cvs commit: src/sys/kern kern_umtx.c src/sys/sys umtx.h

[Robert Watson]

On Mon, 31 Mar 2003, Jeff Roberson wrote:

> > It's great to be getting this.  Can you point me to a document indicating
> > how this will be used by KSE?  Are we going to have "native threads"
> > (thr), KSE, and pthreads?
> 
> I have not written a document for umtx.  I will probably write a man
> page at some point.  I'm not sure that KSE can use umtx directly.  Their
> requirements are different from thr and so this interface is slightly
> tied to the thr idea of threads. 
> 
> Currently you could use umtx to synchronize two processes if I wasn't
> using the PROC_LOCK() in kern_umtx.c.  I could do this if there was
> enough interest.  It sounds sort of neat anyway. 
> 
> The only reason this is tied to threads is because the thr_id_t is the
> "struct thread *".  I dont think kse currently exposes this to userland
> although it could. 
> 
> I'm sure I'll talk with the other kse folks at more depth about whether
> nor not they can use this for the libkse locks. 

Do you have any thoughts on how you might authorize inter-process
synchronization if not implicitly by requiring that synchronization occur
between threads in the same process?  One of the nice things about the
current model is that there are no concerns about authorization...

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert at fledge.watson.org      Network Associates Laboratories