hybrid-graphics-linux team mailing list archive

[Matthias Fulz <olz1983@xxxxxxxxxxxxxx>]

>>
>> Well, don't know why it is considered vga bios is 64k, memory map shows
>> it is 32k in fact, further are ata rom and other devices, so we can skip
>> everything behind 32k - which gives 2 bytes diff, at exactly the same
>> locations as for me - 0x22 & 0x1CC(1C8 in your case, but looks as the
>> same header still)
>>
> 
> Thanks for the information.
> 
> Later on, I'll try to overwrite these bytes you mentioned with the bytes
> from the discrete bios.
> 
> Then I'll check what fglrx is doint, perhaps you're right with the email
> before and fglrx is just validating a checksum of the bios, then this
> method should allready work.

I'm to stupid to find the location of the ATI Bios it seems to change
from boot to boot....

Even I don't understand what the function "acpi_evaluate_object" in
"radeon_atpx_handler.c" is doing so I don't know how / where the Bios is
recived.

Any help is appreciated :)


Also I looked a bit deeper in my DSDT and found following:

                DWordMemory (ResourceProducer, PosDecode, MinFixed, MaxFix
      ed, Cacheable, ReadWrite,
 3801                     0x00000000,         // Granularity
 3802                     0x000C0000,         // Range Minimum
 3803                     0x000C3FFF,         // Range Maximum
 3804                     0x00000000,         // Translation Offset
 3805                     0x00004000,         // Length
 3806                     ,, , AddressRangeMemory, TypeStatic)
 3807

3801 - 3807 are line numbers.

So for me it looks like "TypeStatic" is doing the shadowing and the RO
to the Address we want to write. (It's just a shot in the dark).

Anyone knows if there are other "Types" perhaps IO or something?
I would give it a try if anybody knows keywords which would work.

Best regards

Matthias