[riot-users] xtimer_drift and the cc26x0
ben.l.postman at gmail.com
Fri Jul 5 19:28:21 CEST 2019
I am using the cc26x0 CPU with a custom board, and am getting very poor
performance with xtimer: Sleep times are not always what I expect them to
be, xtimer_now() calls are not monotonic, etc. The xtimer_usleep test
occasionally fails, and xtimer_drift fails every time, with extremely large
drifts and jitters. It is clear that it's some issue with a missed
rollover, because the jitter will jump up all of a sudden and then slowly
decrease, leaving the drift very large.
What is confusing to me, is that although I'm using a custom board, my
timer and xtimer configuration is identical to the cc2650stk (and it's the
same CPU). I see that there was an issue in 2017 with xtimer tests passing
on this board, and that the issue was resolved. So I just wanted to see if
anyone with experience on the cc2650stk could confirm to me that
xtimer_drift still works with that configuration, or if they had any
general debug advice for me.
I know that there are many open issues and PRs for xtimer, but I think the
frequency that I'm seeing errors at indicates that there is something more
fundamental going wrong. The periph_timer test passes for me, but I'm
wondering if I'm somehow configuring the HW timers wrong or something like
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the users