Re: RFC: Fix gdb 5.1 for Linuxthreads

["H . J . Lu" <hjl@lucon.org>]

On Wed, Sep 19, 2001 at 04:04:47PM +0200, Mark Kettenis wrote:
> "H . J . Lu" <hjl@lucon.org> writes:
> 
> > This patch seems to work for me. Any comments? It also fixes debugging
> > the statically linked linuxthreads binaries.
> 
> What's the rationale behind this patch.  Even though it might seem to
> fix your problems, it looks wrong to me.  It seems to add "robustness"
> in the wrong places, such that other assumptions that I made in the
> lin-lwp.c module might be violated.  Certainly the outcome of
> is_cloned() won't be correct in some cases.

This patch tries to show the problem which is calling wait () on cloned
processes. I will buy it is wrong. Can you provide a testcase to show
it like what I did?

> 
> The problem with attaching should be solved in a different way.  And

That is why it is RFC.

> what's wrong with debugging statically linked LinuxThreads binaries?

Just compile my testcase statically and attach it. You can see it
yourself.


H.J.