Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
timing attack on one of those protected m16c mcus, in russian (2019)


english slides on original attack on laptop's microcontroller (with desoldering) https://q3k.org/slides-recon-2018.pdf

That's some fantastic news, thank you for linking this! Sounds like we have a viable way of dumping SMU firmware. Side-channel attacks are so neat!

Now I kinda wanna dump the firmware just to see how the SMU works and what it actually does.
 
I wonder if @barracuda156 non-working Quad can benefit from info gathered here by @Nullcaller ?

I have for now one of those dualcore 2.0Ghz all-PCIE powermacs from "late 2005", so SMU connector might prove to be useful if my memory swap/upgrade trick I plan for next days will not work?

G5s are notoriously impossible to unbork once borked. My guess is that the SMU debug output will be useful for the purposes of understanding what's wrong with the machine, but it will be very much useless for the purposes of actually fixing it.

Now, theoretically, if you extracted the SMU firmware and rewrote it for something less exotic than M16C/28, like STM32... I mean, nothing would prevent you from just straight up manufacturing a replacement motherboard for the G5.

It'd be a tremendous amount of effort to come up with a working prototype of the board, but with a pristine new board, it's a matter of relocating the critical components from the old board to the new. Which is also a challenge in and of itself, but all of it is technically possible (which is the best kind of possible), and about the only way G5s can truly be repaired.

I mean, the current only viable alternative of replacing the motherboard with a different yet equally flawed one from a G5 which somebody gutted to make a Hackintosh can hardly be considered repair.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.