[DPDK/DTS Bug 1500] Create process for adding VFs
bugzilla at dpdk.org
bugzilla at dpdk.org
Thu Jul 25 20:48:10 CEST 2024
https://bugs.dpdk.org/show_bug.cgi?id=1500
Bug ID: 1500
Summary: Create process for adding VFs
Product: DPDK
Version: unspecified
Hardware: All
OS: All
Status: UNCONFIRMED
Severity: normal
Priority: Normal
Component: DTS
Assignee: dev at dpdk.org
Reporter: jspewock at iol.unh.edu
CC: juraj.linkes at pantheon.tech, probb at iol.unh.edu
Target Milestone: ---
VFs are used in multiple of the ethdev test suites that we were planning to
port over into new DTS, but we don't have a method of creating them currently.
Judging by the process that is laid out in multiple test_plans (like
stats_check for example:
https://git.dpdk.org/tools/dts/tree/test_plans/stats_checks_test_plan.rst) it
seems like actually creating the VF and binding it to vfio-pci is pretty
simple. However, because the VFs are on top of the PF that exist on the NIC's
ports, it raises a few questions about where VFs should live in the framework
and how they should be managed, since they get their own PCI addresses.
It might be enough to just add VFs as a list of objects under a port and manage
them that way, but they almost look like ports themselves to testpmd and
devbind. Also, since we currently start testpmd by explicitly allowing the PCI
addresses for the ports in the current testrun, do we also include all VFs
configured on the ports? Do we always want testpmd to see them if they are
configured? There are a few things to discuss regarding how to implement them
properly.
--
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mails.dpdk.org/archives/dev/attachments/20240725/c9194e70/attachment-0001.htm>
More information about the dev
mailing list