netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Eric Dumazet <eric.dumazet@gmail.com>
To: David Miller <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Subject: [PATCH] net: txq_trans_update() helper
Date: Tue, 26 May 2009 07:37:11 +0200	[thread overview]
Message-ID: <4A1B8007.1000703@gmail.com> (raw)
In-Reply-To: <20090518.151152.76033192.davem@davemloft.net>

David Miller a écrit :
> From: Eric Dumazet <dada1@cosmosbay.com>
> Date: Mon, 18 May 2009 09:21:31 +0200
> 
>> Indeed we can do this factorization, just have to take care of not
>> adding a new cache line miss on loopback device, as this one doesnt
>> touch dev->trans_start.
> 
> If you want to test IFF_LOOPBACK or similar, I can accept that.
> Or it can be some NETIF_F_* flag.
 

Or use txq->xmit_lock_owner != -1 test ?

Please note I had to change my email.

Thank you

[PATCH] net: txq_trans_update() helper

We would like to get rid of netdev->trans_start = jiffies; that about all net
drivers have to use in their start_xmit() function, and use txq->trans_start
instead.

This can be done generically in core network, as suggested by David.

Some devices, (particularly loopback) dont need trans_start update, because
they dont have transmit watchdog. We could add a new device flag, or rely
on fact that txq->tran_start can be updated is txq->xmit_lock_owner is
different than -1. Use a helper function to hide our choice.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
 include/linux/netdevice.h |    6 ++++++
 net/core/dev.c            |    3 +++
 net/core/netpoll.c        |    5 ++++-
 net/core/pktgen.c         |    1 +
 net/sched/sch_teql.c      |    1 +
 5 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index f8574e7..a6aa468 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1712,6 +1712,12 @@ static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
 	spin_unlock_bh(&txq->_xmit_lock);
 }
 
+static inline void txq_trans_update(struct netdev_queue *txq)
+{
+	if (txq->xmit_lock_owner != -1)
+		txq->trans_start = jiffies;
+}
+
 /**
  *	netif_tx_lock - grab network device transmit lock
  *	@dev: network device
diff --git a/net/core/dev.c b/net/core/dev.c
index 3942266..29fa4a1 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1698,6 +1698,8 @@ int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
 			skb->dst = NULL;
 		}
 		rc = ops->ndo_start_xmit(skb, dev);
+		if (rc == 0)
+			txq_trans_update(txq);
 		/*
 		 * TODO: if skb_orphan() was called by
 		 * dev->hard_start_xmit() (for example, the unmodified
@@ -1727,6 +1729,7 @@ gso:
 			skb->next = nskb;
 			return rc;
 		}
+		txq_trans_update(txq);
 		if (unlikely(netif_tx_queue_stopped(txq) && skb->next))
 			return NETDEV_TX_BUSY;
 	} while (skb->next);
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index 67b4f3e..35850e6 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -302,8 +302,11 @@ static void netpoll_send_skb(struct netpoll *np, struct sk_buff *skb)
 		for (tries = jiffies_to_usecs(1)/USEC_PER_POLL;
 		     tries > 0; --tries) {
 			if (__netif_tx_trylock(txq)) {
-				if (!netif_tx_queue_stopped(txq))
+				if (!netif_tx_queue_stopped(txq)) {
 					status = ops->ndo_start_xmit(skb, dev);
+					if (status == NETDEV_TX_OK)
+						txq_trans_update(txq);
+					}
 				__netif_tx_unlock(txq);
 
 				if (status == NETDEV_TX_OK)
diff --git a/net/core/pktgen.c b/net/core/pktgen.c
index 0666a82..b8ccd3c 100644
--- a/net/core/pktgen.c
+++ b/net/core/pktgen.c
@@ -3438,6 +3438,7 @@ static __inline__ void pktgen_xmit(struct pktgen_dev *pkt_dev)
 	      retry_now:
 		ret = (*xmit)(pkt_dev->skb, odev);
 		if (likely(ret == NETDEV_TX_OK)) {
+			txq_trans_update(txq);
 			pkt_dev->last_ok = 1;
 			pkt_dev->sofar++;
 			pkt_dev->seq_num++;
diff --git a/net/sched/sch_teql.c b/net/sched/sch_teql.c
index 428a5ef..a886496 100644
--- a/net/sched/sch_teql.c
+++ b/net/sched/sch_teql.c
@@ -308,6 +308,7 @@ restart:
 				if (!netif_tx_queue_stopped(slave_txq) &&
 				    !netif_tx_queue_frozen(slave_txq) &&
 				    slave_ops->ndo_start_xmit(skb, slave) == 0) {
+					txq_trans_update(slave_txq);
 					__netif_tx_unlock(slave_txq);
 					master->slaves = NEXT_SLAVE(q);
 					netif_wake_queue(dev);


  reply	other threads:[~2009-05-26  5:37 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-05-12 19:47 [PATCH] net: tx scalability works : trans_start Eric Dumazet
2009-05-18  3:57 ` David Miller
2009-05-18  7:21   ` Eric Dumazet
2009-05-18 22:11     ` David Miller
2009-05-26  5:37       ` Eric Dumazet [this message]
2009-05-26  5:38         ` [PATCH] net: txq_trans_update() helper David Miller
2009-05-26  5:57           ` Eric Dumazet
2009-05-26  5:58             ` David Miller

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4A1B8007.1000703@gmail.com \
    --to=eric.dumazet@gmail.com \
    --cc=davem@davemloft.net \
    --cc=netdev@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).