[openib-general] Re: ib_cancel_mad and the CM

[Sean Hefty]

Hal Rosenstock wrote:

>>Alternatively, it may be possible to change 
>>ib_cancel_mad to cancel MADs based on a second set of criteria.
> 
> The second set of crtieria could be an incarnation number associated
> with the message, saved and later checked before cancel to make sure it
> is the same. However, I think this means an additional field in struct
> ib_send_wr although I only quickly glanced at this.

I think that would be the case.  Of course, the CM could use the wr_id 
differently, with the drawback that it could no longer store a pointer. 
  E.g. It could store the cm_msg in an idr, with part of the wr_id 
being the incarnation number.  What I'm trying to determine is if this 
is an issue related to just the CM implementation, or if this would be 
something that multiple users would run into.

- Sean