> Hi,
>
> > > vga->con = graphic_console_init(DEVICE(dev), 0, &qxl_ops, qxl);
> > > + qxl->id = qemu_console_get_index(vga->con); /* == channel_id */
> > > + if (qxl->id != 0) {
> > > + error_setg(errp, "primary qxl-vga device must be console 0 "
> > > + "(first display device on the command line)");
> > > + return;
> > > + }
> >
> > In the comment this seems no more required so why testing it?
>
> I'd expect it'll confuse guests.
>
> If you are sure this is not the case feel free to send a patch dropping
> this check.
>
Had a look more deep into Qemu code. Both qxl_realize_primary and
qxl_realize_secondary calls qxl_realize_common which calls
qemu_spice_add_display_interface to register the interface to spice.
Both qxl_realize_primary and qxl_realize_secondary set qxl->id
(first with 0 second from 1 as code below was doing).
There are 2 fields "id": PCIQXLDevice->id and PCIQXLDevice->ssd.qxl.id,
having PCIQXLDevice->ssd.qxl.id == qemu_console_get_index(PCIQXLDevice->vga->con)
== channel_id (PCIQXLDevice->ssd.qxl.id is set in qemu_spice_add_display_interface).
Looks like PCIQXLDevice->id, beside to distinguish if primary or not is
only used for tracing. Would not be better to remove PCIQXLDevice->id
at all after adding the have_vga flag?
> > > static void qxl_realize_secondary(PCIDevice *dev, Error **errp)
> > > {
> > > - static int device_id = 1;
> > > PCIQXLDevice *qxl = PCI_QXL(dev);
> > >
> > > - qxl->id = device_id++;
> > > qxl_init_ramsize(qxl);
> > > memory_region_init_ram(&qxl->vga.vram, OBJECT(dev), "qxl.vgavram",
> > > qxl->vga.vram_size, &error_fatal);
> > > qxl->vga.vram_ptr = memory_region_get_ram_ptr(&qxl->vga.vram);
> > > qxl->vga.con = graphic_console_init(DEVICE(dev), 0, &qxl_ops, qxl);
> > > + qxl->id = qemu_console_get_index(qxl->vga.con); /* == channel_id */
> > >
> >
> > As these IDs must be contiguous this means that there must be the
> > requirement
> > that if there is a qxl interface only qxl interfaces are used and no other
> > console which seems to me wrong.
>
> Hmm. Didn't know this is a requirement. But as far I know qxl->id is
> also expected to be the channel_id. So with qxl and non-qxl mixed we
> are in trouble no matter what:
>
> * with the patch we break the contiguous id requirement.
PCIQXLDevice->id? No reason actually to be contiguous.
PCIQXLDevice->ssd.qxl.id is already qemu_console_get_index so potentially
not contiguous.
> * without the patch we break the qxl->id == channel_id requirement.
Yes, can be broken although I suspect for other reasons this is true.
>
> So, what now?
>
> cheers,
> Gerd
>
I would personally add the "have_vga" field and use a single id field,
this will make sure to have the same consistent value.
Frediano
_______________________________________________
Spice-devel mailing list
Spice-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/spice-devel