Re: [PATCH] IB/cma: Use inner P_Key to determine netdev

[Haggai Eran <haggaie-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>]

On 20/10/2015 19:44, Jason Gunthorpe wrote:
> On Tue, Oct 20, 2015 at 09:45:27AM +0300, Haggai Eran wrote:
>> On 19/10/2015 21:19, Jason Gunthorpe wrote:
>>> On Mon, Oct 19, 2015 at 09:09:25PM +0300, Haggai Eran wrote:
>>>> When discussing the patches to demux ids in rdma_cm instead of ib_cm, it
>>>> was decided that it is best to use the P_Key value in the packet headers
>>>> [1]. However, some drivers are currently unable to send correct P_Key in
>>>> GMP headers.
>>>
>>> You should explicitly describe the broken drivers in the commit text.
>> These are mlx5 and ipath. I'll add them to the commit message.
> 
> And ipath is actually ipath, the obsolete driver being removed, not
> qib? (ie we don't care about it?)
Right. qib is fine.

>> The remaining issue is that it doesn't respect the
>> ib_send_wr.ud.pkey_index field when sending.
> 
> But this is a very serious bug, to the point the mis-labeled packets
> may not even be delivered in some cases - you really care about the
> sub case where mis-labeled packets happen to be deliverable but don't
> parse right?
I understand the issue in mlx5 is serious but we've lived with it a long
time. In this patch I only wanted to work around it for the purpose of
fixing the regression introduced in 4.3.

> Well, don't forget to undo this patch when the mlx5 driver is fixed..
Of course.

Thanks,
Haggai
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html