RT task affected by disk access; also does not appear to sleep for requested period

Ryan Turner Ryan.Turner at virginia.edu
Wed Jul 6 18:38:22 CEST 2005

I've posted a log to:


I would have run the tests for longer, but it crashed after 1 hour 22
minutes, with the message:

/usr/fusion/bin/rtai-load: line 247: 27115 Segmentation fault   $suflag
$* $cmdargs

It is also strange to note that the cat /proc/interrupts command (or
rather, bash) crashed several times (after only a few hundred iterations
in some cases) and proceded to log me out of the terminal.  The cause of
the crash was not always identical, but in one instance the message was:

malloc: ../bash/execute_cmd.c:261: assertion botched
free: called with unallocated block argument
last command: ...
Stopping myself...

In addition, ping -f -s1500 <mygateway> crashed twice with a
segmentation fault.

Very strange that these standard programs are crashing!  This
installation has otherwise been very stable.  Any input is appreciated
on both the results of the stress test (as found in the log above) and
on the cause of these peculiar crashes.


Ryan Turner wrote:

>Sorry for that, added another 's'; so it would be:
>And yes, I ran each test individually.  I will try again running all
>simultaneously (I will call it latency-stress.out).
>Paolo Mantegazza wrote:
>> The above urls are not accessible. The output is in nanos anyway.
>>Notice I do not understand why you have 4 urls. I meant you had to
>>apply all the loads I told you at once, that's the way I do true
>>stress testing over a whole day at least.
>>RTAI mailing list
>>RTAI at rtai.org

Ryan Turner <Ryan.Turner at virginia.edu>
University of Virginia - Computer Engineering

More information about the Rtai mailing list