search

9elements / picoprog

Apache License 2.0
5 stars 4 forks source link

faulty read #20

Open code4brot opened 2 months ago

code4brot commented 2 months ago

We've had several wrong reads using picoprog (+ osfc.io PCB enclosure).

This happened on reading several T540p chips. I think it's Macronix, will send the exact IC specificer(s) tomorrow when I have the boards under my fingers.

Misbehavior took place in 2 forms:

  1. successful read, followup read -> "NO EEPROM", dis- + reconnect IDC connector, successful read, followup read -> "NO EEPROM", ...

  2. 4M chip got read as 8M (with -r flag even 8M saved to disk)

this happened on both flashprog and flashrom.

Note on 1.) successful reads between fuckups seem consistent, sha1sum is equal.

Note on 2.) The clips didn't move a bit. We only found that one by reverting to ch341a (after ordering new clips of course).

Haven't tried any of this against the original pico-serprog firmware. Will try soon and report back!

code4brot commented 2 months ago

The chip was a MX25L3273E.

ArthurHeymans commented 2 months ago

This board has 2 flash chips right? Are you pulling CS high on the other chip?

code4brot commented 2 months ago

Yep it does. No I'm not, at least not knowingly. I leave the other chip alone when reading the 4M chip.

ArthurHeymans commented 2 months ago

images (6).jpeg

If CS is not explicitly pulled high on the other chip it might be confused as its left floating. Please try that.

MDr164 commented 1 week ago

I found out that there is an issue on some chips with regards to how chunking is done right now. I got a potential fix but I don't have a chip here that actually shows that behavior. I'll push the update to the chunking this week, maybe give it another try after that.