[openib-general] Re: Re: ipoib: outstanding patches

[Michael S. Tsirkin]

Quoting r. Roland Dreier <rdreier at cisco.com>:
> Subject: Re: Re: ipoib: outstanding patches
> 
>     Michael> We cant leave the destructor set after unloading the
>     Michael> module.
> 
>     Michael> This is because ops structure is not per neighbour and
>     Michael> not per device.  So while kernel destroys all neighbours
>     Michael> it created for our device, the ops structure is shared
>     Michael> with neighbours for other devices.
> 
> Ugh.  So we're relying on the fact that no one else is sticking
> something in the ->ha member where we stash the ipoib_neigh.

Ugh. Right. With the all neighbour list we can do a lookup there
so make it safe.

> Basically we're getting lucky that this works at all.
> 
> As you said in your other mail, it seems like a more fundamental
> reorganization of this neighbour destructor stuff is required.

Right. [Sound of busy typing ]

-- 
MST