Re: traced process hangs [Was: No Subject]
Nate Eldredge (nate@cartsys.com)
Mon, 05 Apr 1999 19:18:43 -0700
> There appears to be a bug in 2.2.5 with the ptrace stuff. It appears to
> only happen on SMP systems. I have found that the process being straced
> ends up eventually getting stuck in the stopped state. It's quite easy to
> reproduce:
>
> [sroot@oof:/]# strace yes > /dev/null
>
> ...wait about a minute or so...
>
> (in another console)
>
> [sroot@oof:/]# ps auxw | grep yes
> root 11521 15.4 0.3 936 396 3 S 17:59 0:34 strace yes
> root 11522 49.7 0.2 836 268 3 T 17:59 1:51 yes
> root 11533 0.0 0.2 952 312 4 S 18:01 0:00 grep yes
>
> kill -CONT 11522 would make it continue again as normal until the next
> time it occurs.
>
> (Using strace version 3.1.0.1)
Are you sure that `yes' actually is getting stopped? In the sense of,
"No more y's appear." Because the T state also occurs normally for a
traced process (when it makes a syscall, it gets stopped and the
debugger/tracer gets to poke it around for a while until it allows it to
resume).
If it really does get stopped, it could be because strace is hanging
somewhere. You might try tracing strace and see what it's up to.
You might also consider trying the Debian strace, which has quite a few
recent changes wrt 3.1.0.1.
--
Nate Eldredge
nate@cartsys.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/