Subject: Re: CVS commit: src/sys/arch/powerpc
To: None <shige@NetBSD.org>
From: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
List: source-changes
Date: 01/23/2005 06:17:55
In article <20050123.060225.41632322.shige@FreeBSD.org@FreeBSD.org>
shige@NetBSD.org wrote:

> In 405GP/GPr manual, there is the following statement:
>    The two PPC405GP/GPr UARTs (serial ports) can be clocked individually,
>    either from an extrnal serial clock or from an internally generated
>    serial clock. The internally generated clock is derived from the CPU clock,
>    and is CPU/n, where n rangesfrom 1 to 32. ...
> 
> OpenBlockS266 and walnut board is the former.
> But there may be a machine in the latter case.
> In this case, consinit should be in powerpc/ibm4xx/consinit.c.

No. It isn't consinit(), but com_obp_cnattach() which should
be in powerpc/ibm4xx/dev/com_opb.c.
consinit() on some machines with ibm4xx could have VGA or
other console devices other than internal com ports,
so it can't be in MI place.

If both on-chip com ports could have different frequency,
com_opb_cnattach() should have some arg to pass the frequency
used on the console port, and MD consinit() should call it
with proper settings.
---
Izumi Tsutsui
tsutsui@ceres.dti.ne.jp