Re: something's wrong

[Thomas Klausner <wiz%netbsd.org@localhost>]

On Tue, Nov 05, 2013 at 10:22:29AM +0000, Chavdar Ivanov wrote:
> I'll try some bisection.

My bisection so far points to a change on the 23rd.

I couldn't reproduce the problem (yet?) with a kernel from 'cvs up -D
20131023', but I had a hang with one from 'cvs up -D 20131024'.

The changes from the 23rd are:

7766   L Oct 23 To source-chang (0.4K) CVS commit: src/sys/dev/usb
just a config change

7767   L Oct 23 To source-chang (0.4K) CVS commit: src/sys/dev/pci
gffb, which I don't use

7768   L Oct 23 To source-chang (0.4K) CVS commit: src/sys/arch/macppc/conf
kernel config only

7769   L Oct 23 To source-chang (0.3K) CVS commit: src/doc
docs

7770   L Oct 23 To source-chang (0.4K) CVS commit: src/sys/dev/pci
gffb copyright change

7771   L Oct 23 To source-chang (0.4K) CVS commit: src/sys/arch/amd64/conf
+ xhci (which I've removed in my kernel)

7772   L Oct 23 To source-chang (0.4K) CVS commit: src/sys/arch/i386/conf
same

7773   L Oct 23 To source-chang (0.6K) CVS commit: src
use MODULE_CLASS_MISC for Lua modules
which looks harmless enough and doesn't really touch the kernel

7774   L Oct 23 To source-chang (3.1K) CVS commit: src/sys

Use the MI "pcu" framework for bookkeeping of npx/fpu states on x86.
This reduces the amount of MD code enormously, and makes it easier
to implement support for newer CPU features which require more fpu
state, or for fpu usage by the kernel.
For access to FPU state across CPUs, an xcall kthread is used now
rather than a dedicated IPI.
No user visible changes intended.

7775   L Oct 23 To source-chang (0.5K) CVS commit: src/sys/arch/arm/arm
different architecture

So currently it looks to me like the problem might come from the MI
pcu framework commit. Matthias, does that sound possible? Could you
please check the patch again if it might introduce these symptoms?
(see thread for details)
 Thomas

> Chavdar
> 
> On 5 November 2013 10:07, David Brownlee <abs%netbsd.org@localhost> wrote:
> > On 5 November 2013 08:27, Chavdar Ivanov <ci4ic4%gmail.com@localhost> wrote:
> >> Tried - the screen stays as it was, no movement whatsoever. No network
> >> activity, the keyboard is also dead - does not respond to CtrlAltEsc,
> >> even the lights are off and CapsLock/NumLock does not trigger
> >> anything.
> >
> > I've noticed a recent netbsd-6 xen DOM0 hanging on similar big
> > compiles (firefox24/25) if run while the DOMUs are active. Possibly
> > unrelated, but just as a data point.
> 
> 
> 
> -- 
> ----
>