HP DL145 Opteron G3/G4 MicroCode Updates

[oXYnary]

I hex compared and for the most part yours still seems to have more and has some matches. Going to try it tonight. Though I'm starting to wonder. Can AGESA not even be included on some motherboards? I see no mention in the bios of it anywhere in the original while all the other Phoenix I have found have it in the cpu update section. Could it be alternatively placed in the ACPI section hidden?

In short is AGESA in some form even required period to work with AMD processors? If not I assume maybe what might be happening is when I replaced the microcode and AGESA from another motherboard into the update00 section. The AGESA was ignored because there was no calls to it from somewhere else?

[TheWiz]

I am still looking at this so I would hold off on any future testing. AGESA most certainly is a requirnment for AMD CPUs K8 and up, which are most modern boards these days anyways. Once I disassemble your ACPI and try to isolate a couple things, I'll try to give you a more descriptive answer.

bear with me!
TheWiz

[oXYnary]

Thanks Wiz.

I did try 1234's micro code and it had no effect.

[1234s282]

Thanks Wiz.

I did try 1234's micro code and it had no effect.

Sorry it didn't work , thanks for the feedback as i can remove the mod to prevent others from flashing it by mistake

Admin

[oXYnary]

So I got ahold of the correct 2376 meant for 2 way systems only. Still wont boot. Had to try though.

Here is the thing. If AGEAS is needed no matter the manufacturer. Then where is it? In all other phoenix bios's I saw it was in the update00 section. But using 1234's microcode only bios that he replaced the default update00 section entirely with only microcode updates should have made the system brick. It did not however and it still sensed/senses the default cpu.

Which again creates a quandry.
When did AGEAS come into play? The only other thing I can think of is when first building the bios for this system AGEAS was not out yet (mid 96) so they never included it in any of the updates because of trying to integrate it would be too much hassle. That would also explain why they never came out with any updates for the cpus it could take.

That or AGEAS is somewhere else in this bios. But searching for the term AGEAS brings up nothing in the hexeditor. Means nothing I realize as they could have stripped the header.

[oXYnary]

Well I did some more digging. First an overall note for you guys.

Softpaq? Is just compressed bios. You can still uncompress it with

Code:

softpaq.exe -u filename newfilename

So if people say they have softpaq bios there is still a chance you can mod it.

Now new zip with newer files and updates per below.
http://207.246.157.76/HPDL145G3ProjectV2.zip




So with that info. I was able to uncompress the HP DL165 G5/G6 Bios which is the same chipset. It supports up to 6 core processors. Unfortunately it is an AMI bios.

But here is info for you from me testing around.

A: AMI AGESA is in the 1b section in three parts.
-RUN_CSEG: Think thats the core code.
-AGESA_CSEG: Think thats the CPU naming code for Bios front end
-AGESA_DSEG: ?

**(These are extracted under the 1b folder in the Dl165G5_6 Bios folder).

B: There's nothing that comes close to comparing to any of these files in the HP DL145 Bios. Nor no mention of even text saying Agesa anywhere.

C: The other Phoenix Bios (Tyan 4985) clearly has an Agesa section in it under update00 (microcode section). But that section only matches "Run_Cseg". Agesa Cseg and Agesa_dseg from ami do not seemingly exist?

D: Simply replacing the entire Microcode section of original bios with another Phoenix Bios with Agesa in that section (Tyan 4985) does not "initiate" Agesa. (See AG3_5_2000.ROM under Modded Bios _ It flashed but made no change to processor detection).


Conclusions:
Agesa needs to be initiated somewhere else in the bios to then tell it to look in the microcode in Phoenix. I have no idea where to look thus far. My theory at this point is the DL145g3 DOES NOT have that insertion and was "hardlocked" to a specific series of processors.

Option 1: Ignore ACPI and insert all correct AGESA code where needed in DL145 g3Phoenix Bios. Agesa is not initializing/present. Messing with ACPI would be pointless.

Problems:
1.Figuring out where AGESA is called into play to insert.
2. Figuring out where custom code is thats hardwired to particular cpus.


Option 2. Taking the DL165 Bios and going the other direction and sticking the 145 items onto it. Both Boards share the same chipset and a somewhat similiar layout. Deleting the option roms not present like raid and HP NIC. Then replacing with options roms from 145 like Broadcom Ethernet.

Problems
1: Not same byte size. AMI: 1024. Phoenix: 1027
2: How to flash unto chip?
3: Will currently installed flash chip work with the AMI?
4: PCI irqs and such match enough to not crash or brick system?
5: Will it disable the Server Manager firmware? Not important to me but very possibly important to others.
6: Will require reinstall on some systems and people using Windows Server may be told system illegal since the bios will not match old type.


Thoughts?

[TheWiz]

-AGSA-DSEG I believe is a description segment ie. All the string names of proccessors. In regards to your options yes your flash chip can be flashed VIA AMI, posting is another issue. If it did work, chances are the OS may still function normally as I think it pulls the mobo S/N and not the BIOS revision, otherwise anytime you flashed a new BIOS you'd run into some issues. First off though, can't find the ACPITBLs for these bioses in raw extracted forum, can you add them to the zip for me?

Thanks,
TheWiz

[oXYnary]

-AGSA-DSEG I believe is a description segment ie. All the string names of proccessors. In regards to your options yes your flash chip can be flashed VIA AMI, posting is another issue. If it did work, chances are the OS may still function normally as I think it pulls the mobo S/N and not the BIOS revision, otherwise anytime you flashed a new BIOS you'd run into some issues. First off though, can't find the ACPITBLs for these bioses in raw extracted forum, can you add them to the zip for me?

Thanks,
TheWiz

For the AMI ones? They should be in there already.

Look under:
-AGESA_Micro_ACPI_Tyan-3992-E. That has the acpi as well.

-I also did it for the HP DL165 g5 under.. Original_Bios\HPDL165G5_6\1b extract . That has all the different 1b modules extracted.

I did notice I didnt have the phoenix from the original 145 bios. Here is a separate file for that (will add into main portion on next upload):
http://207.246.157.76/145g3_ACPI.zip

What Im saying though is I don't think ACPI at least alone or if at all is the issue as the original HP bios has no reference at all to AGESA

[TheWiz]

Yeah I know but I thought it would be wise to compile your ACPI tables in C and then compare them so we can at least get a diff report on the similarities of the boards were looking at. It is very well possible the CPU support is written into the main bios module as opposed to an exterior one.

[oXYnary]

Yeah I know but I thought it would be wise to compile your ACPI tables in C and then compare them so we can at least get a diff report on the similarities of the boards were looking at. It is very well possible the CPU support is written into the main bios module as opposed to an exterior one.

Just read up the acpi aml file is NOT in the 1b section of the AMI bios. Unfortunately Im up a creek without a paddle looking for the acpi aml section in the AMI at this point. However here is the dl145 aml converted for you into an asl file. That might at least help with figuring out if indeed the cpus are hardlocked. http://207.246.157.76/Dl145G3_AmL.zip

Help?