Re: kern/46885: NetBSD 6.0_RC1 spontaneously reboots as kernel starts to load

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

The following reply was made to PR kern/46885; it has been noted by GNATS.

From: Manuel Bouyer <bouyer%antioche.eu.org@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: kern-bug-people%NetBSD.org@localhost, gnats-admin%NetBSD.org@localhost, 
netbsd-bugs%NetBSD.org@localhost
Subject: Re: kern/46885: NetBSD 6.0_RC1 spontaneously reboots as kernel
 starts to load
Date: Sun, 2 Sep 2012 00:39:04 +0200

 On Sat, Sep 01, 2012 at 11:13:44PM +0200, Manuel Bouyer wrote:
 > I can confirm this on a DELL optiplex with a "Intel(R) Celeron(R) CPU 
 > 2.80GHz".
 > 
 > 
 > > [...]
 > > Compiling a kernel with options DIAGNOSTIC commented out and options DEBUG
 > > uncommented and testing that works perfectly as well!
 > 
 > I looked at low-level code that would have
 > #if defined(DIAGNOSTIC) || defined(DEBUG)
 > but didn't spot anything (the only file I found in i386 and x86 which
 > have something like that is apmcall.S, which is not compiled in).
 
 I did some experiments:
 - a kernel with the files from i386/i386 and *acpi*.o compiled -DDEBUG
   hangs instead of resetting
 - a kernel with kern_*.o,  *acpi*.o and uvm*.o compiled with -DDEBUG
   (and everything else without -DDEBUG) resets just after the first line
   of green message (which I think is about symbols, but it resets too fast
   to really be sure)
 - a kernel with everything but kern_*.o,  *acpi*.o and uvm*.o compiled with
   -DDEBUG boots
 
 So, at first glance, it looks like DEBUG or DIAGNOSTIC doesn't make it
 work because it changes code, but because it changes sizes.
 
 -- 
 Manuel Bouyer <bouyer%antioche.eu.org@localhost>
      NetBSD: 26 ans d'experience feront toujours la difference
 --