Re: [Gumstix-users] Atomic operators

[Adam M. <mcl...@gm...>]

They're being treated as data registers.  One thread writes to them,
two threads read from them.  So each thread has a pointer to an object
that contains the array; one of them calls set methods, others call
get methods.


On 8/25/06, Dave Hylands <dhy...@gm...> wrote:
> Hi Adam,
>
> >  I have a large array of 'long int's that several threads are
> > accessing quite often.  I'd rather not cede performance to a number of
> > mutexes.  As I understand, one way around this is to ensure that the
> > assignment operator for a value is atomic, and this may be possible
> > for single instance of a primitive type via an assembler call.  Does
> > anyone know if it's possible to atomically assign a 'long' value on
> > the PXA like this?
>
> On the pxa, assignment of a 32-bit long that's aligned on a 32-bit
> boundary should be atomic.
>
> But there are bigger issues.
>
> What are you doing with the array of longs? How are they being manipulated?
> You'll probably also need to ensure that any data being shared by
> multiple threads is declared volatile.
>
> --
> Dave Hylands
> Vancouver, BC, Canada
> http://www.DaveHylands.com/
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> gumstix-users mailing list
> gum...@li...
> https://lists.sourceforge.net/lists/listinfo/gumstix-users
>