netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Eric Dumazet <dada1@cosmosbay.com>
To: David Miller <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Subject: [PATCH, resent] [NET] reduce per cpu ram used for loopback stats
Date: Wed, 18 Oct 2006 19:00:03 +0200	[thread overview]
Message-ID: <200610181900.03583.dada1@cosmosbay.com> (raw)
In-Reply-To: <200610181835.48744.dada1@cosmosbay.com>

[-- Attachment #1: Type: text/plain, Size: 403 bytes --]

Sorry David, the previous attachment had a missing / in one filename

[NET] reduce per cpu ram used for loopback device stats

We dont need a full struct net_device_stats (currently 23 long : 184 bytes on 
x86_64) per possible CPU, but only two counters : bytes and packets

We save few CPU cycles too in loopback_xmit() not updating 4 fields, but 2. 

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>

[-- Attachment #2: loopback.patch --]
[-- Type: text/plain, Size: 1894 bytes --]

--- linux/drivers/net/loopback.c	2006-10-18 17:28:20.000000000 +0200
+++ linux-ed/drivers/net/loopback.c	2006-10-18 18:26:41.000000000 +0200
@@ -58,7 +58,11 @@
 #include <linux/tcp.h>
 #include <linux/percpu.h>
 
-static DEFINE_PER_CPU(struct net_device_stats, loopback_stats);
+struct pcpu_lstats {
+	unsigned long packets;
+	unsigned long bytes;
+};
+static DEFINE_PER_CPU(struct pcpu_lstats, pcpu_lstats);
 
 #define LOOPBACK_OVERHEAD (128 + MAX_HEADER + 16 + 16)
 
@@ -128,7 +132,7 @@
  */
 static int loopback_xmit(struct sk_buff *skb, struct net_device *dev)
 {
-	struct net_device_stats *lb_stats;
+	struct pcpu_lstats *lb_stats;
 
 	skb_orphan(skb);
 
@@ -149,11 +153,9 @@
 #endif
 	dev->last_rx = jiffies;
 
-	lb_stats = &per_cpu(loopback_stats, get_cpu());
-	lb_stats->rx_bytes += skb->len;
-	lb_stats->tx_bytes = lb_stats->rx_bytes;
-	lb_stats->rx_packets++;
-	lb_stats->tx_packets = lb_stats->rx_packets;
+	lb_stats = &per_cpu(pcpu_lstats, get_cpu());
+	lb_stats->bytes += skb->len;
+	lb_stats->packets++;
 	put_cpu();
 
 	netif_rx(skb);
@@ -166,20 +168,21 @@
 static struct net_device_stats *get_stats(struct net_device *dev)
 {
 	struct net_device_stats *stats = &loopback_stats;
+	unsigned long bytes = 0;
+	unsigned long packets = 0;
 	int i;
 
-	memset(stats, 0, sizeof(struct net_device_stats));
-
 	for_each_possible_cpu(i) {
-		struct net_device_stats *lb_stats;
+		const struct pcpu_lstats *lb_stats;
 
-		lb_stats = &per_cpu(loopback_stats, i);
-		stats->rx_bytes   += lb_stats->rx_bytes;
-		stats->tx_bytes   += lb_stats->tx_bytes;
-		stats->rx_packets += lb_stats->rx_packets;
-		stats->tx_packets += lb_stats->tx_packets;
+		lb_stats = &per_cpu(pcpu_lstats, i);
+		bytes   += lb_stats->bytes;
+		packets += lb_stats->packets;
 	}
-
+	stats->rx_packets = packets;
+	stats->tx_packets = packets;
+	stats->rx_bytes = bytes;
+	stats->tx_bytes = bytes;
 	return stats;
 }
 

  reply	other threads:[~2006-10-18 17:00 UTC|newest]

Thread overview: 32+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-10-17  0:53 [PATCH] Bound TSO defer time (resend) John Heffner
2006-10-17  3:20 ` Stephen Hemminger
2006-10-17  4:18   ` John Heffner
2006-10-17  5:35     ` David Miller
2006-10-17 12:22       ` John Heffner
2006-10-19  3:39         ` David Miller
2006-10-17 12:58       ` [PATCH] [NET] Size listen hash tables using backlog hint Eric Dumazet Hi
2006-10-18  7:38         ` [PATCH] [NET] inet_peer : group together avl_left, avl_right, v4daddr to speedup lookups on some CPUS Eric Dumazet
2006-10-18 16:35           ` [PATCH] [NET] reduce per cpu ram used for loopback stats Eric Dumazet
2006-10-18 17:00             ` Eric Dumazet [this message]
2006-10-19  3:53               ` [PATCH, resent] " David Miller
2006-10-19  3:53             ` [PATCH] " David Miller
2006-10-19  3:44           ` [PATCH] [NET] inet_peer : group together avl_left, avl_right, v4daddr to speedup lookups on some CPUS David Miller
2006-10-19 10:57           ` Eric Dumazet
2006-10-19 15:45             ` [PATCH] [NET] One NET_INC_STATS() could be NET_INC_STATS_BH in tcp_v4_err() Eric Dumazet
2006-10-20  7:22               ` David Miller
2006-10-20 14:21                 ` Arnaldo Carvalho de Melo
2006-10-20  7:28             ` [PATCH] [NET] inet_peer : group together avl_left, avl_right, v4daddr to speedup lookups on some CPUS David Miller
2006-10-19  3:31         ` [PATCH] [NET] Size listen hash tables using backlog hint David Miller
2006-10-19  4:54           ` Stephen Hemminger
2006-10-19  5:08             ` David Miller
2006-10-19  5:12           ` Eric Dumazet
2006-10-19  6:12             ` David Miller
2006-10-19  6:34               ` Eric Dumazet
2006-10-19  6:57                 ` David Miller
2006-10-19  8:29                   ` Eric Dumazet
2006-10-19  8:41                     ` David Miller
2006-10-19  9:11                       ` Eric Dumazet
2006-10-19  9:27         ` Eric Dumazet
2006-10-20  7:27           ` David Miller
2006-10-18 15:37     ` [PATCH] Bound TSO defer time (resend) Andi Kleen
2006-10-18 16:40       ` Stephen Hemminger

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=200610181900.03583.dada1@cosmosbay.com \
    --to=dada1@cosmosbay.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).