Kaspar Schleiser
Kaspar Schleiser
I'm hitting this again, after fixing it last time with a fake threshold. > Would having a `--no-thresholds` flag or something be useful here? yes, I think so. I'm happy...
@dependabot rebase
Fixed with #734.
> The heuristic can fail if despite the stack overflow the canary is still there (hitting the value by chance is 1 to 2^32-1 on 32 bit systems). The heuristic...
> As far as I recall these MPUs (or at least some of them) only allow to protect a limited number of memory segments, i.e., cannot be used for arbitrary...
I also experimented with enlarging the canary value to a proper redzone (bringing non-mpu checks closer to what the mpu does), but never got to PR that. [this](https://github.com/kaspar030/RIOT/tree/redzone) is the...
> I like the idea, but I don't like zero being the magic number there. I guess re-using the stack test approach makes sense. > Or is the context switch...
I do see this is basically fixing three things, should I split this up? Or would that triple testing?
Interesting!