Re: [Qemu-devel] [PATCH v1 12/13] q35: fill in usb pci slots with -usb

[Jason Baron <jbaron@redhat.com>]

On Tue, Oct 30, 2012 at 05:19:01PM +0100, Gerd Hoffmann wrote:
> On 10/30/12 16:19, Jason Baron wrote:
> > On Tue, Oct 30, 2012 at 07:34:26AM +0100, Gerd Hoffmann wrote:
> >>   Hi,
> >>
> >>> +            uhci_devname[sizeof(uhci_devname) - 2] = ((char)'1') + i;
> >>
> >> snprintf(devname, sizeof(devname), "...%d", i) is more readable.
> > 
> > ok.
> > 
> >>
> >>> +            qdev_prop_set_string(usb_qdev, "masterbus", "ich9-usb-bus.0");
> >>
> >> Any reason why you rename the usb bus?
> >>
> > 
> > I wasn't sure if the user created usb devices on the command-line via
> > -device if that would break naming here. Thus, I added a 'private' name.
> > If the naming is stable, that works. It would be 'usb-bus.0', in that
> > case?
> 
> "usb.0" would be the default name, but you don't need to know it, you
> can just look up what qdev created.  See here:
> 
> http://www.kraxel.org/cgit/qemu/commit/?h=rebase/usb-next&id=70b9867011c4793787c5acee3d2005a6bc951f59

yes, much better :)

> 
> [ This is part of the "usb patch queue" patch series posted today,
>   depending on how the qom discussions go and how fast it goes in
>   you might just call the function the patch provides.  Or do
>   something simliar in pc_q35.c and I'll drop the patch. ]
> 
> -usb for -M pc creates a "usb.0" bus too, so I don't expect trouble.
> 

I think your patch, is a generally useful helper function. Thus, I plan to
incorporate something similar to your patch, but less general in pc_q35.c. So
usb can get testing, and when your patch lands I will drop the extra usb
bits in pc_q35.c.

Thanks,

-Jason