Currently, on failure to link with DPDK, the configure script provides
an error message to update the PKG_CONFIG_PATH even though the cause of
failure was missing dependencies. Improve the error message to include this
scenario.
Signed-off-by: Sunil Pai G
---
v2-> v3: Fix sentence.
v1-> v2: Impro
Currently, on failure to link with DPDK, the configure script provides
an error message to update the PKG_CONFIG_PATH even though the cause of
failure was missing dependencies. Improve the error message to include this
scenario.
Signed-off-by: Sunil Pai G
---
acinclude.m4 | 3 ++-
1 file changed
Bleep bloop. Greetings lin huang, I am a robot and I have tried out your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
ERROR: Author linhuang needs to sign off.
WARNING: Unexpected sign-offs from developers who are no
From: linhuang
Userspace tunnel doesn't have a valid device in the kernel. So
get_ifindex() function (ioctl) always get error during
adding a port, deleting a port or updating a port status.
The info log is
"2021-08-29T09:17:39.830Z|00059|netdev_linux|INFO|ioctl(SIOCGIFINDEX)
on vxlan_sys_4789 d
On 11/24/21 22:23, Maxime Coquelin wrote:
> This series introduces a new HXPS Tx mode alognside existing
> XPS and static modes. The goal is to provide a mode where all
> the transmit queues are used, whatever the number of PMD
> threads. This may be used with Vhost-user ports, where the
> guest ap
Hi David,
On 12/7/21 21:37, David Marchand wrote:
Hey Maxime,
On Wed, Nov 24, 2021 at 10:24 PM Maxime Coquelin
wrote:
HXPS feature will enable steering Tx packets on transmist
transmit*
queues based on their hashes. In order to test the feature,
"their hashes" is ambiguous.
s/their h
Hey Maxime,
On Wed, Nov 24, 2021 at 10:24 PM Maxime Coquelin
wrote:
>
> HXPS feature will enable steering Tx packets on transmist
transmit*
> queues based on their hashes. In order to test the feature,
"their hashes" is ambiguous.
> it is needed to be able to get the per-queue statistics for
The netlink policy unit test contains test fixture data that is
subject to endianness and currently fails on big endian systems.
Store the fixture data in a struct to ensure proper byte order for
the header data.
Also fix improper style for sizeof with expressions.
Fixes: bfee9f6c0115 ("netlink:
On Tue, Dec 7, 2021 at 8:01 AM Frode Nordahl
wrote:
>
> The netlink policy unit test contains test fixture data that is
> subject to endianness and currently fails on big endian systems.
>
> Store the fixture data in a struct to ensure proper byte order for
> the header data.
>
> Also fix improper
Now that there is a segmentation in software as a fall back in
case a netdev doesn't support TCP segmentation offloading (TSO),
enable it by default on all possible netdevs.
This patch showcase the idea, but it can't really be applied
because it doesn't support encapsulated packets yet. Either it
This provides a software implementation in the case
the egress netdev doesn't support segmentation in hardware.
This is an _untested_ patch to showcase the proposed solution.
The challenge here is to guarantee packet ordering in the
original batch that may be full of TSO packets. Each TSO
packet
Currently OVS will calculate the segment size based on the
MTU of the egress port. That usually happens to be correct
when the ports share the same MTU, but that is not always true.
Therefore, if the segment size is provided, then use that and
make sure the over sized packets are dropped.
Signed-
The netdev receiving packets is supposed to provide the flags
indicating if the L4 csum was verified and it is OK or BAD,
otherwise the stack will check when appropriate by software.
If the packet comes with good checksum, then postpone the
checksum calculation to the egress device if needed.
Whe
The netdev receiving packets is supposed to provide the flags
indicating if the IP csum was verified and it is OK or BAD,
otherwise the stack will check when appropriate by software.
If the packet comes with good checksum, then postpone the
checksum calculation to the egress device if needed.
Whe
Add a new command to show the offloading features of
each data path port.
Signed-off-by: Flavio Leitner
---
lib/dpif-netdev-unixctl.man | 5
lib/dpif-netdev.c | 58 +
lib/netdev-provider.h | 3 ++
lib/netdev.c| 35 +++
This helps to identify when it is about the flags or
the packet itself.
Signed-off-by: Flavio Leitner
---
lib/conntrack.c | 8
lib/dp-packet.c | 2 +-
lib/dp-packet.h | 10 +-
lib/ipf.c | 4 ++--
lib/netdev-native-tnl.c | 4 ++--
lib/net
Document the implementation of netdev hardware offloading
in userspace datapath.
Signed-off-by: Flavio Leitner
---
Documentation/automake.mk | 1 +
Documentation/topics/index.rst| 1 +
Documentation/topics/nic-offloads.rst | 95 +++
3 files changed,
Rename to better represent their flags.
Signed-off-by: Flavio Leitner
---
lib/dp-packet.h| 21 +++--
lib/netdev-linux.c | 14 +++---
lib/netdev.c | 18 --
3 files changed, 22 insertions(+), 31 deletions(-)
diff --git a/lib/dp-packet.h b/lib/dp-p
Rename to dp_packet_ol_tx_ipv4 to align the flag.
Signed-off-by: Flavio Leitner
---
lib/conntrack.c| 4 ++--
lib/dp-packet.h| 2 +-
lib/ipf.c | 6 +++---
lib/netdev-linux.c | 2 +-
4 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/lib/conntrack.c b/lib/conntrack.c
i
Currently 'p' and 'b' and used for packets, so use
a convention that struct dp_packet is 'p' and
struct dp_packet_batch is 'b'.
Some comments needed new formatting to not pass the
80 column.
Some variables were using 'p' or 'b' were renamed
as well.
There should be no functional change with this
Rename to dp_packet_ol_tcp_seg, because that is less
redundant and allows other protocols.
Signed-off-by: Flavio Leitner
---
lib/dp-packet.h| 2 +-
lib/netdev-linux.c | 2 +-
lib/netdev.c | 4 ++--
3 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/lib/dp-packet.h b/lib/dp-
The name correlates better with the flag names.
Signed-off-by: Flavio Leitner
---
lib/conntrack.c| 8
lib/dp-packet.h| 28 ++--
lib/ipf.c | 6 +++---
lib/netdev-dpdk.c | 24
lib/netdev-linux.c | 24
This represents better the state and use the same
convention as the flags.
Signed-off-by: Flavio Leitner
---
lib/conntrack.c | 4 ++--
lib/dp-packet.h | 4 ++--
lib/ipf.c | 2 +-
lib/netdev-native-tnl.c | 4 ++--
4 files changed, 7 insertions(+), 7 deletions(-)
dif
Use the 'NETDEV_OFFLOAD_' prefix in the flags to indicate
we are talking about hardware offloading capabilities.
Signed-off-by: Flavio Leitner
---
lib/netdev-dpdk.c | 20 ++--
lib/netdev-linux.c| 10 +-
lib/netdev-provider.h | 10 +-
lib/netdev.c
It seems csum is more common and shorter.
Signed-off-by: Flavio Leitner
---
lib/dp-packet.h | 72 +--
lib/netdev-dpdk.c | 16 +-
lib/netdev-linux.c| 8 ++---
lib/netdev-provider.h | 8 ++---
lib/netdev.c | 6 ++--
5 files
It seems csum is more common and shorter.
Signed-off-by: Flavio Leitner
---
lib/conntrack.c | 12 ++--
lib/dp-packet.h | 10 +-
lib/ipf.c | 4 ++--
lib/netdev-native-tnl.c | 4 ++--
4 files changed, 15 insertions(+), 15 deletions(-)
diff --git a/l
This patch series is at RFC stage, though some of the renaming
changes could go in independently of the rest of the series.
The goal is to enable NIC csum and segmentation offloading by
default in OVS userspace data path with and without DPDK.
Other Linux software devices like tap (br) or socket (
Talking about patches 1 and 2 - they've got totally no negative impact,
it's an optimization for HW VTEP scenario - I'd like them to be included
as a part of 21.12.
For patch #3 there is absolutely no affect for users who use either only
stateless ACLs or only stateful.
For users, who do mix
Greetings,
The fall conference will be starting at 17:00 UTC. In case the
joining information didn't come through via the EventBrite email, or in
case you lost the joining information, here are the instructions for
joining Day 1.
NOTE: Day 2 will use a different event link.
To join, select from
On Tue, Dec 7, 2021 at 3:58 AM Vladislav Odintsov wrote:
>
> On 01.12.2021 15:56, Vladislav Odintsov wrote:
> > Currently if user has a stateless and statetul ACLs (allow-stateless and
> > allow-related) in one port group or in one logical switch simultaneously,
> > the stateless rules whould take
On 12/7/21 16:36, Mohammad Heib wrote:
> Hi Ilya,
>
> Thank you for reviewing the patches.
>
> i fully agree that is a code duplication but I wasn't sure if I can
> include those files in nlmon code since those are kernel-headers files.
> so I sent an email to OVS-DISCUSS mailing list but didn't
On 7 Dec 2021, at 16:31, Ilya Maximets wrote:
> On 12/7/21 15:34, Eelco Chaudron wrote:
>>
>>
>> On 7 Dec 2021, at 15:17, Mike Pattrick wrote:
>>
>>> On Tue, Dec 7, 2021 at 8:54 AM Ilya Maximets wrote:
On 11/19/21 15:35, Mike Pattrick wrote:
> This commit adds a basic packet meta
Hi Ilya,
Thank you for reviewing the patches.
i fully agree that is a code duplication but I wasn't sure if I can
include those files in nlmon code since those are kernel-headers files.
so I sent an email to OVS-DISCUSS mailing list but didn't get a
helpful answer:
email link:
https://www.mail-ar
On 12/7/21 15:34, Eelco Chaudron wrote:
>
>
> On 7 Dec 2021, at 15:17, Mike Pattrick wrote:
>
>> On Tue, Dec 7, 2021 at 8:54 AM Ilya Maximets wrote:
>>>
>>> On 11/19/21 15:35, Mike Pattrick wrote:
This commit adds a basic packet metadata macro to the already existing
macros in ovs_gdb
On 11/24/21 14:32, lin huang wrote:
> From: linhuang
>
> Userspace tunnel doesn't have a valid device in the kernel. So
> get_ifindex() function (ioctl) always get error during
> adding a port, deleting a port or updating a port status.
>
> The info log is
> "2021-08-29T09:17:39.830Z|00059|netde
On 7 Dec 2021, at 15:17, Mike Pattrick wrote:
> On Tue, Dec 7, 2021 at 8:54 AM Ilya Maximets wrote:
>>
>> On 11/19/21 15:35, Mike Pattrick wrote:
>>> This commit adds a basic packet metadata macro to the already existing
>>> macros in ovs_gdb.py, ovs_dump_packets will print out information abo
On 12/6/21 18:11, David Marchand wrote:
> On Tue, Nov 23, 2021 at 3:01 PM Kevin Traynor wrote:
>>
>> It seems that on slow system with high concurrency and cpu contention
>> time/warp is not accurate enough for the ALB unit tests with the minimum
>> time/warp that was used to hit an amount of even
On Tue, Dec 7, 2021 at 8:54 AM Ilya Maximets wrote:
>
> On 11/19/21 15:35, Mike Pattrick wrote:
> > This commit adds a basic packet metadata macro to the already existing
> > macros in ovs_gdb.py, ovs_dump_packets will print out information about
> > one or more packets. It feeds packets into tcpd
On 11/22/21 00:39, Mohammad Heib wrote:
> These two patches aim to make nlmon tool more generic
> and extend its functionality to support TC flowers/actions
> monitoring and parsing.
>
> This change will improve the visibility of the communication
> between the OVS and the TC subsystem and can be
On 11/19/21 15:35, Mike Pattrick wrote:
> This commit adds a basic packet metadata macro to the already existing
> macros in ovs_gdb.py, ovs_dump_packets will print out information about
> one or more packets. It feeds packets into tcpdump, and the user can
> pass in tcpdump options to modify how p
This patch avoids calculating the software hash of the packet again
if the optimized miniflow-extract hit and has already calculated the
packet hash. In cases of scalar miniflow extract, the normal hashing
calculation is performed.
Signed-off-by: Kumar Amber
---
lib/dpif-netdev-avx512.c | 6 +++-
This commit adds IPv6 profile specific hashing which
uses fixed offsets into the packet to improve hashing
perforamnce.
Hash value is autovalidated by MFEX autovalidator.
Signed-off-by: Kumar Amber
Signed-off-by: Harry van Haaren
Co-authored-by: Harry van Haaren
---
v4:
- Use pre-defined hash
This patch adds the per profile AVX512 opt hashing to autovalidator
for validating the hash values against the scalar hash.
Signed-off-by: Kumar Amber
---
lib/dpif-netdev-private-extract.c | 12
1 file changed, 12 insertions(+)
diff --git a/lib/dpif-netdev-private-extract.c
b/lib/
Add AVX512 Ipv6 optimized profile for vlan/IPv6/UDP and
vlan/IPv6/TCP.
MFEX autovalidaton test-case already has the IPv6 support for
validating against the scalar mfex.
Signed-off-by: Kumar Amber
Signed-off-by: Harry van Haaren
Co-authored-by: Harry van Haaren
---
v2:
- Fix check-patch sign-o
This commit adds IPv4 profile specific hashing which
uses fixed offsets into the packet to improve hashing
perforamnce.
Signed-off-by: Kumar Amber
Signed-off-by: Harry van Haaren
Co-authored-by: Harry van Haaren
---
v4:
- Use pre-defined hash length values
v3:
- Fix check-patch sign-offs
---
Add AVX512 IPv6 optimized profile for IPv6/UDP and
IPv6/TCP.
MFEX autovalidaton test-case already has the IPv6 support for
validating against the scalar mfex.
Signed-off-by: Kumar Amber
Signed-off-by: Harry van Haaren
Co-authored-by: Harry van Haaren
---
v4:
- Rebase to master
v2:
- Fix CI bu
---
v4:
- rebase to master.
- use static key lenghts for different packet types.
v3:
- rebase to master.
v2:
- fix the CI build.
- fix check-patch for co-author.
---
The patch-set introduces AVX512 optimizations of IPv6
traffic profiles and hashing improvements for all AVX512
supported traffic pro
On 01.12.2021 15:56, Vladislav Odintsov wrote:
Currently if user has a stateless and statetul ACLs (allow-stateless and
allow-related) in one port group or in one logical switch simultaneously,
the stateless rules whould take precedence.
This patch series adds support for mixing all the ACLs type
48 matches
Mail list logo
