2005-04-16 18:20:36 -04:00
|
|
|
/*
|
|
|
|
* linux/fs/inode.c
|
|
|
|
*
|
|
|
|
* (C) 1997 Linus Torvalds
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/fs.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/dcache.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/quotaops.h>
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/writeback.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/backing-dev.h>
|
|
|
|
#include <linux/wait.h>
|
2009-09-22 19:43:50 -04:00
|
|
|
#include <linux/rwsem.h>
|
2005-04-16 18:20:36 -04:00
|
|
|
#include <linux/hash.h>
|
|
|
|
#include <linux/swap.h>
|
|
|
|
#include <linux/security.h>
|
|
|
|
#include <linux/pagemap.h>
|
|
|
|
#include <linux/cdev.h>
|
|
|
|
#include <linux/bootmem.h>
|
[PATCH] inotify
inotify is intended to correct the deficiencies of dnotify, particularly
its inability to scale and its terrible user interface:
* dnotify requires the opening of one fd per each directory
that you intend to watch. This quickly results in too many
open files and pins removable media, preventing unmount.
* dnotify is directory-based. You only learn about changes to
directories. Sure, a change to a file in a directory affects
the directory, but you are then forced to keep a cache of
stat structures.
* dnotify's interface to user-space is awful. Signals?
inotify provides a more usable, simple, powerful solution to file change
notification:
* inotify's interface is a system call that returns a fd, not SIGIO.
You get a single fd, which is select()-able.
* inotify has an event that says "the filesystem that the item
you were watching is on was unmounted."
* inotify can watch directories or files.
Inotify is currently used by Beagle (a desktop search infrastructure),
Gamin (a FAM replacement), and other projects.
See Documentation/filesystems/inotify.txt.
Signed-off-by: Robert Love <rml@novell.com>
Cc: John McCutchan <ttb@tentacle.dhs.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-12 17:06:03 -04:00
|
|
|
#include <linux/inotify.h>
|
2009-05-21 17:01:26 -04:00
|
|
|
#include <linux/fsnotify.h>
|
2006-01-09 23:52:17 -05:00
|
|
|
#include <linux/mount.h>
|
2009-01-06 10:20:54 -05:00
|
|
|
#include <linux/async.h>
|
2009-06-08 19:50:45 -04:00
|
|
|
#include <linux/posix_acl.h>
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* This is needed for the following functions:
|
|
|
|
* - inode_has_buffers
|
|
|
|
* - invalidate_inode_buffers
|
|
|
|
* - invalidate_bdev
|
|
|
|
*
|
|
|
|
* FIXME: remove all knowledge of the buffer layer from this file
|
|
|
|
*/
|
|
|
|
#include <linux/buffer_head.h>
|
|
|
|
|
|
|
|
/*
|
|
|
|
* New inode.c implementation.
|
|
|
|
*
|
|
|
|
* This implementation has the basic premise of trying
|
|
|
|
* to be extremely low-overhead and SMP-safe, yet be
|
|
|
|
* simple enough to be "obviously correct".
|
|
|
|
*
|
|
|
|
* Famous last words.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* inode dynamic allocation 1999, Andrea Arcangeli <andrea@suse.de> */
|
|
|
|
|
|
|
|
/* #define INODE_PARANOIA 1 */
|
|
|
|
/* #define INODE_DEBUG 1 */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Inode lookup is no longer as critical as it used to be:
|
|
|
|
* most of the lookups are going to be through the dcache.
|
|
|
|
*/
|
|
|
|
#define I_HASHBITS i_hash_shift
|
|
|
|
#define I_HASHMASK i_hash_mask
|
|
|
|
|
2006-03-26 04:37:24 -05:00
|
|
|
static unsigned int i_hash_mask __read_mostly;
|
|
|
|
static unsigned int i_hash_shift __read_mostly;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Each inode can be on two separate lists. One is
|
|
|
|
* the hash list of the inode, used for lookups. The
|
|
|
|
* other linked list is the "type" list:
|
|
|
|
* "in_use" - valid inode, i_count > 0, i_nlink > 0
|
|
|
|
* "dirty" - as "in_use" but also dirty
|
|
|
|
* "unused" - valid inode, i_count = 0
|
|
|
|
*
|
|
|
|
* A "dirty" list is maintained for each super block,
|
|
|
|
* allowing for low-overhead inode sync() operations.
|
|
|
|
*/
|
|
|
|
|
|
|
|
LIST_HEAD(inode_in_use);
|
|
|
|
LIST_HEAD(inode_unused);
|
2006-03-26 04:37:24 -05:00
|
|
|
static struct hlist_head *inode_hashtable __read_mostly;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* A simple spinlock to protect the list manipulations.
|
|
|
|
*
|
|
|
|
* NOTE! You also have to own the lock if you change
|
|
|
|
* the i_state of an inode while it is in use..
|
|
|
|
*/
|
|
|
|
DEFINE_SPINLOCK(inode_lock);
|
|
|
|
|
|
|
|
/*
|
2009-09-22 19:43:50 -04:00
|
|
|
* iprune_sem provides exclusion between the kswapd or try_to_free_pages
|
2005-04-16 18:20:36 -04:00
|
|
|
* icache shrinking path, and the umount path. Without this exclusion,
|
|
|
|
* by the time prune_icache calls iput for the inode whose pages it has
|
|
|
|
* been invalidating, or by the time it calls clear_inode & destroy_inode
|
|
|
|
* from its final dispose_list, the struct super_block they refer to
|
|
|
|
* (for inode->i_sb->s_op) may already have been freed and reused.
|
2009-09-22 19:43:50 -04:00
|
|
|
*
|
|
|
|
* We make this an rwsem because the fastpath is icache shrinking. In
|
|
|
|
* some cases a filesystem may be doing a significant amount of work in
|
|
|
|
* its inode reclaim code, so this should improve parallelism.
|
2005-04-16 18:20:36 -04:00
|
|
|
*/
|
2009-09-22 19:43:50 -04:00
|
|
|
static DECLARE_RWSEM(iprune_sem);
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Statistics gathering..
|
|
|
|
*/
|
|
|
|
struct inodes_stat_t inodes_stat;
|
|
|
|
|
2009-03-31 10:05:54 -04:00
|
|
|
static struct kmem_cache *inode_cachep __read_mostly;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
2007-10-17 02:30:44 -04:00
|
|
|
static void wake_up_inode(struct inode *inode)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Prevent speculative execution through spin_unlock(&inode_lock);
|
|
|
|
*/
|
|
|
|
smp_mb();
|
|
|
|
wake_up_bit(&inode->i_state, __I_LOCK);
|
|
|
|
}
|
|
|
|
|
2008-10-30 02:32:23 -04:00
|
|
|
/**
|
|
|
|
* inode_init_always - perform inode structure intialisation
|
2009-01-06 17:41:13 -05:00
|
|
|
* @sb: superblock inode belongs to
|
|
|
|
* @inode: inode to initialise
|
2008-10-30 02:32:23 -04:00
|
|
|
*
|
|
|
|
* These are initializations that need to be done on every inode
|
|
|
|
* allocation as the fields are not initialised by slab allocation.
|
|
|
|
*/
|
2009-08-07 13:38:25 -04:00
|
|
|
int inode_init_always(struct super_block *sb, struct inode *inode)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2006-06-28 07:26:44 -04:00
|
|
|
static const struct address_space_operations empty_aops;
|
2009-09-21 20:01:11 -04:00
|
|
|
static const struct inode_operations empty_iops;
|
2006-03-28 04:56:41 -05:00
|
|
|
static const struct file_operations empty_fops;
|
2009-03-31 10:05:54 -04:00
|
|
|
struct address_space *const mapping = &inode->i_data;
|
2008-10-30 02:32:23 -04:00
|
|
|
|
|
|
|
inode->i_sb = sb;
|
|
|
|
inode->i_blkbits = sb->s_blocksize_bits;
|
|
|
|
inode->i_flags = 0;
|
|
|
|
atomic_set(&inode->i_count, 1);
|
|
|
|
inode->i_op = &empty_iops;
|
|
|
|
inode->i_fop = &empty_fops;
|
|
|
|
inode->i_nlink = 1;
|
2008-12-09 09:34:39 -05:00
|
|
|
inode->i_uid = 0;
|
|
|
|
inode->i_gid = 0;
|
2008-10-30 02:32:23 -04:00
|
|
|
atomic_set(&inode->i_writecount, 0);
|
|
|
|
inode->i_size = 0;
|
|
|
|
inode->i_blocks = 0;
|
|
|
|
inode->i_bytes = 0;
|
|
|
|
inode->i_generation = 0;
|
2005-04-16 18:20:36 -04:00
|
|
|
#ifdef CONFIG_QUOTA
|
2008-10-30 02:32:23 -04:00
|
|
|
memset(&inode->i_dquot, 0, sizeof(inode->i_dquot));
|
2005-04-16 18:20:36 -04:00
|
|
|
#endif
|
2008-10-30 02:32:23 -04:00
|
|
|
inode->i_pipe = NULL;
|
|
|
|
inode->i_bdev = NULL;
|
|
|
|
inode->i_cdev = NULL;
|
|
|
|
inode->i_rdev = 0;
|
|
|
|
inode->dirtied_when = 0;
|
2009-02-04 09:06:57 -05:00
|
|
|
|
|
|
|
if (security_inode_alloc(inode))
|
2009-08-07 13:38:25 -04:00
|
|
|
goto out;
|
2008-10-30 02:32:23 -04:00
|
|
|
spin_lock_init(&inode->i_lock);
|
|
|
|
lockdep_set_class(&inode->i_lock, &sb->s_type->i_lock_key);
|
|
|
|
|
|
|
|
mutex_init(&inode->i_mutex);
|
|
|
|
lockdep_set_class(&inode->i_mutex, &sb->s_type->i_mutex_key);
|
|
|
|
|
|
|
|
init_rwsem(&inode->i_alloc_sem);
|
|
|
|
lockdep_set_class(&inode->i_alloc_sem, &sb->s_type->i_alloc_sem_key);
|
|
|
|
|
|
|
|
mapping->a_ops = &empty_aops;
|
|
|
|
mapping->host = inode;
|
|
|
|
mapping->flags = 0;
|
2009-01-06 17:39:23 -05:00
|
|
|
mapping_set_gfp_mask(mapping, GFP_HIGHUSER_MOVABLE);
|
2008-10-30 02:32:23 -04:00
|
|
|
mapping->assoc_mapping = NULL;
|
|
|
|
mapping->backing_dev_info = &default_backing_dev_info;
|
|
|
|
mapping->writeback_index = 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the block_device provides a backing_dev_info for client
|
|
|
|
* inodes then use that. Otherwise the inode share the bdev's
|
|
|
|
* backing_dev_info.
|
|
|
|
*/
|
|
|
|
if (sb->s_bdev) {
|
|
|
|
struct backing_dev_info *bdi;
|
|
|
|
|
2009-09-15 03:43:56 -04:00
|
|
|
bdi = sb->s_bdev->bd_inode->i_mapping->backing_dev_info;
|
2008-10-30 02:32:23 -04:00
|
|
|
mapping->backing_dev_info = bdi;
|
|
|
|
}
|
|
|
|
inode->i_private = NULL;
|
|
|
|
inode->i_mapping = mapping;
|
2009-06-08 19:50:45 -04:00
|
|
|
#ifdef CONFIG_FS_POSIX_ACL
|
|
|
|
inode->i_acl = inode->i_default_acl = ACL_NOT_CACHED;
|
|
|
|
#endif
|
2008-10-30 02:32:23 -04:00
|
|
|
|
2009-05-21 17:01:26 -04:00
|
|
|
#ifdef CONFIG_FSNOTIFY
|
|
|
|
inode->i_fsnotify_mask = 0;
|
|
|
|
#endif
|
|
|
|
|
2009-08-07 13:38:25 -04:00
|
|
|
return 0;
|
|
|
|
out:
|
|
|
|
return -ENOMEM;
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
2008-10-30 02:32:23 -04:00
|
|
|
EXPORT_SYMBOL(inode_init_always);
|
|
|
|
|
|
|
|
static struct inode *alloc_inode(struct super_block *sb)
|
|
|
|
{
|
|
|
|
struct inode *inode;
|
|
|
|
|
|
|
|
if (sb->s_op->alloc_inode)
|
|
|
|
inode = sb->s_op->alloc_inode(sb);
|
|
|
|
else
|
|
|
|
inode = kmem_cache_alloc(inode_cachep, GFP_KERNEL);
|
|
|
|
|
2009-08-07 13:38:25 -04:00
|
|
|
if (!inode)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
if (unlikely(inode_init_always(sb, inode))) {
|
|
|
|
if (inode->i_sb->s_op->destroy_inode)
|
|
|
|
inode->i_sb->s_op->destroy_inode(inode);
|
|
|
|
else
|
|
|
|
kmem_cache_free(inode_cachep, inode);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
return inode;
|
2008-10-30 02:32:23 -04:00
|
|
|
}
|
2005-04-16 18:20:36 -04:00
|
|
|
|
2009-08-07 13:38:29 -04:00
|
|
|
void __destroy_inode(struct inode *inode)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2006-04-02 07:38:18 -04:00
|
|
|
BUG_ON(inode_has_buffers(inode));
|
2005-04-16 18:20:36 -04:00
|
|
|
security_inode_free(inode);
|
2009-05-21 17:01:26 -04:00
|
|
|
fsnotify_inode_delete(inode);
|
2009-06-08 19:50:45 -04:00
|
|
|
#ifdef CONFIG_FS_POSIX_ACL
|
|
|
|
if (inode->i_acl && inode->i_acl != ACL_NOT_CACHED)
|
|
|
|
posix_acl_release(inode->i_acl);
|
|
|
|
if (inode->i_default_acl && inode->i_default_acl != ACL_NOT_CACHED)
|
|
|
|
posix_acl_release(inode->i_default_acl);
|
|
|
|
#endif
|
2009-08-07 13:38:29 -04:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__destroy_inode);
|
|
|
|
|
|
|
|
void destroy_inode(struct inode *inode)
|
|
|
|
{
|
|
|
|
__destroy_inode(inode);
|
2005-04-16 18:20:36 -04:00
|
|
|
if (inode->i_sb->s_op->destroy_inode)
|
|
|
|
inode->i_sb->s_op->destroy_inode(inode);
|
|
|
|
else
|
|
|
|
kmem_cache_free(inode_cachep, (inode));
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* These are initializations that only need to be done
|
|
|
|
* once, because the fields are idempotent across use
|
|
|
|
* of the inode, so let the slab aware of that.
|
|
|
|
*/
|
|
|
|
void inode_init_once(struct inode *inode)
|
|
|
|
{
|
|
|
|
memset(inode, 0, sizeof(*inode));
|
|
|
|
INIT_HLIST_NODE(&inode->i_hash);
|
|
|
|
INIT_LIST_HEAD(&inode->i_dentry);
|
|
|
|
INIT_LIST_HEAD(&inode->i_devices);
|
|
|
|
INIT_RADIX_TREE(&inode->i_data.page_tree, GFP_ATOMIC);
|
2008-07-25 22:45:32 -04:00
|
|
|
spin_lock_init(&inode->i_data.tree_lock);
|
2005-04-16 18:20:36 -04:00
|
|
|
spin_lock_init(&inode->i_data.i_mmap_lock);
|
|
|
|
INIT_LIST_HEAD(&inode->i_data.private_list);
|
|
|
|
spin_lock_init(&inode->i_data.private_lock);
|
|
|
|
INIT_RAW_PRIO_TREE_ROOT(&inode->i_data.i_mmap);
|
|
|
|
INIT_LIST_HEAD(&inode->i_data.i_mmap_nonlinear);
|
|
|
|
i_size_ordered_init(inode);
|
[PATCH] inotify
inotify is intended to correct the deficiencies of dnotify, particularly
its inability to scale and its terrible user interface:
* dnotify requires the opening of one fd per each directory
that you intend to watch. This quickly results in too many
open files and pins removable media, preventing unmount.
* dnotify is directory-based. You only learn about changes to
directories. Sure, a change to a file in a directory affects
the directory, but you are then forced to keep a cache of
stat structures.
* dnotify's interface to user-space is awful. Signals?
inotify provides a more usable, simple, powerful solution to file change
notification:
* inotify's interface is a system call that returns a fd, not SIGIO.
You get a single fd, which is select()-able.
* inotify has an event that says "the filesystem that the item
you were watching is on was unmounted."
* inotify can watch directories or files.
Inotify is currently used by Beagle (a desktop search infrastructure),
Gamin (a FAM replacement), and other projects.
See Documentation/filesystems/inotify.txt.
Signed-off-by: Robert Love <rml@novell.com>
Cc: John McCutchan <ttb@tentacle.dhs.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-12 17:06:03 -04:00
|
|
|
#ifdef CONFIG_INOTIFY
|
|
|
|
INIT_LIST_HEAD(&inode->inotify_watches);
|
2006-03-23 06:00:30 -05:00
|
|
|
mutex_init(&inode->inotify_mutex);
|
[PATCH] inotify
inotify is intended to correct the deficiencies of dnotify, particularly
its inability to scale and its terrible user interface:
* dnotify requires the opening of one fd per each directory
that you intend to watch. This quickly results in too many
open files and pins removable media, preventing unmount.
* dnotify is directory-based. You only learn about changes to
directories. Sure, a change to a file in a directory affects
the directory, but you are then forced to keep a cache of
stat structures.
* dnotify's interface to user-space is awful. Signals?
inotify provides a more usable, simple, powerful solution to file change
notification:
* inotify's interface is a system call that returns a fd, not SIGIO.
You get a single fd, which is select()-able.
* inotify has an event that says "the filesystem that the item
you were watching is on was unmounted."
* inotify can watch directories or files.
Inotify is currently used by Beagle (a desktop search infrastructure),
Gamin (a FAM replacement), and other projects.
See Documentation/filesystems/inotify.txt.
Signed-off-by: Robert Love <rml@novell.com>
Cc: John McCutchan <ttb@tentacle.dhs.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-12 17:06:03 -04:00
|
|
|
#endif
|
2009-05-21 17:01:26 -04:00
|
|
|
#ifdef CONFIG_FSNOTIFY
|
|
|
|
INIT_HLIST_HEAD(&inode->i_fsnotify_mark_entries);
|
|
|
|
#endif
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(inode_init_once);
|
|
|
|
|
2008-07-25 22:45:34 -04:00
|
|
|
static void init_once(void *foo)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2009-03-31 10:05:54 -04:00
|
|
|
struct inode *inode = (struct inode *) foo;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
2007-05-17 01:10:57 -04:00
|
|
|
inode_init_once(inode);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* inode_lock must be held
|
|
|
|
*/
|
2009-03-31 10:05:54 -04:00
|
|
|
void __iget(struct inode *inode)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
if (atomic_read(&inode->i_count)) {
|
|
|
|
atomic_inc(&inode->i_count);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
atomic_inc(&inode->i_count);
|
2007-10-17 02:30:44 -04:00
|
|
|
if (!(inode->i_state & (I_DIRTY|I_SYNC)))
|
2005-04-16 18:20:36 -04:00
|
|
|
list_move(&inode->i_list, &inode_in_use);
|
|
|
|
inodes_stat.nr_unused--;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clear_inode - clear an inode
|
|
|
|
* @inode: inode to clear
|
|
|
|
*
|
|
|
|
* This is called by the filesystem to tell us
|
|
|
|
* that the inode is no longer useful. We just
|
|
|
|
* terminate it with extreme prejudice.
|
|
|
|
*/
|
|
|
|
void clear_inode(struct inode *inode)
|
|
|
|
{
|
|
|
|
might_sleep();
|
|
|
|
invalidate_inode_buffers(inode);
|
2009-03-31 10:05:54 -04:00
|
|
|
|
2006-04-02 07:38:18 -04:00
|
|
|
BUG_ON(inode->i_data.nrpages);
|
|
|
|
BUG_ON(!(inode->i_state & I_FREEING));
|
|
|
|
BUG_ON(inode->i_state & I_CLEAR);
|
2007-10-17 02:30:44 -04:00
|
|
|
inode_sync_wait(inode);
|
2009-01-26 10:45:12 -05:00
|
|
|
vfs_dq_drop(inode);
|
2007-05-08 03:25:52 -04:00
|
|
|
if (inode->i_sb->s_op->clear_inode)
|
2005-04-16 18:20:36 -04:00
|
|
|
inode->i_sb->s_op->clear_inode(inode);
|
2006-09-27 04:50:48 -04:00
|
|
|
if (S_ISBLK(inode->i_mode) && inode->i_bdev)
|
2005-04-16 18:20:36 -04:00
|
|
|
bd_forget(inode);
|
2006-09-27 04:50:49 -04:00
|
|
|
if (S_ISCHR(inode->i_mode) && inode->i_cdev)
|
2005-04-16 18:20:36 -04:00
|
|
|
cd_forget(inode);
|
|
|
|
inode->i_state = I_CLEAR;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(clear_inode);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* dispose_list - dispose of the contents of a local list
|
|
|
|
* @head: the head of the list to free
|
|
|
|
*
|
|
|
|
* Dispose-list gets a local list with local inodes in it, so it doesn't
|
|
|
|
* need to worry about list corruption and SMP locks.
|
|
|
|
*/
|
|
|
|
static void dispose_list(struct list_head *head)
|
|
|
|
{
|
|
|
|
int nr_disposed = 0;
|
|
|
|
|
|
|
|
while (!list_empty(head)) {
|
|
|
|
struct inode *inode;
|
|
|
|
|
Introduce a handy list_first_entry macro
There are many places in the kernel where the construction like
foo = list_entry(head->next, struct foo_struct, list);
are used.
The code might look more descriptive and neat if using the macro
list_first_entry(head, type, member) \
list_entry((head)->next, type, member)
Here is the macro itself and the examples of its usage in the generic code.
If it will turn out to be useful, I can prepare the set of patches to
inject in into arch-specific code, drivers, networking, etc.
Signed-off-by: Pavel Emelianov <xemul@openvz.org>
Signed-off-by: Kirill Korotaev <dev@openvz.org>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Zach Brown <zach.brown@oracle.com>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: John McCutchan <ttb@tentacle.dhs.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Ram Pai <linuxram@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-05-08 03:30:19 -04:00
|
|
|
inode = list_first_entry(head, struct inode, i_list);
|
2005-04-16 18:20:36 -04:00
|
|
|
list_del(&inode->i_list);
|
|
|
|
|
|
|
|
if (inode->i_data.nrpages)
|
|
|
|
truncate_inode_pages(&inode->i_data, 0);
|
|
|
|
clear_inode(inode);
|
2005-07-12 16:58:12 -04:00
|
|
|
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
hlist_del_init(&inode->i_hash);
|
|
|
|
list_del_init(&inode->i_sb_list);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
|
|
|
|
wake_up_inode(inode);
|
2005-04-16 18:20:36 -04:00
|
|
|
destroy_inode(inode);
|
|
|
|
nr_disposed++;
|
|
|
|
}
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
inodes_stat.nr_inodes -= nr_disposed;
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Invalidate all inodes for a device.
|
|
|
|
*/
|
|
|
|
static int invalidate_list(struct list_head *head, struct list_head *dispose)
|
|
|
|
{
|
|
|
|
struct list_head *next;
|
|
|
|
int busy = 0, count = 0;
|
|
|
|
|
|
|
|
next = head->next;
|
|
|
|
for (;;) {
|
2009-03-31 10:05:54 -04:00
|
|
|
struct list_head *tmp = next;
|
|
|
|
struct inode *inode;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We can reschedule here without worrying about the list's
|
|
|
|
* consistency because the per-sb list of inodes must not
|
2009-09-22 19:43:50 -04:00
|
|
|
* change during umount anymore, and because iprune_sem keeps
|
2005-04-16 18:20:36 -04:00
|
|
|
* shrink_icache_memory() away.
|
|
|
|
*/
|
|
|
|
cond_resched_lock(&inode_lock);
|
|
|
|
|
|
|
|
next = next->next;
|
|
|
|
if (tmp == head)
|
|
|
|
break;
|
|
|
|
inode = list_entry(tmp, struct inode, i_sb_list);
|
2009-03-11 16:17:36 -04:00
|
|
|
if (inode->i_state & I_NEW)
|
|
|
|
continue;
|
2005-04-16 18:20:36 -04:00
|
|
|
invalidate_inode_buffers(inode);
|
|
|
|
if (!atomic_read(&inode->i_count)) {
|
|
|
|
list_move(&inode->i_list, dispose);
|
fs: new inode i_state corruption fix
There was a report of a data corruption
http://lkml.org/lkml/2008/11/14/121. There is a script included to
reproduce the problem.
During testing, I encountered a number of strange things with ext3, so I
tried ext2 to attempt to reduce complexity of the problem. I found that
fsstress would quickly hang in wait_on_inode, waiting for I_LOCK to be
cleared, even though instrumentation showed that unlock_new_inode had
already been called for that inode. This points to memory scribble, or
synchronisation problme.
i_state of I_NEW inodes is not protected by inode_lock because other
processes are not supposed to touch them until I_LOCK (and I_NEW) is
cleared. Adding WARN_ON(inode->i_state & I_NEW) to sites where we modify
i_state revealed that generic_sync_sb_inodes is picking up new inodes from
the inode lists and passing them to __writeback_single_inode without
waiting for I_NEW. Subsequently modifying i_state causes corruption. In
my case it would look like this:
CPU0 CPU1
unlock_new_inode() __sync_single_inode()
reg <- inode->i_state
reg -> reg & ~(I_LOCK|I_NEW) reg <- inode->i_state
reg -> inode->i_state reg -> reg | I_SYNC
reg -> inode->i_state
Non-atomic RMW on CPU1 overwrites CPU0 store and sets I_LOCK|I_NEW again.
Fix for this is rather than wait for I_NEW inodes, just skip over them:
inodes concurrently being created are not subject to data integrity
operations, and should not significantly contribute to dirty memory
either.
After this change, I'm unable to reproduce any of the added warnings or
hangs after ~1hour of running. Previously, the new warnings would start
immediately and hang would happen in under 5 minutes.
I'm also testing on ext3 now, and so far no problems there either. I
don't know whether this fixes the problem reported above, but it fixes a
real problem for me.
Cc: "Jorge Boncompte [DTI2]" <jorge@dti2.net>
Reported-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: <stable@kernel.org>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-03-12 17:31:38 -04:00
|
|
|
WARN_ON(inode->i_state & I_NEW);
|
2005-04-16 18:20:36 -04:00
|
|
|
inode->i_state |= I_FREEING;
|
|
|
|
count++;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
busy = 1;
|
|
|
|
}
|
|
|
|
/* only unused inodes may be cached with i_count zero */
|
|
|
|
inodes_stat.nr_unused -= count;
|
|
|
|
return busy;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* invalidate_inodes - discard the inodes on a device
|
|
|
|
* @sb: superblock
|
|
|
|
*
|
|
|
|
* Discard all of the inodes for a given superblock. If the discard
|
|
|
|
* fails because there are busy inodes then a non zero value is returned.
|
|
|
|
* If the discard is successful all the inodes have been discarded.
|
|
|
|
*/
|
2009-03-31 10:05:54 -04:00
|
|
|
int invalidate_inodes(struct super_block *sb)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
int busy;
|
|
|
|
LIST_HEAD(throw_away);
|
|
|
|
|
2009-09-22 19:43:50 -04:00
|
|
|
down_write(&iprune_sem);
|
2005-04-16 18:20:36 -04:00
|
|
|
spin_lock(&inode_lock);
|
[PATCH] inotify
inotify is intended to correct the deficiencies of dnotify, particularly
its inability to scale and its terrible user interface:
* dnotify requires the opening of one fd per each directory
that you intend to watch. This quickly results in too many
open files and pins removable media, preventing unmount.
* dnotify is directory-based. You only learn about changes to
directories. Sure, a change to a file in a directory affects
the directory, but you are then forced to keep a cache of
stat structures.
* dnotify's interface to user-space is awful. Signals?
inotify provides a more usable, simple, powerful solution to file change
notification:
* inotify's interface is a system call that returns a fd, not SIGIO.
You get a single fd, which is select()-able.
* inotify has an event that says "the filesystem that the item
you were watching is on was unmounted."
* inotify can watch directories or files.
Inotify is currently used by Beagle (a desktop search infrastructure),
Gamin (a FAM replacement), and other projects.
See Documentation/filesystems/inotify.txt.
Signed-off-by: Robert Love <rml@novell.com>
Cc: John McCutchan <ttb@tentacle.dhs.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-07-12 17:06:03 -04:00
|
|
|
inotify_unmount_inodes(&sb->s_inodes);
|
2009-05-21 17:01:58 -04:00
|
|
|
fsnotify_unmount_inodes(&sb->s_inodes);
|
2005-04-16 18:20:36 -04:00
|
|
|
busy = invalidate_list(&sb->s_inodes, &throw_away);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
|
|
|
|
dispose_list(&throw_away);
|
2009-09-22 19:43:50 -04:00
|
|
|
up_write(&iprune_sem);
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
return busy;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(invalidate_inodes);
|
|
|
|
|
|
|
|
static int can_unuse(struct inode *inode)
|
|
|
|
{
|
|
|
|
if (inode->i_state)
|
|
|
|
return 0;
|
|
|
|
if (inode_has_buffers(inode))
|
|
|
|
return 0;
|
|
|
|
if (atomic_read(&inode->i_count))
|
|
|
|
return 0;
|
|
|
|
if (inode->i_data.nrpages)
|
|
|
|
return 0;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Scan `goal' inodes on the unused list for freeable ones. They are moved to
|
|
|
|
* a temporary list and then are freed outside inode_lock by dispose_list().
|
|
|
|
*
|
|
|
|
* Any inodes which are pinned purely because of attached pagecache have their
|
|
|
|
* pagecache removed. We expect the final iput() on that inode to add it to
|
|
|
|
* the front of the inode_unused list. So look for it there and if the
|
|
|
|
* inode is still freeable, proceed. The right inode is found 99.9% of the
|
|
|
|
* time in testing on a 4-way.
|
|
|
|
*
|
|
|
|
* If the inode has metadata buffers attached to mapping->private_list then
|
|
|
|
* try to remove them.
|
|
|
|
*/
|
|
|
|
static void prune_icache(int nr_to_scan)
|
|
|
|
{
|
|
|
|
LIST_HEAD(freeable);
|
|
|
|
int nr_pruned = 0;
|
|
|
|
int nr_scanned;
|
|
|
|
unsigned long reap = 0;
|
|
|
|
|
2009-09-22 19:43:50 -04:00
|
|
|
down_read(&iprune_sem);
|
2005-04-16 18:20:36 -04:00
|
|
|
spin_lock(&inode_lock);
|
|
|
|
for (nr_scanned = 0; nr_scanned < nr_to_scan; nr_scanned++) {
|
|
|
|
struct inode *inode;
|
|
|
|
|
|
|
|
if (list_empty(&inode_unused))
|
|
|
|
break;
|
|
|
|
|
|
|
|
inode = list_entry(inode_unused.prev, struct inode, i_list);
|
|
|
|
|
|
|
|
if (inode->i_state || atomic_read(&inode->i_count)) {
|
|
|
|
list_move(&inode->i_list, &inode_unused);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (inode_has_buffers(inode) || inode->i_data.nrpages) {
|
|
|
|
__iget(inode);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
if (remove_inode_buffers(inode))
|
2007-02-10 04:45:39 -05:00
|
|
|
reap += invalidate_mapping_pages(&inode->i_data,
|
|
|
|
0, -1);
|
2005-04-16 18:20:36 -04:00
|
|
|
iput(inode);
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
|
|
|
|
if (inode != list_entry(inode_unused.next,
|
|
|
|
struct inode, i_list))
|
|
|
|
continue; /* wrong inode or list_empty */
|
|
|
|
if (!can_unuse(inode))
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
list_move(&inode->i_list, &freeable);
|
fs: new inode i_state corruption fix
There was a report of a data corruption
http://lkml.org/lkml/2008/11/14/121. There is a script included to
reproduce the problem.
During testing, I encountered a number of strange things with ext3, so I
tried ext2 to attempt to reduce complexity of the problem. I found that
fsstress would quickly hang in wait_on_inode, waiting for I_LOCK to be
cleared, even though instrumentation showed that unlock_new_inode had
already been called for that inode. This points to memory scribble, or
synchronisation problme.
i_state of I_NEW inodes is not protected by inode_lock because other
processes are not supposed to touch them until I_LOCK (and I_NEW) is
cleared. Adding WARN_ON(inode->i_state & I_NEW) to sites where we modify
i_state revealed that generic_sync_sb_inodes is picking up new inodes from
the inode lists and passing them to __writeback_single_inode without
waiting for I_NEW. Subsequently modifying i_state causes corruption. In
my case it would look like this:
CPU0 CPU1
unlock_new_inode() __sync_single_inode()
reg <- inode->i_state
reg -> reg & ~(I_LOCK|I_NEW) reg <- inode->i_state
reg -> inode->i_state reg -> reg | I_SYNC
reg -> inode->i_state
Non-atomic RMW on CPU1 overwrites CPU0 store and sets I_LOCK|I_NEW again.
Fix for this is rather than wait for I_NEW inodes, just skip over them:
inodes concurrently being created are not subject to data integrity
operations, and should not significantly contribute to dirty memory
either.
After this change, I'm unable to reproduce any of the added warnings or
hangs after ~1hour of running. Previously, the new warnings would start
immediately and hang would happen in under 5 minutes.
I'm also testing on ext3 now, and so far no problems there either. I
don't know whether this fixes the problem reported above, but it fixes a
real problem for me.
Cc: "Jorge Boncompte [DTI2]" <jorge@dti2.net>
Reported-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: <stable@kernel.org>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-03-12 17:31:38 -04:00
|
|
|
WARN_ON(inode->i_state & I_NEW);
|
2005-04-16 18:20:36 -04:00
|
|
|
inode->i_state |= I_FREEING;
|
|
|
|
nr_pruned++;
|
|
|
|
}
|
|
|
|
inodes_stat.nr_unused -= nr_pruned;
|
2006-06-30 04:55:45 -04:00
|
|
|
if (current_is_kswapd())
|
|
|
|
__count_vm_events(KSWAPD_INODESTEAL, reap);
|
|
|
|
else
|
|
|
|
__count_vm_events(PGINODESTEAL, reap);
|
2005-04-16 18:20:36 -04:00
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
|
|
|
|
dispose_list(&freeable);
|
2009-09-22 19:43:50 -04:00
|
|
|
up_read(&iprune_sem);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* shrink_icache_memory() will attempt to reclaim some unused inodes. Here,
|
|
|
|
* "unused" means that no dentries are referring to the inodes: the files are
|
|
|
|
* not open and the dcache references to those inodes have already been
|
|
|
|
* reclaimed.
|
|
|
|
*
|
|
|
|
* This function is passed the number of inodes to scan, and it returns the
|
|
|
|
* total number of remaining possibly-reclaimable inodes.
|
|
|
|
*/
|
2005-10-21 03:20:48 -04:00
|
|
|
static int shrink_icache_memory(int nr, gfp_t gfp_mask)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
if (nr) {
|
|
|
|
/*
|
|
|
|
* Nasty deadlock avoidance. We may hold various FS locks,
|
|
|
|
* and we don't want to recurse into the FS that called us
|
|
|
|
* in clear_inode() and friends..
|
2009-03-31 10:05:54 -04:00
|
|
|
*/
|
2005-04-16 18:20:36 -04:00
|
|
|
if (!(gfp_mask & __GFP_FS))
|
|
|
|
return -1;
|
|
|
|
prune_icache(nr);
|
|
|
|
}
|
|
|
|
return (inodes_stat.nr_unused / 100) * sysctl_vfs_cache_pressure;
|
|
|
|
}
|
|
|
|
|
2007-07-17 07:03:17 -04:00
|
|
|
static struct shrinker icache_shrinker = {
|
|
|
|
.shrink = shrink_icache_memory,
|
|
|
|
.seeks = DEFAULT_SEEKS,
|
|
|
|
};
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
static void __wait_on_freeing_inode(struct inode *inode);
|
|
|
|
/*
|
|
|
|
* Called with the inode lock held.
|
|
|
|
* NOTE: we are not increasing the inode-refcount, you must call __iget()
|
|
|
|
* by hand after calling find_inode now! This simplifies iunique and won't
|
|
|
|
* add any additional branch in the common code.
|
|
|
|
*/
|
2009-03-31 10:05:54 -04:00
|
|
|
static struct inode *find_inode(struct super_block *sb,
|
|
|
|
struct hlist_head *head,
|
|
|
|
int (*test)(struct inode *, void *),
|
|
|
|
void *data)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
struct hlist_node *node;
|
2009-03-31 10:05:54 -04:00
|
|
|
struct inode *inode = NULL;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
repeat:
|
2008-04-29 03:59:40 -04:00
|
|
|
hlist_for_each_entry(inode, node, head, i_hash) {
|
2005-04-16 18:20:36 -04:00
|
|
|
if (inode->i_sb != sb)
|
|
|
|
continue;
|
|
|
|
if (!test(inode, data))
|
|
|
|
continue;
|
2005-06-23 03:09:01 -04:00
|
|
|
if (inode->i_state & (I_FREEING|I_CLEAR|I_WILL_FREE)) {
|
2005-04-16 18:20:36 -04:00
|
|
|
__wait_on_freeing_inode(inode);
|
|
|
|
goto repeat;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
return node ? inode : NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* find_inode_fast is the fast path version of find_inode, see the comment at
|
|
|
|
* iget_locked for details.
|
|
|
|
*/
|
2009-03-31 10:05:54 -04:00
|
|
|
static struct inode *find_inode_fast(struct super_block *sb,
|
|
|
|
struct hlist_head *head, unsigned long ino)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
struct hlist_node *node;
|
2009-03-31 10:05:54 -04:00
|
|
|
struct inode *inode = NULL;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
repeat:
|
2008-04-29 03:59:40 -04:00
|
|
|
hlist_for_each_entry(inode, node, head, i_hash) {
|
2005-04-16 18:20:36 -04:00
|
|
|
if (inode->i_ino != ino)
|
|
|
|
continue;
|
|
|
|
if (inode->i_sb != sb)
|
|
|
|
continue;
|
2005-06-23 03:09:01 -04:00
|
|
|
if (inode->i_state & (I_FREEING|I_CLEAR|I_WILL_FREE)) {
|
2005-04-16 18:20:36 -04:00
|
|
|
__wait_on_freeing_inode(inode);
|
|
|
|
goto repeat;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
return node ? inode : NULL;
|
|
|
|
}
|
|
|
|
|
2008-10-30 02:35:24 -04:00
|
|
|
static unsigned long hash(struct super_block *sb, unsigned long hashval)
|
|
|
|
{
|
|
|
|
unsigned long tmp;
|
|
|
|
|
|
|
|
tmp = (hashval * (unsigned long)sb) ^ (GOLDEN_RATIO_PRIME + hashval) /
|
|
|
|
L1_CACHE_BYTES;
|
|
|
|
tmp = tmp ^ ((tmp ^ GOLDEN_RATIO_PRIME) >> I_HASHBITS);
|
|
|
|
return tmp & I_HASHMASK;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
__inode_add_to_lists(struct super_block *sb, struct hlist_head *head,
|
|
|
|
struct inode *inode)
|
|
|
|
{
|
|
|
|
inodes_stat.nr_inodes++;
|
|
|
|
list_add(&inode->i_list, &inode_in_use);
|
|
|
|
list_add(&inode->i_sb_list, &sb->s_inodes);
|
|
|
|
if (head)
|
|
|
|
hlist_add_head(&inode->i_hash, head);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* inode_add_to_lists - add a new inode to relevant lists
|
2009-01-06 17:41:13 -05:00
|
|
|
* @sb: superblock inode belongs to
|
|
|
|
* @inode: inode to mark in use
|
2008-10-30 02:35:24 -04:00
|
|
|
*
|
|
|
|
* When an inode is allocated it needs to be accounted for, added to the in use
|
|
|
|
* list, the owning superblock and the inode hash. This needs to be done under
|
|
|
|
* the inode_lock, so export a function to do this rather than the inode lock
|
|
|
|
* itself. We calculate the hash list to add to here so it is all internal
|
|
|
|
* which requires the caller to have already set up the inode number in the
|
|
|
|
* inode to add.
|
|
|
|
*/
|
|
|
|
void inode_add_to_lists(struct super_block *sb, struct inode *inode)
|
|
|
|
{
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(sb, inode->i_ino);
|
|
|
|
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
__inode_add_to_lists(sb, head, inode);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(inode_add_to_lists);
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
/**
|
|
|
|
* new_inode - obtain an inode
|
|
|
|
* @sb: superblock
|
|
|
|
*
|
2007-07-17 07:03:05 -04:00
|
|
|
* Allocates a new inode for given superblock. The default gfp_mask
|
2009-01-06 17:39:23 -05:00
|
|
|
* for allocations related to inode->i_mapping is GFP_HIGHUSER_MOVABLE.
|
2007-07-17 07:03:05 -04:00
|
|
|
* If HIGHMEM pages are unsuitable or it is known that pages allocated
|
|
|
|
* for the page cache are not reclaimable or migratable,
|
|
|
|
* mapping_set_gfp_mask() must be called with suitable flags on the
|
|
|
|
* newly created inode's mapping
|
|
|
|
*
|
2005-04-16 18:20:36 -04:00
|
|
|
*/
|
|
|
|
struct inode *new_inode(struct super_block *sb)
|
|
|
|
{
|
inode numbering: make static counters in new_inode and iunique be 32 bits
The problems are:
- on filesystems w/o permanent inode numbers, i_ino values can be larger
than 32 bits, which can cause problems for some 32 bit userspace programs on
a 64 bit kernel. We can't do anything for filesystems that have actual
>32-bit inode numbers, but on filesystems that generate i_ino values on the
fly, we should try to have them fit in 32 bits. We could trivially fix this
by making the static counters in new_inode and iunique 32 bits, but...
- many filesystems call new_inode and assume that the i_ino values they are
given are unique. They are not guaranteed to be so, since the static
counter can wrap. This problem is exacerbated by the fix for #1.
- after allocating a new inode, some filesystems call iunique to try to get
a unique i_ino value, but they don't actually add their inodes to the
hashtable, and so they're still not guaranteed to be unique if that counter
wraps.
This patch set takes the simpler approach of simply using iunique and hashing
the inodes afterward. Christoph H. previously mentioned that he thought that
this approach may slow down lookups for filesystems that currently hash their
inodes.
The questions are:
1) how much would this slow down lookups for these filesystems?
2) is it enough to justify adding more infrastructure to avoid it?
What might be best is to start with this approach and then only move to using
IDR or some other scheme if these extra inodes in the hashtable prove to be
problematic.
I've done some cursory testing with this patch and the overhead of hashing and
unhashing the inodes with pipefs is pretty low -- just a few seconds of system
time added on to the creation and destruction of 10 million pipes (very
similar to the overhead that the IDR approach would add).
The hard thing to measure is what effect this has on other filesystems. I'm
open to ways to try and gauge this.
Again, I've only converted pipefs as an example. If this approach is
acceptable then I'll start work on patches to convert other filesystems.
With a pretty-much-worst-case microbenchmark provided by Eric Dumazet
<dada1@cosmosbay.com>:
hashing patch (pipebench):
sys 1m15.329s
sys 1m16.249s
sys 1m17.169s
unpatched (pipebench):
sys 1m9.836s
sys 1m12.541s
sys 1m14.153s
Which works out to 1.05642174294555027017. So ~5-6% slowdown.
This patch:
When a 32-bit program that was not compiled with large file offsets does a
stat and gets a st_ino value back that won't fit in the 32 bit field, glibc
(correctly) generates an EOVERFLOW error. We can't do anything about fs's
with larger permanent inode numbers, but when we generate them on the fly, we
ought to try and have them fit within a 32 bit field.
This patch takes the first step toward this by making the static counters in
these two functions be 32 bits.
[jlayton@redhat.com: mention that it's only the case for 32bit, non-LFS stat]
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-05-08 03:32:29 -04:00
|
|
|
/*
|
|
|
|
* On a 32bit, non LFS stat() call, glibc will generate an EOVERFLOW
|
|
|
|
* error if st_ino won't fit in target struct field. Use 32bit counter
|
|
|
|
* here to attempt to avoid that.
|
|
|
|
*/
|
|
|
|
static unsigned int last_ino;
|
2009-03-31 10:05:54 -04:00
|
|
|
struct inode *inode;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
spin_lock_prefetch(&inode_lock);
|
2009-03-31 10:05:54 -04:00
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
inode = alloc_inode(sb);
|
|
|
|
if (inode) {
|
|
|
|
spin_lock(&inode_lock);
|
2008-10-30 02:35:24 -04:00
|
|
|
__inode_add_to_lists(sb, NULL, inode);
|
2005-04-16 18:20:36 -04:00
|
|
|
inode->i_ino = ++last_ino;
|
|
|
|
inode->i_state = 0;
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
}
|
|
|
|
return inode;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(new_inode);
|
|
|
|
|
|
|
|
void unlock_new_inode(struct inode *inode)
|
|
|
|
{
|
2007-10-13 19:38:33 -04:00
|
|
|
#ifdef CONFIG_DEBUG_LOCK_ALLOC
|
2007-10-16 00:47:54 -04:00
|
|
|
if (inode->i_mode & S_IFDIR) {
|
|
|
|
struct file_system_type *type = inode->i_sb->s_type;
|
|
|
|
|
2009-06-04 09:26:49 -04:00
|
|
|
/* Set new key only if filesystem hasn't already changed it */
|
|
|
|
if (!lockdep_match_class(&inode->i_mutex,
|
|
|
|
&type->i_mutex_key)) {
|
|
|
|
/*
|
|
|
|
* ensure nobody is actually holding i_mutex
|
|
|
|
*/
|
|
|
|
mutex_destroy(&inode->i_mutex);
|
|
|
|
mutex_init(&inode->i_mutex);
|
|
|
|
lockdep_set_class(&inode->i_mutex,
|
|
|
|
&type->i_mutex_dir_key);
|
|
|
|
}
|
2007-10-16 00:47:54 -04:00
|
|
|
}
|
2007-10-13 19:38:33 -04:00
|
|
|
#endif
|
2005-04-16 18:20:36 -04:00
|
|
|
/*
|
2009-09-21 20:01:06 -04:00
|
|
|
* This is special! We do not need the spinlock when clearing I_LOCK,
|
|
|
|
* because we're guaranteed that nobody else tries to do anything about
|
|
|
|
* the state of the inode when it is locked, as we just created it (so
|
|
|
|
* there can be no old holders that haven't tested I_LOCK).
|
|
|
|
* However we must emit the memory barrier so that other CPUs reliably
|
|
|
|
* see the clearing of I_LOCK after the other inode initialisation has
|
|
|
|
* completed.
|
2005-04-16 18:20:36 -04:00
|
|
|
*/
|
2009-09-21 20:01:06 -04:00
|
|
|
smp_mb();
|
fs: new inode i_state corruption fix
There was a report of a data corruption
http://lkml.org/lkml/2008/11/14/121. There is a script included to
reproduce the problem.
During testing, I encountered a number of strange things with ext3, so I
tried ext2 to attempt to reduce complexity of the problem. I found that
fsstress would quickly hang in wait_on_inode, waiting for I_LOCK to be
cleared, even though instrumentation showed that unlock_new_inode had
already been called for that inode. This points to memory scribble, or
synchronisation problme.
i_state of I_NEW inodes is not protected by inode_lock because other
processes are not supposed to touch them until I_LOCK (and I_NEW) is
cleared. Adding WARN_ON(inode->i_state & I_NEW) to sites where we modify
i_state revealed that generic_sync_sb_inodes is picking up new inodes from
the inode lists and passing them to __writeback_single_inode without
waiting for I_NEW. Subsequently modifying i_state causes corruption. In
my case it would look like this:
CPU0 CPU1
unlock_new_inode() __sync_single_inode()
reg <- inode->i_state
reg -> reg & ~(I_LOCK|I_NEW) reg <- inode->i_state
reg -> inode->i_state reg -> reg | I_SYNC
reg -> inode->i_state
Non-atomic RMW on CPU1 overwrites CPU0 store and sets I_LOCK|I_NEW again.
Fix for this is rather than wait for I_NEW inodes, just skip over them:
inodes concurrently being created are not subject to data integrity
operations, and should not significantly contribute to dirty memory
either.
After this change, I'm unable to reproduce any of the added warnings or
hangs after ~1hour of running. Previously, the new warnings would start
immediately and hang would happen in under 5 minutes.
I'm also testing on ext3 now, and so far no problems there either. I
don't know whether this fixes the problem reported above, but it fixes a
real problem for me.
Cc: "Jorge Boncompte [DTI2]" <jorge@dti2.net>
Reported-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: <stable@kernel.org>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-03-12 17:31:38 -04:00
|
|
|
WARN_ON((inode->i_state & (I_LOCK|I_NEW)) != (I_LOCK|I_NEW));
|
2005-04-16 18:20:36 -04:00
|
|
|
inode->i_state &= ~(I_LOCK|I_NEW);
|
|
|
|
wake_up_inode(inode);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(unlock_new_inode);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This is called without the inode lock held.. Be careful.
|
|
|
|
*
|
|
|
|
* We no longer cache the sb_flags in i_flags - see fs.h
|
|
|
|
* -- rmk@arm.uk.linux.org
|
|
|
|
*/
|
2009-03-31 10:05:54 -04:00
|
|
|
static struct inode *get_new_inode(struct super_block *sb,
|
|
|
|
struct hlist_head *head,
|
|
|
|
int (*test)(struct inode *, void *),
|
|
|
|
int (*set)(struct inode *, void *),
|
|
|
|
void *data)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2009-03-31 10:05:54 -04:00
|
|
|
struct inode *inode;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
inode = alloc_inode(sb);
|
|
|
|
if (inode) {
|
2009-03-31 10:05:54 -04:00
|
|
|
struct inode *old;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
/* We released the lock, so.. */
|
|
|
|
old = find_inode(sb, head, test, data);
|
|
|
|
if (!old) {
|
|
|
|
if (set(inode, data))
|
|
|
|
goto set_failed;
|
|
|
|
|
2008-10-30 02:35:24 -04:00
|
|
|
__inode_add_to_lists(sb, head, inode);
|
2005-04-16 18:20:36 -04:00
|
|
|
inode->i_state = I_LOCK|I_NEW;
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
|
|
|
|
/* Return the locked inode with I_NEW set, the
|
|
|
|
* caller is responsible for filling in the contents
|
|
|
|
*/
|
|
|
|
return inode;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Uhhuh, somebody else created the same inode under
|
|
|
|
* us. Use the old inode instead of the one we just
|
|
|
|
* allocated.
|
|
|
|
*/
|
|
|
|
__iget(old);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
destroy_inode(inode);
|
|
|
|
inode = old;
|
|
|
|
wait_on_inode(inode);
|
|
|
|
}
|
|
|
|
return inode;
|
|
|
|
|
|
|
|
set_failed:
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
destroy_inode(inode);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* get_new_inode_fast is the fast path version of get_new_inode, see the
|
|
|
|
* comment at iget_locked for details.
|
|
|
|
*/
|
2009-03-31 10:05:54 -04:00
|
|
|
static struct inode *get_new_inode_fast(struct super_block *sb,
|
|
|
|
struct hlist_head *head, unsigned long ino)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2009-03-31 10:05:54 -04:00
|
|
|
struct inode *inode;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
inode = alloc_inode(sb);
|
|
|
|
if (inode) {
|
2009-03-31 10:05:54 -04:00
|
|
|
struct inode *old;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
/* We released the lock, so.. */
|
|
|
|
old = find_inode_fast(sb, head, ino);
|
|
|
|
if (!old) {
|
|
|
|
inode->i_ino = ino;
|
2008-10-30 02:35:24 -04:00
|
|
|
__inode_add_to_lists(sb, head, inode);
|
2005-04-16 18:20:36 -04:00
|
|
|
inode->i_state = I_LOCK|I_NEW;
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
|
|
|
|
/* Return the locked inode with I_NEW set, the
|
|
|
|
* caller is responsible for filling in the contents
|
|
|
|
*/
|
|
|
|
return inode;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Uhhuh, somebody else created the same inode under
|
|
|
|
* us. Use the old inode instead of the one we just
|
|
|
|
* allocated.
|
|
|
|
*/
|
|
|
|
__iget(old);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
destroy_inode(inode);
|
|
|
|
inode = old;
|
|
|
|
wait_on_inode(inode);
|
|
|
|
}
|
|
|
|
return inode;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* iunique - get a unique inode number
|
|
|
|
* @sb: superblock
|
|
|
|
* @max_reserved: highest reserved inode number
|
|
|
|
*
|
|
|
|
* Obtain an inode number that is unique on the system for a given
|
|
|
|
* superblock. This is used by file systems that have no natural
|
|
|
|
* permanent inode numbering system. An inode number is returned that
|
|
|
|
* is higher than the reserved limit but unique.
|
|
|
|
*
|
|
|
|
* BUGS:
|
|
|
|
* With a large number of inodes live on the file system this function
|
|
|
|
* currently becomes quite slow.
|
|
|
|
*/
|
|
|
|
ino_t iunique(struct super_block *sb, ino_t max_reserved)
|
|
|
|
{
|
inode numbering: make static counters in new_inode and iunique be 32 bits
The problems are:
- on filesystems w/o permanent inode numbers, i_ino values can be larger
than 32 bits, which can cause problems for some 32 bit userspace programs on
a 64 bit kernel. We can't do anything for filesystems that have actual
>32-bit inode numbers, but on filesystems that generate i_ino values on the
fly, we should try to have them fit in 32 bits. We could trivially fix this
by making the static counters in new_inode and iunique 32 bits, but...
- many filesystems call new_inode and assume that the i_ino values they are
given are unique. They are not guaranteed to be so, since the static
counter can wrap. This problem is exacerbated by the fix for #1.
- after allocating a new inode, some filesystems call iunique to try to get
a unique i_ino value, but they don't actually add their inodes to the
hashtable, and so they're still not guaranteed to be unique if that counter
wraps.
This patch set takes the simpler approach of simply using iunique and hashing
the inodes afterward. Christoph H. previously mentioned that he thought that
this approach may slow down lookups for filesystems that currently hash their
inodes.
The questions are:
1) how much would this slow down lookups for these filesystems?
2) is it enough to justify adding more infrastructure to avoid it?
What might be best is to start with this approach and then only move to using
IDR or some other scheme if these extra inodes in the hashtable prove to be
problematic.
I've done some cursory testing with this patch and the overhead of hashing and
unhashing the inodes with pipefs is pretty low -- just a few seconds of system
time added on to the creation and destruction of 10 million pipes (very
similar to the overhead that the IDR approach would add).
The hard thing to measure is what effect this has on other filesystems. I'm
open to ways to try and gauge this.
Again, I've only converted pipefs as an example. If this approach is
acceptable then I'll start work on patches to convert other filesystems.
With a pretty-much-worst-case microbenchmark provided by Eric Dumazet
<dada1@cosmosbay.com>:
hashing patch (pipebench):
sys 1m15.329s
sys 1m16.249s
sys 1m17.169s
unpatched (pipebench):
sys 1m9.836s
sys 1m12.541s
sys 1m14.153s
Which works out to 1.05642174294555027017. So ~5-6% slowdown.
This patch:
When a 32-bit program that was not compiled with large file offsets does a
stat and gets a st_ino value back that won't fit in the 32 bit field, glibc
(correctly) generates an EOVERFLOW error. We can't do anything about fs's
with larger permanent inode numbers, but when we generate them on the fly, we
ought to try and have them fit within a 32 bit field.
This patch takes the first step toward this by making the static counters in
these two functions be 32 bits.
[jlayton@redhat.com: mention that it's only the case for 32bit, non-LFS stat]
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-05-08 03:32:29 -04:00
|
|
|
/*
|
|
|
|
* On a 32bit, non LFS stat() call, glibc will generate an EOVERFLOW
|
|
|
|
* error if st_ino won't fit in target struct field. Use 32bit counter
|
|
|
|
* here to attempt to avoid that.
|
|
|
|
*/
|
|
|
|
static unsigned int counter;
|
2005-04-16 18:20:36 -04:00
|
|
|
struct inode *inode;
|
2007-05-08 03:29:48 -04:00
|
|
|
struct hlist_head *head;
|
2005-04-16 18:20:36 -04:00
|
|
|
ino_t res;
|
2007-05-08 03:29:48 -04:00
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
spin_lock(&inode_lock);
|
2007-05-08 03:29:48 -04:00
|
|
|
do {
|
|
|
|
if (counter <= max_reserved)
|
|
|
|
counter = max_reserved + 1;
|
2005-04-16 18:20:36 -04:00
|
|
|
res = counter++;
|
2007-05-08 03:29:48 -04:00
|
|
|
head = inode_hashtable + hash(sb, res);
|
2005-04-16 18:20:36 -04:00
|
|
|
inode = find_inode_fast(sb, head, res);
|
2007-05-08 03:29:48 -04:00
|
|
|
} while (inode != NULL);
|
|
|
|
spin_unlock(&inode_lock);
|
2005-04-16 18:20:36 -04:00
|
|
|
|
2007-05-08 03:29:48 -04:00
|
|
|
return res;
|
|
|
|
}
|
2005-04-16 18:20:36 -04:00
|
|
|
EXPORT_SYMBOL(iunique);
|
|
|
|
|
|
|
|
struct inode *igrab(struct inode *inode)
|
|
|
|
{
|
|
|
|
spin_lock(&inode_lock);
|
2007-02-10 04:44:59 -05:00
|
|
|
if (!(inode->i_state & (I_FREEING|I_CLEAR|I_WILL_FREE)))
|
2005-04-16 18:20:36 -04:00
|
|
|
__iget(inode);
|
|
|
|
else
|
|
|
|
/*
|
|
|
|
* Handle the case where s_op->clear_inode is not been
|
|
|
|
* called yet, and somebody is calling igrab
|
|
|
|
* while the inode is getting freed.
|
|
|
|
*/
|
|
|
|
inode = NULL;
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
return inode;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(igrab);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* ifind - internal function, you want ilookup5() or iget5().
|
|
|
|
* @sb: super block of file system to search
|
|
|
|
* @head: the head of the list to search
|
|
|
|
* @test: callback used for comparisons between inodes
|
|
|
|
* @data: opaque data pointer to pass to @test
|
2005-07-13 04:10:44 -04:00
|
|
|
* @wait: if true wait for the inode to be unlocked, if false do not
|
2005-04-16 18:20:36 -04:00
|
|
|
*
|
|
|
|
* ifind() searches for the inode specified by @data in the inode
|
|
|
|
* cache. This is a generalized version of ifind_fast() for file systems where
|
|
|
|
* the inode number is not sufficient for unique identification of an inode.
|
|
|
|
*
|
|
|
|
* If the inode is in the cache, the inode is returned with an incremented
|
|
|
|
* reference count.
|
|
|
|
*
|
|
|
|
* Otherwise NULL is returned.
|
|
|
|
*
|
|
|
|
* Note, @test is called with the inode_lock held, so can't sleep.
|
|
|
|
*/
|
2006-01-08 04:05:21 -05:00
|
|
|
static struct inode *ifind(struct super_block *sb,
|
2005-04-16 18:20:36 -04:00
|
|
|
struct hlist_head *head, int (*test)(struct inode *, void *),
|
2005-07-13 04:10:44 -04:00
|
|
|
void *data, const int wait)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
struct inode *inode;
|
|
|
|
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
inode = find_inode(sb, head, test, data);
|
|
|
|
if (inode) {
|
|
|
|
__iget(inode);
|
|
|
|
spin_unlock(&inode_lock);
|
2005-07-13 04:10:44 -04:00
|
|
|
if (likely(wait))
|
|
|
|
wait_on_inode(inode);
|
2005-04-16 18:20:36 -04:00
|
|
|
return inode;
|
|
|
|
}
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* ifind_fast - internal function, you want ilookup() or iget().
|
|
|
|
* @sb: super block of file system to search
|
|
|
|
* @head: head of the list to search
|
|
|
|
* @ino: inode number to search for
|
|
|
|
*
|
|
|
|
* ifind_fast() searches for the inode @ino in the inode cache. This is for
|
|
|
|
* file systems where the inode number is sufficient for unique identification
|
|
|
|
* of an inode.
|
|
|
|
*
|
|
|
|
* If the inode is in the cache, the inode is returned with an incremented
|
|
|
|
* reference count.
|
|
|
|
*
|
|
|
|
* Otherwise NULL is returned.
|
|
|
|
*/
|
2006-01-08 04:05:21 -05:00
|
|
|
static struct inode *ifind_fast(struct super_block *sb,
|
2005-04-16 18:20:36 -04:00
|
|
|
struct hlist_head *head, unsigned long ino)
|
|
|
|
{
|
|
|
|
struct inode *inode;
|
|
|
|
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
inode = find_inode_fast(sb, head, ino);
|
|
|
|
if (inode) {
|
|
|
|
__iget(inode);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
wait_on_inode(inode);
|
|
|
|
return inode;
|
|
|
|
}
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2005-07-13 04:10:44 -04:00
|
|
|
* ilookup5_nowait - search for an inode in the inode cache
|
2005-04-16 18:20:36 -04:00
|
|
|
* @sb: super block of file system to search
|
|
|
|
* @hashval: hash value (usually inode number) to search for
|
|
|
|
* @test: callback used for comparisons between inodes
|
|
|
|
* @data: opaque data pointer to pass to @test
|
|
|
|
*
|
|
|
|
* ilookup5() uses ifind() to search for the inode specified by @hashval and
|
|
|
|
* @data in the inode cache. This is a generalized version of ilookup() for
|
|
|
|
* file systems where the inode number is not sufficient for unique
|
|
|
|
* identification of an inode.
|
|
|
|
*
|
|
|
|
* If the inode is in the cache, the inode is returned with an incremented
|
2005-07-13 04:10:44 -04:00
|
|
|
* reference count. Note, the inode lock is not waited upon so you have to be
|
|
|
|
* very careful what you do with the returned inode. You probably should be
|
|
|
|
* using ilookup5() instead.
|
|
|
|
*
|
|
|
|
* Otherwise NULL is returned.
|
|
|
|
*
|
|
|
|
* Note, @test is called with the inode_lock held, so can't sleep.
|
|
|
|
*/
|
|
|
|
struct inode *ilookup5_nowait(struct super_block *sb, unsigned long hashval,
|
|
|
|
int (*test)(struct inode *, void *), void *data)
|
|
|
|
{
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(sb, hashval);
|
|
|
|
|
|
|
|
return ifind(sb, head, test, data, 0);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(ilookup5_nowait);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* ilookup5 - search for an inode in the inode cache
|
|
|
|
* @sb: super block of file system to search
|
|
|
|
* @hashval: hash value (usually inode number) to search for
|
|
|
|
* @test: callback used for comparisons between inodes
|
|
|
|
* @data: opaque data pointer to pass to @test
|
|
|
|
*
|
|
|
|
* ilookup5() uses ifind() to search for the inode specified by @hashval and
|
|
|
|
* @data in the inode cache. This is a generalized version of ilookup() for
|
|
|
|
* file systems where the inode number is not sufficient for unique
|
|
|
|
* identification of an inode.
|
|
|
|
*
|
|
|
|
* If the inode is in the cache, the inode lock is waited upon and the inode is
|
|
|
|
* returned with an incremented reference count.
|
2005-04-16 18:20:36 -04:00
|
|
|
*
|
|
|
|
* Otherwise NULL is returned.
|
|
|
|
*
|
|
|
|
* Note, @test is called with the inode_lock held, so can't sleep.
|
|
|
|
*/
|
|
|
|
struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
|
|
|
|
int (*test)(struct inode *, void *), void *data)
|
|
|
|
{
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(sb, hashval);
|
|
|
|
|
2005-07-13 04:10:44 -04:00
|
|
|
return ifind(sb, head, test, data, 1);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(ilookup5);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* ilookup - search for an inode in the inode cache
|
|
|
|
* @sb: super block of file system to search
|
|
|
|
* @ino: inode number to search for
|
|
|
|
*
|
|
|
|
* ilookup() uses ifind_fast() to search for the inode @ino in the inode cache.
|
|
|
|
* This is for file systems where the inode number is sufficient for unique
|
|
|
|
* identification of an inode.
|
|
|
|
*
|
|
|
|
* If the inode is in the cache, the inode is returned with an incremented
|
|
|
|
* reference count.
|
|
|
|
*
|
|
|
|
* Otherwise NULL is returned.
|
|
|
|
*/
|
|
|
|
struct inode *ilookup(struct super_block *sb, unsigned long ino)
|
|
|
|
{
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(sb, ino);
|
|
|
|
|
|
|
|
return ifind_fast(sb, head, ino);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(ilookup);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* iget5_locked - obtain an inode from a mounted file system
|
|
|
|
* @sb: super block of file system
|
|
|
|
* @hashval: hash value (usually inode number) to get
|
|
|
|
* @test: callback used for comparisons between inodes
|
|
|
|
* @set: callback used to initialize a new struct inode
|
|
|
|
* @data: opaque data pointer to pass to @test and @set
|
|
|
|
*
|
|
|
|
* iget5_locked() uses ifind() to search for the inode specified by @hashval
|
|
|
|
* and @data in the inode cache and if present it is returned with an increased
|
|
|
|
* reference count. This is a generalized version of iget_locked() for file
|
|
|
|
* systems where the inode number is not sufficient for unique identification
|
|
|
|
* of an inode.
|
|
|
|
*
|
|
|
|
* If the inode is not in cache, get_new_inode() is called to allocate a new
|
|
|
|
* inode and this is returned locked, hashed, and with the I_NEW flag set. The
|
|
|
|
* file system gets to fill it in before unlocking it via unlock_new_inode().
|
|
|
|
*
|
|
|
|
* Note both @test and @set are called with the inode_lock held, so can't sleep.
|
|
|
|
*/
|
|
|
|
struct inode *iget5_locked(struct super_block *sb, unsigned long hashval,
|
|
|
|
int (*test)(struct inode *, void *),
|
|
|
|
int (*set)(struct inode *, void *), void *data)
|
|
|
|
{
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(sb, hashval);
|
|
|
|
struct inode *inode;
|
|
|
|
|
2005-07-13 04:10:44 -04:00
|
|
|
inode = ifind(sb, head, test, data, 1);
|
2005-04-16 18:20:36 -04:00
|
|
|
if (inode)
|
|
|
|
return inode;
|
|
|
|
/*
|
|
|
|
* get_new_inode() will do the right thing, re-trying the search
|
|
|
|
* in case it had to block at any point.
|
|
|
|
*/
|
|
|
|
return get_new_inode(sb, head, test, set, data);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(iget5_locked);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* iget_locked - obtain an inode from a mounted file system
|
|
|
|
* @sb: super block of file system
|
|
|
|
* @ino: inode number to get
|
|
|
|
*
|
|
|
|
* iget_locked() uses ifind_fast() to search for the inode specified by @ino in
|
|
|
|
* the inode cache and if present it is returned with an increased reference
|
|
|
|
* count. This is for file systems where the inode number is sufficient for
|
|
|
|
* unique identification of an inode.
|
|
|
|
*
|
|
|
|
* If the inode is not in cache, get_new_inode_fast() is called to allocate a
|
|
|
|
* new inode and this is returned locked, hashed, and with the I_NEW flag set.
|
|
|
|
* The file system gets to fill it in before unlocking it via
|
|
|
|
* unlock_new_inode().
|
|
|
|
*/
|
|
|
|
struct inode *iget_locked(struct super_block *sb, unsigned long ino)
|
|
|
|
{
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(sb, ino);
|
|
|
|
struct inode *inode;
|
|
|
|
|
|
|
|
inode = ifind_fast(sb, head, ino);
|
|
|
|
if (inode)
|
|
|
|
return inode;
|
|
|
|
/*
|
|
|
|
* get_new_inode_fast() will do the right thing, re-trying the search
|
|
|
|
* in case it had to block at any point.
|
|
|
|
*/
|
|
|
|
return get_new_inode_fast(sb, head, ino);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(iget_locked);
|
|
|
|
|
2008-12-30 01:48:21 -05:00
|
|
|
int insert_inode_locked(struct inode *inode)
|
|
|
|
{
|
|
|
|
struct super_block *sb = inode->i_sb;
|
|
|
|
ino_t ino = inode->i_ino;
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(sb, ino);
|
|
|
|
|
|
|
|
inode->i_state |= I_LOCK|I_NEW;
|
|
|
|
while (1) {
|
2009-05-13 14:13:40 -04:00
|
|
|
struct hlist_node *node;
|
|
|
|
struct inode *old = NULL;
|
2008-12-30 01:48:21 -05:00
|
|
|
spin_lock(&inode_lock);
|
2009-05-13 14:13:40 -04:00
|
|
|
hlist_for_each_entry(old, node, head, i_hash) {
|
|
|
|
if (old->i_ino != ino)
|
|
|
|
continue;
|
|
|
|
if (old->i_sb != sb)
|
|
|
|
continue;
|
|
|
|
if (old->i_state & (I_FREEING|I_CLEAR|I_WILL_FREE))
|
|
|
|
continue;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (likely(!node)) {
|
2008-12-30 01:48:21 -05:00
|
|
|
hlist_add_head(&inode->i_hash, head);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
__iget(old);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
wait_on_inode(old);
|
|
|
|
if (unlikely(!hlist_unhashed(&old->i_hash))) {
|
|
|
|
iput(old);
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
|
|
|
iput(old);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(insert_inode_locked);
|
|
|
|
|
|
|
|
int insert_inode_locked4(struct inode *inode, unsigned long hashval,
|
|
|
|
int (*test)(struct inode *, void *), void *data)
|
|
|
|
{
|
|
|
|
struct super_block *sb = inode->i_sb;
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(sb, hashval);
|
|
|
|
|
|
|
|
inode->i_state |= I_LOCK|I_NEW;
|
|
|
|
|
|
|
|
while (1) {
|
2009-05-13 14:13:40 -04:00
|
|
|
struct hlist_node *node;
|
|
|
|
struct inode *old = NULL;
|
|
|
|
|
2008-12-30 01:48:21 -05:00
|
|
|
spin_lock(&inode_lock);
|
2009-05-13 14:13:40 -04:00
|
|
|
hlist_for_each_entry(old, node, head, i_hash) {
|
|
|
|
if (old->i_sb != sb)
|
|
|
|
continue;
|
|
|
|
if (!test(old, data))
|
|
|
|
continue;
|
|
|
|
if (old->i_state & (I_FREEING|I_CLEAR|I_WILL_FREE))
|
|
|
|
continue;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (likely(!node)) {
|
2008-12-30 01:48:21 -05:00
|
|
|
hlist_add_head(&inode->i_hash, head);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
__iget(old);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
wait_on_inode(old);
|
|
|
|
if (unlikely(!hlist_unhashed(&old->i_hash))) {
|
|
|
|
iput(old);
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
|
|
|
iput(old);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(insert_inode_locked4);
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
/**
|
|
|
|
* __insert_inode_hash - hash an inode
|
|
|
|
* @inode: unhashed inode
|
|
|
|
* @hashval: unsigned long value used to locate this object in the
|
|
|
|
* inode_hashtable.
|
|
|
|
*
|
|
|
|
* Add an inode to the inode hash for this superblock.
|
|
|
|
*/
|
|
|
|
void __insert_inode_hash(struct inode *inode, unsigned long hashval)
|
|
|
|
{
|
|
|
|
struct hlist_head *head = inode_hashtable + hash(inode->i_sb, hashval);
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
hlist_add_head(&inode->i_hash, head);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__insert_inode_hash);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* remove_inode_hash - remove an inode from the hash
|
|
|
|
* @inode: inode to unhash
|
|
|
|
*
|
|
|
|
* Remove an inode from the superblock.
|
|
|
|
*/
|
|
|
|
void remove_inode_hash(struct inode *inode)
|
|
|
|
{
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
hlist_del_init(&inode->i_hash);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(remove_inode_hash);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Tell the filesystem that this inode is no longer of any interest and should
|
|
|
|
* be completely destroyed.
|
|
|
|
*
|
|
|
|
* We leave the inode in the inode hash table until *after* the filesystem's
|
|
|
|
* ->delete_inode completes. This ensures that an iget (such as nfsd might
|
|
|
|
* instigate) will always find up-to-date information either in the hash or on
|
|
|
|
* disk.
|
|
|
|
*
|
|
|
|
* I_FREEING is set so that no-one will take a new reference to the inode while
|
|
|
|
* it is being deleted.
|
|
|
|
*/
|
2009-01-09 10:04:15 -05:00
|
|
|
void generic_delete_inode(struct inode *inode)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2007-02-12 03:55:41 -05:00
|
|
|
const struct super_operations *op = inode->i_sb->s_op;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
2009-01-09 10:04:15 -05:00
|
|
|
list_del_init(&inode->i_list);
|
|
|
|
list_del_init(&inode->i_sb_list);
|
fs: new inode i_state corruption fix
There was a report of a data corruption
http://lkml.org/lkml/2008/11/14/121. There is a script included to
reproduce the problem.
During testing, I encountered a number of strange things with ext3, so I
tried ext2 to attempt to reduce complexity of the problem. I found that
fsstress would quickly hang in wait_on_inode, waiting for I_LOCK to be
cleared, even though instrumentation showed that unlock_new_inode had
already been called for that inode. This points to memory scribble, or
synchronisation problme.
i_state of I_NEW inodes is not protected by inode_lock because other
processes are not supposed to touch them until I_LOCK (and I_NEW) is
cleared. Adding WARN_ON(inode->i_state & I_NEW) to sites where we modify
i_state revealed that generic_sync_sb_inodes is picking up new inodes from
the inode lists and passing them to __writeback_single_inode without
waiting for I_NEW. Subsequently modifying i_state causes corruption. In
my case it would look like this:
CPU0 CPU1
unlock_new_inode() __sync_single_inode()
reg <- inode->i_state
reg -> reg & ~(I_LOCK|I_NEW) reg <- inode->i_state
reg -> inode->i_state reg -> reg | I_SYNC
reg -> inode->i_state
Non-atomic RMW on CPU1 overwrites CPU0 store and sets I_LOCK|I_NEW again.
Fix for this is rather than wait for I_NEW inodes, just skip over them:
inodes concurrently being created are not subject to data integrity
operations, and should not significantly contribute to dirty memory
either.
After this change, I'm unable to reproduce any of the added warnings or
hangs after ~1hour of running. Previously, the new warnings would start
immediately and hang would happen in under 5 minutes.
I'm also testing on ext3 now, and so far no problems there either. I
don't know whether this fixes the problem reported above, but it fixes a
real problem for me.
Cc: "Jorge Boncompte [DTI2]" <jorge@dti2.net>
Reported-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: <stable@kernel.org>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-03-12 17:31:38 -04:00
|
|
|
WARN_ON(inode->i_state & I_NEW);
|
2009-01-09 10:04:15 -05:00
|
|
|
inode->i_state |= I_FREEING;
|
|
|
|
inodes_stat.nr_inodes--;
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
security_inode_delete(inode);
|
|
|
|
|
|
|
|
if (op->delete_inode) {
|
|
|
|
void (*delete)(struct inode *) = op->delete_inode;
|
|
|
|
if (!is_bad_inode(inode))
|
2009-01-26 10:45:12 -05:00
|
|
|
vfs_dq_init(inode);
|
2005-09-09 16:01:29 -04:00
|
|
|
/* Filesystems implementing their own
|
|
|
|
* s_op->delete_inode are required to call
|
|
|
|
* truncate_inode_pages and clear_inode()
|
|
|
|
* internally */
|
2005-04-16 18:20:36 -04:00
|
|
|
delete(inode);
|
2005-09-09 16:01:29 -04:00
|
|
|
} else {
|
|
|
|
truncate_inode_pages(&inode->i_data, 0);
|
2005-04-16 18:20:36 -04:00
|
|
|
clear_inode(inode);
|
2005-09-09 16:01:29 -04:00
|
|
|
}
|
2005-04-16 18:20:36 -04:00
|
|
|
spin_lock(&inode_lock);
|
|
|
|
hlist_del_init(&inode->i_hash);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
wake_up_inode(inode);
|
2006-04-02 07:38:18 -04:00
|
|
|
BUG_ON(inode->i_state != I_CLEAR);
|
2005-04-16 18:20:36 -04:00
|
|
|
destroy_inode(inode);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(generic_delete_inode);
|
|
|
|
|
2009-09-18 16:05:44 -04:00
|
|
|
/**
|
|
|
|
* generic_detach_inode - remove inode from inode lists
|
|
|
|
* @inode: inode to remove
|
|
|
|
*
|
|
|
|
* Remove inode from inode lists, write it if it's dirty. This is just an
|
|
|
|
* internal VFS helper exported for hugetlbfs. Do not use!
|
|
|
|
*
|
|
|
|
* Returns 1 if inode should be completely destroyed.
|
|
|
|
*/
|
|
|
|
int generic_detach_inode(struct inode *inode)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
struct super_block *sb = inode->i_sb;
|
|
|
|
|
|
|
|
if (!hlist_unhashed(&inode->i_hash)) {
|
2007-10-17 02:30:44 -04:00
|
|
|
if (!(inode->i_state & (I_DIRTY|I_SYNC)))
|
2005-04-16 18:20:36 -04:00
|
|
|
list_move(&inode->i_list, &inode_unused);
|
|
|
|
inodes_stat.nr_unused++;
|
2007-05-08 03:25:52 -04:00
|
|
|
if (sb->s_flags & MS_ACTIVE) {
|
2005-06-23 03:09:01 -04:00
|
|
|
spin_unlock(&inode_lock);
|
2009-09-18 16:05:44 -04:00
|
|
|
return 0;
|
2005-06-23 03:09:01 -04:00
|
|
|
}
|
fs: new inode i_state corruption fix
There was a report of a data corruption
http://lkml.org/lkml/2008/11/14/121. There is a script included to
reproduce the problem.
During testing, I encountered a number of strange things with ext3, so I
tried ext2 to attempt to reduce complexity of the problem. I found that
fsstress would quickly hang in wait_on_inode, waiting for I_LOCK to be
cleared, even though instrumentation showed that unlock_new_inode had
already been called for that inode. This points to memory scribble, or
synchronisation problme.
i_state of I_NEW inodes is not protected by inode_lock because other
processes are not supposed to touch them until I_LOCK (and I_NEW) is
cleared. Adding WARN_ON(inode->i_state & I_NEW) to sites where we modify
i_state revealed that generic_sync_sb_inodes is picking up new inodes from
the inode lists and passing them to __writeback_single_inode without
waiting for I_NEW. Subsequently modifying i_state causes corruption. In
my case it would look like this:
CPU0 CPU1
unlock_new_inode() __sync_single_inode()
reg <- inode->i_state
reg -> reg & ~(I_LOCK|I_NEW) reg <- inode->i_state
reg -> inode->i_state reg -> reg | I_SYNC
reg -> inode->i_state
Non-atomic RMW on CPU1 overwrites CPU0 store and sets I_LOCK|I_NEW again.
Fix for this is rather than wait for I_NEW inodes, just skip over them:
inodes concurrently being created are not subject to data integrity
operations, and should not significantly contribute to dirty memory
either.
After this change, I'm unable to reproduce any of the added warnings or
hangs after ~1hour of running. Previously, the new warnings would start
immediately and hang would happen in under 5 minutes.
I'm also testing on ext3 now, and so far no problems there either. I
don't know whether this fixes the problem reported above, but it fixes a
real problem for me.
Cc: "Jorge Boncompte [DTI2]" <jorge@dti2.net>
Reported-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: <stable@kernel.org>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-03-12 17:31:38 -04:00
|
|
|
WARN_ON(inode->i_state & I_NEW);
|
2005-06-23 03:09:01 -04:00
|
|
|
inode->i_state |= I_WILL_FREE;
|
|
|
|
spin_unlock(&inode_lock);
|
2005-04-16 18:20:36 -04:00
|
|
|
write_inode_now(inode, 1);
|
|
|
|
spin_lock(&inode_lock);
|
fs: new inode i_state corruption fix
There was a report of a data corruption
http://lkml.org/lkml/2008/11/14/121. There is a script included to
reproduce the problem.
During testing, I encountered a number of strange things with ext3, so I
tried ext2 to attempt to reduce complexity of the problem. I found that
fsstress would quickly hang in wait_on_inode, waiting for I_LOCK to be
cleared, even though instrumentation showed that unlock_new_inode had
already been called for that inode. This points to memory scribble, or
synchronisation problme.
i_state of I_NEW inodes is not protected by inode_lock because other
processes are not supposed to touch them until I_LOCK (and I_NEW) is
cleared. Adding WARN_ON(inode->i_state & I_NEW) to sites where we modify
i_state revealed that generic_sync_sb_inodes is picking up new inodes from
the inode lists and passing them to __writeback_single_inode without
waiting for I_NEW. Subsequently modifying i_state causes corruption. In
my case it would look like this:
CPU0 CPU1
unlock_new_inode() __sync_single_inode()
reg <- inode->i_state
reg -> reg & ~(I_LOCK|I_NEW) reg <- inode->i_state
reg -> inode->i_state reg -> reg | I_SYNC
reg -> inode->i_state
Non-atomic RMW on CPU1 overwrites CPU0 store and sets I_LOCK|I_NEW again.
Fix for this is rather than wait for I_NEW inodes, just skip over them:
inodes concurrently being created are not subject to data integrity
operations, and should not significantly contribute to dirty memory
either.
After this change, I'm unable to reproduce any of the added warnings or
hangs after ~1hour of running. Previously, the new warnings would start
immediately and hang would happen in under 5 minutes.
I'm also testing on ext3 now, and so far no problems there either. I
don't know whether this fixes the problem reported above, but it fixes a
real problem for me.
Cc: "Jorge Boncompte [DTI2]" <jorge@dti2.net>
Reported-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: <stable@kernel.org>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-03-12 17:31:38 -04:00
|
|
|
WARN_ON(inode->i_state & I_NEW);
|
2005-06-23 03:09:01 -04:00
|
|
|
inode->i_state &= ~I_WILL_FREE;
|
2005-04-16 18:20:36 -04:00
|
|
|
inodes_stat.nr_unused--;
|
|
|
|
hlist_del_init(&inode->i_hash);
|
|
|
|
}
|
|
|
|
list_del_init(&inode->i_list);
|
|
|
|
list_del_init(&inode->i_sb_list);
|
fs: new inode i_state corruption fix
There was a report of a data corruption
http://lkml.org/lkml/2008/11/14/121. There is a script included to
reproduce the problem.
During testing, I encountered a number of strange things with ext3, so I
tried ext2 to attempt to reduce complexity of the problem. I found that
fsstress would quickly hang in wait_on_inode, waiting for I_LOCK to be
cleared, even though instrumentation showed that unlock_new_inode had
already been called for that inode. This points to memory scribble, or
synchronisation problme.
i_state of I_NEW inodes is not protected by inode_lock because other
processes are not supposed to touch them until I_LOCK (and I_NEW) is
cleared. Adding WARN_ON(inode->i_state & I_NEW) to sites where we modify
i_state revealed that generic_sync_sb_inodes is picking up new inodes from
the inode lists and passing them to __writeback_single_inode without
waiting for I_NEW. Subsequently modifying i_state causes corruption. In
my case it would look like this:
CPU0 CPU1
unlock_new_inode() __sync_single_inode()
reg <- inode->i_state
reg -> reg & ~(I_LOCK|I_NEW) reg <- inode->i_state
reg -> inode->i_state reg -> reg | I_SYNC
reg -> inode->i_state
Non-atomic RMW on CPU1 overwrites CPU0 store and sets I_LOCK|I_NEW again.
Fix for this is rather than wait for I_NEW inodes, just skip over them:
inodes concurrently being created are not subject to data integrity
operations, and should not significantly contribute to dirty memory
either.
After this change, I'm unable to reproduce any of the added warnings or
hangs after ~1hour of running. Previously, the new warnings would start
immediately and hang would happen in under 5 minutes.
I'm also testing on ext3 now, and so far no problems there either. I
don't know whether this fixes the problem reported above, but it fixes a
real problem for me.
Cc: "Jorge Boncompte [DTI2]" <jorge@dti2.net>
Reported-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: <stable@kernel.org>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-03-12 17:31:38 -04:00
|
|
|
WARN_ON(inode->i_state & I_NEW);
|
2005-06-23 03:09:01 -04:00
|
|
|
inode->i_state |= I_FREEING;
|
2005-04-16 18:20:36 -04:00
|
|
|
inodes_stat.nr_inodes--;
|
|
|
|
spin_unlock(&inode_lock);
|
2009-09-18 16:05:44 -04:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(generic_detach_inode);
|
|
|
|
|
|
|
|
static void generic_forget_inode(struct inode *inode)
|
|
|
|
{
|
|
|
|
if (!generic_detach_inode(inode))
|
|
|
|
return;
|
2005-04-16 18:20:36 -04:00
|
|
|
if (inode->i_data.nrpages)
|
|
|
|
truncate_inode_pages(&inode->i_data, 0);
|
|
|
|
clear_inode(inode);
|
[PATCH] fix nr_unused accounting, and avoid recursing in iput with I_WILL_FREE set
list_move(&inode->i_list, &inode_in_use);
} else {
list_move(&inode->i_list, &inode_unused);
+ inodes_stat.nr_unused++;
}
}
wake_up_inode(inode);
Are you sure the above diff is correct? It was added somewhere between
2.6.5 and 2.6.8. I think it's wrong.
The only way I can imagine the i_count to be zero in the above path, is
that I_WILL_FREE is set. And if I_WILL_FREE is set, then we must not
increase nr_unused. So I believe the above change is buggy and it will
definitely overstate the number of unused inodes and it should be backed
out.
Note that __writeback_single_inode before calling __sync_single_inode, can
drop the spinlock and we can have both the dirty and locked bitflags clear
here:
spin_unlock(&inode_lock);
__wait_on_inode(inode);
iput(inode);
XXXXXXX
spin_lock(&inode_lock);
}
use inode again here
a construct like the above makes zero sense from a reference counting
standpoint.
Either we don't ever use the inode again after the iput, or the
inode_lock should be taken _before_ executing the iput (i.e. a __iput
would be required). Taking the inode_lock after iput means the iget was
useless if we keep using the inode after the iput.
So the only chance the 2.6 was safe to call __writeback_single_inode
with the i_count == 0, is that I_WILL_FREE is set (I_WILL_FREE will
prevent the VM to free the inode in XXXXX).
Potentially calling the above iput with I_WILL_FREE was also wrong
because it would recurse in iput_final (the second mainline bug).
The below (untested) patch fixes the nr_unused accounting, avoids recursing
in iput when I_WILL_FREE is set and makes sure (with the BUG_ON) that we
don't corrupt memory and that all holders that don't set I_WILL_FREE, keeps
a reference on the inode!
Signed-off-by: Andrea Arcangeli <andrea@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 18:03:05 -05:00
|
|
|
wake_up_inode(inode);
|
2005-04-16 18:20:36 -04:00
|
|
|
destroy_inode(inode);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Normal UNIX filesystem behaviour: delete the
|
|
|
|
* inode when the usage count drops to zero, and
|
|
|
|
* i_nlink is zero.
|
|
|
|
*/
|
2005-07-07 20:56:03 -04:00
|
|
|
void generic_drop_inode(struct inode *inode)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
if (!inode->i_nlink)
|
|
|
|
generic_delete_inode(inode);
|
|
|
|
else
|
|
|
|
generic_forget_inode(inode);
|
|
|
|
}
|
2005-07-07 20:56:03 -04:00
|
|
|
EXPORT_SYMBOL_GPL(generic_drop_inode);
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
/*
|
|
|
|
* Called when we're dropping the last reference
|
2009-03-31 10:05:54 -04:00
|
|
|
* to an inode.
|
2005-04-16 18:20:36 -04:00
|
|
|
*
|
|
|
|
* Call the FS "drop()" function, defaulting to
|
|
|
|
* the legacy UNIX filesystem behaviour..
|
|
|
|
*
|
|
|
|
* NOTE! NOTE! NOTE! We're called with the inode lock
|
|
|
|
* held, and the drop function is supposed to release
|
|
|
|
* the lock!
|
|
|
|
*/
|
|
|
|
static inline void iput_final(struct inode *inode)
|
|
|
|
{
|
2007-02-12 03:55:41 -05:00
|
|
|
const struct super_operations *op = inode->i_sb->s_op;
|
2005-04-16 18:20:36 -04:00
|
|
|
void (*drop)(struct inode *) = generic_drop_inode;
|
|
|
|
|
|
|
|
if (op && op->drop_inode)
|
|
|
|
drop = op->drop_inode;
|
|
|
|
drop(inode);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2009-03-31 10:05:54 -04:00
|
|
|
* iput - put an inode
|
2005-04-16 18:20:36 -04:00
|
|
|
* @inode: inode to put
|
|
|
|
*
|
|
|
|
* Puts an inode, dropping its usage count. If the inode use count hits
|
|
|
|
* zero, the inode is then freed and may also be destroyed.
|
|
|
|
*
|
|
|
|
* Consequently, iput() can sleep.
|
|
|
|
*/
|
|
|
|
void iput(struct inode *inode)
|
|
|
|
{
|
|
|
|
if (inode) {
|
|
|
|
BUG_ON(inode->i_state == I_CLEAR);
|
|
|
|
|
|
|
|
if (atomic_dec_and_lock(&inode->i_count, &inode_lock))
|
|
|
|
iput_final(inode);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(iput);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* bmap - find a block number in a file
|
|
|
|
* @inode: inode of file
|
|
|
|
* @block: block to find
|
|
|
|
*
|
|
|
|
* Returns the block number on the device holding the inode that
|
|
|
|
* is the disk block number for the block of the file requested.
|
|
|
|
* That is, asked for block 4 of inode 1 the function will return the
|
2009-03-31 10:05:54 -04:00
|
|
|
* disk block relative to the disk start that holds that block of the
|
2005-04-16 18:20:36 -04:00
|
|
|
* file.
|
|
|
|
*/
|
2009-03-31 10:05:54 -04:00
|
|
|
sector_t bmap(struct inode *inode, sector_t block)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
sector_t res = 0;
|
|
|
|
if (inode->i_mapping->a_ops->bmap)
|
|
|
|
res = inode->i_mapping->a_ops->bmap(inode->i_mapping, block);
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(bmap);
|
|
|
|
|
2009-03-26 13:32:14 -04:00
|
|
|
/*
|
|
|
|
* With relative atime, only update atime if the previous atime is
|
|
|
|
* earlier than either the ctime or mtime or if at least a day has
|
|
|
|
* passed since the last atime update.
|
|
|
|
*/
|
|
|
|
static int relatime_need_update(struct vfsmount *mnt, struct inode *inode,
|
|
|
|
struct timespec now)
|
|
|
|
{
|
|
|
|
|
|
|
|
if (!(mnt->mnt_flags & MNT_RELATIME))
|
|
|
|
return 1;
|
|
|
|
/*
|
|
|
|
* Is mtime younger than atime? If yes, update atime:
|
|
|
|
*/
|
|
|
|
if (timespec_compare(&inode->i_mtime, &inode->i_atime) >= 0)
|
|
|
|
return 1;
|
|
|
|
/*
|
|
|
|
* Is ctime younger than atime? If yes, update atime:
|
|
|
|
*/
|
|
|
|
if (timespec_compare(&inode->i_ctime, &inode->i_atime) >= 0)
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Is the previous atime value older than a day? If yes,
|
|
|
|
* update atime:
|
|
|
|
*/
|
|
|
|
if ((long)(now.tv_sec - inode->i_atime.tv_sec) >= 24*60*60)
|
|
|
|
return 1;
|
|
|
|
/*
|
|
|
|
* Good, we can skip the atime update:
|
|
|
|
*/
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
/**
|
2006-01-09 23:52:03 -05:00
|
|
|
* touch_atime - update the access time
|
|
|
|
* @mnt: mount the inode is accessed on
|
2006-02-01 06:06:57 -05:00
|
|
|
* @dentry: dentry accessed
|
2005-04-16 18:20:36 -04:00
|
|
|
*
|
|
|
|
* Update the accessed time on an inode and mark it for writeback.
|
|
|
|
* This function automatically handles read only file systems and media,
|
|
|
|
* as well as the "noatime" flag and inode specific "noatime" markers.
|
|
|
|
*/
|
2006-01-09 23:52:03 -05:00
|
|
|
void touch_atime(struct vfsmount *mnt, struct dentry *dentry)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2006-01-09 23:52:03 -05:00
|
|
|
struct inode *inode = dentry->d_inode;
|
2005-04-16 18:20:36 -04:00
|
|
|
struct timespec now;
|
|
|
|
|
2008-02-15 17:37:41 -05:00
|
|
|
if (inode->i_flags & S_NOATIME)
|
2009-09-18 16:05:47 -04:00
|
|
|
return;
|
2007-02-10 04:44:49 -05:00
|
|
|
if (IS_NOATIME(inode))
|
2009-09-18 16:05:47 -04:00
|
|
|
return;
|
2006-12-13 03:34:33 -05:00
|
|
|
if ((inode->i_sb->s_flags & MS_NODIRATIME) && S_ISDIR(inode->i_mode))
|
2009-09-18 16:05:47 -04:00
|
|
|
return;
|
2006-12-13 03:34:34 -05:00
|
|
|
|
2008-02-15 17:37:41 -05:00
|
|
|
if (mnt->mnt_flags & MNT_NOATIME)
|
2009-09-18 16:05:47 -04:00
|
|
|
return;
|
2008-02-15 17:37:41 -05:00
|
|
|
if ((mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode))
|
2009-09-18 16:05:47 -04:00
|
|
|
return;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
now = current_fs_time(inode->i_sb);
|
2009-03-26 13:32:14 -04:00
|
|
|
|
|
|
|
if (!relatime_need_update(mnt, inode, now))
|
2009-09-18 16:05:47 -04:00
|
|
|
return;
|
2009-03-26 13:32:14 -04:00
|
|
|
|
2006-12-13 03:34:34 -05:00
|
|
|
if (timespec_equal(&inode->i_atime, &now))
|
2009-09-18 16:05:47 -04:00
|
|
|
return;
|
|
|
|
|
|
|
|
if (mnt_want_write(mnt))
|
|
|
|
return;
|
2006-12-13 03:34:34 -05:00
|
|
|
|
|
|
|
inode->i_atime = now;
|
|
|
|
mark_inode_dirty_sync(inode);
|
2008-02-15 17:37:41 -05:00
|
|
|
mnt_drop_write(mnt);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
2006-01-09 23:52:03 -05:00
|
|
|
EXPORT_SYMBOL(touch_atime);
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
/**
|
2006-01-09 23:52:01 -05:00
|
|
|
* file_update_time - update mtime and ctime time
|
|
|
|
* @file: file accessed
|
2005-04-16 18:20:36 -04:00
|
|
|
*
|
2006-01-09 23:52:01 -05:00
|
|
|
* Update the mtime and ctime members of an inode and mark the inode
|
|
|
|
* for writeback. Note that this function is meant exclusively for
|
|
|
|
* usage in the file write path of filesystems, and filesystems may
|
|
|
|
* choose to explicitly ignore update via this function with the
|
2009-04-02 09:23:37 -04:00
|
|
|
* S_NOCMTIME inode flag, e.g. for network filesystem where these
|
2006-01-09 23:52:01 -05:00
|
|
|
* timestamps are handled by the server.
|
2005-04-16 18:20:36 -04:00
|
|
|
*/
|
|
|
|
|
2006-01-09 23:52:01 -05:00
|
|
|
void file_update_time(struct file *file)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
2006-12-08 05:36:35 -05:00
|
|
|
struct inode *inode = file->f_path.dentry->d_inode;
|
2005-04-16 18:20:36 -04:00
|
|
|
struct timespec now;
|
2009-09-18 16:05:48 -04:00
|
|
|
enum { S_MTIME = 1, S_CTIME = 2, S_VERSION = 4 } sync_it = 0;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
2009-09-18 16:05:48 -04:00
|
|
|
/* First try to exhaust all avenues to not sync */
|
2005-04-16 18:20:36 -04:00
|
|
|
if (IS_NOCMTIME(inode))
|
|
|
|
return;
|
2008-02-15 17:37:43 -05:00
|
|
|
|
2005-04-16 18:20:36 -04:00
|
|
|
now = current_fs_time(inode->i_sb);
|
2009-09-18 16:05:48 -04:00
|
|
|
if (!timespec_equal(&inode->i_mtime, &now))
|
|
|
|
sync_it = S_MTIME;
|
2005-04-16 18:20:36 -04:00
|
|
|
|
2009-09-18 16:05:48 -04:00
|
|
|
if (!timespec_equal(&inode->i_ctime, &now))
|
|
|
|
sync_it |= S_CTIME;
|
2006-01-09 23:52:01 -05:00
|
|
|
|
2009-09-18 16:05:48 -04:00
|
|
|
if (IS_I_VERSION(inode))
|
|
|
|
sync_it |= S_VERSION;
|
2008-01-28 23:58:27 -05:00
|
|
|
|
2009-09-18 16:05:48 -04:00
|
|
|
if (!sync_it)
|
|
|
|
return;
|
|
|
|
|
|
|
|
/* Finally allowed to write? Takes lock. */
|
|
|
|
if (mnt_want_write_file(file))
|
|
|
|
return;
|
|
|
|
|
|
|
|
/* Only change inode inside the lock region */
|
|
|
|
if (sync_it & S_VERSION)
|
|
|
|
inode_inc_iversion(inode);
|
|
|
|
if (sync_it & S_CTIME)
|
|
|
|
inode->i_ctime = now;
|
|
|
|
if (sync_it & S_MTIME)
|
|
|
|
inode->i_mtime = now;
|
|
|
|
mark_inode_dirty_sync(inode);
|
2008-02-15 17:37:43 -05:00
|
|
|
mnt_drop_write(file->f_path.mnt);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
2006-01-09 23:52:01 -05:00
|
|
|
EXPORT_SYMBOL(file_update_time);
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
int inode_needs_sync(struct inode *inode)
|
|
|
|
{
|
|
|
|
if (IS_SYNC(inode))
|
|
|
|
return 1;
|
|
|
|
if (S_ISDIR(inode->i_mode) && IS_DIRSYNC(inode))
|
|
|
|
return 1;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(inode_needs_sync);
|
|
|
|
|
|
|
|
int inode_wait(void *word)
|
|
|
|
{
|
|
|
|
schedule();
|
|
|
|
return 0;
|
|
|
|
}
|
2008-11-10 01:06:05 -05:00
|
|
|
EXPORT_SYMBOL(inode_wait);
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
/*
|
2005-07-12 16:58:10 -04:00
|
|
|
* If we try to find an inode in the inode hash while it is being
|
|
|
|
* deleted, we have to wait until the filesystem completes its
|
|
|
|
* deletion before reporting that it isn't found. This function waits
|
|
|
|
* until the deletion _might_ have completed. Callers are responsible
|
|
|
|
* to recheck inode state.
|
|
|
|
*
|
|
|
|
* It doesn't matter if I_LOCK is not set initially, a call to
|
|
|
|
* wake_up_inode() after removing from the hash list will DTRT.
|
|
|
|
*
|
2005-04-16 18:20:36 -04:00
|
|
|
* This is called with inode_lock held.
|
|
|
|
*/
|
|
|
|
static void __wait_on_freeing_inode(struct inode *inode)
|
|
|
|
{
|
|
|
|
wait_queue_head_t *wq;
|
|
|
|
DEFINE_WAIT_BIT(wait, &inode->i_state, __I_LOCK);
|
|
|
|
wq = bit_waitqueue(&inode->i_state, __I_LOCK);
|
|
|
|
prepare_to_wait(wq, &wait.wait, TASK_UNINTERRUPTIBLE);
|
|
|
|
spin_unlock(&inode_lock);
|
|
|
|
schedule();
|
|
|
|
finish_wait(wq, &wait.wait);
|
|
|
|
spin_lock(&inode_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static __initdata unsigned long ihash_entries;
|
|
|
|
static int __init set_ihash_entries(char *str)
|
|
|
|
{
|
|
|
|
if (!str)
|
|
|
|
return 0;
|
|
|
|
ihash_entries = simple_strtoul(str, &str, 0);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("ihash_entries=", set_ihash_entries);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Initialize the waitqueues and inode hash table.
|
|
|
|
*/
|
|
|
|
void __init inode_init_early(void)
|
|
|
|
{
|
|
|
|
int loop;
|
|
|
|
|
|
|
|
/* If hashes are distributed across NUMA nodes, defer
|
|
|
|
* hash allocation until vmalloc space is available.
|
|
|
|
*/
|
|
|
|
if (hashdist)
|
|
|
|
return;
|
|
|
|
|
|
|
|
inode_hashtable =
|
|
|
|
alloc_large_system_hash("Inode-cache",
|
|
|
|
sizeof(struct hlist_head),
|
|
|
|
ihash_entries,
|
|
|
|
14,
|
|
|
|
HASH_EARLY,
|
|
|
|
&i_hash_shift,
|
|
|
|
&i_hash_mask,
|
|
|
|
0);
|
|
|
|
|
|
|
|
for (loop = 0; loop < (1 << i_hash_shift); loop++)
|
|
|
|
INIT_HLIST_HEAD(&inode_hashtable[loop]);
|
|
|
|
}
|
|
|
|
|
2007-10-17 02:26:30 -04:00
|
|
|
void __init inode_init(void)
|
2005-04-16 18:20:36 -04:00
|
|
|
{
|
|
|
|
int loop;
|
|
|
|
|
|
|
|
/* inode slab cache */
|
2006-03-24 06:16:09 -05:00
|
|
|
inode_cachep = kmem_cache_create("inode_cache",
|
|
|
|
sizeof(struct inode),
|
|
|
|
0,
|
|
|
|
(SLAB_RECLAIM_ACCOUNT|SLAB_PANIC|
|
|
|
|
SLAB_MEM_SPREAD),
|
2007-07-19 21:11:58 -04:00
|
|
|
init_once);
|
2007-07-17 07:03:17 -04:00
|
|
|
register_shrinker(&icache_shrinker);
|
2005-04-16 18:20:36 -04:00
|
|
|
|
|
|
|
/* Hash may have been set up in inode_init_early */
|
|
|
|
if (!hashdist)
|
|
|
|
return;
|
|
|
|
|
|
|
|
inode_hashtable =
|
|
|
|
alloc_large_system_hash("Inode-cache",
|
|
|
|
sizeof(struct hlist_head),
|
|
|
|
ihash_entries,
|
|
|
|
14,
|
|
|
|
0,
|
|
|
|
&i_hash_shift,
|
|
|
|
&i_hash_mask,
|
|
|
|
0);
|
|
|
|
|
|
|
|
for (loop = 0; loop < (1 << i_hash_shift); loop++)
|
|
|
|
INIT_HLIST_HEAD(&inode_hashtable[loop]);
|
|
|
|
}
|
|
|
|
|
|
|
|
void init_special_inode(struct inode *inode, umode_t mode, dev_t rdev)
|
|
|
|
{
|
|
|
|
inode->i_mode = mode;
|
|
|
|
if (S_ISCHR(mode)) {
|
|
|
|
inode->i_fop = &def_chr_fops;
|
|
|
|
inode->i_rdev = rdev;
|
|
|
|
} else if (S_ISBLK(mode)) {
|
|
|
|
inode->i_fop = &def_blk_fops;
|
|
|
|
inode->i_rdev = rdev;
|
|
|
|
} else if (S_ISFIFO(mode))
|
|
|
|
inode->i_fop = &def_fifo_fops;
|
|
|
|
else if (S_ISSOCK(mode))
|
|
|
|
inode->i_fop = &bad_sock_fops;
|
|
|
|
else
|
2009-09-18 16:05:43 -04:00
|
|
|
printk(KERN_DEBUG "init_special_inode: bogus i_mode (%o) for"
|
|
|
|
" inode %s:%lu\n", mode, inode->i_sb->s_id,
|
|
|
|
inode->i_ino);
|
2005-04-16 18:20:36 -04:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(init_special_inode);
|