On Tuesday 19 February 2008 18:02, Tang, Changqing wrote:
> > >
> > I did this for ibv_srq, ibv_qp, and ibv_cq.
> >
> > ibv_srq and ibv_qp -- had to.
> >
> > ibv_cq -- just in case.
> >
> > ibv_context -- does not use cond.  If mutex changes, we're
> > cooked anyway.
> 
> I don't have many different OS, but on one of our machine:
> 
> Linux mpia32-5 2.6.16.21-0.8-smp #1 SMP Mon Jul 3 18:25:39 UTC 2006 i686
> i686 i386 GNU/Linux
> 
> sizeof(mutex)=24, sizeof(cond)=48
> 
> For other machines (64bit):
> 
> sizeof(mutex)=40, sizeof(cond)=48
> 
> So mutex size is really changed.
> 
> --CQ

This is not a problem.  I don't expect an app compiled against a 32-bit library
to work against a 64-bit library.

The case we are dealing with here is if, on the same host, an upgrade results in
binary incompatibility.

- Jack
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general

Reply via email to