Flip the HAP bit

Zitat von ziddey im Beitrag #16

---

Flip the HAP bit

---

https://github.com/corna/me_cleaner/wiki...ltMeDisable-bit

Is ME chipset-specific? What if I just downgrade it with a different program?

Looks like the project is a fail.

First, used CoffeeTime to add microcode 906E9 which I think is the correct one for my Pentium G4620.
Won't boot. When pressing power button, the power LED flashes a split second, CPU and PSU fans come on for about 3-seconds and turns off.

2nd attempt, disable HAP bit. Had to use the me_cleaner in pull request #282 to successfully disable it on ME12. https://github.com/corna/me_cleaner/pull/282
Booted with my Core i3 9100. Confirmed in BIOS, ME FW shows 0.0.0.0, so successfully disabled. Boots into Windows just fine.

Swapped to G4620. Same problem as before. Power LED flashes a split second, CPU and PSU fans come on for about 3-seconds and turns off.

Any ideas?

Did some more reading. Looks like my symptoms closely resemble that of an incompatible ME.

Hypothesis: One website about me_cleaner describes, "a mechanism that can disable Intel ME after hardware is initialized and the main processor starts". Maybe the soft-disable is happening too late to prevent ME from detecting incompatible hardware and aborting POST. Perhaps downgrading or removing ME is the solution.

Tried UEFITool alpha 58 (newest?), it can't do anything with this board's BIOS. No replace or save options. This board's "ME region" uses something called IFWI, I guess as opposed to FPT of ME11. I'm guessing that's why UEFITool can't modify it.

So right now, I'm at a dead end.

Another update on downgrading a motherboard to work with SKL and KBL if anyone is interested.

Picked up a Gigabyte H310M-A 2.0 <-- 2.0 is key, because it's the 22nm version of H310M that uses ME11.
Used CoffeeTime to add KBL microcode. Flashed using a modified efiflash.exe. Didn't work due to ME11.8 CON apparently not liking KBL.
Used CoffeeTime to downgrade to ME11.7 CON.
Booted up first time, it's working happily with my $20 G4620. So far, no problems. So it looks like 22nm 3xx chipsets with the right ME aren't a problem to work with KBL.

Still wish there was some way to downgrade 14nm chipsets to ME11, but maybe that's impossible. All 14nm chipsets use ME12 and all 22nm use ME11 regardless of age, so ME version may be tied to chipset.

ME is not independend of age, e.g. ME 14 runs on Union Point just as well in case of H410 and B460. But the other way round is impossible: The code for 14 nm silicon was just not there in old MEs and apparently no one has found a way to mod the new MEs for different CPUs or to get the old MEs compatible. (Which is a pitty because I am 99.9% that all physical interfaces of RKL are 22 nm compatible and only ME keeps Core i 11000 from running on B460 and I also see a 60% chance that anyhting running on B460 will run on Z370 and on Z170.)
#IdidntgetPalmCovebutIwouldacceptCypressinstead >:-)

I wonder if there were a way to delete ME from BIOS entirely, if it would be possible to boot an unsupported CPU. HAP disable apparently happens too late.

Boot up code/CPU initialisation is part of IME and strictly necesarry. Replaycing this is the holy grail of all UEFI modding, but this would not only require to reverse engeneer some of the low levelst code at all, but also getting the replacement to run on the IME cores in the PCH (from where the CPU gets initialised) which are in themselves another well kept secret and which are on top the physcial foundation for virtually any hardware-assisted security function in the system and as such hardened against tempering as much as possible. There have been a few successfull exploits for security holes in high level functions, but I know of no one even coming close to understanding the fundaments of Intel current IME technology. (There has been a huge shift with the introduction of Z170. Older IME has been cracked, but did not even use the same command set.)

HAP in comparision is nothing else than a "please shut down command", which deactivates all the IME parts that are not necessary for the system to continue working. A bit like pressing "stop" on a CD player whereas booting a CPU without IME is comparable to attempting to hearing the music by holding an LP in your hand.

Zitat von GnatGoSplat im Beitrag #22

---

Another update on downgrading a motherboard to work with SKL and KBL if anyone is interested.

Picked up a Gigabyte H310M-A 2.0 <-- 2.0 is key, because it's the 22nm version of H310M that uses ME11.
Used CoffeeTime to add KBL microcode. Flashed using a modified efiflash.exe. Didn't work due to ME11.8 CON apparently not liking KBL.
Used CoffeeTime to downgrade to ME11.7 CON.
Booted up first time, it's working happily with my $20 G4620. So far, no problems. So it looks like 22nm 3xx chipsets with the right ME aren't a problem to work with KBL.

Still wish there was some way to downgrade 14nm chipsets to ME11, but maybe that's impossible. All 14nm chipsets use ME12 and all 22nm use ME11 regardless of age, so ME version may be tied to chipset.

---

Zitat von meredito im Beitrag #26

---

Hi there. I'm in a similar situation and really want to try it but with a 6700k instead. Did you have to do the pin mod and did you do it without a spi programmer? Thanks..

---

Zitat von GnatGoSplat im Beitrag #27

---

Zitat von meredito im Beitrag #26

---

Hi there. I'm in a similar situation and really want to try it but with a 6700k instead. Did you have to do the pin mod and did you do it without a spi programmer? Thanks..

---

Are you using an H310M-A 2.0? If so, no pinmod necessary and no programmer necessary. I did use a hacked Efiflash.exe that someone posted on this forum awhile back. That's necessary to flash hacked BIOS because the unhacked one won't flash a modded BIOS.

---

I think it is possible now
Long story short.
My distributor insisted that Gigabyte B365M Gaming HD Rev 1.0 support 6th, 7th, 8th and 9th Gen Intel CPU.
So I ordered 1 for testing and it did support i5-7400.
I couldn't test fully as my is CPU half broken (Can access BIOS and DOS only).
I dumped the BIOS with CH341A programmer.
Any BIOS expert care to analyse the BIOS dump?
By the way how is it even possible?
Manufacturer: Gigabyte
Model: B365M Gaming HD
Rev.: 1.0
BIOS Ver.: F20
Bios Date: 29th July 2021