DTS WG Meeting Minutes - September 26, 2024

DTS WG Meeting Minutes - September 26, 2024

[Patrick Robb]

#####################################################################
September 26, 2024
Attendees
* Patrick Robb
* Jeremy Spewock
* Juraj Linkeš
* Tomas Durovec
* Luca Vizzarro

#####################################################################
Minutes

=====================================================================
General Discussion
* Juraj and Jeremy both have their last days at their respective
companies (ARM, UNH-IOL) working on DTS this Friday
   * how/who will inherit some of the testsuites?
      * Jeremy:
         * Open testsuites: UNH team will take these as they have the
hardware and know how to wrap these up
         * VF discussion
            * Writing PF/VF testsuites: There were a couple ideas from
the end of the previous meeting. It may be possible to have PFs and
VFs always run from the same suite (reduced code duplication) and
choose PF or VF mode (or both) based on a configuration value. If we
go this route Patrick believes it would be ideal for the testsuite
result to be represented in a distinct way in the output such that it
is clear whether a PF run or VF run happened.
            * Best policy for discovering VFs and creating them?
Current patch uses VFs if they are already there, makes them
otherwise.
               * This also causes a bug if we want to use more VF
interfaces than currently exist (if any exist, we will not make any
more)
         * Add driver binding to TG:
            * Updated to only copying over the binding script to TG
note - this is ready.
         * XML-RPC Server:
            * Need to update some of the docstrings
         * Verbose parser:
            * Jeremy will update if possible, otherwise the UNH guys
are able to inherit this
         * Packet addressing:
            * Juraj had some comments, no big news
         * Jeremy is rebasing all his patches on next-dts
      * Juraj
         * Capabilities patch:
            * One final idea: an enum of a name was assigned a
function - that function checked whether the capability was supported
or not. And elsewhere he added the ability to add another function
which would allow for configuring something before checking the
capability. These two things should be combined into the single enum.
Juraj’s idea is to extend this so that the enum will contain a
capabilities object.
         * RFC was submitted for external DPDK build and stats
            * Juraj and Tomas will submit one new version today or tomorrow
         * Patch discussions
* External DPDK build and output json
* Per testsuite config
   * V1 submission creates config file which inherits from base config
file for specific testsuites
      * Alex used this to write his RFC (was not submitted) for the
check_port_speed testsuite, just checking that the user defined device
speed matches what DPDK returns
* We should start using dts-dev on slack more often - it seems to
allow for quicker conversation between devs than the mailing list
   * Of course, this is for high level stuff - more comprehensive
patch review needs to be on the mailing list
* Driver binding on tg:
https://patchwork.dpdk.org/project/dpdk/patch/20240812172251.41131-2-jspewock@iol.unh.edu/

=====================================================================
Bugzilla discussions
* None

=====================================================================
* Any other business
* Next meeting Oct 10, 2024