Dave Jones wrote:

>On Fri, Apr 26, 2002 at 10:00:50PM -0500, Corey Minyard wrote:
>
>Hi Corey,
>
> > --- ./arch/i386/kernel/pci-pc.c.smbios	Fri Apr 26 10:59:55 2002
> > +++ ./arch/i386/kernel/pci-pc.c	Fri Apr 26 11:00:13 2002
> > @@ -1293,6 +1293,10 @@
> >  	return;
> >  }
> >  
> > +#ifdef CONFIG_SMBIOS
> > +extern void smbios_init(void);
> > +#endif
> > +
> >  void __init pcibios_init(void)
> >  {
> >  	int quad;
> > @@ -1322,6 +1326,10 @@
> >  
> >  	pcibios_fixup_irqs();
> >  	pcibios_resource_survey();
> > +
> > +#ifdef CONFIG_SMBIOS
> > +	smbios_init();
> > +#endif
> >  
> >  #ifdef CONFIG_PCI_BIOS
> >  	if ((pci_probe & PCI_BIOS_SORT) && !(pci_probe & PCI_NO_SORT))
>
>Any reason for initialising it there instead of using a subsys_initcall
>from smbios.c ?
>
Can you control other orders within the subsystem?  Other things that 
come later in this subsystem might need it.  Actually, it might be 
needed before this, I've moved it to before the PCI initialization, 
since this contains interrupt information and possible PCI information.

>
>
> > +void __init hexdump (u8 *data, int len)
>
>Something this generic sounding should be static.
>
It's actually supposed to be in a #ifdef DEBUG, and it should be static, 
too.

>
>
> > +	char str[80];
>
>Worth adding a if (len>80) return here in case of crap biosen?
>Or am I overly paranoid?
>
It's printing 16 hex characters every time, so it doesn't matter.  And 
it's only debug code, anyway.

I've attached another patch.

-Corey