-
Notifications
You must be signed in to change notification settings - Fork 67
Mlx5 offload enable #2676
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Mlx5 offload enable #2676
Conversation
80aa366
to
ad11c46
Compare
A link to the thread that started this PR. |
The pipelines failed with:
This is because the initrd now contains a phantom cpio piece, I solved this in my kernel upgrade PR here, probably will solve this issue: 2be94c2. |
Build action triggered: https://github.com/flatcar/scripts/actions/runs/14363396624 |
dc13801
to
5dc24e0
Compare
On my local machine I was able to pass the build packages and build images. I am hopeful that this will also be the case in CI. |
The arm64 build failed due to CONFIG_SWITCHDEV being in the amd64-only config, and CONFIG_VFIO_PCI_{VGA,IGD} in commonconfig actually being x86 only. |
Also exectued the |
I do not. Thank you for the help. Right now the tests are failing, does that mean things have gotten too big? Did I miss documentation somewhere on a size limitation? |
I would like to help here, but I am not sure what to do. |
@jepio Did it fail because of this? |
Tests run fine actually, devcontainer fails (that's expected) and the PR tests can't post messages to this issue (also expected). If the code review of this PR is good, we can merge. Build and tests can be considered good. |
Signed-off-by: Jeremi Piotrowski <[email protected]>
Signed-off-by: Jeremi Piotrowski <[email protected]>
Requires for mlx5 hardware offload on arm64 too. Signed-off-by: Jeremi Piotrowski <[email protected]>
These options are x86 only. Signed-off-by: Jeremi Piotrowski <[email protected]>
Signed-off-by: Jeremi Piotrowski <[email protected]>
300ead6
to
f39ee44
Compare
I've rebased to address the conflict, and I'm now running this under our internal CI, including on Azure. |
Signed-off-by: James Le Cuirot <[email protected]>
I remembered what Thilo said in the meeting, so I've now cancelled that and started it again after bumping the Azure machine sizes to v6. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That successfully passed. I'm slightly concerned about the 331KB kernel image increase, but we have sufficient space before I resolve that issue.
[Enable Hardware Offloads in Mellanox driver ]
The configuration changes in this PR will allow for hardware offloading of traffic control and connection tracking operations. There is a netdev conf paper that talks about these features.
How to use
These kernel configuration changes will allow the hardware offload features of the Mellanox nics to be used. The usage of these features comes from user space. For those interested in containerized workloads, one possible way to use these is similar to Red Hat Openshift allows this configuration (via Open vSwitch). Outside of ovs, these offloads can be configured via the
tc
ornft
commands.Testing done
I have compiled these changes, and booted a physical node equipped with a mellanox Cx-7.
changelog/
directory (user-facing change, bug fix, security fix, update)/boot
and/usr
size, packages, list files for any missing binaries, kernel modules, config files, kernel modules, etc.