Re: [PATCH 1/1] iser-target: Fix handling of RDMA_CV_EVENT_ADDR_CHANGE

[Jason Gunthorpe <jgg@nvidia.com>]

On Thu, Jul 22, 2021 at 12:54:26PM -0700, Sagi Grimberg wrote:
> 
> > > > > What is this trying to do anyhow? If the addr has truely changed why
> > > > > does the bind fail?
> > > > 
> > > > When the active physical link member of bonding interface in active-standby
> > > > mode gets faulty, the standby link will represent the assigned addresses on
> > > > behalf of the active link.
> > > > Therefore, RDMA communication manager will notify iSER target with
> > > > RDMA_CM_EVENT_ADDR_CHANGE.
> > > 
> > > Ah, here is my recollection...
> > > 
> > > However I think that if we move that into a work, we should do it
> > > periodically until we succeed or until the endpoint is torn down so
> > > we can handle address removed and restore use-cases.
> > 
> > That soudns better, but still I would say this shouldn't even happen
> > in the first place, check the address and don't initiate rebinding if
> > it hasn't changed.
> 
> But don't you need to setup a new cm_id for the this notification? It
> will remain active?

AFAIK the existing listening ID remains, the notification is
informative, it doesn't indicate any CM state has changed.
 
> Also it's a bit cumbersome to match addresses in some cases like multi
> address interfaces. Almost seems easier to setup a new one...

How so? There is only one address passed to bind. If you create
multiple CM ids to cover all addresses then you need to run a set
algorithm to figure out what cm_ids to destroy and which to
create.

Jason