On Sat, Apr 24, 2010 at 08:26:45PM +0200, Patrick Georgi wrote:
Am 24.04.2010 19:43, schrieb xdrudis:
They might just use a watchdog:
- BIOS 1 sets a flag
- BIOS 1 configures the watchdog to trigger when it's not touched within
2 seconds (or whatever). watchdog would reboot the system then
- BIOS 1 jumps in BIOS 2
- BIOS 2 does whatever it needs to do to consider itself "safe"
- Meanwhile, BIOS 2 touches the watchdog every so often
- BIOS 2 deactivates the watchdog
In this scenario, coreboot would have to know how to tell the watchdog to reset its countdown, and how to disable the watchdog, to safely use the Dual BIOS feature.
Ok. I'm rereading the link Gigabyte gave me, which does not explain enough or I don't understand it enough, but it might be this scenario you explain
http://www.gigabyte.com.tw/FileList/NewTech/2006_motherboard_newtech/article...
(the URL says 2006 but it was given to me in a mail in early March 2010)
I've noticed they say it reboots before running the other BIOS, it's not just a jump. How would that work ? would it be some flag in CMOS ? This is better, I guess in that it gives both BIOSes the same initial state.
It also says the original BIOS checks both BIOS copies, but I guess it doesn't matter since it will only run if coreboot fails, and then you have to reflash it anyway.
The feature supposedly shouldn't just guard against non-Gigabyte images, but against issues with their own images, too - and those would have the right signature, and thus would pass any such test.
I'd be really amazed if they'd add another chip (that actually costs money) and then only implement an incomplete protection scheme with it.
Ok. It makes sense. Thank you for explaining.