[dpdk-dev] [PATCH v3 4/4] hash: modify lookup bulk pipeline

From: Byron Marohn This patch replaces the pipelined rte_hash lookup mechanism with a loop-and-jump model, which performs significantly better, especially for smaller table sizes and smaller table occupancies. Signed-off-by: Byron Marohn Signed-off-by: Saikrishna

[dpdk-dev] [PATCH v3 3/4] hash: add vectorized comparison

From: Byron Marohn In lookup bulk function, the signatures of all entries are compared against the signature of the key that is being looked up. Now that all the signatures are together, they can be compared with vector instructions (SSE, AVX2), achieving higher lookup

[dpdk-dev] [PATCH v3 2/4] hash: reorganize bucket structure

From: Byron Marohn Move current signatures of all entries together in the bucket and same with all alternative signatures, instead of having current and alternative signatures together per entry in the bucket. This will be benefitial in the next commits, where a

[dpdk-dev] [PATCH v3 1/4] hash: reorder hash structure

In order to optimize lookup performance, hash structure is reordered, so all fields used for lookup will be in the first cache line. Signed-off-by: Pablo de Lara --- lib/librte_hash/rte_cuckoo_hash.h | 12 +++- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git

[dpdk-dev] [PATCH v3 0/4] Cuckoo hash lookup enhancements

This patchset improves lookup performance on the current hash library by changing the existing lookup bulk pipeline, with an improved pipeline, based on a loop-and-jump model, instead of the current 4-stage 2-entry pipeline. Also, x86 vectorized intrinsics are used to improve performance when

[dpdk-dev] [PATCH v3 0/4] Cuckoo hash lookup enhancements

This patchset improves lookup performance on the current hash library by changing the existing lookup bulk pipeline, with an improved pipeline, based on a loop-and-jump model, instead of the current 4-stage 2-entry pipeline. Also, x86 vectorized intrinsics are used to improve performance when

[dpdk-dev] [PATCH v2] net/virtio: fix xstats name issue

On Tue, Sep 06, 2016 at 03:50:12PM +0800, Zhiyong Yang wrote: > The patch fixes some xstats name issues Please, state **clearly** what the issue is: it's far away from being enough just mentioning "fix a issue" without actually telling what it is. For this case, you could describe the issue

[dpdk-dev] [PATCH] maintainers: claim v16.07 as a stable release

On Thu, Jul 28, 2016 at 08:21:47PM +0800, Yuanhan Liu wrote: > Claim v16.07 as a stable release: only bug fix patches are allowed. > And volunteer to be the maintainer. > > Signed-off-by: Yuanhan Liu Hi all, I will apply this patch shortly; this would literally mean I'm about to start applying

[dpdk-dev] [RFC] net/ixgbe: start rxtx after all nic config done.

if start rxtx before flow director config, will cause driver can't receive packets from nic. Signed-off-by: wang wei --- drivers/net/ixgbe/ixgbe_ethdev.c | 60 1 file changed, 30 insertions(+), 30 deletions(-) diff --git

[dpdk-dev] [PATCH 5/6] vhost: add a flag to enable Tx zero copy

On Tue, Sep 06, 2016 at 09:42:38AM +, Xu, Qian Q wrote: > Another interesting thing to me is the ZERO-COPY settings. If I have 2 vhost, > and 1 is set as > Zero-copy=0, and another is set zero-copy=1, so the vhost will take it as > Zero-copy > Enabled for all vhost, or for one vhost. The

[dpdk-dev] [PATCH 5/6] vhost: add a flag to enable Tx zero copy

On Tue, Sep 06, 2016 at 09:00:14AM +, Xu, Qian Q wrote: > Just curious about the naming: vhost USER TX Zero copy. In fact, it's Vhost > RX zero-copy > For virtio, it's Virtio TX zero-copy. So, I wonder why we call it as Vhost TX > ZERO-COPY, > Any comments? It's just that "Tx zero copy"

[dpdk-dev] [PATCH 2/3] net/virtio_user: fix wrong sequence of messages

On Tue, Sep 06, 2016 at 03:54:30PM +0800, Tan, Jianfeng wrote: > Hi Yuanhan, > > > On 9/6/2016 2:42 PM, Yuanhan Liu wrote: > >On Fri, Aug 05, 2016 at 11:36:42AM +, Jianfeng Tan wrote: > >>When virtio_user is used with VPP's native vhost user, it cannot > >>send/receive any packets. > >> >

[dpdk-dev] [PATCH v3] virtio: enable indirect descriptors feature

>From be1210e77f0f9072ccb8e6970552596b6780a44c Mon Sep 17 00:00:00 2001 From: Pierre Pfister Date: Fri, 2 Sep 2016 16:24:57 +0200 Subject: [PATCH] virtio: enable indirect descriptors feature Virtio indirect descriptors are supported by the data-path but the feature bit is

[dpdk-dev] [PATCH 2/3] net/virtio_user: fix wrong sequence of messages

Hi Yuanhan, On 9/6/2016 2:42 PM, Yuanhan Liu wrote: > On Fri, Aug 05, 2016 at 11:36:42AM +, Jianfeng Tan wrote: >> When virtio_user is used with VPP's native vhost user, it cannot >> send/receive any packets. >> >> The root cause is that vpp-vhost-user translates the message >>

[dpdk-dev] [PATCH v2] net/virtio: fix xstats name issue

The patch fixes some xstats name issues and make the xstats name conform to the definition of etherStatsPkts1024to1518Octets in rfc2819 page 23. Fixes: 76d4c652e07d ("virtio: add extended stats") --- Changes in v2: modify commit summary. add the fixline. Signed-off-by: Zhiyong Yang ---

[dpdk-dev] [PATCH 2/2] virtio: support IOMMU platform

Oops, seems I failed to send it out (I happened to know it when I was checking the patchwork status: I didn't find my comment). I also happened to find out that I failed to receive quite many emails, including some patches. Seems something went wrong :( --yliu On Mon, Sep 05, 2016 at

[dpdk-dev] [PATCH] kni: add support for core_id param in single threaded mode

On 9/6/2016 3:14 PM, Ferruh Yigit wrote: > On 9/6/2016 12:25 PM, Vladyslav Buslov wrote: >> Allow binding KNI thread to specific core in single threaded mode >> by setting core_id and force_bind config parameters. > > Thanks, patch does exactly what we talked, and as I tested it works fine. > >

[dpdk-dev] [PATCH] kni: add support for core_id param in single threaded mode

On 9/6/2016 3:14 PM, Ferruh Yigit wrote: > On 9/6/2016 12:25 PM, Vladyslav Buslov wrote: >> Allow binding KNI thread to specific core in single threaded mode >> by setting core_id and force_bind config parameters. > > Thanks, patch does exactly what we talked, and as I tested it works fine. > >

[dpdk-dev] [PATCH] kni: add support for core_id param in single threaded mode

On 9/6/2016 12:25 PM, Vladyslav Buslov wrote: > Allow binding KNI thread to specific core in single threaded mode > by setting core_id and force_bind config parameters. Thanks, patch does exactly what we talked, and as I tested it works fine. 1) There are a few comments, can you please find them

[dpdk-dev] [PATCH] maintainers: claim v16.07 as a stable release

> -Original Message- > From: Yuanhan Liu [mailto:yuanhan.liu at linux.intel.com] > Sent: Tuesday, September 6, 2016 1:24 PM > To: dev at dpdk.org > Cc: Thomas Monjalon ; Mcnamara, John > > Subject: Re: [PATCH] maintainers: claim v16.07 as a stable release > > On Thu, Jul 28, 2016 at

[dpdk-dev] [PATCH 2/3] net/virtio_user: fix wrong sequence of messages

On Fri, Aug 05, 2016 at 11:36:42AM +, Jianfeng Tan wrote: > When virtio_user is used with VPP's native vhost user, it cannot > send/receive any packets. > > The root cause is that vpp-vhost-user translates the message > VHOST_USER_SET_FEATURES as puting this device into init state, > aka,

[dpdk-dev] [PATCH] kni: add support for core_id param in single threaded mode

Ferruh, Thanks for suggestions. I'll try to provide new patch this week. Regards, Vladyslav -Original Message- From: Ferruh Yigit [mailto:ferruh.yi...@intel.com] Sent: Tuesday, September 06, 2016 5:31 PM To: Vladyslav Buslov Cc: dev at dpdk.org Subject: Re: [dpdk-dev] [PATCH] kni: add

[dpdk-dev] [PATCH 1/3] net/virtio_user: fix queue pair not enabled

On Fri, Aug 05, 2016 at 11:36:41AM +, Jianfeng Tan wrote: > When virtio_user is used with OVS-DPDK (with mq disabled), it cannot > receive any packets. > > It's because when vhost provides VHOST_USER_GET_PROTOCOL_FEATURES, > all queue pairs are initialized in the disabled state. Quote >

[dpdk-dev] [PATCH] kni: add support for core_id param in single threaded mode

Allow binding KNI thread to specific core in single threaded mode by setting core_id and force_bind config parameters. Signed-off-by: Vladyslav Buslov --- lib/librte_eal/linuxapp/kni/kni_misc.c | 48 ++ 1 file changed, 32 insertions(+), 16 deletions(-) diff

[dpdk-dev] [PATCH] add support for core_id param in single threaded mode

Hi Ferruh, According to your suggestions I implemented dynamic KNI thread affinity patch in single threaded mode. It reuses core_id and force_bind config parameters from multi threaded mode. Regards, Vladyslav Vladyslav Buslov (1): kni: add support for core_id param in single threaded mode

[dpdk-dev] virtio kills qemu VM after stopping/starting ports

> -Original Message- > From: Yuanhan Liu [mailto:yuanhan.liu at linux.intel.com] > Sent: Monday, September 05, 2016 12:10 AM > To: Kyle Larose > Cc: dev at dpdk.org; huawei.xie at intel.com; jianfeng.tan at intel.com > Subject: Re: virtio kills qemu VM after stopping/starting ports > >

[dpdk-dev] meter: excess token bucket update in srtcm

Hi Kannan, Thank you for your reply. I will submit the patch on similar lines I had used for my test. Regards, Nikhil On 6 September 2016 at 10:40, Ramia, Kannan Babu < kannan.babu.ramia at intel.com> wrote: > Hi Nikhil > > You could submit a patch, something like that below logic > > If(

[dpdk-dev] [PATCH] net/kni: add KNI PMD

Add KNI PMD which wraps librte_kni for ease of use. KNI PMD can be used as any regular PMD to send / receive packets to the Linux networking stack. Signed-off-by: Ferruh Yigit --- config/common_base | 1 + config/common_linuxapp | 1 +

[dpdk-dev] [PATCH] doc: add limitations for i40e PMD

This patch adds "Limitations or Known issues" section for i40e PMD, including two items: 1. MPLS packet classification on X710/XL710 2. 16 Byte Descriptor cannot be used on DPDK VF Signed-off-by: Jingjing Wu --- doc/guides/nics/i40e.rst | 33 + 1 file changed, 33

[dpdk-dev] meter: excess token bucket update in srtcm

Hi Nikhil, It also looks to me that you are right. Sorry, my mistake when translating the RFC into code. Challenge in fixing this is how to code it using minimal number of branches ... Thanks, Cristian > -Original Message- > From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of

[dpdk-dev] [PATCH v2] virtio: enable indirect descriptors feature

On 09/05/2016 11:08 PM, Stephen Hemminger wrote: > On Mon, 5 Sep 2016 16:24:13 +0200 > Maxime Coquelin wrote: > >> Thanks Pierre for sending the fix. >> >> Minor comments below: >> >> On 09/05/2016 08:52 AM, Pierre Pfister (ppfister) wrote: >>> Indirect descriptors support was disabled by

[dpdk-dev] meter: excess token bucket update in srtcm

Hi, Can someone please comment on this? Nikhil On 31 August 2016 at 15:32, Nikhil Jagtap wrote: > Hi, > > As per srTCM RFC 2697, we should be updating the E bucket only after the C > bucket overflows. > "Thereafter, the token counts Tc and Te are updated CIR times per second > as follows: >

[dpdk-dev] [PATCH 5/6] vhost: add a flag to enable Tx zero copy

Another interesting thing to me is the ZERO-COPY settings. If I have 2 vhost, and 1 is set as Zero-copy=0, and another is set zero-copy=1, so the vhost will take it as Zero-copy Enabled for all vhost, or for one vhost. Does the vhost allow such usage? Or we need Enforce all vhost zero-copy to

[dpdk-dev] [PATCH v2 5/5] net/i40e: implement new Rx checksum flag

Add CKSUM_GOOD flag to distinguish a good checksum from an unknown one. Signed-off-by: Xiao Wang --- drivers/net/i40e/i40e_rxtx.c | 6 ++ 1 file changed, 6 insertions(+) diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c index 554d167..b49d9dc 100644 ---

[dpdk-dev] [PATCH v2 4/5] net/ixgbe: implement new Rx checksum flag

Add CKSUM_GOOD flag to distinguish a good checksum from an unknown one. Signed-off-by: Xiao Wang --- drivers/net/ixgbe/ixgbe_rxtx.c | 4 +++- drivers/net/ixgbe/ixgbe_rxtx_vec_sse.c | 30 -- 2 files changed, 27 insertions(+), 7 deletions(-) diff --git

[dpdk-dev] [PATCH v2 3/5] net/e1000: implement new Rx checksum flag

Add CKSUM_GOOD flag to distinguish a good checksum from an unknown one. Signed-off-by: Xiao Wang --- drivers/net/e1000/igb_rxtx.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/net/e1000/igb_rxtx.c b/drivers/net/e1000/igb_rxtx.c index 9d80a0b..bc33aed 100644 ---

[dpdk-dev] [PATCH v2 2/5] net/fm10k: implement new Rx checksum flag

Add CKSUM_GOOD flag to distinguish a good checksum from an unknown one. Since vector Rx supports checksum offload, this patch removes the hw_ip_checksum check in fm10k_rx_vec_condition_check(). Signed-off-by: Xiao Wang --- drivers/net/fm10k/fm10k_rxtx.c | 4

[dpdk-dev] [PATCH v2 1/5] net/fm10k: fix Rx checksum flags

A previous patch removed some necessary lines about Rx checksum offload by mistake, this patch adds them back. Fixes: 6046898f5097 ("net/mbuf: remove unused Rx error flags") Signed-off-by: Xiao Wang --- drivers/net/fm10k/fm10k_rxtx.c | 10 ++ 1 file changed, 10 insertions(+) diff

[dpdk-dev] [PATCH v2 0/5] implement new Rx checksum flag

v2: * Removed hw_ip_checksum check in fm10k_rx_vec_condition_check(). * Defined CKSUM_SHIFT for SSE bits shift. * Changed commit title from "add back Rx checksum offload" to "fix Rx checksum flags". * Added new cksum flag support for ixgbe vector Rx, based on patch

[dpdk-dev] [PATCH v2] kni: Bug fix in module_init and module_exit

On 9/2/2016 11:23 AM, Vincent Guo wrote: > Fix pernet calls when HAVE_SIMPLIFIED_PERNET_OPERATIONS is not set. > > Fixes: e6734d21b4e1 ("kni: fix build with kernel 2.6.32") > > Signed-off-by: Vincent Guo Only patch subject should start with lowercase, Thomas would you mind fixing this while

[dpdk-dev] [PATCH 5/6] vhost: add a flag to enable Tx zero copy

Just curious about the naming: vhost USER TX Zero copy. In fact, it's Vhost RX zero-copy For virtio, it's Virtio TX zero-copy. So, I wonder why we call it as Vhost TX ZERO-COPY, Any comments? -Original Message- From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Yuanhan Liu Sent:

[dpdk-dev] [PATCH v2] virtio: enable indirect descriptors feature

On Tue, 6 Sep 2016 06:49:26 + "Pierre Pfister (ppfister)" wrote: > > Le 5 sept. 2016 ? 23:08, Stephen Hemminger > > a ?crit : > > > > On Mon, 5 Sep 2016 16:24:13 +0200 > > Maxime Coquelin wrote: > > > >> Thanks Pierre for sending the fix. > >> > >> Minor comments below: > >> > >> On

[dpdk-dev] [PATCH v2] virtio: enable indirect descriptors feature

> Le 5 sept. 2016 ? 23:08, Stephen Hemminger a > ?crit : > > On Mon, 5 Sep 2016 16:24:13 +0200 > Maxime Coquelin wrote: > >> Thanks Pierre for sending the fix. >> >> Minor comments below: >> >> On 09/05/2016 08:52 AM, Pierre Pfister (ppfister) wrote: >>> Indirect descriptors support was

[dpdk-dev] [PATCH v2] sched: fix releasing enqueued packets

rte_sched_port_free should release only enqueued packets of all queues. Previous behavior is that enqueued and already dequeued packets of only first 4 queues are released. Fixes: 61383240 ("sched: release enqueued mbufs when freeing port") Signed-off-by: Hiroyuki Mikita --- v2: * use