search for: netdev_tx_busy

This effectively reverts 99ffc696d10b28580fe93441d627cf290ac4484c "virtio: wean net driver off NETDEV_TX_BUSY". The complexity of queuing an skb (setting a tasklet to re-xmit) is questionable, especially once we get rid of the other reason for the tasklet in the next patch. If the skb won't fit in the tx queue, just return NETDEV_TX_BUSY. It might be frowned upon, but it's common and not go...

This effectively reverts 99ffc696d10b28580fe93441d627cf290ac4484c "virtio: wean net driver off NETDEV_TX_BUSY". The complexity of queuing an skb (setting a tasklet to re-xmit) is questionable, especially once we get rid of the other reason for the tasklet in the next patch. If the skb won't fit in the tx queue, just return NETDEV_TX_BUSY. It might be frowned upon, but it's common and not go...

This paraphrases DaveM (and steals some of his words) explaining why a device shouldn't return NETDEV_TX_BUSY, even though it looks so inviting to driver authors. See http://www.spinics.net/lists/netdev/msg322350.html Inspired-by: David Miller <davem at davemloft.net> Signed-off-by: Rusty Russell <rusty at rustcorp.com.au> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h in...

Herbert tells me that returning NETDEV_TX_BUSY from hard_start_xmit is seen as a poor thing to do; we should cache the packet and stop the queue. Signed-off-by: Rusty Russell <rusty at rustcorp.com.au> --- drivers/net/virtio_net.c | 57 +++++++++++++++++++++++++++++++---------------- 1 file changed, 38 insertions(+), 19 deletions(-)...

Herbert tells me that returning NETDEV_TX_BUSY from hard_start_xmit is seen as a poor thing to do; we should cache the packet and stop the queue. Signed-off-by: Rusty Russell <rusty at rustcorp.com.au> --- drivers/net/virtio_net.c | 57 +++++++++++++++++++++++++++++++---------------- 1 file changed, 38 insertions(+), 19 deletions(-)...

...e). - * Since most packets only take 1 or 2 ring slots - * this means 16 slots are typically wasted. + /* If running out of space, stop queue to avoid getting packets that we + * are then unable to transmit. + * An alternative would be to force queuing layer to requeue the skb by + * returning NETDEV_TX_BUSY. However, NETDEV_TX_BUSY should not be + * returned in a normal path of operation: it means that driver is not + * maintaining the TX queue stop/start state properly, and causes + * the stack to do a non-trivial amount of useless work. + * Since most packets only take 1 or 2 ring slots, stoppin...

...e). - * Since most packets only take 1 or 2 ring slots - * this means 16 slots are typically wasted. + /* If running out of space, stop queue to avoid getting packets that we + * are then unable to transmit. + * An alternative would be to force queuing layer to requeue the skb by + * returning NETDEV_TX_BUSY. However, NETDEV_TX_BUSY should not be + * returned in a normal path of operation: it means that driver is not + * maintaining the TX queue stop/start state properly, and causes + * the stack to do a non-trivial amount of useless work. + * Since most packets only take 1 or 2 ring slots, stoppin...

...r_queue(vi, sq - vi->sq)) > >>> + check_sq_full_and_disable(vi, dev, sq); > >>> + > >> > >> Sorry if I missed something obvious here. > >> > >> As the comment in start_xmit(), the current skb is added to the sq->vq, so > >> NETDEV_TX_BUSY can not be returned. > >> > >> /* If running out of space, stop queue to avoid getting packets that we > >> * are then unable to transmit. > >> * An alternative would be to force queuing layer to requeue the skb by > >> * returning NETDEV_TX_BUSY....

Herbert tells me that returning NETDEV_TX_BUSY from hard_start_xmit is seen as a poor thing to do; we should cache the packet and stop the queue. Signed-off-by: Rusty Russell <rusty at rustcorp.com.au> --- drivers/net/virtio_net.c | 58 ++++++++++++++++++++++++++++++++--------------- 1 file changed, 40 insertions(+), 18 deletions(-)...

Herbert tells me that returning NETDEV_TX_BUSY from hard_start_xmit is seen as a poor thing to do; we should cache the packet and stop the queue. Signed-off-by: Rusty Russell <rusty at rustcorp.com.au> --- drivers/net/virtio_net.c | 58 ++++++++++++++++++++++++++++++++--------------- 1 file changed, 40 insertions(+), 18 deletions(-)...

The original comment was not really informative or funny as well as sexist. Replace it with a better explanation of why the driver does stop and what the impacts are. Signed-off-by: Stephen Hemminger <stephen at networkplumber.org> --- a/drivers/net/virtio_net.c 2015-03-24 15:20:25.174671000 -0700 +++ b/drivers/net/virtio_net.c 2015-03-24 16:17:28.478525333 -0700 @@ -939,8 +939,12 @@

The original comment was not really informative or funny as well as sexist. Replace it with a better explanation of why the driver does stop and what the impacts are. Signed-off-by: Stephen Hemminger <stephen at networkplumber.org> --- a/drivers/net/virtio_net.c 2015-03-24 15:20:25.174671000 -0700 +++ b/drivers/net/virtio_net.c 2015-03-24 16:17:28.478525333 -0700 @@ -939,8 +939,12 @@

If the queue of xdp xmit is not an independent queue, then when the xdp xmit used all the desc, the xmit from the __dev_queue_xmit() may encounter the following error. net ens4: Unexpected TXQ (0) queue failure: -28 This patch adds a check whether sq is full in XDP Xmit. Thanks. Xuan Zhuo (2): virtio_net: separate the logic of checking whether sq is full virtio_net: add checking sq is full

If the queue of xdp xmit is not an independent queue, then when the xdp xmit used all the desc, the xmit from the __dev_queue_xmit() may encounter the following error. net ens4: Unexpected TXQ (0) queue failure: -28 This patch adds a check whether sq is full in XDP Xmit. Thanks. v1: 1. rename to check_sq_full_and_disable 2. reorder some funcs to avoid declaration Xuan Zhuo (3):

...dev); - goto again; + if (net_ratelimit()) { + if (likely(capacity == -ENOMEM)) + dev_warn(&dev->dev, + "TX queue failure: out of memory\n"); + else + dev_warn(&dev->dev, + "Unexpected TX queue failure: %d\n", + capacity); } return NETDEV_TX_BUSY; } -- 1.7.1.12.g42b7f

...dev); - goto again; + if (net_ratelimit()) { + if (likely(capacity == -ENOMEM)) + dev_warn(&dev->dev, + "TX queue failure: out of memory\n"); + else + dev_warn(&dev->dev, + "Unexpected TX queue failure: %d\n", + capacity); } return NETDEV_TX_BUSY; } -- 1.7.1.12.g42b7f

...queueing new ones. */ free_old_xmit_skbs(vi); - /* Put new one in send queue and do transmit */ - __skb_queue_head(&vi->send, skb); + /* Try to transmit */ capacity = xmit_skb(vi, skb); /* This can happen with OOM and indirect buffers. */ @@ -532,8 +531,11 @@ again: } return NETDEV_TX_BUSY; } - vi->svq->vq_ops->kick(vi->svq); + + /* Put new one in send queue */ + __skb_queue_head(&vi->send, skb); + /* Don't wait up for transmitted skbs to be freed. */ skb_orphan(skb); nf_reset(skb); -- 1.6.5.rc2

...queueing new ones. */ free_old_xmit_skbs(vi); - /* Put new one in send queue and do transmit */ - __skb_queue_head(&vi->send, skb); + /* Try to transmit */ capacity = xmit_skb(vi, skb); /* This can happen with OOM and indirect buffers. */ @@ -532,8 +531,11 @@ again: } return NETDEV_TX_BUSY; } - vi->svq->vq_ops->kick(vi->svq); + + /* Put new one in send queue */ + __skb_queue_head(&vi->send, skb); + /* Don't wait up for transmitted skbs to be freed. */ skb_orphan(skb); nf_reset(skb); -- 1.6.5.rc2

...). DaveM points out that there are advantages to doing it generally (it's more likely to be on same CPU than after xmit), and I couldn't find any new starvation issues in simple benchmarking here. This patch adds skb_orphan to the start of dev_hard_start_xmit(): it can be premature in the NETDEV_TX_BUSY case, but that's uncommon. I removed the drivers' skb_orphan(), though it's harmless. Signed-off-by: Rusty Russell <rusty at rustcorp.com.au> Cc: Divy Le Ray <divy at chelsio.com> Cc: Roland Dreier <rolandd at cisco.com> Cc: Pavel Emelianov <xemul at openvz.org&gt...

...). DaveM points out that there are advantages to doing it generally (it's more likely to be on same CPU than after xmit), and I couldn't find any new starvation issues in simple benchmarking here. This patch adds skb_orphan to the start of dev_hard_start_xmit(): it can be premature in the NETDEV_TX_BUSY case, but that's uncommon. I removed the drivers' skb_orphan(), though it's harmless. Signed-off-by: Rusty Russell <rusty at rustcorp.com.au> Cc: Divy Le Ray <divy at chelsio.com> Cc: Roland Dreier <rolandd at cisco.com> Cc: Pavel Emelianov <xemul at openvz.org&gt...