Kernel 'bisection'

[Gene Heskett]

On Saturday, February 14, 2015 07:57:28 AM Liam Proven wrote:
> On 14 February 2015 at 02:59, Chris <cpollock at embarqmail.com> wrote:
> > Who can give me some pointers on doing a kernel bisection on either
> > 3.13.0-35 or -36 kernels? Ever since one of those two were installed,
> > I'm on -45 now, my system will lockup (black screen, cursor moves,
> > but CTRL-ALT-F1 will not take me to a terminal log-in. This is the
> > error I see - Feb 13 19:05:22 localhost kernel: [807775.808019]
> > [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed...
> > render ring idle. I've reported this last year on Launchpad as bug -
> > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1402331 I've
> > jumped through all the hoops that have been suggested there except
> > doing a kernel bisection which seems to have me confused. I'm really
> > getting a bit tired of these continued 'lockups'.
> 
> I have to ask. WTF is a "kernel bisection"?
> 
> You want to... cut your kernel in half? Is this some kind of Solomonic
> judgement? Because, you know, like half a baby doesn't work, half a
> kernel won't, either.

You Liam, I would think would know what that is, but when you are building 
your own kernels, and the next version you build doesn't work, you backup 
to an older version that does work, then apply half of the patches in the 
same sequence as they were applied to get from one that works to one that 
doesn't to build an intermediate kernel & see if it works, if not, next 
pass only applies 1/4 of the patches, or if it does work, apply 1/4 of the 
remaining patches.  In 8 or maybe 9 such builds you can nail it down to 
the specific patch that broke it.  Successive approximation, works well 
for finding bad patches.  Works well if you are following the -rc# as they 
are released, but it does get cumbersome to find whu a 2.6.32 works, but a 
3.18 doesn't. That may take twice the builds.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
US V Castleman, SCOTUS, Mar 2014 is grounds for Impeaching SCOTUS