Am 18.04.2013 um 10:30 schrieb Peter Maydell <peter.mayd...@linaro.org>:

> On 18 April 2013 07:49, Marc Zyngier <m...@misterjones.org> wrote:
>> If you need an early console, why not simply wire the 8250 emulation in
>> kvmtool to be useable from the MMIO bus? I reckon this would solve your
>> problem in a more elegant way...
> 
> The other approach I thought of would be something involving
> defining a hypercall interface for console I/O, in the same
> way that we have hypercalls for "start cpu"/"stop cpu"/etc.
> Is there any mileage in considering that approach, or is it
> a non-starter?

It's exactly what we had for the s390-virtio machine. Virtio-console as console 
device plus a hypercall for early printk.

It was a mess.

Trying to inject character output that lands in machine context, where 
hypercalls get handled, inside of a specific virtio-console device, which owns 
the char output, is hard. We haven't found any good solution in qemu to layer 
this properly.

The closest approach to something workable was to create 2 char outputs and mux 
them together, like you usually would mus monitor and serial. Good luck muxing 
that one again with the monitor :).

I'd rather spare you guys from going through the same pain.


Alex

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Reply via email to