DPDK patches and discussions
 help / color / mirror / Atom feed
From: Konstantin Ananyev <konstantin.ananyev@huawei.com>
To: Stephen Hemminger <stephen@networkplumber.org>,
	Trevor Tao <taozj888@163.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: RE: [PATCH v4 1/2] examples/l3fwd: relax RSS requirement with option
Date: Mon, 13 Nov 2023 20:01:34 +0000	[thread overview]
Message-ID: <d5a1724b44ca4282aa5f6faaa2d4301b@huawei.com> (raw)
In-Reply-To: <20231113095323.2c4064cf@hermes.local>



> > > Now the port Rx mq_mode had been set to RTE_ETH_MQ_RX_RSS
> > > by default, but some hw and/or virtual interface does not
> > > support the RSS and offload mode presupposed, e.g., some
> > > virtio interfaces in the cloud don't support
> > > RSS and the error msg may like:
> > >
> > > virtio_dev_configure(): RSS support requested but not supported by
> > > the device
> > > Port0 dev_configure = -95
> > >
> > > So to enable the l3fwd running in that environment, the Rx mode requirement
> > > can be relaxed to reflect the hardware feature reality here, and the l3fwd
> > > can run smoothly then.
> > >
> > > An option named "disable-rss" is added to disable the RX RSS explicitly,
> > > and it's disabled by default.
> > >
> > > Signed-off-by: Trevor Tao <taozj888@163.com>
> >
> > Why is running with > 1 rx queue and RSS disabled useful?
> > What happens is all packets arrive on 1st queue and you burn a thread
> > polling an always empty queue.
> >
> > I would prefer not adding yet another command line option and
> > just "do the right thing". If number of rx queues > 1, check that device
> > supports RSS before using it. If device does not support RSS give
> > an error and exit.

I think that's pretty much what we have right now in l3fwd.
As I understand, the rational was: some virtual NICs does not support RSS (vritio?),
but there is still a desire to run l3fwd app on such devices.
I presume for performance comparison purposes.
So the idea was to relax NIC requirements and allow to run l3fwd with disabled RSS.
I suppose the main question here: what then will be the mechanism to distribute
packets over different RX queues?
From my understanding with '--disable-rss' option enabled it is undefined and 
depends on particular NIC.
In case of virtio, my guess - it will be hypervisor counterpart (vhost) to determine packet distribution.

> 
> Something like this maybe:
> diff --git a/examples/l3fwd/main.c b/examples/l3fwd/main.c
> index 6063eb139900..c747ef8aadfb 100644
> --- a/examples/l3fwd/main.c
> +++ b/examples/l3fwd/main.c
> @@ -1257,8 +1257,15 @@ l3fwd_poll_resource_setup(void)
>  		local_port_conf.rx_adv_conf.rss_conf.rss_hf &=
>  			dev_info.flow_type_rss_offloads;
> 
> -		if (dev_info.max_rx_queues == 1)
> +		if (nb_rx_queue > dev_info.max_rx_queues)
> +			rte_exit(EXIT_FAILURE, "Port %u only supports %u queues\n",
> +				 portid, dev_info.max_rx_queues);
> +
> +		if (nb_rx_queue == 1)
>  			local_port_conf.rxmode.mq_mode = RTE_ETH_MQ_RX_NONE;
> +		else if (!(dev_info.rx_offload_capa & RTE_ETH_RX_OFFLOAD_RSS_HASH))
> +			rte_exit(EXIT_FAILURE, "Port %u does not support RSS but %u queues requested\n",
> +				 portid, nb_rx_queue);
> 
>  		if (local_port_conf.rx_adv_conf.rss_conf.rss_hf !=
>  				port_conf.rx_adv_conf.rss_conf.rss_hf) {


  reply	other threads:[~2023-11-13 20:01 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-11-13 16:02 [PATCH v4 0/2] example/l3fwd: relax l3fwd rx RSS/Offload if needed Trevor Tao
2023-11-13 16:02 ` [PATCH v4 1/2] examples/l3fwd: relax RSS requirement with option Trevor Tao
2023-11-13 16:34   ` Stephen Hemminger
2023-11-13 17:53     ` Stephen Hemminger
2023-11-13 20:01       ` Konstantin Ananyev [this message]
2023-11-13 16:02 ` [PATCH v4 2/2] examples/l3fwd: relax RX Offload " Trevor Tao

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=d5a1724b44ca4282aa5f6faaa2d4301b@huawei.com \
    --to=konstantin.ananyev@huawei.com \
    --cc=dev@dpdk.org \
    --cc=stephen@networkplumber.org \
    --cc=taozj888@163.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).