Comment 42 for bug 2038249

Hi Timo,

Thanks for looking into this.

> Wouldn't it be less risky to just backport 8.0.4 instead of 24/36 commits to mantic/jammy?

So so; and there's more involved.

There are near-future steps being taken in that direction already (SRU exception to be discussed in the upcoming Engineering Sprint), but that is likely for a time-frame longer than these SRUs (which are much wanted sooner), considering the test requirements and preparation work that is needed.

The regression risk in this case is mostly with the Jammy GA/5.15 kernel, since its HWE kernel (6.5) and Mantic GA/6.5 kernel are already unusable with their current crash version.

To that extent, AFAICT from the individual patch reviews, they are all built with backwards compatibility, so as long as crash with the Jammy 5.15 GA kernel is well tested (and the detailed test plans should ensure this), that should be OK.

Therefore, under the assumption/understanding that the regression risk is indeed low (and there are benefits of making crash usable at all with the 6.5-based kernel HWE/GA kernels), and additionally considering the engineering time already put into this (if results are aligned with SRU policy), it would seem reasonable to move forward with the SRUs instead of MREs.

Thanks again.