Hi Stephen,
On 10/30/24 22:56, Stephen Hemminger wrote:
The current tap device is slow both due to architectural choices and the
overhead of Linux system calls. I am exploring a how to fix that but some
of the choices require some tradeoffs. Which leads to some open questions:
1. DPDK tap also
On Fri, 1 Nov 2024 10:07:20 +
liwencheng wrote:
> +def unbind_platform_one(dev_name):
> +filename = "/sys/bus/platform/devices/%s/driver" % dev_name
> +
> +if exists(filename):
> +try:
> +f = open(os.path.join(filename, "unbind"), "w")
> +except OSError as
This patch mainly adds functions for bind and unbind platform devices,
such as bind_platform_one and unbind_platform_one.
Signed-off-by: liwencheng
---
usertools/dpdk-devbind.py | 101 +-
1 file changed, 99 insertions(+), 2 deletions(-)
diff --git a/u
To optimize Rx/Tx burst process,
add NEON vector instructions on arm architecture.
Signed-off-by: liwencheng
---
drivers/net/macb/macb_rxtx.c | 6 +
drivers/net/macb/macb_rxtx_vec_neon.c | 673 ++
drivers/net/macb/meson.build | 4 +
3 files c
On 10/31/2024 2:24 AM, lihuisong (C) wrote:
> 在 2024/10/30 23:42, Lukas Sismis 写道:
>> Remove rte_experimental macros from the stable functions
>> as they have been part of the stable API since 23.11.
>>
>> Signed-off-by: Lukas Sismis
>
> Acked-by: Huisong Li
>
Carrying from other thread:
Acked-
Hi, all maintainers,
Kindly ping for reviews.
Thanks,
Jie Hai
On 2024/7/19 17:04, Jie Hai wrote:
For hns3 NIC, when KEEP_CRC offload is enabled, the CRC data is
still be stripped in rare scenarios. Some users of hns3 are
already using this feature. So driver has to recaculate packet CRC.
In ad
On Tue, Nov 05, 2024 at 09:32:36AM +0100, David Marchand wrote:
> On Tue, Nov 5, 2024 at 4:19 AM Andre Muezerie
> wrote:
> >
> > MSVC does not support VLAs, replace VLAs with standard C arrays
> > or alloca(). alloca() is available for all toolchain/platform
> > combinations officially supported b
On 10/30/2024 9:29 AM, Jie Hai wrote:
> From: Dengdui Huang
>
> The hardware determines the priority of the flow rule based on the position
> of the rule in the hardware flow director table. Lower index denotes higher
> priority (it means when a packet matches multiple indexes, the smaller
> inde
On 11/4/2024 9:44 PM, Jochen Behrens wrote:
> On 11/4/24 02:52, Morten Brørup wrote:
>> Virtual hardware version 6 supports larger max MTU, but the device
>> information (dev_info) did not reflect this, so it could not be used.
>>
>> Fixes: b1584dd0affe ("net/vmxnet3: support version 6")
>>
>> Sig
On 11/4/2024 10:52 AM, Morten Brørup wrote:
> With virtual hardware version 6, the max number of RX queues was increased
> to VMXNET3_EXT_MAX_RX_QUEUES (32) from VMXNET3_MAX_RX_QUEUES (16),
> similarly, the max number of TX queues was increased to
> VMXNET3_EXT_MAX_TX_QUEUES (32) from VMXNET3_MAX_T
On Tue, Nov 05, 2024 at 09:19:25AM +0100, David Marchand wrote:
> Hello Andre,
>
> Thanks for taking over this series.
>
> On Tue, Nov 5, 2024 at 4:18 AM Andre Muezerie
> wrote:
> >
> > From: Konstantin Ananyev
> >
> > 1) ./lib/eal/linux/eal_interrupts.c:1073:16
> > : warning: ISO C90 forbi
Hi, all maintainers,
Kindly ping for review.
Thanks,
Jie Hai
On 2024/10/18 9:08, Jie Hai wrote:
Add options to support configuring RX burst size and cache size
of mbuf mempoool.
--
v4:
1. fix help info.
v3:
1. add Acked-bys.
2. fix compile error.
--
Jie Hai (2):
examples/l3fwd: add opt
Hi, all maintainers,
Kindly ping for review.
Thanks,
Jie Hai
On 2024/10/12 16:40, Jie Hai wrote:
Now the Rx burst size is fixed to MAX_PKT_BURST (32). This
parameter needs to be modified in some performance optimization
scenarios. So an option '--burst' is added to set the burst size
explicit
On 11/4/2024 1:34 AM, Chaoyong He wrote:
> This patch series fix one problem imported by mistake and add support of
> two new APIs.
>
> ---
> v4:
> * Also update the 'nfp.ini' file.
> v3:
> * Adapt to the multiple PF firmware.
> v2:
> * Add one missing commit.
> ---
>
> Chaoyong He (4):
> net/n
On 11/4/2024 2:29 AM, Jiawen Wu wrote:
> Fix some bugs in txgbe/ngbe, and optimize the Tx flow.
>
> v2 -> v3:
> - Reset Tx desc error statistics in .stats_reset().
> - Add comment for Tx desc error count.
>
> v1 -> v2:
> - Fix build errors on ARM platform.
>
> Jiawen Wu (13):
> net/txgbe: fix
On 11/4/2024 11:58 AM, Junlong Wang wrote:
> v10:
> -
> move zxdh under Wind River in MAINTAINERS and add myself as the maintainer
> and add experimental into MAINTAINERS/driver file,elease notes.
> - changed DPDK syntax is to have return value in a separate line.
> - Add a keyword in lo
#
October 31, 2024
Attendees
1. Patrick Robb
2. Paul Szczepanek
3. Aaron Conole
#
Minutes
=
G
On Wed, Nov 6, 2024 at 2:54 AM Morten Brørup wrote:
>
> > From: Stephen Hemminger [mailto:step...@networkplumber.org]
> > Sent: Wednesday, 6 November 2024 00.26
> >
> > On Wed, 6 Nov 2024 00:22:19 +0100
> > Morten Brørup wrote:
> >
> > > From what I understand, the TAP io_uring PMD only supports
#
October 25, 2024
Attendees
* Patrick Robb
* Paul Szczepanek
* Luca Vizzarro
* Thomas Wilks
#
Minutes
==
[Public]
Snipped
>
>
> The current tap device is slow both due to architectural choices and the
> overhead of
> Linux system calls. I am exploring a how to fix that but some of the choices
> require
> some tradeoffs. Which leads to some open questions:
>
> 1. DPDK tap also support tunnel (TUN) m
On 11/5/2024 9:39 AM, Junlong Wang wrote:
>> > > >> Hi Junlong,
>> > > >
>> > > >> I can see not all of the eth_dev_ops implemented, and datapath not
>> > > >> implemented, so driver is not functional right now.
>> > > >
>> > > >> What happens if you want to run testpmd with the current state of th
On 11/4/2024 9:30 AM, Rybalchenko, Kirill wrote:
>
>> -Original Message-
>> From: Ye, MingjinX
>> Sent: Tuesday 15 October 2024 09:22
>> To: dev@dpdk.org
>> Cc: Ye, MingjinX ; Su, Simei ;
>> Wenjun Wu ; Rybalchenko, Kirill
>>
>> Subject: [PATCH v7] examples/ptpclient: add frequency adjus
> From: Stephen Hemminger [mailto:step...@networkplumber.org]
> Sent: Wednesday, 6 November 2024 00.26
>
> On Wed, 6 Nov 2024 00:22:19 +0100
> Morten Brørup wrote:
>
> > From what I understand, the TAP io_uring PMD only supports one RX
> queue per port and one TX queue per port (i.e. per TAP int
> From: Stephen Hemminger [mailto:step...@networkplumber.org]
> Sent: Tuesday, 5 November 2024 19.59
>
> On Sat, 2 Nov 2024 23:28:49 +0100
> Morten Brørup wrote:
>
> > > > >
> > > > > Probably the hardest part of using io_uring is figuring out how
> to
> > > > > collect
> > > > > completions. Th
On 05. 11. 24 17:50, Morten Brørup wrote:
From: Stephen Hemminger [mailto:step...@networkplumber.org]
Sent: Tuesday, 5 November 2024 16.55
On Tue, 5 Nov 2024 09:49:39 +0100
Morten Brørup wrote:
I suspect AF_PACKET provides an intermediate step which can buffer
more
or spread out the work.
Hello John, Patrick,
On Thu, Oct 24, 2024 at 5:20 PM Robin Jarry wrote:
>
> Here are fixes for three coverity issues:
This series fixes two coverity issues. The fixes look correct.
But the Coverity reports are not obvious to me.
Is there a way to pass those fixes through Coverity to confirm the
Hello,
On Wed, Oct 30, 2024 at 5:00 AM Chengwen Feng wrote:
> diff --git a/drivers/net/tap/rte_eth_tap.c b/drivers/net/tap/rte_eth_tap.c
> index 36b06b3ac5..9366cb39b8 100644
> --- a/drivers/net/tap/rte_eth_tap.c
> +++ b/drivers/net/tap/rte_eth_tap.c
> @@ -2484,19 +2484,19 @@ rte_pmd_tap_probe(st
On Sat, 2 Nov 2024 23:28:49 +0100
Morten Brørup wrote:
> > > >
> > > > Probably the hardest part of using io_uring is figuring out how to
> > > > collect
> > > > completions. The simplest way would be to handle all completions rx
> > and
> > > > tx
> > > > in the rx_burst function.
> > >
>
On 11/4/2024 9:43 AM, 王颢 wrote:
> Dear Ferruh,
>
> I am currently making the changes you mentioned. I would like to clarify a
> few questions with you.
>
>> +F: drivers/net/r8169
>> +
>>
>
> Please sort alphabetically for 'Realtek'
>
> ===> As I can see in the MAINTAINERS file, each PMD in the
This patch adds a new test suite that is designed to test the stopping
and modification of port queues at runtime. Specifically, there are
test cases that display the ports ability to stop some queues but still
send and receive traffic on others, as well as the ability to configure
the ring size of
This patch ports over a test suite from old DTS that tests the
implementation of runtime port queue configuration of PMDs. It also
includes necessary framework changes to allow this to happen such as:
* Sending multiple packets and not capturing the result
* Only adjusting addresses of packets if
This patch adds methods for querying and modifying port queue state and
configuration. In addition to this, it also adds the ability to capture
the forwarding statistics that get outputted when you send the "stop"
command in testpmd. Querying of port queue information is handled
through a TextParse
> From: Stephen Hemminger [mailto:step...@networkplumber.org]
> Sent: Tuesday, 5 November 2024 16.55
>
> On Tue, 5 Nov 2024 09:49:39 +0100
> Morten Brørup wrote:
>
> > >
> > > I suspect AF_PACKET provides an intermediate step which can buffer
> more
> > > or spread out the work.
> >
> > Agree. I
> From: Robin Jarry [mailto:rja...@redhat.com]
> Sent: Tuesday, 5 November 2024 15.13
>
> IPv6 headers are always aligned on a 2-bytes grid. Reflect this in the
> structure definition.
>
> Signed-off-by: Robin Jarry
> ---
>
> Notes:
> v2: added changelog entry
>
> doc/guides/rel_notes/rel
IPv6 headers are always aligned on a 2-bytes grid. Reflect this in the
structure definition.
Signed-off-by: Robin Jarry
---
Notes:
v2: added changelog entry
doc/guides/rel_notes/release_24_11.rst | 6 ++
lib/net/rte_ip6.h | 6 +++---
2 files changed, 9 insertions(+
Hi Nicolas,
On 10/30/24 19:56, Nicolas Chautru wrote:
Allowing ring memory allocation whose end address is aligned with 64 MB.
Previous logic was off by one.
Fixes: 060e76729302 ("baseband/acc100: add queue configuration")
Cc: sta...@dpdk.org
Signed-off-by: Nicolas Chautru
---
drivers/baseb
> -Original Message-
> From: pbhagavat...@marvell.com
> Sent: Friday, November 1, 2024 1:08 PM
> To: Jerin Jacob ; chcch...@163.com; Kiran Kumar
> Kokkilagadda ; Nithin Kumar Dabilpuram
> ; Zhirun Yan
> Cc: dev@dpdk.org; Pavan Nikhilesh Bhagavatula
>
> Subject: [PATCH v2] graph: fix m
On Fri, Nov 1, 2024 at 8:54 AM wrote:
>
> From: Pavan Nikhilesh
>
> Free memory allocated for the node when xstats memory
> allocation fails.
>
> Coverity issue: 445529
> Fixes: 070db97e017b ("graph: support node xstats")
>
> Signed-off-by: Pavan Nikhilesh
Reviewed-by: David Marchand
Huichao,
On Tue, Nov 5, 2024 at 12:08 PM Bruce Richardson
wrote:
>
> On Tue, Nov 05, 2024 at 11:49:52AM +0100, David Marchand wrote:
> > On Tue, Nov 5, 2024 at 11:20 AM Morten Brørup
> > wrote:
> > > > But there is still the question of packed structures with MSVC.
> > > > Tyler proposal seemed to rely o
> From: Morten Brørup [mailto:m...@smartsharesystems.com]
> Sent: Tuesday, 5 November 2024 13.28
>
> > From: Robin Jarry [mailto:rja...@redhat.com]
> > Sent: Tuesday, 5 November 2024 13.25
> >
> > IPv6 headers are always aligned on a 2-bytes grid. Reflect this in
> the
> > structure definition.
>
> From: Robin Jarry [mailto:rja...@redhat.com]
> Sent: Tuesday, 5 November 2024 13.25
>
> IPv6 headers are always aligned on a 2-bytes grid. Reflect this in the
> structure definition.
>
> Signed-off-by: Robin Jarry
> ---
> lib/net/rte_ip6.h | 6 +++---
> 1 file changed, 3 insertions(+), 3 dele
IPv6 headers are always aligned on a 2-bytes grid. Reflect this in the
structure definition.
Signed-off-by: Robin Jarry
---
lib/net/rte_ip6.h | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/lib/net/rte_ip6.h b/lib/net/rte_ip6.h
index 3ae38811b27c..fea7d7b0cd73 100644
--
On Tue, Nov 05, 2024 at 10:14:19AM +, Soumyadeep Hore wrote:
> In ICE PMD, previously the ready bitmap checking before reading
> PHY timestamp was not present. This caused incorrect Tx
> timestamping.
>
> The ready bitmap checking is enabled and PHY timestamp is read once
> the ready bitmap gi
The callbacks(rx_cb) member of struct interface was unused inside
dumpcap, but was taking up a lot of memory space, since it was scaled
according to RTE_MAX_QUEUES_PER_PORT, which is 1k by default. Save
memory by removing the whole array.
Fixes: cbb44143be74 ("app/dumpcap: add new packet capture a
In ICE PMD, previously the ready bitmap checking before reading
PHY timestamp was not present. This caused incorrect Tx
timestamping.
The ready bitmap checking is enabled and PHY timestamp is read once
the ready bitmap gives positive value.
Fixes: 881169950d80 ("net/ice/base: implement initial PT
> From: Bruce Richardson [mailto:bruce.richard...@intel.com]
> Sent: Tuesday, 5 November 2024 12.03
>
> On Tue, Nov 05, 2024 at 11:49:52AM +0100, David Marchand wrote:
> > On Tue, Nov 5, 2024 at 11:20 AM Morten Brørup
> wrote:
> > > > But there is still the question of packed structures with MSVC
> In order to profile app one needs to store significant amount of samples
> somewhere for an analysis later on. Since trace library supports
> storing data in a CTF format lets take advantage of that and add a
> dedicated PMU tracepoint.
>
> Signed-off-by: Tomasz Duszynski
> ---
> app/test/te
On Tue, Nov 05, 2024 at 11:49:52AM +0100, David Marchand wrote:
> On Tue, Nov 5, 2024 at 11:20 AM Morten Brørup
> wrote:
> > > But there is still the question of packed structures with MSVC.
> > > Tyler proposal seemed to rely on the current __rte_packed conventional
> > > position.
> > > https:/
> Add support for programming PMU counters and reading their values
> in runtime bypassing kernel completely.
>
> This is especially useful in cases where CPU cores are isolated
> i.e run dedicated tasks. In such cases one cannot use standard
> perf utility without sacrificing latency and perfo
On Tue, Nov 5, 2024 at 11:20 AM Morten Brørup
wrote:
> > But there is still the question of packed structures with MSVC.
> > Tyler proposal seemed to rely on the current __rte_packed conventional
> > position.
> > https://patchwork.dpdk.org/project/dpdk/patch/1713225913-20792-2-git-
> > send-emai
Enhance example code to allow topology based lcores API, while
retaining default behaviour.
- helloworld: allow lcoes to send hello to lcores in selected topology.
- l2fwd: allow use of IO lcores topology.
- skeleton: choose the lcore from IO topology which has more ports.
v4 changes:
- cross
Add `Topology` section into eal_init documentation.
Signed-off-by: Vipin Varghese
---
.../prog_guide/env_abstraction_layer.rst | 22 +++
1 file changed, 22 insertions(+)
diff --git a/doc/guides/prog_guide/env_abstraction_layer.rst
b/doc/guides/prog_guide/env_abstraction_la
add functional test cases to validate topology supported lcore
API.
v4 cahnges:
- add MACRO for triggering tests if topology is enabled.
- add test cases for
* rte_lcore_is_main_in_domain
* rte_lcore_cpuset_in_domain
* MACRO RTE_LCORE_FORN_NEXT_DOMAIN
* MACRO RTE_LCORE_FORN_WORKER_NE
Introduce topology aware lcore mapping into lcore API.
With higher core density, more and more cores are categorized
into various chiplets based on IO (memory and PCIe) and
Last Level Cache (mainly L3).
Using hwloc library, the dpdk available lcores can be grouped
into various groups nameley L1, L
This patch introduces improvements for NUMA topology awareness in
relation to DPDK logical cores. The goal is to expose API which allows
users to select optimal logical cores for any application. These logical
cores can be selected from various NUMA domains like CPU and I/O.
Change Summary:
- Int
> From: David Marchand [mailto:david.march...@redhat.com]
> Sent: Tuesday, 5 November 2024 10.28
>
> On Tue, Nov 5, 2024 at 10:09 AM Morten Brørup
> wrote:
> > > diff --git a/lib/net/rte_ip4.h b/lib/net/rte_ip4.h
> > > index 4dd0058cc5..f9b832 100644
> > > --- a/lib/net/rte_ip4.h
> > > +++ b/
On Tue, Nov 05, 2024 at 09:59:11AM +0100, David Marchand wrote:
> The new function breaks compilation with -Wcast-align.
>
> In file included from /home/runner/work/ovs/ovs/dpdk-dir/include/rte_ip.h:9:
> /home/runner/work/ovs/ovs/dpdk-dir/include/rte_ip4.h:191:10:
> error: cast from 'const u
> From: Robin Jarry [mailto:rja...@redhat.com]
> Sent: Tuesday, 5 November 2024 10.37
>
> Morten Brørup, Nov 05, 2024 at 10:09:
> > For consistency, could one of you - David or Robin - please also
> > 2-byte align the IPv6 header structure?
>
> I can send a patch but I wonder if this is really ne
> > > >> Hi Junlong,
> > > >
> > > >> I can see not all of the eth_dev_ops implemented, and datapath not
> > > >> implemented, so driver is not functional right now.
> > > >
> > > >> What happens if you want to run testpmd with the current state of the
> > > >> driver, I assume it crashes?
> > > >
Hi Andre,
> As per guidance technical board meeting 2024/04/17. This series
> removes the use of VLAs from code built for Windows for all 3
> toolchains. If there are additional opportunities to convert VLAs
> to regular C arrays please provide the details for incorporation
> into the series.
>
Morten Brørup, Nov 05, 2024 at 10:09:
For consistency, could one of you - David or Robin - please also
2-byte align the IPv6 header structure?
I can send a patch but I wonder if this is really necessary after the
RC-1 has passed?
We don't have any gcc warning related to IPv6 unaligned access
On Tue, Nov 5, 2024 at 10:09 AM Morten Brørup
wrote:
> > diff --git a/lib/net/rte_ip4.h b/lib/net/rte_ip4.h
> > index 4dd0058cc5..f9b832 100644
> > --- a/lib/net/rte_ip4.h
> > +++ b/lib/net/rte_ip4.h
> > @@ -39,7 +39,7 @@ extern "C" {
> > /**
> > * IPv4 Header
> > */
> > -struct rte_ipv4
> From: David Marchand [mailto:david.march...@redhat.com]
> Sent: Tuesday, 5 November 2024 09.59
>
> The new function breaks compilation with -Wcast-align.
>
> In file included from /home/runner/work/ovs/ovs/dpdk-
> dir/include/rte_ip.h:9:
> /home/runner/work/ovs/ovs/dpdk-dir/include/rte_ip4.h:19
The new function breaks compilation with -Wcast-align.
In file included from /home/runner/work/ovs/ovs/dpdk-dir/include/rte_ip.h:9:
/home/runner/work/ovs/ovs/dpdk-dir/include/rte_ip4.h:191:10:
error: cast from 'const uint8_t *' (aka 'const unsigned char *')
to 'const unaligned_uint
To optimize Rx/Tx burst process, add NEON vector instructions
on ARM architecture.
Signed-off-by: liwencheng
---
drivers/net/macb/macb_rxtx.c | 6 +
drivers/net/macb/macb_rxtx_vec_neon.c | 673 ++
drivers/net/macb/meson.build | 4 +
3 files c
> From: Stephen Hemminger [mailto:step...@networkplumber.org]
> Sent: Wednesday, 30 October 2024 17.07
>
> On Wed, 30 Oct 2024 16:40:10 +0100
> Lukáš Šišmiš wrote:
>
> > On 30. 10. 24 16:20, Stephen Hemminger wrote:
> > > On Wed, 30 Oct 2024 14:58:40 +0100
> > > Lukáš Šišmiš wrote:
> > >
> > >>
This patch adds functions for binding and unbinding platform devices,
such as bind_platform_one and unbind_platform_one.
Signed-off-by: liwencheng
---
usertools/dpdk-devbind.py | 101 +-
1 file changed, 99 insertions(+), 2 deletions(-)
diff --git a/us
On Tue, Nov 5, 2024 at 4:19 AM Andre Muezerie
wrote:
>
> MSVC does not support VLAs, replace VLAs with standard C arrays
> or alloca(). alloca() is available for all toolchain/platform
> combinations officially supported by DPDK.
This commitlog does not reflect what is being changed in the patch.
Hello Andre,
Thanks for taking over this series.
On Tue, Nov 5, 2024 at 4:18 AM Andre Muezerie
wrote:
>
> From: Konstantin Ananyev
>
> 1) ./lib/eal/linux/eal_interrupts.c:1073:16
> : warning: ISO C90 forbids variable length array ‘events’
>
> eal_intr_handle_interrupts() is called by eal_in
69 matches
Mail list logo
