Re: [Qemu-devel] [PATCH 2/6] Remove monitor.c dependency on CONFIG_I8259

[David Gibson]

On Tue, Mar 31, 2015 at 11:57:28AM +0200, Markus Armbruster wrote:
> David Gibson <address@hidden> writes:
> 
> > On Mon, Mar 30, 2015 at 10:37:45AM +0200, Markus Armbruster wrote:
> [...]
> >> This adds "info irq" and "info pic" to the targets that didn't have them
> >> before, implemented by i8259.c's hmp_info_pic() and hmp_info_irq().
> >> They do nothing unless the machine has an "isa-i8259" device.
> >> 
> >> Cases:
> >> 
> >> 1. If the machine has one, and it's the only interrupt controller, the
> >> commands work as advertized.
> >> 
> >> 2. If the machine doesn't have one, the commands are empty promises.
> >> 
> >> 3. If the machine has one, but it's not the only interrupt controller,
> >> the commands confidently claim the i8259 is all there is.
> >> Misinformation.
> >> 
> >> Cases 2 and 3 are common, case 1 is rare.
> >> 
> >> We can:
> >> 
> >> A. Fix the commands to cover all interrupt controllers.
> >> 
> >> B. Fix them to warn the user about missing interrupt controllers.
> >> 
> >>    We can approximate this by warning always, because it's almost never
> >>    the only interrupt controller anyway :)
> >> 
> >> C. Rip 'em both out and be done with it.
> >> 
> >> D. Do nothing.
> >> 
> >> E. Provide them as is on all targets.
> >> 
> >>    Spread the badness fairly.
> >> 
> >> I vote for C or B.  A seems not worthwhile.
> >
> > I'd love to do C, if we can get confirmation that no-one's really
> > using the existing HMP commands.  That would make a bunch of things
> > simpler.
> 
> I gave it a shot, let's see how people react.

Heh, you beat me to it.

Thanks for the detailed analysis.

-- 
David Gibson                    | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au  | minimalist, thank you.  NOT _the_ _other_
                                | _way_ _around_!
http://www.ozlabs.org/~dgibson

pgpCYsQGq0rM1.pgp
Description: PGP signature