Hi Baptiste,
I am sorry, but after the original failure, I tried tweaking a few
things like timer intervals, length of text in the print statement
etc, but the only difference was in when it failed, e.g. 50minutes,
1hour 10minutes etc.
RIOT is/was one of many RTOSes I've been looking at, and the
"/RIOT/tests/vtimer_msg"
program was the second program I ran, after the hello world program,
and frankly I had no time to set up a debugging environment, so I can
only guess at the cause. It could be a trap (pointer access to invalid
memory etc), a Uart buffer bug, getting locked at empty/full, but not
so likely. My final guess is a Semaphore/Mutex lockup bug, because the
Uart output is never garbled, is complete, and in my case always ended
with
sec=31 min=5 hour=1
sec=32 min=5 hour=1
these lines suspiciously missing from the final lines of output. But
who knows...
the vtimer problems are (infamously) well known at the moment. We are currently putting a timer task-force in place that will focus on re-designing the timer infrastructure. So expect some further information on this soon!