On Sat, 4 Jun 2005, Christoph Hellwig wrote:

hch> On Fri, Jun 03, 2005 at 03:32:37PM -0700, Tom Duffy wrote:
hch> > This patch is HUGE.  Sorry.  Not much I could do to split it.
hch> > 
hch> > It makes a superclass for all the dat objects, and subclasses
hch> > dat-provider (dapl) off of it.  No more opaque handles.  No more magic
hch> > checking.  I actually found a few bugs from the compiler because of it
hch> > (more than reported to openib).
hch> > 
hch> > The dapl_header struct is renamed dapl_common.  Most of the _handle
hch> > variables are renamed removing _handle.  It removes the
hch> > get_handle_type() function since this is no longer used.  Also, it
hch> > unimplements [get|set]_consumer_context (for now) since nobody was using
hch> > it.  DAT_BAD_HANDLE() goes as it becomes a simple NULL check.
hch> 
hch> And event that should go away in most cases.  In the kernel we should't
hch> check for NULL pointers all over the place, if client programmers don't
hch> stick to the APIs they deserve to get panics.

Any objections to removing the NULL pointer checks of API parameters?
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general

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

Reply via email to