It has taken me a while to find ROM pack where nothing is changed within data/command tables. I lack the expertise to crack this so I present it hoping you or another may.
If you compare the tables list you will see ROMs are nearly identical except for fewer elements then any others I have.Collated info on Signature (Click to show)
So we know from this info
by The Stilt:-
The signature is actually a hash of the hashes of the protected blocks.
Then in the PM I sent we have Lordkag info
It is most likely done on the Legacy ROM, the only thing left to determine is its boundaries and how it transforms from SHA-1 (160 bits) to Some_Hash (128 bits).
(+ note at end of post).
If the 8-bit checksum and CRC can change without affecting the hash, it would imply that the hash must exclude them from calculation.
Next you can see some of my testing for boundaries in this post
113-C88100-107 date change post
113-C8800100-107 code revision change not post
113-C8800100-107 padding area between legacy & efi change post
Above tests are "pure UEFI" mode, any modified ROM regardless of mod, mobo will only post with CSM=ON, SB=OFF.
As on Hawaii/Fiji we have no issue with driver detecting signature incorrect for modified ROM my testing has been for UEFI purposes (ie "pure UEFI mode on mobo).
Testing was done on all the same hardware/system only ROM on Fury X was being changed. The test was mobo must post with display output and UEFI usable whilst CSM=Off, SB=ON and FB=ON, SSD/HDD was disconnected for this testing.
Now the Sapphire Tri-X 290 OC ROM pack:-
i) legacy ROM checksum differs but is not part of signature make up from my Fiji testing, see note.
ii) date change I know is irrelevant in context of signature as tested this. Bios P/N and the 2 other text differences between ROMs I will check.
iii) as the hex value which enables UEFI in a ROM is after ROM code revision from my test info in spoiler I would deem editing there is protected by signature, but will test this.
iv) next BIOS_IDTF changes, see note further on.
v) then from 0x269 we have the bios signature change and the rest of the ROM hex for hex is identical until UEFI/GOP module. Which from Lordkag info and my testing of changing padding between legacy and uefi sections is outside of boundaries it covers.
Tri-X_290_OC_ROMs.zip 147k .zip file
Note:- BIOS_IDTF I have been noting on Fiji when I flash an updated stock ROM from Sapphire/AMD and compare dump it will change during flashing process and legacy checksum is also updated to reflect this change.
*** edit ***
point ii) above BIOS P/N I'm deeming is protected as changing that = no post.Edited by gupsterg - 7/8/16 at 11:27am