[dpdk-dev] [PATCH 1/3 v2] ethdev: add rx offload to drop error packets

[Nipun Gupta]

> -----Original Message-----
> From: Jerin Jacob <jerinjacobk at gmail.com>
> Sent: Tuesday, October 6, 2020 6:44 PM
> To: Nipun Gupta <nipun.gupta at nxp.com>
> Cc: Stephen Hemminger <stephen at networkplumber.org>; dpdk-dev
> <dev at dpdk.org>; Thomas Monjalon <thomas at monjalon.net>; Ferruh Yigit
> <ferruh.yigit at intel.com>; Andrew Rybchenko <arybchenko at solarflare.com>;
> Hemant Agrawal <hemant.agrawal at nxp.com>; Sachin Saxena
> <sachin.saxena at nxp.com>; Rohit Raj <rohit.raj at nxp.com>
> Subject: Re: [dpdk-dev] [PATCH 1/3 v2] ethdev: add rx offload to drop error
> packets
> 
> On Tue, Oct 6, 2020 at 6:40 PM Nipun Gupta <nipun.gupta at nxp.com> wrote:
> >
> >
> >
> > > -----Original Message-----
> > > From: Jerin Jacob <jerinjacobk at gmail.com>
> > > Sent: Tuesday, October 6, 2020 5:31 PM
> > > To: Nipun Gupta <nipun.gupta at nxp.com>
> > > Cc: Stephen Hemminger <stephen at networkplumber.org>; dpdk-dev
> > > <dev at dpdk.org>; Thomas Monjalon <thomas at monjalon.net>; Ferruh Yigit
> > > <ferruh.yigit at intel.com>; Andrew Rybchenko
> <arybchenko at solarflare.com>;
> > > Hemant Agrawal <hemant.agrawal at nxp.com>; Sachin Saxena
> > > <sachin.saxena at nxp.com>; Rohit Raj <rohit.raj at nxp.com>
> > > Subject: Re: [dpdk-dev] [PATCH 1/3 v2] ethdev: add rx offload to drop error
> > > packets
> > >
> > > On Tue, Oct 6, 2020 at 4:07 PM Nipun Gupta <nipun.gupta at nxp.com> wrote:
> > > >
> > > >
> > > >
> > > > > -----Original Message-----
> > > > > From: Jerin Jacob <jerinjacobk at gmail.com>
> > > > > Sent: Monday, October 5, 2020 9:40 PM
> > > > > To: Stephen Hemminger <stephen at networkplumber.org>
> > > > > Cc: Nipun Gupta <nipun.gupta at nxp.com>; dpdk-dev <dev at dpdk.org>;
> > > Thomas
> > > > > Monjalon <thomas at monjalon.net>; Ferruh Yigit
> <ferruh.yigit at intel.com>;
> > > > > Andrew Rybchenko <arybchenko at solarflare.com>; Hemant Agrawal
> > > > > <hemant.agrawal at nxp.com>; Sachin Saxena <sachin.saxena at nxp.com>;
> > > Rohit
> > > > > Raj <rohit.raj at nxp.com>
> > > > > Subject: Re: [dpdk-dev] [PATCH 1/3 v2] ethdev: add rx offload to drop
> error
> > > > > packets
> > > > >
> > > > > On Mon, Oct 5, 2020 at 9:05 PM Stephen Hemminger
> > > > > <stephen at networkplumber.org> wrote:
> > > > > >
> > > > > > On Mon,  5 Oct 2020 12:45:04 +0530
> > > > > > nipun.gupta at nxp.com wrote:
> > > > > >
> > > > > > > From: Nipun Gupta <nipun.gupta at nxp.com>
> > > > > > >
> > > > > > > This change adds a RX offload capability, which once enabled,
> > > > > > > hardware will drop the packets in case there of any error in
> > > > > > > the packet such as L3 checksum error or L4 checksum.
> > > > >
> > > > > IMO, Providing additional support up to the level to choose the errors
> > > > > to drops give more control to the application. Meaning,
> > > > > L1 errors such as FCS error
> > > > > L2 errors ..
> > > > > L3 errors such checksum
> > > > > i.e ethdev spec need to have  error level supported by PMD and the
> > > > > application can set the layers interested to drop.
> > > >
> > > > Agree, but 'DEV_RX_OFFLOAD_ERR_PKT_DROP' shall also be there to drop
> all
> > > the
> > > > error packets? Maybe we can rename it to
> > > DEV_RX_OFFLOAD_ALL_ERR_PKT_DROP.
> > >
> > > IMHO,  we introduce such shortcut for a single flag for all err drop
> > > then we can not change the scheme
> > > without an API/ABI break.
> >
> > Are the following offloads fine:
> >         DEV_RX_OFFLOAD_L1_FCS_ERR_PKT_DROP
> >         DEV_RX_OFFLOAD_L3_CSUM_ERR_PKT_DROP
> >         DEV_RX_OFFLOAD_L4_CSUM_ERR_PKT_DROP
> >         DEV_RX_OFFLOAD_ALL_ERR_PKT_DROP
> >
> > Please let me know in case I need to add any other too.
> 
> I think, single offload flags and some config/capability structure to
> define the additional
> layer selection would be good, instead of adding a lot of new offload flags.


+/**
+ * A structure used to enable/disable error packet drop on Rx.
+ */
+struct rte_rx_err_pkt_drop_conf {
+       /** enable/disable all RX error packet drop.
+        * 0 (default) - disable, 1 enable
+        */
+       uint32_t all:1;
+};
+
 /**
  * A structure used to configure an Ethernet port.
  * Depending upon the RX multi-queue mode, extra advanced
@@ -1236,6 +1246,8 @@ struct rte_eth_conf {
        uint32_t dcb_capability_en;
        struct rte_fdir_conf fdir_conf; /**< FDIR configuration. DEPRECATED */
        struct rte_intr_conf intr_conf; /**< Interrupt mode configuration. */
+       struct rte_rx_err_pkt_drop_conf err_pkt_drop_conf;
+       /**< RX error packet drop configuration. */

Is this the kind of changes you are talking about?

Also, more changes will be there in 'struct rte_eth_dev_info' structure, defining
additional separate capability something like 'uint64_t rx_err_drop_offload_capa'.

Regards,
Nipun

> 
> 
> > Ill send a v3.
> >
> > Thanks,
> > Nipun
> >
> > >
> > > >
> > > > Currently we have not planned to add separate knobs for separate error in
> > > > the driver, maybe we can define them separately, or we need have them in
> > > > this series itself?
> > >
> > > I think, ethdev API can have the capability on what are levels it
> > > supported, in your
> > > driver case, you can express the same.
> > >
> > >
> > > >
> > > > >
> > > > > > >
> > > > > > > Signed-off-by: Nipun Gupta <nipun.gupta at nxp.com>
> > > > > > > Signed-off-by: Rohit Raj <rohit.raj at nxp.com>
> > > > > > > ---
> > > > > > > These patches are based over series:
> > > > > > >
> > > > >
> > >
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpatchwo
> > > > >
> > >
> rk.dpdk.org%2Fpatch%2F78630%2F&data=02%7C01%7Cnipun.gupta%40nx
> > > > >
> > >
> p.com%7C90b516fd465c48945e7008d869492b3e%7C686ea1d3bc2b4c6fa92cd9
> > > > >
> > >
> 9c5c301635%7C0%7C0%7C637375110263097933&sdata=RBQswMBsfpM6
> > > > > nyKur%2FaHvOMvNK7RU%2BRyhHt%2FXBsP1OM%3D&reserved=0
> > > > > > >
> > > > > > > Changes in v2:
> > > > > > >  - Add support in DPAA1 driver (patch 2/3)
> > > > > > >  - Add support and config parameter in testpmd (patch 3/3)
> > > > > > >
> > > > > > >  lib/librte_ethdev/rte_ethdev.h | 1 +
> > > > > > >  1 file changed, 1 insertion(+)
> > > > > >
> > > > > > Maybe this should be an rte_flow match/action which would then make
> it
> > > > > > more flexible?
> > > > >
> > > > > I think, it is not based on any Patten matching. So IMO, it should be best
> if it
> > > > > is part of RX offload.
> > > > >
> > > > > >
> > > > > > There is not much of a performance gain for this in real life and
> > > > > > if only one driver supports it then I am not convinced this is needed.
> > > > >
> > > > > Marvell HW has this feature.