Hello all,
the cryptic subject of this message is due to my efforts to verify that Flashrom is producing correct images.
After much practical testing I have been able to conclude positively that the Award Bios 4.51 of a FIC mainboard is rewriting the blocks
0x0003800 - 0x00038fff cleartext on system characteristica, 0x0003a00 - 0x0003afff pci-table.
at each reboot. Now I wonder if also Coreboot is using this opportunity under some circumstances. It is not that I think it is a commendable thing to do, instead it is a rather hazardous endeavour in my eyes.
Best regards,
Mats Erik Andersson
we don't rewrite flash. Yes, this is hazardous I think as well.
ron
On 01.10.2008 19:40, ron minnich wrote:
we don't rewrite flash. Yes, this is hazardous I think as well.
We may do that soon, but I promise it will be done in the most careful way possible. That means regardless of when the write to a flash chip fails, you can still boot and still have your configuration data. Why? NVRAM is not really unlimited storage. Expect a proposal and skeleton code in the next few days.
Regards, Carl-Daniel
on 01/10/2008 19:04 Mats Erik Andersson said the following:
Hello all,
the cryptic subject of this message is due to my efforts to verify that Flashrom is producing correct images.
After much practical testing I have been able to conclude positively that the Award Bios 4.51 of a FIC mainboard is rewriting the blocks
0x0003800 - 0x00038fff cleartext on system characteristica, 0x0003a00 - 0x0003afff pci-table.
I think that this is related to ESCD updates. It seems that some proprietary BIOSes contain flash writing routines which sometimes are even exposed for external use (via special signatures).
at each reboot. Now I wonder if also Coreboot is using this opportunity under some circumstances. It is not that I think it is a commendable thing to do, instead it is a rather hazardous endeavour in my eyes.
Best regards,
Mats Erik Andersson
-- coreboot mailing list: coreboot@coreboot.org http://www.coreboot.org/mailman/listinfo/coreboot