linux/fs/sysfs/file.c

759 lines
19 KiB
C
Raw Normal View History

/*
* fs/sysfs/file.c - sysfs regular (text) file implementation
*
* Copyright (c) 2001-3 Patrick Mochel
* Copyright (c) 2007 SUSE Linux Products GmbH
* Copyright (c) 2007 Tejun Heo <teheo@suse.de>
*
* This file is released under the GPLv2.
*
* Please see Documentation/filesystems/sysfs.txt for more information.
*/
#include <linux/module.h>
#include <linux/kobject.h>
#include <linux/kallsyms.h>
#include <linux/slab.h>
#include <linux/fsnotify.h>
#include <linux/namei.h>
#include <linux/poll.h>
#include <linux/list.h>
#include <linux/mutex.h>
#include <linux/limits.h>
#include <asm/uaccess.h>
#include "sysfs.h"
/* used in crash dumps to help with debugging */
static char last_sysfs_file[PATH_MAX];
void sysfs_printk_last_file(void)
{
printk(KERN_EMERG "last sysfs file: %s\n", last_sysfs_file);
}
/*
* There's one sysfs_buffer for each open file and one
* sysfs_open_dirent for each sysfs_dirent with one or more open
* files.
*
* filp->private_data points to sysfs_buffer and
* sysfs_dirent->s_attr.open points to sysfs_open_dirent. s_attr.open
* is protected by sysfs_open_dirent_lock.
*/
static DEFINE_SPINLOCK(sysfs_open_dirent_lock);
struct sysfs_open_dirent {
atomic_t refcnt;
atomic_t event;
wait_queue_head_t poll;
struct list_head buffers; /* goes through sysfs_buffer.list */
};
struct sysfs_buffer {
size_t count;
loff_t pos;
char * page;
const struct sysfs_ops * ops;
struct mutex mutex;
int needs_read_fill;
int event;
struct list_head list;
};
/**
* fill_read_buffer - allocate and fill buffer from object.
* @dentry: dentry pointer.
* @buffer: data buffer for file.
*
* Allocate @buffer->page, if it hasn't been already, then call the
* kobject's show() method to fill the buffer with this attribute's
* data.
* This is called only once, on the file's first read unless an error
* is returned.
*/
static int fill_read_buffer(struct dentry * dentry, struct sysfs_buffer * buffer)
{
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
struct sysfs_dirent *attr_sd = dentry->d_fsdata;
struct kobject *kobj = attr_sd->s_parent->s_dir.kobj;
const struct sysfs_ops * ops = buffer->ops;
int ret = 0;
ssize_t count;
if (!buffer->page)
buffer->page = (char *) get_zeroed_page(GFP_KERNEL);
if (!buffer->page)
return -ENOMEM;
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
/* need attr_sd for attr and ops, its parent for kobj */
if (!sysfs_get_active(attr_sd))
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
return -ENODEV;
buffer->event = atomic_read(&attr_sd->s_attr.open->event);
count = ops->show(kobj, attr_sd->s_attr.attr, buffer->page);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
sysfs_put_active(attr_sd);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
/*
* The code works fine with PAGE_SIZE return but it's likely to
* indicate truncated result or overflow in normal use cases.
*/
if (count >= (ssize_t)PAGE_SIZE) {
print_symbol("fill_read_buffer: %s returned bad count\n",
(unsigned long)ops->show);
/* Try to struggle along */
count = PAGE_SIZE - 1;
}
if (count >= 0) {
buffer->needs_read_fill = 0;
buffer->count = count;
} else {
ret = count;
}
return ret;
}
/**
* sysfs_read_file - read an attribute.
* @file: file pointer.
* @buf: buffer to fill.
* @count: number of bytes to read.
* @ppos: starting offset in file.
*
* Userspace wants to read an attribute file. The attribute descriptor
* is in the file's ->d_fsdata. The target object is in the directory's
* ->d_fsdata.
*
* We call fill_read_buffer() to allocate and fill the buffer from the
* object's show() method exactly once (if the read is happening from
* the beginning of the file). That should fill the entire buffer with
* all the data the object has to offer for that attribute.
* We then call flush_read_buffer() to copy the buffer to userspace
* in the increments specified.
*/
static ssize_t
sysfs_read_file(struct file *file, char __user *buf, size_t count, loff_t *ppos)
{
struct sysfs_buffer * buffer = file->private_data;
ssize_t retval = 0;
mutex_lock(&buffer->mutex);
if (buffer->needs_read_fill || *ppos == 0) {
retval = fill_read_buffer(file->f_path.dentry,buffer);
if (retval)
goto out;
}
pr_debug("%s: count = %zd, ppos = %lld, buf = %s\n",
__func__, count, *ppos, buffer->page);
retval = simple_read_from_buffer(buf, count, ppos, buffer->page,
buffer->count);
out:
mutex_unlock(&buffer->mutex);
return retval;
}
/**
* fill_write_buffer - copy buffer from userspace.
* @buffer: data buffer for file.
* @buf: data from user.
* @count: number of bytes in @userbuf.
*
* Allocate @buffer->page if it hasn't been already, then
* copy the user-supplied buffer into it.
*/
static int
fill_write_buffer(struct sysfs_buffer * buffer, const char __user * buf, size_t count)
{
int error;
if (!buffer->page)
buffer->page = (char *)get_zeroed_page(GFP_KERNEL);
if (!buffer->page)
return -ENOMEM;
if (count >= PAGE_SIZE)
count = PAGE_SIZE - 1;
error = copy_from_user(buffer->page,buf,count);
buffer->needs_read_fill = 1;
/* if buf is assumed to contain a string, terminate it by \0,
so e.g. sscanf() can scan the string easily */
buffer->page[count] = 0;
return error ? -EFAULT : count;
}
/**
* flush_write_buffer - push buffer to kobject.
* @dentry: dentry to the attribute
* @buffer: data buffer for file.
* @count: number of bytes
*
* Get the correct pointers for the kobject and the attribute we're
* dealing with, then call the store() method for the attribute,
* passing the buffer that we acquired in fill_write_buffer().
*/
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
static int
flush_write_buffer(struct dentry * dentry, struct sysfs_buffer * buffer, size_t count)
{
struct sysfs_dirent *attr_sd = dentry->d_fsdata;
struct kobject *kobj = attr_sd->s_parent->s_dir.kobj;
const struct sysfs_ops * ops = buffer->ops;
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
int rc;
/* need attr_sd for attr and ops, its parent for kobj */
if (!sysfs_get_active(attr_sd))
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
return -ENODEV;
rc = ops->store(kobj, attr_sd->s_attr.attr, buffer->page, count);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
sysfs_put_active(attr_sd);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
return rc;
}
/**
* sysfs_write_file - write an attribute.
* @file: file pointer
* @buf: data to write
* @count: number of bytes
* @ppos: starting offset
*
* Similar to sysfs_read_file(), though working in the opposite direction.
* We allocate and fill the data from the user in fill_write_buffer(),
* then push it to the kobject in flush_write_buffer().
* There is no easy way for us to know if userspace is only doing a partial
* write, so we don't support them. We expect the entire buffer to come
* on the first write.
* Hint: if you're writing a value, first read the file, modify only the
* the value you're changing, then write entire buffer back.
*/
static ssize_t
sysfs_write_file(struct file *file, const char __user *buf, size_t count, loff_t *ppos)
{
struct sysfs_buffer * buffer = file->private_data;
ssize_t len;
mutex_lock(&buffer->mutex);
len = fill_write_buffer(buffer, buf, count);
if (len > 0)
len = flush_write_buffer(file->f_path.dentry, buffer, len);
if (len > 0)
*ppos += len;
mutex_unlock(&buffer->mutex);
return len;
}
/**
* sysfs_get_open_dirent - get or create sysfs_open_dirent
* @sd: target sysfs_dirent
* @buffer: sysfs_buffer for this instance of open
*
* If @sd->s_attr.open exists, increment its reference count;
* otherwise, create one. @buffer is chained to the buffers
* list.
*
* LOCKING:
* Kernel thread context (may sleep).
*
* RETURNS:
* 0 on success, -errno on failure.
*/
static int sysfs_get_open_dirent(struct sysfs_dirent *sd,
struct sysfs_buffer *buffer)
{
struct sysfs_open_dirent *od, *new_od = NULL;
retry:
spin_lock_irq(&sysfs_open_dirent_lock);
if (!sd->s_attr.open && new_od) {
sd->s_attr.open = new_od;
new_od = NULL;
}
od = sd->s_attr.open;
if (od) {
atomic_inc(&od->refcnt);
list_add_tail(&buffer->list, &od->buffers);
}
spin_unlock_irq(&sysfs_open_dirent_lock);
if (od) {
kfree(new_od);
return 0;
}
/* not there, initialize a new one and retry */
new_od = kmalloc(sizeof(*new_od), GFP_KERNEL);
if (!new_od)
return -ENOMEM;
atomic_set(&new_od->refcnt, 0);
atomic_set(&new_od->event, 1);
init_waitqueue_head(&new_od->poll);
INIT_LIST_HEAD(&new_od->buffers);
goto retry;
}
/**
* sysfs_put_open_dirent - put sysfs_open_dirent
* @sd: target sysfs_dirent
* @buffer: associated sysfs_buffer
*
* Put @sd->s_attr.open and unlink @buffer from the buffers list.
* If reference count reaches zero, disassociate and free it.
*
* LOCKING:
* None.
*/
static void sysfs_put_open_dirent(struct sysfs_dirent *sd,
struct sysfs_buffer *buffer)
{
struct sysfs_open_dirent *od = sd->s_attr.open;
unsigned long flags;
spin_lock_irqsave(&sysfs_open_dirent_lock, flags);
list_del(&buffer->list);
if (atomic_dec_and_test(&od->refcnt))
sd->s_attr.open = NULL;
else
od = NULL;
spin_unlock_irqrestore(&sysfs_open_dirent_lock, flags);
kfree(od);
}
static int sysfs_open_file(struct inode *inode, struct file *file)
{
struct sysfs_dirent *attr_sd = file->f_path.dentry->d_fsdata;
struct kobject *kobj = attr_sd->s_parent->s_dir.kobj;
struct sysfs_buffer *buffer;
const struct sysfs_ops *ops;
int error = -EACCES;
char *p;
p = d_path(&file->f_path, last_sysfs_file, sizeof(last_sysfs_file));
if (p)
memmove(last_sysfs_file, p, strlen(p) + 1);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
/* need attr_sd for attr and ops, its parent for kobj */
if (!sysfs_get_active(attr_sd))
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
return -ENODEV;
/* every kobject with an attribute needs a ktype assigned */
if (kobj->ktype && kobj->ktype->sysfs_ops)
ops = kobj->ktype->sysfs_ops;
else {
WARN(1, KERN_ERR "missing sysfs attribute operations for "
"kobject: %s\n", kobject_name(kobj));
goto err_out;
}
/* File needs write support.
* The inode's perms must say it's ok,
* and we must have a store method.
*/
if (file->f_mode & FMODE_WRITE) {
if (!(inode->i_mode & S_IWUGO) || !ops->store)
goto err_out;
}
/* File needs read support.
* The inode's perms must say it's ok, and we there
* must be a show method for it.
*/
if (file->f_mode & FMODE_READ) {
if (!(inode->i_mode & S_IRUGO) || !ops->show)
goto err_out;
}
/* No error? Great, allocate a buffer for the file, and store it
* it in file->private_data for easy access.
*/
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
error = -ENOMEM;
buffer = kzalloc(sizeof(struct sysfs_buffer), GFP_KERNEL);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
if (!buffer)
goto err_out;
mutex_init(&buffer->mutex);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
buffer->needs_read_fill = 1;
buffer->ops = ops;
file->private_data = buffer;
/* make sure we have open dirent struct */
error = sysfs_get_open_dirent(attr_sd, buffer);
if (error)
goto err_free;
/* open succeeded, put active references */
sysfs_put_active(attr_sd);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
return 0;
err_free:
kfree(buffer);
err_out:
sysfs_put_active(attr_sd);
return error;
}
static int sysfs_release(struct inode *inode, struct file *filp)
{
struct sysfs_dirent *sd = filp->f_path.dentry->d_fsdata;
struct sysfs_buffer *buffer = filp->private_data;
sysfs_put_open_dirent(sd, buffer);
if (buffer->page)
free_page((unsigned long)buffer->page);
kfree(buffer);
return 0;
}
/* Sysfs attribute files are pollable. The idea is that you read
* the content and then you use 'poll' or 'select' to wait for
* the content to change. When the content changes (assuming the
* manager for the kobject supports notification), poll will
* return POLLERR|POLLPRI, and select will return the fd whether
* it is waiting for read, write, or exceptions.
* Once poll/select indicates that the value has changed, you
* need to close and re-open the file, or seek to 0 and read again.
* Reminder: this only works for attributes which actively support
* it, and it is not possible to test an attribute from userspace
* to see if it supports poll (Neither 'poll' nor 'select' return
* an appropriate error code). When in doubt, set a suitable timeout value.
*/
static unsigned int sysfs_poll(struct file *filp, poll_table *wait)
{
struct sysfs_buffer * buffer = filp->private_data;
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
struct sysfs_dirent *attr_sd = filp->f_path.dentry->d_fsdata;
struct sysfs_open_dirent *od = attr_sd->s_attr.open;
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
/* need parent for the kobj, grab both */
if (!sysfs_get_active(attr_sd))
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
goto trigger;
poll_wait(filp, &od->poll, wait);
sysfs_put_active(attr_sd);
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
if (buffer->event != atomic_read(&od->event))
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
goto trigger;
return DEFAULT_POLLMASK;
sysfs: implement sysfs_dirent active reference and immediate disconnect sysfs: implement sysfs_dirent active reference and immediate disconnect Opening a sysfs node references its associated kobject, so userland can arbitrarily prolong lifetime of a kobject which complicates lifetime rules in drivers. This patch implements active reference and makes the association between kobject and sysfs immediately breakable. Now each sysfs_dirent has two reference counts - s_count and s_active. s_count is a regular reference count which guarantees that the containing sysfs_dirent is accessible. As long as s_count reference is held, all sysfs internal fields in sysfs_dirent are accessible including s_parent and s_name. The newly added s_active is active reference count. This is acquired by invoking sysfs_get_active() and it's the caller's responsibility to ensure sysfs_dirent itself is accessible (should be holding s_count one way or the other). Dereferencing sysfs_dirent to access objects out of sysfs proper requires active reference. This includes access to the associated kobjects, attributes and ops. The active references can be drained and denied by calling sysfs_deactivate(). All active sysfs_dirents must be deactivated after deletion but before the default reference is dropped. This enables immediate disconnect of sysfs nodes. Once a sysfs_dirent is deleted, it won't access any entity external to sysfs proper. Because attr/bin_attr ops access both the node itself and its parent for kobject, they need to hold active references to both. sysfs_get/put_active_two() helpers are provided to help grabbing both references. Parent's is acquired first and released last. Unlike other operations, mmapped area lingers on after mmap() is finished and the module implement implementing it and kobj need to stay referenced till all the mapped pages are gone. This is accomplished by holding one set of active references to the bin_attr and its parent if there have been any mmap during lifetime of an openfile. The references are dropped when the openfile is released. This change makes sysfs lifetime rules independent from both kobject's and module's. It not only fixes several race conditions caused by sysfs not holding onto the proper module when referencing kobject, but also helps fixing and simplifying lifetime management in driver model and drivers by taking sysfs out of the equation. Please read the following message for more info. http://article.gmane.org/gmane.linux.kernel/510293 Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 18:45:16 +00:00
trigger:
buffer->needs_read_fill = 1;
return DEFAULT_POLLMASK|POLLERR|POLLPRI;
}
void sysfs_notify_dirent(struct sysfs_dirent *sd)
{
struct sysfs_open_dirent *od;
unsigned long flags;
spin_lock_irqsave(&sysfs_open_dirent_lock, flags);
od = sd->s_attr.open;
if (od) {
atomic_inc(&od->event);
wake_up_interruptible(&od->poll);
}
spin_unlock_irqrestore(&sysfs_open_dirent_lock, flags);
}
EXPORT_SYMBOL_GPL(sysfs_notify_dirent);
void sysfs_notify(struct kobject *k, const char *dir, const char *attr)
{
struct sysfs_dirent *sd = k->sd;
mutex_lock(&sysfs_mutex);
if (sd && dir)
sysfs: Implement sysfs tagged directory support. The problem. When implementing a network namespace I need to be able to have multiple network devices with the same name. Currently this is a problem for /sys/class/net/*, /sys/devices/virtual/net/*, and potentially a few other directories of the form /sys/ ... /net/*. What this patch does is to add an additional tag field to the sysfs dirent structure. For directories that should show different contents depending on the context such as /sys/class/net/, and /sys/devices/virtual/net/ this tag field is used to specify the context in which those directories should be visible. Effectively this is the same as creating multiple distinct directories with the same name but internally to sysfs the result is nicer. I am calling the concept of a single directory that looks like multiple directories all at the same path in the filesystem tagged directories. For the networking namespace the set of directories whose contents I need to filter with tags can depend on the presence or absence of hotplug hardware or which modules are currently loaded. Which means I need a simple race free way to setup those directories as tagged. To achieve a reace free design all tagged directories are created and managed by sysfs itself. Users of this interface: - define a type in the sysfs_tag_type enumeration. - call sysfs_register_ns_types with the type and it's operations - sysfs_exit_ns when an individual tag is no longer valid - Implement mount_ns() which returns the ns of the calling process so we can attach it to a sysfs superblock. - Implement ktype.namespace() which returns the ns of a syfs kobject. Everything else is left up to sysfs and the driver layer. For the network namespace mount_ns and namespace() are essentially one line functions, and look to remain that. Tags are currently represented a const void * pointers as that is both generic, prevides enough information for equality comparisons, and is trivial to create for current users, as it is just the existing namespace pointer. The work needed in sysfs is more extensive. At each directory or symlink creating I need to check if the directory it is being created in is a tagged directory and if so generate the appropriate tag to place on the sysfs_dirent. Likewise at each symlink or directory removal I need to check if the sysfs directory it is being removed from is a tagged directory and if so figure out which tag goes along with the name I am deleting. Currently only directories which hold kobjects, and symlinks are supported. There is not enough information in the current file attribute interfaces to give us anything to discriminate on which makes it useless, and there are no potential users which makes it an uninteresting problem to solve. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-30 18:31:26 +00:00
/* Only directories are tagged, so no need to pass
* a tag explicitly.
*/
sd = sysfs_find_dirent(sd, NULL, dir);
if (sd && attr)
sysfs: Implement sysfs tagged directory support. The problem. When implementing a network namespace I need to be able to have multiple network devices with the same name. Currently this is a problem for /sys/class/net/*, /sys/devices/virtual/net/*, and potentially a few other directories of the form /sys/ ... /net/*. What this patch does is to add an additional tag field to the sysfs dirent structure. For directories that should show different contents depending on the context such as /sys/class/net/, and /sys/devices/virtual/net/ this tag field is used to specify the context in which those directories should be visible. Effectively this is the same as creating multiple distinct directories with the same name but internally to sysfs the result is nicer. I am calling the concept of a single directory that looks like multiple directories all at the same path in the filesystem tagged directories. For the networking namespace the set of directories whose contents I need to filter with tags can depend on the presence or absence of hotplug hardware or which modules are currently loaded. Which means I need a simple race free way to setup those directories as tagged. To achieve a reace free design all tagged directories are created and managed by sysfs itself. Users of this interface: - define a type in the sysfs_tag_type enumeration. - call sysfs_register_ns_types with the type and it's operations - sysfs_exit_ns when an individual tag is no longer valid - Implement mount_ns() which returns the ns of the calling process so we can attach it to a sysfs superblock. - Implement ktype.namespace() which returns the ns of a syfs kobject. Everything else is left up to sysfs and the driver layer. For the network namespace mount_ns and namespace() are essentially one line functions, and look to remain that. Tags are currently represented a const void * pointers as that is both generic, prevides enough information for equality comparisons, and is trivial to create for current users, as it is just the existing namespace pointer. The work needed in sysfs is more extensive. At each directory or symlink creating I need to check if the directory it is being created in is a tagged directory and if so generate the appropriate tag to place on the sysfs_dirent. Likewise at each symlink or directory removal I need to check if the sysfs directory it is being removed from is a tagged directory and if so figure out which tag goes along with the name I am deleting. Currently only directories which hold kobjects, and symlinks are supported. There is not enough information in the current file attribute interfaces to give us anything to discriminate on which makes it useless, and there are no potential users which makes it an uninteresting problem to solve. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-30 18:31:26 +00:00
sd = sysfs_find_dirent(sd, NULL, attr);
if (sd)
sysfs_notify_dirent(sd);
mutex_unlock(&sysfs_mutex);
}
EXPORT_SYMBOL_GPL(sysfs_notify);
const struct file_operations sysfs_file_operations = {
.read = sysfs_read_file,
.write = sysfs_write_file,
.llseek = generic_file_llseek,
.open = sysfs_open_file,
.release = sysfs_release,
.poll = sysfs_poll,
};
int sysfs_add_file_mode(struct sysfs_dirent *dir_sd,
const struct attribute *attr, int type, mode_t amode)
{
umode_t mode = (amode & S_IALLUGO) | S_IFREG;
sysfs: restructure add/remove paths and fix inode update The original add/remove code had the following problems. * parent's timestamps are updated on dentry instantiation. this is incorrect with reclaimable files. * updating parent's timestamps isn't synchronized. * parent nlink update assumes the inode is accessible which won't be true once directory dentries are made reclaimable. This patch restructures add/remove paths to resolve the above problems. Add/removal are done in the following steps. 1. sysfs_addrm_start() : acquire locks including sysfs_mutex and other resources. 2-a. sysfs_add_one() : add new sd. linking the new sd into the children list is caller's responsibility. 2-b. sysfs_remove_one() : remove a sd. unlinking the sd from the children list is caller's responsibility. 3. sysfs_addrm_finish() : release all resources and clean up. Steps 2-a and/or 2-b can be repeated multiple times. Parent's inode is looked up during sysfs_addrm_start(). If available (always at the moment), it's pinned and nlink is updated as sd's are added and removed. Timestamps are updated during finish if any sd has been added or removed. If parent's inode is not available during start, sysfs_mutex ensures that parent inode is not created till add/remove is complete. All the complexity is contained inside the helper functions. Especially, dentry/inode handling is properly hidden from the rest of sysfs which now mostly operate on sysfs_dirents. As an added bonus, codes which use these helpers to add and remove sysfs_dirents are now more structured and simpler. Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 19:27:24 +00:00
struct sysfs_addrm_cxt acxt;
struct sysfs_dirent *sd;
int rc;
sd = sysfs_new_dirent(attr->name, mode, type);
if (!sd)
return -ENOMEM;
sd->s_attr.attr = (void *)attr;
sysfs_dirent_init_lockdep(sd);
sysfs: restructure add/remove paths and fix inode update The original add/remove code had the following problems. * parent's timestamps are updated on dentry instantiation. this is incorrect with reclaimable files. * updating parent's timestamps isn't synchronized. * parent nlink update assumes the inode is accessible which won't be true once directory dentries are made reclaimable. This patch restructures add/remove paths to resolve the above problems. Add/removal are done in the following steps. 1. sysfs_addrm_start() : acquire locks including sysfs_mutex and other resources. 2-a. sysfs_add_one() : add new sd. linking the new sd into the children list is caller's responsibility. 2-b. sysfs_remove_one() : remove a sd. unlinking the sd from the children list is caller's responsibility. 3. sysfs_addrm_finish() : release all resources and clean up. Steps 2-a and/or 2-b can be repeated multiple times. Parent's inode is looked up during sysfs_addrm_start(). If available (always at the moment), it's pinned and nlink is updated as sd's are added and removed. Timestamps are updated during finish if any sd has been added or removed. If parent's inode is not available during start, sysfs_mutex ensures that parent inode is not created till add/remove is complete. All the complexity is contained inside the helper functions. Especially, dentry/inode handling is properly hidden from the rest of sysfs which now mostly operate on sysfs_dirents. As an added bonus, codes which use these helpers to add and remove sysfs_dirents are now more structured and simpler. Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2007-06-13 19:27:24 +00:00
sysfs_addrm_start(&acxt, dir_sd);
rc = sysfs_add_one(&acxt, sd);
sysfs_addrm_finish(&acxt);
if (rc)
sysfs_put(sd);
return rc;
}
int sysfs_add_file(struct sysfs_dirent *dir_sd, const struct attribute *attr,
int type)
{
return sysfs_add_file_mode(dir_sd, attr, type, attr->mode);
}
/**
* sysfs_create_file - create an attribute file for an object.
* @kobj: object we're creating for.
* @attr: attribute descriptor.
*/
int sysfs_create_file(struct kobject * kobj, const struct attribute * attr)
{
BUG_ON(!kobj || !kobj->sd || !attr);
return sysfs_add_file(kobj->sd, attr, SYSFS_KOBJ_ATTR);
}
int sysfs_create_files(struct kobject *kobj, const struct attribute **ptr)
{
int err = 0;
int i;
for (i = 0; ptr[i] && !err; i++)
err = sysfs_create_file(kobj, ptr[i]);
if (err)
while (--i >= 0)
sysfs_remove_file(kobj, ptr[i]);
return err;
}
/**
* sysfs_add_file_to_group - add an attribute file to a pre-existing group.
* @kobj: object we're acting for.
* @attr: attribute descriptor.
* @group: group name.
*/
int sysfs_add_file_to_group(struct kobject *kobj,
const struct attribute *attr, const char *group)
{
struct sysfs_dirent *dir_sd;
int error;
if (group)
sysfs: Implement sysfs tagged directory support. The problem. When implementing a network namespace I need to be able to have multiple network devices with the same name. Currently this is a problem for /sys/class/net/*, /sys/devices/virtual/net/*, and potentially a few other directories of the form /sys/ ... /net/*. What this patch does is to add an additional tag field to the sysfs dirent structure. For directories that should show different contents depending on the context such as /sys/class/net/, and /sys/devices/virtual/net/ this tag field is used to specify the context in which those directories should be visible. Effectively this is the same as creating multiple distinct directories with the same name but internally to sysfs the result is nicer. I am calling the concept of a single directory that looks like multiple directories all at the same path in the filesystem tagged directories. For the networking namespace the set of directories whose contents I need to filter with tags can depend on the presence or absence of hotplug hardware or which modules are currently loaded. Which means I need a simple race free way to setup those directories as tagged. To achieve a reace free design all tagged directories are created and managed by sysfs itself. Users of this interface: - define a type in the sysfs_tag_type enumeration. - call sysfs_register_ns_types with the type and it's operations - sysfs_exit_ns when an individual tag is no longer valid - Implement mount_ns() which returns the ns of the calling process so we can attach it to a sysfs superblock. - Implement ktype.namespace() which returns the ns of a syfs kobject. Everything else is left up to sysfs and the driver layer. For the network namespace mount_ns and namespace() are essentially one line functions, and look to remain that. Tags are currently represented a const void * pointers as that is both generic, prevides enough information for equality comparisons, and is trivial to create for current users, as it is just the existing namespace pointer. The work needed in sysfs is more extensive. At each directory or symlink creating I need to check if the directory it is being created in is a tagged directory and if so generate the appropriate tag to place on the sysfs_dirent. Likewise at each symlink or directory removal I need to check if the sysfs directory it is being removed from is a tagged directory and if so figure out which tag goes along with the name I am deleting. Currently only directories which hold kobjects, and symlinks are supported. There is not enough information in the current file attribute interfaces to give us anything to discriminate on which makes it useless, and there are no potential users which makes it an uninteresting problem to solve. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-30 18:31:26 +00:00
dir_sd = sysfs_get_dirent(kobj->sd, NULL, group);
else
dir_sd = sysfs_get(kobj->sd);
if (!dir_sd)
return -ENOENT;
error = sysfs_add_file(dir_sd, attr, SYSFS_KOBJ_ATTR);
sysfs_put(dir_sd);
return error;
}
EXPORT_SYMBOL_GPL(sysfs_add_file_to_group);
/**
* sysfs_chmod_file - update the modified mode value on an object attribute.
* @kobj: object we're acting for.
* @attr: attribute descriptor.
* @mode: file permissions.
*
*/
int sysfs_chmod_file(struct kobject *kobj, struct attribute *attr, mode_t mode)
{
struct sysfs_dirent *sd;
struct iattr newattrs;
int rc;
mutex_lock(&sysfs_mutex);
rc = -ENOENT;
sysfs: Implement sysfs tagged directory support. The problem. When implementing a network namespace I need to be able to have multiple network devices with the same name. Currently this is a problem for /sys/class/net/*, /sys/devices/virtual/net/*, and potentially a few other directories of the form /sys/ ... /net/*. What this patch does is to add an additional tag field to the sysfs dirent structure. For directories that should show different contents depending on the context such as /sys/class/net/, and /sys/devices/virtual/net/ this tag field is used to specify the context in which those directories should be visible. Effectively this is the same as creating multiple distinct directories with the same name but internally to sysfs the result is nicer. I am calling the concept of a single directory that looks like multiple directories all at the same path in the filesystem tagged directories. For the networking namespace the set of directories whose contents I need to filter with tags can depend on the presence or absence of hotplug hardware or which modules are currently loaded. Which means I need a simple race free way to setup those directories as tagged. To achieve a reace free design all tagged directories are created and managed by sysfs itself. Users of this interface: - define a type in the sysfs_tag_type enumeration. - call sysfs_register_ns_types with the type and it's operations - sysfs_exit_ns when an individual tag is no longer valid - Implement mount_ns() which returns the ns of the calling process so we can attach it to a sysfs superblock. - Implement ktype.namespace() which returns the ns of a syfs kobject. Everything else is left up to sysfs and the driver layer. For the network namespace mount_ns and namespace() are essentially one line functions, and look to remain that. Tags are currently represented a const void * pointers as that is both generic, prevides enough information for equality comparisons, and is trivial to create for current users, as it is just the existing namespace pointer. The work needed in sysfs is more extensive. At each directory or symlink creating I need to check if the directory it is being created in is a tagged directory and if so generate the appropriate tag to place on the sysfs_dirent. Likewise at each symlink or directory removal I need to check if the sysfs directory it is being removed from is a tagged directory and if so figure out which tag goes along with the name I am deleting. Currently only directories which hold kobjects, and symlinks are supported. There is not enough information in the current file attribute interfaces to give us anything to discriminate on which makes it useless, and there are no potential users which makes it an uninteresting problem to solve. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-30 18:31:26 +00:00
sd = sysfs_find_dirent(kobj->sd, NULL, attr->name);
if (!sd)
goto out;
newattrs.ia_mode = (mode & S_IALLUGO) | (sd->s_mode & ~S_IALLUGO);
newattrs.ia_valid = ATTR_MODE;
rc = sysfs_sd_setattr(sd, &newattrs);
out:
mutex_unlock(&sysfs_mutex);
return rc;
}
EXPORT_SYMBOL_GPL(sysfs_chmod_file);
/**
* sysfs_remove_file - remove an object attribute.
* @kobj: object we're acting for.
* @attr: attribute descriptor.
*
* Hash the attribute name and kill the victim.
*/
void sysfs_remove_file(struct kobject * kobj, const struct attribute * attr)
{
sysfs: Implement sysfs tagged directory support. The problem. When implementing a network namespace I need to be able to have multiple network devices with the same name. Currently this is a problem for /sys/class/net/*, /sys/devices/virtual/net/*, and potentially a few other directories of the form /sys/ ... /net/*. What this patch does is to add an additional tag field to the sysfs dirent structure. For directories that should show different contents depending on the context such as /sys/class/net/, and /sys/devices/virtual/net/ this tag field is used to specify the context in which those directories should be visible. Effectively this is the same as creating multiple distinct directories with the same name but internally to sysfs the result is nicer. I am calling the concept of a single directory that looks like multiple directories all at the same path in the filesystem tagged directories. For the networking namespace the set of directories whose contents I need to filter with tags can depend on the presence or absence of hotplug hardware or which modules are currently loaded. Which means I need a simple race free way to setup those directories as tagged. To achieve a reace free design all tagged directories are created and managed by sysfs itself. Users of this interface: - define a type in the sysfs_tag_type enumeration. - call sysfs_register_ns_types with the type and it's operations - sysfs_exit_ns when an individual tag is no longer valid - Implement mount_ns() which returns the ns of the calling process so we can attach it to a sysfs superblock. - Implement ktype.namespace() which returns the ns of a syfs kobject. Everything else is left up to sysfs and the driver layer. For the network namespace mount_ns and namespace() are essentially one line functions, and look to remain that. Tags are currently represented a const void * pointers as that is both generic, prevides enough information for equality comparisons, and is trivial to create for current users, as it is just the existing namespace pointer. The work needed in sysfs is more extensive. At each directory or symlink creating I need to check if the directory it is being created in is a tagged directory and if so generate the appropriate tag to place on the sysfs_dirent. Likewise at each symlink or directory removal I need to check if the sysfs directory it is being removed from is a tagged directory and if so figure out which tag goes along with the name I am deleting. Currently only directories which hold kobjects, and symlinks are supported. There is not enough information in the current file attribute interfaces to give us anything to discriminate on which makes it useless, and there are no potential users which makes it an uninteresting problem to solve. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-30 18:31:26 +00:00
sysfs_hash_and_remove(kobj->sd, NULL, attr->name);
}
void sysfs_remove_files(struct kobject * kobj, const struct attribute **ptr)
{
int i;
for (i = 0; ptr[i]; i++)
sysfs_remove_file(kobj, ptr[i]);
}
/**
* sysfs_remove_file_from_group - remove an attribute file from a group.
* @kobj: object we're acting for.
* @attr: attribute descriptor.
* @group: group name.
*/
void sysfs_remove_file_from_group(struct kobject *kobj,
const struct attribute *attr, const char *group)
{
struct sysfs_dirent *dir_sd;
if (group)
sysfs: Implement sysfs tagged directory support. The problem. When implementing a network namespace I need to be able to have multiple network devices with the same name. Currently this is a problem for /sys/class/net/*, /sys/devices/virtual/net/*, and potentially a few other directories of the form /sys/ ... /net/*. What this patch does is to add an additional tag field to the sysfs dirent structure. For directories that should show different contents depending on the context such as /sys/class/net/, and /sys/devices/virtual/net/ this tag field is used to specify the context in which those directories should be visible. Effectively this is the same as creating multiple distinct directories with the same name but internally to sysfs the result is nicer. I am calling the concept of a single directory that looks like multiple directories all at the same path in the filesystem tagged directories. For the networking namespace the set of directories whose contents I need to filter with tags can depend on the presence or absence of hotplug hardware or which modules are currently loaded. Which means I need a simple race free way to setup those directories as tagged. To achieve a reace free design all tagged directories are created and managed by sysfs itself. Users of this interface: - define a type in the sysfs_tag_type enumeration. - call sysfs_register_ns_types with the type and it's operations - sysfs_exit_ns when an individual tag is no longer valid - Implement mount_ns() which returns the ns of the calling process so we can attach it to a sysfs superblock. - Implement ktype.namespace() which returns the ns of a syfs kobject. Everything else is left up to sysfs and the driver layer. For the network namespace mount_ns and namespace() are essentially one line functions, and look to remain that. Tags are currently represented a const void * pointers as that is both generic, prevides enough information for equality comparisons, and is trivial to create for current users, as it is just the existing namespace pointer. The work needed in sysfs is more extensive. At each directory or symlink creating I need to check if the directory it is being created in is a tagged directory and if so generate the appropriate tag to place on the sysfs_dirent. Likewise at each symlink or directory removal I need to check if the sysfs directory it is being removed from is a tagged directory and if so figure out which tag goes along with the name I am deleting. Currently only directories which hold kobjects, and symlinks are supported. There is not enough information in the current file attribute interfaces to give us anything to discriminate on which makes it useless, and there are no potential users which makes it an uninteresting problem to solve. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-30 18:31:26 +00:00
dir_sd = sysfs_get_dirent(kobj->sd, NULL, group);
else
dir_sd = sysfs_get(kobj->sd);
if (dir_sd) {
sysfs: Implement sysfs tagged directory support. The problem. When implementing a network namespace I need to be able to have multiple network devices with the same name. Currently this is a problem for /sys/class/net/*, /sys/devices/virtual/net/*, and potentially a few other directories of the form /sys/ ... /net/*. What this patch does is to add an additional tag field to the sysfs dirent structure. For directories that should show different contents depending on the context such as /sys/class/net/, and /sys/devices/virtual/net/ this tag field is used to specify the context in which those directories should be visible. Effectively this is the same as creating multiple distinct directories with the same name but internally to sysfs the result is nicer. I am calling the concept of a single directory that looks like multiple directories all at the same path in the filesystem tagged directories. For the networking namespace the set of directories whose contents I need to filter with tags can depend on the presence or absence of hotplug hardware or which modules are currently loaded. Which means I need a simple race free way to setup those directories as tagged. To achieve a reace free design all tagged directories are created and managed by sysfs itself. Users of this interface: - define a type in the sysfs_tag_type enumeration. - call sysfs_register_ns_types with the type and it's operations - sysfs_exit_ns when an individual tag is no longer valid - Implement mount_ns() which returns the ns of the calling process so we can attach it to a sysfs superblock. - Implement ktype.namespace() which returns the ns of a syfs kobject. Everything else is left up to sysfs and the driver layer. For the network namespace mount_ns and namespace() are essentially one line functions, and look to remain that. Tags are currently represented a const void * pointers as that is both generic, prevides enough information for equality comparisons, and is trivial to create for current users, as it is just the existing namespace pointer. The work needed in sysfs is more extensive. At each directory or symlink creating I need to check if the directory it is being created in is a tagged directory and if so generate the appropriate tag to place on the sysfs_dirent. Likewise at each symlink or directory removal I need to check if the sysfs directory it is being removed from is a tagged directory and if so figure out which tag goes along with the name I am deleting. Currently only directories which hold kobjects, and symlinks are supported. There is not enough information in the current file attribute interfaces to give us anything to discriminate on which makes it useless, and there are no potential users which makes it an uninteresting problem to solve. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-30 18:31:26 +00:00
sysfs_hash_and_remove(dir_sd, NULL, attr->name);
sysfs_put(dir_sd);
}
}
EXPORT_SYMBOL_GPL(sysfs_remove_file_from_group);
struct sysfs_schedule_callback_struct {
struct list_head workq_list;
struct kobject *kobj;
void (*func)(void *);
void *data;
struct module *owner;
struct work_struct work;
};
sysfs: don't use global workqueue in sysfs_schedule_callback() A sysfs attribute using sysfs_schedule_callback() to commit suicide may end up calling device_unregister(), which will eventually call a driver's ->remove function. Drivers may call flush_scheduled_work() in their shutdown routines, in which case lockdep will complain with something like the following: ============================================= [ INFO: possible recursive locking detected ] 2.6.29-rc8-kk #1 --------------------------------------------- events/4/56 is trying to acquire lock: (events){--..}, at: [<ffffffff80257fc0>] flush_workqueue+0x0/0xa0 but task is already holding lock: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 other info that might help us debug this: 3 locks held by events/4/56: #0: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #1: (&ss->work){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #2: (pci_remove_rescan_mutex){--..}, at: [<ffffffff803c10d1>] remove_callback+0x21/0x40 stack backtrace: Pid: 56, comm: events/4 Not tainted 2.6.29-rc8-kk #1 Call Trace: [<ffffffff8026dfcd>] validate_chain+0xb7d/0x1260 [<ffffffff8026eade>] __lock_acquire+0x42e/0xa40 [<ffffffff8026f148>] lock_acquire+0x58/0x80 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff8025800d>] flush_workqueue+0x4d/0xa0 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff80258070>] flush_scheduled_work+0x10/0x20 [<ffffffffa0144065>] e1000_remove+0x55/0xfe [e1000e] [<ffffffff8033ee30>] ? sysfs_schedule_callback_work+0x0/0x50 [<ffffffff803bfeb2>] pci_device_remove+0x32/0x70 [<ffffffff80441da9>] __device_release_driver+0x59/0x90 [<ffffffff80441edb>] device_release_driver+0x2b/0x40 [<ffffffff804419d6>] bus_remove_device+0xa6/0x120 [<ffffffff8043e46b>] device_del+0x12b/0x190 [<ffffffff8043e4f6>] device_unregister+0x26/0x70 [<ffffffff803ba969>] pci_stop_dev+0x49/0x60 [<ffffffff803baab0>] pci_remove_bus_device+0x40/0xc0 [<ffffffff803c10d9>] remove_callback+0x29/0x40 [<ffffffff8033ee4f>] sysfs_schedule_callback_work+0x1f/0x50 [<ffffffff8025769a>] run_workqueue+0x15a/0x230 [<ffffffff80257648>] ? run_workqueue+0x108/0x230 [<ffffffff8025846f>] worker_thread+0x9f/0x100 [<ffffffff8025bce0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff802583d0>] ? worker_thread+0x0/0x100 [<ffffffff8025b89d>] kthread+0x4d/0x80 [<ffffffff8020d4ba>] child_rip+0xa/0x20 [<ffffffff8020cebc>] ? restore_args+0x0/0x30 [<ffffffff8025b850>] ? kthread+0x0/0x80 [<ffffffff8020d4b0>] ? child_rip+0x0/0x20 Although we know that the device_unregister path will never acquire a lock that a driver might try to acquire in its ->remove, in general we should never attempt to flush a workqueue from within the same workqueue, and lockdep rightly complains. So as long as sysfs attributes cannot commit suicide directly and we are stuck with this callback mechanism, put the sysfs callbacks on their own workqueue instead of the global one. This has the side benefit that if a suicidal sysfs attribute kicks off a long chain of ->remove callbacks, we no longer induce a long delay on the global queue. This also fixes a missing module_put in the error path introduced by sysfs-only-allow-one-scheduled-removal-callback-per-kobj.patch. We never destroy the workqueue, but I'm not sure that's a problem. Reported-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Tested-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Signed-off-by: Alex Chiang <achiang@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-03-25 21:11:36 +00:00
static struct workqueue_struct *sysfs_workqueue;
static DEFINE_MUTEX(sysfs_workq_mutex);
static LIST_HEAD(sysfs_workq);
static void sysfs_schedule_callback_work(struct work_struct *work)
{
struct sysfs_schedule_callback_struct *ss = container_of(work,
struct sysfs_schedule_callback_struct, work);
(ss->func)(ss->data);
kobject_put(ss->kobj);
module_put(ss->owner);
mutex_lock(&sysfs_workq_mutex);
list_del(&ss->workq_list);
mutex_unlock(&sysfs_workq_mutex);
kfree(ss);
}
/**
* sysfs_schedule_callback - helper to schedule a callback for a kobject
* @kobj: object we're acting for.
* @func: callback function to invoke later.
* @data: argument to pass to @func.
* @owner: module owning the callback code
*
* sysfs attribute methods must not unregister themselves or their parent
* kobject (which would amount to the same thing). Attempts to do so will
* deadlock, since unregistration is mutually exclusive with driver
* callbacks.
*
* Instead methods can call this routine, which will attempt to allocate
* and schedule a workqueue request to call back @func with @data as its
* argument in the workqueue's process context. @kobj will be pinned
* until @func returns.
*
* Returns 0 if the request was submitted, -ENOMEM if storage could not
* be allocated, -ENODEV if a reference to @owner isn't available,
* -EAGAIN if a callback has already been scheduled for @kobj.
*/
int sysfs_schedule_callback(struct kobject *kobj, void (*func)(void *),
void *data, struct module *owner)
{
struct sysfs_schedule_callback_struct *ss, *tmp;
if (!try_module_get(owner))
return -ENODEV;
mutex_lock(&sysfs_workq_mutex);
list_for_each_entry_safe(ss, tmp, &sysfs_workq, workq_list)
if (ss->kobj == kobj) {
sysfs: don't use global workqueue in sysfs_schedule_callback() A sysfs attribute using sysfs_schedule_callback() to commit suicide may end up calling device_unregister(), which will eventually call a driver's ->remove function. Drivers may call flush_scheduled_work() in their shutdown routines, in which case lockdep will complain with something like the following: ============================================= [ INFO: possible recursive locking detected ] 2.6.29-rc8-kk #1 --------------------------------------------- events/4/56 is trying to acquire lock: (events){--..}, at: [<ffffffff80257fc0>] flush_workqueue+0x0/0xa0 but task is already holding lock: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 other info that might help us debug this: 3 locks held by events/4/56: #0: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #1: (&ss->work){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #2: (pci_remove_rescan_mutex){--..}, at: [<ffffffff803c10d1>] remove_callback+0x21/0x40 stack backtrace: Pid: 56, comm: events/4 Not tainted 2.6.29-rc8-kk #1 Call Trace: [<ffffffff8026dfcd>] validate_chain+0xb7d/0x1260 [<ffffffff8026eade>] __lock_acquire+0x42e/0xa40 [<ffffffff8026f148>] lock_acquire+0x58/0x80 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff8025800d>] flush_workqueue+0x4d/0xa0 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff80258070>] flush_scheduled_work+0x10/0x20 [<ffffffffa0144065>] e1000_remove+0x55/0xfe [e1000e] [<ffffffff8033ee30>] ? sysfs_schedule_callback_work+0x0/0x50 [<ffffffff803bfeb2>] pci_device_remove+0x32/0x70 [<ffffffff80441da9>] __device_release_driver+0x59/0x90 [<ffffffff80441edb>] device_release_driver+0x2b/0x40 [<ffffffff804419d6>] bus_remove_device+0xa6/0x120 [<ffffffff8043e46b>] device_del+0x12b/0x190 [<ffffffff8043e4f6>] device_unregister+0x26/0x70 [<ffffffff803ba969>] pci_stop_dev+0x49/0x60 [<ffffffff803baab0>] pci_remove_bus_device+0x40/0xc0 [<ffffffff803c10d9>] remove_callback+0x29/0x40 [<ffffffff8033ee4f>] sysfs_schedule_callback_work+0x1f/0x50 [<ffffffff8025769a>] run_workqueue+0x15a/0x230 [<ffffffff80257648>] ? run_workqueue+0x108/0x230 [<ffffffff8025846f>] worker_thread+0x9f/0x100 [<ffffffff8025bce0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff802583d0>] ? worker_thread+0x0/0x100 [<ffffffff8025b89d>] kthread+0x4d/0x80 [<ffffffff8020d4ba>] child_rip+0xa/0x20 [<ffffffff8020cebc>] ? restore_args+0x0/0x30 [<ffffffff8025b850>] ? kthread+0x0/0x80 [<ffffffff8020d4b0>] ? child_rip+0x0/0x20 Although we know that the device_unregister path will never acquire a lock that a driver might try to acquire in its ->remove, in general we should never attempt to flush a workqueue from within the same workqueue, and lockdep rightly complains. So as long as sysfs attributes cannot commit suicide directly and we are stuck with this callback mechanism, put the sysfs callbacks on their own workqueue instead of the global one. This has the side benefit that if a suicidal sysfs attribute kicks off a long chain of ->remove callbacks, we no longer induce a long delay on the global queue. This also fixes a missing module_put in the error path introduced by sysfs-only-allow-one-scheduled-removal-callback-per-kobj.patch. We never destroy the workqueue, but I'm not sure that's a problem. Reported-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Tested-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Signed-off-by: Alex Chiang <achiang@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-03-25 21:11:36 +00:00
module_put(owner);
mutex_unlock(&sysfs_workq_mutex);
return -EAGAIN;
}
mutex_unlock(&sysfs_workq_mutex);
sysfs: don't use global workqueue in sysfs_schedule_callback() A sysfs attribute using sysfs_schedule_callback() to commit suicide may end up calling device_unregister(), which will eventually call a driver's ->remove function. Drivers may call flush_scheduled_work() in their shutdown routines, in which case lockdep will complain with something like the following: ============================================= [ INFO: possible recursive locking detected ] 2.6.29-rc8-kk #1 --------------------------------------------- events/4/56 is trying to acquire lock: (events){--..}, at: [<ffffffff80257fc0>] flush_workqueue+0x0/0xa0 but task is already holding lock: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 other info that might help us debug this: 3 locks held by events/4/56: #0: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #1: (&ss->work){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #2: (pci_remove_rescan_mutex){--..}, at: [<ffffffff803c10d1>] remove_callback+0x21/0x40 stack backtrace: Pid: 56, comm: events/4 Not tainted 2.6.29-rc8-kk #1 Call Trace: [<ffffffff8026dfcd>] validate_chain+0xb7d/0x1260 [<ffffffff8026eade>] __lock_acquire+0x42e/0xa40 [<ffffffff8026f148>] lock_acquire+0x58/0x80 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff8025800d>] flush_workqueue+0x4d/0xa0 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff80258070>] flush_scheduled_work+0x10/0x20 [<ffffffffa0144065>] e1000_remove+0x55/0xfe [e1000e] [<ffffffff8033ee30>] ? sysfs_schedule_callback_work+0x0/0x50 [<ffffffff803bfeb2>] pci_device_remove+0x32/0x70 [<ffffffff80441da9>] __device_release_driver+0x59/0x90 [<ffffffff80441edb>] device_release_driver+0x2b/0x40 [<ffffffff804419d6>] bus_remove_device+0xa6/0x120 [<ffffffff8043e46b>] device_del+0x12b/0x190 [<ffffffff8043e4f6>] device_unregister+0x26/0x70 [<ffffffff803ba969>] pci_stop_dev+0x49/0x60 [<ffffffff803baab0>] pci_remove_bus_device+0x40/0xc0 [<ffffffff803c10d9>] remove_callback+0x29/0x40 [<ffffffff8033ee4f>] sysfs_schedule_callback_work+0x1f/0x50 [<ffffffff8025769a>] run_workqueue+0x15a/0x230 [<ffffffff80257648>] ? run_workqueue+0x108/0x230 [<ffffffff8025846f>] worker_thread+0x9f/0x100 [<ffffffff8025bce0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff802583d0>] ? worker_thread+0x0/0x100 [<ffffffff8025b89d>] kthread+0x4d/0x80 [<ffffffff8020d4ba>] child_rip+0xa/0x20 [<ffffffff8020cebc>] ? restore_args+0x0/0x30 [<ffffffff8025b850>] ? kthread+0x0/0x80 [<ffffffff8020d4b0>] ? child_rip+0x0/0x20 Although we know that the device_unregister path will never acquire a lock that a driver might try to acquire in its ->remove, in general we should never attempt to flush a workqueue from within the same workqueue, and lockdep rightly complains. So as long as sysfs attributes cannot commit suicide directly and we are stuck with this callback mechanism, put the sysfs callbacks on their own workqueue instead of the global one. This has the side benefit that if a suicidal sysfs attribute kicks off a long chain of ->remove callbacks, we no longer induce a long delay on the global queue. This also fixes a missing module_put in the error path introduced by sysfs-only-allow-one-scheduled-removal-callback-per-kobj.patch. We never destroy the workqueue, but I'm not sure that's a problem. Reported-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Tested-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Signed-off-by: Alex Chiang <achiang@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-03-25 21:11:36 +00:00
if (sysfs_workqueue == NULL) {
sysfs_workqueue = create_singlethread_workqueue("sysfsd");
sysfs: don't use global workqueue in sysfs_schedule_callback() A sysfs attribute using sysfs_schedule_callback() to commit suicide may end up calling device_unregister(), which will eventually call a driver's ->remove function. Drivers may call flush_scheduled_work() in their shutdown routines, in which case lockdep will complain with something like the following: ============================================= [ INFO: possible recursive locking detected ] 2.6.29-rc8-kk #1 --------------------------------------------- events/4/56 is trying to acquire lock: (events){--..}, at: [<ffffffff80257fc0>] flush_workqueue+0x0/0xa0 but task is already holding lock: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 other info that might help us debug this: 3 locks held by events/4/56: #0: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #1: (&ss->work){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #2: (pci_remove_rescan_mutex){--..}, at: [<ffffffff803c10d1>] remove_callback+0x21/0x40 stack backtrace: Pid: 56, comm: events/4 Not tainted 2.6.29-rc8-kk #1 Call Trace: [<ffffffff8026dfcd>] validate_chain+0xb7d/0x1260 [<ffffffff8026eade>] __lock_acquire+0x42e/0xa40 [<ffffffff8026f148>] lock_acquire+0x58/0x80 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff8025800d>] flush_workqueue+0x4d/0xa0 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff80258070>] flush_scheduled_work+0x10/0x20 [<ffffffffa0144065>] e1000_remove+0x55/0xfe [e1000e] [<ffffffff8033ee30>] ? sysfs_schedule_callback_work+0x0/0x50 [<ffffffff803bfeb2>] pci_device_remove+0x32/0x70 [<ffffffff80441da9>] __device_release_driver+0x59/0x90 [<ffffffff80441edb>] device_release_driver+0x2b/0x40 [<ffffffff804419d6>] bus_remove_device+0xa6/0x120 [<ffffffff8043e46b>] device_del+0x12b/0x190 [<ffffffff8043e4f6>] device_unregister+0x26/0x70 [<ffffffff803ba969>] pci_stop_dev+0x49/0x60 [<ffffffff803baab0>] pci_remove_bus_device+0x40/0xc0 [<ffffffff803c10d9>] remove_callback+0x29/0x40 [<ffffffff8033ee4f>] sysfs_schedule_callback_work+0x1f/0x50 [<ffffffff8025769a>] run_workqueue+0x15a/0x230 [<ffffffff80257648>] ? run_workqueue+0x108/0x230 [<ffffffff8025846f>] worker_thread+0x9f/0x100 [<ffffffff8025bce0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff802583d0>] ? worker_thread+0x0/0x100 [<ffffffff8025b89d>] kthread+0x4d/0x80 [<ffffffff8020d4ba>] child_rip+0xa/0x20 [<ffffffff8020cebc>] ? restore_args+0x0/0x30 [<ffffffff8025b850>] ? kthread+0x0/0x80 [<ffffffff8020d4b0>] ? child_rip+0x0/0x20 Although we know that the device_unregister path will never acquire a lock that a driver might try to acquire in its ->remove, in general we should never attempt to flush a workqueue from within the same workqueue, and lockdep rightly complains. So as long as sysfs attributes cannot commit suicide directly and we are stuck with this callback mechanism, put the sysfs callbacks on their own workqueue instead of the global one. This has the side benefit that if a suicidal sysfs attribute kicks off a long chain of ->remove callbacks, we no longer induce a long delay on the global queue. This also fixes a missing module_put in the error path introduced by sysfs-only-allow-one-scheduled-removal-callback-per-kobj.patch. We never destroy the workqueue, but I'm not sure that's a problem. Reported-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Tested-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Signed-off-by: Alex Chiang <achiang@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-03-25 21:11:36 +00:00
if (sysfs_workqueue == NULL) {
module_put(owner);
return -ENOMEM;
}
}
ss = kmalloc(sizeof(*ss), GFP_KERNEL);
if (!ss) {
module_put(owner);
return -ENOMEM;
}
kobject_get(kobj);
ss->kobj = kobj;
ss->func = func;
ss->data = data;
ss->owner = owner;
INIT_WORK(&ss->work, sysfs_schedule_callback_work);
INIT_LIST_HEAD(&ss->workq_list);
mutex_lock(&sysfs_workq_mutex);
list_add_tail(&ss->workq_list, &sysfs_workq);
mutex_unlock(&sysfs_workq_mutex);
sysfs: don't use global workqueue in sysfs_schedule_callback() A sysfs attribute using sysfs_schedule_callback() to commit suicide may end up calling device_unregister(), which will eventually call a driver's ->remove function. Drivers may call flush_scheduled_work() in their shutdown routines, in which case lockdep will complain with something like the following: ============================================= [ INFO: possible recursive locking detected ] 2.6.29-rc8-kk #1 --------------------------------------------- events/4/56 is trying to acquire lock: (events){--..}, at: [<ffffffff80257fc0>] flush_workqueue+0x0/0xa0 but task is already holding lock: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 other info that might help us debug this: 3 locks held by events/4/56: #0: (events){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #1: (&ss->work){--..}, at: [<ffffffff80257648>] run_workqueue+0x108/0x230 #2: (pci_remove_rescan_mutex){--..}, at: [<ffffffff803c10d1>] remove_callback+0x21/0x40 stack backtrace: Pid: 56, comm: events/4 Not tainted 2.6.29-rc8-kk #1 Call Trace: [<ffffffff8026dfcd>] validate_chain+0xb7d/0x1260 [<ffffffff8026eade>] __lock_acquire+0x42e/0xa40 [<ffffffff8026f148>] lock_acquire+0x58/0x80 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff8025800d>] flush_workqueue+0x4d/0xa0 [<ffffffff80257fc0>] ? flush_workqueue+0x0/0xa0 [<ffffffff80258070>] flush_scheduled_work+0x10/0x20 [<ffffffffa0144065>] e1000_remove+0x55/0xfe [e1000e] [<ffffffff8033ee30>] ? sysfs_schedule_callback_work+0x0/0x50 [<ffffffff803bfeb2>] pci_device_remove+0x32/0x70 [<ffffffff80441da9>] __device_release_driver+0x59/0x90 [<ffffffff80441edb>] device_release_driver+0x2b/0x40 [<ffffffff804419d6>] bus_remove_device+0xa6/0x120 [<ffffffff8043e46b>] device_del+0x12b/0x190 [<ffffffff8043e4f6>] device_unregister+0x26/0x70 [<ffffffff803ba969>] pci_stop_dev+0x49/0x60 [<ffffffff803baab0>] pci_remove_bus_device+0x40/0xc0 [<ffffffff803c10d9>] remove_callback+0x29/0x40 [<ffffffff8033ee4f>] sysfs_schedule_callback_work+0x1f/0x50 [<ffffffff8025769a>] run_workqueue+0x15a/0x230 [<ffffffff80257648>] ? run_workqueue+0x108/0x230 [<ffffffff8025846f>] worker_thread+0x9f/0x100 [<ffffffff8025bce0>] ? autoremove_wake_function+0x0/0x40 [<ffffffff802583d0>] ? worker_thread+0x0/0x100 [<ffffffff8025b89d>] kthread+0x4d/0x80 [<ffffffff8020d4ba>] child_rip+0xa/0x20 [<ffffffff8020cebc>] ? restore_args+0x0/0x30 [<ffffffff8025b850>] ? kthread+0x0/0x80 [<ffffffff8020d4b0>] ? child_rip+0x0/0x20 Although we know that the device_unregister path will never acquire a lock that a driver might try to acquire in its ->remove, in general we should never attempt to flush a workqueue from within the same workqueue, and lockdep rightly complains. So as long as sysfs attributes cannot commit suicide directly and we are stuck with this callback mechanism, put the sysfs callbacks on their own workqueue instead of the global one. This has the side benefit that if a suicidal sysfs attribute kicks off a long chain of ->remove callbacks, we no longer induce a long delay on the global queue. This also fixes a missing module_put in the error path introduced by sysfs-only-allow-one-scheduled-removal-callback-per-kobj.patch. We never destroy the workqueue, but I'm not sure that's a problem. Reported-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Tested-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Signed-off-by: Alex Chiang <achiang@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-03-25 21:11:36 +00:00
queue_work(sysfs_workqueue, &ss->work);
return 0;
}
EXPORT_SYMBOL_GPL(sysfs_schedule_callback);
EXPORT_SYMBOL_GPL(sysfs_create_file);
EXPORT_SYMBOL_GPL(sysfs_remove_file);
EXPORT_SYMBOL_GPL(sysfs_remove_files);
EXPORT_SYMBOL_GPL(sysfs_create_files);