Re: [PATCH RESEND] infiniband:core:Add needed error path in cm_init_av_by_path

[Jason Gunthorpe <jgunthorpe@obsidianresearch.com>]

On Wed, Dec 16, 2015 at 11:26:39AM +0100, Michael Wang wrote:
> 
> On 12/15/2015 06:30 PM, Jason Gunthorpe wrote:
> > On Tue, Dec 15, 2015 at 05:38:34PM +0100, Michael Wang wrote:
> >> The hop_limit is only suggest that the package allowed to be
> >> routed, not have to, correct?
> > 
> > If the hop limit is >= 2 (?) then the GRH is mandatory. The
> > SM will return this information in the PathRecord if it determines a
> > GRH is required. The whole stack follows this protocol.
> > 
> > The GRH is optional for in-subnet communications.
> 
> Thanks for the explain :-)
> 
> I've rechecked the ib_init_ah_from_path() again, and found it
> still set IB_AH_GRH when the GID cache missing, but with:

How do you mean?

                ah_attr->ah_flags = IB_AH_GRH;
                ah_attr->grh.dgid = rec->dgid;

                ret = ib_find_cached_gid(device, &rec->sgid, ndev, &port_num,
                                         &gid_index);
                if (ret) {
                        if (ndev)
                                dev_put(ndev);
                        return ret;
                }

If find_cached_gid fails then ib_init_ah_from_path also fails.

Is there a case where ib_find_cached_gid can succeed but not return
good data?

I agree it would read nicer if the ah_flags and gr.dgid was moved
after the ib_find_cached_gid

> BTW, cma_sidr_rep_handler() also call ib_init_ah_from_path() with out
> a check on return.

That sounds like a problem.

Jason