[EXT] Re: [PATCH 0/2] add telemetry for inline IPsec

[Bruce Richardson]

On Thu, Feb 10, 2022 at 10:04:22AM +0000, Ankur Dwivedi wrote:
> Hi Bruce,
> 
> Please see my comment inline.
> 
> Regards,
> Ankur
> 
> >-----Original Message-----
> >From: Bruce Richardson <bruce.richardson at intel.com>
> >Sent: Thursday, February 10, 2022 2:42 PM
> >To: Ankur Dwivedi <adwivedi at marvell.com>
> >Cc: dev at dpdk.org; Nithin Kumar Dabilpuram <ndabilpuram at marvell.com>;
> >Kiran Kumar Kokkilagadda <kirankumark at marvell.com>; Sunil Kumar Kori
> ><skori at marvell.com>; Satha Koteswara Rao Kottidi
> ><skoteshwar at marvell.com>; Jerin Jacob Kollanukkaran
> ><jerinj at marvell.com>; Anoob Joseph <anoobj at marvell.com>; Tejasree
> >Kondoj <ktejasree at marvell.com>
> >Subject: [EXT] Re: [PATCH 0/2] add telemetry for inline IPsec
> >
> >External Email
> >
> >----------------------------------------------------------------------
> >On Thu, Feb 10, 2022 at 12:23:32PM +0530, Ankur Dwivedi wrote:
> >> Adds telemetry function for inline ipsec in net CNXK driver.
> >>
> >> This series depends on the following patch series:
> >>
> >> Series: Adding new features and improvements in cnxk crypto PMD
> >> https://urldefense.proofpoint.com/v2/url?u=http-3A__patches.dpdk.org_p
> >> roject_dpdk_list_-3Fseries-
> >3D21433&d=DwIBAg&c=nKjWec2b6R0mOyPaz7xtfQ&r
> >>
> >=ILjiNF3GF25y6QdHZUxMl6JrStU0MIuCtO5dMzn3Ybk&m=szcLmV4s_NiVroE4B
> >vC5yW6
> >>
> >litxm6Pfo2ymX7QhE1qDSpKTa56lnfmMqfnohRfI6&s=VtKHr9SAo3R0g7lGPoSJlg
> >xVxn
> >> r2t_TYhHbfH7rS4sE&e=
> >>
> >> Ankur Dwivedi (2): net/cnxk: add telemetry for inline IPsec for cn9k
> >> net/cnxk: add telemetry for inline IPsec for cn10k
> >>
> >Out of interest, is there any of this telemetry that could be made generic and
> >apply at a device-class or library level rather than being tied to the cnxk driver
> >only?
> I could think of that the callback function registered in rte_telemetry_register_cmd() can be a generic one. The net device (eth_dev) may have ops registered, which can be called depending on dev->tx_offloads and dev->rx_offloads flags. However the data to be returned by telemetry might depend on the PMD.
> Please let me know if this is possible.

Thanks. So if I understand you correctly, the information returned would
vary based on the offload flags only? If that is the case, it does sound
like it could be generic, where the offload flags field in the return
information identify what other fields are present in the reply. That would
then make the telemetry generic and possible to implement at the ethdev
level. Is that all correct?