Re: Simplified Steps for Building a Loadable module on -CURRENT

On Friday 01 September 2006 20:57, Pyun YongHyeon wrote:
> On Fri, Sep 01, 2006 at 12:14:28PM -0400, John Baldwin wrote:
>  > On Thursday 31 August 2006 23:59, Pyun YongHyeon wrote:
>  > > On Thu, Aug 31, 2006 at 03:28:13PM -0400, John Baldwin wrote:
>  > >  > On Thursday 31 August 2006 06:22, Pyun YongHyeon wrote:
>  > >  > > On Thu, Aug 31, 2006 at 05:32:13PM +0800, LI Xin wrote:
>  > >  > >  > Pyun YongHyeon wrote:
>  > >  > >  > > On Wed, Aug 30, 2006 at 03:12:59PM -0700, David Christensen 
wrote:
>  > >  > >  > >  > I've been able to successfully build drivers in the past 
as
>  > >  > >  > >  > loadable modules but I'm getting some kernel panics 
>  > with -CURRENT
>  > >  > >  > >  > when installing a module using kldload now where things 
used to 
>  > >  > >  > > 
>  > >  > >  > > I wonder you encountered the same panic I have been seeing on 
>  > CURRENT.
>  > >  > >  > > I get "Fatal trap 30" message when I load em(4) module with 
>  > kldload.
>  > >  > >  > 
>  > >  > >  > What does Fatal trap 30 mean in these places?  I get some 
strange 
>  > fatal
>  > >  > >  > trap 30's in acpi_cpi_idle, but I can not imagine how can these 
>  > >  > happen :-(
>  > >  > >  > 
>  > >  > > 
>  > >  > > Don't know what's cause of the panic since it used to work ok.
>  > >  > > See 
>  > >  > 
>  > 
http://lists.freebsd.org/pipermail/freebsd-current/2006-August/065243.html
>  > >  > 
>  > >  > Trap 30 means an IDT vector fired that we didn't expect.  In this 
case, I 
>  > >  > think it may only happen on SMP, and it maybe that the interrupt 
gets 
>  > sent to 
>  > > 
>  > > Yes, it's SMP(i386).
>  > 
>  > Can you try disabling SMP via kern.smp.disabled?
>  > 
> 
> Thank you. Setting kern.smp.disabled fixed the panic.

s/fixed/masked/.  It still needs fixing.  The only thing I can think of the 
smp_rendezvous() change should have fixed. :(

-- 
John Baldwin