> Hi Long, > > Assume that if "count > MANA_MBUF_BULK", and int the first iteration of the > loop 'mana_post_rx_wqe()' failed, but in second iteration it is successful, > this > will cause function to return success in spite of failure in first iteration. > > As mbufs posted Rx queue, it may be OK to consider above case as success, > but since 'count' number not posted this may be misleading. > I just want to double check if this is done intentionally.
You are correct about the first loop may fail and second may succeed. I'm sending v5 to fail the queue when this happens. This is fatal and we should stop the queue. Thanks, Long > > > With the limitation of VLA code become more complex, and if there is no > performance benefit of the allocating 'mbufs' array from stack, you may > prefer to switch back to allocating dynamic memory, up to you. >
