#avr Logs

Mar 19 2021

04:37 AM vgtw_ is now known as vgtw
05:09 AM exp: back on this issue today, I'm not even sure where I should go next. I have a different board with a 646 on it where I can break out all the JTAG pins, but given that the writing code calls directly into avr-gcc i doubt i'm going to see anything shocking
05:16 AM exp: it seems more like I've missed something in a datasheet somewhere, and so have the bootloader authors, but so far I've found nothing referring to banks or mappings or extra locking
07:08 AM twnqx: exp: where did you get your 646s?
07:36 AM exp: twnqx: digikey, mouser, farnell
07:36 AM exp: none from anywhere i wouldn't trust, but it's a thought i've had too
07:37 AM exp: but i can flash that space with a usbasp, it's only SPM that's failing
07:38 AM twnqx: interesting
08:25 AM exp: twnqx: indeed, i double checked and yep i can read and write that segment with usbasp no concerns, but with SAM it cannot be written and remains 0xff up to the bootloader boundary
08:25 AM exp: i double checked the datasheet too and the largest NRWW is 8k bytes / 4k words
08:26 AM exp: so it shouldn't ever overlap 56k
08:33 AM LeoNerd: In general, you know that external programmers can write flash areas the chip itself cannot selfprogram, right?
08:34 AM LeoNerd: I don't know that chip in particular, but e.g. the newer UPDI-based ones have three zones; bootloader, applicaation, appdata. Each area can only be written by code running in /previous/ areas.. never itself
08:34 AM exp: LeoNerd: yeah that's the NRWW zone
08:34 AM exp: the addresses in the datasheet go to 0x6FFF words
08:41 AM exp: sorry that was quite without context, was on an exceedingly long meeting
08:42 AM exp: the datasheet specifies two zones, application and bootloader, the bootloader segment extends 4k words down from the end, so finishes at ~56k, regardless i also have the lock fuse set to 0xff so no protection should be active
08:43 AM exp: I've stripped the bootloader code down to the bare avr-gcc calls now and it still doesn't function, and there's no sign of any mistake in what avrdude is sending, so i am somewhat at a loss
08:43 AM exp: i'm going to write an example bootloader that simply flashes a monotonic sequence using the bare boot_page_etc functions and post on avrfreaks or wherever if that fails
10:13 AM McDonaldsWiFi is now known as PublicWiFi
12:08 PM [1]MrMobius is now known as MrMobius
01:28 PM [1]MrMobius is now known as MrMobius
08:47 PM rue_mohr: hmm, what the normal internal clock rate for a m8, 1mhz?
10:07 PM rue_mohr: you guys are really bad at answering simeple ones eh?
10:07 PM rue_mohr: "guys"
10:08 PM rue_mohr: but still, cmon, I mean, like really.