[openib-general] Re: Re: madvise MADV_DONTFORK/MADV_DOFORK

Gleb Natapov glebn at voltaire.com
Mon Feb 20 01:03:23 PST 2006


On Wed, Feb 15, 2006 at 10:23:31AM +0200, Michael S. Tsirkin wrote:
> Quoting r. Gleb Natapov <glebn at voltaire.com>:
> > Subject: Re: Re: madvise MADV_DONTFORK/MADV_DOFORK
> > 
> > On Tue, Feb 14, 2006 at 06:42:36PM +0200, Michael S. Tsirkin wrote:
> > > All, MADV_DONTFORK patch is now part of the -mm tree.
> > > Everyone who's interested in fork support, please test 2.6.16-rc3-mm1 and
> > > publish the results here and on lkml.
> > > 
> > Good news!
> > 
> > Should call to madvise be the part of reg_mr call?
> 
> Probably no - MPI should have to do it.
> 
It had come to my attention that there is file memory.c in libibverbs
that implements refcounting for mlock. I think it was meant to be used
from reg_mr() (since interface is hidden) back when mlock was needed for
kernel bug workaround. If it was good idea back than why not now?
Alternatively we can make this interface public for application to use
explicitly.

Roland can you please tell the history behind memory.c?


--
			Gleb.



More information about the general mailing list