libkse and SMP (was Re: USB bulk read & pthreads)

Daniel Eischen eischen at
Fri May 23 13:33:18 PDT 2003

On Fri, 23 May 2003, Julian Elischer wrote:

> On Fri, 23 May 2003, Terry Lambert wrote:
> > Daniel Eischen wrote:
> > 
> > This is handy to know; so basically, my expectation from
> > reading the code around PTHREAD_SCOPE_SYSTEM was correct:
> > a single CPU system with PTHREAD_SCOPE_PROCESS (the default)
> > can still get itself blocked in the kernel by a single
> > blocking call (as in the USB bulk read device issue).
> No you are completely wrong..
> Each PTHREAD_SCOPE_PROCESS gets its OWN KSE to run on.
> that thread may block but other threads may run unimpeded.

I'm not sure if what you meant here, but here's a (hopefully)
clearer explanation.

All threads that are created with PTHREAD_SCOPE_PROCESS
(the default) will run in the same (initial) KSEG.  The
initial KSEG will have as many KSEs as CPUs by default.
When a scope process thread blocks in the kernel, upcalls
are made to the originating KSE and a new thread is scheduled.
When scope process threads unblock in the kernel, upcalls
are made to one or more of the same KSEs within the initial
KSEG to notify the library that the threads can be resumed.

Each scope system thread gets its own KSE/KSEG pair in which
to run.

Dan Eischen

More information about the freebsd-hackers mailing list