All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Michael S. Tsirkin" <mst@redhat.com>
To: linux-kernel@vger.kernel.org
Cc: Rusty Russell <rusty@rustcorp.com.au>,
	Carsten Otte <cotte@de.ibm.com>,
	Christian Borntraeger <borntraeger@de.ibm.com>,
	linux390@de.ibm.com, Martin Schwidefsky <schwidefsky@de.ibm.com>,
	Heiko Carstens <heiko.carstens@de.ibm.com>,
	Shirley Ma <xma@us.ibm.com>,
	lguest@lists.ozlabs.org, linux-kernel@vger.kernel.org,
	virtualization@lists.linux-foundation.org,
	netdev@vger.kernel.org, linux-s390@vger.kernel.org,
	kvm@vger.kernel.org, Krishna Kumar <krkumar2@in.ibm.com>,
	Tom Lendacky <tahm@linux.vnet.ibm.com>,
	steved@us.ibm.com, habanero@linux.vnet.ibm.com
Subject: [PATCH 18/18] virtio_net: limit xmit polling
Date: Wed, 4 May 2011 23:53:13 +0300	[thread overview]
Message-ID: <5964e2f3d6aac5cd48f467848eed6570517470ef.1304541919.git.mst@redhat.com> (raw)
In-Reply-To: <cover.1304541918.git.mst@redhat.com>

Current code might introduce a lot of latency variation
if there are many pending bufs at the time we
attempt to transmit a new one. This is bad for
real-time applications and can't be good for TCP either.

Free up just enough to both clean up all buffers
eventually and to be able to xmit the next packet.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
 drivers/net/virtio_net.c |   18 +++++++++++-------
 1 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index f33c92b..9982bd7 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -509,17 +509,23 @@ again:
 	return received;
 }
 
-static void free_old_xmit_skbs(struct virtnet_info *vi)
+static bool free_old_xmit_skbs(struct virtnet_info *vi, int capacity)
 {
 	struct sk_buff *skb;
 	unsigned int len;
+	bool c;
+	/* We try to free up at least 2 skbs per one sent, so that we'll get
+	 * all of the memory back if they are used fast enough. */
+	int n = 2;
 
-	while ((skb = virtqueue_get_buf(vi->svq, &len)) != NULL) {
+	while ((c = virtqueue_get_capacity(vi->svq) >= capacity) && --n > 0 &&
+	       (skb = virtqueue_get_buf(vi->svq, &len)) != NULL) {
 		pr_debug("Sent skb %p\n", skb);
 		vi->dev->stats.tx_bytes += skb->len;
 		vi->dev->stats.tx_packets++;
 		dev_kfree_skb_any(skb);
 	}
+	return c;
 }
 
 static int xmit_skb(struct virtnet_info *vi, struct sk_buff *skb)
@@ -574,8 +580,8 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
 	struct virtnet_info *vi = netdev_priv(dev);
 	int capacity;
 
-	/* Free up any pending old buffers before queueing new ones. */
-	free_old_xmit_skbs(vi);
+	/* Free enough pending old buffers to enable queueing new ones. */
+	free_old_xmit_skbs(vi, 2+MAX_SKB_FRAGS);
 
 	/* Try to transmit */
 	capacity = xmit_skb(vi, skb);
@@ -609,9 +615,7 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
 		netif_stop_queue(dev);
 		if (unlikely(!virtqueue_enable_cb_delayed(vi->svq))) {
 			/* More just got used, free them then recheck. */
-			free_old_xmit_skbs(vi);
-			capacity = virtqueue_get_capacity(vi->svq);
-			if (capacity >= 2+MAX_SKB_FRAGS) {
+			if (!likely(free_old_xmit_skbs(vi, 2+MAX_SKB_FRAGS))) {
 				netif_start_queue(dev);
 				virtqueue_disable_cb(vi->svq);
 			}
-- 
1.7.5.53.gc233e

WARNING: multiple messages have this Message-ID (diff)
From: "Michael S. Tsirkin" <mst@redhat.com>
To: linux-kernel@vger.kernel.org
Cc: Krishna Kumar <krkumar2@in.ibm.com>,
	Carsten Otte <cotte@de.ibm.com>,
	lguest@lists.ozlabs.org, Shirley Ma <xma@us.ibm.com>,
	kvm@vger.kernel.org, linux-s390@vger.kernel.org,
	netdev@vger.kernel.org, habanero@linux.vnet.ibm.com,
	Heiko Carstens <heiko.carstens@de.ibm.com>,
	linux-kernel@vger.kernel.org,
	virtualization@lists.linux-foundation.org, steved@us.ibm.com,
	Christian Borntraeger <borntraeger@de.ibm.com>,
	Tom Lendacky <tahm@linux.vnet.ibm.com>,
	Martin Schwidefsky <schwidefsky@de.ibm.com>,
	linux390@de.ibm.com
Subject: [PATCH 18/18] virtio_net: limit xmit polling
Date: Wed, 4 May 2011 23:53:13 +0300	[thread overview]
Message-ID: <5964e2f3d6aac5cd48f467848eed6570517470ef.1304541919.git.mst@redhat.com> (raw)
In-Reply-To: <cover.1304541918.git.mst@redhat.com>

Current code might introduce a lot of latency variation
if there are many pending bufs at the time we
attempt to transmit a new one. This is bad for
real-time applications and can't be good for TCP either.

Free up just enough to both clean up all buffers
eventually and to be able to xmit the next packet.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
 drivers/net/virtio_net.c |   18 +++++++++++-------
 1 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index f33c92b..9982bd7 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -509,17 +509,23 @@ again:
 	return received;
 }
 
-static void free_old_xmit_skbs(struct virtnet_info *vi)
+static bool free_old_xmit_skbs(struct virtnet_info *vi, int capacity)
 {
 	struct sk_buff *skb;
 	unsigned int len;
+	bool c;
+	/* We try to free up at least 2 skbs per one sent, so that we'll get
+	 * all of the memory back if they are used fast enough. */
+	int n = 2;
 
-	while ((skb = virtqueue_get_buf(vi->svq, &len)) != NULL) {
+	while ((c = virtqueue_get_capacity(vi->svq) >= capacity) && --n > 0 &&
+	       (skb = virtqueue_get_buf(vi->svq, &len)) != NULL) {
 		pr_debug("Sent skb %p\n", skb);
 		vi->dev->stats.tx_bytes += skb->len;
 		vi->dev->stats.tx_packets++;
 		dev_kfree_skb_any(skb);
 	}
+	return c;
 }
 
 static int xmit_skb(struct virtnet_info *vi, struct sk_buff *skb)
@@ -574,8 +580,8 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
 	struct virtnet_info *vi = netdev_priv(dev);
 	int capacity;
 
-	/* Free up any pending old buffers before queueing new ones. */
-	free_old_xmit_skbs(vi);
+	/* Free enough pending old buffers to enable queueing new ones. */
+	free_old_xmit_skbs(vi, 2+MAX_SKB_FRAGS);
 
 	/* Try to transmit */
 	capacity = xmit_skb(vi, skb);
@@ -609,9 +615,7 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
 		netif_stop_queue(dev);
 		if (unlikely(!virtqueue_enable_cb_delayed(vi->svq))) {
 			/* More just got used, free them then recheck. */
-			free_old_xmit_skbs(vi);
-			capacity = virtqueue_get_capacity(vi->svq);
-			if (capacity >= 2+MAX_SKB_FRAGS) {
+			if (!likely(free_old_xmit_skbs(vi, 2+MAX_SKB_FRAGS))) {
 				netif_start_queue(dev);
 				virtqueue_disable_cb(vi->svq);
 			}
-- 
1.7.5.53.gc233e

WARNING: multiple messages have this Message-ID (diff)
From: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Krishna Kumar <krkumar2@in.ibm.com>,
	Carsten Otte <cotte@de.ibm.com>,
	lguest@lists.ozlabs.org, Shirley Ma <xma@us.ibm.com>,
	kvm@vger.kernel.org, linux-s390@vger.kernel.org,
	netdev@vger.kernel.org, habanero@linux.vnet.ibm.com,
	Heiko Carstens <heiko.carstens@de.ibm.com>,
	linux-kernel@vger.kernel.org,
	virtualization@lists.linux-foundation.org, steved@us.ibm.com,
	Christian Borntraeger <borntraeger@de.ibm.com>,
	Tom Lendacky <tahm@linux.vnet.ibm.com>,
	Martin Schwidefsky <schwidefsky@de.ibm.com>,
	linux390@de.ibm.com
Subject: [PATCH 18/18] virtio_net: limit xmit polling
Date: Wed, 4 May 2011 23:53:13 +0300	[thread overview]
Message-ID: <5964e2f3d6aac5cd48f467848eed6570517470ef.1304541919.git.mst@redhat.com> (raw)
In-Reply-To: <cover.1304541918.git.mst@redhat.com>

Current code might introduce a lot of latency variation
if there are many pending bufs at the time we
attempt to transmit a new one. This is bad for
real-time applications and can't be good for TCP either.

Free up just enough to both clean up all buffers
eventually and to be able to xmit the next packet.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
 drivers/net/virtio_net.c |   18 +++++++++++-------
 1 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index f33c92b..9982bd7 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -509,17 +509,23 @@ again:
 	return received;
 }
 
-static void free_old_xmit_skbs(struct virtnet_info *vi)
+static bool free_old_xmit_skbs(struct virtnet_info *vi, int capacity)
 {
 	struct sk_buff *skb;
 	unsigned int len;
+	bool c;
+	/* We try to free up at least 2 skbs per one sent, so that we'll get
+	 * all of the memory back if they are used fast enough. */
+	int n = 2;
 
-	while ((skb = virtqueue_get_buf(vi->svq, &len)) != NULL) {
+	while ((c = virtqueue_get_capacity(vi->svq) >= capacity) && --n > 0 &&
+	       (skb = virtqueue_get_buf(vi->svq, &len)) != NULL) {
 		pr_debug("Sent skb %p\n", skb);
 		vi->dev->stats.tx_bytes += skb->len;
 		vi->dev->stats.tx_packets++;
 		dev_kfree_skb_any(skb);
 	}
+	return c;
 }
 
 static int xmit_skb(struct virtnet_info *vi, struct sk_buff *skb)
@@ -574,8 +580,8 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
 	struct virtnet_info *vi = netdev_priv(dev);
 	int capacity;
 
-	/* Free up any pending old buffers before queueing new ones. */
-	free_old_xmit_skbs(vi);
+	/* Free enough pending old buffers to enable queueing new ones. */
+	free_old_xmit_skbs(vi, 2+MAX_SKB_FRAGS);
 
 	/* Try to transmit */
 	capacity = xmit_skb(vi, skb);
@@ -609,9 +615,7 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
 		netif_stop_queue(dev);
 		if (unlikely(!virtqueue_enable_cb_delayed(vi->svq))) {
 			/* More just got used, free them then recheck. */
-			free_old_xmit_skbs(vi);
-			capacity = virtqueue_get_capacity(vi->svq);
-			if (capacity >= 2+MAX_SKB_FRAGS) {
+			if (!likely(free_old_xmit_skbs(vi, 2+MAX_SKB_FRAGS))) {
 				netif_start_queue(dev);
 				virtqueue_disable_cb(vi->svq);
 			}
-- 
1.7.5.53.gc233e

  parent reply	other threads:[~2011-05-04 20:53 UTC|newest]

Thread overview: 144+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-05-04 20:50 [PATCH 00/18] virtio and vhost-net performance enhancements Michael S. Tsirkin
2011-05-04 20:50 ` Michael S. Tsirkin
2011-05-04 20:50 ` [PATCH 01/18] virtio: 64 bit features Michael S. Tsirkin
2011-05-04 20:50 ` Michael S. Tsirkin
2011-05-04 20:50   ` Michael S. Tsirkin
2011-05-04 20:50 ` [PATCH 02/18] virtio_test: update for " Michael S. Tsirkin
2011-05-04 20:50   ` Michael S. Tsirkin
2011-05-04 20:50   ` Michael S. Tsirkin
2011-05-04 20:50 ` [PATCH 03/18] vhost: fix " Michael S. Tsirkin
2011-05-04 20:50   ` Michael S. Tsirkin
2011-05-04 20:50   ` Michael S. Tsirkin
2011-05-04 20:51 ` [PATCH 04/18] virtio: don't delay avail index update Michael S. Tsirkin
2011-05-04 20:51 ` Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 20:51 ` [PATCH 05/18] virtio: used event index interface Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 21:56   ` Tom Lendacky
2011-05-04 21:56   ` Tom Lendacky
2011-05-05  9:38     ` Michael S. Tsirkin
2011-05-05  9:38     ` Michael S. Tsirkin
2011-05-04 20:51 ` [PATCH 06/18] virtio_ring: avail " Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-09  4:13   ` Rusty Russell
2011-05-09  4:13   ` Rusty Russell
2011-05-09  4:13     ` Rusty Russell
2011-05-15 12:47     ` Michael S. Tsirkin
2011-05-15 12:47     ` Michael S. Tsirkin
2011-05-15 12:47       ` Michael S. Tsirkin
2011-05-16  6:23       ` Rusty Russell
2011-05-16  6:23       ` Rusty Russell
2011-05-16  6:23         ` Rusty Russell
2011-05-17  6:00         ` Michael S. Tsirkin
2011-05-17  6:00         ` Michael S. Tsirkin
2011-05-17  6:00           ` Michael S. Tsirkin
2011-05-18  0:08           ` Rusty Russell
2011-05-18  0:08           ` Rusty Russell
2011-05-18  0:08             ` Rusty Russell
2011-05-04 20:51 ` [PATCH 07/18] virtio ring: inline function to check for events Michael S. Tsirkin
2011-05-04 20:51 ` Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-05  8:34   ` Stefan Hajnoczi
2011-05-05  8:56     ` Michael S. Tsirkin
2011-05-05  8:56     ` Michael S. Tsirkin
2011-05-05  8:34   ` Stefan Hajnoczi
2011-05-04 20:51 ` [PATCH 08/18] virtio_ring: support for used_event idx feature Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-09  4:17   ` Rusty Russell
2011-05-09  4:17   ` Rusty Russell
2011-05-09  4:17     ` Rusty Russell
2011-05-15 12:47     ` Michael S. Tsirkin
2011-05-15 12:47     ` Michael S. Tsirkin
2011-05-15 12:47       ` Michael S. Tsirkin
2011-05-04 20:51 ` [PATCH 09/18] virtio: use avail_event index Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 21:58   ` Tom Lendacky
2011-05-04 21:58   ` Tom Lendacky
2011-05-05  9:34     ` Michael S. Tsirkin
2011-05-05  9:34     ` Michael S. Tsirkin
2011-05-09  4:33   ` Rusty Russell
2011-05-09  4:33   ` Rusty Russell
2011-05-09  4:33     ` Rusty Russell
2011-05-15 13:55     ` Michael S. Tsirkin
2011-05-15 13:55     ` Michael S. Tsirkin
2011-05-15 13:55       ` Michael S. Tsirkin
2011-05-16  7:12       ` Rusty Russell
2011-05-16  7:12       ` Rusty Russell
2011-05-16  7:12         ` Rusty Russell
2011-05-17  6:10         ` Michael S. Tsirkin
2011-05-17  6:10           ` Michael S. Tsirkin
2011-05-18  0:19           ` Rusty Russell
2011-05-18  0:19           ` Rusty Russell
2011-05-18  0:19             ` Rusty Russell
2011-05-18  5:43             ` Michael S. Tsirkin
2011-05-18  5:43             ` Michael S. Tsirkin
2011-05-18  5:43               ` Michael S. Tsirkin
2011-05-19  7:27             ` Michael S. Tsirkin
2011-05-19  7:27             ` Michael S. Tsirkin
2011-05-19  7:27               ` Michael S. Tsirkin
2011-05-17  6:10         ` Michael S. Tsirkin
2011-05-17 16:23         ` Tom Lendacky
2011-05-17 16:23         ` Tom Lendacky
2011-05-04 20:51 ` [PATCH 10/18] vhost: utilize used_event index Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 20:51   ` Michael S. Tsirkin
2011-05-04 20:52 ` [PATCH 11/18] vhost: support avail_event idx Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:52 ` [PATCH 12/18] virtio_test: support used_event index Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:52 ` [PATCH 13/18] virtio_test: avail_event index support Michael S. Tsirkin
2011-05-04 20:52 ` Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:52 ` [PATCH 14/18] virtio: add api for delayed callbacks Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-09  5:57   ` Rusty Russell
2011-05-09  5:57   ` Rusty Russell
2011-05-09  5:57     ` Rusty Russell
2011-05-15 12:48     ` Michael S. Tsirkin
2011-05-15 12:48     ` Michael S. Tsirkin
2011-05-15 12:48       ` Michael S. Tsirkin
2011-05-16  7:13       ` Rusty Russell
2011-05-16  7:13       ` Rusty Russell
2011-05-16  7:13         ` Rusty Russell
2011-05-19  7:24         ` Michael S. Tsirkin
2011-05-19  7:24           ` Michael S. Tsirkin
2011-05-20  7:43           ` Rusty Russell
2011-05-20  7:43           ` Rusty Russell
2011-05-20  7:43             ` Rusty Russell
2011-05-19  7:24         ` Michael S. Tsirkin
2011-05-04 20:52 ` [PATCH 15/18] virtio_net: delay TX callbacks Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:52 ` Michael S. Tsirkin
2011-05-04 20:52 ` [PATCH 16/18] virtio_ring: Add capacity check API Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:52   ` Michael S. Tsirkin
2011-05-04 20:53 ` [PATCH 17/18] virtio_net: fix TX capacity checks using new API Michael S. Tsirkin
2011-05-04 20:53   ` Michael S. Tsirkin
2011-05-04 20:53   ` Michael S. Tsirkin
2011-05-04 20:53 ` Michael S. Tsirkin [this message]
2011-05-04 20:53   ` [PATCH 18/18] virtio_net: limit xmit polling Michael S. Tsirkin
2011-05-04 20:53   ` Michael S. Tsirkin
2011-05-05 15:07 ` [PATCH 0/3] virtio and vhost-net performance enhancements Michael S. Tsirkin
2011-05-05 15:07   ` Michael S. Tsirkin
2011-05-05 15:08   ` [PATCH 1/3] virtio: fix avail event support Michael S. Tsirkin
2011-05-05 15:08   ` Michael S. Tsirkin
2011-05-05 15:08     ` Michael S. Tsirkin
2011-05-05 15:08   ` [PATCH 2/3] virtio_ring: check used_event offset Michael S. Tsirkin
2011-05-05 15:08     ` Michael S. Tsirkin
2011-05-05 15:08   ` Michael S. Tsirkin
2011-05-05 15:08   ` [PATCH 3/3] virtio_ring: need_event api comment fix Michael S. Tsirkin
2011-05-05 15:08   ` Michael S. Tsirkin
2011-05-05 15:08     ` Michael S. Tsirkin
2011-05-09  5:59     ` Rusty Russell
2011-05-09  5:59     ` Rusty Russell
2011-05-09  5:59       ` Rusty Russell
2011-05-05 15:07 ` [PATCH 0/3] virtio and vhost-net performance enhancements Michael S. Tsirkin
2011-05-11 17:10 ` [PATCH 00/18] " Krishna Kumar2
2011-05-11 17:10 ` Krishna Kumar2

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=5964e2f3d6aac5cd48f467848eed6570517470ef.1304541919.git.mst@redhat.com \
    --to=mst@redhat.com \
    --cc=borntraeger@de.ibm.com \
    --cc=cotte@de.ibm.com \
    --cc=habanero@linux.vnet.ibm.com \
    --cc=heiko.carstens@de.ibm.com \
    --cc=krkumar2@in.ibm.com \
    --cc=kvm@vger.kernel.org \
    --cc=lguest@lists.ozlabs.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-s390@vger.kernel.org \
    --cc=linux390@de.ibm.com \
    --cc=netdev@vger.kernel.org \
    --cc=rusty@rustcorp.com.au \
    --cc=schwidefsky@de.ibm.com \
    --cc=steved@us.ibm.com \
    --cc=tahm@linux.vnet.ibm.com \
    --cc=virtualization@lists.linux-foundation.org \
    --cc=xma@us.ibm.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.