[ofa-general] Re: [PATCH 4/4] [RFC] IPoIB/cm: Add connected mode support for devices without SRQs

[Roland Dreier]

 > 1. Do not understand the usage and purpose of recv_count (something new that
 > you have introduced). Can you please explain. However, the suspicion being 
 > that if somehow the if clause is executed, the rx_ring gets freed and so 
 > all the skb pointers are bogus. I have commented out this segment of code.

We need to keep track of the number of pending receives on a QP so
that we wait until all have been flushed before destroying the QP.
recv_count starts at ipoib_recvq_size and is decremented every time we
get an error completion and don't repost the receive, and when it
reaches 0 we move the QP to the reap list.

 > 2. The call to ipoib_cm_alloc_rx_skb() in ipoib_cm_handle_rx_wc() uses an
 > index value of 0 (hard coded) which is incorrect for no srq. I have changed
 > that to index instead.

Yes, I saw that and fixed it.  Actually I finished rewriting things so
that there is no more index and no static table any more.  I haven't
posted the patches yet because I wanted to read it over a little more,
but if you re-pull my for-2.6.25 branch you will get them.

I have a feeling that this second bug might be the cause of the
problems you are seeing.

 - R.