PSA: Don't update your Ryzen Mainboards to AGESA 1.0.7.2 yet

[Azulath]

Thanks for the warning. However, I prefer to wait on new BIOSes anyway
That being said, a new BIOS isn’t available for my Taichi yet.

[catsay]

I’m currently poking at the two latest Firmware files for my ASRock X370 Gaming K4 board.
The major thing I’ve noticed is that a volume image responsible for CSM (Legacy Boot & Devices on UEFI) has been removed in BIOS version 4.10 versus 3.40

On the right we have Firmware 3.4 and on the Left 4.10

Notice the 5C3DB34C-75BE-4641-B064-C6247ABE1FF6 File that’s present in 3.40 but not in 4.10

It contains a few DXE (Driver Execution Environment) Files.

Let’s see if they’ve been moved to elsewhere in the file structure.

EDIT:

Ok so the CsmDxe File has been merged into a different part of the volume image, same GUID. It’s also smaller.
It appears that LegacyInterrupt, LegacyRegion and CsmVideo have been removed or likely rolled into one of the other modules. I’ll have a more complete analysis once I’ve redesigned my tools to handle this AMD Firmware better.

Ps Lots of interesting PSP related code

[Azulath]

That’s why I love this form, normal people don’t do this. (I mean this as a compliment )

Edit: Apparently ASRock has removed the BIOS from it’s page.

@catsay Shouldn’t this be AGESA 1.0.0.7, or did I get something wrong?

[catsay]

Technically yes It should be Agesa 1.0.0.7. But that got vetod since this BIOS is a major release, rather than a minor increment as the previous ones where.

Hence why ASRock is also using v4.10 instead of v3.50 on my mainboard as the version designator.

[Azulath]

Thanks for clarifying that,

[catsay]

I’m sort of having to relearn how UEFI and PEIM / PPI interfaces all work.
Thankfully though all the Driver files inside UEFI are PE (Portable Executable) Format. Or Terse Executable format. Same as Microsoft uses.

Portable Executable

This slide and presentation where really helpful in me getting started.

eu-14-Kovah-Analyzing-UEFI-BIOSes-From-Attacker-And-Defender-Viewpoints.pdf

Only difference was adapting the tools to work around the AMD Differences compared to Intel.

For example the Management Engine block in the Firmware structure isn’t then used for the equivalent ARM PSP code. It’s laid out somewhat differently and the Headers for some blocks of firmware aren’t known yet so I’m treating them as data padding.

If I can finally get a proper SPI Flash Rom clipper I might try and read the Firmware directly out of the flash rom on a mainboard.

It seems the flashing process decodes part of the firmware before writing it to the chip.

[campdude]

The AGESA 1.0.7.2 helped my system reach spd and rated ram speed. I read about it being warned about bugs but… if you are not getting the spd to work properly prior… you will try out the new bios anyhow.