[dpdk-dev] [PATCH v2 2/2] ethdev: introduce Tx queue offloads API
konstantin.ananyev at intel.com
Tue Sep 12 16:26:38 CEST 2017
> -----Original Message-----
> From: Andrew Rybchenko [mailto:arybchenko at solarflare.com]
> Sent: Tuesday, September 12, 2017 11:28 AM
> To: Shahaf Shuler <shahafs at mellanox.com>; Jerin Jacob <jerin.jacob at caviumnetworks.com>
> Cc: Ananyev, Konstantin <konstantin.ananyev at intel.com>; Stephen Hemminger <stephen at networkplumber.org>; Thomas Monjalon
> <thomas at monjalon.net>; dev at dpdk.org; Zhang, Helin <helin.zhang at intel.com>; Wu, Jingjing <jingjing.wu at intel.com>
> Subject: Re: [dpdk-dev] [PATCH v2 2/2] ethdev: introduce Tx queue offloads API
> On 09/12/2017 11:03 AM, Shahaf Shuler wrote:
> > OK, well understood the requirement for such flags. Thanks for your replies.
> > I think that for simplicity I will add two more flags on the Tx offloads capabilities:
> > DEV_TX_OFFLOADS _MULTI_MEMPOOL <** Device supports transmission of mbufs from multiple mempools. */
> > DEV_TX_OFFLOADS_INDIRECT_MBUFS <** Device support transmission of indirect mbufs. */
> Indirect mbufs is just an example when reference counters are required.
> Direct mbufs may use reference counters as well.
Personally, I still in favor to move these 2 flags away from TX_OFFLOADS.
But if people think it would be really helpfull to keep them, should we have then:
DEV_TX_OFFLOADS_FAST_FREE (or whatever then name will be) -
it would mean the same what (NOMULTIMEMP | NOREFCOUNT) means now.
> > Those caps can be reported by the PMD as per-port/per-queue offloads. Application will choose how to set those. When not set - PMD
> can assume all mbufs has ref_cnt = 1 and the same mempool.
> > Any objection?
More information about the dev