From: Stephen Hemminger <step...@networkplumber.org>
Sent: Tuesday, May 20, 2025 8:47 PM
To: Kiran Kumar Kokkilagadda <kirankum...@marvell.com>
Cc: Aman Singh <aman.deep.si...@intel.com>; Thomas Monjalon
<tho...@monjalon.net>; Ferruh Yigit <ferruh.yi...@amd.com>; Andrew Rybchenko
<andrew.rybche...@oktetlabs.ru>; dev@dpdk.org
Subject: [EXTERNAL] Re: [PATCH v6 1/2] ethdev: support RSS based on RoCEv2
header
On Mon, 5 May 2025 11: 27: 15 +0530 <kirankumark@ marvell. com> wrote: > @@
-149,6 +151,7 @@ const struct rss_type_info rss_type_table[] = { > {
"l3-dst-only", RTE_ETH_RSS_L3_DST_ONLY }, > { "l3-src-only",
RTE_ETH_RSS_L3_SRC_ONLY
On Mon, 5 May 2025 11:27:15 +0530
<kirankum...@marvell.com<mailto:kirankum...@marvell.com>> wrote:
> @@ -149,6 +151,7 @@ const struct rss_type_info rss_type_table[] = {
> { "l3-dst-only", RTE_ETH_RSS_L3_DST_ONLY },
> { "l3-src-only", RTE_ETH_RSS_L3_SRC_ONLY },
> { "ipv6-flow-label", RTE_ETH_RSS_IPV6_FLOW_LABEL },
> + { "ib_bth", RTE_ETH_RSS_IB_BTH },
Please follow convention of other options. Use minus instead of underscore.
Consistency matters.
Ack. Will change it.
> diff --git a/doc/guides/rel_notes/release_25_07.rst
> b/doc/guides/rel_notes/release_25_07.rst
> index 093b85d206..d1293893be 100644
> --- a/doc/guides/rel_notes/release_25_07.rst
> +++ b/doc/guides/rel_notes/release_25_07.rst
> @@ -55,6 +55,10 @@ New Features
> Also, make sure to start the actual text at the margin.
> =======================================================
>
> +* **Added new RSS offload types for IB_BTH in RSS flow.**
> +
> + Added ``RTE_ETH_RSS_IB_BTH`` macro so that the IB BTH header can be
> used as input set for RSS.
> +
Please add reference to the relevant standard and expand the acronym
at least one place in the documentation.
Ack. Will do.
> diff --git a/lib/ethdev/rte_ethdev.h b/lib/ethdev/rte_ethdev.h
> index ea7f8c4a1a..edcd6b75b5 100644
> --- a/lib/ethdev/rte_ethdev.h
> +++ b/lib/ethdev/rte_ethdev.h
> @@ -602,6 +602,7 @@ struct rte_eth_rss_conf {
>
> #define RTE_ETH_RSS_L2TPV2 RTE_BIT64(36)
> #define RTE_ETH_RSS_IPV6_FLOW_LABEL RTE_BIT64(37)
> +#define RTE_ETH_RSS_IB_BTH RTE_BIT64(38)
>
Why is the acronym IB_BTH rather than ROCEV2?
Wouldn't it be clearer if it followed convention?
This is to keep in sync with RTE_FLOW_TYPE_IB_BTH
