#####################################################################
September 18, 2025
Attendees
   1. Patrick Robb
   2. Aaron Conole
   3. Luca Vizzarro
   4. Sadashiva Rao
   5. Paul Szczepanek
   6. Aaron Conole
   
#####################################################################
Minutes

=====================================================================
General Announcements
   * UNH guys met with governing board about the DTS work this year and priorities going forward in remaining 2025 time and in 2026 (pending next contract).
   * We have published the new DTS instructional videos, showing how to setup and run the framework, and how to write a testcase.
        * Should produce a part 2 video soon showing some of the minor details like how to use the logger etc.
   * OvS conference: Call for papers now, ending Monday
CI Status

---------------------------------------------------------------------
AWS Lab
   * None
   
---------------------------------------------------------------------
UNH-IOL Community Lab
   * PW-CI: Working on standing up an instance of pw-ci, so we can start to poll for some patches and better understand if this can be used in our lab
        * Adam made a graphic about how pw-ci works and sent it to Aaron to see whether our understanding is correct
        * It could also be committed to the repo if that was considered a value
   * DTS:
        * Mlx5_core: A patch has been merged which changes how the device reports MTU, causing the MTU DTS testsuite to fail. Patrick needs to check what the issue is and send the info to the submitter.
   * Sadashiva asked what platforms/tests are being run at the community lab:
        * Intel i40e and ice testing:
        * Performance:
            * Single core forwarding test
        * Functional testing:
            * Various ethdev tests found in https://git.dpdk.org/tools/dts/tree/tests
        * Cryptodev testing:
            * Still using legacy DTS for this
        * Patrick to send a summary to Satashiva for Intel based testing at the Community Lab

---------------------------------------------------------------------
Intel Lab
   * None

---------------------------------------------------------------------
Github Actions Robot
   * Aaron has sent out to a few CI people an initial patch porting pw-ci over to Python.
        * Adam and Patrick from UNH will do a review
        * Coderabbit and sourcery.ai:
            * Reporting into DPDK patchwork
        * Consensus from tech board is that it is sometimes catching issues, but also making a lot of “noise.” Unclear whether it is a positive or negative value.
            * Possibly going to disable coderabbit for now and enable a competing system.

---------------------------------------------------------------------
Loongson Lab
   * None
   
=====================================================================
DTS Improvements & Test Development
   * Aaron’s AI CI systems are sending some reviews for DTS patches:
        * https://github.com/ovsrobot/dpdk/pull/49
   * DTS API Series:
        * Looks good. Patrick to merge.
        * Want to ensure that there are no bugs introduced to testsuites through this patch. UNH guys to run testing on our hardware.
   * Testsuite docs: https://patchwork.dpdk.org/project/dpdk/patch/20250910194259.1027220-2-paul.szczepanek@arm.com/
        * Depends on the API update series, which is being merged today.
   * Virtio testing: Dean is working on a virtio forwarding testsuite to start
        * Contains virtio testing testcases both with the backend vhost in –no-pci mode and also others that have the backend vhost forward packets from physical NICs
        * V1 Submitted.
   * Single_core_forwarding
        * Different drivers require different testpmd flags to optimize performance. So, reading SUT node config from get_ctx() and switching testpmd args based on this.
            * Without this change, some NICs will perform below their forwarding baseline
        * Build args for Intel:
            * Intel uses a special flag for the rx descriptor byte size, which would need to be included in the user’s test_run.yaml for the SUT to build “correctly.”
   * https://patchwork.dpdk.org/project/dpdk/patch/20250910122749.8277-1-abailey@iol.unh.edu/
        * We need to refactor the bound_for_kernel logic
        * Patrick needs to do a review the v5
   * Patrick to do a review for the tx_offload framework support and the mbuf_fast_free series
        * https://patchwork.dpdk.org/project/dpdk/list/?series=36153
        * Tech board is still discussing the right default behavior for mbuf_fast_free, which will determine how the testcase should be written
               
Any other business
      * Next Meeting is Oct 2, 2025