[dpdk-dev] [PATCH 1/2] net/softnic: enable flow classification function

Singh, Jasvinder jasvinder.singh at intel.com
Fri Dec 8 10:53:09 CET 2017


Hi Ferruh,

> -----Original Message-----
> From: Yigit, Ferruh
> Sent: Thursday, December 7, 2017 9:03 PM
> To: Singh, Jasvinder <jasvinder.singh at intel.com>; dev at dpdk.org
> Cc: Dumitrescu, Cristian <cristian.dumitrescu at intel.com>; Thomas Monjalon
> <thomas at monjalon.net>
> Subject: Re: [PATCH 1/2] net/softnic: enable flow classification function
> 
> On 11/30/2017 12:08 PM, Jasvinder Singh wrote:
> > Enables flow classification on softnic rx path so that proceding
> > functions of the packet processing pipeline such as metering and
> > policing could be implemented.
> >
> > Example: Create "soft" port for "hard" port "0000:02:00.1", enable the
> > Flow Classification (FC) feature with default
> > settings:
> >   --vdev 'net_softnic0,hard_name=0000:02:00.1,soft_fc=on'
> >
> > Signed-off-by: Jasvinder Singh <jasvinder.singh at intel.com>
> 
> <...>
> 
> > +#define PMD_PARAM_SOFT_FC
> 	"soft_fc"
> > +#define PMD_PARAM_SOFT_FC_NB_RULES
> 	"soft_fc_nb_rules"
> > +#define PMD_PARAM_SOFT_FC_FLOW_KEY_SIZE
> 	"soft_fc_flow_key_size"
> > +#define PMD_PARAM_SOFT_FC_FLOW_KEY_OFFSET
> 	"soft_fc_flow_key_offset"
> > +#define PMD_PARAM_SOFT_FC_FLOW_KEY_MASK
> 	"soft_fc_flow_key_mask"
> > +
> >  #define PMD_PARAM_HARD_NAME
> 	"hard_name"
> >  #define PMD_PARAM_HARD_TX_QUEUE_ID
> 	"hard_tx_queue_id"
> 
> I wouldn't mention if you won't be already sending a new version but can
> you please align them :)

Yes, will take care of alignment in next version.

 
> <...>
> 
> > @@ -60,7 +60,9 @@ _LDLIBS-y += -L$(RTE_SDK_BIN)/lib  #
> >  _LDLIBS-$(CONFIG_RTE_LIBRTE_FLOW_CLASSIFY)  += -lrte_flow_classify
> >  _LDLIBS-$(CONFIG_RTE_LIBRTE_PIPELINE)       += -lrte_pipeline
> > +_LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)            += --whole-archive
> >  _LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)          += -lrte_table
> > +_LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)            += --no-whole-archive
> 
> I think it is better to move rte_table between whole-archive flags, but I
> vaguely remember Thomas preferred this in the past, so adding him for
> comment.

Yes, I followed what Thomas suggested once.  The order should be protected and flags should be inserted without changing the order.

Thanks,
Jasvinder


More information about the dev mailing list