DTS WG Meeting Minutes - July 3, 2024

[Patrick Robb <probb@iol.unh.edu>]

#####################################################################
Attendees
* Patrick Robb
* Nicholas Pratte
* Dean Marx
* Alex
* Nicholas Pratte
* Paul Szczepanek

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

=====================================================================
General Announcements
* Patrick will send out an email after this meeting, requesting that
DTS be moved to Thursdays at 13:00 UTC. So, the same time as the CI
Testing meetings, on the off weeks.
* Juraj/David said that voting for a DTS next-* git tree could happen
over the mailing list instead of live at a tech board meeting
* Patrick emailed Thomas about testpmd usage, and he confirmed for
ethdev tests we should not use any of the dpdk example apps, only
testpmd (example: no l2fwd example app for recreating the l2fwd test)

=====================================================================
Patch discussions
* Port blocklist testsuite:
https://patchwork.dpdk.org/project/dpdk/list/?series=32286
   * Patrick Robbwill test on hardware today and do a review
* Testrun and node configuration split:
https://patchwork.dpdk.org/project/dpdk/patch/20240610193410.17968-2-npratte@iol.unh.edu/
   * Should submit a recheck to clean up the false ci fail for github robot
   * Does this need to be rebased after execution was renamed?
      * Nick has already done it locally, and he will submit it later today
* Remove Excess attributes from user config:
https://patchwork.dpdk.org/project/dpdk/list/?series=32161
   * Need to send out a new version which consolidates some patch files
* Mac filter testsuite:
https://patchwork.dpdk.org/project/dpdk/list/?series=32373
   * Looking for reviews, but has been run on hardware at UNH
   * Passing for Intel, Broadcom
   * There is a mac address pool, and it is possible to add more
addresses for Mellanox than you are supposed to be able to, so one
testcase is failing for this check
      * Will double check the mlnx failure manually, but if it is a
bug Nick will create a bugzilla ticket and CC Raslan and Thomas
* Skip testsuites on capabilities:
https://patchwork.dpdk.org/project/dpdk/list/?series=31721
* Nick has started work on the IPGRE testsuite
* Queue start/stop testsuite
   * Does run, will submit as is now
   * Later, once the capability patch is available, should add a
capability check for live starting and stopping queues
* Jeremy working on the rxtx_offload suite now
* Jeremy has to submit a new version for the dynamic queue testsuite
* Scatter suite depends on the capabilities series, but is updated and
ready for reviews
* Xml-rpc replacement:
   * Needs more reviews. Juraj already provided one.
* Improve interactive shell output gathering series:
   * Has had some reviews, is ready to go

=====================================================================
Bugzilla discussions
* MTU enforcement: https://bugs.dpdk.org/show_bug.cgi?id=1476
   * Writing the testsuite according to the definition from
kernel.org. So, providing for the 1500 byte l3 packet, 14 byte (base)
ethernet header, 4 byte vlan tag == 1518 packet size allowed for a
1500 MTU.
* Vlan filtering: https://bugs.dpdk.org/show_bug.cgi?id=1464
   * Dean went through the verbose debug logs for each NIC. Didn’t
find much other than a line in the i40e logs saying “cannot disable
default mac vlan filter”
* When you start testpmd with a non-standard (1500) –max-pkt-len with
mellanox nics, testpmd states that the MTU has updated, but it never
actually updates https://bugs.dpdk.org/show_bug.cgi?id=1470
   * Runtime command for set-max-pkt-len does work correctly, it’s
only the testpmd arg which doesn’t work

=====================================================================
Any other business
* Next meeting Jul 17, 2024