Re: [stable] Removing or restricting timer_stats

[Greg KH <greg@kroah.com>]

On Wed, Apr 19, 2017 at 08:01:18AM -0700, Kees Cook wrote:
> On Wed, Apr 19, 2017 at 7:57 AM, Greg KH <greg@kroah.com> wrote:
> > On Wed, Apr 19, 2017 at 07:54:34AM -0700, Kees Cook wrote:
> >> On Wed, Apr 19, 2017 at 4:50 AM, Greg KH <greg@kroah.com> wrote:
> >> > On Tue, Apr 18, 2017 at 03:25:05AM +0100, Ben Hutchings wrote:
> >> >> The timer_stats feature was removed upstream by:
> >> >>
> >> >> commit dfb4357da6ddbdf57d583ba64361c9d792b0e0b1
> >> >> Author: Kees Cook <keescook@chromium.org>
> >> >> Date:   Wed Feb 8 11:26:59 2017 -0800
> >> >>
> >> >>     time: Remove CONFIG_TIMER_STATS
> >> >>
> >> >> I'm hesitant to propose removing a feature in stable, even if it is
> >> >> redundant.  What I've done for Debian stable is to restrict it to the
> >> >> initial pid namespace (see attached).  Would that be a reasonable
> >> >> alternative change for stable branches?
> >> >
> >> > I don't mind removing things in stable as that's what happened in
> >> > Linus's tree, and it was removed for a reason.  We've done it before,
> >> > and I'm more hesitant to apply something that works a bit
> >> > "differently".
> >> >
> >> > Kees, any objection for me just taking the "full" patch in the stable
> >> > kernels?
> >>
> >> Arjan said it would break powertop, IIRC.
> >
> > So we broke it in 4.11?  That doesn't seem to sound reasonable, how are
> > you getting away with that?  :)
> 
> tglx seemed to think that the same information was available elsewhere?
> 
> Thread was here: https://patchwork.kernel.org/patch/9561519/

Getting the same info elsewhere is fine, breaking working tools is not.

I'm running powertop on 4.11-rc6 here and there doesn't seem to be
anything "broken".  And, as this is removed there, I don't see why
backporting it should matter.  Either it breaks someone, or it doesn't,
the kernel version shouldn't matter...

thanks,

greg k-h