Levi Khatskevitch wrote:
>
> Hi,
>
> A strange problem I encountered when assembling one of our brand new
> "Cachier" boxes (MS7308e M/B + DoC Mil.). The new box took a half second
> longer to upack the kernel image and then poweroff() didn't make system
> shutdown. I tried the original Ollie's code from acpi_shutdown.c - no effect.
> Then while figuring out how different the box was from our three working
> prototypes I inserted the original BIOS chip, booted up with and it then
> replaced it back with the DoC and the probem was fixed. After I shorted out
> the CMOS memory situation repeated.
>
> I couldn't imagine that after removing the original BIOS chip there is any
> code on the M/B that uses or relies on CMOS, this case proved me wrong. Now
> I'm worried that our users will encounter the same problem when the CMOS
> battery dies. If that happens then pressing the power button will cause the
> system to halt (my module) but the green power light will be still on making
> it look like the system is still running.
>
> I looks like original BIOS makes settings that are then stored in CMOS that
> LinuxBIOS doesn't. And some part of system, what I can't image and it's not
> LinuxBIOS, that relies on those CMOS setting. I'm very intrested in find out
> the answer and I'll appriciate if anyone tries to reproduce and/or fix the
> problem.
>
Yes, we have found the same problem here. The APC register which is in
the
CMOS RAM controls some "misc" stuff of the chipset. One is the LAN stuff
which we suspect that cause Ron's problem. And the delay in boot time is
another problem.
The southbridge.c in the CVS has addressed the LAN part. I am working on
other issues.
Ollie
