From: Chengguang Xu
Date: Tue, 26 Jun 2018 09:16:31 +0800
> sizeof() will return unsigned value so in the error check
> negative error code will be always larger than sizeof().
>
> Fixes: a0d8e02c35ff ("nfp: add support for reading nffw info")
>
> Signed-off-by: Chengguang Xu
> Acked-by: Jakub
From: Roopa Prabhu
Date: Mon, 25 Jun 2018 20:32:53 -0700
> From: Roopa Prabhu
>
> In systems where neigh gc thresh holds are set to high values,
> admin deleted neigh entries (eg ip neigh flush or ip neigh del) can
> linger around in NUD_FAILED state for a long time until periodic gc kicks
> in
On 6/26/2018 11:05 PM, Jiri Pirko wrote:
Wed, Jun 27, 2018 at 02:04:31AM CEST, xiyou.wangc...@gmail.com wrote:
On Tue, Jun 26, 2018 at 1:01 AM Jiri Pirko wrote:
Create dummy device with clsact first:
# ip link add type dummy
# tc qdisc add dev dummy0 clsact
There is no template assigned by
Wed, Jun 27, 2018 at 02:04:31AM CEST, xiyou.wangc...@gmail.com wrote:
>On Tue, Jun 26, 2018 at 1:01 AM Jiri Pirko wrote:
>> Create dummy device with clsact first:
>> # ip link add type dummy
>> # tc qdisc add dev dummy0 clsact
>>
>> There is no template assigned by default:
>> # tc filter template
On Sun, Jun 24, 2018 at 11:23:46AM +0300, Leon Romanovsky wrote:
> From: Leon Romanovsky
>
> Any error status returned by FW will trigger similar
> to the following error message in the dmesg.
>
> [ 55.884355] mlx5_core :00:04.0: mlx5_cmd_check:712:(pid 555):
> ALLOC_UAR(0x802) op_mod(0x0) f
On Tue, Jun 26, 2018 at 02:39:21PM -0600, Jason Gunthorpe wrote:
> On Tue, Jun 26, 2018 at 07:21:26AM +0300, Leon Romanovsky wrote:
> > On Mon, Jun 25, 2018 at 03:34:38PM -0600, Jason Gunthorpe wrote:
> > > On Sun, Jun 24, 2018 at 11:23:41AM +0300, Leon Romanovsky wrote:
> > > > From: Leon Romanovs
On Tue, Jun 26, 2018 at 6:34 PM, David Miller wrote:
> From: "Jason A. Donenfeld"
> Date: Tue, 26 Jun 2018 01:39:32 +0200
>
>> Two rules with different values of suppress_prefix or suppress_ifgroup
>> are not the same. This fixes an -EEXIST when running:
>>
>>$ ip -4 rule add table main suppr
From: Simon Horman
Metadata may be NULL for one of two reasons:
* Missing user input
* Failure to allocate the metadata dst
Disambiguate these case by returning -EINVAL for the former and -ENOMEM
for the latter rather than -EINVAL for both cases.
This is in preparation for using extended ack to
Hi,
Simon & Pieter say:
This set adds Geneve Options support to the TC tunnel key action.
It provides the plumbing required to configure Geneve variable length
options. The options can be configured in the form CLASS:TYPE:DATA,
where CLASS is represented as a 16bit hexadecimal value, TYPE as an
From: Simon Horman
Add extended ack support for the tunnel key action by using NL_SET_ERR_MSG
during validation of user input.
Cc: Alexander Aring
Signed-off-by: Simon Horman
Signed-off-by: Pieter Jansen van Vuuren
Reviewed-by: Jakub Kicinski
Reviewed-by: David Ahern
---
net/sched/act_tunn
From: Simon Horman
Allow setting tunnel options using the act_tunnel_key action.
Options are expressed as class:type:data and multiple options
may be listed using a comma delimiter.
# ip link add name geneve0 type geneve dstport 0 external
# tc qdisc add dev eth0 ingress
# tc filter add dev
From: Pieter Jansen van Vuuren
Check the tunnel option type stored in tunnel flags when creating options
for tunnels. Thereby ensuring we do not set geneve, vxlan or erspan tunnel
options on interfaces that are not associated with them.
Make sure all users of the infrastructure set correct flags
Hi Lawrence,
Thank you for the patch! Perhaps something to improve:
[auto build test WARNING on net-next/master]
url:
https://github.com/0day-ci/linux/commits/Lawrence-Brakmo/tcp-force-cwnd-at-least-2-in-tcp_cwnd_reduction/20180627-095533
reproduce:
# apt-get install sparse
m
From: Masahiro Yamada
What we want here is to embed a user-space program into the kernel.
Instead of the complex ELF magic, let's simply wrap it in the assembly
with the '.incbin' directive.
Signed-off-by: Masahiro Yamada
Signed-off-by: Alexei Starovoitov
---
I think this patch should 'fix' bp
Hi Pieter,
Thank you for the patch! Perhaps something to improve:
[auto build test WARNING on net-next/master]
url:
https://github.com/0day-ci/linux/commits/Jakub-Kicinski/net-Geneve-options-support-for-TC-act_tunnel_key/20180627-030036
reproduce:
# apt-get install sparse
mak
From: Jiong Wang
Map read has been supported on NFP, this patch enables optimization
for memcpy from map to packet.
This patch also fixed one latent bug which will cause copying from
unexpected address once memcpy for map pointer enabled. The fixed
code path was not exercised before.
Reported-
CQE has all the meta data information from HW.
Make it available to the driver xdp handlers.
Signed-off-by: Saeed Mahameed
---
drivers/net/ethernet/mellanox/mlx5/core/en.h| 9 ++---
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 15 +--
2 files changed, 15 insertions(+),
Implement XDP meta data hash and vlan support.
1. on xdp setup ndo: add support for XDP_QUERY_META_FLAGS and return the
two supported flags
2. use xdp_data_meta_set_hash and xdp_data_meta_set_vlan helpers to fill
in the meta data fileds.
Signed-off-by: Saeed Mahameed
---
.../net/ethernet/mellan
Make xdp flags and meta data info avaiable to RQs data path,
to enable xdp meta data offloads in next two patches.
Signed-off-by: Saeed Mahameed
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 10 +++-
.../net/ethernet/mellanox/mlx5/core/en_main.c | 52 +++
.../net/ethernet/m
A user space application can request to enable a specific set of meta data
to be reported in every xdp buffer provided to the xdp program.
When meta_data flags are required, XDP devices must respond to
XDP_QUERY_META_FLAGS command with all the meta data flags the device
actually supports, and the
Add a new program (prog_num = 4) that will not parse packets and will
use the meta data hash to spread/redirect traffic into different cpus.
For the new program we set on bpf_set_link_xdp_fd:
xdp_flags |= XDP_FLAGS_META_HASH | XDP_FLAGS_META_VLAN;
On mlx5 it will succeed since mlx5 alread
The idea from this patch is to define a well known structure for XDP meta
data fields format and offset placement inside the xdp data meta buffer.
For that driver will need some static information to know what to
provide and where, enters struct xdp_md_info and xdp_md_info_arr:
struct xdp_md_info
Hello,
Although it is far from being close to completion, this series provides
the means and infrastructure to enable device drivers to share packets
meta data information and accelerations with XDP programs,
such as hash, stripped vlan, checksum, flow mark, packet header types,
etc ..
The serie
When using dctcp and doing RPCs, if the last packet of a request is
ECN marked as having seen congestion (CE), the sender can decrease its
cwnd to 1. As a result, it will only send one packet when a new request
is sent. In some instances this results in high tail latencies.
For example, in one set
Pass through mode is to allow packets in MAP format to be passed
on to the stack. rmnet driver can be used to process and demultiplex
these packets. Note that pass through mode can be enabled when the
device is in raw ip mode only.
Signed-off-by: Subash Abhinov Kasiviswanathan
---
drivers/net/us
Hi Lawrence,
Thank you for the patch! Perhaps something to improve:
[auto build test WARNING on net-next/master]
url:
https://github.com/0day-ci/linux/commits/Lawrence-Brakmo/tcp-force-cwnd-at-least-2-in-tcp_cwnd_reduction/20180627-095533
config: x86_64-randconfig-x003-201825 (attached as .c
When using dctcp and doing RPCs, if the last packet of a request is
ECN marked as having seen congestion (CE), the sender can decrease its
cwnd to 1. As a result, it will only send one packet when a new request
is sent. In some instances this results in high tail latencies.
For example, in one set
From: Jakub Kicinski
Date: Mon, 25 Jun 2018 20:36:26 -0700
> This series brings two fixes to TC filter/action offload code.
> Pieter fixes matching MPLS packets when the match is purely on
> the MPLS ethertype and none of the MPLS fields are used.
> John provides a fix for offload of shared block
From: Petr Machata
Date: Tue, 26 Jun 2018 02:05:51 +0200
> This patchset adds a test for ECMP and weighted ECMP between two GRE
> tunnels.
>
> In patches #1 and #2, the function multipath_eval() is first moved from
> router_multipath.sh to lib.sh for ease of reuse, and then fixed up.
>
> In pat
On receving an incomplete message, the existing code stores the
remaining length of the cloned skb in the early_eaten field instead of
incrementing the value returned by __strp_recv. This defers invocation
of sock_rfree for the current skb until the next invocation of
__strp_recv, which returns ear
From: "Jason A. Donenfeld"
Date: Tue, 26 Jun 2018 01:39:32 +0200
> Two rules with different values of suppress_prefix or suppress_ifgroup
> are not the same. This fixes an -EEXIST when running:
>
>$ ip -4 rule add table main suppress_prefixlength 0
>
> Signed-off-by: Jason A. Donenfeld
> F
From: Sowmini Varadhan
Date: Tue, 26 Jun 2018 12:28:22 -0400
> On (06/26/18 10:53), Sowmini Varadhan wrote:
>> Date: Tue, 26 Jun 2018 10:53:23 -0400
>> From: Sowmini Varadhan
>> To: David Miller
>> Cc: netdev@vger.kernel.org, rds-de...@oss.oracle.com
>> Subject: Re: [rds-devel] [PATCH net-next]
On 6/26/2018 4:04 AM, Willem de Bruijn wrote:
> On Mon, Jun 25, 2018 at 7:06 PM Amritha Nambiar
> wrote:
>>
>> This patch adds support to pick Tx queue based on the Rx queue(s) map
>> configuration set by the admin through the sysfs attribute
>> for each Tx queue. If the user configuration for rec
Hi Dave,
This pull request includes fixes for mlx5 netdev and core driver,
for more information please see the tag log below and for.
Please pull and let me know if there's any issue.
For -stable:
For -stable v4.0
net/mlx5: Fix incorrect raw command length parsing
For -stable v4.6
net/
From: Or Gerlitz
In smartnic env, the host (PF) driver might not be an e-switch
manager, hence the switchdev mode representors are running on
the embedded cpu (EC) and not at the host.
As such, we should avoid dealing with vport representors if
not being esw manager.
Fixes: b5ca15ad7e61 ('IB/ml
From: Shay Agroskin
The driver allocates wrong size (due to wrong struct name) when issuing
a query/set request to NIC's register.
Fixes: d8880795dabf ("net/mlx5e: Implement DCBNL IEEE max rate")
Signed-off-by: Shay Agroskin
Signed-off-by: Saeed Mahameed
---
drivers/net/ethernet/mellanox/mlx5
From: Eli Cohen
In smartnic env, if the host (PF) driver is not an e-switch manager, we
are not allowed to apply eswitch ports setups such as vlan (VST),
spoof-checks, min/max rate or state.
Make sure we are eswitch manager when coming to issue these callbacks
and err otherwise.
Also fix the de
From: Alex Vesker
The NULL character was not set correctly for the string containing
the command length, this caused failures reading the output of the
command due to a random length. The fix is to initialize the output
length string.
Fixes: e126ba97dba9 ("mlx5: Add driver for Mellanox Connect-I
From: Alex Vesker
The command interface can work in two modes: Events and Polling.
In the general case, each time we invoke a command, a work is
queued to handle it.
When working in events, the interrupt handler completes the
command execution. On the other hand, when working in polling
mode, th
From: Or Gerlitz
In smartnic env, the host (PF) driver might not be an e-switch
manager, hence the FW will err on driver attempts to deal with
setting/unsetting the eswitch and as a result the overall setup
of sriov will fail.
Fix that by avoiding the operation if e-switch management is not
allo
From: Or Gerlitz
The check for cpu hit statistics was not returning immediate false for
any non vport rep netdev and hence we crashed (say on mlx5 probed VFs) if
user-space tool was calling into any possible netdev in the system.
Fix that by doing a proper check before dereferencing.
Fixes: 1d4
From: Eli Cohen
Manipulating of the MPFS requires eswitch manager capabilities.
Fixes: eeb66cdb6826 ('net/mlx5: Separate between E-Switch and MPFS')
Signed-off-by: Eli Cohen
Reviewed-by: Or Gerlitz
Signed-off-by: Saeed Mahameed
---
drivers/net/ethernet/mellanox/mlx5/core/lib/mpfs.c | 9 +
From: Or Gerlitz
In smartnic env, the host (PF) driver might not be an e-switch
manager, hence the switchdev mode representors are running on
the embedded cpu (EC) and not at the host.
As such, we should avoid dealing with vport representors if
not being esw manager.
While here, make sure to di
On 6/26/2018 3:55 AM, Willem de Bruijn wrote:
> On Mon, Jun 25, 2018 at 7:06 PM Amritha Nambiar
> wrote:
>>
>> Extend transmit queue sysfs attribute to configure Rx queue(s) map
>> per Tx queue. By default no receive queues are configured for the
>> Tx queue.
>>
>> - /sys/class/net/eth0/queues/tx-
On Mon, Jun 25, 2018 at 7:27 AM, Jesper Dangaard Brouer
wrote:
> XDP_TX requires also changing the MAC-addrs, else some hardware
> may drop the TX packet before reaching the wire. This was
> observed with driver mlx5.
>
> If xdp_rxq_info select --action XDP_TX the swapmac functionality
> is activ
Remove a bunch of unused fields from struct pch_gbe_adapter. Among these
polling_netdev, config_space & led_status are entirely unused.
ethtool_lock is initialized but we never attempt to acquire the lock, so
that is effectively unused too. A msg_enable field was documented but
missing, so drop tha
On 06/26/2018 11:17 AM, Edward Cree wrote:
> Just calls netif_receive_skb() in a loop.
...
> +void netif_receive_skb_list(struct sk_buff_head *list)
Please use a standard list_head and standard list operators.
(In all your patches)
1) We do not want to carry a spinlock_t + count per list..
Allow the pch_gbe driver to be built on MIPS platforms, allowing its use
on the MIPS Boston development board.
Signed-off-by: Paul Burton
Cc: Andrew Lunn
Cc: David S. Miller
Cc: netdev@vger.kernel.org
---
Changes in v7: None
drivers/net/ethernet/oki-semi/pch_gbe/Kconfig | 2 +-
1 file chang
The pch_gbe driver uses an irq_sem variable to implement a sempahore
that seems to inconsistently count the number of times we enable or
disable interrupts, and only write to the interrupt enable register when
this count hits 0. This makes absolutely no sense to me, both from the
perspective of how
From: Andrew Lunn
Convert this driver to use the mdio bus and phylib infrastructure. It
will then use the common AT803X PHY driver, rather than use its own
code. Have the shared code also handle the GPIO used to reset the PHY.
Over all, these changes should make it easier to use other PHYs with
We should now be able to cope with the PHY entering hibernation, ie.
ceasing to provide the RX clock, whilst the ethernet link is down.
Remove the code responsible for disabling the AR8031 PHY's hibernation
feature, allowing the PHY to enter its low power hibernation state.
Signed-off-by: Paul Bu
Allow the ptp_pch driver to be built on MIPS platforms in preparation
for use on the MIPS Boston board.
Signed-off-by: Paul Burton
Acked-by: Richard Cochran
Cc: Andrew Lunn
Cc: David S. Miller
Cc: netdev@vger.kernel.org
---
Changes in v7: None
drivers/ptp/Kconfig | 2 +-
1 file changed, 1 i
The pch_gbe driver currently probes for the PHY ID & configures the PHY
every time the MAC is reset, even though we know that the PHY won't have
changed since the last MAC reset [1].
This patch moves the PHY probe to instead happen only once when the
driver is probed, saving time & moving us close
Currently pch_gbe_reset() performs a number of tasks:
1) Calls pch_gbe_reset_hw(), which:
1a) Reads the MAC address from the hardware, even though we already
did that in pch_gbe_open() & it should not have changed.
1b) Writes to the RESET register to reset the MAC.
1c)
When using a PHY connected via RGMII, as the pch_gbe driver presumes is
the case, the RX clock is provided by the PHY to the MAC. Various PHYs,
including both the AR8031 used by the Minnowboard & the RTL8211E used by
the MIPS Boston development board, will stop generating the RX clock
when the ethe
pch_gbe_set_multi() loops through each unused MAC address register,
masking them one by one & waiting for a bit to clear indicating that the
change has taken effect before zeroing out the MAC register.
This is needlessly inefficient. We can instead set all the desired mask
bits with a single write
This patch moves the pch_gbe_watchdog() function lower in pch_gbe_main.c
in order to allow use of other functions in the next patch, without
requiring lots of forward declarations. Doing this as a separate patch
makes it clearer what actually changed in the next patch.
The function is unmodified e
This series cleans up & reworks the pch_gbe driver such that it no
longer contains PHY-specific code, converts it to phylib & enables it to
be built on MIPS systems for use with the MIPS Boston development board.
Unfortunately I don't have access to a Minnowboard, which the driver
contains some pl
On Tue, Jun 26, 2018 at 1:01 AM Jiri Pirko wrote:
> Create dummy device with clsact first:
> # ip link add type dummy
> # tc qdisc add dev dummy0 clsact
>
> There is no template assigned by default:
> # tc filter template show dev dummy0 ingress
>
> Add a template of type flower allowing to insert
On 6/26/2018 3:58 AM, Willem de Bruijn wrote:
> On Mon, Jun 25, 2018 at 7:06 PM Amritha Nambiar
> wrote:
>>
>> Change 'skc_tx_queue_mapping' field in sock_common structure from
>> 'int' to 'unsigned short' type with 0 indicating unset and
>> a positive queue value being set. This way it is consist
On 6/25/2018 8:25 PM, Alexander Duyck wrote:
> On Mon, Jun 25, 2018 at 6:34 PM, Tom Herbert wrote:
>>
>>
>> On Mon, Jun 25, 2018 at 11:04 AM, Amritha Nambiar
>> wrote:
>>>
>>> Change 'skc_tx_queue_mapping' field in sock_common structure from
>>> 'int' to 'unsigned short' type with 0 indicating un
On Mon, Jun 25, 2018 at 7:27 AM, Jesper Dangaard Brouer
wrote:
> There is a cost associated with reading the packet data payload
> that this test ignored. Add option --read to allow enabling
> reading part of the payload.
>
> This sample/tool helps us analyse an issue observed with a NIC
> mlx5 (
From: David Ahern
For ACLs implemented using either FIB rules or FIB entries, the BPF
program needs the FIB lookup status to be able to drop the packet.
Since the bpf_fib_lookup API has not reached a released kernel yet,
change the return code to contain an encoding of the FIB lookup
result and r
On Mon, Jun 25, 2018 at 11:04 AM, Amritha Nambiar
wrote:
> Signed-off-by: Amritha Nambiar
Acked-by: Tom Herbert
> ---
> Documentation/ABI/testing/sysfs-class-net-queues | 11
> Documentation/networking/scaling.txt | 57
> ++
> 2 files changed, 58 inse
On Mon, Jun 25, 2018 at 11:04 AM, Amritha Nambiar
wrote:
> Use static_key for XPS maps to reduce the cost of extra map checks,
> similar to how it is used for RPS and RFS. This includes static_key
> 'xps_needed' for XPS and another for 'xps_rxqs_needed' for XPS using
> Rx queues map.
>
Acked-by:
On Mon, Jun 25, 2018 at 11:04 AM, Amritha Nambiar
wrote:
> Refactor XPS code to support Tx queue selection based on
> CPU(s) map or Rx queue(s) map.
>
> Signed-off-by: Amritha Nambiar
> ---
> include/linux/cpumask.h | 11 ++
> include/linux/netdevice.h | 100 +
> net/cor
On Mon, Jun 25, 2018 at 5:25 AM, Toke Høiland-Jørgensen wrote:
> Add an example program showing how to sample packets from XDP using the
> perf event buffer. The example userspace program just prints the ethernet
> header for every packet sampled.
>
> Reviewed-by: Jakub Kicinski
> Signed-off-by:
On Mon, Jun 25, 2018 at 5:25 AM, Toke Høiland-Jørgensen wrote:
> Add two new helper functions to trace_helpers that supports polling
> multiple perf file descriptors for events. These are used to the XDP
> perf_event_output example, which needs to work with one perf fd per CPU.
>
> Reviewed-by: Ja
On Tue, 26 Jun 2018 17:57:08 +0300, Or Gerlitz wrote:
> > Forwarded Message
> > Subject: [PATCH 0/6] offload Linux LAG devices to the TC datapath
> > Date: Thu, 21 Jun 2018 14:35:55 +0100
> > From: John Hurley
> > To: d...@openvswitch.org, r...@mellanox.com, g...@mellanox.com,
>
On Sun, Jun 24, 2018 at 8:54 PM, Jakub Kicinski
wrote:
> From: Jiong Wang
>
> NFP doesn't have integer divide instruction, this patch use reciprocal
> algorithm (the basic one, reciprocal_div) to emulate it.
>
> For each u32 divide, we would need 11 instructions to finish the operation.
>
> 7 (
On Sun, Jun 24, 2018 at 8:54 PM, Jakub Kicinski
wrote:
> From: Jiong Wang
>
> NFP supports u16 and u32 multiplication. Multiplication is done 8-bits per
> step, therefore we need 2 steps for u16 and 4 steps for u32.
>
> We also need one start instruction to initialize the sequence and one or
> tw
TCP Fast Open is triggered by sys_sendmsg with MSG_FASTOPEN flag for
SOCK_STREAM socket.
Even though it's sys_sendmsg, it eventually calls __inet_stream_connect
the same way sys_connect does for TCP. __inet_stream_connect, in turn,
already has BPF hooks for sys_connect.
That means TFO is already
On Tue, 26 Jun 2018 10:07:54 -0700, Shannon Nelson wrote:
> Implement the IPsec/XFRM offload API for testing.
>
> Signed-off-by: Shannon Nelson
> ---
> V2 - addressed formatting comments from Jakub Kicinski
> V3 - a couple more little xmas tree nits
Thank you! :)
Reviewed-by: Jakub Kicinski
The offload_handle should be an opaque data cookie for the driver
to use, much like the data cookie for a timer or alarm callback.
Thus, the XFRM stack should not be checking for non-zero, because
the driver might use that to store an array reference, which could
be zero, or some other zero but mea
On Tue, 26 Jun 2018 09:12:17 +0200, Jiri Pirko wrote:
> Tue, Jun 26, 2018 at 09:00:45AM CEST, jakub.kicin...@netronome.com wrote:
> >On Mon, Jun 25, 2018 at 11:43 PM, Jiri Pirko wrote:
> >> Tue, Jun 26, 2018 at 06:58:50AM CEST, jakub.kicin...@netronome.com wrote:
> >>>On Mon, 25 Jun 2018 23:01
On Sun, 24 Jun 2018 20:54:21 -0700, Jakub Kicinski wrote:
> + * NOTE: because we are using "reciprocal_value_adv" which doesn't
> + * support dividend with MSB set, so we need to JIT separate NFP
> + * sequence to handle such case. It could be a simple sequence if there
> + * is
On Mon, 25 Jun 2018 23:21:10 -0700, Song Liu wrote:
> > +struct reciprocal_value_adv reciprocal_value_adv(u32 d, u8 prec)
> > +{
> > + struct reciprocal_value_adv R;
> > + u32 l, post_shift;
> > + u64 mhigh, mlow;
> > +
> > + l = fls(d - 1);
> > + post_shift = l;
> > +
On Tue, Jun 26, 2018 at 11:15 AM, Edward Cree wrote:
>
> This patch series adds the capability for the network stack to receive a
> list of packets and process them as a unit, rather than handling each
> packet singly in sequence. This is done by factoring out the existing
> datapath code at e
On Tue, Jun 26, 2018 at 07:21:26AM +0300, Leon Romanovsky wrote:
> On Mon, Jun 25, 2018 at 03:34:38PM -0600, Jason Gunthorpe wrote:
> > On Sun, Jun 24, 2018 at 11:23:41AM +0300, Leon Romanovsky wrote:
> > > From: Leon Romanovsky
> > >
> > > Hi,
> > >
> > > This is bunch of patches trigged by runni
On 6/26/18 12:53 PM, Jakub Kicinski wrote:
> From: Simon Horman
>
> Add extended ack support for the tunnel key action by using NL_SET_ERR_MSG
> during validation of user input.
>
> Cc: David Ahern
> Cc: Alexander Aring
> Signed-off-by: Simon Horman
> Signed-off-by: Pieter Jansen van Vuuren
From: David Ahern
Have one extack message for parsing and validating.
Signed-off-by: David Ahern
---
lib/nlattr.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/nlattr.c b/lib/nlattr.c
index dfa55c873c13..e335bcafa9e4 100644
--- a/lib/nlattr.c
+++ b/lib/nlattr.c
@@
On Tue, Jun 26, 2018 at 07:01:32PM +, Vadim Pasternak wrote:
>
>
> > -Original Message-
> > From: Andrew Lunn [mailto:and...@lunn.ch]
> > Sent: Tuesday, June 26, 2018 9:18 PM
> > To: Vadim Pasternak
> > Cc: Guenter Roeck ; linux...@vger.kernel.org;
> > netdev@vger.kernel.org; rui.zh.
> -Original Message-
> From: Andrew Lunn [mailto:and...@lunn.ch]
> Sent: Tuesday, June 26, 2018 9:18 PM
> To: Vadim Pasternak
> Cc: Guenter Roeck ; linux...@vger.kernel.org;
> netdev@vger.kernel.org; rui.zh...@intel.com; edubez...@gmail.com;
> j...@resnulli.us
> Subject: Re: [patch net-
From: Pieter Jansen van Vuuren
Check the tunnel option type stored in tunnel flags when creating options
for tunnels. Thereby ensuring we do not set geneve, vxlan or erspan tunnel
options on interfaces that are not associated with them.
Make sure all users of the infrastructure set correct flags
Hi,
Simon & Pieter say:
This set adds Geneve Options support to the TC tunnel key action.
It provides the plumbing required to configure Geneve variable length
options. The options can be configured in the form CLASS:TYPE:DATA,
where CLASS is represented as a 16bit hexadecimal value, TYPE as an
From: Simon Horman
Add extended ack support for the tunnel key action by using NL_SET_ERR_MSG
during validation of user input.
Cc: David Ahern
Cc: Alexander Aring
Signed-off-by: Simon Horman
Signed-off-by: Pieter Jansen van Vuuren
Reviewed-by: Jakub Kicinski
---
net/sched/act_tunnel_key.c
From: Simon Horman
Allow setting tunnel options using the act_tunnel_key action.
Options are expressed as class:type:data and multiple options
may be listed using a comma delimiter.
# ip link add name geneve0 type geneve dstport 0 external
# tc qdisc add dev eth0 ingress
# tc filter add dev
From: Simon Horman
Metadata may be NULL for one of two reasons:
* Missing user input
* Failure to allocate the metadata dst
Disambiguate these case by returning -EINVAL for the former and -ENOMEM
for the latter rather than -EINVAL for both cases.
This is in preparation for using extended ack to
On Tue, Jun 26, 2018 at 04:47:05PM +, Vadim Pasternak wrote:
>
>
> > -Original Message-
> > From: Guenter Roeck [mailto:li...@roeck-us.net]
> > Sent: Tuesday, June 26, 2018 7:33 PM
> > To: Vadim Pasternak
> > Cc: Andrew Lunn ; da...@davemloft.net;
> > netdev@vger.kernel.org; rui.zh..
When JITing an eBPF program on x86_64, also JIT a list_func that calls the
bpf_func in a loop. Since this is a direct call, it should perform better
than indirect-calling bpf_func in a loop.
Since getting the percpu 'redirect_info' variable is ugly and magic, pass
a pointer into the list_func a
Deals with all the pre- and post-amble to the BPF program itself, which is
still called one packet at a time.
Involves some fiddly percpu variables to cope with XDP_REDIRECT handling.
Signed-off-by: Edward Cree
---
include/linux/filter.h | 10 +++
net/core/dev.c | 165 +
Adds listified versions of the eBPF interpreter functions, and uses them when
the single func is not JITed. If the single func is JITed (and the list
func is not, which currently it never is), then use the single func since
the cost of interpreting is probably much worse than the cost of the ex
Generally the check should be very cheap, as the sk_buff_head is in cache.
Signed-off-by: Edward Cree
---
net/core/dev.c | 8 ++--
net/ipv4/ip_input.c | 2 ++
2 files changed, 8 insertions(+), 2 deletions(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index f0eb00e9fb57..11f80d4502b9
ip_rcv_finish_core(), if it does not drop, sets skb->dst by either early
demux or route lookup. The last step, calling dst_input(skb), is left to
the caller; in the listified case, we split to form sublists with a common
dst, but then ip_sublist_rcv_finish() just calls dst_input(skb) in a loop.
Also involved adding a way to run a netfilter hook over a list of packets.
Rather than attempting to make netfilter know about lists (which would be
a major project in itself) we just let it call the regular okfn (in this
case ip_rcv_finish()) for any packets it steals, and have it give us back
__netif_receive_skb_taps() does a depressingly large amount of per-packet
work that can't easily be listified, because the another_round looping
makes it nontrivial to slice up into smaller functions.
Fortunately, most of that work disappears in the fast path:
* Hardware devices generally don't
netif_receive_skb_list_internal() now processes a list and hands it
on to the next function.
Signed-off-by: Edward Cree
---
net/core/dev.c | 73 ++
1 file changed, 69 insertions(+), 4 deletions(-)
diff --git a/net/core/dev.c b/net/core/de
First example of a layer splitting the list (rather than merely taking
individual packets off it).
Signed-off-by: Edward Cree
---
net/core/dev.c | 46 ++
1 file changed, 38 insertions(+), 8 deletions(-)
diff --git a/net/core/dev.c b/net/core/dev.c
in
Adds iterator skb_queue_for_each() to run over a list without modifying it.
Signed-off-by: Edward Cree
---
include/linux/skbuff.h | 16
include/trace/events/net.h | 7 +++
net/core/dev.c | 4 +++-
3 files changed, 26 insertions(+), 1 deletion(-)
diff --git
> However, I have some concerns on this matter.
> Our hardware provides bulk reading of the modules temperature, means
> I can get all inputs by one hardware request, which is important optimization.
> Reading each module individually will be resulted in huge overhead and will
> require maybe some
1 - 100 of 206 matches
Mail list logo
