following is a problematic configuration:

 VM1: virtio-net device connected to macvtap0@eth0
 VM2: e1000 device connect to macvtap1@eth0

The problem is is that virtio-net supports checksum offloading
and thus sends the packets to the host with CHECKSUM_PARTIAL set.
On the other hand, e1000 does not support any acceleration.

For small TCP packets (and this includes the 3-way handshake),
e1000 ends up receiving packets that only have a partial checksum
set.  This causes TCP to fail checksum validation and to drop
packets.  As a result tcp connections can not be established.

Commit 3e4f8b787370978733ca6cae452720a4f0c296b8
macvtap: Perform GSO on forwarding path.
fixes this issue for large packets wthat will end up undergoing GSO.
This commit adds a check for the non-GSO case and attempts to
compute the checksum for partially checksummed packets in the
non-GSO case.

CC: Daniel Lezcano <daniel.lezcano@free.fr>
CC: Patrick McHardy <kaber@trash.net>
CC: Andrian Nord <nightnord@gmail.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Jason Wang <jasowang@redhat.com>
Signed-off-by: Vlad Yasevich <vyasevic@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agotcp: increment retransmit counters in tlp and fast open
Eric Dumazet [Tue, 29 Apr 2014 05:00:29 +0000 (22:00 -0700)]
tcp: increment retransmit counters in tlp and fast open

Both TLP and Fast Open call __tcp_retransmit_skb() instead of
tcp_retransmit_skb() to avoid changing tp->retrans_out.

This has the side effect of missing SNMP counters increments as well
as tcp_info tcpi_total_retrans updates.

Fix this by moving the stats increments of into __tcp_retransmit_skb()

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: sxgbe: Added set function for interrupt on complete
Byungho An [Tue, 29 Apr 2014 04:15:27 +0000 (13:15 +0900)]
net: sxgbe: Added set function for interrupt on complete

This patch adds set_rx_int_on_com function for interrupt when
dma is completed.

Signed-off-by: Byungho An <bh74.an@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: sxgbe: Added rxqueue enable function
Byungho An [Tue, 29 Apr 2014 04:15:17 +0000 (13:15 +0900)]
net: sxgbe: Added rxqueue enable function

This patch adds rxqueue enable function according to number of rxqueue
and adds rxqueue disable function for removing.

Signed-off-by: Byungho An <bh74.an@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: sxgbe: sw reset moved to probe function
Byungho An [Tue, 29 Apr 2014 04:15:15 +0000 (13:15 +0900)]
net: sxgbe: sw reset moved to probe function

This patch moves sw reset to probe function because
sw reset is needed early stage before open function.

Signed-off-by: Byungho An <bh74.an@samsung.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agocxgb4: Decode PCIe Gen3 link speed
Roland Dreier [Tue, 29 Apr 2014 00:36:20 +0000 (17:36 -0700)]
cxgb4: Decode PCIe Gen3 link speed

Add handling for " 8 GT/s" in print_port_info().

Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoAltera TSE: Add missing include to silence sparse warnings
Tobias Klauser [Mon, 28 Apr 2014 21:23:13 +0000 (23:23 +0200)]
Altera TSE: Add missing include to silence sparse warnings

This fixes the following sparse warnings:

drivers/net/ethernet/altera/altera_msgdma.c:23:5: warning: symbol 'msgdma_initialize' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:28:6: warning: symbol 'msgdma_uninitialize' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:32:6: warning: symbol 'msgdma_reset' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:77:6: warning: symbol 'msgdma_disable_rxirq' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:83:6: warning: symbol 'msgdma_enable_rxirq' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:89:6: warning: symbol 'msgdma_disable_txirq' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:95:6: warning: symbol 'msgdma_enable_txirq' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:101:6: warning: symbol 'msgdma_clear_rxirq' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:107:6: warning: symbol 'msgdma_clear_txirq' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:114:5: warning: symbol 'msgdma_tx_buffer' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:129:5: warning: symbol 'msgdma_tx_completions' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:154:5: warning: symbol 'msgdma_add_rx_desc' was not declared. Should it be static?
drivers/net/ethernet/altera/altera_msgdma.c:181:5: warning: symbol 'msgdma_rx_status' was not declared. Should it be static?

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Acked-by: Vince Bridgers <vbridgers2013@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoAltera TSE: Fix DMA secriptor length initialization
Tobias Klauser [Mon, 28 Apr 2014 21:23:01 +0000 (23:23 +0200)]
Altera TSE: Fix DMA secriptor length initialization

sgdma_descrip is a function name as well as the name of a struct. In
sgdma_initialize(), we should initialize the descriptor length field
with the actual length of a descriptor not with the size of the
function. In order to prevent such things from happening in the future,
rename the function to sgdma_setup_descrip().

Found by sparse which yields the following warning:

drivers/net/ethernet/altera/altera_sgdma.c:74:30: warning: expression using sizeof on a function

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agortl8192cu: Fix unbalanced irq enable in error path of rtl92cu_hw_init()
Ben Hutchings [Sat, 26 Apr 2014 20:59:04 +0000 (21:59 +0100)]
rtl8192cu: Fix unbalanced irq enable in error path of rtl92cu_hw_init()

Fixes: a53268be0cb9 ('rtlwifi: rtl8192cu: Fix too long disable of IRQs')
Cc: stable@vger.kernel.org
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
10 years agortlwifi: rtl8192se: Fix regression due to commit 1bf4bbb
Larry Finger [Fri, 25 Apr 2014 15:05:43 +0000 (10:05 -0500)]
rtlwifi: rtl8192se: Fix regression due to commit 1bf4bbb

Beginning with kernel 3.13, this driver fails on some systems. The problem
was bisected to:

Commit 1bf4bbb4024dcdab5e57634dd8ae1072d42a53ac
Author: Felix Fietkau <nbd@openwrt.org>
Title: mac80211: send control port protocol frames to the VO queue

There is noting wrong with the above commit. The regression occurs because
V0 queue on RTL8192SE cards uses priority 6, not the usual 7. The fix is to
modify the rtl8192se routine that sets the correct transmit queue.

Bug: https://bugzilla.kernel.org/show_bug.cgi?id=74541

Reported-by: Alex Miller <almiller_1@yahoo.co.uk>
Tested-by: Alex Miller <almiller_1@yahoo.co.uk>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: Stable <stable@vger.kernel.org> [3.13+]
Signed-off-by: John W. Linville <linville@tuxdriver.com>
10 years agoath9k: remove tid->paused flag
Felix Fietkau [Mon, 28 Apr 2014 16:32:12 +0000 (18:32 +0200)]
ath9k: remove tid->paused flag

There are some corner cases where the driver could get stuck with a full
tid queue that is paused, leading to a software tx queue hang.

Since the tx queueing rework, pausing per-tid queues on aggregation
session setup is no longer necessary. The driver will assign sequence
numbers to buffered frames when a new session is established, in order
to get the correct starting sequence number.

mac80211 prevents new frames from entering the queue during setup.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
10 years agoath9k_hw: do not lower ANI setting below default on AR913x
Felix Fietkau [Sun, 27 Apr 2014 14:37:39 +0000 (16:37 +0200)]
ath9k_hw: do not lower ANI setting below default on AR913x

When the amount of noise fluctuates strongly, low immunity settings
can sometimes disrupt signal detection on AR913x chips. When that
happens, no OFDM/CCK errors are reported anymore, and ANI tunes the
radio to the lowest immunity settings.
Usually rx/tx fails as well in that case.

To fix this, keep noise immunity settings at or above ANI default level,
which will keep radio parameters at or above INI values.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
10 years agonet: bonding: Fix format string mismatch in bond_sysfs.c
Masanari Iida [Mon, 28 Apr 2014 15:41:21 +0000 (00:41 +0900)]
net: bonding: Fix format string mismatch in bond_sysfs.c

Fix format string mismatch in bonding_show_min_links().

Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: ipv6: more places need LOOPBACK_IFINDEX for flowi6_iif
Julian Anastasov [Mon, 28 Apr 2014 07:51:56 +0000 (10:51 +0300)]
net: ipv6: more places need LOOPBACK_IFINDEX for flowi6_iif

To properly match iif in ip rules we have to provide
LOOPBACK_IFINDEX in flowi6_iif, not 0. Some ip6mr_fib_lookup
and fib6_rule_lookup callers need such fix.

Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: sctp: Don't transition to PF state when transport has exhausted 'Path.Max.Retrans'.
Karl Heiss [Fri, 25 Apr 2014 18:26:30 +0000 (14:26 -0400)]
net: sctp: Don't transition to PF state when transport has exhausted 'Path.Max.Retrans'.

Don't transition to the PF state on every strike after 'Path.Max.Retrans'.
Per draft-ietf-tsvwg-sctp-failover-03 Section 5.1.6:

   Additional (PMR - PFMR) consecutive timeouts on a PF destination
   confirm the path failure, upon which the destination transitions to the
   Inactive state.  As described in [RFC4960], the sender (i) SHOULD notify
   ULP about this state transition, and (ii) transmit heartbeats to the
   Inactive destination at a lower frequency as described in Section 8.3 of
   [RFC4960].

This also prevents sending SCTP_ADDR_UNREACHABLE to the user as the state
bounces between SCTP_INACTIVE and SCTP_PF for each subsequent strike.

Signed-off-by: Karl Heiss <kheiss@gmail.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoslip: fix spinlock variant
Oliver Hartkopp [Sat, 26 Apr 2014 19:18:32 +0000 (21:18 +0200)]
slip: fix spinlock variant

With commit cc9fa74e2a ("slip/slcan: added locking in wakeup function") a
formerly missing locking was added to slip.c and slcan.c by Andre Naujoks.

Alexander Stein contributed the fix 367525c8c2 ("can: slcan: Fix spinlock
variant") as the kernel lock debugging advised to use spin_lock_bh() instead
of just using spin_lock().

This fix has to be applied to the same code section in slip.c for the same
reason too.

Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoMerge branch 'qmi_wwan'
David S. Miller [Mon, 28 Apr 2014 02:05:22 +0000 (22:05 -0400)]
Merge branch 'qmi_wwan'

Bjørn Mork says:

====================
qmi_wwan: adding a few new devices

Adding some new and some old QMI devices.  A similar series for
usb-serial support will be posted as well.

These new device ID patches should also go to all maintained stable kernels
where they apply. I believe that should be stable-3.10 and newer.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: qmi_wwan: add a number of Dell devices
Bjørn Mork [Fri, 25 Apr 2014 17:00:34 +0000 (19:00 +0200)]
net: qmi_wwan: add a number of Dell devices

Dan writes:

"The Dell drivers use the same configuration for PIDs:

81A2: Dell Wireless 5806 Gobi(TM) 4G LTE Mobile Broadband Card
81A3: Dell Wireless 5570 HSPA+ (42Mbps) Mobile Broadband Card
81A4: Dell Wireless 5570e HSPA+ (42Mbps) Mobile Broadband Card
81A8: Dell Wireless 5808 Gobi(TM) 4G LTE Mobile Broadband Card
81A9: Dell Wireless 5808e Gobi(TM) 4G LTE Mobile Broadband Card

These devices are all clearly Sierra devices, but are also definitely
Gobi-based.  The A8 might be the MC7700/7710 and A9 is likely a MC7750.

>From DellGobi5kSetup.exe from the Dell drivers:

usbif0: serial/firmware loader?
usbif2: nmea
usbif3: modem/ppp
usbif8: net/QMI"

Reported-by: AceLan Kao <acelan.kao@canonical.com>
Reported-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: qmi_wwan: add a number of CMOTech devices
Bjørn Mork [Fri, 25 Apr 2014 17:00:33 +0000 (19:00 +0200)]
net: qmi_wwan: add a number of CMOTech devices

A number of older CMOTech modems are based on Qualcomm
chips and exporting a QMI/wwan function.

Reported-by: Lars Melin <larsm17@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: qmi_wwan: add Alcatel L800MA
Bjørn Mork [Fri, 25 Apr 2014 17:00:32 +0000 (19:00 +0200)]
net: qmi_wwan: add Alcatel L800MA

Device interface layout:
0: ff/ff/ff - serial
1: ff/00/00 - serial AT+PPP
2: ff/ff/ff - QMI/wwan
3: 08/06/50 - storage

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: qmi_wwan: add Olivetti Olicard 500
Bjørn Mork [Fri, 25 Apr 2014 17:00:31 +0000 (19:00 +0200)]
net: qmi_wwan: add Olivetti Olicard 500

Device interface layout:
0: ff/ff/ff - serial
1: ff/ff/ff - serial AT+PPP
2: 08/06/50 - storage
3: ff/ff/ff - serial
4: ff/ff/ff - QMI/wwan

Reported-by: Julio Araujo <julio.araujo@wllctel.com.br>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: qmi_wwan: add Sierra Wireless MC7305/MC7355
Bjørn Mork [Fri, 25 Apr 2014 17:00:30 +0000 (19:00 +0200)]
net: qmi_wwan: add Sierra Wireless MC7305/MC7355

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: qmi_wwan: add Sierra Wireless MC73xx
Bjørn Mork [Fri, 25 Apr 2014 17:00:29 +0000 (19:00 +0200)]
net: qmi_wwan: add Sierra Wireless MC73xx

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agonet: qmi_wwan: add Sierra Wireless EM7355
Bjørn Mork [Fri, 25 Apr 2014 17:00:28 +0000 (19:00 +0200)]
net: qmi_wwan: add Sierra Wireless EM7355

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agosctp: reset flowi4_oif parameter on route lookup
Xufeng Zhang [Fri, 25 Apr 2014 08:55:41 +0000 (16:55 +0800)]
sctp: reset flowi4_oif parameter on route lookup

commit 813b3b5db83 (ipv4: Use caller's on-stack flowi as-is
in output route lookups.) introduces another regression which
is very similar to the problem of commit e6b45241c (ipv4: reset
flowi parameters on route connect) wants to fix:
Before we call ip_route_output_key() in sctp_v4_get_dst() to
get a dst that matches a bind address as the source address,
we have already called this function previously and the flowi
parameters have been initialized including flowi4_oif, so when
we call this function again, the process in __ip_route_output_key()
will be different because of the setting of flowi4_oif, and we'll
get a networking device which corresponds to the inputted flowi4_oif
as the output device, this is wrong because we'll never hit this
place if the previously returned source address of dst match one
of the bound addresses.

To reproduce this problem, a vlan setting is enough:
  # ifconfig eth0 up
  # route del default
  # vconfig add eth0 2
  # vconfig add eth0 3
  # ifconfig eth0.2 10.0.1.14 netmask 255.255.255.0
  # route add default gw 10.0.1.254 dev eth0.2
  # ifconfig eth0.3 10.0.0.14 netmask 255.255.255.0
  # ip rule add from 10.0.0.14 table 4
  # ip route add table 4 default via 10.0.0.254 src 10.0.0.14 dev eth0.3
  # sctp_darn -H 10.0.0.14 -P 36422 -h 10.1.4.134 -p 36422 -s -I
You'll detect that all the flow are routed to eth0.2(10.0.1.254).

Signed-off-by: Xufeng Zhang <xufeng.zhang@windriver.com>
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoarc_emac: add clock handling
Heiko Stübner [Fri, 25 Apr 2014 08:06:13 +0000 (10:06 +0200)]
arc_emac: add clock handling

This adds ability for the arc_emac to really handle its supplying clock.
To get the needed clock-frequency either a real clock or the previous
clock-frequency property must be provided.

Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Tested-by: Max Schwarz <max.schwarz@online.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoarc_emac: fix probe error path
Heiko Stübner [Fri, 25 Apr 2014 08:03:29 +0000 (10:03 +0200)]
arc_emac: fix probe error path

The probe function at the moment only frees the netdev but does not disconnect
the phy or removes the mdio bus it registered.

Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agobridge: Handle IFLA_ADDRESS correctly when creating bridge device
Toshiaki Makita [Fri, 25 Apr 2014 08:01:18 +0000 (17:01 +0900)]
bridge: Handle IFLA_ADDRESS correctly when creating bridge device

When bridge device is created with IFLA_ADDRESS, we are not calling
br_stp_change_bridge_id(), which leads to incorrect local fdb
management and bridge id calculation, and prevents us from receiving
frames on the bridge device.

Reported-by: Tom Gundersen <teg@jklm.no>
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agocxgb4: Update Kconfig to include Chelsio T5 adapter
Hariprasad Shenai [Fri, 25 Apr 2014 12:30:12 +0000 (18:00 +0530)]
cxgb4: Update Kconfig to include Chelsio T5 adapter

Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoi40e: fix Timesync Tx interrupt handler code
Jacob Keller [Fri, 25 Apr 2014 01:05:03 +0000 (18:05 -0700)]
i40e: fix Timesync Tx interrupt handler code

This patch fixes the PTP Tx timestamp interrupt handler. The original
code misinterpreted the interrupt handler design. We were clearing the
ena_mask bit for the Timesync interrupts. This is done to indicate that
the interrupt will be handled in a scheduled work item (instead of
immediately) and that work item is responsible for re-enabling the
interrupts. However, the Tx timestamp was being handled immediately and
nothing was ever re-enabling it. This resulted in a single interrupt
working for the life of the driver.

This patch fixes the issue by instead clearing the bit from icr0 which
is used to indicate that the interrupt was immediately handled and can
be re-enabled right away. This patch also clears up a related issue due
to writing the PRTTSYN_STAT_0 register, which was unintentionally
clearing the cause bits for Timesync interrupts.

Change-ID: I057bd70d53c302f60fab78246989cbdfa469d83b
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Acked-by: Anjali Singhai Jain <anjali.singhai@intel.com>
Acked-by: Shannon Nelson <shannon.nelson@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoMerge tag 'linux-can-fixes-for-3.15-20140424' of git://gitorious.org/linux-can/linux-can
David S. Miller [Sat, 26 Apr 2014 16:29:19 +0000 (12:29 -0400)]
Merge tag 'linux-can-fixes-for-3.15-20140424' of git://gitorious.org/linux-can/linux-can

Marc Kleine-Budde says:

====================
this is a pull request for net/master, for the v3.15 release cycle, consisting
of 26 patches.

Thomas Gleixner contributes 21 patches for the c_can driver, which address
several shortcomings in the driver like hardware initialisation, concurrency,
message ordering and poor performance. Two patches Oliver Hartkopp, one adds a
missing lock to the sja1000_isa driver, the other one fixes the return value in
the generic bit time configuration function. And finally a patch by Alexander
Stein, that fixes the slcan driver to use the correct spinlock variant.

To make it 26 patches, Wolfgang Grandegger patch for the c_can_pci
driver, which enables the bus master only for MSI and a patch by
Wolfram Sang, which converts the 'instance' in the c_can driver to the
proper type.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoMerge branch 'altera_tse'
David S. Miller [Sat, 26 Apr 2014 16:27:14 +0000 (12:27 -0400)]
Merge branch 'altera_tse'

Vince Bridgers says:

====================
This series of patches addresses a handful of issues found in testing
and reported by users of the Altera Triple Speed Ethernet soft IP.

The patches address the following issues (in summary)

1) The SGDMA soft IP was found to incorrectly process receive packets
   when the target physical address of the receive buffer was on
   a boundary that's not 32-bit aligned. One of the patches addresses
   this issue.
2) The pause quanta was not being set by the driver, one patch of this
   series sets the pause quanta to the IEEE defined default value
   since the hardware reset value is 0.
3) An issue in a error recovery path of the probe routine caused a
   kernel panic in the event a phy was probed and could not be found.
   A patch addresses this issue.
4) A change was made to the driver name for Ethtool support, and
   comments added to support an addition to Ethtool to support
   the Altera Triple Speed Ethernet controller.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoAltera TSE: Change driver name used by Ethtool
Vince Bridgers [Thu, 24 Apr 2014 21:58:11 +0000 (16:58 -0500)]
Altera TSE: Change driver name used by Ethtool

This patch changes the name used by Ethtool to something more
conventional in preparation for TSE Ethtool register dump
support to be added in the near future.

Signed-off-by: Vince Bridgers <vbridgers2013@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoAltera TSE: Fix Panic in probe routine when phy probe fails
Vince Bridgers [Thu, 24 Apr 2014 21:58:10 +0000 (16:58 -0500)]
Altera TSE: Fix Panic in probe routine when phy probe fails

This patch addresses a fault in the error recovery path of the probe
routine where the netdev structure was not being unregistered properly
leading to a panic only when the phy probe failed.

Abbreviated panic stack seen is as follows:

(free_netdev+0xXX) from (altera_tse_probe+0xXX)
(altera_tse_probe+0xXX) from (platform_drv_probe+0xXX)
(platform_drv_probe+0xXX) from (driver_probe_device+0xXX)
(driver_probe_device+0xXX) from (__driver_attach+0xXX)
(__driver_attach+0xXX) from (bus_for_each_dev+0xXX)
(bus_for_each_dev+0xXX) from (driver_attach+0xXX)
(driver_attach+0xXX) from (bus_add_driver+0xXX)
(bus_add_driver+0xXX) from (driver_register+0xXX)
(driver_register+0xXX) from (__platform_driver_register+0xXX)
(__platform_driver_register+0xXX) from (altera_tse_driver_init+0xXX)
(altera_tse_driver_init+0xXX) from (do_one_initcall+0xXX)
(do_one_initcall+0xXX) from (kernel_init_freeable+0xXX)
(kernel_init_freeable+0xXX) from (kernel_init+0xXX)
(kernel_init+0xXX) from (ret_from_fork+0xXX)

Signed-off-by: Vince Bridgers <vbridgers2013@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoAltera TSE: Set the Pause Quanta value to the IEEE default value
Vince Bridgers [Thu, 24 Apr 2014 21:58:09 +0000 (16:58 -0500)]
Altera TSE: Set the Pause Quanta value to the IEEE default value

This patch initializes the pause quanta set for transmitted pause frames
to the IEEE specified default of 0xffff.

Signed-off-by: Vince Bridgers <vbridgers2013@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoAltera TSE: Work around unaligned DMA receive packet issue with Altera SGDMA
Vince Bridgers [Thu, 24 Apr 2014 21:58:08 +0000 (16:58 -0500)]
Altera TSE: Work around unaligned DMA receive packet issue with Altera SGDMA

This patch works around a recently discovered unaligned receive dma problem
with the Altera SGMDA. The Altera SGDMA component cannot be configured to
DMA data to unaligned addresses for receive packet operations from the
Triple Speed Ethernet component because of a potential data transfer
corruption that can occur. This patch addresses this issue by
utilizing the shift 16 bits feature of the Altera Triple Speed Ethernet
component and modifying the receive buffer physical addresses accordingly
such that the target receive DMA address is always aligned on a 32-bit
boundary.

Signed-off-by: Vince Bridgers <vbridgers2013@gmail.com>
Tested-by: Matthew Gerlach <mgerlach@altera.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agoMerge branch 'bnx2x-net'
David S. Miller [Sat, 26 Apr 2014 16:16:18 +0000 (12:16 -0400)]
Merge branch 'bnx2x-net'

Yuval Mintz says:

====================
bnx2x: SRIOV bug fixes

This series contains 3 SRIOV bug fixes, 2 of which are regressions starting
with commit 2dc33bbc "bnx2x: Remove the sriov VFOP mechanism".
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agobnx2x: Fix failure to configure VF multicast filters
Narender Kumar [Thu, 24 Apr 2014 16:29:54 +0000 (19:29 +0300)]
bnx2x: Fix failure to configure VF multicast filters

Commit 2dc33bbc "bnx2x: Remove the sriov VFOP mechanism" caused a regression,
preventing VFs from configuring multicast filters.

Signed-off-by: Naredner Kumar <narender.kumar@qlogic.com>
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>
10 years agobnx2x: Fix vlan credit issues for VFs
Yuval Mintz [Thu, 24 Apr 2014 16:29:53 +0000 (19:29 +0300)]
bnx2x: Fix vlan credit issues for VFs

Starting with commit 2dc33bbc "bnx2x: Remove the sriov VFOP mechanism",
the bnx2x started enforcing vlan credits for all vlan configurations.
This exposed 2 issues:
  - Vlan credits are not returned once a VF is removed; this causes a leak
    of credits, and eventually will lead to VFs with no vlan credits.
  - A vlan credit must be set aside for the Hypervisor to use, and should
    not be visible to the VF.

Although linux VFs at the moment do not support vlan configuration [from the
VF side] which causes them to be resilient to this sort of issue, Windows VF
over linux hypervisors might fail to load as the vlan credits become depleted.

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>
10 years agobnx2x: Memory leak during VF removal
Yuval Mintz [Thu, 24 Apr 2014 16:29:52 +0000 (19:29 +0300)]
bnx2x: Memory leak during VF removal

When removing a VF interface, the driver fails to release that VF's mailbox
and bulletin board allocated memory.

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>
10 years agoMerge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetoot...
John W. Linville [Fri, 25 Apr 2014 12:57:17 +0000 (08:57 -0400)]
Merge branch 'for-upstream' of git://git./linux/kernel/git/bluetooth/bluetooth

10 years agoBluetooth: Add support for Lite-on [04ca:3007]
Mohammed Habibulla [Thu, 17 Apr 2014 18:37:13 +0000 (11:37 -0700)]
Bluetooth: Add support for Lite-on [04ca:3007]

Add support for the AR9462 chip

T:  Bus=01 Lev=01 Prnt=01 Port=03 Cnt=03 Dev#=  3 Spd=12   MxCh= 0
D:  Ver= 1.10 Cls=e0(wlcon) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=04ca ProdID=3007 Rev= 0.01
C:* #Ifs= 2 Cfg#= 1 Atr=e0 MxPwr=100mA
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=82(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=   0 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=   0 Ivl=1ms
I:  If#= 1 Alt= 1 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=   9 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=   9 Ivl=1ms
I:  If#= 1 Alt= 2 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=  17 Ivl=1ms
E:  Ad=03(O) 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=83(I) Atr=01(Isoc) MxPS=  25 Ivl=1ms
E:  Ad=03(O) 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=83(I) Atr=01(Isoc) MxPS=  33 Ivl=1ms
E:  Ad=03(O) 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=83(I) Atr=01(Isoc) MxPS=  49 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=  49 Ivl=1ms

Signed-off-by: Mohammed Habibulla <moch@chromium.org>
Cc: stable@vger.kernel.org
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
10 years agoBluetooth: Fix redundant encryption request for reauthentication
Johan Hedberg [Fri, 11 Apr 2014 19:02:32 +0000 (12:02 -0700)]
Bluetooth: Fix redundant encryption request for reauthentication

When we're performing reauthentication (in order to elevate the
security level from an unauthenticated key to an authenticated one) we
do not need to issue any encryption command once authentication
completes. Since the trigger for the encryption HCI command is the
ENCRYPT_PEND flag this flag should not be set in this scenario.
Instead, the REAUTH_PEND flag takes care of all necessary steps for
reauthentication.

Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Cc: stable@vger.kernel.org
10 years agoBluetooth: Fix triggering BR/EDR L2CAP Connect too early
Johan Hedberg [Fri, 11 Apr 2014 19:02:31 +0000 (12:02 -0700)]
Bluetooth: Fix triggering BR/EDR L2CAP Connect too early

Commit 1c2e004183178 introduced an event handler for the encryption key
refresh complete event with the intent of fixing some LE/SMP cases.
However, this event is shared with BR/EDR and there we actually want to
act only on the auth_complete event (which comes after the key refresh).

If we do not do this we may trigger an L2CAP Connect Request too early
and cause the remote side to return a security block error.

Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Cc: stable@vger.kernel.org
10 years agoRevert "Bluetooth: Enable autosuspend for Intel Bluetooth device"
Marcel Holtmann [Tue, 22 Apr 2014 21:04:16 +0000 (14:04 -0700)]
Revert "Bluetooth: Enable autosuspend for Intel Bluetooth device"

This reverts commit d2bee8fb6e18f6116aada39851918473761f7ab1.

Enabling autosuspend for Intel Bluetooth devices has been shown to not
work reliable. It does work for some people with certain combinations
of USB host controllers, but for others it puts the device to sleep and
it will not wake up for any event.

These events can be important ones like HCI Inquiry Complete or HCI
Connection Request. The events will arrive as soon as you poke the
device with a new command, but that is not something we can do in
these cases.

Initially there were patches to the xHCI USB controller that fixed
this for some people, but not for all. This could be well a problem
somewhere in the USB subsystem or in the USB host controllers or
just plain a hardware issue somewhere. At this moment we just do
not know and the only safe action is to revert this patch.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Cc: Tedd Ho-Jeong An <tedd.an@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
10 years agobrcmfmac: Fix brcmf_chip_ai_coredisable not applying reset bits to BCMA_IOCTL
Hans de Goede [Wed, 23 Apr 2014 10:20:55 +0000 (12:20 +0200)]
brcmfmac: Fix brcmf_chip_ai_coredisable not applying reset bits to BCMA_IOCTL

brcmfmac has been broken on my cubietruck with a BCM43362:

brcmfmac: brcmf_chip_recognition: found AXI chip: BCM43362, rev=1
brcmfmac: brcmf_c_preinit_dcmds: Firmware version = wl0:
        Apr 22 2013 14:50:00 version 5.90.195.89.6 FWID 01-b30a427d

since commit 53036261033: "brcmfmac: update core reset and disable routines".

The problem is that since this commit brcmf_chip_ai_resetcore no longer sets
BCMA_IOCTL itself before bringing the core out of reset, instead relying on
brcmf_chip_ai_coredisable to do so. But brcmf_chip_ai_coredisable is a nop
of the chip is already in reset. This patch modifies brcmf_chip_ai_coredisable
to always set BCMA_IOCTL even if the core is already in reset.

This fixes brcmfmac hanging in firmware loading on my board.

Cc: stable@vger.kernel.org # v3.14
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
10 years agoath9k: fix race in setting ATH_OP_INVALID
Rajkumar Manoharan [Wed, 23 Apr 2014 09:37:57 +0000 (15:07 +0530)]
ath9k: fix race in setting ATH_OP_INVALID

The commit "ath9k: move sc_flags to ath_common" moved setting
ATH_OP_INVALID flag below ieee80211_register_hw. This is causing
the flag never being cleared randomly as the drv_start is called
prior to setting flag. Fix this by setting the flag prior to
register_hw.

Signed-off-by: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
10 years agoipv6: fib: fix fib dump restart
Kumar Sundararajan [Thu, 24 Apr 2014 13:48:53 +0000 (09:48 -0400)]
ipv6: fib: fix fib dump restart

When the ipv6 fib changes during a table dump, the walk is
restarted and the number of nodes dumped are skipped. But the existing
code doesn't advance to the next node after a node is skipped. This can
cause the dump to loop or produce lots of duplicates when the fib
is modified during the dump.

This change advances the walk to the next node if the current node is
skipped after a restart.

Signed-off-by: Kumar Sundararajan <kumar@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
10 years agocan: slcan: Fix spinlock variant
Alexander Stein [Tue, 15 Apr 2014 14:51:04 +0000 (16:51 +0200)]
can: slcan: Fix spinlock variant

slc_xmit is called within softirq context and locks sl->lock, but
slcan_write_wakeup is not softirq context, so we need to use
spin_[un]lock_bh!
Detected using kernel lock debugging mechanism.

Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
Acked-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: fix return value from can_get_bittiming()
Oliver Hartkopp [Wed, 23 Apr 2014 20:10:55 +0000 (22:10 +0200)]
can: fix return value from can_get_bittiming()

When trying to set a data bitrate on non CAN FD devices the 'ip' tool
answers with:

RTNETLINK answers: Unknown error 524

Rename '-ENOTSUPP' to '-EOPNOTSUPP' so that 'ip' answers correctly:

       RTNETLINK answers: Operation not supported

Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: sja1000_isa: add locking for indirect register access mode
Oliver Hartkopp [Tue, 15 Apr 2014 17:30:00 +0000 (19:30 +0200)]
can: sja1000_isa: add locking for indirect register access mode

When accessing the SJA1000 controller registers in the indirect access mode,
writing the register number and reading/writing the data has to be an atomic
attempt.

As the sja1000_isa driver is an old style driver with a fixed number of
instances the locking variable depends on the same index like all the other
configuration elements given on the module command line.

As a positive side effect dev->dev_id is populated by the instance index,
which was missing in 3e66d0138c05d9 ("can: populate netdev::dev_id for udev
discrimination").

Reported-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can_pci: enable PCI bus master only for MSI
Wolfgang Grandegger [Thu, 3 Apr 2014 18:41:24 +0000 (20:41 +0200)]
can: c_can_pci: enable PCI bus master only for MSI

Coverity complains that c_can_pci_probe() calls pci_enable_msi() without
checking the result:

CID 712278 (#1 of 1): Unchecked return value (CHECKED_RETURN) 3. check_return:
Calling pci_enable_msi_block without checking return value (as is done
elsewhere 88 out of 105 times).
 88        pci_enable_msi(pdev);

This is CID 712278.

Signed-off-by: Wolfgang Grandegger <wg@grandegger.com>
Reported-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: use proper type for 'instance'
Wolfram Sang [Thu, 17 Apr 2014 08:57:18 +0000 (10:57 +0200)]
can: c_can: use proper type for 'instance'

Commit 6439fbce1075 (can: c_can: fix error checking of priv->instance in
probe()) found the warning but applied a suboptimal solution. Since, both
pdev->id and of_alias_get_id() return integers, it makes sense to convert the
variable to an integer and avoid the cast.

Signed-off-by: Wolfram Sang <wsa@sang-engineering.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Speed up tx buffer invalidation
Thomas Gleixner [Fri, 11 Apr 2014 08:13:22 +0000 (08:13 +0000)]
can: c_can: Speed up tx buffer invalidation

It's suffcient to kill the TXIE bit in the message control register
even if the documentation of C and D CAN says that it's not allowed to
do that while MSGVAL is set. Reality tells a different story and this
change gives us another 2% of CPU back for not waiting on I/O.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Remove tx locking
Thomas Gleixner [Fri, 11 Apr 2014 08:13:22 +0000 (08:13 +0000)]
can: c_can: Remove tx locking

Mark suggested to use one IF for the softirq and the other for the
xmit function to avoid the xmit lock.

That requires to write the frame into the interface first, then handle
the echo skb and store the dlc before committing the TX request to the
message ram.

We use an atomic to handle the active buffers instead of reading the
MSGVAL register as thats way faster especially on PCH/x86.

Suggested-by: Mark <mark5@del-llc.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Use proper u32 variables in c_can_write_msg_object()
Thomas Gleixner [Fri, 11 Apr 2014 08:13:21 +0000 (08:13 +0000)]
can: c_can: Use proper u32 variables in c_can_write_msg_object()

Instead of obfuscating the code by artificial 16 bit splits use the
proper 32 bit assignments and split the result when writing to the
interface.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Cleanup c_can_write_msg_object()
Thomas Gleixner [Fri, 11 Apr 2014 08:13:21 +0000 (08:13 +0000)]
can: c_can: Cleanup c_can_write_msg_object()

Remove the MASK from the TX transfer side.

Make the code readable and get rid of the annoying IFX_WRITE_XXX_16BIT
macros which are just obfuscating the code.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Cleanup c_can_msg_obj_put/get()
Thomas Gleixner [Fri, 11 Apr 2014 08:13:20 +0000 (08:13 +0000)]
can: c_can: Cleanup c_can_msg_obj_put/get()

Sigh!

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Cleanup c_can_inval_msg_object()
Thomas Gleixner [Fri, 11 Apr 2014 08:13:19 +0000 (08:13 +0000)]
can: c_can: Cleanup c_can_inval_msg_object()

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Cleanup setup of receive buffers
Thomas Gleixner [Fri, 11 Apr 2014 08:13:18 +0000 (08:13 +0000)]
can: c_can: Cleanup setup of receive buffers

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Cleanup c_can_read_msg_object()
Thomas Gleixner [Fri, 11 Apr 2014 08:13:18 +0000 (08:13 +0000)]
can: c_can: Cleanup c_can_read_msg_object()

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Cleanup irq enable/disable
Thomas Gleixner [Fri, 11 Apr 2014 08:13:17 +0000 (08:13 +0000)]
can: c_can: Cleanup irq enable/disable

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Work around C_CAN RX wreckage
Thomas Gleixner [Fri, 11 Apr 2014 08:13:17 +0000 (08:13 +0000)]
can: c_can: Work around C_CAN RX wreckage

Alexander reported that the new optimized handling of the RX fifo
causes random packet loss on Intel PCH C_CAN hardware.

After a few fruitless debugging sessions I got hold of a PCH (eg20t)
afflicted system. That machine does not have the CAN interface wired
up, but it was possible to reproduce the issue with the HW loopback
mode.

As Alexander observed correctly, clearing the NewDat flag along with
reading out the message buffer causes that issue on C_CAN, while D_CAN
handles that correctly.

Instead of restoring the original message buffer handling horror the
following workaround solves the issue:

    transfer buffer to IF without clearing the NewDat
    handle the message
    clear NewDat bit

That's similar to the original code but conditional for C_CAN.

I really wonder why all user manuals (C_CAN, Intel PCH and some more)
recommend to clear the NewDat bit right away. The knows it all Oracle
operated by Gurgle does not unearth any useful information either. I
simply cannot believe that we are the first to uncover that HW issue.

Reported-and-tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Disable rx split as workaround
Thomas Gleixner [Fri, 11 Apr 2014 08:13:16 +0000 (08:13 +0000)]
can: c_can: Disable rx split as workaround

The RX buffer split causes packet loss in the hardware:

What happens is:

RX Packet 1 --> message buffer 1 (newdat bit is not cleared)
RX Packet 2 --> message buffer 2 (newdat bit is not cleared)
RX Packet 3 --> message buffer 3 (newdat bit is not cleared)
RX Packet 4 --> message buffer 4 (newdat bit is not cleared)
RX Packet 5 --> message buffer 5 (newdat bit is not cleared)
RX Packet 6 --> message buffer 6 (newdat bit is not cleared)
RX Packet 7 --> message buffer 7 (newdat bit is not cleared)
RX Packet 8 --> message buffer 8 (newdat bit is not cleared)

Clear newdat bit in message buffer 1
Clear newdat bit in message buffer 2
Clear newdat bit in message buffer 3
Clear newdat bit in message buffer 4
Clear newdat bit in message buffer 5
Clear newdat bit in message buffer 6
Clear newdat bit in message buffer 7
Clear newdat bit in message buffer 8

Now if during that clearing of newdat bits, a new message comes in,
the HW gets confused and drops it.

It does not matter how many of them you clear. I put a delay between
clear of buffer 1 and buffer 2 which was long enough that the message
should have been queued either in buffer 1 or buffer 9. But it did not
show up anywhere. The next message ended up in buffer 1. So the
hardware lost a packet of course without telling it via one of the
error handlers.

That does not happen on all clear newdat bit events. I see one of 10k
packets dropped in the scenario which allows us to reproduce. But the
trace looks always the same.

Not splitting the RX Buffer avoids the packet loss but can cause
reordering. It's hard to trigger, but it CAN happen.

With that mode we use the HW as it was probably designed for. We read
from the buffer 1 upwards and clear the buffer as we get the
message. That's how all microcontrollers use it. So I assume that the
way we handle the buffers was never really tested. According to the
public documentation it should just work :)

Let the user decide which evil is the lesser one.

[ Oliver Hartkopp: Provided a sane config option and help text and
  made me switch to favour potential and unlikely reordering over
  packet loss ]

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Get rid of pointless interrupts
Thomas Gleixner [Fri, 11 Apr 2014 08:13:15 +0000 (08:13 +0000)]
can: c_can: Get rid of pointless interrupts

The driver handles pointlessly TWO interrupts per packet. The reason
is that it enables the status interrupt which fires for each rx and tx
packet and it enables the per message object interrupts as well.

The status interrupt merily acks or in case of D_CAN ignores the TX/RX
state and then the message object interrupt fires.

The message objects interrupts are only useful if all message objects
have hardware filters activated.

But we don't have that and its not simple to implement in that driver
without rewriting it completely.

So we can ditch the message object interrupts and handle the RX/TX
right away from the status interrupt. Instead of TWO we handle ONE.

Note: We must keep the TXIE/RXIE bits in the message buffers because
the status interrupt alone is not reliable enough in corner cases.

If we ever have the need for HW filtering, then this code needs a
complete overhaul and we can think about it then. For now we prefer a
lower interrupt load.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Avoid status register update for D_CAN
Thomas Gleixner [Fri, 11 Apr 2014 08:13:15 +0000 (08:13 +0000)]
can: c_can: Avoid status register update for D_CAN

On D_CAN the RXOK, TXOK and LEC bits are cleared/set on read of the
status register. No need to update them.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Simplify buffer reenabling
Thomas Gleixner [Fri, 11 Apr 2014 08:13:14 +0000 (08:13 +0000)]
can: c_can: Simplify buffer reenabling

Instead of writing to the message object we can simply clear the
NewDat bit with the get method.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Always update error stats
Thomas Gleixner [Fri, 11 Apr 2014 08:13:13 +0000 (08:13 +0000)]
can: c_can: Always update error stats

If the allocation of the error skb fails, we still want to see the
error statistics.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Fix berr reporting
Thomas Gleixner [Fri, 11 Apr 2014 08:13:13 +0000 (08:13 +0000)]
can: c_can: Fix berr reporting

Reading the LEC type with

  return (mode & ENABLED) && (status & LEC_MASK);

is not guaranteed to return (status & LEC_MASK) if the enabled bit in
mode is set. It's guaranteed to return 0 or !=0.

Remove the inline function and call unconditionally into the
berr_handling code and return early when the reporting is disabled.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Handle state change correctly
Thomas Gleixner [Fri, 11 Apr 2014 08:13:12 +0000 (08:13 +0000)]
can: c_can: Handle state change correctly

If the allocation of an error skb fails, the state change handling
returns w/o doing any work. That leaves the interface in a wreckaged
state as the internal status is wrong.

Split the interface handling and the skb handling.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Do not access skb after net_receive_skb()
Thomas Gleixner [Fri, 11 Apr 2014 08:13:12 +0000 (08:13 +0000)]
can: c_can: Do not access skb after net_receive_skb()

There is no guarantee that the skb is in the same state after calling
net_receive_skb(). It might be freed or reused. Not really harmful as
its a read access, except you turn on the proper debugging options
which catch a use after free.

The whole can subsystem is full of this. Copy and paste ....

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
10 years agocan: c_can: Make bus off interrupt disable logic work
Thomas Gleixner [Fri, 11 Apr 2014 08:13:11 +0000 (08:13 +0000)]
can: c_can: Make bus off interrupt disable logic work

The state change handler is called with device interrupts disabled
already. So no point in disabling them again when we enter bus off
state.

But what's worse is that we reenable the interrupts at the end of NAPI
poll unconditionally. So c_can_start() which is called from the
restart timer can trigger interrupts which confuse the hell out of the
half reinitialized driver/hw.

Remove the pointless device interrupt disable in the BUS_OFF handler
and prevent reenabling the device interrupts at the end of the poll
routine when the current state is BUS_OFF.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>