IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
Misc fixes all over the place.
Revert of virtio used length validation series: the approach taken does
not seem to work, breaking too many guests in the process. We'll need to
do length validation using some other approach.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
-----BEGIN PGP SIGNATURE-----
iQFDBAABCAAtFiEEXQn9CHHI+FuUyooNKB8NuNKNVGkFAmGe0sEPHG1zdEByZWRo
YXQuY29tAAoJECgfDbjSjVRp8WEH/imDIq1iduDeAuvFnmrm5eEO9w3wzXCT4NiG
8Pla241FzQ1pEFEAne16KP0+SlLhj7P0oc5FR8vkYvxxuyneDbCzcS2M1kYMOpA1
ry28PuObAnekzE/WXxvC031ozB5Zb/FL54gmw+/1EdAOdMGL0CdQ1aJxREBHRTBo
p4ZHr83GA2D2C/IyKCsgQ8cB9ZrMqImTQQ4vRD89HoFBp+GH2u2Di1iyXEWuOqdI
n1+7M9jjbyW8A+N1bkOicpShS/6UcyJQOOcg8kvUQOV6srVkYhfaiWC/CbOP2g73
8PKK+/K2Htf92s6RdvDUPSKmvqGR/4KPZWPtWThXBYXGgWul0uI=
=q6tO
-----END PGP SIGNATURE-----
Merge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost
Pull vhost,virtio,vdpa bugfixes from Michael Tsirkin:
"Misc fixes all over the place.
Revert of virtio used length validation series: the approach taken
does not seem to work, breaking too many guests in the process. We'll
need to do length validation using some other approach"
[ This merge also ends up reverting commit f7a36b03a732 ("vsock/virtio:
suppress used length validation"), which came in through the
networking tree in the meantime, and was part of that whole used
length validation series - Linus ]
* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
vdpa_sim: avoid putting an uninitialized iova_domain
vhost-vdpa: clean irqs before reseting vdpa device
virtio-blk: modify the value type of num in virtio_queue_rq()
vhost/vsock: cleanup removing `len` variable
vhost/vsock: fix incorrect used length reported to the guest
Revert "virtio_ring: validate used buffer length"
Revert "virtio-net: don't let virtio core to validate used length"
Revert "virtio-blk: don't let virtio core to validate used length"
Revert "virtio-scsi: don't let virtio core to validate used buffer length"
Use the architecture independent Kconfig option PAGE_SIZE_LESS_THAN_64KB
to indicate that VMXNET3 requires a page size smaller than 64kB.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Current driver version is able to handle only one bridge at time.
Configuring two bridges on two different ports would end up shorting this
bridges by HW. To reproduce it:
ip l a name br0 type bridge
ip l a name br1 type bridge
ip l s dev br0 up
ip l s dev br1 up
ip l s lan1 master br0
ip l s dev lan1 up
ip l s lan2 master br1
ip l s dev lan2 up
Ping on lan1 and get response on lan2, which should not happen.
This happened, because current driver version is storing one global "Port VLAN
Membership" and applying it to all ports which are members of any
bridge.
To solve this issue, we need to handle each port separately.
This patch is dropping the global port member storage and calculating
membership dynamically depending on STP state and bridge participation.
Note: STP support was broken before this patch and should be fixed
separately.
Fixes: c2e866911e25 ("net: dsa: microchip: break KSZ9477 DSA driver into two files")
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Link: https://lore.kernel.org/r/20211126123926.2981028-1-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The driver doesn't support RX timestamping for non-PTP packets, but it
declares that it does. Restrict the reported RX filters to PTP v2 over
L2 and over L4.
Fixes: 4e3b0468e6d7 ("net: mscc: PTP Hardware Clock (PHC) support")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
IEEE 1588 support was declared too soon for the Ocelot switch. Out of
reset, this switch does not apply any special treatment for PTP packets,
i.e. when an event message is received, the natural tendency is to
forward it by MAC DA/VLAN ID. This poses a problem when the ingress port
is under a bridge, since user space application stacks (written
primarily for endpoint ports, not switches) like ptp4l expect that PTP
messages are always received on AF_PACKET / AF_INET sockets (depending
on the PTP transport being used), and never being autonomously
forwarded. Any forwarding, if necessary (for example in Transparent
Clock mode) is handled in software by ptp4l. Having the hardware forward
these packets too will cause duplicates which will confuse endpoints
connected to these switches.
So PTP over L2 barely works, in the sense that PTP packets reach the CPU
port, but they reach it via flooding, and therefore reach lots of other
unwanted destinations too. But PTP over IPv4/IPv6 does not work at all.
This is because the Ocelot switch have a separate destination port mask
for unknown IP multicast (which PTP over IP is) flooding compared to
unknown non-IP multicast (which PTP over L2 is) flooding. Specifically,
the driver allows the CPU port to be in the PGID_MC port group, but not
in PGID_MCIPV4 and PGID_MCIPV6. There are several presentations from
Allan Nielsen which explain that the embedded MIPS CPU on Ocelot
switches is not very powerful at all, so every penny they could save by
not allowing flooding to the CPU port module matters. Unknown IP
multicast did not make it.
The de facto consensus is that when a switch is PTP-aware and an
application stack for PTP is running, switches should have some sort of
trapping mechanism for PTP packets, to extract them from the hardware
data path. This avoids both problems:
(a) PTP packets are no longer flooded to unwanted destinations
(b) PTP over IP packets are no longer denied from reaching the CPU since
they arrive there via a trap and not via flooding
It is not the first time when this change is attempted. Last time, the
feedback from Allan Nielsen and Andrew Lunn was that the traps should
not be installed by default, and that PTP-unaware switching may be
desired for some use cases:
https://patchwork.ozlabs.org/project/netdev/patch/20190813025214.18601-5-yangbo.lu@nxp.com/
To address that feedback, the present patch adds the necessary packet
traps according to the RX filter configuration transmitted by user space
through the SIOCSHWTSTAMP ioctl. Trapping is done via VCAP IS2, where we
keep 5 filters, which are amended each time RX timestamping is enabled
or disabled on a port:
- 1 for PTP over L2
- 2 for PTP over IPv4 (UDP ports 319 and 320)
- 2 for PTP over IPv6 (UDP ports 319 and 320)
The cookie by which these filters (invisible to tc) are identified is
strategically chosen such that it does not collide with the filters used
for the ocelot-8021q tagging protocol by the Felix driver, or with the
MRP traps set up by the Ocelot library.
Other alternatives were considered, like patching user space to do
something, but there are so many ways in which PTP packets could be made
to reach the CPU, generically speaking, that "do what?" is a very valid
question. The ptp4l program from the linuxptp stack already attempts to
do something: it calls setsockopt(IP_ADD_MEMBERSHIP) (and
PACKET_ADD_MEMBERSHIP, respectively) which translates in both cases into
a dev_mc_add() on the interface, in the kernel:
https://github.com/richardcochran/linuxptp/blob/v3.1.1/udp.c#L73https://github.com/richardcochran/linuxptp/blob/v3.1.1/raw.c
Reality shows that this is not sufficient in case the interface belongs
to a switchdev driver, as dev_mc_add() does not show the intention to
trap a packet to the CPU, but rather the intention to not drop it (it is
strictly for RX filtering, same as promiscuous does not mean to send all
traffic to the CPU, but to not drop traffic with unknown MAC DA). This
topic is a can of worms in itself, and it would be great if user space
could just stay out of it.
On the other hand, setting up PTP traps privately within the driver is
not new by any stretch of the imagination:
https://elixir.bootlin.com/linux/v5.16-rc2/source/drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.c#L833https://elixir.bootlin.com/linux/v5.16-rc2/source/drivers/net/dsa/hirschmann/hellcreek.c#L1050https://elixir.bootlin.com/linux/v5.16-rc2/source/include/linux/dsa/sja1105.h#L21
So this is the approach taken here as well. The difference here being
that we prepare and destroy the traps per port, dynamically at runtime,
as opposed to driver init time, because apparently, PTP-unaware
forwarding is a use case.
Fixes: 4e3b0468e6d7 ("net: mscc: PTP Hardware Clock (PHC) support")
Reported-by: Po Liu <po.liu@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
VCAP (Versatile Content Aware Processor) is the TCAM-based engine behind
tc flower offload on ocelot, among other things. The ingress port mask
on which VCAP rules match is present as a bit field in the actual key of
the rule. This means that it is possible for a rule to be shared among
multiple source ports. When the rule is added one by one on each desired
port, that the ingress port mask of the key must be edited and rewritten
to hardware.
But the API in ocelot_vcap.c does not allow for this. For one thing,
ocelot_vcap_filter_add() and ocelot_vcap_filter_del() are not symmetric,
because ocelot_vcap_filter_add() works with a preallocated and
prepopulated filter and programs it to hardware, and
ocelot_vcap_filter_del() does both the job of removing the specified
filter from hardware, as well as kfreeing it. That is to say, the only
option of editing a filter in place, which is to delete it, modify the
structure and add it back, does not work because it results in
use-after-free.
This patch introduces ocelot_vcap_filter_replace, which trivially
reprograms a VCAP entry to hardware, at the exact same index at which it
existed before, without modifying any list or allocating any memory.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The ocelot driver, when asked to timestamp all receiving packets, 1588
v1 or NTP, says "nah, here's 1588 v2 for you".
According to this discussion:
https://patchwork.kernel.org/project/netdevbpf/patch/20211104133204.19757-8-martin.kaistra@linutronix.de/#24577647
drivers that downgrade from a wider request to a narrower response (or
even a response where the intersection with the request is empty) are
buggy, and should return -ERANGE instead. This patch fixes that.
Fixes: 4e3b0468e6d7 ("net: mscc: PTP Hardware Clock (PHC) support")
Suggested-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently, HNS3 driver doesn't clear the reset flags of components after
successfully executing reset, it causes userspace info of
"Components reset" and "Components not reset" is incorrect.
So fix this problem by clear corresponding reset flag after reset process.
Fixes: ddccc5e368a3 ("net: hns3: add support for triggering reset by ethtool")
Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently, when user queries page pool info by debugfs command
"cat page_pool_info", the cnt of allocated page for page pool may be
incorrect because of memory inconsistency problem caused by compiler
optimization.
So this patch uses READ_ONCE() to read value of pages_state_hold_cnt to
fix this problem.
Fixes: 850bfb912a6d ("net: hns3: debugfs add support dumping page pool info")
Signed-off-by: Hao Chen <chenhao288@hisilicon.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When page pool is not enabled, its address value is still NULL and page
pool should not be accessed, so add a check for it.
Fixes: 850bfb912a6d ("net: hns3: debugfs add support dumping page pool info")
Signed-off-by: Hao Chen <chenhao288@hisilicon.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When PF is set to multi-TCs and configured mapping relationship between
priorities and TCs, the hardware will active these settings for this PF
and its VFs.
In this case when VF just uses one TC and its rx packets contain priority,
and if the priority is not mapped to TC0, as other TCs of VF is not valid,
hardware always put this kind of packets to the queue 0. It cause this kind
of packets of VF can not be used RSS function.
To fix this problem, set tc mode of all unused TCs of VF to the setting of
TC0, then rx packet with priority which map to unused TC will be direct to
TC0.
Fixes: e2cb1dec9779 ("net: hns3: Add HNS3 VF HCL(Hardware Compatibility Layer) Support")
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
If the variable 'p_bit->flags' is always 0,
the loop condition is always 0.
The variable 'j' may be greater than or equal to 32.
At this time, the array 'p_aeu->bits[32]' may be out
of bound.
Signed-off-by: zhangyue <zhangyue1@kylinos.cn>
Link: https://lore.kernel.org/r/20211125113610.273841-1-zhangyue1@kylinos.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The Tx queues were not disabled in situations where the driver needed to
stop the interface to apply a new configuration. This could result in a
kernel panic when doing any of the 3 following actions:
* reconfiguring the number of queues (ethtool -L)
* reconfiguring the size of the ring buffers (ethtool -G)
* installing/removing an XDP program (ip l set dev ethX xdp)
Prevent the panic by making sure netif_tx_disable is called when stopping
an interface.
Without this patch, the following kernel panic can be observed when doing
any of the actions above:
Unable to handle kernel paging request at virtual address ffff80001238d040
[....]
Call trace:
dwmac4_set_addr+0x8/0x10
dev_hard_start_xmit+0xe4/0x1ac
sch_direct_xmit+0xe8/0x39c
__dev_queue_xmit+0x3ec/0xaf0
dev_queue_xmit+0x14/0x20
[...]
[ end trace 0000000000000002 ]---
Fixes: 5fabb01207a2d ("net: stmmac: Add initial XDP support")
Fixes: aa042f60e4961 ("net: stmmac: Add support to Ethtool get/set ring parameters")
Fixes: 0366f7e06a6be ("net: stmmac: add ethtool support for get/set channels")
Signed-off-by: Yannick Vignon <yannick.vignon@nxp.com>
Link: https://lore.kernel.org/r/20211124154731.1676949-1-yannick.vignon@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The issue happened randomly in runtime. The message "Link is Down" is
popped but soon it recovered to "Link is Up".
The "Link is Down" results from the incorrect read data for reading the
PHY register via MDIO bus. The correct sequence for reading the data
shall be:
1. fire the command
2. wait for command done (this step was missing)
3. wait for data idle
4. read data from data register
Cc: stable@vger.kernel.org
Fixes: f160e99462c6 ("net: phy: Add mdio-aspeed")
Reviewed-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Dylan Hung <dylan_hung@aspeedtech.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://lore.kernel.org/r/20211125024432.15809-1-dylan_hung@aspeedtech.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Oleksandr brought a bug report where netpoll causes trace
messages in the log on igb.
Danielle brought this back up as still occurring, so we'll try
again.
[22038.710800] ------------[ cut here ]------------
[22038.710801] igb_poll+0x0/0x1440 [igb] exceeded budget in poll
[22038.710802] WARNING: CPU: 12 PID: 40362 at net/core/netpoll.c:155 netpoll_poll_dev+0x18a/0x1a0
As Alex suggested, change the driver to return work_done at the
exit of napi_poll, which should be safe to do in this driver
because it is not polling multiple queues in this single napi
context (multiple queues attached to one MSI-X vector). Several
other drivers contain the same simple sequence, so I hope
this will not create new problems.
Fixes: 16eb8815c235 ("igb: Refactor clean_rx_irq to reduce overhead and improve performance")
Reported-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Reported-by: Danielle Ratson <danieller@nvidia.com>
Suggested-by: Alexander Duyck <alexander.duyck@gmail.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Tested-by: Danielle Ratson <danieller@nvidia.com>
Link: https://lore.kernel.org/r/20211123204000.1597971-1-jesse.brandeburg@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
On mv88e6xxx 1G/2.5G PCS, the SerDes register 4.2001.2 has the following
description:
This register bit indicates when link was lost since the last
read. For the current link status, read this register
back-to-back.
Thus to get current link state, we need to read the register twice.
But doing that in the link change interrupt handler would lead to
potentially ignoring link down events, which we really want to avoid.
Thus this needs to be solved in phylink's resolve, by retriggering
another resolve in the event when PCS reports link down and previous
link was up, and by re-reading PCS state if the previous link was down.
The wrong value is read when phylink requests change from sgmii to
2500base-x mode, and link won't come up. This fixes the bug.
Fixes: 9525ae83959b ("phylink: add phylink infrastructure")
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
On PHY state change the phylink_resolve() function can read stale
information from the MAC and report incorrect link speed and duplex to
the kernel message log.
Example with a Marvell 88X3310 PHY connected to a SerDes port on Marvell
88E6393X switch:
- PHY driver triggers state change due to PHY interface mode being
changed from 10gbase-r to 2500base-x due to copper change in speed
from 10Gbps to 2.5Gbps, but the PHY itself either hasn't yet changed
its interface to the host, or the interrupt about loss of SerDes link
hadn't arrived yet (there can be a delay of several milliseconds for
this), so we still think that the 10gbase-r mode is up
- phylink_resolve()
- phylink_mac_pcs_get_state()
- this fills in speed=10g link=up
- interface mode is updated to 2500base-x but speed is left at 10Gbps
- phylink_major_config()
- interface is changed to 2500base-x
- phylink_link_up()
- mv88e6xxx_mac_link_up()
- .port_set_speed_duplex()
- speed is set to 10Gbps
- reports "Link is Up - 10Gbps/Full" to dmesg
Afterwards when the interrupt finally arrives for mv88e6xxx, another
resolve is forced in which we get the correct speed from
phylink_mac_pcs_get_state(), but since the interface is not being
changed anymore, we don't call phylink_major_config() but only
phylink_mac_config(), which does not set speed/duplex anymore.
To fix this, we need to force the link down and trigger another resolve
on PHY interface change event.
Fixes: 9525ae83959b ("phylink: add phylink infrastructure")
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Usage of phy_ethtool_get_link_ksettings() in the link status change
handler isn't needed, and in combination with the referenced change
it results in a deadlock. Simply remove the call and replace it with
direct access to phydev->speed. The duplex argument of
lan743x_phy_update_flowcontrol() isn't used and can be removed.
Fixes: c10a485c3de5 ("phy: phy_ethtool_ksettings_get: Lock the phy for consistency")
Reported-by: Alessandro B Maurici <abmaurici@gmail.com>
Tested-by: Alessandro B Maurici <abmaurici@gmail.com>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/40e27f76-0ba3-dcef-ee32-a78b9df38b0f@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This reverts commit 816625c13652cef5b2c49082d652875da6f2ad7a.
Attempts to validate length in the core did not work out.
We'll drop them, so revert the dependent changes in drivers.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently mvpp2_xdp_setup won't allow attaching XDP program if
mtu > ETH_DATA_LEN (1500).
The mvpp2_change_mtu on the other hand checks whether
MVPP2_RX_PKT_SIZE(mtu) > MVPP2_BM_LONG_PKT_SIZE.
These two checks are semantically different.
Moreover this limit can be increased to MVPP2_MAX_RX_BUF_SIZE, since in
mvpp2_rx we have
xdp.data = data + MVPP2_MH_SIZE + MVPP2_SKB_HEADROOM;
xdp.frame_sz = PAGE_SIZE;
Change the checks to check whether
mtu > MVPP2_MAX_RX_BUF_SIZE
Fixes: 07dd0a7aae7f ("mvpp2: add basic XDP support")
Signed-off-by: Marek Behún <kabel@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Calling ipa_cmd_pipeline_clear() after stopping the channel
underlying the AP<-modem RX endpoint can lead to a deadlock.
This occurs in the ->runtime_suspend device power operation for the
IPA driver. While this callback is in progress, any other requests
for power will block until the callback returns.
Stopping the AP<-modem RX channel does not prevent the modem from
sending another packet to this endpoint. If a packet arrives for an
RX channel when the channel is stopped, an SUSPEND IPA interrupt
condition will be pending. Handling an IPA interrupt requires
power, so ipa_isr_thread() calls pm_runtime_get_sync() first thing.
The problem occurs because a "pipeline clear" command will not
complete while such a SUSPEND interrupt condition exists. So the
SUSPEND IPA interrupt handler won't proceed until it gets power;
that won't happen until the ->runtime_suspend callback (and its
"pipeline clear" command) completes; and that can't happen while
the SUSPEND interrupt condition exists.
It turns out that in this case there is no need to use the "pipeline
clear" command. There are scenarios in which clearing the pipeline
is required while suspending, but those are not (yet) supported
upstream. So a simple fix, avoiding the potential deadlock, is to
stop calling ipa_cmd_pipeline_clear() in ipa_endpoint_suspend().
This removes the only user of ipa_cmd_pipeline_clear(), so get rid
of that function. It can be restored again whenever it's needed.
This is basically a manual revert along with an explanation for
commit 6cb63ea6a39ea ("net: ipa: introduce ipa_cmd_tag_process()").
Fixes: 6cb63ea6a39ea ("net: ipa: introduce ipa_cmd_tag_process()")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
During the process of driver probing, probe function should return < 0
for failure, otherwise kernel will treat value == 0 as success.
Therefore, we should set err to -EINVAL when
adapter->registered_device_map is NULL. Otherwise kernel will assume
that driver has been successfully probed and will cause unexpected
errors.
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The original changes brakes MAC address assignment on older chip
versions (see bug report [0]), and it brakes random MAC assignment.
is_valid_ether_addr() requires that its argument is word-aligned.
Add the missing alignment to array mac_addr.
[0] https://bugzilla.kernel.org/show_bug.cgi?id=215087
Fixes: 1c5d09d58748 ("ethernet: r8169: use eth_hw_addr_set()")
Reported-by: Richard Herbert <rherbert@sympatico.ca>
Tested-by: Richard Herbert <rherbert@sympatico.ca>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2021-11-22
Maciej Fijalkowski says:
Here are the two fixes for issues around ethtool's set_channels()
callback for ice driver. Both are related to XDP resources. First one
corrects the size of vsi->txq_map that is used to track the usage of Tx
resources and the second one prevents the wrong refcounting of bpf_prog.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
The IPA setup_complete flag is set at the end of ipa_setup(), when
the setup phase of initialization has completed successfully. This
occurs as part of driver probe processing, or (if "modem-init" is
specified in the DTS file) it is triggered by the "ipa-setup-ready"
SMP2P interrupt generated by the modem.
In the latter case, it's possible for driver shutdown (or remove) to
begin while setup processing is underway, and this can't be allowed.
The problem is that the setup_complete flag is not adequate to signal
that setup is underway.
If setup_complete is set, it will never be un-set, so that case is
not a problem. But if setup_complete is false, there's a chance
setup is underway.
Because setup is triggered by an interrupt on a "modem-init" system,
there is a simple way to ensure the value of setup_complete is safe
to read. The threaded handler--if it is executing--will complete as
part of a request to disable the "ipa-modem-ready" interrupt. This
means that ipa_setup() (which is called from the handler) will run
to completion if it was underway, or will never be called otherwise.
The request to disable the "ipa-setup-ready" interrupt is currently
made within ipa_modem_stop(). Instead, disable the interrupt
outside that function in the two places it's called. In the case of
ipa_remove(), this ensures the setup_complete flag is safe to read
before we read it.
Rename ipa_smp2p_disable() to be ipa_smp2p_irq_disable_setup(), to be
more specific about its effect.
Fixes: 530f9216a953 ("soc: qcom: ipa: AP/modem communications")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
We currently maintain a "disabled" Boolean flag to determine whether
the "ipa-setup-ready" SMP2P IRQ handler does anything. That flag
must be accessed under protection of a mutex.
Instead, disable the SMP2P interrupt when requested, which prevents
the interrupt handler from ever being called. More importantly, it
synchronizes a thread disabling the interrupt with the completion of
the interrupt handler in case they run concurrently.
Use the IPA setup_complete flag rather than the disabled flag in the
handler to determine whether to ignore any interrupts arriving after
the first.
Rename the "disabled" flag to be "setup_disabled", to be specific
about its purpose.
Fixes: 530f9216a953 ("soc: qcom: ipa: AP/modem communications")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When processing port up/down events generated by the device's firmware,
the driver protects itself from events reported for non-existent local
ports, but not the CPU port (local port 0), which exists, but lacks a
netdev.
This can result in a NULL pointer dereference when calling
netif_carrier_{on,off}().
Fix this by bailing early when processing an event reported for the CPU
port. Problem was only observed when running on top of a buggy emulator.
Fixes: 28b1987ef506 ("mlxsw: spectrum: Register CPU port with devlink")
Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The driver fails to load with old firmware versions that cannot report
the maximum number of RIF MAC profiles [1].
Fix this by defaulting to a maximum of a single profile in such
situations, as multiple profiles are not supported by old firmware
versions.
[1]
mlxsw_spectrum 0000:03:00.0: cannot register bus device
mlxsw_spectrum: probe of 0000:03:00.0 failed with error -5
Fixes: 1c375ffb2efab ("mlxsw: spectrum_router: Expose RIF MAC profiles to devlink resource")
Signed-off-by: Danielle Ratson <danieller@nvidia.com>
Reported-by: Vadim Pasternak <vadimp@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
MIPS/IA64 define END as assembly function ending, which conflict
with END definition in slip.h, just undef it at first
Reported-by: lkp@intel.com
Signed-off-by: Huang Pei <huangpei@loongson.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
MIPS/IA64 define END as assembly function ending, which conflict
with END definition in mkiss.c, just undef it at first
Reported-by: lkp@intel.com
Signed-off-by: Huang Pei <huangpei@loongson.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
The change to eth_hw_addr_set() caused gcc to correctly spot a
bug that was introduced in an earlier incorrect fix:
In file included from include/linux/etherdevice.h:21,
from drivers/net/ethernet/ni/nixge.c:7:
In function '__dev_addr_set',
inlined from 'eth_hw_addr_set' at include/linux/etherdevice.h:319:2,
inlined from 'nixge_probe' at drivers/net/ethernet/ni/nixge.c:1286:3:
include/linux/netdevice.h:4648:9: error: 'memcpy' reading 6 bytes from a region of size 0 [-Werror=stringop-overread]
4648 | memcpy(dev->dev_addr, addr, len);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
As nixge_get_nvmem_address() can return either NULL or an error
pointer, the NULL check is wrong, and we can end up reading from
ERR_PTR(-EOPNOTSUPP), which gcc knows to contain zero readable
bytes.
Make the function always return an error pointer again but fix
the check to match that.
Fixes: f3956ebb3bf0 ("ethernet: use eth_hw_addr_set() instead of ether_addr_copy()")
Fixes: abcd3d6fc640 ("net: nixge: Fix error path for obtaining mac address")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Function axspi_read_status calls:
ret = spi_write_then_read(ax_spi->spi, ax_spi->cmd_buf, 1,
(u8 *)&status, 3);
status is a pointer to a struct spi_status, which is 3-byte wide:
struct spi_status {
u16 isr;
u8 status;
};
But &status is the pointer to this pointer, and spi_write_then_read does
not dereference this parameter:
int spi_write_then_read(struct spi_device *spi,
const void *txbuf, unsigned n_tx,
void *rxbuf, unsigned n_rx)
Therefore axspi_read_status currently receive a SPI response in the
pointer status, which overwrites 24 bits of the pointer.
Thankfully, on Little-Endian systems, the pointer is only used in
le16_to_cpus(&status->isr);
... which is a no-operation. So there, the overwritten pointer is not
dereferenced. Nevertheless on Big-Endian systems, this can lead to
dereferencing pointers after their 24 most significant bits were
overwritten. And in all systems this leads to possible use of
uninitialized value in functions calling spi_write_then_read which
expect status to be initialized when the function returns.
Moreover function axspi_read_status (and macro AX_READ_STATUS) do not
seem to be used anywhere. So currently this seems to be dead code. Fix
the issue anyway so that future code works properly when using function
axspi_read_status.
Fixes: a97c69ba4f30 ("net: ax88796c: ASIX AX88796C SPI Ethernet Adapter Driver")
Signed-off-by: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Acked-by: Łukasz Stelmach <l.stelmach@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, when user space emits SIOCSHWTSTAMP ioctl calls such as
enabling/disabling timestamping or changing filter settings, the driver
reads the current CLOCK_REALTIME value and programming this into the
NIC's hardware clock. This might be necessary during system
initialization, but at runtime, when the PTP clock has already been
synchronized to a grandmaster, a reset of the timestamp settings might
result in a clock jump. Furthermore, if the clock is also controlled by
phc2sys in automatic mode (where the UTC offset is queried from ptp4l),
that UTC-to-TAI offset (currently 37 seconds in 2021) would be
temporarily reset to 0, and it would take a long time for phc2sys to
readjust so that CLOCK_REALTIME and the PHC are apart by 37 seconds
again.
To address the issue, we introduce a new function called
stmmac_init_tstamp_counter(), which gets called during ndo_open().
It contains the code snippet moved from stmmac_hwtstamp_set() that
manages the time synchronization. Besides, the sub second increment
configuration is also moved here since the related values are hardware
dependent and runtime invariant.
Furthermore, the hardware clock must be kept running even when no time
stamping mode is selected in order to retain the synchronized time base.
That way, timestamping can be enabled again at any time only with the
need to compensate the clock's natural drifting.
As a side effect, this patch fixes the issue that ptp_clock_info::enable
can be called before SIOCSHWTSTAMP and the driver (which looks at
priv->systime_flags) was not prepared to handle that ordering.
Fixes: 92ba6888510c ("stmmac: add the support for PTP hw clock driver")
Reported-by: Michael Olbrich <m.olbrich@pengutronix.de>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Holger Assmann <h.assmann@pengutronix.de>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use nn->tlv_caps.me_freq_mhz instead of nn->me_freq_mhz to check whether
rx-usecs/tx-usecs is valid.
This is because nn->tlv_caps.me_freq_mhz represents the clock_freq (MHz) of
the flow processing cores (FPC) on the NIC. While nn->me_freq_mhz is not
be set.
Fixes: ce991ab6662a ("nfp: read ME frequency from vNIC ctrl memory")
Signed-off-by: Diana Wang <na.wang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
qca8k has a global MTU, so its tracking the MTU per port to make sure
that the largest MTU gets applied.
Since it uses the frame size instead of MTU the driver MTU change function
will then add the size of Ethernet header and checksum on top of MTU.
The driver currently populates the per port MTU size as Ethernet frame
length + checksum which equals 1518.
The issue is that then MTU change function will go through all of the
ports, find the largest MTU and apply the Ethernet header + checksum on
top of it again, so for a desired MTU of 1500 you will end up with 1536.
This is obviously incorrect, so to correct it populate the per port struct
MTU with just the MTU and not include the Ethernet header + checksum size
as those will be added by the MTU change function.
Fixes: f58d2598cf70 ("net: dsa: qca8k: implement the port MTU callbacks")
Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With SGMII phy the internal delay is always applied to the PAD0 config.
This is caused by the falling edge configuration that hardcode the reg
to PAD0 (as the falling edge bits are present only in PAD0 reg)
Move the delay configuration before the reg overwrite to correctly apply
the delay.
Fixes: cef08115846e ("net: dsa: qca8k: set internal delay also for sgmii")
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a VF goes through a reset, it's possible for the VF's feature set
to change. For example it may lose the VIRTCHNL_VF_OFFLOAD_VLAN
capability after VF reset. Unfortunately, the driver doesn't correctly
deal with this situation and errors are seen from downing/upping the
interface and/or moving the interface in/out of a network namespace.
When setting the interface down/up we see the following errors after the
VIRTCHNL_VF_OFFLOAD_VLAN capability was taken away from the VF:
ice 0000:51:00.1: VF 1 failed opcode 12, retval: -64 iavf 0000:51:09.1:
Failed to add VLAN filter, error IAVF_NOT_SUPPORTED ice 0000:51:00.1: VF
1 failed opcode 13, retval: -64 iavf 0000:51:09.1: Failed to delete VLAN
filter, error IAVF_NOT_SUPPORTED
These add/delete errors are happening because the VLAN filters are
tracked internally to the driver and regardless of the VLAN_ALLOWED()
setting the driver tries to delete/re-add them over virtchnl.
Fix the delete failure by making sure to delete any VLAN filter tracking
in the driver when a removal request is made, while preventing the
virtchnl request. This makes it so the driver's VLAN list is up to date
and the errors are
Fix the add failure by making sure the check for VLAN_ALLOWED() during
reset is done after the VF receives its capability list from the PF via
VIRTCHNL_OP_GET_VF_RESOURCES. If VLAN functionality is not allowed, then
prevent requesting re-adding the filters over virtchnl.
When moving the interface into a network namespace we see the following
errors after the VIRTCHNL_VF_OFFLOAD_VLAN capability was taken away from
the VF:
iavf 0000:51:09.1 enp81s0f1v1: NIC Link is Up Speed is 25 Gbps Full Duplex
iavf 0000:51:09.1 temp_27: renamed from enp81s0f1v1
iavf 0000:51:09.1 mgmt: renamed from temp_27
iavf 0000:51:09.1 dev27: set_features() failed (-22); wanted 0x020190001fd54833, left 0x020190001fd54bb3
These errors are happening because we aren't correctly updating the
netdev capabilities and dealing with ndo_fix_features() and
ndo_set_features() correctly.
Fix this by only reporting errors in the driver's ndo_set_features()
callback when VIRTCHNL_VF_OFFLOAD_VLAN is not allowed and any attempt to
enable the VLAN features is made. Also, make sure to disable VLAN
insertion, filtering, and stripping since the VIRTCHNL_VF_OFFLOAD_VLAN
flag applies to all of them and not just VLAN stripping.
Also, after we process the capabilities in the VF reset path, make sure
to call netdev_update_features() in case the capabilities have changed
in order to update the netdev's feature set to match the VF's actual
capabilities.
Lastly, make sure to always report success on VLAN filter delete when
VIRTCHNL_VF_OFFLOAD_VLAN is not supported. The changed flow in
iavf_del_vlans() allows the stack to delete previosly existing VLAN
filters even if VLAN filtering is not allowed. This makes it so the VLAN
filter list is up to date.
Fixes: 8774370d268f ("i40e/i40evf: support for VF VLAN tag stripping control")
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently iavf adapter statistics are refreshed only in a
watchdog task, triggered approximately every two seconds,
which causes some ethtool requests to return outdated values.
Add explicit statistics refresh when requested by ethtool -S.
Fixes: b476b0030e61 ("iavf: Move commands processing to the separate function")
Signed-off-by: Jan Sokolowski <jan.sokolowski@intel.com>
Signed-off-by: Jedrzej Jagielski <jedrzej.jagielski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
System hangs if close the interface is called from the kernel during
the interface is in resetting state.
During resetting operation the link is closing but kernel didn't
know it and it tried to close this interface again what sometimes
led to deadlock.
Inform kernel about current state of interface
and turn off the flag IFF_UP when interface is closing until reset
is finished.
Previously it was most likely to hang the system when kernel
(network manager) tried to close the interface in the same time
when interface was in resetting state because of deadlock.
Fixes: 3c8e0b989aa1 ("i40vf: don't stop me now")
Signed-off-by: Jaroslaw Gawin <jaroslawx.gawin@intel.com>
Signed-off-by: Jedrzej Jagielski <jedrzej.jagielski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Resolve being able to change static values on VF when adaptive interrupt
moderation is enabled.
This problem is fixed by checking the interrupt settings is not
a combination of change of static value while adaptive interrupt
moderation is turned on.
Without this fix, the user would be able to change static values
on VF with adaptive moderation enabled.
Fixes: 65e87c0398f5 ("i40evf: support queue-specific settings for interrupt moderation")
Signed-off-by: Nitesh B Venkatesh <nitesh.b.venkatesh@intel.com>
Tested-by: George Kuruvinakunnel <george.kuruvinakunnel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This bug report came up when we were testing the device driver
by fuzzing. It shows that buf1_len can get underflowed and be
0xfffffffc (4294967292).
This bug is triggerable with a compromised/malfunctioning device.
We found the bug through QEMU emulation tested the patch with
emulation. We did NOT test it on real hardware.
Attached is the bug report by fuzzing.
BUG: KASAN: use-after-free in stmmac_napi_poll_rx+0x1c08/0x36e0 [stmmac]
Read of size 4294967292 at addr ffff888016358000 by task ksoftirqd/0/9
CPU: 0 PID: 9 Comm: ksoftirqd/0 Tainted: G W 5.6.0 #1
Call Trace:
dump_stack+0x76/0xa0
print_address_description.constprop.0+0x16/0x200
? stmmac_napi_poll_rx+0x1c08/0x36e0 [stmmac]
? stmmac_napi_poll_rx+0x1c08/0x36e0 [stmmac]
__kasan_report.cold+0x37/0x7c
? stmmac_napi_poll_rx+0x1c08/0x36e0 [stmmac]
kasan_report+0xe/0x20
check_memory_region+0x15a/0x1d0
memcpy+0x20/0x50
stmmac_napi_poll_rx+0x1c08/0x36e0 [stmmac]
? stmmac_suspend+0x850/0x850 [stmmac]
? __next_timer_interrupt+0xba/0xf0
net_rx_action+0x363/0xbd0
? call_timer_fn+0x240/0x240
? __switch_to_asm+0x40/0x70
? napi_busy_loop+0x520/0x520
? __schedule+0x839/0x15a0
__do_softirq+0x18c/0x634
? takeover_tasklets+0x5f0/0x5f0
run_ksoftirqd+0x15/0x20
smpboot_thread_fn+0x2f1/0x6b0
? smpboot_unregister_percpu_thread+0x160/0x160
? __kthread_parkme+0x80/0x100
? smpboot_unregister_percpu_thread+0x160/0x160
kthread+0x2b5/0x3b0
? kthread_create_on_node+0xd0/0xd0
ret_from_fork+0x22/0x40
Reported-by: Brendan Dolan-Gavitt <brendandg@nyu.edu>
Signed-off-by: Zekun Shen <bruceshenzk@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We found this bug while fuzzing the device driver. Using and freeing
the dangling pointer buff->skb would cause use-after-free and
double-free.
This bug is triggerable with compromised/malfunctioning devices. We
found the bug with QEMU emulation and tested the patch by emulation.
We did NOT test on a real device.
Attached is the bug report.
BUG: KASAN: double-free or invalid-free in consume_skb+0x6c/0x1c0
Call Trace:
dump_stack+0x76/0xa0
print_address_description.constprop.0+0x16/0x200
? consume_skb+0x6c/0x1c0
kasan_report_invalid_free+0x61/0xa0
? consume_skb+0x6c/0x1c0
__kasan_slab_free+0x15e/0x170
? consume_skb+0x6c/0x1c0
kfree+0x8c/0x230
consume_skb+0x6c/0x1c0
aq_ring_tx_clean+0x5c2/0xa80 [atlantic]
aq_vec_poll+0x309/0x5d0 [atlantic]
? _sub_I_65535_1+0x20/0x20 [atlantic]
? __next_timer_interrupt+0xba/0xf0
net_rx_action+0x363/0xbd0
? call_timer_fn+0x240/0x240
? __switch_to_asm+0x34/0x70
? napi_busy_loop+0x520/0x520
? net_tx_action+0x379/0x720
__do_softirq+0x18c/0x634
? takeover_tasklets+0x5f0/0x5f0
run_ksoftirqd+0x15/0x20
smpboot_thread_fn+0x2f1/0x6b0
? smpboot_unregister_percpu_thread+0x160/0x160
? __kthread_parkme+0x80/0x100
? smpboot_unregister_percpu_thread+0x160/0x160
kthread+0x2b5/0x3b0
? kthread_create_on_node+0xd0/0xd0
ret_from_fork+0x22/0x40
Reported-by: Brendan Dolan-Gavitt <brendandg@nyu.edu>
Signed-off-by: Zekun Shen <bruceshenzk@gmail.com>
Reviewed-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Return NOTIFY_DONE (dont't care) for switchdev notifications
that prestera driver don't know how to handle them.
With introduction of SWITCHDEV_BRPORT_[UN]OFFLOADED switchdev
events, the driver rejects adding swport to bridge operation
which is handled by prestera_bridge_port_join() func. The root
cause of this is that prestera driver returns error (EOPNOTSUPP)
in prestera_switchdev_blk_event() handler for unknown swdev
events. This causes switchdev_bridge_port_offload() to fail
when adding port to bridge in prestera_bridge_port_join().
Fixes: 957e2235e526 ("net: make switchdev_bridge_port_{,unoffload} loosely coupled with the bridge")
Signed-off-by: Volodymyr Mytnyk <vmytnyk@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The definition of macro MOTO_SROM_BUG is:
#define MOTO_SROM_BUG (lp->active == 8 && (get_unaligned_le32(
dev->dev_addr) & 0x00ffffff) == 0x3e0008)
and the if statement
if (MOTO_SROM_BUG) lp->active = 0;
using this macro indicates lp->active could be 8. If lp->active is 8 and
the second comparison of this macro is false. lp->active will remain 8 in:
lp->phy[lp->active].gep = (*p ? p : NULL); p += (2 * (*p) + 1);
lp->phy[lp->active].rst = (*p ? p : NULL); p += (2 * (*p) + 1);
lp->phy[lp->active].mc = get_unaligned_le16(p); p += 2;
lp->phy[lp->active].ana = get_unaligned_le16(p); p += 2;
lp->phy[lp->active].fdx = get_unaligned_le16(p); p += 2;
lp->phy[lp->active].ttm = get_unaligned_le16(p); p += 2;
lp->phy[lp->active].mci = *p;
However, the length of array lp->phy is 8, so array overflows can occur.
To fix these possible array overflows, we first check lp->active and then
return -EINVAL if it is greater or equal to ARRAY_SIZE(lp->phy) (i.e. 8).
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Teng Qi <starmiku1207184332@gmail.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
In line 5001, if all id in the array 'lp->phy[8]' is not 0, when the
'for' end, the 'k' is 8.
At this time, the array 'lp->phy[8]' may be out of bound.
Signed-off-by: zhangyue <zhangyue1@kylinos.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
queue
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2021-11-17
This series contains updates to i40e driver only.
Eryk adds accounting for VLAN header in packet size when VF port VLAN is
configured. He also fixes TC queue distribution when the user has changed
queue counts as well as for configuration of VF ADQ which caused dropped
packets.
Michal adds tracking for when a VSI is being released to prevent null
pointer dereference when managing filters.
Karen ensures PF successfully initiates VF requested reset which could
cause a call trace otherwise.
Jedrzej moves validation of channel queue value earlier to prevent
partial configuration when the value is invalid.
Grzegorz corrects the reported error when adding filter fails.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
