Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Error parsing kernel release number in ifcfg module #5087
When I install ka-lite in my ARM device, I can't get it to start without crashing, due to a parsin problem in the ifcfg module, which is printing this error instead:
Looking at the relevant code it seems to me that there are two issues:
Branch or installer method
The crash is happening both in the stable version installable with
The wrong comparison against the 3.3 version seems to be wrong regardless of the underlying system.
How to reproduce
Alternatively replace this line with something
added a commit
Apr 15, 2016
referenced this issue
Apr 15, 2016
No problem, happy to help! For the record, I've just closed the pull request and proposed the same change in the python-ifcfg module instead, following the recommendation from #5088 (comment).
Would rather keep this open while KA Lite is not actually fixed, though, just to keep track of it properly.
@benjaoming Not very rigorous (but hopefully good enough), I've did this to confirm that the fix works:
Additionally, I added a couple of prints in
@MCGallaspy It looks like the #5090 Pull Request is all that is left from your suggestion in #5088 (comment), so I think this can probably be closed as soon as the new release is out and tested to work fine in a more rigorous way (which I volunteer for helping with)
Thanks @mariospr ! I see that the PR also passes automated tests, so I'm quite confident it's a safe merge for a patch release.
Could you weigh in about how you see the severity of the issue? I think there's gonna be a discussion about whether or not to immediately release a patch version. This error has been around forever so the real question is if the sort of Linux Kernel version string is very common or perhaps has become so :)
Cool, thanks for your help and (extremely) timely response!
I'm not really a kernel expert so I can't tell how common is to have kernels reporting only the VERSION and MAJOR REVISION numbers as their release number string via the
However, I've checked this with one of our Kernel guys and he explained to me that "the minor version is something used by people in several non-standard ways" and so that it can't really be trusted (actually, it seems that ubuntu does a poor job reporting that number), so it's probably wise not to rely on it at all, as if it was always being reported as part of the kernel's release number (output of
But on top of that, the patch also fixed a wrong use of the
All of this being a long way of saying that... the issues fixed are important, but only if you either run ifcfg in a OS not reporting the kernel version (which will crash) or if you run it in OS with a 3.x kernel >= 3.10 (which will treat as < 3.3). Still, we (Endless) would particularly appreciate if it could make it to the next stable release, for obvious reasons :-).
Cool, thanks for explaining it thoroughly! I think it's fair to aim for a 0.16.3 within the next ~1 week's time to accommodate the need! I'll let you know as soon as it's done, and this ticket should remain until we're resolved on when we get the fixed released.
Great! Happy to see my very first contribution to this project so quickly and efficiently dealt with :-). Let me know if you need anything from my side. Otherwise I'll wait until the release is done for testing it again in my (problematic) ARM hardware. Thanks!