ron minnich wrote:
I'm a little concerned that we broke the lar zeroing code again. that would explain the bogus ACPI bits.
we're breaking things too often. I think we may need to stop committing until we get a test. So we may need a signoff, "Tested-by", and no commits until that is seen.
We don't even need to do manual testing. We have a fully automated test system for that:
http://www.coreboot.org/Distributed_and_Automated_Testsystem
v3 is not supported yet, but it should be simpler than v2 was.
You can browse the code here: http://coreboot.org/viewvc/?root=Testsystem
All we need is someone with the hardware, possibly an X10 power switch, a bios savior and another machine with connection to the internet to fetch and execute the tests. More information on the requirements are in the papers on above site.