Re: [PATCH RFC 3/9] console: add EHCI debug port based serial console

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Tue, Jul 31, 2012 at 08:10:25AM +0100, Jan Beulich wrote:
> >>> On 30.07.12 at 22:51, Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> wrote:
> >> +    set_fixmap_nocache(FIX_EHCI_DBGP, dbgp->bar_val);
> > 
> > Should this have dbgp->bar_val & PAGE_MASK ?
> 
> No, the set_fixmap_nocache() implementation shifts out the low
> bits.
> 
> >> +    case PHYSDEVOP_DBGP_RESET_PREPARE:
> > 
> > Oh, you have a handoff protocol!
> > 
> > So how does this work without using this? Meaning if one uses Xen hypervisor
> > EHCI with a pvops kernel that does not implement this?
> 
> The hypervisor will lose access to its console (and the next console
> access will cause a brief hang that does no other harm). That's the
> same behavior as for native Linux if you ripped out the handoff
> protocol (and the effect is even specified that way by the
> specification). Plus - no-one is required to use the new functionality
> when (s)he's aware that Dom0 doesn't cope.

Sure.
> 
> >> +        return ehci_dbgp_external_startup(&ehci_dbgp) ?: 1;
> > 
> > Oh, this is new. With the ?: it just passes on the return value from
> > ehci_dbgp_external_startup?
> 
> Yes, but it's not new. Just grep for the construct in Xen or Linux -
> it's being used in a number of places, and has been supported by
> gcc virtually forever.

Ok, never hit upon it until now. Either way, you can tag on Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
if you would like.

Thanks!
> 
> Jan