lkml.org 
[lkml]   [2006]   [Mar]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH] deinline some larger functions from netdevice.h
Date
Hi Andrew,

Network folks did non comment on these two patches, let me try
submitting them to you instead.

I hunted down some large inlines. This patch address those found
in netdevice.h.

On a allyesconfig'ured kernel:

Size  Uses Wasted Name and definition
===== ==== ====== ================================================
   95  162  12075 netif_wake_queue      include/linux/netdevice.h
  129   86   9265 dev_kfree_skb_any     include/linux/netdevice.h
  127   56   5885 netif_device_attach   include/linux/netdevice.h
   73   86   4505 dev_kfree_skb_irq     include/linux/netdevice.h
   46   60   1534 netif_device_detach   include/linux/netdevice.h
  119   16   1485 __netif_rx_schedule   include/linux/netdevice.h
  143    5    492 netif_rx_schedule     include/linux/netdevice.h
   81    7    366 netif_schedule        include/linux/netdevice.h
netif_wake_queue is big because __netif_schedule is a big inline:

static inline void __netif_schedule(struct net_device *dev)
{
        if (!test_and_set_bit(__LINK_STATE_SCHED, &dev->state)) {
                unsigned long flags;
                struct softnet_data *sd;
                local_irq_save(flags);
                sd = &__get_cpu_var(softnet_data);
                dev->next_sched = sd->output_queue;
                sd->output_queue = dev;
                raise_softirq_irqoff(NET_TX_SOFTIRQ);
                local_irq_restore(flags);
        }
}
static inline void netif_wake_queue(struct net_device *dev)
{
#ifdef CONFIG_NETPOLL_TRAP
        if (netpoll_trap())
                return;
#endif
        if (test_and_clear_bit(__LINK_STATE_XOFF, &dev->state))
                __netif_schedule(dev);
}
By de-inlining __netif_schedule we are saving a lot of text
at each callsite of netif_wake_queue and netif_schedule.
__netif_rx_schedule is also big, and it makes more sense to keep
both of them out of line.

Patch also deinlines dev_kfree_skb_any. We can deinline dev_kfree_skb_irq
instead... oh well.

netif_device_attach/detach are not hot paths, we can deinline them too.

Signed-off-by: Denis Vlasenko <vda@ilport.com.ua>
--
vda
diff -urpN linux-2.6.16.org/include/linux/netdevice.h linux-2.6.16.deinline2/include/linux/netdevice.h
--- linux-2.6.16.org/include/linux/netdevice.h Mon Mar 20 07:53:29 2006
+++ linux-2.6.16.deinline2/include/linux/netdevice.h Mon Mar 27 13:46:15 2006
@@ -594,20 +594,7 @@ DECLARE_PER_CPU(struct softnet_data,soft

#define HAVE_NETIF_QUEUE

-static inline void __netif_schedule(struct net_device *dev)
-{
- if (!test_and_set_bit(__LINK_STATE_SCHED, &dev->state)) {
- unsigned long flags;
- struct softnet_data *sd;
-
- local_irq_save(flags);
- sd = &__get_cpu_var(softnet_data);
- dev->next_sched = sd->output_queue;
- sd->output_queue = dev;
- raise_softirq_irqoff(NET_TX_SOFTIRQ);
- local_irq_restore(flags);
- }
-}
+extern void __netif_schedule(struct net_device *dev);

static inline void netif_schedule(struct net_device *dev)
{
@@ -671,13 +658,7 @@ static inline void dev_kfree_skb_irq(str
/* Use this variant in places where it could be invoked
* either from interrupt or non-interrupt context.
*/
-static inline void dev_kfree_skb_any(struct sk_buff *skb)
-{
- if (in_irq() || irqs_disabled())
- dev_kfree_skb_irq(skb);
- else
- dev_kfree_skb(skb);
-}
+extern void dev_kfree_skb_any(struct sk_buff *skb);

#define HAVE_NETIF_RX 1
extern int netif_rx(struct sk_buff *skb);
@@ -735,22 +716,9 @@ static inline int netif_device_present(s
return test_bit(__LINK_STATE_PRESENT, &dev->state);
}

-static inline void netif_device_detach(struct net_device *dev)
-{
- if (test_and_clear_bit(__LINK_STATE_PRESENT, &dev->state) &&
- netif_running(dev)) {
- netif_stop_queue(dev);
- }
-}
+extern void netif_device_detach(struct net_device *dev);

-static inline void netif_device_attach(struct net_device *dev)
-{
- if (!test_and_set_bit(__LINK_STATE_PRESENT, &dev->state) &&
- netif_running(dev)) {
- netif_wake_queue(dev);
- __netdev_watchdog_up(dev);
- }
-}
+extern void netif_device_attach(struct net_device *dev);

/*
* Network interface message level settings
@@ -818,20 +786,7 @@ static inline int netif_rx_schedule_prep
* already been called and returned 1.
*/

-static inline void __netif_rx_schedule(struct net_device *dev)
-{
- unsigned long flags;
-
- local_irq_save(flags);
- dev_hold(dev);
- list_add_tail(&dev->poll_list, &__get_cpu_var(softnet_data).poll_list);
- if (dev->quota < 0)
- dev->quota += dev->weight;
- else
- dev->quota = dev->weight;
- __raise_softirq_irqoff(NET_RX_SOFTIRQ);
- local_irq_restore(flags);
-}
+extern void __netif_rx_schedule(struct net_device *dev);

/* Try to reschedule poll. Called by irq handler. */

diff -urpN linux-2.6.16.org/net/core/dev.c linux-2.6.16.deinline2/net/core/dev.c
--- linux-2.6.16.org/net/core/dev.c Mon Mar 20 07:53:29 2006
+++ linux-2.6.16.deinline2/net/core/dev.c Mon Mar 27 13:47:00 2006
@@ -1073,6 +1073,70 @@ void dev_queue_xmit_nit(struct sk_buff *
rcu_read_unlock();
}

+
+void __netif_schedule(struct net_device *dev)
+{
+ if (!test_and_set_bit(__LINK_STATE_SCHED, &dev->state)) {
+ unsigned long flags;
+ struct softnet_data *sd;
+
+ local_irq_save(flags);
+ sd = &__get_cpu_var(softnet_data);
+ dev->next_sched = sd->output_queue;
+ sd->output_queue = dev;
+ raise_softirq_irqoff(NET_TX_SOFTIRQ);
+ local_irq_restore(flags);
+ }
+}
+EXPORT_SYMBOL(__netif_schedule);
+
+void __netif_rx_schedule(struct net_device *dev)
+{
+ unsigned long flags;
+
+ local_irq_save(flags);
+ dev_hold(dev);
+ list_add_tail(&dev->poll_list, &__get_cpu_var(softnet_data).poll_list);
+ if (dev->quota < 0)
+ dev->quota += dev->weight;
+ else
+ dev->quota = dev->weight;
+ __raise_softirq_irqoff(NET_RX_SOFTIRQ);
+ local_irq_restore(flags);
+}
+EXPORT_SYMBOL(__netif_rx_schedule);
+
+void dev_kfree_skb_any(struct sk_buff *skb)
+{
+ if (in_irq() || irqs_disabled())
+ dev_kfree_skb_irq(skb);
+ else
+ dev_kfree_skb(skb);
+}
+EXPORT_SYMBOL(dev_kfree_skb_any);
+
+
+/* Hot-plugging. */
+void netif_device_detach(struct net_device *dev)
+{
+ if (test_and_clear_bit(__LINK_STATE_PRESENT, &dev->state) &&
+ netif_running(dev)) {
+ netif_stop_queue(dev);
+ }
+}
+EXPORT_SYMBOL(netif_device_detach);
+
+void netif_device_attach(struct net_device *dev)
+{
+ if (!test_and_set_bit(__LINK_STATE_PRESENT, &dev->state) &&
+ netif_running(dev)) {
+ netif_wake_queue(dev);
+ __netdev_watchdog_up(dev);
+ }
+}
+EXPORT_SYMBOL(netif_device_attach);
+
+
/*
* Invalidate hardware checksum when packet is to be mangled, and
* complete checksum manually on outgoing path.
\
 
 \ /
  Last update: 2006-03-30 09:25    [from the cache]
©2003-2014 Jasper Spaans. Advertise on this site