linux/fs/sysfs/file.c

1303 lines
31 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 <linux/uaccess.h>
#include <linux/seq_file.h>
#include <linux/mm.h>
#include "sysfs.h"
/*
* There's one sysfs_open_file for each open file and one sysfs_open_dirent
* for each sysfs_dirent with one or more open files.
*
* sysfs_dirent->s_attr.open points to sysfs_open_dirent. s_attr.open is
* protected by sysfs_open_dirent_lock.
*
* filp->private_data points to seq_file whose ->private points to
* sysfs_open_file. sysfs_open_files are chained at
* sysfs_open_dirent->files, which is protected by sysfs_open_file_mutex.
*/
static DEFINE_SPINLOCK(sysfs_open_dirent_lock);
static DEFINE_MUTEX(sysfs_open_file_mutex);
struct sysfs_open_dirent {
atomic_t refcnt;
atomic_t event;
wait_queue_head_t poll;
struct list_head files; /* goes through sysfs_open_file.list */
};
static struct sysfs_open_file *sysfs_of(struct file *file)
{
return ((struct seq_file *)file->private_data)->private;
}
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
/*
* Determine the kernfs_ops for the given sysfs_dirent. This function must
* be called while holding an active reference.
*/
static const struct kernfs_ops *kernfs_ops(struct sysfs_dirent *sd)
{
if (!sysfs_ignore_lockdep(sd))
lockdep_assert_held(sd);
return sd->s_attr.ops;
}
/*
* Determine ktype->sysfs_ops for the given sysfs_dirent. This function
* must be called while holding an active reference.
*/
static const struct sysfs_ops *sysfs_file_ops(struct sysfs_dirent *sd)
{
struct kobject *kobj = sd->s_parent->priv;
sysfs: make sysfs_file_ops() follow ignore_lockdep flag 375b611e60 ("sysfs: remove sysfs_buffer->ops") introduced sysfs_file_ops() which determines the associated file operation of a given sysfs_dirent. As file ops access should be protected by an active reference, the new function includes a lockdep assertion on the sysfs_dirent; unfortunately, I forgot to take attr->ignore_lockdep flag into account and the lockdep assertion trips spuriously for files which opt out from active reference lockdep checking. # cat /sys/devices/pci0000:00/0000:00:01.2/usb1/authorized ------------[ cut here ]------------ WARNING: CPU: 1 PID: 540 at /work/os/work/fs/sysfs/file.c:79 sysfs_file_ops+0x4e/0x60() Modules linked in: CPU: 1 PID: 540 Comm: cat Not tainted 3.11.0-work+ #3 Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 0000000000000009 ffff880016205c08 ffffffff81ca0131 0000000000000000 ffff880016205c40 ffffffff81096d0d ffff8800166cb898 ffff8800166f6f60 ffffffff8125a220 ffff880011ab1ec0 ffff88000aff0c78 ffff880016205c50 Call Trace: [<ffffffff81ca0131>] dump_stack+0x4e/0x82 [<ffffffff81096d0d>] warn_slowpath_common+0x7d/0xa0 [<ffffffff81096dea>] warn_slowpath_null+0x1a/0x20 [<ffffffff8125994e>] sysfs_file_ops+0x4e/0x60 [<ffffffff8125a274>] sysfs_open_file+0x54/0x300 [<ffffffff811df612>] do_dentry_open.isra.17+0x182/0x280 [<ffffffff811df820>] finish_open+0x30/0x40 [<ffffffff811f0623>] do_last+0x503/0xd90 [<ffffffff811f0f6b>] path_openat+0xbb/0x6d0 [<ffffffff811f23ba>] do_filp_open+0x3a/0x90 [<ffffffff811e09a9>] do_sys_open+0x129/0x220 [<ffffffff811e0abe>] SyS_open+0x1e/0x20 [<ffffffff81caf3c2>] system_call_fastpath+0x16/0x1b ---[ end trace aa48096b111dafdb ]--- Rename fs/sysfs/dir.c::ignore_lockdep() to sysfs_ignore_lockdep() and move it to fs/sysfs/sysfs.h and make sysfs_file_ops() skip lockdep assertion if sysfs_ignore_lockdep() is true. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-10-14 13:27:11 +00:00
if (!sysfs_ignore_lockdep(sd))
lockdep_assert_held(sd);
return kobj->ktype ? kobj->ktype->sysfs_ops : NULL;
}
/*
* Reads on sysfs are handled through seq_file, which takes care of hairy
* details like buffering and seeking. The following function pipes
* sysfs_ops->show() result through seq_file.
*/
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
static int sysfs_kf_seq_show(struct seq_file *sf, void *v)
{
struct sysfs_open_file *of = sf->private;
struct kobject *kobj = of->sd->s_parent->priv;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
const struct sysfs_ops *ops = sysfs_file_ops(of->sd);
ssize_t count;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
char *buf;
/* acquire buffer and ensure that it's >= PAGE_SIZE */
count = seq_get_buf(sf, &buf);
if (count < PAGE_SIZE) {
seq_commit(sf, -1);
return 0;
}
/*
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
* Invoke show(). Control may reach here via seq file lseek even
* if @ops->show() isn't implemented.
*/
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
if (ops->show) {
count = ops->show(kobj, of->sd->priv, buf);
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
if (count < 0)
return 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
/*
* 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;
}
seq_commit(sf, count);
return 0;
}
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
static ssize_t sysfs_kf_bin_read(struct sysfs_open_file *of, char *buf,
size_t count, loff_t pos)
{
struct bin_attribute *battr = of->sd->priv;
struct kobject *kobj = of->sd->s_parent->priv;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
loff_t size = file_inode(of->file)->i_size;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
if (!count)
return 0;
if (size) {
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
if (pos > size)
return 0;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
if (pos + count > size)
count = size - pos;
}
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
if (!battr->read)
return -EIO;
return battr->read(of->file, kobj, battr, buf, pos, count);
}
static void *kernfs_seq_start(struct seq_file *sf, loff_t *ppos)
{
struct sysfs_open_file *of = sf->private;
const struct kernfs_ops *ops;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
/*
* @of->mutex nests outside active ref and is just to ensure that
* the ops aren't called concurrently for the same open file.
*/
mutex_lock(&of->mutex);
if (!sysfs_get_active(of->sd))
return ERR_PTR(-ENODEV);
ops = kernfs_ops(of->sd);
if (ops->seq_start) {
return ops->seq_start(sf, ppos);
} else {
/*
* The same behavior and code as single_open(). Returns
* !NULL if pos is at the beginning; otherwise, NULL.
*/
return NULL + !*ppos;
}
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
}
static void *kernfs_seq_next(struct seq_file *sf, void *v, loff_t *ppos)
{
struct sysfs_open_file *of = sf->private;
const struct kernfs_ops *ops = kernfs_ops(of->sd);
if (ops->seq_next) {
return ops->seq_next(sf, v, ppos);
} else {
/*
* The same behavior and code as single_open(), always
* terminate after the initial read.
*/
++*ppos;
return NULL;
}
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
}
static void kernfs_seq_stop(struct seq_file *sf, void *v)
{
struct sysfs_open_file *of = sf->private;
const struct kernfs_ops *ops = kernfs_ops(of->sd);
if (ops->seq_stop)
ops->seq_stop(sf, v);
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
sysfs_put_active(of->sd);
mutex_unlock(&of->mutex);
}
static int kernfs_seq_show(struct seq_file *sf, void *v)
{
struct sysfs_open_file *of = sf->private;
of->event = atomic_read(&of->sd->s_attr.open->event);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
return of->sd->s_attr.ops->seq_show(sf, v);
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
}
static const struct seq_operations kernfs_seq_ops = {
.start = kernfs_seq_start,
.next = kernfs_seq_next,
.stop = kernfs_seq_stop,
.show = kernfs_seq_show,
};
/*
* As reading a bin file can have side-effects, the exact offset and bytes
* specified in read(2) call should be passed to the read callback making
* it difficult to use seq_file. Implement simplistic custom buffering for
* bin files.
*/
static ssize_t kernfs_file_direct_read(struct sysfs_open_file *of,
char __user *user_buf, size_t count,
loff_t *ppos)
{
ssize_t len = min_t(size_t, count, PAGE_SIZE);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
const struct kernfs_ops *ops;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
char *buf;
buf = kmalloc(len, GFP_KERNEL);
if (!buf)
return -ENOMEM;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
/*
* @of->mutex nests outside active ref and is just to ensure that
* the ops aren't called concurrently for the same open file.
*/
mutex_lock(&of->mutex);
if (!sysfs_get_active(of->sd)) {
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
len = -ENODEV;
mutex_unlock(&of->mutex);
goto out_free;
}
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
ops = kernfs_ops(of->sd);
if (ops->read)
len = ops->read(of, buf, len, *ppos);
else
len = -EINVAL;
sysfs_put_active(of->sd);
mutex_unlock(&of->mutex);
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
if (len < 0)
goto out_free;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
if (copy_to_user(user_buf, buf, len)) {
len = -EFAULT;
goto out_free;
}
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
*ppos += len;
out_free:
kfree(buf);
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
return len;
}
/**
* kernfs_file_read - kernfs vfs read callback
* @file: file pointer
* @user_buf: data to write
* @count: number of bytes
* @ppos: starting offset
*/
static ssize_t kernfs_file_read(struct file *file, char __user *user_buf,
size_t count, loff_t *ppos)
{
struct sysfs_open_file *of = sysfs_of(file);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
if (of->sd->s_flags & SYSFS_FLAG_HAS_SEQ_SHOW)
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
return seq_read(file, user_buf, count, ppos);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
else
return kernfs_file_direct_read(of, user_buf, count, ppos);
}
/* kernfs write callback for regular sysfs files */
static ssize_t sysfs_kf_write(struct sysfs_open_file *of, char *buf,
size_t count, loff_t pos)
{
const struct sysfs_ops *ops = sysfs_file_ops(of->sd);
struct kobject *kobj = of->sd->s_parent->priv;
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 (!count)
return 0;
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 ops->store(kobj, of->sd->priv, buf, count);
}
/* kernfs write callback for bin sysfs files */
static ssize_t sysfs_kf_bin_write(struct sysfs_open_file *of, char *buf,
size_t count, loff_t pos)
{
struct bin_attribute *battr = of->sd->priv;
struct kobject *kobj = of->sd->s_parent->priv;
loff_t size = file_inode(of->file)->i_size;
if (size) {
if (size <= pos)
return 0;
count = min_t(ssize_t, count, size - pos);
}
if (!count)
return 0;
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 (!battr->write)
return -EIO;
return battr->write(of->file, kobj, battr, buf, pos, count);
}
/**
* kernfs_file_write - kernfs vfs write callback
* @file: file pointer
* @user_buf: data to write
* @count: number of bytes
* @ppos: starting offset
*
* Copy data in from userland and pass it to the matching kernfs write
* operation.
*
* 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 kernfs_file_write(struct file *file, const char __user *user_buf,
size_t count, loff_t *ppos)
{
struct sysfs_open_file *of = sysfs_of(file);
ssize_t len = min_t(size_t, count, PAGE_SIZE);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
const struct kernfs_ops *ops;
char *buf;
buf = kmalloc(len + 1, GFP_KERNEL);
if (!buf)
return -ENOMEM;
if (copy_from_user(buf, user_buf, len)) {
len = -EFAULT;
goto out_free;
}
buf[len] = '\0'; /* guarantee string termination */
/*
* @of->mutex nests outside active ref and is just to ensure that
* the ops aren't called concurrently for the same open file.
*/
mutex_lock(&of->mutex);
if (!sysfs_get_active(of->sd)) {
mutex_unlock(&of->mutex);
len = -ENODEV;
goto out_free;
}
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
ops = kernfs_ops(of->sd);
if (ops->write)
len = ops->write(of, buf, len, *ppos);
else
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
len = -EINVAL;
sysfs_put_active(of->sd);
mutex_unlock(&of->mutex);
if (len > 0)
*ppos += len;
out_free:
kfree(buf);
return len;
}
static int sysfs_kf_bin_mmap(struct sysfs_open_file *of,
struct vm_area_struct *vma)
{
struct bin_attribute *battr = of->sd->priv;
struct kobject *kobj = of->sd->s_parent->priv;
if (!battr->mmap)
return -ENODEV;
return battr->mmap(of->file, kobj, battr, vma);
}
static void kernfs_vma_open(struct vm_area_struct *vma)
{
struct file *file = vma->vm_file;
struct sysfs_open_file *of = sysfs_of(file);
if (!of->vm_ops)
return;
if (!sysfs_get_active(of->sd))
return;
if (of->vm_ops->open)
of->vm_ops->open(vma);
sysfs_put_active(of->sd);
}
static int kernfs_vma_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
struct file *file = vma->vm_file;
struct sysfs_open_file *of = sysfs_of(file);
int ret;
if (!of->vm_ops)
return VM_FAULT_SIGBUS;
if (!sysfs_get_active(of->sd))
return VM_FAULT_SIGBUS;
ret = VM_FAULT_SIGBUS;
if (of->vm_ops->fault)
ret = of->vm_ops->fault(vma, vmf);
sysfs_put_active(of->sd);
return ret;
}
static int kernfs_vma_page_mkwrite(struct vm_area_struct *vma,
struct vm_fault *vmf)
{
struct file *file = vma->vm_file;
struct sysfs_open_file *of = sysfs_of(file);
int ret;
if (!of->vm_ops)
return VM_FAULT_SIGBUS;
if (!sysfs_get_active(of->sd))
return VM_FAULT_SIGBUS;
ret = 0;
if (of->vm_ops->page_mkwrite)
ret = of->vm_ops->page_mkwrite(vma, vmf);
else
file_update_time(file);
sysfs_put_active(of->sd);
return ret;
}
static int kernfs_vma_access(struct vm_area_struct *vma, unsigned long addr,
void *buf, int len, int write)
{
struct file *file = vma->vm_file;
struct sysfs_open_file *of = sysfs_of(file);
int ret;
if (!of->vm_ops)
return -EINVAL;
if (!sysfs_get_active(of->sd))
return -EINVAL;
ret = -EINVAL;
if (of->vm_ops->access)
ret = of->vm_ops->access(vma, addr, buf, len, write);
sysfs_put_active(of->sd);
return ret;
}
#ifdef CONFIG_NUMA
static int kernfs_vma_set_policy(struct vm_area_struct *vma,
struct mempolicy *new)
{
struct file *file = vma->vm_file;
struct sysfs_open_file *of = sysfs_of(file);
int ret;
if (!of->vm_ops)
return 0;
if (!sysfs_get_active(of->sd))
return -EINVAL;
ret = 0;
if (of->vm_ops->set_policy)
ret = of->vm_ops->set_policy(vma, new);
sysfs_put_active(of->sd);
return ret;
}
static struct mempolicy *kernfs_vma_get_policy(struct vm_area_struct *vma,
unsigned long addr)
{
struct file *file = vma->vm_file;
struct sysfs_open_file *of = sysfs_of(file);
struct mempolicy *pol;
if (!of->vm_ops)
return vma->vm_policy;
if (!sysfs_get_active(of->sd))
return vma->vm_policy;
pol = vma->vm_policy;
if (of->vm_ops->get_policy)
pol = of->vm_ops->get_policy(vma, addr);
sysfs_put_active(of->sd);
return pol;
}
static int kernfs_vma_migrate(struct vm_area_struct *vma,
const nodemask_t *from, const nodemask_t *to,
unsigned long flags)
{
struct file *file = vma->vm_file;
struct sysfs_open_file *of = sysfs_of(file);
int ret;
if (!of->vm_ops)
return 0;
if (!sysfs_get_active(of->sd))
return 0;
ret = 0;
if (of->vm_ops->migrate)
ret = of->vm_ops->migrate(vma, from, to, flags);
sysfs_put_active(of->sd);
return ret;
}
#endif
static const struct vm_operations_struct kernfs_vm_ops = {
.open = kernfs_vma_open,
.fault = kernfs_vma_fault,
.page_mkwrite = kernfs_vma_page_mkwrite,
.access = kernfs_vma_access,
#ifdef CONFIG_NUMA
.set_policy = kernfs_vma_set_policy,
.get_policy = kernfs_vma_get_policy,
.migrate = kernfs_vma_migrate,
#endif
};
static int kernfs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
struct sysfs_open_file *of = sysfs_of(file);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
const struct kernfs_ops *ops;
int rc;
mutex_lock(&of->mutex);
rc = -ENODEV;
if (!sysfs_get_active(of->sd))
goto out_unlock;
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
ops = kernfs_ops(of->sd);
if (ops->mmap)
rc = ops->mmap(of, vma);
if (rc)
goto out_put;
/*
* PowerPC's pci_mmap of legacy_mem uses shmem_zero_setup()
* to satisfy versions of X which crash if the mmap fails: that
* substitutes a new vm_file, and we don't then want bin_vm_ops.
*/
if (vma->vm_file != file)
goto out_put;
rc = -EINVAL;
if (of->mmapped && of->vm_ops != vma->vm_ops)
goto out_put;
/*
* It is not possible to successfully wrap close.
* So error if someone is trying to use close.
*/
rc = -EINVAL;
if (vma->vm_ops && vma->vm_ops->close)
goto out_put;
rc = 0;
of->mmapped = 1;
of->vm_ops = vma->vm_ops;
vma->vm_ops = &kernfs_vm_ops;
out_put:
sysfs_put_active(of->sd);
out_unlock:
mutex_unlock(&of->mutex);
return rc;
}
/**
* sysfs_get_open_dirent - get or create sysfs_open_dirent
* @sd: target sysfs_dirent
* @of: sysfs_open_file for this instance of open
*
* If @sd->s_attr.open exists, increment its reference count;
* otherwise, create one. @of is chained to the files 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_open_file *of)
{
struct sysfs_open_dirent *od, *new_od = NULL;
retry:
mutex_lock(&sysfs_open_file_mutex);
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(&of->list, &od->files);
}
spin_unlock_irq(&sysfs_open_dirent_lock);
mutex_unlock(&sysfs_open_file_mutex);
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->files);
goto retry;
}
/**
* sysfs_put_open_dirent - put sysfs_open_dirent
* @sd: target sysfs_dirent
* @of: associated sysfs_open_file
*
* Put @sd->s_attr.open and unlink @of from the files list. If
* reference count reaches zero, disassociate and free it.
*
* LOCKING:
* None.
*/
static void sysfs_put_open_dirent(struct sysfs_dirent *sd,
struct sysfs_open_file *of)
{
struct sysfs_open_dirent *od = sd->s_attr.open;
unsigned long flags;
mutex_lock(&sysfs_open_file_mutex);
spin_lock_irqsave(&sysfs_open_dirent_lock, flags);
if (of)
list_del(&of->list);
if (atomic_dec_and_test(&od->refcnt))
sd->s_attr.open = NULL;
else
od = NULL;
spin_unlock_irqrestore(&sysfs_open_dirent_lock, flags);
mutex_unlock(&sysfs_open_file_mutex);
kfree(od);
}
static int kernfs_file_open(struct inode *inode, struct file *file)
{
struct sysfs_dirent *attr_sd = file->f_path.dentry->d_fsdata;
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
const struct kernfs_ops *ops;
struct sysfs_open_file *of;
sysfs: use a separate locking class for open files depending on mmap The following two commits implemented mmap support in the regular file path and merged bin file support into the regular path. 73d9714627ad ("sysfs: copy bin mmap support from fs/sysfs/bin.c to fs/sysfs/file.c") 3124eb1679b2 ("sysfs: merge regular and bin file handling") After the merge, the following commands trigger a spurious lockdep warning. "test-mmap-read" simply mmaps the file and dumps the content. $ cat /sys/block/sda/trace/act_mask $ test-mmap-read /sys/devices/pci0000\:00/0000\:00\:03.0/resource0 4096 ====================================================== [ INFO: possible circular locking dependency detected ] 3.12.0-work+ #378 Not tainted ------------------------------------------------------- test-mmap-read/567 is trying to acquire lock: (&of->mutex){+.+.+.}, at: [<ffffffff8120a8df>] sysfs_bin_mmap+0x4f/0x120 but task is already holding lock: (&mm->mmap_sem){++++++}, at: [<ffffffff8114b399>] vm_mmap_pgoff+0x49/0xa0 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #3 (&mm->mmap_sem){++++++}: ... -> #2 (sr_mutex){+.+.+.}: ... -> #1 (&bdev->bd_mutex){+.+.+.}: ... -> #0 (&of->mutex){+.+.+.}: ... other info that might help us debug this: Chain exists of: &of->mutex --> sr_mutex --> &mm->mmap_sem Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&mm->mmap_sem); lock(sr_mutex); lock(&mm->mmap_sem); lock(&of->mutex); *** DEADLOCK *** 1 lock held by test-mmap-read/567: #0: (&mm->mmap_sem){++++++}, at: [<ffffffff8114b399>] vm_mmap_pgoff+0x49/0xa0 stack backtrace: CPU: 3 PID: 567 Comm: test-mmap-read Not tainted 3.12.0-work+ #378 Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 ffffffff81ed41a0 ffff880009441bc8 ffffffff81611ad2 ffffffff81eccb80 ffff880009441c08 ffffffff8160f215 ffff880009441c60 ffff880009c75208 0000000000000000 ffff880009c751e0 ffff880009c75208 ffff880009c74ac0 Call Trace: [<ffffffff81611ad2>] dump_stack+0x4e/0x7a [<ffffffff8160f215>] print_circular_bug+0x2b0/0x2bf [<ffffffff8109ca0a>] __lock_acquire+0x1a3a/0x1e60 [<ffffffff8109d6ba>] lock_acquire+0x9a/0x1d0 [<ffffffff81615547>] mutex_lock_nested+0x67/0x3f0 [<ffffffff8120a8df>] sysfs_bin_mmap+0x4f/0x120 [<ffffffff8115d363>] mmap_region+0x3b3/0x5b0 [<ffffffff8115d8ae>] do_mmap_pgoff+0x34e/0x3d0 [<ffffffff8114b3ba>] vm_mmap_pgoff+0x6a/0xa0 [<ffffffff8115be3e>] SyS_mmap_pgoff+0xbe/0x250 [<ffffffff81008282>] SyS_mmap+0x22/0x30 [<ffffffff8161a4d2>] system_call_fastpath+0x16/0x1b This happens because one file nests sr_mutex, which nests mm->mmap_sem under it, under of->mutex while mmap implementation naturally nests of->mutex under mm->mmap_sem. The warning is false positive as of->mutex is per open-file and the two paths belong to two different files. This warning didn't trigger before regular and bin file supports were merged because only bin file supported mmap and the other side of locking happened only on regular files which used equivalent but separate locking. It'd be best if we give separate locking classes per file but we can't easily do that. Let's differentiate on ->mmap() for now. Later we'll add explicit file operations struct and can add per-ops lockdep key there. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Dave Jones <davej@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-17 02:17:36 +00:00
bool has_read, has_write, has_mmap;
int error = -EACCES;
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;
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
ops = kernfs_ops(attr_sd);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
has_read = ops->seq_show || ops->read || ops->mmap;
has_write = ops->write || ops->mmap;
has_mmap = ops->mmap;
/* check perms and supported operations */
if ((file->f_mode & FMODE_WRITE) &&
(!(inode->i_mode & S_IWUGO) || !has_write))
goto err_out;
if ((file->f_mode & FMODE_READ) &&
(!(inode->i_mode & S_IRUGO) || !has_read))
goto err_out;
/* allocate a sysfs_open_file for the file */
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;
of = kzalloc(sizeof(struct sysfs_open_file), GFP_KERNEL);
if (!of)
goto err_out;
sysfs: use a separate locking class for open files depending on mmap The following two commits implemented mmap support in the regular file path and merged bin file support into the regular path. 73d9714627ad ("sysfs: copy bin mmap support from fs/sysfs/bin.c to fs/sysfs/file.c") 3124eb1679b2 ("sysfs: merge regular and bin file handling") After the merge, the following commands trigger a spurious lockdep warning. "test-mmap-read" simply mmaps the file and dumps the content. $ cat /sys/block/sda/trace/act_mask $ test-mmap-read /sys/devices/pci0000\:00/0000\:00\:03.0/resource0 4096 ====================================================== [ INFO: possible circular locking dependency detected ] 3.12.0-work+ #378 Not tainted ------------------------------------------------------- test-mmap-read/567 is trying to acquire lock: (&of->mutex){+.+.+.}, at: [<ffffffff8120a8df>] sysfs_bin_mmap+0x4f/0x120 but task is already holding lock: (&mm->mmap_sem){++++++}, at: [<ffffffff8114b399>] vm_mmap_pgoff+0x49/0xa0 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #3 (&mm->mmap_sem){++++++}: ... -> #2 (sr_mutex){+.+.+.}: ... -> #1 (&bdev->bd_mutex){+.+.+.}: ... -> #0 (&of->mutex){+.+.+.}: ... other info that might help us debug this: Chain exists of: &of->mutex --> sr_mutex --> &mm->mmap_sem Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&mm->mmap_sem); lock(sr_mutex); lock(&mm->mmap_sem); lock(&of->mutex); *** DEADLOCK *** 1 lock held by test-mmap-read/567: #0: (&mm->mmap_sem){++++++}, at: [<ffffffff8114b399>] vm_mmap_pgoff+0x49/0xa0 stack backtrace: CPU: 3 PID: 567 Comm: test-mmap-read Not tainted 3.12.0-work+ #378 Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 ffffffff81ed41a0 ffff880009441bc8 ffffffff81611ad2 ffffffff81eccb80 ffff880009441c08 ffffffff8160f215 ffff880009441c60 ffff880009c75208 0000000000000000 ffff880009c751e0 ffff880009c75208 ffff880009c74ac0 Call Trace: [<ffffffff81611ad2>] dump_stack+0x4e/0x7a [<ffffffff8160f215>] print_circular_bug+0x2b0/0x2bf [<ffffffff8109ca0a>] __lock_acquire+0x1a3a/0x1e60 [<ffffffff8109d6ba>] lock_acquire+0x9a/0x1d0 [<ffffffff81615547>] mutex_lock_nested+0x67/0x3f0 [<ffffffff8120a8df>] sysfs_bin_mmap+0x4f/0x120 [<ffffffff8115d363>] mmap_region+0x3b3/0x5b0 [<ffffffff8115d8ae>] do_mmap_pgoff+0x34e/0x3d0 [<ffffffff8114b3ba>] vm_mmap_pgoff+0x6a/0xa0 [<ffffffff8115be3e>] SyS_mmap_pgoff+0xbe/0x250 [<ffffffff81008282>] SyS_mmap+0x22/0x30 [<ffffffff8161a4d2>] system_call_fastpath+0x16/0x1b This happens because one file nests sr_mutex, which nests mm->mmap_sem under it, under of->mutex while mmap implementation naturally nests of->mutex under mm->mmap_sem. The warning is false positive as of->mutex is per open-file and the two paths belong to two different files. This warning didn't trigger before regular and bin file supports were merged because only bin file supported mmap and the other side of locking happened only on regular files which used equivalent but separate locking. It'd be best if we give separate locking classes per file but we can't easily do that. Let's differentiate on ->mmap() for now. Later we'll add explicit file operations struct and can add per-ops lockdep key there. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Dave Jones <davej@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-17 02:17:36 +00:00
/*
* The following is done to give a different lockdep key to
* @of->mutex for files which implement mmap. This is a rather
* crude way to avoid false positive lockdep warning around
* mm->mmap_sem - mmap nests @of->mutex under mm->mmap_sem and
* reading /sys/block/sda/trace/act_mask grabs sr_mutex, under
* which mm->mmap_sem nests, while holding @of->mutex. As each
* open file has a separate mutex, it's okay as long as those don't
* happen on the same file. At this point, we can't easily give
* each file a separate locking class. Let's differentiate on
* whether the file has mmap or not for now.
*/
if (has_mmap)
mutex_init(&of->mutex);
else
mutex_init(&of->mutex);
of->sd = attr_sd;
of->file = file;
/*
* Always instantiate seq_file even if read access doesn't use
* seq_file or is not requested. This unifies private data access
* and readable regular files are the vast majority anyway.
*/
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
if (ops->seq_show)
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
error = seq_open(file, &kernfs_seq_ops);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
else
error = seq_open(file, NULL);
if (error)
goto err_free;
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
((struct seq_file *)file->private_data)->private = of;
/* seq_file clears PWRITE unconditionally, restore it if WRITE */
if (file->f_mode & FMODE_WRITE)
file->f_mode |= FMODE_PWRITE;
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
/* make sure we have open dirent struct */
error = sysfs_get_open_dirent(attr_sd, of);
if (error)
goto err_close;
/* 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_close:
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
seq_release(inode, file);
err_free:
kfree(of);
err_out:
sysfs_put_active(attr_sd);
return error;
}
static int kernfs_file_release(struct inode *inode, struct file *filp)
{
struct sysfs_dirent *sd = filp->f_path.dentry->d_fsdata;
struct sysfs_open_file *of = sysfs_of(filp);
sysfs_put_open_dirent(sd, of);
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
seq_release(inode, filp);
kfree(of);
return 0;
}
void sysfs_unmap_bin_file(struct sysfs_dirent *sd)
{
struct sysfs_open_dirent *od;
struct sysfs_open_file *of;
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
if (!(sd->s_flags & SYSFS_FLAG_HAS_MMAP))
return;
spin_lock_irq(&sysfs_open_dirent_lock);
od = sd->s_attr.open;
if (od)
atomic_inc(&od->refcnt);
spin_unlock_irq(&sysfs_open_dirent_lock);
if (!od)
return;
mutex_lock(&sysfs_open_file_mutex);
list_for_each_entry(of, &od->files, list) {
struct inode *inode = file_inode(of->file);
unmap_mapping_range(inode->i_mapping, 0, 0, 1);
}
mutex_unlock(&sysfs_open_file_mutex);
sysfs_put_open_dirent(sd, NULL);
}
/* 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 kernfs_file_poll(struct file *filp, poll_table *wait)
{
struct sysfs_open_file *of = sysfs_of(filp);
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 (of->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:
return DEFAULT_POLLMASK|POLLERR|POLLPRI;
}
/**
* kernfs_notify - notify a kernfs file
* @sd: file to notify
*
* Notify @sd such that poll(2) on @sd wakes up.
*/
void kernfs_notify(struct sysfs_dirent *sd)
{
struct sysfs_open_dirent *od;
unsigned long flags;
spin_lock_irqsave(&sysfs_open_dirent_lock, flags);
if (!WARN_ON(sysfs_type(sd) != SYSFS_KOBJ_ATTR)) {
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(kernfs_notify);
void sysfs_notify(struct kobject *k, const char *dir, const char *attr)
{
struct sysfs_dirent *sd = k->sd, *tmp;
if (sd && dir)
sd = sysfs_get_dirent(sd, dir);
else
sysfs_get(sd);
if (sd && attr) {
tmp = sysfs_get_dirent(sd, attr);
sysfs_put(sd);
sd = tmp;
}
if (sd) {
kernfs_notify(sd);
sysfs_put(sd);
}
}
EXPORT_SYMBOL_GPL(sysfs_notify);
const struct file_operations kernfs_file_operations = {
sysfs, kernfs: prepare read path for kernfs We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch rearranges read path so that the kernfs and sysfs parts are separate. * Regular file read path is refactored such that kernfs_seq_start/next/stop/show() handle all the boilerplate work including locking and updating event count for poll, while sysfs_kf_seq_show() deals with interaction with kobj show method. * Bin file read path is refactored such that kernfs_file_direct_read() handles all the boilerplate work including buffer management and locking, while sysfs_kf_bin_read() deals with interaction with bin_attribute read method. kernfs_file_read() is added. It invokes either the seq_file or direct read path depending on the file type. This will eventually allow using the same file_operations for both file types, which is necessary to separate out kernfs. While this patch changes the order of some operations, it shouldn't change any visible behavior. v2: Dropped unnecessary zeroing of @count from sysfs_kf_seq_show(). Add comments explaining single_open() behavior. Both suggested by Pavel. v3: seq_stop() is called even after seq_start() failed. kernfs_seq_start() updated so that it doesn't unlock sysfs_open_file->mutex on failure so that kernfs_seq_stop() doesn't try to unlock an already unlocked mutex. Reported by Fengguang. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Pavel Machek <pavel@ucw.cz> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:16 +00:00
.read = kernfs_file_read,
.write = kernfs_file_write,
sysfs: use generic_file_llseek() for sysfs_file_operations 13c589d5b0ac6 ("sysfs: use seq_file when reading regular files") converted regular sysfs files to use seq_file. The commit substituted generic_file_llseek() with seq_lseek() for llseek implementation. Before the change, all regular sysfs files were allowed to seek to any position in [0, PAGE_SIZE] as the file size is always PAGE_SIZE and generic_file_llseek() allows any seeking inside the range under file size; however, seq_lseek()'s behavior is different. It traverses the output by repeatedly invoking ->show() until it reaches the target offset or traversal indicates EOF. As seq_files are fully dynamic and may not end at all, it doesn't support seeking from the end (SEEK_END). Apparently, there are userland tools which uses SEEK_END to discover the buffer size to use and the switch to seq_lseek() disturbs them as SEEK_END fails with -EINVAL. The only benefits of using seq_lseek() instead of generic_file_llseek() are * Early failure. If traversing to certain file position should fail, seq_lseek() will report such failures on lseek(2) instead of the following read/write operations. * EOF detection. While SEEK_END is not supported, SEEK_SET/CUR + large offset can be used to detect eof - eof at the time of the seek anyway as the file size may change dynamically. Both aren't necessary for sysfs or prospect kernfs users. Revert to genefic_file_llseek() and preserve the original behavior. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Heiko Carstens <heiko.carstens@de.ibm.com> Link: https://lkml.kernel.org/r/20131031114358.GA5551@osiris Tested-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-01 17:16:53 +00:00
.llseek = generic_file_llseek,
.mmap = kernfs_file_mmap,
.open = kernfs_file_open,
.release = kernfs_file_release,
.poll = kernfs_file_poll,
};
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
static const struct kernfs_ops sysfs_file_kfops_empty = {
};
static const struct kernfs_ops sysfs_file_kfops_ro = {
.seq_show = sysfs_kf_seq_show,
};
static const struct kernfs_ops sysfs_file_kfops_wo = {
.write = sysfs_kf_write,
};
static const struct kernfs_ops sysfs_file_kfops_rw = {
.seq_show = sysfs_kf_seq_show,
.write = sysfs_kf_write,
};
static const struct kernfs_ops sysfs_bin_kfops_ro = {
.read = sysfs_kf_bin_read,
};
static const struct kernfs_ops sysfs_bin_kfops_wo = {
.write = sysfs_kf_bin_write,
};
static const struct kernfs_ops sysfs_bin_kfops_rw = {
.read = sysfs_kf_bin_read,
.write = sysfs_kf_bin_write,
.mmap = sysfs_kf_bin_mmap,
};
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
int sysfs_add_file_mode_ns(struct sysfs_dirent *dir_sd,
const struct attribute *attr, bool is_bin,
umode_t mode, const void *ns)
{
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
const struct kernfs_ops *ops;
struct sysfs_dirent *sd;
loff_t size;
if (!is_bin) {
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
struct kobject *kobj = dir_sd->priv;
const struct sysfs_ops *sysfs_ops = kobj->ktype->sysfs_ops;
/* every kobject with an attribute needs a ktype assigned */
if (WARN(!sysfs_ops, KERN_ERR
"missing sysfs attribute operations for kobject: %s\n",
kobject_name(kobj)))
return -EINVAL;
if (sysfs_ops->show && sysfs_ops->store)
ops = &sysfs_file_kfops_rw;
else if (sysfs_ops->show)
ops = &sysfs_file_kfops_ro;
else if (sysfs_ops->store)
ops = &sysfs_file_kfops_wo;
else
ops = &sysfs_file_kfops_empty;
size = PAGE_SIZE;
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
} else {
struct bin_attribute *battr = (void *)attr;
if ((battr->read && battr->write) || battr->mmap)
ops = &sysfs_bin_kfops_rw;
else if (battr->read)
ops = &sysfs_bin_kfops_ro;
else if (battr->write)
ops = &sysfs_bin_kfops_wo;
else
ops = &sysfs_file_kfops_empty;
size = battr->size;
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
}
sd = kernfs_create_file_ns(dir_sd, attr->name, mode, size,
ops, (void *)attr, ns);
if (IS_ERR(sd)) {
if (PTR_ERR(sd) == -EEXIST)
sysfs_warn_dup(dir_sd, attr->name);
return PTR_ERR(sd);
}
return 0;
}
/**
* kernfs_create_file_ns - create a file
* @parent: directory to create the file in
* @name: name of the file
* @mode: mode of the file
* @size: size of the file
* @ops: kernfs operations for the file
* @priv: private data for the file
* @ns: optional namespace tag of the file
*
* Returns the created node on success, ERR_PTR() value on error.
*/
struct sysfs_dirent *kernfs_create_file_ns(struct sysfs_dirent *parent,
const char *name,
umode_t mode, loff_t size,
const struct kernfs_ops *ops,
void *priv, const void *ns)
{
struct sysfs_addrm_cxt acxt;
struct sysfs_dirent *sd;
int rc;
sd = sysfs_new_dirent(name, (mode & S_IALLUGO) | S_IFREG,
SYSFS_KOBJ_ATTR);
if (!sd)
return ERR_PTR(-ENOMEM);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
sd->s_attr.ops = ops;
sd->s_attr.size = size;
sd->s_ns = ns;
sd->priv = priv;
sysfs_dirent_init_lockdep(sd);
sysfs, kernfs: introduce kernfs_ops We're in the process of separating out core sysfs functionality into kernfs which will deal with sysfs_dirents directly. This patch introduces kernfs_ops which hosts methods kernfs users implement and updates fs/sysfs/file.c such that sysfs_kf_*() functions populate kernfs_ops and kernfs_file_*() functions call the matching entries from kernfs_ops. kernfs_ops contains the following groups of methods. * seq_show() - for kernfs files which use seq_file for reads. * read() - for direct read implementations. Used iff seq_show() is not implemented. * write() - for writes. * mmap() - for mmaps. Notes: * sysfs_elem_attr->ops is added so that kernfs_ops can be accessed from sysfs_dirent. kernfs_ops() helper is added to verify locking and access the field. * SYSFS_FLAG_HAS_(SEQ_SHOW|MMAP) added. sd->s_attr->ops is accessible only while holding active_ref and there are cases where we want to take different actions depending on which ops are implemented. These two flags cache whether the two ops are implemented for those. * kernfs_file_*() no longer test sysfs type but chooses different behaviors depending on which methods in kernfs_ops are implemented. The conversions are trivial except for the open path. As kernfs_file_open() now decides whether to allow read/write accesses depending on the kernfs_ops implemented, the presence of methods in kobjs and attribute_bin should be propagated to kernfs_ops. sysfs_add_file_mode_ns() is updated so that it propagates presence / absence of the callbacks through _empty, _ro, _wo, _rw kernfs_ops. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-11-28 19:54:21 +00:00
/*
* sd->s_attr.ops is accesible only while holding active ref. We
* need to know whether some ops are implemented outside active
* ref. Cache their existence in flags.
*/
if (ops->seq_show)
sd->s_flags |= SYSFS_FLAG_HAS_SEQ_SHOW;
if (ops->mmap)
sd->s_flags |= SYSFS_FLAG_HAS_MMAP;
sysfs_addrm_start(&acxt);
rc = sysfs_add_one(&acxt, sd, parent);
sysfs_addrm_finish(&acxt);
if (rc) {
sysfs_put(sd);
return ERR_PTR(rc);
}
return sd;
}
int sysfs_add_file(struct sysfs_dirent *dir_sd, const struct attribute *attr,
bool is_bin)
{
return sysfs_add_file_mode_ns(dir_sd, attr, is_bin, attr->mode, NULL);
}
/**
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
* sysfs_create_file_ns - create an attribute file for an object with custom ns
* @kobj: object we're creating for
* @attr: attribute descriptor
* @ns: namespace the new file should belong to
*/
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
int sysfs_create_file_ns(struct kobject *kobj, const struct attribute *attr,
const void *ns)
{
BUG_ON(!kobj || !kobj->sd || !attr);
return sysfs_add_file_mode_ns(kobj->sd, attr, false, attr->mode, ns);
}
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
EXPORT_SYMBOL_GPL(sysfs_create_file_ns);
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;
}
EXPORT_SYMBOL_GPL(sysfs_create_files);
/**
* 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)
dir_sd = sysfs_get_dirent(kobj->sd, group);
else
dir_sd = sysfs_get(kobj->sd);
if (!dir_sd)
return -ENOENT;
error = sysfs_add_file(dir_sd, attr, false);
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, const struct attribute *attr,
umode_t mode)
{
struct sysfs_dirent *sd;
struct iattr newattrs;
int rc;
sd = sysfs_get_dirent(kobj->sd, attr->name);
if (!sd)
return -ENOENT;
newattrs.ia_mode = (mode & S_IALLUGO) | (sd->s_mode & ~S_IALLUGO);
newattrs.ia_valid = ATTR_MODE;
rc = kernfs_setattr(sd, &newattrs);
sysfs_put(sd);
return rc;
}
EXPORT_SYMBOL_GPL(sysfs_chmod_file);
/**
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
* sysfs_remove_file_ns - remove an object attribute with a custom ns tag
* @kobj: object we're acting for
* @attr: attribute descriptor
* @ns: namespace tag of the file to remove
*
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
* Hash the attribute name and namespace tag and kill the victim.
*/
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
void sysfs_remove_file_ns(struct kobject *kobj, const struct attribute *attr,
const void *ns)
{
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
struct sysfs_dirent *dir_sd = kobj->sd;
kernfs_remove_by_name_ns(dir_sd, attr->name, ns);
}
sysfs: make attr namespace interface less convoluted sysfs ns (namespace) implementation became more convoluted than necessary while trying to hide ns information from visible interface. The relatively recent attr ns support is a good example. * attr ns tag is determined by sysfs_ops->namespace() callback while dir tag is determined by kobj_type->namespace(). The placement is arbitrary. * Instead of performing operations with explicit ns tag, the namespace callback is routed through sysfs_attr_ns(), sysfs_ops->namespace(), class_attr_namespace(), class_attr->namespace(). It's not simpler in any sense. The only thing this convolution does is traversing the whole stack backwards. The namespace callbacks are unncessary because the operations involved are inherently synchronous. The information can be provided in in straight-forward top-down direction and reversing that direction is unnecessary and against basic design principles. This backward interface is unnecessarily convoluted and hinders properly separating out sysfs from driver model / kobject for proper layering. This patch updates attr ns support such that * sysfs_ops->namespace() and class_attr->namespace() are dropped. * sysfs_{create|remove}_file_ns(), which take explicit @ns param, are added and sysfs_{create|remove}_file() are now simple wrappers around the ns aware functions. * ns handling is dropped from sysfs_chmod_file(). Nobody uses it at this point. sysfs_chmod_file_ns() can be added later if necessary. * Explicit @ns is propagated through class_{create|remove}_file_ns() and netdev_class_{create|remove}_file_ns(). * driver/net/bonding which is currently the only user of attr namespace is updated to use netdev_class_{create|remove}_file_ns() with @bh->net as the ns tag instead of using the namespace callback. This patch should be an equivalent conversion without any functional difference. It makes the code easier to follow, reduces lines of code a bit and helps proper separation and layering. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Kay Sievers <kay@vrfy.org> Acked-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-09-12 02:29:04 +00:00
EXPORT_SYMBOL_GPL(sysfs_remove_file_ns);
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]);
}
EXPORT_SYMBOL_GPL(sysfs_remove_files);
/**
* 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)
dir_sd = sysfs_get_dirent(kobj->sd, group);
else
dir_sd = sysfs_get(kobj->sd);
if (dir_sd) {
kernfs_remove_by_name(dir_sd, attr->name);
sysfs_put(dir_sd);
}
}
EXPORT_SYMBOL_GPL(sysfs_remove_file_from_group);
/**
* sysfs_create_bin_file - create binary file for object.
* @kobj: object.
* @attr: attribute descriptor.
*/
int sysfs_create_bin_file(struct kobject *kobj,
const struct bin_attribute *attr)
{
BUG_ON(!kobj || !kobj->sd || !attr);
return sysfs_add_file(kobj->sd, &attr->attr, true);
}
EXPORT_SYMBOL_GPL(sysfs_create_bin_file);
/**
* sysfs_remove_bin_file - remove binary file for object.
* @kobj: object.
* @attr: attribute descriptor.
*/
void sysfs_remove_bin_file(struct kobject *kobj,
const struct bin_attribute *attr)
{
kernfs_remove_by_name(kobj->sd, attr->attr.name);
}
EXPORT_SYMBOL_GPL(sysfs_remove_bin_file);
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);