-
It seems I was lucky
10/26/2016 at 18:54 • 0 commentsI got several reports that flash chip in T100 is designed to work at maximum 2V VCC. So either I was lucky or motherboard has good overvoltage protection. Exact chip used in mine T100 (I've checked today) is Winbond W25Q64FWIG (datasheet) which is not supposed to get more than 2V. However it survived flashing twice :)
Also, after the sucessfull resurrection I found a .docx file which seems to be a service center document- http://docslide.nl/documents/t100ta-fw-update.html
Document has embedded archives, one of them contains an old (2xx) experimental/debugging UEFI which is a full 8MB file suitable for debricking (bonus- you get service software for all sensors onboard). -
What possibly went wrong with downgrade
01/05/2016 at 12:52 • 0 commentsI finally had some time to take a look at flash dump and to compare it with "untouched" UEFI image.
Best guess- NVRAM was unreadable for the UEFI.
It seems that NVRAM is stored within the UEFI starting at 0x10358- untouched 314 image on the left, backup on the right:
I also guess that default setup values are stored just before NVRAM, starting (possibly) at 0x10060
However data dumped form the flash chip differs from the default image:
- several bytes also in (possibly) default NVRAM data area have different values (eg. 0x10064, 0x1007B and some other bytes after that)
- starting at 0x1010F data seems to be almost the same in both images but written with a growing offset- at the beginning it's just one byte but later (0x10358) offset grows to 10 bytes
(314 image on the left side , dump on the right)
But this is exactly how 227 image looks like (as before- original on the left side, dump on the right):
My guess:
Asus WinFlash utility moved NVRAM data (and default setup values) to the place compatible with provided UEFI image and after the reboot UEFI's built-in flasher was responsible for image update. But in my case it refused to work (although I forced WinFlash not to check image build date) and I'm not surprised that T100 didn't want to boot without access to proper NVRAM data.
If only there was a good old "clear cmos" jumper somewhere......