Re: [PATCH] earlyprintk: Clean up pciserial

[Peter Zijlstra <peterz@infradead.org>]

On Thu, Aug 25, 2022 at 02:51:43PM -0500, Bjorn Helgaas wrote:
> On Thu, Aug 25, 2022 at 09:36:35PM +0200, Peter Zijlstra wrote:
> > On Thu, Aug 25, 2022 at 12:39:59PM -0500, Bjorn Helgaas wrote:
> > 
> > > > -	   (((classcode >> 8) & 0xff) != 0x02)) /* 16550 I/F at BAR0 */ {
> > 
> > > > diff --git a/include/linux/pci_ids.h b/include/linux/pci_ids.h
> > > > index 6feade66efdb..1d2c7df8cd41 100644
> > > > --- a/include/linux/pci_ids.h
> > > > +++ b/include/linux/pci_ids.h
> > > > @@ -75,6 +75,9 @@
> > > >  #define PCI_CLASS_COMMUNICATION_MODEM	0x0703
> > > >  #define PCI_CLASS_COMMUNICATION_OTHER	0x0780
> > > >  
> > > > +/* I/F for SERIAL/MODEM */
> > > 
> > > I/F?  Grep says it's fairly common, but doesn't seem completely
> > > obvious.  I guess it means "interface"?
> > 
> > It does, I carried the nomenclature from the above line; happy to change
> > it if you want.
> > 
> >   https://wiki.osdev.org/PCI#Class_Codes
> > 
> > Calls it 'Prog IF'.
> 
> Personally, I would spell it out just to make it less arcane.  I
> wouldn't know where to find the definition, and the PCI specs don't
> seem to use the abbreviation.  But no big deal either way.

Ok,

/* Interface for SERIAL/MODEM */

it is.