Re: [PATCH] Kbuild: Move -Wmaybe-uninitialized to W=1

[Ingo Molnar]

* Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:

> On Thu, Jul 28, 2016 at 12:03 PM, Linus Torvalds
> <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> >
> > Once we get to the point that the warning is no longer useful, and is more 
> > pain than gain, it gets disabled.
> 
> Btw, I have a suspicion that you didn't realize that "-Wmaybe-uninitialized" is 
> separate from "-Wuninitialized" (which is *not* disabled).

I very much know the difference, as you can see from the commit IDs I cited.

> The "maybe-uninitialized" warning is literally gcc saying "I haven't really 
> followed all the logic, but from my broken understanding it isn't _obvious_ that 
> it is initialized".
> 
> And the problem is that a lot of gcc optimization choices basically move the 
> pointer of "obvious". So the warning is a bit random to begin with. And when the 
> gcc people screw thigns up, things go to hell in a handbasket.

Fair enough.

Thanks,

	Ingo