Quote:EDIT: My bios chip might be a W74M12JW with a HMAC!
Yeah, this is the BIOS chip. The other G513/G517 models on the original thread have this same chip:
https://www.bios-mods.com/forum/Thread-R...#pid188379
The link for the documentation for this chip went down, but I have a copy, which I have attached to this post. This chip has two features that could make hardware flashing the chip more difficult: HMAC authentication (as you mentioned) and a required voltage modulation on the /CS pin using a pull-up resister to enable read/write functionality. Software flashing appears to not be feasible with this chip based on previous tests.
I am hoping that the manufacturer did not actually implement the optional HMAC authentication capabilities of the BIOS chip (similarly to how Intel BootGuard usually isn't enabled). If they did, I am not sure how to bypass that. I am pretty sure though that at least the voltage modulation will be required here.
If you want to go ahead with this, I can modify this BIOS image. First, we will need to get the voltage modulation part down. To test if we are successful, we will need to get two consistent BIOS dumps in a row. This means two things:
1.) The 2 dumps are identical to each other with a hex comparison.
2.) The data inside forms a coherent BIOS image that can be opened and extracted in UEFITool.
I imagine that getting identical dumps will mean that we met the voltage requirement for activating the BIOS chip for reads/writes. I imagine that if the data also makes sense, then HMAC isn't implemented. Or it's only implemented for writes, but the fact that AFU has software read permissions to the chip suggests that they might not block hardware reading attempts with an SPI programmer.
On the other thread with this BIOS chip, we have inconsistent, mostly garbage data. This suggests that the chip isn't getting enough voltage for proper reads (as we haven't tried modulating the voltage yet). Based on occasional identical segments of data from dumps, I'm leaning toward HMAC not being implemented- at least for hardware dumping/flashing - which would constitute good news for us.
Let me know if you want to proceed. CH341A SPI programmers tend to be $10-$11 USD, a (potentially necessary) better clip for connecting to the BIOS chip is usually around $20-$25 USD, and I imagine equipment for a pull-up resister might be less than $5 USD. You will also need a 1.8V converter module for the CH341A. This is a few dollars. The better clip is only required if the included one doesn't make a good connection, but it might be a good idea for our purposes since there are multiple things that could cause a bad BIOS dump here.
If we can get past the voltage requirement and find that there is no HMAC implemented, I can provide a BIOS mod for not only your 2 notebooks, but also 6 others on the other thread.