Thursday 8 January 2015

Mystery solved!

I discovered that the mystery bootloader (see previous post) is from a Mediatek (urgh, again?!) MT6260. Oddly, flashable through USB1.1 instead of serial like I was expecting.

The download agent detected the external flash as a 32Mbit Winbond, and read it without trouble. I get the feeling there's a bit missing as I assume there to be internal flash in the SoC... but I don't know yet.

Wednesday 7 January 2015

Mystery bootloader & Work in progress

Does anyone recognise this SoC bootloader output? The actual IC is hidden under a sealed metal can that I don't really want to have to remove.There's a chance it is a Mediatek MT6xxx part, or an ST32*. It's part of a chinese clone of a GPRS module and doesn't output much more than this:

XT897 UART. "Can't be done" according to 'the internet'. Sure it can:
(disclaimer: work in progress; and only currently possible after the bootloader has finished)

I'll detail it once I figure out how the hardware bit is supposed to work (as opposed to how it is above - e.g. hold this wire here for 2 seconds, once that changes to this, swap that wire to this pin, etc)

And finally, just to show a reader who is having MTK trouble... framebuffer console is definitely possible :) The UART was invaluable in being able to write the LCM driver, although a good plan B is cause a kernel panic on purpose, and the android kernel dumps the last chunk of its ring buffer to a fixed location in RAM, which is accessible after a reboot (into 'recovery', for example)

This is also a W-I-P with the aim of eventually getting an up-to-date upstream+patches kernel running on this hardware (MT6582).