[Xen-ia64-devel] RE: [PATCH] fixed serial console support

[Alex Williamson <alex.williamson@xxxxxx>]

On Wed, 2005-11-30 at 13:52 -0800, Magenheimer, Dan (HP Labs Fort
Collins) wrote:
> Is there one place (or a small number of places) where
> some Xen-specific code could be added to Linux to
> disable the 8250 (using "if (running_on_xen)" without
> CONFIG_SERIAL_8250=n?   It would be nice to retain
> transparent paravirtualization if possible.

   It really seems like we need some way to hide individual devices.  I
would only feel comfortable hiding a UART using "running_on_xen" for
guest domain if we could do it such that only the console UART is
hidden.  Are there any hypervisor callbacks we can use to tell a
privileged domain the physical address of the console UART?  If so we
might be able to exit out of registering the serial device fairly
easily.  Maybe it could be done via an "ignore_uart=" boot parameter
passed into dom0(?)  Otherwise we'd need to trap the ioremap in the
hypervisor or mess with SAL PCI calls for PCI UARTs and ACPI namespace
for non-PCI.  I'm sure all of these have fairly serious problems and
implications.

>  Also,
> will this eliminate the need to have CONFIG_VT=y
> on dom0 and CONFIG_VT=n on domU?

   I don't know, I never really understood the problem there.  Xen
provides a ttyS (serial) console device to dom0  and a tty (vt) device
to non-dom0 domains.  I would therefore have expected domU to need
CONFIG_VT=y.  What was the problem with CONFIG_VT on domU before?
Thanks,

        Alex

-- 
Alex Williamson                             HP Linux & Open Source Lab


_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel