Re: [patch] Fix attaching to Linux stopped processes

[Daniel Jacobowitz <drow@false.org>]

On Mon, Sep 18, 2006 at 12:12:28AM +0200, Jan Kratochvil wrote:
> Hi,
> 
> sleep 1h& pid=$!; kill -STOP $pid; gdb sleep $pid
> 	->
> Attaching to program: /bin/sleep, process 20768
> ../../gdb/linux-nat.c:1057: internal-error: linux_nat_attach: Assertion `pid == GET_PID (inferior_ptid) && WIFSTOPPED (status) && WSTOPSIG (status) == SIGSTOP' failed.

> +  /* Do not check `WSTOPSIG (status) == SIGSTOP' as the status may be
> +     arbitrary - depending on the signal that stopped the processes.
> +     If the process was running we get SIGSTOP, if it was already stopped
> +     by SIGSTOP we get 0.  The value gets used for `PTRACE_CONT'.  */

That's a very strange behavior.  It doesn't make much sense to me, and
it isn't familiar - which is strange because I've been all over the
kernel code for this.  So I checked.  Sure enough, on my 2.6.18-rc4
installation, this is _not_ what happens.  Instead, the traditional
thing happens: wait hangs and no new event is returned.

Is this new in kernel.org?  Or is it a Red Hat kernel patch?  Where
did it come from?

-- 
Daniel Jacobowitz
CodeSourcery