[VPP-2091] Memif crashes VPP in container with jumbo frames

[vvalderrv]

Description

When first reported, VPP was crashing.

Here [0] is an example core. The exact place of the crash is not always the same, although I think memif_interface_tx_zc_inline is always involved when CSIT tests sees VPP crashing on the host machine. My experiments with VPP code were not able to fix the issue, but some edits moved the crash to "container" VPP process (not configured to create core on crash).

I will continue trying to simplify the CSIT setup. The current setup is like this [1] for the container VPP and this [2] for the host VPP.

[0] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/vpp-csit-verify-perf-master-ubuntu2204-x86_64-2n-clx/117/csit_current/0/log.html.gz#s1-s1-s1-s1-s1-t2-k3-k4-k1

[1] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/vpp-csit-verify-perf-master-ubuntu2204-x86_64-2n-clx/117/csit_current/0/log.html.gz#s1-s1-s1-s1-s1-t2-k2-k8-k8-k3-k1

[2] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/vpp-csit-verify-perf-master-ubuntu2204-x86_64-2n-clx/117/csit_current/0/log.html.gz#s1-s1-s1-s1-s1-t2-k3-k2

Assignee

Unassigned

Reporter

Vratko Polak

Comments

• vrpolak (Mon, 18 Nov 2024 11:52:02 +0000): In rls2410, I see crash [4] involving mlx5_tx_burst_mt_empw.

[4] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-coverage-2410-2n-icx/13/log.html.gz#s1-s1-s1-s1-s1-t6-k3-k4-k1

• vrpolak (Tue, 30 Jul 2024 11:42:48 +0000): In rls2406 [3], only the VPP inside container is crashing, so no core is easily available.

[3] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-coverage-2406-2n-clx/25/log.html.gz#s1-s1-s1-s1-s1-t6-k3

Original issue: https://jira.fd.io/browse/VPP-2091