interesting. Not sure what is up here but I have seen that many VGA bioses copy themselves to RAM and then modify their own code. Actually, I have seen many BIOSes that do this self modifying code hack; they use it to change behavior for a warm restart; copy themselves to shadow RAM and then change a byte or two so the jump target is different.. I've no idea if that's what is going on here but thought I'd mention it.
ron
On Tue, Jun 2, 2015 at 4:17 PM Vladimir quickcracktime@gmail.com wrote:
I retrieved my vgabios via Linux kernel 3.19 (ubuntu 15.04) using this method - http://www.coreboot.org/VGA_support#Retrieval_via_Linux_kernel ; then I downloaded latest sources of SeaBIOS & Coreboot and built Coreboot using compiled SeaBIOS payload as well as vgabios file ( 1002,990b )
After coreboot installation I have decided to retrieve vgabios again, using exactly the same kernel and OS, and then diff'ed two vgabios files out of curiosity. To my surprise, they are slightly different! vgabios_before.bin - SHA1 checksum: e4d320eb278b0118c46e2e470e7154b12c41966d vgabios__after.bin - SHA1 checksum: a9e2ed569bbaaea283b5380a5f6c44fc4efc3da4 Here is a report about 3 bytes difference between them - http://www.diffnow.com/?report=2kwq3 (wait a few seconds while it loads)
Then I teardown a laptop, and check if vgabios inside coreboot's image (flashed in chip) is 3 bytes different as well, but there was no difference against the original vgabios.
So, it appears that, while loading vgabios from a flash chip, coreboot modifies it slightly. Although this tiny difference is not causing any graphical glitches or problems for me, this could be a result of a bug - which is not necessarily limited to my hardware ( Lenovo G505s ) , and maybe could lead to some other problems
Please tell your opinion, is it a bug or I am wrong at something?
Best regards, Vladimir Shipovalov -- coreboot mailing list: coreboot@coreboot.org http://www.coreboot.org/mailman/listinfo/coreboot
-
ron minnich