Re: [dpdk-dev] [PATCH v4] net/ice: fix invalid RSS type

["Su, Simei" <simei.su@intel.com>]

Hi, Qi

> -----Original Message-----
> From: Zhang, Qi Z <qi.z.zhang@intel.com>
> Sent: Wednesday, July 15, 2020 10:13 PM
> To: Su, Simei <simei.su@intel.com>
> Cc: dev@dpdk.org; Guo, Junfeng <junfeng.guo@intel.com>; Guo, Jia
> <jia.guo@intel.com>; Xu, Ting <ting.xu@intel.com>
> Subject: RE: [PATCH v4] net/ice: fix invalid RSS type
> 
> 
> 
> > -----Original Message-----
> > From: Su, Simei <simei.su@intel.com>
> > Sent: Wednesday, July 15, 2020 6:26 PM
> > To: Zhang, Qi Z <qi.z.zhang@intel.com>
> > Cc: dev@dpdk.org; Guo, Junfeng <junfeng.guo@intel.com>; Guo, Jia
> > <jia.guo@intel.com>; Xu, Ting <ting.xu@intel.com>; Su, Simei
> > <simei.su@intel.com>
> > Subject: [PATCH v4] net/ice: fix invalid RSS type
> >
> > When a RSS rule with only SRC/DST_ONLY or IPV6 prefix RSS type, it
> > should return failure. Besides, when a RSS rule with symmetric hash
> > function, the RSS type shouldn't carry with SRC/DST_ONLY.
> > This patch adds invalid RSS type check for the two cases.
> >
> > Fixes: dfdc589f6ee0 ("net/ice: refactor PF hash flow")
> >
> > Signed-off-by: Simei Su <simei.su@intel.com>
> 
> > ---
> >
> > v4:
> > * Modify check logic for symmetric case.
> >
> > v3:
> > * Add invalid RSS type check for symmetric case.
> > * Consider IPV6 prefix.
> > * Refine commit log.
> >
> > v2:
> > * Add specific macro value in check rather than hard code.
> > ---
> >  drivers/net/ice/ice_hash.c | 28 ++++++++++++++++++++++++++++
> >  1 file changed, 28 insertions(+)
> >
> > diff --git a/drivers/net/ice/ice_hash.c b/drivers/net/ice/ice_hash.c
> > index e57feff..009ea1f 100644
> > --- a/drivers/net/ice/ice_hash.c
> > +++ b/drivers/net/ice/ice_hash.c
> > @@ -1049,13 +1049,41 @@ struct ice_hash_match_type
> > ice_hash_type_list[] = {
> >  			combine_type = ETH_RSS_L2_SRC_ONLY |
> >  					ETH_RSS_L2_DST_ONLY |
> >  					RTE_ETH_RSS_L3_PRE32    |
> > +					RTE_ETH_RSS_L3_PRE40	|
> >  					RTE_ETH_RSS_L3_PRE48    |
> > +					RTE_ETH_RSS_L3_PRE56	|
> >  					RTE_ETH_RSS_L3_PRE64    |
> > +					RTE_ETH_RSS_L3_PRE96	|
> >  					ETH_RSS_L3_SRC_ONLY |
> >  					ETH_RSS_L3_DST_ONLY |
> >  					ETH_RSS_L4_SRC_ONLY |
> >  					ETH_RSS_L4_DST_ONLY;
> 
> Do we support 48, 56 and 96?  Will they already be rejected during previous
> check?

  We only support 32, 48 and 64 currently.

> 
> Better to refactor the code to avoid duplicate, for example:
> 
>  rss_attr_src_dst = ETH_RSS_L2_DST_ONLY | ETH_RSS_L2_SRC_ONLY....
> 
>  rss_attr_l3_pre = ETH_RSS_L3_PRE32 | PRE_40 | ...
> 
>  rss_attr_all = rss_attr_v6_prefix | rss_attr_src_dst
> 
   
   Ok, I will refine it. Thanks.

Br
Simei

> 
> >
> > +			/* Check if only SRC/DST_ONLY or ipv6 prefix exists. */
> > +			if ((rss_type & ~combine_type) == 0)
> > +				return rte_flow_error_set(error, ENOTSUP,
> > +					RTE_FLOW_ERROR_TYPE_ACTION, action,
> > +					"invalid rss types");
> > +
> > +			/**
> > +			 * Check if SRC/DST_ONLY is set for SYMMETRIC_TOEPLITZ
> > +			 * hash function.
> > +			 */
> > +			if (rss->func ==
> > +				RTE_ETH_HASH_FUNCTION_SYMMETRIC_TOEPLITZ) {
> > +				if (rss_type & (ETH_RSS_L2_SRC_ONLY |
> > +						ETH_RSS_L2_DST_ONLY |
> > +						ETH_RSS_L3_SRC_ONLY |
> > +						ETH_RSS_L3_DST_ONLY |
> > +						ETH_RSS_L4_SRC_ONLY |
> > +						ETH_RSS_L4_DST_ONLY))
> > +					return rte_flow_error_set(error,
> > +						ENOTSUP,
> > +						RTE_FLOW_ERROR_TYPE_ACTION,
> > +						action,
> > +						"invalid rss types");
> > +			}
> > +
> >  			/* Check if rss types match pattern. */
> >  			if (rss_type & ~combine_type & ~m->eth_rss_hint) {
> >  				return rte_flow_error_set(error,
> > --
> > 1.8.3.1
>