[openib-general] Re: RFC: ib_cache_event problems

[Michael S. Tsirkin]

Quoting r. Roland Dreier <rdreier at cisco.com>:
> Subject: Re: RFC: ib_cache_event problems
> 
>  > > Ah, that looks like the bug I guess.  What's the situation?  SM clears
>  > > P_Key table and then later readds a P_Key?
> 
>  > Any ideas on how to fix this?
> 
> Does it work to just start the pkey_task if ipoib_ib_dev_flush() wants
> for a P_Key that's not there?  Or is it trickier?

If this works, why is dev_up playing with pkey_chek_presence at all?
Can we kill all of this then?

        ipoib_pkey_dev_check_presence(dev);

        if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) {
                ipoib_dbg(priv, "PKEY is not assigned.\n");
                return 0;
        }

It seems we must avoid joining multicast groups while key isn't assigned ...


-- 
MST