Re: Debugging Thinkpad T430s occasional suspend failure.

[Ingo Molnar <mingo@kernel.org>]

* Paul E. McKenney <paulmck@linux.vnet.ibm.com> wrote:

> On Mon, Feb 18, 2013 at 09:41:30AM +0100, Ingo Molnar wrote:
> > 
> > * H. Peter Anvin <hpa@zytor.com> wrote:
> > 
> > > On 02/16/2013 11:46 AM, Linus Torvalds wrote:
> > > >Adding Peter Anvin to the people, just in case he sees what's wrong
> > > >with the system call stub generation that keeps excessively old object
> > > >files around. If it's easy to fix, it might be worth trying to make it
> > > >ok to switch from i386 to x86-64 and back in the same tree.
> > > 
> > > I have not been able to reproduce this; it seems to Just 
> > > Work[TM].
> > > 
> > > The syscall header stuff is definitely not to blame: it 
> > > doesn't even *see* the CONFIG_ settings; instead they are only 
> > > used to determine which subset of files to create, but the 
> > > files themselves are configuration-independent.  This is by 
> > > design.
> > > 
> > > As such, without an actual known to fail test case there isn't 
> > > much I can do.
> > 
> > I tried (based on the versions given by Paul):
> > 
> >   git checkout v3.7-rc7
> >   make ARCH=i386 defconfig
> >   make -j64 bzImage
> > 
> >   git checkout v3.8-rc7
> >   make ARCH=x86_64 olddefconfig
> >   make -j64 bzImage
> > 
> > but it built just fine. The build bug might depend on the 
> > specific config file, or might depend on tooling details?
> 
> The problem is that the git tree I used is one that I build out of
> very infrequently, and so all I really know about the previous
> build was that it was done a long time ago.  :-(

Unless you recreated the Git repo from scratch, you could find 
clues about past versions used in .git/logs/HEAD.

Except if you used it to pull bits frequently - but built rarely 
out of it. In that case the information is probably lost.

(Don't worry about it too much - someone will eventually hit it 
again during bisection or so.)

Thanks,

	Ingo