We bought original Supermicro motherboard
H8QGI-F. Installed on it 4 AMD Opteron processors
family15h (model OS6234WKTCGGU - 12 cores). And then properly installed 2
dimms per socket (one per node).
We configured coreboot for this board. But it
doesn't work. Memtest passes ok, but system failes in AmdCpuPost (log in
attachment).
It seems like system reboots in function
"Relinquish control of APs". Is it unplanned reset? After it system find only
one node and then failed and stuck.
In this "Relinquish control of APs" we do
halt function on every AP in "for" cycle. I input debug message in this cycle
and figured out, that reset happens after second halt function. I tried to
reverse cycle, but again system reset happend on second halt function. Maybe
this little info can help to determine the problem.
If i comment this function, there will be no reboot
and code passes far far away, but according to debug log in this case
system work on only one core.
I tried to use current coreboot version and version
when Kerry Sheh initially commited code for this board like working and
tested (http://review.coreboot.org/#/c/567/).
They both have same behaviour for me.
I need some help to figured out the
problem
Best regards,
Aladyshev Konstantin