Replace a check of magic number 4095 with VLAN_N_VID. This
makes it obvious that a later check against VLAN_N_VID is
always true and can be removed.
Change-ID: I28998f127a61a529480ce63d8a07e266f6c63b7b
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This label is unnecessary, as are jumping to a block that checks aq_ret
and then immediately skipping it and returning. So just jump straight to
the error_param and remove this unnecessary label.
Also use goto error_param even in the last check for style consistency.
Change-ID: If487c7d10c4048e37c594e5eca167693aaed45f6
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This change makes it so that we are much more robust about defining what we
can and cannot offload. Previously we were performing no checks. This
should bring us up to parity with the i40e PF driver.
In addition the device only supports GSO as long as the MSS is 64 or
greater. We were not checking this so an MSS less than that was resulting
in Tx hangs.
Change-ID: If533553ec92fc6ba694eab6ac81fdaf3004f3592
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This change makes it so that we are much more robust about defining what we
can and cannot offload. Previously we were just checking for the L4 tunnel
header length, however there are other fields we should be verifying as
there are multiple scenarios in which we cannot perform hardware offloads.
In addition the device only supports GSO as long as the MSS is 64 or
greater. We were not checking this so an MSS less than that was resulting
in Tx hangs.
Change-ID: I5e2fd5f3075c73601b4b36327b771c64fcb6c31b
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
We want to set the other "err" variable here so that we can return it
later. My version of GCC misses this issue but I caught it with a
static checker.
Fixes: 9f70eb339f ("net: ethernet: renesas: ravb: fix fixed-link phydev leaks")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Reviewed-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When streaming a lot of data and the RZ/A1 can't keep up, some status bits
will get set that are not being checked or cleared which cause the
following messages and the Ethernet driver to stop working. This
patch fixes that issue.
irq 21: nobody cared (try booting with the "irqpoll" option)
handlers:
[<c036b71c>] sh_eth_interrupt
Disabling IRQ #21
Fixes: db893473d3 ("sh_eth: Add support for r7s72100")
Signed-off-by: Chris Brandt <chris.brandt@renesas.com>
Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
__bcmgenet_tx_reclaim() and bcmgenet_free_rx_buffers() are not using the
same struct device during unmap that was used for the map operation,
which makes DMA-API debugging warn about it. Fix this by always using
&priv->pdev->dev throughout the driver, using an identical device
reference for all map/unmap calls.
Fixes: 1c1008c793 ("net: bcmgenet: add main driver file")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Armada 3700 is a new ARMv8 SoC from Marvell using same network controller
as older Armada 370/38x/XP. There are however some differences that
needed taking into account when adding support for it:
* open default MBUS window to 4GB of DRAM - Armada 3700 SoC's Mbus
configuration for network controller has to be done on two levels:
global and per-port. The first one is inherited from the
bootloader. The latter can be opened in a default way, leaving
arbitration to the bus controller. Hence filled mbus_dram_target_info
structure is not needed
* make per-CPU operation optional - Recent patches adding RSS and XPS
support for Armada 38x/XP enabled per-CPU operation of the controller
by default. Contrary to older SoC's Armada 3700 SoC's network
controller is not capable of per-CPU processing due to interrupt lines'
connectivity. This patch restores non-per-CPU operation, which is now
optional and depends on neta_armada3700 flag value in mvneta_port
structure. In order not to complicate the code, separate interrupt
subroutine is implemented.
For now, on the Armada 3700, RSS is disabled as the current
implementation depend on the per cpu interrupts.
[gregory.clement@free-electrons.com: extract from a larger patch, replace
some ifdef and port to net-next for v4.10]
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Actually only the mvneta_bm support is not 64-bits compatible.
The mvneta code itself can run on 64-bits architecture.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Until now the virtual address of the received buffer were stored in the
cookie field of the rx descriptor. However, this field is 32-bits only
which prevents to use the driver on a 64-bits architecture.
With this patch the virtual address is stored in an array not shared with
the hardware (no more need to use the DMA API). Thanks to this, it is
possible to use cache contrary to the access of the rx descriptor member.
The change is done in the swbm path only because the hwbm uses the cookie
field, this also means that currently the hwbm is not usable in 64-bits.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Reviewed-by: Jisheng Zhang <jszhang@marvell.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For HWBM all buffers are allocated in mvneta_bm_construct() and in runtime
they are put into descriptors by hardware. There is no need to fill them
at this point.
Suggested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For small frame reuse the phys_addr variable instead of accessing the
uncacheable value in the rx descriptor.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
My recent commit to get more precise rx/tx counters in ndo_get_stats64()
can lead to crashes at device dismantle, as Jesper found out.
We must prevent mlx4_en_fold_software_stats() trying to access
tx/rx rings if they are deleted.
Fix this by adding a test against priv->port_up in
mlx4_en_fold_software_stats()
Calling mlx4_en_fold_software_stats() from mlx4_en_stop_port()
allows us to eventually broadcast the latest/current counters to
rtnetlink monitors.
Fixes: 40931b8511 ("mlx4: give precise rx/tx bytes/packets counters")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-and-bisected-by: Jesper Dangaard Brouer <brouer@redhat.com>
Tested-by: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Tariq Toukan <tariqt@mellanox.com>
Cc: Saeed Mahameed <saeedm@dev.mellanox.co.il>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Transmit queue timeout issue is seen in two cases
- Due to a race condition btw setting stop_queue at xmit()
and checking for stopped_queue in NAPI poll routine, at times
transmission from a SQ comes to a halt. This is fixed
by using barriers and also added a check for SQ free descriptors,
incase SQ is stopped and there are only CQE_RX i.e no CQE_TX.
- Contrary to an assumption, a HW errata where HW doesn't stop transmission
even though there are not enough CQEs available for a CQE_TX is
not fixed in T88 pass 2.x. This results in a Qset error with
'CQ_WR_FULL' stalling transmission. This is fixed by adjusting
RXQ's RED levels for CQ level such that there is always enough
space left for CQE_TXs.
Signed-off-by: Sunil Goutham <sgoutham@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When ndo_setup_tc is called with an egress_dev flag set, it means that
the ndo call was executed on the mirred action (egress) device and not
on the ingress device.
In order to support this kind of ndo_setup_tc call, and insert the
correct decap rule to the hardware, the uplink device on the same eswitch
should be found.
Currently, we use this resolution between the mirred device and the
uplink on the same eswitch to offload vxlan shared device decap rules.
Signed-off-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace the representor private data to a net_device pointer holding the
representor netdevice, instead of void pointer holding mlx5e_priv.
It will be used by a new eswitch service function, returning the uplink representor
netdevice.
Signed-off-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The VF Representor udp tunnel ndo entries were removed by mistake,
return them.
Fixes: 370bad0f9a ('net/mlx5e: Support HW (offloaded) and SW counters for SRIOV switchdev mode')
Signed-off-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds out of order packet handling for hardware offloaded
iSCSI. Out of order packet handling requires driver buffer allocation
and assistance.
Signed-off-by: Arun Easi <arun.easi@cavium.com>
Signed-off-by: Yuval Mintz <yuval.mintz@cavium.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
This adds the backbone required for the various HW initalizations
which are necessary for the iSCSI driver (qedi) for QLogic FastLinQ
4xxxx line of adapters - FW notification, resource initializations, etc.
Signed-off-by: Arun Easi <arun.easi@cavium.com>
Signed-off-by: Yuval Mintz <yuval.mintz@cavium.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
The driver already uses its private lock for synchronization between xmit
and xmit completion handler making the additional use of the xmit_lock
unnecessary.
Furthermore the driver does not set NETIF_F_LLTX resulting in xmit to be
called with the xmit_lock held and then taking the private lock while xmit
completion handler does the reverse, first take the private lock, then the
xmit_lock.
Fix these issues by not taking the xmit_lock in the tx completion handler.
Signed-off-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
An explicit dma sync for device directly after mapping as well as an
explicit dma sync for cpu directly before unmapping is unnecessary and
costly on the hotpath. So remove these calls.
Signed-off-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is already using the %pM printf extension; might as well also use
%ph to make the code smaller.
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
When CONFIG_INET is disabled, the new selftest results in a link
error:
drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.o: In function `mlx5e_test_loopback':
en_selftest.c:(.text.mlx5e_test_loopback+0x2ec): undefined reference to `ip_send_check'
en_selftest.c:(.text.mlx5e_test_loopback+0x34c): undefined reference to `udp4_hwcsum'
This hides the specific test in that configuration.
Fixes: 0952da791c ("net/mlx5e: Add support for loopback selftest")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
After 326fe02d1e ("net/mlx4_en: protect ring->xdp_prog with rcu_read_lock"),
the rcu_read_lock() in bpf_prog_run_xdp() is superfluous, since callers
need to hold rcu_read_lock() already to make sure BPF program doesn't
get released in the background.
Thus, drop it from bpf_prog_run_xdp(), as it can otherwise be misleading.
Still keeping the bpf_prog_run_xdp() is useful as it allows for grepping
in XDP supported drivers and to keep the typecheck on the context intact.
For mlx4, this means we don't have a double rcu_read_lock() anymore. nfp can
just make use of bpf_prog_run_xdp(), too. For qede, just move rcu_read_lock()
out of the helper. When the driver gets atomic replace support, this will
move to call-sites eventually.
mlx5 needs actual fixing as it has the same issue as described already in
326fe02d1e ("net/mlx4_en: protect ring->xdp_prog with rcu_read_lock"),
that is, we're under RCU bh at this time, BPF programs are released via
call_rcu(), and call_rcu() != call_rcu_bh(), so we need to properly mark
read side as programs can get xchg()'ed in mlx5e_xdp_set() without queue
reset.
Fixes: 86994156c7 ("net/mlx5e: XDP fast RX drop bpf programs support")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Handling flow encap entry should be inside tc del flow
and is only relevant for offloaded eswitch TC rules.
Fixes: 11a457e9b6c1 ("net/mlx5e: Add basic TC tunnel set action for SRIOV offloads")
Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Change the function that deletes offloaded TC rule to get
struct mlx5e_tc_flow instance which contains both the flow
handle and flow attributes. This is a cleanup needed for
downstream patches, it doesn't change any functionality.
Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
According to the reverse unwinding principle, on delete time we should
first handle deletion of the steering rule and later handle the vlan
deletion from the eswitch.
Fixes: 8b32580df1 ("net/mlx5e: Add TC vlan action for SRIOV offloads")
Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We will never find a flow with the same cookie as cls_flower always
allocates a new flow and the cookie is the allocated memory address.
Fixes: e3a2b7ed01 ("net/mlx5e: Support offload cls_flower with drop action")
Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In Striding RQ implementation, we used a single UMR
(User-Mode Memory Registration) memory key for all RQs.
When the product of RQs number*size gets high, we hit a
limitation of u16 field size in FW.
Here we move to using a UMR memory key per RQ, so we can
scale to any number of rings, with the maximum buffer
size in each.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In next patch we are going to create a UMR MKey per RQ, we need
mlx5e_create_umr_mkey declared before mlx5e_create_rq.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add new type of struct mlx5_frag_buf which is used to allocate fragmented
buffers rather than contiguous, and make the Completion Queues (CQs) use
it as they are big (default of 2MB per CQ in Striding RQ).
This fixes the failures of type:
"mlx5e_open_locked: mlx5e_open_channels failed, -12"
due to dma_zalloc_coherent insufficient contiguous coherent memory to
satisfy the driver's request when the user tries to setup more or larger
rings.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reported-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure to deregister and free any fixed-link phy registered during
probe on probe errors and on driver unbind by adding a new glue helper
function.
Drop the of-node reference taken in the same path also on late probe
errors (and not just on driver unbind) by moving the put from
stmmac_dvr_remove() to the new helper.
Fixes: 277323814e ("stmmac: add fixed-link device-tree support")
Fixes: 4613b279be ("ethernet: stmicro: stmmac: add missing of_node_put
after calling of_parse_phandle")
Signed-off-by: Johan Hovold <johan@kernel.org>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix the OF-helper function header to reflect that the function no longer
has a platform-data parameter.
Fixes: b0003ead75 ("stmmac: make stmmac_probe_config_dt return the
platform data struct")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure to disable clocks before returning on late probe errors.
Fixes: 566e825162 ("net: stmmac: add a glue driver for the Amlogic
Meson 8b / GXBB DWMAC")
Signed-off-by: Johan Hovold <johan@kernel.org>
Acked-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure to call any exit() callback to undo the effect of init()
before returning on late probe errors.
Fixes: cf3f047b9a ("stmmac: move hw init in the probe (v2)")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure to disable runtime PM, power down the PHY, and disable clocks
before returning on late probe errors.
Fixes: 27ffefd2d1 ("stmmac: dwmac-rk: create a new probe function")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure to disable clocks before returning on late probe errors.
Fixes: 8387ee21f9 ("stmmac: dwmac-sti: turn setup callback into a
probe function")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure to call stmmac_dvr_remove() before returning on late probe
errors so that memory is freed, clocks are disabled, and the netdev is
deregistered before its resources go away.
Fixes: 3c201b5a84 ("net: stmmac: socfpga: Remove re-registration of
reset controller")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for the (optional) SGMII PCS functionality of the Altera
TSE MAC. If the phy-mode is set to 'sgmii' then we attempt to discover
and initialise the PCS so that the MAC can communicate to the PHY.
The PCS IP block provides a scratch register for testing presence of
the PCS, which is mapped into one of the two MDIO spaces present in
the MAC's register space. Once we have determined that the scratch
register is functioning, we attempt to initialise the PCS to
auto-negotiate an SGMII link with the PHY. There is no need to monitor
or manage the SGMII link beyond this, since the normal PHY MDIO will
then be used to monitor the media layer.
The Altera TSE MAC has only one way in which it can be configured with an
SGMII PCS, and as such, this patch only looks to the phy-mode to select
whether or not to attempt to initialise the PCS registers. During
initialisation, we report the PCS's equivalent of a PHY ID register.
This can be parameterised during the IP instantiation and is often left
as '0x00000000' which is not an error.
Signed-off-by: Neill Whillans <neill.whillans@codethink.co.uk>
Reviewed-by: Daniel Silverstone <daniel.silverstone@codethink.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the case of IPIP and SIT tunnel frames the outer transport header
offset is actually set to the same offset as the inner transport header.
This results in the lco_csum call not doing any checksum computation over
the inner IPv4/v6 header data.
In order to account for that I am updating the code so that we determine
the location to start the checksum ourselves based on the location of the
IPv4 header and the length.
Fixes: b83e30104b ("ixgbe/ixgbevf: Add support for GSO partial")
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the case of IPIP and SIT tunnel frames the outer transport header
offset is actually set to the same offset as the inner transport header.
This results in the lco_csum call not doing any checksum computation over
the inner IPv4/v6 header data.
In order to account for that I am updating the code so that we determine
the location to start the checksum ourselves based on the location of the
IPv4 header and the length.
Fixes: e10715d3e9 ("igb/igbvf: Add support for GSO partial")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It's no longer used now that Falcon is gone.
Also remove a reference in a comment to an ioctl that doesn't exist.
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Easy enough for Falcon users to enable it when making oldconfig.
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Defalconisation removed one of the string arguments, but missed the
corresponding %s.
Fixes: 5a6681e22c ("sfc: separate out SFC4000 ("Falcon") support into new sfc-falcon driver")
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
netif_set_real_num_tx/rx_queues() are required to be called with rtnl_lock
taken, otherwise ASSERT_RTNL() warning will be triggered - which happens
now during System resume from suspend:
cpsw_resume()
|- cpsw_ndo_open()
|- netif_set_real_num_tx/rx_queues()
|- ASSERT_RTNL();
Hence, fix it by surrounding cpsw_ndo_open() by rtnl_lock/unlock() calls.
Cc: Dave Gerlach <d-gerlach@ti.com>
Cc: Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
Fixes: commit e05107e6b7 ("net: ethernet: ti: cpsw: add multi queue support")
Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
Tested-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In alloc_cmd_box(), pci_pool_alloc() followed by memset will be
replaced by pci_pool_zalloc()
Signed-off-by: Souptick joarder <jrdr.linux@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Check budget fullness only after it's updated and update
channel mask only once to keep budget balance between channels.
It's also needed for farther changes.
Signed-off-by: Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
