2007-10-04 11:43:52

by Takenori Nagano

[permalink] [raw]
Subject: [PATCH 1/2] add tunable_notifier function

This patch adds new notifier function tunable_notifier_chain. Its base is
atomic_notifier_chain.

Thanks,

---

Signed-off-by: Takenori Nagano <[email protected]>

---
diff -uprN linux-2.6.23-rc9.orig/include/linux/notifier.h
linux-2.6.23-rc9/include/linux/notifier.h
--- linux-2.6.23-rc9.orig/include/linux/notifier.h 2007-10-02 12:24:52.000000000
+0900
+++ linux-2.6.23-rc9/include/linux/notifier.h 2007-10-03 14:48:04.288000000 +0900
@@ -13,6 +13,7 @@
#include <linux/mutex.h>
#include <linux/rwsem.h>
#include <linux/srcu.h>
+#include <linux/kobject.h>

/*
* Notifier chains are of four types:
@@ -53,6 +54,14 @@ struct notifier_block {
int priority;
};

+struct tunable_notifier_block {
+ struct notifier_block *nb;
+ struct tunable_notifier_head *head;
+ struct dentry *dir;
+ struct dentry *pri_dentry;
+ struct dentry *desc_dentry;
+};
+
struct atomic_notifier_head {
spinlock_t lock;
struct notifier_block *head;
@@ -63,6 +72,13 @@ struct blocking_notifier_head {
struct notifier_block *head;
};

+struct tunable_notifier_head {
+ spinlock_t lock;
+ struct notifier_block *head;
+ char *name;
+ struct dentry *dir;
+};
+
struct raw_notifier_head {
struct notifier_block *head;
};
@@ -81,6 +97,12 @@ struct srcu_notifier_head {
init_rwsem(&(name)->rwsem); \
(name)->head = NULL; \
} while (0)
+#define TUNABLE_INIT_NOTIFIER(val1, val2) do { \
+ spin_lock_init(&(val1)->lock); \
+ (val1)->head = NULL; \
+ (val1)->name = val2; \
+ (val1)->dir = NULL; \
+ } while (0)
#define RAW_INIT_NOTIFIER_HEAD(name) do { \
(name)->head = NULL; \
} while (0)
@@ -96,6 +118,11 @@ extern void srcu_init_notifier_head(stru
#define BLOCKING_NOTIFIER_INIT(name) { \
.rwsem = __RWSEM_INITIALIZER((name).rwsem), \
.head = NULL }
+#define TUNABLE_NOTIFIER_INIT(val1, val2) { \
+ .lock =__SPIN_LOCK_UNLOCKED(val1.lock), \
+ .head = NULL, \
+ .name = val2, \
+ .dir = NULL }
#define RAW_NOTIFIER_INIT(name) { \
.head = NULL }
/* srcu_notifier_heads cannot be initialized statically */
@@ -106,6 +133,9 @@ extern void srcu_init_notifier_head(stru
#define BLOCKING_NOTIFIER_HEAD(name) \
struct blocking_notifier_head name = \
BLOCKING_NOTIFIER_INIT(name)
+#define TUNABLE_NOTIFIER_HEAD(name, val) \
+ struct tunable_notifier_head name = \
+ TUNABLE_NOTIFIER_INIT(name, val)
#define RAW_NOTIFIER_HEAD(name) \
struct raw_notifier_head name = \
RAW_NOTIFIER_INIT(name)
@@ -116,6 +146,8 @@ extern int atomic_notifier_chain_registe
struct notifier_block *nb);
extern int blocking_notifier_chain_register(struct blocking_notifier_head *nh,
struct notifier_block *nb);
+extern int tunable_notifier_chain_register(struct tunable_notifier_head *nh,
+ struct tunable_notifier_block *nb, char *name, char *desc);
extern int raw_notifier_chain_register(struct raw_notifier_head *nh,
struct notifier_block *nb);
extern int srcu_notifier_chain_register(struct srcu_notifier_head *nh,
@@ -125,6 +157,8 @@ extern int atomic_notifier_chain_unregis
struct notifier_block *nb);
extern int blocking_notifier_chain_unregister(struct blocking_notifier_head *nh,
struct notifier_block *nb);
+extern int tunable_notifier_chain_unregister(struct tunable_notifier_head *nh,
+ struct tunable_notifier_block *nb);
extern int raw_notifier_chain_unregister(struct raw_notifier_head *nh,
struct notifier_block *nb);
extern int srcu_notifier_chain_unregister(struct srcu_notifier_head *nh,
@@ -138,6 +172,10 @@ extern int blocking_notifier_call_chain(
unsigned long val, void *v);
extern int __blocking_notifier_call_chain(struct blocking_notifier_head *nh,
unsigned long val, void *v, int nr_to_call, int *nr_calls);
+extern int tunable_notifier_call_chain(struct tunable_notifier_head *nh,
+ unsigned long val, void *v);
+extern int __tunable_notifier_call_chain(struct tunable_notifier_head *nh,
+ unsigned long val, void *v, int nr_to_call, int *nr_calls);
extern int raw_notifier_call_chain(struct raw_notifier_head *nh,
unsigned long val, void *v);
extern int __raw_notifier_call_chain(struct raw_notifier_head *nh,
diff -uprN linux-2.6.23-rc9.orig/kernel/sys.c linux-2.6.23-rc9/kernel/sys.c
--- linux-2.6.23-rc9.orig/kernel/sys.c 2007-10-02 12:24:52.000000000 +0900
+++ linux-2.6.23-rc9/kernel/sys.c 2007-10-03 14:48:15.160000000 +0900
@@ -38,6 +38,7 @@
#include <linux/syscalls.h>
#include <linux/kprobes.h>
#include <linux/user_namespace.h>
+#include <linux/debugfs.h>

#include <asm/uaccess.h>
#include <asm/io.h>
@@ -393,6 +394,234 @@ int blocking_notifier_call_chain(struct
EXPORT_SYMBOL_GPL(blocking_notifier_call_chain);

/*
+ * Tunable notifier chain routines. Registration and unregistration
+ * use a spinlock, and call_chain is synchronized by RCU (no locks).
+ * User can change the list order to use /sys/kernel/debug/list-name/.
+ */
+
+static ssize_t priority_read(struct file *file,
+ char __user *user_buf, size_t count, loff_t *ppos)
+{
+ struct tunable_notifier_block *n = file->f_dentry->d_inode->i_private;
+ int priority = n->nb->priority;
+ char buf[64], *s;
+
+ s = buf;
+ s += sprintf(s, "%d\n", priority);
+
+ return simple_read_from_buffer((void __user *)user_buf, count,
+ ppos, buf, s - buf);
+}
+
+static ssize_t priority_write(struct file *file,
+ const char __user *user_buf, size_t count, loff_t *ppos)
+{
+ struct tunable_notifier_block *n = file->f_dentry->d_inode->i_private;
+ struct tunable_notifier_head *nh = n->head;
+ char *buf, *end;
+ int ret = -ENOMEM, priority;
+ unsigned long tmp, flags;
+
+ buf = kmalloc(count + 1, GFP_KERNEL);
+ if (!buf)
+ goto out;
+
+ buf[count] = 0;
+ ret = -EFAULT;
+ if (copy_from_user(buf, user_buf, count))
+ goto out_free;
+
+ ret = -EINVAL;
+ tmp = simple_strtoul(buf, &end, 10);
+ if ((end == buf) || (tmp > INT_MAX))
+ goto out_free;
+
+ priority = (int)tmp;
+ n->nb->priority = priority;
+
+ spin_lock_irqsave(&nh->lock, flags);
+ ret = notifier_chain_unregister(&nh->head, n->nb);
+ if (ret)
+ goto out_unlock;
+ ret = notifier_chain_register(&nh->head, n->nb);
+ if (!ret)
+ ret = count;
+
+out_unlock:
+ spin_unlock_irqrestore(&nh->lock, flags);
+out_free:
+ kfree(buf);
+out:
+ return ret;
+
+}
+
+static const struct file_operations pri_fops = {
+ .read = priority_read,
+ .write = priority_write,
+};
+
+static ssize_t description_read(struct file *file,
+ char __user *user_buf, size_t count, loff_t *ppos)
+{
+ char *desc = file->f_dentry->d_inode->i_private;
+ int avail = strlen(desc);
+
+ return simple_read_from_buffer(user_buf, count, ppos, desc, avail);
+}
+
+static const struct file_operations desc_fops = {
+ .read = description_read,
+};
+
+/**
+ * tunable_notifier_chain_register - Add notifier to an tunable notifier chain
+ * @nh: Pointer to head of the tunable notifier chain
+ * @n: New entry in notifier chain
+ * @name: Pointer to the name of this notifier chain
+ * @desc: Pointer to the description of new entry
+ *
+ * Adds a notifier to an tunable notifier chain and makes control dir.
+ *
+ * Returns zero on success or %-ENODEV on failure.
+ */
+
+int tunable_notifier_chain_register(struct tunable_notifier_head *nh,
+ struct tunable_notifier_block *n, char *name, char *desc)
+{
+ unsigned long flags;
+ int ret = -EINVAL;
+ struct dentry *nh_dir, *nb_dir, *pri_dentry, *desc_dentry = NULL;
+
+ if (!name)
+ goto nb_fail;
+
+ ret = -ENOMEM;
+ if (!nh->dir) {
+ nh_dir = debugfs_create_dir(nh->name, NULL);
+ if (!nh_dir)
+ return ret;
+ nh->dir = nh_dir;
+ } else
+ nh_dir = nh->dir;
+
+ nb_dir = debugfs_create_dir(name, nh_dir);
+ if (!nb_dir)
+ goto nb_fail;
+ n->dir = nb_dir;
+
+ pri_dentry = debugfs_create_file("priority",0600, nb_dir, n, &pri_fops);
+ if (!pri_dentry)
+ goto pri_fail;
+ n->pri_dentry = pri_dentry;
+
+ if (desc) {
+ desc_dentry = debugfs_create_file("description", 0400, nb_dir,
+ desc, &desc_fops);
+ if (!desc_dentry)
+ goto desc_fail;
+ n->desc_dentry = desc_dentry;
+ }
+
+ spin_lock_irqsave(&nh->lock, flags);
+ ret = notifier_chain_register(&nh->head, n->nb);
+ spin_unlock_irqrestore(&nh->lock, flags);
+
+ if (ret)
+ goto reg_fail;
+
+ n->head = nh;
+
+ return ret;
+
+reg_fail:
+ debugfs_remove(desc_dentry);
+desc_fail:
+ debugfs_remove(pri_dentry);
+pri_fail:
+ debugfs_remove(nb_dir);
+nb_fail:
+ return ret;
+}
+
+EXPORT_SYMBOL_GPL(tunable_notifier_chain_register);
+
+/**
+ * tunable_notifier_chain_unregister - Remove notifier from a tunable notifier
chain
+ * @nh: Pointer to head of the tunable notifier chain
+ * @n: Entry to remove from notifier chain
+ *
+ * Removes a notifier from a tunable notifier chain.
+ *
+ * Retunrns zero on success or %-ENOENT on failure.
+ */
+
+int tunable_notifier_chain_unregister(struct tunable_notifier_head *nh,
+ struct tunable_notifier_block *n)
+{
+ unsigned long flags;
+ int ret;
+
+ spin_lock_irqsave(&nh->lock, flags);
+ ret = notifier_chain_unregister(&nh->head, n->nb);
+ spin_unlock_irqrestore(&nh->lock, flags);
+ synchronize_rcu();
+
+ if (ret)
+ return ret;
+
+ debugfs_remove(n->desc_dentry);
+ debugfs_remove(n->pri_dentry);
+ debugfs_remove(n->dir);
+
+ return 0;
+}
+
+EXPORT_SYMBOL_GPL(tunable_notifier_chain_unregister);
+
+/**
+ * __tunable_notifier_call_chain - Call functions in a tunable notifier chain
+ * @nh: Pointer to head of the tunable notifier chain
+ * @val: Value passed unmodified to notifier function
+ * @v: Pointer passed unmodified to notifier function
+ * @nt_to_call: See the comment for notifier_call_chain
+ * @nr_calls: See the comment for notifier_call_chain
+ *
+ * Calls each function in a notifier chain in turn. The functions
+ * run in an atomic context, so they must not block.
+ * This routine uses RCU to synchronize with changes to the chain.
+ *
+ * If the return value of the notifier can be and'ed
+ * with %NOTIFY_STOP_MASK then tunable_notifier_call_chain()
+ * will return immediately, with the return value of
+ * the notifier function which halted execution.
+ * Otherwise the return value is the return value
+ * of the last notifier function called.
+ */
+
+int __kprobes __tunable_notifier_call_chain(struct tunable_notifier_head *nh,
+ unsigned long val, void *v,
+ int nr_to_call, int *nr_calls)
+{
+ int ret;
+
+ rcu_read_lock();
+ ret = notifier_call_chain(&nh->head, val, v, nr_to_call, nr_calls);
+ rcu_read_unlock();
+ return ret;
+}
+
+EXPORT_SYMBOL_GPL(__tunable_notifier_call_chain);
+
+int __kprobes tunable_notifier_call_chain(struct tunable_notifier_head *nh,
+ unsigned long val, void *v)
+{
+ return __tunable_notifier_call_chain(nh, val, v, -1, NULL);
+}
+
+EXPORT_SYMBOL_GPL(tunable_notifier_call_chain);
+
+/*
* Raw notifier chain routines. There is no protection;
* the caller must provide it. Use at your own risk!
*/


2007-10-04 16:18:00

by Randy Dunlap

[permalink] [raw]
Subject: Re: [PATCH 1/2] add tunable_notifier function

On Thu, 04 Oct 2007 20:38:34 +0900 Takenori Nagano wrote:

> This patch adds new notifier function tunable_notifier_chain. Its base is
> atomic_notifier_chain.
>
> Thanks,
>
> ---
>
> Signed-off-by: Takenori Nagano <[email protected]>
>
> ---

> diff -uprN linux-2.6.23-rc9.orig/kernel/sys.c linux-2.6.23-rc9/kernel/sys.c
> --- linux-2.6.23-rc9.orig/kernel/sys.c 2007-10-02 12:24:52.000000000 +0900
> +++ linux-2.6.23-rc9/kernel/sys.c 2007-10-03 14:48:15.160000000 +0900
> @@ -38,6 +38,7 @@
> #include <linux/syscalls.h>
> #include <linux/kprobes.h>
> #include <linux/user_namespace.h>
> +#include <linux/debugfs.h>
>
> #include <asm/uaccess.h>
> #include <asm/io.h>
> @@ -393,6 +394,234 @@ int blocking_notifier_call_chain(struct

> +/**
> + * tunable_notifier_chain_register - Add notifier to an tunable notifier chain
> + * @nh: Pointer to head of the tunable notifier chain
> + * @n: New entry in notifier chain
> + * @name: Pointer to the name of this notifier chain

Is @name the name of a notifier chain or of the new notifier entry?


> + * @desc: Pointer to the description of new entry
> + *
> + * Adds a notifier to an tunable notifier chain and makes control dir.
> + *
> + * Returns zero on success or %-ENODEV on failure.
> + */
> +
> +int tunable_notifier_chain_register(struct tunable_notifier_head *nh,
> + struct tunable_notifier_block *n, char *name, char *desc)
> +{
> + unsigned long flags;
> + int ret = -EINVAL;
> + struct dentry *nh_dir, *nb_dir, *pri_dentry, *desc_dentry = NULL;
> +
> + if (!name)
> + goto nb_fail;
> +
> + ret = -ENOMEM;
> + if (!nh->dir) {
> + nh_dir = debugfs_create_dir(nh->name, NULL);
> + if (!nh_dir)
> + return ret;
> + nh->dir = nh_dir;
> + } else
> + nh_dir = nh->dir;
> +
> + nb_dir = debugfs_create_dir(name, nh_dir);
> + if (!nb_dir)
> + goto nb_fail;
> + n->dir = nb_dir;
> +
> + pri_dentry = debugfs_create_file("priority",0600, nb_dir, n, &pri_fops);
> + if (!pri_dentry)
> + goto pri_fail;
> + n->pri_dentry = pri_dentry;
> +
> + if (desc) {
> + desc_dentry = debugfs_create_file("description", 0400, nb_dir,
> + desc, &desc_fops);
> + if (!desc_dentry)
> + goto desc_fail;
> + n->desc_dentry = desc_dentry;
> + }
> +
> + spin_lock_irqsave(&nh->lock, flags);
> + ret = notifier_chain_register(&nh->head, n->nb);
> + spin_unlock_irqrestore(&nh->lock, flags);
> +
> + if (ret)
> + goto reg_fail;
> +
> + n->head = nh;
> +
> + return ret;
> +
> +reg_fail:
> + debugfs_remove(desc_dentry);
> +desc_fail:
> + debugfs_remove(pri_dentry);
> +pri_fail:
> + debugfs_remove(nb_dir);
> +nb_fail:
> + return ret;
> +}
> +
> +EXPORT_SYMBOL_GPL(tunable_notifier_chain_register);

---
~Randy

2007-10-05 05:08:45

by Takenori Nagano

[permalink] [raw]
Subject: Re: [PATCH 1/2] add tunable_notifier function

Randy Dunlap wrote:
> On Thu, 04 Oct 2007 20:38:34 +0900 Takenori Nagano wrote:
>> diff -uprN linux-2.6.23-rc9.orig/kernel/sys.c linux-2.6.23-rc9/kernel/sys.c
>> --- linux-2.6.23-rc9.orig/kernel/sys.c 2007-10-02 12:24:52.000000000 +0900
>> +++ linux-2.6.23-rc9/kernel/sys.c 2007-10-03 14:48:15.160000000 +0900
>> @@ -38,6 +38,7 @@
>> #include <linux/syscalls.h>
>> #include <linux/kprobes.h>
>> #include <linux/user_namespace.h>
>> +#include <linux/debugfs.h>
>>
>> #include <asm/uaccess.h>
>> #include <asm/io.h>
>> @@ -393,6 +394,234 @@ int blocking_notifier_call_chain(struct
>
>> +/**
>> + * tunable_notifier_chain_register - Add notifier to an tunable notifier chain
>> + * @nh: Pointer to head of the tunable notifier chain
>> + * @n: New entry in notifier chain
>> + * @name: Pointer to the name of this notifier chain
>
> Is @name the name of a notifier chain or of the new notifier entry?

Hi Randy,

@name: Pointer to the name of the new notifier entry.

I'll change the explanation.

Thanks,

2007-10-05 05:42:19

by Vivek Goyal

[permalink] [raw]
Subject: Re: [PATCH 1/2] add tunable_notifier function

On Thu, Oct 04, 2007 at 08:38:34PM +0900, Takenori Nagano wrote:
> This patch adds new notifier function tunable_notifier_chain. Its base is
> atomic_notifier_chain.
>
> Thanks,
>
> ---
>
> Signed-off-by: Takenori Nagano <[email protected]>
>
> ---
> diff -uprN linux-2.6.23-rc9.orig/include/linux/notifier.h
> linux-2.6.23-rc9/include/linux/notifier.h
> --- linux-2.6.23-rc9.orig/include/linux/notifier.h 2007-10-02 12:24:52.000000000
> +0900
> +++ linux-2.6.23-rc9/include/linux/notifier.h 2007-10-03 14:48:04.288000000 +0900
> @@ -13,6 +13,7 @@
> #include <linux/mutex.h>
> #include <linux/rwsem.h>
> #include <linux/srcu.h>
> +#include <linux/kobject.h>
>
> /*
> * Notifier chains are of four types:
> @@ -53,6 +54,14 @@ struct notifier_block {
> int priority;
> };
>
> +struct tunable_notifier_block {
> + struct notifier_block *nb;
> + struct tunable_notifier_head *head;
> + struct dentry *dir;
> + struct dentry *pri_dentry;
> + struct dentry *desc_dentry;
> +};
> +

Should this be tunable_atomic_notifier_block? I think there are two kind
of lists. One where handlers have to be atomic and other one where handlers
can be blocking one. I think you are making atomic one tunable. If that's
the case it should be reflected in the naming everywhere.

Thanks
Vivek

2007-10-05 05:52:00

by Takenori Nagano

[permalink] [raw]
Subject: Re: [PATCH 1/2] add tunable_notifier function

Vivek Goyal wrote:
> On Thu, Oct 04, 2007 at 08:38:34PM +0900, Takenori Nagano wrote:
>> This patch adds new notifier function tunable_notifier_chain. Its base is
>> atomic_notifier_chain.
>>
>> Thanks,
>>
>> ---
>>
>> Signed-off-by: Takenori Nagano <[email protected]>
>>
>> ---
>> diff -uprN linux-2.6.23-rc9.orig/include/linux/notifier.h
>> linux-2.6.23-rc9/include/linux/notifier.h
>> --- linux-2.6.23-rc9.orig/include/linux/notifier.h 2007-10-02 12:24:52.000000000
>> +0900
>> +++ linux-2.6.23-rc9/include/linux/notifier.h 2007-10-03 14:48:04.288000000 +0900
>> @@ -13,6 +13,7 @@
>> #include <linux/mutex.h>
>> #include <linux/rwsem.h>
>> #include <linux/srcu.h>
>> +#include <linux/kobject.h>
>>
>> /*
>> * Notifier chains are of four types:
>> @@ -53,6 +54,14 @@ struct notifier_block {
>> int priority;
>> };
>>
>> +struct tunable_notifier_block {
>> + struct notifier_block *nb;
>> + struct tunable_notifier_head *head;
>> + struct dentry *dir;
>> + struct dentry *pri_dentry;
>> + struct dentry *desc_dentry;
>> +};
>> +
>
> Should this be tunable_atomic_notifier_block? I think there are two kind
> of lists. One where handlers have to be atomic and other one where handlers
> can be blocking one. I think you are making atomic one tunable. If that's
> the case it should be reflected in the naming everywhere.

Hi Vivek,

Yes, it based on atomic_notifier_list. I think your opinion is reasonable.
I'll change the name tunable_notifier to tunable_atomic_notifier.

Thanks,

Takenori Nagano <[email protected]>

2007-10-05 12:54:52

by Eric W. Biederman

[permalink] [raw]
Subject: Re: [PATCH 1/2] add tunable_notifier function

Takenori Nagano <[email protected]> writes:

> This patch adds new notifier function tunable_notifier_chain. Its base is
> atomic_notifier_chain.



> +/**
> + * tunable_notifier_chain_register - Add notifier to an tunable notifier chain
> + * @nh: Pointer to head of the tunable notifier chain
> + * @n: New entry in notifier chain
> + * @name: Pointer to the name of this notifier chain
> + * @desc: Pointer to the description of new entry
> + *
> + * Adds a notifier to an tunable notifier chain and makes control dir.
> + *
> + * Returns zero on success or %-ENODEV on failure.
> + */
> +
> +int tunable_notifier_chain_register(struct tunable_notifier_head *nh,
> + struct tunable_notifier_block *n, char *name, char *desc)
> +{
> + unsigned long flags;
> + int ret = -EINVAL;
> + struct dentry *nh_dir, *nb_dir, *pri_dentry, *desc_dentry = NULL;
> +
> + if (!name)
> + goto nb_fail;
> +
> + ret = -ENOMEM;
> + if (!nh->dir) {
> + nh_dir = debugfs_create_dir(nh->name, NULL);

Hmm. debugfs does not appear to be an appropriate place to create
files that are supposed to be part of a serious user space interface.

Eric