[openib-general] Re: [PATCH] ipoib: tx ring overrun

[Roland Dreier]

    Michael> Dont try to post more send work requests if the TX ring
    Michael> is full.  Setting netif_stop_queue is insufficient: linux
    Michael> can still land a tx packet on us.

I'm confused -- is the code in ipoib_start_xmit()

	/*
	 * Check if our queue is stopped.  Since we have the LLTX bit
	 * set, we can't rely on netif_stop_queue() preventing our
	 * xmit function from being called with a full queue.
	 */
	if (unlikely(netif_queue_stopped(dev))) {
		spin_unlock_irqrestore(&priv->tx_lock, flags);
		return NETDEV_TX_BUSY;
	}

not enough to prevent us from trying to queue a TX packet after
stopping the queue?

BTW, I've lost track of the pending IPoIB patches a little bit.  I
have a lot of patches queued for review, and I'm not sure which have
been replaced by new versions, which are critical, etc.  Could you
send me a list of which patches still need to be applied, and which
ones fix problems you hit in testing (vs cosmetic changes, memory
leaks and so on).

 - R.