tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: revivesa status 2008/07/09



>>>>> On Mon, 28 Jul 2008 20:35:08 -0700,
      Jason Thorpe <thorpej%shagadelic.org@localhost> said:

> That said, I'm not sure that has EVER really worked properly on our  
> system, or if supporting it is even really a good idea.  Using some  
> random pthread library can't possibly work -- think  
> "PTHREAD_MUTEX_INITIALIZER".

Not only PTHREAD_*_INITIALIZER, but also some more ABI constraints
must be kept.  It needs careful work, of course.

> Now, if we wanted to have 1:1 threads in libc and a system-supplied  
> libpthread_sa that was fully ABI-compatible with libc's 1:1 threads,  
> that seems workable to me, and also seems like a pretty reasonable way  
> to select which behavior you want (not that I really think we ought to  
> be shipping a libpthread_sa, but I suspect I'll lose that argument  
> "just because").

I still don't understand why it's better to integrate 1:1 thread into
libc.  Just same thing can be achieved by separated libptherad, and it
must be cleaner and easier to enhanace libpthread implementation, even
for 1:1 case (e.g. trying enhanced 1:1 thread).
-- 
soda




Home | Main Index | Thread Index | Old Index