From e188dc02d3a9c911be56eca5aa114fe7e9822d53 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Fri, 3 Feb 2012 14:25:18 +0100 Subject: [PATCH 1/7] vfs: fix d_inode_lookup() dentry ref leak d_inode_lookup() leaks a dentry reference on IS_DEADDIR(). Signed-off-by: Miklos Szeredi CC: stable@vger.kernel.org Signed-off-by: Al Viro --- fs/namei.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/namei.c b/fs/namei.c index 208c6aa4a989..a780ea515c47 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -1095,8 +1095,10 @@ static struct dentry *d_inode_lookup(struct dentry *parent, struct dentry *dentr struct dentry *old; /* Don't create child dentry for a dead directory. */ - if (unlikely(IS_DEADDIR(inode))) + if (unlikely(IS_DEADDIR(inode))) { + dput(dentry); return ERR_PTR(-ENOENT); + } old = inode->i_op->lookup(inode, dentry, nd); if (unlikely(old)) { From 1d6f2097865e64963e90cce04980dce2f9fc023f Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 22 Aug 2011 11:52:28 +0800 Subject: [PATCH 2/7] autofs4 - fix lockdep splat in autofs When recursing down the locks when traversing a tree/list in get_next_positive_dentry() or get_next_positive_subdir() a lock can change from being nested to being a parent which breaks lockdep. This patch tells lockdep about what we did. Signed-off-by: Steven Rostedt Acked-by: Ian Kent Signed-off-by: Al Viro --- fs/autofs4/expire.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/fs/autofs4/expire.c b/fs/autofs4/expire.c index 450f529a4eae..1feb68ecef95 100644 --- a/fs/autofs4/expire.c +++ b/fs/autofs4/expire.c @@ -124,6 +124,7 @@ static struct dentry *get_next_positive_subdir(struct dentry *prev, /* Negative dentry - try next */ if (!simple_positive(q)) { spin_unlock(&p->d_lock); + lock_set_subclass(&q->d_lock.dep_map, 0, _RET_IP_); p = q; goto again; } @@ -186,6 +187,7 @@ static struct dentry *get_next_positive_dentry(struct dentry *prev, /* Negative dentry - try next */ if (!simple_positive(ret)) { spin_unlock(&p->d_lock); + lock_set_subclass(&ret->d_lock.dep_map, 0, _RET_IP_); p = ret; goto again; } From 074b85175a43a23fdbde60f55feea636e0bf0f85 Mon Sep 17 00:00:00 2001 From: Dimitri Sivanich Date: Wed, 8 Feb 2012 12:39:07 -0800 Subject: [PATCH 3/7] vfs: fix panic in __d_lookup() with high dentry hashtable counts When the number of dentry cache hash table entries gets too high (2147483648 entries), as happens by default on a 16TB system, use of a signed integer in the dcache_init() initialization loop prevents the dentry_hashtable from getting initialized, causing a panic in __d_lookup(). Fix this in dcache_init() and similar areas. Signed-off-by: Dimitri Sivanich Acked-by: David S. Miller Cc: Al Viro Signed-off-by: Andrew Morton Signed-off-by: Al Viro --- fs/dcache.c | 8 ++++---- fs/inode.c | 8 ++++---- kernel/pid.c | 4 ++-- mm/page_alloc.c | 1 + net/ipv4/tcp.c | 5 +++-- 5 files changed, 14 insertions(+), 12 deletions(-) diff --git a/fs/dcache.c b/fs/dcache.c index 16a53cc2cc02..fe19ac13f75f 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -2968,7 +2968,7 @@ __setup("dhash_entries=", set_dhash_entries); static void __init dcache_init_early(void) { - int loop; + unsigned int loop; /* If hashes are distributed across NUMA nodes, defer * hash allocation until vmalloc space is available. @@ -2986,13 +2986,13 @@ static void __init dcache_init_early(void) &d_hash_mask, 0); - for (loop = 0; loop < (1 << d_hash_shift); loop++) + for (loop = 0; loop < (1U << d_hash_shift); loop++) INIT_HLIST_BL_HEAD(dentry_hashtable + loop); } static void __init dcache_init(void) { - int loop; + unsigned int loop; /* * A constructor could be added for stable state like the lists, @@ -3016,7 +3016,7 @@ static void __init dcache_init(void) &d_hash_mask, 0); - for (loop = 0; loop < (1 << d_hash_shift); loop++) + for (loop = 0; loop < (1U << d_hash_shift); loop++) INIT_HLIST_BL_HEAD(dentry_hashtable + loop); } diff --git a/fs/inode.c b/fs/inode.c index fb10d86ffad7..d3ebdbe723d0 100644 --- a/fs/inode.c +++ b/fs/inode.c @@ -1651,7 +1651,7 @@ __setup("ihash_entries=", set_ihash_entries); */ void __init inode_init_early(void) { - int loop; + unsigned int loop; /* If hashes are distributed across NUMA nodes, defer * hash allocation until vmalloc space is available. @@ -1669,13 +1669,13 @@ void __init inode_init_early(void) &i_hash_mask, 0); - for (loop = 0; loop < (1 << i_hash_shift); loop++) + for (loop = 0; loop < (1U << i_hash_shift); loop++) INIT_HLIST_HEAD(&inode_hashtable[loop]); } void __init inode_init(void) { - int loop; + unsigned int loop; /* inode slab cache */ inode_cachep = kmem_cache_create("inode_cache", @@ -1699,7 +1699,7 @@ void __init inode_init(void) &i_hash_mask, 0); - for (loop = 0; loop < (1 << i_hash_shift); loop++) + for (loop = 0; loop < (1U << i_hash_shift); loop++) INIT_HLIST_HEAD(&inode_hashtable[loop]); } diff --git a/kernel/pid.c b/kernel/pid.c index ce8e00deaccb..9f08dfabaf13 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -543,12 +543,12 @@ struct pid *find_ge_pid(int nr, struct pid_namespace *ns) */ void __init pidhash_init(void) { - int i, pidhash_size; + unsigned int i, pidhash_size; pid_hash = alloc_large_system_hash("PID", sizeof(*pid_hash), 0, 18, HASH_EARLY | HASH_SMALL, &pidhash_shift, NULL, 4096); - pidhash_size = 1 << pidhash_shift; + pidhash_size = 1U << pidhash_shift; for (i = 0; i < pidhash_size; i++) INIT_HLIST_HEAD(&pid_hash[i]); diff --git a/mm/page_alloc.c b/mm/page_alloc.c index d2186ecb36f7..a13ded1938f0 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -5236,6 +5236,7 @@ void *__init alloc_large_system_hash(const char *tablename, max = ((unsigned long long)nr_all_pages << PAGE_SHIFT) >> 4; do_div(max, bucketsize); } + max = min(max, 0x80000000ULL); if (numentries > max) numentries = max; diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 37755ccc0e96..22ef5f9fd2ff 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -3240,7 +3240,8 @@ void __init tcp_init(void) { struct sk_buff *skb = NULL; unsigned long limit; - int i, max_share, cnt; + int max_share, cnt; + unsigned int i; unsigned long jiffy = jiffies; BUILD_BUG_ON(sizeof(struct tcp_skb_cb) > sizeof(skb->cb)); @@ -3283,7 +3284,7 @@ void __init tcp_init(void) &tcp_hashinfo.bhash_size, NULL, 64 * 1024); - tcp_hashinfo.bhash_size = 1 << tcp_hashinfo.bhash_size; + tcp_hashinfo.bhash_size = 1U << tcp_hashinfo.bhash_size; for (i = 0; i < tcp_hashinfo.bhash_size; i++) { spin_lock_init(&tcp_hashinfo.bhash[i].lock); INIT_HLIST_HEAD(&tcp_hashinfo.bhash[i].chain); From 6b6dc836a195e077e76977b6c020a73de411b46d Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Fri, 10 Feb 2012 11:03:00 +0100 Subject: [PATCH 4/7] vfs: Provide function to get superblock and wait for it to thaw In quota code we need to find a superblock corresponding to a device and wait for superblock to be unfrozen. However this waiting has to happen without s_umount semaphore because that is required for superblock to thaw. So provide a function in VFS for this to keep dances with s_umount where they belong. [AV: implementation switched to saner variant] Signed-off-by: Jan Kara Signed-off-by: Al Viro --- fs/super.c | 22 ++++++++++++++++++++++ include/linux/fs.h | 1 + 2 files changed, 23 insertions(+) diff --git a/fs/super.c b/fs/super.c index 6015c02296b7..6277ec6cb60a 100644 --- a/fs/super.c +++ b/fs/super.c @@ -633,6 +633,28 @@ struct super_block *get_super(struct block_device *bdev) EXPORT_SYMBOL(get_super); +/** + * get_super_thawed - get thawed superblock of a device + * @bdev: device to get the superblock for + * + * Scans the superblock list and finds the superblock of the file system + * mounted on the device. The superblock is returned once it is thawed + * (or immediately if it was not frozen). %NULL is returned if no match + * is found. + */ +struct super_block *get_super_thawed(struct block_device *bdev) +{ + while (1) { + struct super_block *s = get_super(bdev); + if (!s || s->s_frozen == SB_UNFROZEN) + return s; + up_read(&s->s_umount); + vfs_check_frozen(s, SB_FREEZE_WRITE); + put_super(s); + } +} +EXPORT_SYMBOL(get_super_thawed); + /** * get_active_super - get an active reference to the superblock of a device * @bdev: device to get the superblock for diff --git a/include/linux/fs.h b/include/linux/fs.h index 386da09f229d..69cd5bb640f5 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -2496,6 +2496,7 @@ extern void get_filesystem(struct file_system_type *fs); extern void put_filesystem(struct file_system_type *fs); extern struct file_system_type *get_fs_type(const char *name); extern struct super_block *get_super(struct block_device *); +extern struct super_block *get_super_thawed(struct block_device *); extern struct super_block *get_active_super(struct block_device *bdev); extern void drop_super(struct super_block *sb); extern void iterate_supers(void (*)(struct super_block *, void *), void *); From dcdbed853d9fbb0547b781ba676049b87f54129a Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Fri, 10 Feb 2012 11:03:01 +0100 Subject: [PATCH 5/7] quota: Fix deadlock with suspend and quotas This script causes a kernel deadlock: set -e DEVICE=/dev/vg1/linear lvchange -ay $DEVICE mkfs.ext3 $DEVICE mount -t ext3 -o usrquota,grpquota $DEVICE /mnt/test quotacheck -gu /mnt/test umount /mnt/test mount -t ext3 -o usrquota,grpquota $DEVICE /mnt/test quotaon /mnt/test dmsetup suspend $DEVICE setquota -u root 1 2 3 4 /mnt/test & sleep 1 dmsetup resume $DEVICE setquota acquired semaphore s_umount for read and then tried to perform a transaction (and waits because the device is suspended). dmsetup resume tries to acquire s_umount for write before resuming the device (and waits for setquota). Fix the deadlock by grabbing a thawed superblock for quota commands which need it. Reported-by: Mikulas Patocka Signed-off-by: Jan Kara Signed-off-by: Al Viro --- fs/quota/quota.c | 24 +++++++++++++++++++++--- 1 file changed, 21 insertions(+), 3 deletions(-) diff --git a/fs/quota/quota.c b/fs/quota/quota.c index 7898cd688a00..fc2c4388d126 100644 --- a/fs/quota/quota.c +++ b/fs/quota/quota.c @@ -292,11 +292,26 @@ static int do_quotactl(struct super_block *sb, int type, int cmd, qid_t id, } } +/* Return 1 if 'cmd' will block on frozen filesystem */ +static int quotactl_cmd_write(int cmd) +{ + switch (cmd) { + case Q_GETFMT: + case Q_GETINFO: + case Q_SYNC: + case Q_XGETQSTAT: + case Q_XGETQUOTA: + case Q_XQUOTASYNC: + return 0; + } + return 1; +} + /* * look up a superblock on which quota ops will be performed * - use the name of a block device to find the superblock thereon */ -static struct super_block *quotactl_block(const char __user *special) +static struct super_block *quotactl_block(const char __user *special, int cmd) { #ifdef CONFIG_BLOCK struct block_device *bdev; @@ -309,7 +324,10 @@ static struct super_block *quotactl_block(const char __user *special) putname(tmp); if (IS_ERR(bdev)) return ERR_CAST(bdev); - sb = get_super(bdev); + if (quotactl_cmd_write(cmd)) + sb = get_super_thawed(bdev); + else + sb = get_super(bdev); bdput(bdev); if (!sb) return ERR_PTR(-ENODEV); @@ -361,7 +379,7 @@ SYSCALL_DEFINE4(quotactl, unsigned int, cmd, const char __user *, special, pathp = &path; } - sb = quotactl_block(special); + sb = quotactl_block(special, cmds); if (IS_ERR(sb)) { ret = PTR_ERR(sb); goto out; From fcf83067bf6eb101a35620d752bd559d473cfbaa Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 12 Feb 2012 20:56:29 -0500 Subject: [PATCH 6/7] vfs: fix compat_sys_stat() handling of overflows in st_nlink Massaged cp_compat_stat() into form closer to cp_new_stat(); the only real issue had been in handling of st_nlink overflows - native 32bit stat(2) returns -EOVERFLOW in such situations, compat one silently loses upper bits. Signed-off-by: Al Viro --- fs/compat.c | 58 ++++++++++++++++++++++++----------------------------- 1 file changed, 26 insertions(+), 32 deletions(-) diff --git a/fs/compat.c b/fs/compat.c index fa9d721ecfee..07880bae28a9 100644 --- a/fs/compat.c +++ b/fs/compat.c @@ -131,41 +131,35 @@ asmlinkage long compat_sys_utimes(const char __user *filename, struct compat_tim static int cp_compat_stat(struct kstat *stat, struct compat_stat __user *ubuf) { - compat_ino_t ino = stat->ino; - typeof(ubuf->st_uid) uid = 0; - typeof(ubuf->st_gid) gid = 0; - int err; + struct compat_stat tmp; - SET_UID(uid, stat->uid); - SET_GID(gid, stat->gid); - - if ((u64) stat->size > MAX_NON_LFS || - !old_valid_dev(stat->dev) || - !old_valid_dev(stat->rdev)) - return -EOVERFLOW; - if (sizeof(ino) < sizeof(stat->ino) && ino != stat->ino) + if (!old_valid_dev(stat->dev) || !old_valid_dev(stat->rdev)) return -EOVERFLOW; - if (clear_user(ubuf, sizeof(*ubuf))) - return -EFAULT; - - err = __put_user(old_encode_dev(stat->dev), &ubuf->st_dev); - err |= __put_user(ino, &ubuf->st_ino); - err |= __put_user(stat->mode, &ubuf->st_mode); - err |= __put_user(stat->nlink, &ubuf->st_nlink); - err |= __put_user(uid, &ubuf->st_uid); - err |= __put_user(gid, &ubuf->st_gid); - err |= __put_user(old_encode_dev(stat->rdev), &ubuf->st_rdev); - err |= __put_user(stat->size, &ubuf->st_size); - err |= __put_user(stat->atime.tv_sec, &ubuf->st_atime); - err |= __put_user(stat->atime.tv_nsec, &ubuf->st_atime_nsec); - err |= __put_user(stat->mtime.tv_sec, &ubuf->st_mtime); - err |= __put_user(stat->mtime.tv_nsec, &ubuf->st_mtime_nsec); - err |= __put_user(stat->ctime.tv_sec, &ubuf->st_ctime); - err |= __put_user(stat->ctime.tv_nsec, &ubuf->st_ctime_nsec); - err |= __put_user(stat->blksize, &ubuf->st_blksize); - err |= __put_user(stat->blocks, &ubuf->st_blocks); - return err; + memset(&tmp, 0, sizeof(tmp)); + tmp.st_dev = old_encode_dev(stat->dev); + tmp.st_ino = stat->ino; + if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino) + return -EOVERFLOW; + tmp.st_mode = stat->mode; + tmp.st_nlink = stat->nlink; + if (tmp.st_nlink != stat->nlink) + return -EOVERFLOW; + SET_UID(tmp.st_uid, stat->uid); + SET_GID(tmp.st_gid, stat->gid); + tmp.st_rdev = old_encode_dev(stat->rdev); + if ((u64) stat->size > MAX_NON_LFS) + return -EOVERFLOW; + tmp.st_size = stat->size; + tmp.st_atime = stat->atime.tv_sec; + tmp.st_atime_nsec = stat->atime.tv_nsec; + tmp.st_mtime = stat->mtime.tv_sec; + tmp.st_mtime_nsec = stat->mtime.tv_nsec; + tmp.st_ctime = stat->ctime.tv_sec; + tmp.st_ctime_nsec = stat->ctime.tv_nsec; + tmp.st_blocks = stat->blocks; + tmp.st_blksize = stat->blksize; + return copy_to_user(ubuf, &tmp, sizeof(tmp)) ? -EFAULT : 0; } asmlinkage long compat_sys_newstat(const char __user * filename, From 847c9db5cb50841589b8ebd3da0769b1b02fb3b2 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 12 Feb 2012 21:00:05 -0500 Subject: [PATCH 7/7] ocfs2: deal with wraparounds of i_nlink in ocfs2_rename() unfortunately, nlink_t may be smaller than 32 bits and ->i_nlink on ocfs2 can grow up to 0xffffffff; storing it in nlink_t variable will lose upper bits on such architectures. Needs to be made u32, until we get kernel-side nlink_t uniformly 32bit... Signed-off-by: Al Viro --- fs/ocfs2/namei.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c index be244692550d..a9856e3eaaf0 100644 --- a/fs/ocfs2/namei.c +++ b/fs/ocfs2/namei.c @@ -1053,7 +1053,7 @@ static int ocfs2_rename(struct inode *old_dir, handle_t *handle = NULL; struct buffer_head *old_dir_bh = NULL; struct buffer_head *new_dir_bh = NULL; - nlink_t old_dir_nlink = old_dir->i_nlink; + u32 old_dir_nlink = old_dir->i_nlink; struct ocfs2_dinode *old_di; struct ocfs2_dir_lookup_result old_inode_dot_dot_res = { NULL, }; struct ocfs2_dir_lookup_result target_lookup_res = { NULL, };