strace exits after 4096 pagefaults?

[Roland McGrath]

> Roland McGrath said:
> >> "Maximum number (4096) of interrupts exceeded.  Exiting."
> >
> > strace contains no such message.
> >
> 
> Sorry, you're right. The application I was debugging emits this, though
> not unless being straced.

It's possible there is an strace bug that causes this.  However, it's also
possible that there is no way for strace to avoid perturbing the process in
the way that is causing the problem.  For example, just slowing it down
more than it can cope with.  There are other corner cases where the kernel
interfaces necessitate some perturbation of behavior to trace.  

You'll have to try to figure out exactly what conditions your program
experiences differently under strace and not.