AutoQuad Forum

Hi Max,

Thanks for your script, I'll try it later.

I think printf/scanf contributes most part of the problem. The code stops crashing after I increase stack size, but it doesn't produce expected outputs until I apply the aligned(8) fix.

Another thing I'd like to try it to move params io to another low priority task, which only gets executed when it's requested to read/write params. In this case, the stack of other tasks can be reduced.

Regards,
Ye

Quick Update:
I tried to apply aligned(4) for structs with no 64 bit member, e.g. long long and double. Then printf starts to generate rubbish info in sd card. So I'll keep 8 bytes alignment for now.

OK, many updates.... https://github.com/mpaperno/aq_flight_c ... ntegration

"feature_integration" branch has everything so far combined together. There are also separate standalone branches for some of the changes. Summary: https://github.com/mpaperno/aq_flight_c ... CHANGES.md

Ye, the memory changes are also in their own branch, possibly a bit clearer for review: https://github.com/mpaperno/aq_flight_c ... ory_manage (but please make changes against the feature_integration branch.)

Cheers,
-Max

Wow, that's huge amount of work. I'll go through those changes and puts my $0.01 there

Regards,
Ye

It does look like a lot, but hopefully mostly because I broke it down into many smaller pieces...

BTW, I implemented the idea of having all the config param reads/write requests funneled through one thread -- supervisor. It's low-priority, and was already calling most of those routines anyway (eg. when using stick commands to save params). This reduces required COMM stack space quite a bit. I did add some space to SUPERVISOR stack, just in case, but it's not really required.

Anyway, the relevant commit is here: https://github.com/mpaperno/aq_flight_c ... 0cb0dc2dae

-Max

that is a good news, great job!