[COMMIT] Infrastructure Updates 4

Eric W. Biederman ebiederman at lnxi.com
Mon Sep 8 16:33:00 CEST 2003


YhLu <YhLu at tyan.com> writes:

> Eric,
> 
> In the k8/cpufixup.c, you have use 0x3f0000 for TOM ( you said leave 64M for
> ROM and Device IO). In case that if the device need more MMIO range, it will
> produce problems for example AGP. So You need compare mmio_basek with
> 0x3f0000 to see if mmio_basek is smaller than 0x3f0000. ( 0xf0000000 in
> bytes.)

Hmm.  The 64M hole was an initial hard code in raminit.c that was just
a small arbitrary hole so we could access I/O devices until everything is
setup properly.

mmio_basek should be the lowest address at which we have devices.  So cpufixup.c
should be doing the right thing.

Although I have a few pieces to update.

> I have ever tried to move that hard_reset to mainboard.c, and it still does
> not work for s2885. I will try it on s2880 again.

Ok.  It really depends on how your bus is enumerated.

Eric



More information about the coreboot mailing list