IB/hfi1: Correct -EBUSY handling in tx code

The current code mishandles -EBUSY in two ways:
- The flow change doesn't test the return from the flush and runs on to
  process the current packet racing with the wakeup processing
- The -EBUSY handling for a single packet inserts the tx into the txlist
  after the submit call, racing with the same wakeup processing

Fix the first by dropping the skb and returning NETDEV_TX_OK.

Fix the second by insuring the the list entry within the txreq is inited
when allocated.  This enables the sleep routine to detect that the txreq
has used the non-list api and queue the packet to the txlist.

Both flaws can lead to having the flushing thread executing in causing two
threads to manipulate the txlist.

Fixes: d99dc602e2 ("IB/hfi1: Add functions to transmit datagram ipoib packets")
Link: https://lore.kernel.org/r/20200623204321.108092.83898.stgit@awfm-01.aw.intel.com
Reviewed-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
This commit is contained in:
Mike Marciniszyn 2020-06-23 16:43:22 -04:00 committed by Jason Gunthorpe
parent 822fbd3741
commit 82172b7655

View file

@ -369,6 +369,7 @@ static struct ipoib_txreq *hfi1_ipoib_send_dma_common(struct net_device *dev,
tx->priv = priv;
tx->txq = txp->txq;
tx->skb = skb;
INIT_LIST_HEAD(&tx->txreq.list);
hfi1_ipoib_build_ib_tx_headers(tx, txp);
@ -469,6 +470,7 @@ static int hfi1_ipoib_send_dma_single(struct net_device *dev,
ret = hfi1_ipoib_submit_tx(txq, tx);
if (likely(!ret)) {
tx_ok:
trace_sdma_output_ibhdr(tx->priv->dd,
&tx->sdma_hdr.hdr,
ib_is_sc5(txp->flow.sc5));
@ -478,20 +480,8 @@ static int hfi1_ipoib_send_dma_single(struct net_device *dev,
txq->pkts_sent = false;
if (ret == -EBUSY) {
list_add_tail(&tx->txreq.list, &txq->tx_list);
trace_sdma_output_ibhdr(tx->priv->dd,
&tx->sdma_hdr.hdr,
ib_is_sc5(txp->flow.sc5));
hfi1_ipoib_check_queue_depth(txq);
return NETDEV_TX_OK;
}
if (ret == -ECOMM) {
hfi1_ipoib_check_queue_depth(txq);
return NETDEV_TX_OK;
}
if (ret == -EBUSY || ret == -ECOMM)
goto tx_ok;
sdma_txclean(priv->dd, &tx->txreq);
dev_kfree_skb_any(skb);
@ -509,9 +499,17 @@ static int hfi1_ipoib_send_dma_list(struct net_device *dev,
struct ipoib_txreq *tx;
/* Has the flow change ? */
if (txq->flow.as_int != txp->flow.as_int)
(void)hfi1_ipoib_flush_tx_list(dev, txq);
if (txq->flow.as_int != txp->flow.as_int) {
int ret;
ret = hfi1_ipoib_flush_tx_list(dev, txq);
if (unlikely(ret)) {
if (ret == -EBUSY)
++dev->stats.tx_dropped;
dev_kfree_skb_any(skb);
return NETDEV_TX_OK;
}
}
tx = hfi1_ipoib_send_dma_common(dev, skb, txp);
if (IS_ERR(tx)) {
int ret = PTR_ERR(tx);
@ -612,6 +610,9 @@ static int hfi1_ipoib_sdma_sleep(struct sdma_engine *sde,
netif_stop_subqueue(txq->priv->netdev, txq->q_idx);
if (list_empty(&txreq->list))
/* came from non-list submit */
list_add_tail(&txreq->list, &txq->tx_list);
if (list_empty(&txq->wait.list))
iowait_queue(pkts_sent, wait->iow, &sde->dmawait);