Re: Question about CLONE_CLEARTID and thread group leader

[Ingo Molnar]

On Tue, 17 Sep 2002, Jamie Lokier wrote:

> One question has been bothering me for a while: what about the thread
> group leader's stack?  These days, isn't it the case that the group
> leader is supposed to be equivalent to the other threads?  If so, how
> does it exit and release its own stack -- or do we understand that the
> group leader, as a one-off exception, has to block signals before
> exiting?

good question. We might need a new mechanism (new syscall) for a thread to
set its own ->user_tid after it has started up. I'll code it up. The
fastest thread-startup method is still to also have this mechanism
provided by clone() as well - but oviously at exec() time we cannot know
about such issues.

	Ingo