[ofa-general] Re: [PATCHv2 RESEND] IB/IPoIB: Don't let a bad muticast address in the join list stop subsequent joins
Yossi Etigin
yosefe at voltaire.com
Mon Aug 24 09:48:51 PDT 2009
On 24/08/09 19:15, Jason Gunthorpe wrote:
> On Mon, Aug 24, 2009 at 04:51:03PM +0300, Moni Shoua wrote:
>
>> http://lists.openfabrics.org/pipermail/general/2009-July/060496.html
>> The discussion in the link above didn't end with a decision. You were asking
>> about a way to inject illegal mcast addresses from userspace to ib_ipoib and
>> Jason pointed about such (described below). Could you please review the patch?
>
> FWIW, upon looking at this more closely, I would rather see this patch
> of yours fix the timeout problem. This actually has nothing to do with
> illegal addreses but with any situation where the SA returns failure
> (ie MLID exhaustion, etc)
>
> There is already a per-event increasing back off, it just needs a
> little fussing to keep track of time properly and sort the list by
> expiration.
>
> Jason
Are you suggesting to sort the list each time we have add/remove a new entry,
or search for the correct location to insert the new entry? I'm afraid that
would add too much complexity and be inefficient (in O() terms).
Moreover, I believe that moving a failed mcast entry to the end of the list
behaves the same as always joining the least-backoff-value mcast entry (since
everybody start with the same backoff).
BTW Moni - Do send-only joins need the same solution too?
--Yossi
More information about the general
mailing list