Update DPU detection logic.

[crypticC0der]

In recent research with real hardware we have found our previous assumptions to be incorrect, so we are updating it to be controller = 1 and pcipf OR pcivf in the devlink port output 0 1.

[fnordahl]

@crypticC0der the dqlite build error persists, I also checked on main branch. From my perspective this is likely technical debt in dqlite, which has now been forced into error by toolchain changes in resolute.

To unblock this a tactical approach could be:

• Raise an issue with dqlite.
• Reference said issue in snap/snapcraft.yaml as we add CFLAGS: -Wno-error=discarded-qualifiers to build-environment in the dqlite part.

[crypticC0der]

Raise an issue with dqlite.

Why raise an issue when you can raise a PR

[crypticC0der]

the snapcraft change is based on not having this merged canonical/dqlite#875, so when it gets merged and added in a release we can remove it which will be nice

[fnordahl]

Intriguing, now we have a Go crash fatal error: concurrent map writes (LP#2144984). Unsure if this is the source of the intermittent failure of the upgrade test, or if it goes in hand with toolchain and/or dqlite changes.

[fnordahl]

Intriguing, now we have a Go crash fatal error: concurrent map writes (LP#2144984). Unsure if this is the source of the intermittent failure of the upgrade test, or if it goes in hand with toolchain and/or dqlite changes.

Given it just passed after another retry it is likely this is the source of the intermittent test failure \o/. Let's deal with that separately though.

[fnordahl]

Also, LP#2144987

[fnordahl]

LGTM, thanks!