Re: LD linker stubs?

[Thomas Bushnell, BSG]

Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de> writes:

> Can you expand on that?  In particular, I would be grateful if you could
> clarify "cancellation".  Cancellation in pthreads is requesting the
> termination of a thread, where I neither see how that is important in the
> Hurd, nor what is missing for that in the current implementation.

We use it to interrupt RPCs.  The Hurd's "thread_cancel" orders all
future RPCs made by the thread to stop in a controlled way--most
importantly, the RPC that is in progress at roughly the same time as
the cancel is made.

> So I guess you are referring to the interruptable RPCs.  Either we keep the
> current implementation, but put it on top of pthreads rather than cthreads,
> or we change the Hurd to use the pthread infrastructure for that (I guess
> that means using pthread_cleanup_* in the relevant places and using the
> terminating pthread_cancel, rather than using the hurd_condition_wait
> semantics, where a non-zero return value indicates cancellation).

Right.  Switching to pthreads beans doing just that.