Re: [PATCH RESEND] virtio_console: read size from config space during device init

[Filip Hejsek <filip.hejsek@gmail.com>]

On Thu, 2026-06-11 at 03:38 -0400, Michael S. Tsirkin wrote:
> [...]
> > > 
> > > Wait a second. Why is there this rproc test here?
> > > Was not in the original code and commit log says nothing about it.
> > > 
> > 
> > Previously, this code was in config_work_handler(), which was never
> > called for rproc_serial (it's scheduled from config_intr(), which is
> > the config_changed handler only for virtio_console).
> > 
> > Now update_size_from_config() is called unconditionally from
> > virtcons_probe(), so it will be called for rproc_serial too, which
> > doesn't have the F_SIZE feature.
> 
> So why not test it? 

The virtio_console driver implements two similar but distinct virtio
devices: VIRTIO_ID_CONSOLE and VIRTIO_ID_RPROC_SERIAL. Although some of
the implementation code is shared, the devices are different. In
particular, rproc_serial doesn't support multiport nor any of the tty
specific features. This means that the relevant feature bits are not
valid for this device and must not be tested.

I have to admit though that I don't quite understand what the
RPROC_SERIAL device is supposed to be used for. It was added by commit
1b6370463e88b0c1c317de16d7b962acc1dab4f2, which describes it as "a
simple serial connection driver called VIRTIO_ID_RPROC_SERIAL (11) for
communicating with a remote processor in an asymmetric multi-processing
configuration". It seems that it was never standardized, as the virtio
spec only says that its ID is reserved.

> What does "not a valid feature" mean?

I copied the "not a valid feature" comment form other instances in the
same file where a feature is tested, e.g. in resize_console():

	/* Don't test F_SIZE at all if we're rproc: not a valid feature! */
	if (!is_rproc_serial(vdev) &&
	    virtio_has_feature(vdev, VIRTIO_CONSOLE_F_SIZE))
	  hvc_resize(port->cons.hvc, port->cons.ws);


Best regards,
Filip Hejsek
> >