Re: [virtio-comment] Feedback for chapter 4

[Thomas Huth <thuth@linux.vnet.ibm.com>]

On Mon, 20 Jan 2014 14:22:58 +1030
Rusty Russell <rusty@au1.ibm.com> wrote:

> Thomas Huth <thuth@linux.vnet.ibm.com> writes:
> > Here's my feedback for Virtio draft 01, chapter 4:
> 
> Thanks again for the feedback!
> 
> >
> > Page 20 /  PCI Device Layout:
> >
> > - "To configure the device, use I/O and/or memory regions and/or PCI configuration
> >   space of the PCI device."
> >   => That sounds a little bit sparse/confusing. Maybe rather something like:
> >   "To configure the device, it is possible to use the PCI configuration space
> >   and/or to access the configuration data via an I/O and/or MMIO base-address
> >   register."
> 
> I ended up with this:
>         The device is configured via I/O and/or memory regions (though see
>         VIRTIO_PCI_CAP_PCI_CFG for access via the PCI configutation space).

s/configutation/configuration/

[...] 
> > Page 27 / MMIO Device Discovery:
> >
> > - The device tree snippet is obviously an example. That's ok, but I think the
> >   spec should explicitely say so (and maybe add some generic words about the
> >   required properties before the example, too).
> 
> OK, I clarified this:
> 
> Unlike PCI, MMIO provides no generic device discovery.  For each
> device, the guest OS will need to know the location of the registers
> and interrupt(s) used.  The suggested binding for Open Firmware is
> shown in this example:

Technically, this is not an Open Firmware example, because in that case
you likely would have to write Forth code instead :-)
The syntax that is used here is from the ePAPR specification, so I'd
rather talk about an "ePAPR-style device tree" or a "flattened device tree"
example here instead.

>         // EXAMPLE: virtio_block device taking 256 bytes at 0x1e000, interrupt 42.
> 	virtio_block@1e000 {
> 		compatible = "virtio,mmio";
> 		reg = <0x1e000 0x100>;
> 		interrupts = <42>;
> 	}

 Thomas