[LinuxBIOS] LinuxBios on Epia-ML

Christian Sühs chris at suehsi.de
Sun Dec 18 19:54:56 CET 2005


Richard Smith schrieb:
>>>>INSTALL REAL-MODE IDT
>>>>DO THE VGA BIOS
>>>>found VGA: vid=1106, did=3122
>>>>rom base, size: fffc0000
>>>>BAD SIGNATURE 0x8 0x1e
> 
> 
> I missed this earlier.  This indicates that the loader do_vgabios()
> was unable to find the bios extension signature of 0xAA55 using the
> base offset of 0xfffc0000.
> Your video bios was _NOT_ copied to ram.
> 
> 
>>>biosint: # 0x10, eax 0x4f14 ebx 0x18003 ecx 0x1 edx 0x0
>>>biosint: ebp 0x17fa8 esp 0xffa edi 0x0 esi 0xffff70b6
>>>biosint: ip 0x8e22 cs 0x0 flags 0x46
>>>BIOSINT: Unsupport int #0x10
>>
> 
> This is probally the result of vga_enable_console() which is called
> regardless of what happens in do_vgabios() so I think its just running
> bogus code.
> 
> Your rom image does not have the vga bios located correctly.  You must
> locate the video bios in your ROM image such that it will get
> referenced at 0xfffc0000.
> 

Well, it could be that the dd comand to get the video.bios from original 
bios has had the false values for skip. The epia-ml vgabios is smaller 
than the epia-m vgabios and eventually located on another place in memory

However, I have tested the epia-m vgabios extracted with award bios 
editor from the older bios version, and the vga console comes up, if the 
kernel is loaded :D
Unfortunatly, there is no fb0 and X fails.

Doesn't matter, I have also extracted the original vga bios from bios 
and will try it again.

> For a ROM image that is 256k in size that means you need a 64k VGA
> bios image located at offset 0x0.  The linuxbios image can then start
> at offset 0x10000.
> 
> If you do a 'hd' on your ROM image file the very first word you see
> should be 0x55aa which is the 0xaa55 rom signature in little endian.
> 
I will have a look ;)

> Since the build system should have done all this for you I'm guessing
> that your vga bios image is not valid.
>
Jip, i totally agree. The first vgabios i get with dd is shit :D

> --
> Richard A. Smith





More information about the coreboot mailing list