Messages in this thread | | | From | Roland McGrath <> | Subject | Re: [PATCH 2/2] ptrace children revamp | Date | Mon, 14 Apr 2008 18:41:10 -0700 (PDT) |
| |
> Sorry for delay!
There's no need to apologize! I heartily appreciate all your help on this.
> Agreed. I never understood why we need __WNOTHREAD. The same for > ->pdeath_signal in its current "per-thread" form. I think it would be > very nice to kill them both (or send the ->pdeath_signal when the whole > process exits). Then we can place all childrens on one signal->children > list. But this is a bit off-topic for now.
I'm in complete agreement here, but indeed it's to be considered later. I never understood pdeath_signal, but I recall someone piping up before and saying it really was used with its current semantics, so go figure. (Btw, we can move ->children and still keep __WNOTHREAD compatibility. It just has to check p->parent == current for __WNOTHREAD. We can do that if we determine that __WNOTHREAD is purely for anal backward compatibility, and noone cares about the performance difference of __WNOTHREAD calls only looping over a shorter list.)
> I think the 4th patch has a small problem,
Thanks. I think we're moving on to the other variant of the patch now, so I won't worry about fixing up the version we're abandoning.
> Yes! I thought about this too. Actually, I was very sure that this is your > plan from the the very beginning ;)
You were quite right! I somehow tricked myself into trying something inferior first. Silly me. :-)
Thanks, Roland
| |