Comment 113 for bug 1690085

Revision history for this message
In , oyvinds (oyvinds-linux-kernel-bugs) wrote :

(In reply to James Le Cuirot from comment #27)
> I wouldn't call it as a regression as the kernel options involved are merely
> a workaround for the underlying issue.

That's true and we shouldn't have to apply this workaround. Still, on Fedora it's the regression in my opinion because kernel 4.12.x (happens to) work fine and 4.13.x doesn't.

As a end-user my conclusion was simply that 4.13.x is broken and I kept using 4.12.x until I had time to look into why 4.13.x is broken. And it still is for many, my 68 year old mother isn't going to be adding no rcu_nocbs= kernel command line on her laptop or do much beyond using it to check the news and weather. I am fairly sure I wouldn't be able to guide her through doing so even if I spent 10 hours on the phone, I'd have to go there. Luckily it's got some old Intel CPU.

Yes, these are kernel options that some enable and some don't - but in the case of Fedora it's You update, you get new kernel, you're screwed and now your machine hangs randomly - which it previously did not. Isn't that the definition of regression?