DPDK patches and discussions
 help / color / mirror / Atom feed
From: Ivan Boule <ivan.boule@6wind.com>
To: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>,
	"Kulasek, TomaszX" <tomaszx.kulasek@intel.com>,
	"dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH 1/2] ethdev: add buffered tx api
Date: Fri, 12 Feb 2016 17:40:02 +0100	[thread overview]
Message-ID: <56BE0AE2.8080305@6wind.com> (raw)
In-Reply-To: <2601191342CEEE43887BDE71AB97725836B062E2@irsmsx105.ger.corp.intel.com>

On 02/12/2016 12:44 PM, Ananyev, Konstantin wrote:
>
>>
>>> -----Original Message-----
...
>>
>> In that case we don't need to make any changes at rte_ethdev.[h,c] to alloc/free/maintain tx_buffer inside each queue...
>> It all will be upper layer responsibility.
>> So no need to modify existing rte_ethdev structures/code.
>> Again, no need for error callback - caller would check return value and decide what to do with unsent packets in the tx_buffer.
>>
>
> Just to summarise why I think it is better to have tx buffering managed on the app level:
>
> 1. avoid any ABI change.
> 2. Avoid extra changes in rte_ethdev.c: tx_queue_setup/tx_queue_stop.
> 3. Provides much more flexibility to the user:
>     a) where to allocate space for tx_buffer (stack, heap, hugepages, etc).
>     b) user can mix and match plain tx_burst() and   tx_buffer/tx_buffer_flush()
>          in any way he fills it appropriate.
>     c) user can change the size of tx_buffer without stop/re-config/start queue:
>          just allocate new larger(smaller) tx_buffer & copy contents to the new one.
>     d) user can preserve buffered packets through device restart circle:
>          i.e if let say TX hang happened, and user has to do dev_stop/dev_start -
>          contents of tx_buffer will stay unchanged and its contents could be
>          (re-)transmitted after device is up again, or  through different port/queue if needed.
>
> As a drawbacks mentioned - tx error handling becomes less transparent...
> But we can add error handling routine and it's user provided parameter
> into struct rte_eth_dev_tx_buffer', something like that:
>
> +struct rte_eth_dev_tx_buffer {
> +	buffer_tx_error_fn cbfn;
> +	void *userdata;
> +	unsigned nb_pkts;
> +	uint64_t errors;
> +	/**< Total number of queue packets to sent that are dropped. */
> +	struct rte_mbuf *pkts[];
> +};
>
> Konstantin
>

Just to enforce Konstantin's comments.
As a very basic - not to say fundamental - rule, one should avoid adding 
in the PMD RX/TX API any extra processing that can be handled at a 
higher level.
The only and self-sufficient reason is that we must avoid impacting 
performances on the critical path, in particular for those - usually the 
majority of - applications that do not need such extra operations, or 
better implement them at upper level.

Maybe in a not so far future will come a proposal for forking a new open 
source fast-dpdk project aiming at providing API simplicity, 
zero-overhead, modular design, and all those nice properties that every 
one claims to seek :-)

Ivan

  reply	other threads:[~2016-02-12 16:40 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-01-15 14:43 [dpdk-dev] [PATCH 0/2] add support for buffered tx to ethdev Tomasz Kulasek
2016-01-15 14:43 ` [dpdk-dev] [PATCH 1/2] ethdev: add buffered tx api Tomasz Kulasek
2016-01-15 18:13   ` Stephen Hemminger
2016-01-15 18:14   ` Stephen Hemminger
2016-01-15 18:44   ` Ananyev, Konstantin
2016-02-02 10:00     ` Kulasek, TomaszX
2016-02-02 13:49       ` Ananyev, Konstantin
2016-02-09 17:02         ` Kulasek, TomaszX
2016-02-09 23:56           ` Ananyev, Konstantin
2016-02-12 11:44             ` Ananyev, Konstantin
2016-02-12 16:40               ` Ivan Boule [this message]
2016-02-12 17:33                 ` Bruce Richardson
2016-01-15 14:43 ` [dpdk-dev] [PATCH 2/2] examples: sample apps rework to use " Tomasz Kulasek
2016-01-15 18:12 ` [dpdk-dev] [PATCH 0/2] add support for buffered tx to ethdev Stephen Hemminger
2016-02-24 17:08 ` [dpdk-dev] [PATCH v2 " Tomasz Kulasek
2016-02-24 17:08   ` [dpdk-dev] [PATCH v2 1/2] ethdev: add buffered tx api Tomasz Kulasek
2016-03-08 22:52     ` Thomas Monjalon
2016-03-09 13:36       ` Ananyev, Konstantin
2016-03-09 14:25         ` Thomas Monjalon
2016-03-09 15:23           ` Ananyev, Konstantin
2016-03-09 15:26             ` Thomas Monjalon
2016-03-09 15:32               ` Kulasek, TomaszX
2016-03-09 15:37                 ` Thomas Monjalon
2016-03-09 15:42               ` Ananyev, Konstantin
2016-03-09 15:52                 ` Thomas Monjalon
2016-03-09 16:17                   ` Ananyev, Konstantin
2016-03-09 16:21                     ` Thomas Monjalon
2016-03-09 16:35       ` Kulasek, TomaszX
2016-03-09 17:06         ` Thomas Monjalon
2016-03-09 18:12           ` Kulasek, TomaszX
2016-02-24 17:08   ` [dpdk-dev] [PATCH v2 2/2] examples: rework to use " Tomasz Kulasek
2016-02-25 16:17   ` [dpdk-dev] [PATCH v2 0/2] add support for buffered tx to ethdev Ananyev, Konstantin
2016-03-10 10:57   ` [dpdk-dev] [PATCH v3 " Tomasz Kulasek
2016-03-10 10:57     ` [dpdk-dev] [PATCH v3 1/2] ethdev: add buffered tx api Tomasz Kulasek
2016-03-10 16:23       ` Thomas Monjalon
2016-03-10 17:15         ` Kulasek, TomaszX
2016-03-10 10:57     ` [dpdk-dev] [PATCH v3 2/2] examples: rework to use buffered tx Tomasz Kulasek
2016-03-10 11:31     ` [dpdk-dev] [PATCH v3 0/2] add support for buffered tx to ethdev Ananyev, Konstantin
2016-03-10 16:01       ` Jastrzebski, MichalX K
2016-03-10 17:19     ` [dpdk-dev] [PATCH v4 " Tomasz Kulasek
2016-03-10 17:19       ` [dpdk-dev] [PATCH v4 1/2] ethdev: add buffered tx api Tomasz Kulasek
2016-03-10 17:19       ` [dpdk-dev] [PATCH v4 2/2] examples: rework to use buffered tx Tomasz Kulasek
2016-03-11 16:39       ` [dpdk-dev] [PATCH v4 0/2] add support for buffered tx to ethdev Thomas Monjalon
  -- strict thread matches above, loose matches on Subject: below --
2016-01-15 14:26 [dpdk-dev] [PATCH 1/2] ethdev: add buffered tx api Tomasz Kulasek

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=56BE0AE2.8080305@6wind.com \
    --to=ivan.boule@6wind.com \
    --cc=dev@dpdk.org \
    --cc=konstantin.ananyev@intel.com \
    --cc=tomaszx.kulasek@intel.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).