Re: [dpdk-dev] [PATCH 1/2] lib/ethdev: introduce RTE_ETH_DEV_CAPA_FLOW_CREATE_BEFORE_START

[Thomas Monjalon <thomas@monjalon.net>]

There was no follow up to this patch.

Ori, could you give your opinion please?


12/10/2021 14:02, Andrew Rybchenko:
> Summary prefix should be just "ethdev: " and summary should
> be human readable. It should not refer to defines etc.
> 
> On 10/5/21 8:19 PM, Martin Havlik wrote:
> > Not all PMDs allow RTE flow rules to be created before start.
> > This capability will be set for the ones that allow it.
> > 
> > Signed-off-by: Martin Havlik <xhavli56@stud.fit.vutbr.cz>
> > ---
> >  lib/ethdev/rte_ethdev.h | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/lib/ethdev/rte_ethdev.h b/lib/ethdev/rte_ethdev.h
> > index bef24173cf..3115a6fccf 100644
> > --- a/lib/ethdev/rte_ethdev.h
> > +++ b/lib/ethdev/rte_ethdev.h
> > @@ -1448,6 +1448,8 @@ struct rte_eth_conf {
> >  #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
> >  /** Device supports Tx queue setup after device started. */
> >  #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
> > +/** Device supports RTE Flow rule creation before device start. */
> > +#define RTE_ETH_DEV_CAPA_FLOW_CREATE_BEFORE_START 0x00000004
> >  /**@}*/
> >  
> >  /*
> > 
> 
> 
> I think that rte_flow_create() should check
> the capability in !dev->data->dev_started case
> and return error if capability is not advertised.
> The only problem here is to avoid dev_info get on
> each rte_flow_create() call.
>