Skip to contentSkip to conversation

Move some code into __init section when CONFIG_NET_NS=n

PATCH Applied Pavel Emelyanov 3 participants 5 messages Oct 4, 2007 – Oct 8, 2007

Subsystems (3)
Networking Network drivers Proc filesystem
lore
Share & tools
lore

Pavel Emelyanov

author
#0 lore raw

PATCH 1/1 Move some code into __init section when CONFIG_NET_NS=n

With the net namespaces many code leaved the __init section, thus making the kernel occupy more memory than it did before. Since we have a config option that prohibits the namespace creation, the functions that initialize/finalize some netns stuff are simply not needed and can be freed after the boot.

Currently, this is almost not noticeable, since few calls are no longer in __init, but when the namespaces will be merged it will be possible to free more code. I propose to use the __net_init, __net_exit and __net_initdata "attributes" for functions/variables that are not used if the CONFIG_NET_NS is not set to save more space in memory.

Signed-off-by Pavel Emelyanov <[email protected]>

---

include/net/net_namespace.h +9 -0
index 934c840..747170f 100644
@@ -99,6 +99,15 @@ static inline void release_net(struct ne
 #define for_each_net(VAR)				\
 	list_for_each_entry(VAR, &net_namespace_list, list)
 
+#ifdef CONFIG_NET_NS
+#define __net_init
+#define __net_exit
+#define __net_initdata
+#else
+#define __net_init	__init
+#define __net_exit	__exit
+#define __net_initdata	__initdata
+#endif
 
 struct pernet_operations {
 	struct list_head list;
drivers/net/loopback.c +3 -3
index d6997ae..be25aa3 100644
@@ -250,7 +250,7 @@ static void loopback_setup(struct net_de
 }
 
 /* Setup and register the loopback device. */
-static int loopback_net_init(struct net *net)
+static __net_init int loopback_net_init(struct net *net)
 {
 	struct net_device *dev;
 	int err;

@@ -278,14 +278,14 @@ out_free_netdev:
 	goto out;
 }
 
-static void loopback_net_exit(struct net *net)
+static __net_exit void loopback_net_exit(struct net *net)
 {
 	struct net_device *dev = net->loopback_dev;
 
 	unregister_netdev(dev);
 }
 
-static struct pernet_operations loopback_net_ops = {
+static struct pernet_operations __net_initdata loopback_net_ops = {
        .init = loopback_net_init,
        .exit = loopback_net_exit,
 };
fs/proc/proc_net.c +4 -4
index 85cc8e8..2e91fb7 100644
@@ -140,7 +140,7 @@ static struct inode_operations proc_net_
 	.setattr	= proc_net_setattr,
 };
 
-static int proc_net_ns_init(struct net *net)
+static __net_init int proc_net_ns_init(struct net *net)
 {
 	struct proc_dir_entry *root, *netd, *net_statd;
 	int err;

@@ -178,19 +178,19 @@ free_root:
 	goto out;
 }
 
-static void proc_net_ns_exit(struct net *net)
+static __net_exit void proc_net_ns_exit(struct net *net)
 {
 	remove_proc_entry("stat", net->proc_net);
 	remove_proc_entry("net", net->proc_net_root);
 	kfree(net->proc_net_root);
 }
 
-struct pernet_operations proc_net_ns_ops = {
+struct pernet_operations __net_initdata proc_net_ns_ops = {
 	.init = proc_net_ns_init,
 	.exit = proc_net_ns_exit,
 };
 
-int proc_net_init(void)
+int __init proc_net_init(void)
 {
 	proc_net_shadow = proc_mkdir("net", NULL);
 	proc_net_shadow->proc_iops = &proc_net_dir_inode_operations;
net/core/dev.c +8 -8
index d998646..37f8858 100644
@@ -2611,7 +2611,7 @@ static const struct file_operations ptyp
 };
 
 
-static int dev_proc_net_init(struct net *net)
+static int __net_init dev_proc_net_init(struct net *net)
 {
 	int rc = -ENOMEM;
 

@@ -2636,7 +2636,7 @@ out_dev:
 	goto out;
 }
 
-static void dev_proc_net_exit(struct net *net)
+static void __net_exit dev_proc_net_exit(struct net *net)
 {
 	wext_proc_exit(net);
 

@@ -2645,7 +2645,7 @@ static void dev_proc_net_exit(struct net
 	proc_net_remove(net, "dev");
 }
 
-static struct pernet_operations dev_proc_ops = {
+static struct pernet_operations __net_initdata dev_proc_ops = {
 	.init = dev_proc_net_init,
 	.exit = dev_proc_net_exit,
 };

@@ -4250,7 +4250,7 @@ static struct hlist_head *netdev_create_
 }
 
 /* Initialize per network namespace state */
-static int netdev_init(struct net *net)
+static int __net_init netdev_init(struct net *net)
 {
 	INIT_LIST_HEAD(&net->dev_base_head);
 	rwlock_init(&dev_base_lock);

@@ -4271,18 +4271,18 @@ err_name:
 	return -ENOMEM;
 }
 
-static void netdev_exit(struct net *net)
+static void __net_exit netdev_exit(struct net *net)
 {
 	kfree(net->dev_name_head);
 	kfree(net->dev_index_head);
 }
 
-static struct pernet_operations netdev_net_ops = {
+static struct pernet_operations __net_initdata netdev_net_ops = {
 	.init = netdev_init,
 	.exit = netdev_exit,
 };
 
-static void default_device_exit(struct net *net)
+static void __net_exit default_device_exit(struct net *net)
 {
 	struct net_device *dev, *next;
 	/*

@@ -4308,7 +4308,7 @@ static void default_device_exit(struct n
 	rtnl_unlock();
 }
 
-static struct pernet_operations default_device_ops = {
+static struct pernet_operations __net_initdata default_device_ops = {
 	.exit = default_device_exit,
 };
net/core/dev_mcast.c +3 -3
index 896b0ca..15241cf 100644
@@ -273,19 +273,19 @@ static const struct file_operations dev_
 
 #endif
 
-static int dev_mc_net_init(struct net *net)
+static int __net_init dev_mc_net_init(struct net *net)
 {
 	if (!proc_net_fops_create(net, "dev_mcast", 0, &dev_mc_seq_fops))
 		return -ENOMEM;
 	return 0;
 }
 
-static void dev_mc_net_exit(struct net *net)
+static void __net_exit dev_mc_net_exit(struct net *net)
 {
 	proc_net_remove(net, "dev_mcast");
 }
 
-static struct pernet_operations dev_mc_net_ops = {
+static struct pernet_operations __net_initdata dev_mc_net_ops = {
 	.init = dev_mc_net_init,
 	.exit = dev_mc_net_exit,
 };
net/netlink/af_netlink.c +3 -3
index 46eb5ea..3ef3282 100644
@@ -1924,7 +1924,7 @@ static struct net_proto_family netlink_f
 	.owner	= THIS_MODULE,	/* for consistency 8) */
 };
 
-static int netlink_net_init(struct net *net)
+static int __net_init netlink_net_init(struct net *net)
 {
 #ifdef CONFIG_PROC_FS
 	if (!proc_net_fops_create(net, "netlink", 0, &netlink_seq_fops))

@@ -1933,14 +1933,14 @@ static int netlink_net_init(struct net *
 	return 0;
 }
 
-static void netlink_net_exit(struct net *net)
+static void __net_exit netlink_net_exit(struct net *net)
 {
 #ifdef CONFIG_PROC_FS
 	proc_net_remove(net, "netlink");
 #endif
 }
 
-static struct pernet_operations netlink_net_ops = {
+static struct pernet_operations __net_initdata netlink_net_ops = {
 	.init = netlink_net_init,
 	.exit = netlink_net_exit,
 };
#1 lore raw
Quoted 13 lines · Pavel Emelyanov

With the net namespaces many code leaved the __init section, thus making the kernel occupy more memory than it did before. Since we have a config option that prohibits the namespace creation, the functions that initialize/finalize some netns stuff are simply not needed and can be freed after the boot.

Currently, this is almost not noticeable, since few calls are no longer in __init, but when the namespaces will be merged it will be possible to free more code. I propose to use the __net_init, __net_exit and __net_initdata "attributes" for functions/variables that are not used if the CONFIG_NET_NS is not set to save more space in memory.

include/net/net_namespace.h static inline void release_net(struct ne · line 110

Quoted 9 lines
include/net/net_namespace.h +9 -0
+#ifdef CONFIG_NET_NS
+#define __net_init
+#define __net_exit
+#define __net_initdata
+#else
+#define __net_init	__init
+#define __net_exit	__exit
+#define __net_initdata	__initdata
+#endif

Yet another set of double-underscored section annotations is the last thing that is needed, methinks. :)

maintainer
#2 lore raw

From: Alexey Dobriyan <[email protected]>
Date: Thu, 4 Oct 2007 18:02:02 +0400

Quoted 14 lines · Pavel Emelyanov
Quoted 12 lines

With the net namespaces many code leaved the __init section, thus making the kernel occupy more memory than it did before. Since we have a config option that prohibits the namespace creation, the functions that initialize/finalize some netns stuff are simply not needed and can be freed after the boot.

Currently, this is almost not noticeable, since few calls are no longer in __init, but when the namespaces will be merged it will be possible to free more code. I propose to use the __net_init, __net_exit and __net_initdata "attributes" for functions/variables that are not used if the CONFIG_NET_NS is not set to save more space in memory.

Quoted 12 lines

include/net/net_namespace.h static inline void release_net(struct ne · line 110

Quoted 9 lines
include/net/net_namespace.h +9 -0
+#ifdef CONFIG_NET_NS
+#define __net_init
+#define __net_exit
+#define __net_initdata
+#else
+#define __net_init	__init
+#define __net_exit	__exit
+#define __net_initdata	__initdata
+#endif

Yet another set of double-underscored section annotations is the last thing that is needed, methinks. :)

I'm not so sure. I think double-underscores are a powerful deterrant for developers. Over time folks have learned that this prefix in an interface name means "semantics are non-trivial, use with care" and that definitely applies to init sections.

I'm going to apply Pavel's patch, if you want to touch it up do so as a followon patch, thanks!

maintainer
#3 lore raw

From: David Miller <[email protected]>
Date: Mon, 08 Oct 2007 00:17:16 -0700 (PDT)

Quoted 2 lines

I'm going to apply Pavel's patch, if you want to touch it up do so as a followon patch, thanks!

Actually, I take that back, it doesn't even build, I'm therefore removing the patch:

LD .tmp_vmlinux1
.exit.text' referenced in section .init.data' of fs/built-in.o: defined in discarded section .exit.text' of fs/built-in.o .exit.text' referenced in section .init.data' of drivers/built-in.o: defined in discarded section .exit.text' of drivers/built-in.o .exit.text' referenced in section .init.data' of net/built-in.o: defined in discarded section .exit.text' of net/built-in.o .exit.text' referenced in section .init.data' of net/built-in.o: defined in discarded section .exit.text' of net/built-in.o .exit.text' referenced in section .init.data' of net/built-in.o: defined in discarded section .exit.text' of net/built-in.o .exit.text' referenced in section .init.data' of net/built-in.o: defined in discarded section .exit.text' of net/built-in.o .exit.text' referenced in section .init.data' of net/built-in.o: defined in discarded section `.exit.text' of net/built-in.o
make: *** [.tmp_vmlinux1] Error 1

Please fix this up and resubmit if you like.

maintainer Applied
#4 lore raw
In reply to Pavel Emelyanov

From: Pavel Emelyanov <[email protected]>
Date: Thu, 04 Oct 2007 17:54:11 +0400

Quoted 6 lines

With the net namespaces many code leaved the __init section, thus making the kernel occupy more memory than it did before. Since we have a config option that prohibits the namespace creation, the functions that initialize/finalize some netns stuff are simply not needed and can be freed after the boot.

Quoted 8 lines

Currently, this is almost not noticeable, since few calls are no longer in __init, but when the namespaces will be merged it will be possible to free more code. I propose to use the __net_init, __net_exit and __net_initdata "attributes" for functions/variables that are not used if the CONFIG_NET_NS is not set to save more space in memory.

Signed-off-by Pavel Emelyanov <[email protected]>

Applied, thanks Pavel.