Closed solardiz closed 2 months ago
Looks like linux-headers-generic
(linux-headers-5.15.0-102-generic
) brings /lib/modules/5.15.0-102-generic
which confuses the build script.
(Fix is obvious but) I am thinking, how useful is this ARMv7 boot test at all? Maybe it's time to just remove it?
Also, the test imitate real hardware and thus brings linux-firmware
which is 267 MB which takes 12 seconds to just download it on GA, and locally here ~7 minutes..
Perhaps, it's more useful to re-enable i386` test (it's not explained in commit message why it's disabled temporary).
Also, the test imitate real hardware and thus brings
linux-firmware
which is 267 MB which takes 12 seconds to just download it on GA, and locally here ~7 minutes..
[About replacing the test kernel, they have linux-virtual-hwe-22.04
(the only) which would not bring linux-firmware
but it seems to be 6.5.0-27
, which will invalidate 'old kernel testing' goal and requires gcc-12 to build the module for it, thus making replacement not feasible.]
how useful is this ARMv7 boot test at all? Maybe it's time to just remove it?
As long as we do support this arch and it's not hard to keep this test working, I think we should keep it. I guess it may also just happen to uncover (future) LKRG issues not limited to this arch, just due to it being different than other tests.
Perhaps, it's more useful to re-enable i386` test
If we can, sure, let's re-enable it. But that's separate, unrelated to us keeping this test or not.
seems to be 6.5.0-27, which will invalidate 'old kernel testing' goal
I'm not sure we have that goal here. We could test old kernels on other archs.
Overall, I think for now let's go for the "obvious" fix you mentioned. Thank you, @vt-alt!
@vt-alt This has started failing https://github.com/lkrg-org/lkrg/actions/runs/8727807004/job/23946138137