On Thu, May 13, 2010 at 10:35:14AM -0700, Roland Dreier wrote:
> > One reason is that get_src_path_mask() may access ah_lock.
>
> I don't see that:
>
> static u8 get_src_path_mask(struct ib_device *device, u8 port_num)
> {
> struct ib_sa_device *sa_dev;
> struct ib_sa_port *port;
> unsigned long flags;
> u8 src_path_mask;
>
> sa_dev = ib_get_client_data(device, &sa_client);
> if (!sa_dev)
> return 0x7f;
>
> so if we never add the sa_client structure it just returns.
>
I see... so I have no good reason to not treat sa_query.c in the same
manner I did for multicast.c.
By the way, how do you prefer to work out all these comments. Would
you prefer a new patch set or would you prefer to push the fixed
versions to your tree?
Also, there are a few fixes that I pushed to OFED that were pushed
after I sent V8. How would you like me to communicate them?
_______________________________________________
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
