Port-ofppc archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: map_isa_ioregs



Frank Wille wrote:
Tim Rightnour wrote:

On 08.11.07 02:19:25 you wrote:

Are you using a physical console?  It's possible I broke something in that
realm.  I do all my work via serial console.

Not even serial console works here. But I debugged the problem today:

First, when working with physical console, checking for device_type property
"display" fails on the Pegasos. It is called "bootconsole". Additionally,
when the pckbc driver is missing, cninit_kd() would just return, instead of
falling back to the ofwbootcons.

Second, when using a serial terminal, the check for the name-property
"serial" fails, which is "failsafe" here. IMHO this check is not needed at
all, because it is a fallback to the ofwbootcons.

Yeah you should never, ever switch on the "name" property, it's absolutely
99.9% of the time purely informational.

As for the "display" node, if you have a VGA graphics card in there, but
it's in text mode on boot (that means, everyone with a public firmware,
and everyone on Efika who didn't setenv fb-mode) it usually presents itself
as device_type "serial" - it isn't serial, but it *is* an ANSI terminal
emulator so ANSI terminal code for a serial port (as long as you're
outputting using firmware read/write methods) works.

In framebuffer mode, it really does read as display. You can confirm that
the display is a graphics card by the simple fact that it includes a bunch
of PCI properties which according to the PCI standard (class-code etc.)
will remind you that this is a PCI card of a certain programming
interface..

I'm curious about the "bootconsole" thing though. I've never seen that?

--
Matt Sealey <matt%genesi-usa.com@localhost>
Genesi, Manager, Developer Relations



Home | Main Index | Thread Index | Old Index