Re: new kernel resets system - "solution"

[Kurt Schreiner <ks%ub.uni-mainz.de@localhost>]

On Fri, Feb 05, 2010 at 11:06:09PM +0000, Quentin Garnier wrote:
> On Fri, Feb 05, 2010 at 10:36:07PM +0100, Kurt Schreiner wrote:
> > Hi,
> > 
> > reverting 
> > 
> >  sys/kern/kern_mutex.c
> >  sys/kern/kern_rwlock.c
> >  sys/sys/mutex.h
> >  sys/sys/rwlock.h
> > 
> > to yesterday's versions yields a working kernel again.
> > (at least on i386...)
> 
> Would the following "fix" it by any chance?
> 
> Index: kern_mutex.c
> ===================================================================
> RCS file: /cvsroot/src/sys/kern/kern_mutex.c,v
> retrieving revision 1.47
> diff -u -r1.47 kern_mutex.c
> --- kern_mutex.c      5 Feb 2010 06:43:16 -0000       1.47
> +++ kern_mutex.c      5 Feb 2010 23:03:58 -0000
> @@ -181,7 +181,8 @@
>      (((mtx)->mtx_owner & MUTEX_BIT_SPIN) == 0)
>  
>  #define      MUTEX_DEBUG_P(mtx)      (((mtx)->mtx_owner & MUTEX_BIT_NODEBUG) 
> == 0)
> -#if defined(LOCKDEBUG)
> +//#if defined(LOCKDEBUG)
> +#if 1
>  #define      MUTEX_OWNED(owner)              (((owner) & ~MUTEX_BIT_NODEBUG) 
> != 0)
>  #define      MUTEX_INHERITDEBUG(new, old)    (new) |= (old) & 
> MUTEX_BIT_NODEBUG
>  #else /* defined(LOCKDEBUG) */
Hm, sorry, spooke to soon, i.e. I had left "options LOCKDEBUG" in my
kernel config. With only the "fix" above I run into:

panic: uao_create: hashinit swhash failed

bt shows: (2 finger copy ;)

breakpoint
panic
uao_create(3fc000000,2,240,1000,c0554440,2,0,c0348745,ffc00000,c2ac4000) at
netbsd:uao_create+0x26b
uvm_init(c052e5e0,2,0,2,0,...,0) at netbsd:uvm_init+0x88
main(0x,c01002ca,0,...,0) at netbsd:main+0x51


Kurt