Re: [PATCH 2/2] x86/mtrr: Refactor PAT initialization code

From: Ingo Molnar
Date: Fri Mar 11 2016 - 04:14:07 EST

* Ingo Molnar <mingo@xxxxxxxxxx> wrote:

> * Toshi Kani <toshi.kani@xxxxxxx> wrote:
> > MTRR manages PAT initialization as it implements a rendezvous
> > handler that initializes PAT as part of MTRR initialization.
> >
> > When CPU does not support MTRR, ex. qemu32 virtual CPU, MTRR
> > simply skips PAT init, which causes PAT left enabled without
> > initialization. [...]
> What practical effects does this have to the user? Does the kernel crash?

Btw., I find this omission _highly_ annoying: describing what negative effects a
bug _causes in practice_ is the most important part of a changelog. How on earth
can an experienced contributor omit such an important component from a patch

Most readers of changelogs couldn't care less about technical details of how the
bug is fixed (of course others will read it so it's nice to have too), but what
symptoms a bug causes, how serious is it, whether it should be backported are like
super important compared to everything else you wrote - and both the description
and the changelogs are totally silent on those topics ...

I've seen this in other PAT patches - please try to improve this.