aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
Diffstat (limited to 'fs')
-rw-r--r--fs/anon_inodes.c4
-rw-r--r--fs/compat_ioctl.c2
-rw-r--r--fs/eventfd.c2
-rw-r--r--fs/eventpoll.c2
-rw-r--r--fs/ext3/inode.c8
-rw-r--r--fs/ext3/namei.c28
-rw-r--r--fs/ext3/resize.c35
-rw-r--r--fs/ext3/super.c19
-rw-r--r--fs/ext4/ext4.h6
-rw-r--r--fs/ext4/inode.c36
-rw-r--r--fs/ext4/mballoc.c6
-rw-r--r--fs/ext4/super.c5
-rw-r--r--fs/file_table.c4
-rw-r--r--fs/internal.h7
-rw-r--r--fs/jbd/journal.c2
-rw-r--r--fs/jbd2/journal.c3
-rw-r--r--fs/jfs/super.c2
-rw-r--r--fs/namei.c8
-rw-r--r--fs/nfsd/nfsfh.c2
-rw-r--r--fs/ocfs2/Kconfig10
-rw-r--r--fs/ocfs2/Makefile7
-rw-r--r--fs/ocfs2/acl.c4
-rw-r--r--fs/ocfs2/acl.h22
-rw-r--r--fs/ocfs2/alloc.c10
-rw-r--r--fs/ocfs2/alloc.h5
-rw-r--r--fs/ocfs2/cluster/heartbeat.c6
-rw-r--r--fs/ocfs2/cluster/nodemanager.c51
-rw-r--r--fs/ocfs2/cluster/nodemanager.h7
-rw-r--r--fs/ocfs2/cluster/quorum.c16
-rw-r--r--fs/ocfs2/dlm/dlmrecovery.c18
-rw-r--r--fs/ocfs2/extent_map.c25
-rw-r--r--fs/ocfs2/namei.c6
-rw-r--r--fs/ocfs2/ocfs2.h8
-rw-r--r--fs/ocfs2/ocfs2_fs.h2
-rw-r--r--fs/ocfs2/refcounttree.c150
-rw-r--r--fs/ocfs2/stack_user.c2
-rw-r--r--fs/ocfs2/super.c95
-rw-r--r--fs/ocfs2/symlink.c2
-rw-r--r--fs/ocfs2/xattr.c6
-rw-r--r--fs/ocfs2/xattr.h2
-rw-r--r--fs/open.c13
-rw-r--r--fs/quota/dquot.c288
-rw-r--r--fs/quota/quota_v2.c9
-rw-r--r--fs/signalfd.c2
-rw-r--r--fs/stat.c10
-rw-r--r--fs/super.c3
-rw-r--r--fs/sysfs/bin.c6
-rw-r--r--fs/timerfd.c2
48 files changed, 573 insertions, 395 deletions
diff --git a/fs/anon_inodes.c b/fs/anon_inodes.c
index 2c994591f4d7..9f0bf13291e5 100644
--- a/fs/anon_inodes.c
+++ b/fs/anon_inodes.c
@@ -121,13 +121,13 @@ struct file *anon_inode_getfile(const char *name,
121 d_instantiate(path.dentry, anon_inode_inode); 121 d_instantiate(path.dentry, anon_inode_inode);
122 122
123 error = -ENFILE; 123 error = -ENFILE;
124 file = alloc_file(&path, FMODE_READ | FMODE_WRITE, fops); 124 file = alloc_file(&path, OPEN_FMODE(flags), fops);
125 if (!file) 125 if (!file)
126 goto err_dput; 126 goto err_dput;
127 file->f_mapping = anon_inode_inode->i_mapping; 127 file->f_mapping = anon_inode_inode->i_mapping;
128 128
129 file->f_pos = 0; 129 file->f_pos = 0;
130 file->f_flags = O_RDWR | (flags & O_NONBLOCK); 130 file->f_flags = flags & (O_ACCMODE | O_NONBLOCK);
131 file->f_version = 0; 131 file->f_version = 0;
132 file->private_data = priv; 132 file->private_data = priv;
133 133
diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c
index 14cbc831422a..332dd00f0894 100644
--- a/fs/compat_ioctl.c
+++ b/fs/compat_ioctl.c
@@ -1600,8 +1600,6 @@ static long do_ioctl_trans(int fd, unsigned int cmd,
1600 case KDSKBMETA: 1600 case KDSKBMETA:
1601 case KDSKBLED: 1601 case KDSKBLED:
1602 case KDSETLED: 1602 case KDSETLED:
1603 /* SG stuff */
1604 case SG_SET_TRANSFORM:
1605 /* AUTOFS */ 1603 /* AUTOFS */
1606 case AUTOFS_IOC_READY: 1604 case AUTOFS_IOC_READY:
1607 case AUTOFS_IOC_FAIL: 1605 case AUTOFS_IOC_FAIL:
diff --git a/fs/eventfd.c b/fs/eventfd.c
index 8b47e4200e65..d26402ff06ea 100644
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -339,7 +339,7 @@ struct file *eventfd_file_create(unsigned int count, int flags)
339 ctx->flags = flags; 339 ctx->flags = flags;
340 340
341 file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx, 341 file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx,
342 flags & EFD_SHARED_FCNTL_FLAGS); 342 O_RDWR | (flags & EFD_SHARED_FCNTL_FLAGS));
343 if (IS_ERR(file)) 343 if (IS_ERR(file))
344 eventfd_free_ctx(ctx); 344 eventfd_free_ctx(ctx);
345 345
diff --git a/fs/eventpoll.c b/fs/eventpoll.c
index 366c503f9657..bd056a5b4efc 100644
--- a/fs/eventpoll.c
+++ b/fs/eventpoll.c
@@ -1206,7 +1206,7 @@ SYSCALL_DEFINE1(epoll_create1, int, flags)
1206 * a file structure and a free file descriptor. 1206 * a file structure and a free file descriptor.
1207 */ 1207 */
1208 error = anon_inode_getfd("[eventpoll]", &eventpoll_fops, ep, 1208 error = anon_inode_getfd("[eventpoll]", &eventpoll_fops, ep,
1209 flags & O_CLOEXEC); 1209 O_RDWR | (flags & O_CLOEXEC));
1210 if (error < 0) 1210 if (error < 0)
1211 ep_free(ep); 1211 ep_free(ep);
1212 1212
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index ad14227f509e..455e6e6e5cb9 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -970,7 +970,7 @@ static int ext3_get_block(struct inode *inode, sector_t iblock,
970 if (max_blocks > DIO_MAX_BLOCKS) 970 if (max_blocks > DIO_MAX_BLOCKS)
971 max_blocks = DIO_MAX_BLOCKS; 971 max_blocks = DIO_MAX_BLOCKS;
972 handle = ext3_journal_start(inode, DIO_CREDITS + 972 handle = ext3_journal_start(inode, DIO_CREDITS +
973 2 * EXT3_QUOTA_TRANS_BLOCKS(inode->i_sb)); 973 EXT3_MAXQUOTAS_TRANS_BLOCKS(inode->i_sb));
974 if (IS_ERR(handle)) { 974 if (IS_ERR(handle)) {
975 ret = PTR_ERR(handle); 975 ret = PTR_ERR(handle);
976 goto out; 976 goto out;
@@ -3146,8 +3146,8 @@ int ext3_setattr(struct dentry *dentry, struct iattr *attr)
3146 3146
3147 /* (user+group)*(old+new) structure, inode write (sb, 3147 /* (user+group)*(old+new) structure, inode write (sb,
3148 * inode block, ? - but truncate inode update has it) */ 3148 * inode block, ? - but truncate inode update has it) */
3149 handle = ext3_journal_start(inode, 2*(EXT3_QUOTA_INIT_BLOCKS(inode->i_sb)+ 3149 handle = ext3_journal_start(inode, EXT3_MAXQUOTAS_INIT_BLOCKS(inode->i_sb)+
3150 EXT3_QUOTA_DEL_BLOCKS(inode->i_sb))+3); 3150 EXT3_MAXQUOTAS_DEL_BLOCKS(inode->i_sb)+3);
3151 if (IS_ERR(handle)) { 3151 if (IS_ERR(handle)) {
3152 error = PTR_ERR(handle); 3152 error = PTR_ERR(handle);
3153 goto err_out; 3153 goto err_out;
@@ -3239,7 +3239,7 @@ static int ext3_writepage_trans_blocks(struct inode *inode)
3239#ifdef CONFIG_QUOTA 3239#ifdef CONFIG_QUOTA
3240 /* We know that structure was already allocated during vfs_dq_init so 3240 /* We know that structure was already allocated during vfs_dq_init so
3241 * we will be updating only the data blocks + inodes */ 3241 * we will be updating only the data blocks + inodes */
3242 ret += 2*EXT3_QUOTA_TRANS_BLOCKS(inode->i_sb); 3242 ret += EXT3_MAXQUOTAS_TRANS_BLOCKS(inode->i_sb);
3243#endif 3243#endif
3244 3244
3245 return ret; 3245 return ret;
diff --git a/fs/ext3/namei.c b/fs/ext3/namei.c
index aad6400c9b77..7b0e44f7d66f 100644
--- a/fs/ext3/namei.c
+++ b/fs/ext3/namei.c
@@ -1699,7 +1699,7 @@ static int ext3_create (struct inode * dir, struct dentry * dentry, int mode,
1699retry: 1699retry:
1700 handle = ext3_journal_start(dir, EXT3_DATA_TRANS_BLOCKS(dir->i_sb) + 1700 handle = ext3_journal_start(dir, EXT3_DATA_TRANS_BLOCKS(dir->i_sb) +
1701 EXT3_INDEX_EXTRA_TRANS_BLOCKS + 3 + 1701 EXT3_INDEX_EXTRA_TRANS_BLOCKS + 3 +
1702 2*EXT3_QUOTA_INIT_BLOCKS(dir->i_sb)); 1702 EXT3_MAXQUOTAS_INIT_BLOCKS(dir->i_sb));
1703 if (IS_ERR(handle)) 1703 if (IS_ERR(handle))
1704 return PTR_ERR(handle); 1704 return PTR_ERR(handle);
1705 1705
@@ -1733,7 +1733,7 @@ static int ext3_mknod (struct inode * dir, struct dentry *dentry,
1733retry: 1733retry:
1734 handle = ext3_journal_start(dir, EXT3_DATA_TRANS_BLOCKS(dir->i_sb) + 1734 handle = ext3_journal_start(dir, EXT3_DATA_TRANS_BLOCKS(dir->i_sb) +
1735 EXT3_INDEX_EXTRA_TRANS_BLOCKS + 3 + 1735 EXT3_INDEX_EXTRA_TRANS_BLOCKS + 3 +
1736 2*EXT3_QUOTA_INIT_BLOCKS(dir->i_sb)); 1736 EXT3_MAXQUOTAS_INIT_BLOCKS(dir->i_sb));
1737 if (IS_ERR(handle)) 1737 if (IS_ERR(handle))
1738 return PTR_ERR(handle); 1738 return PTR_ERR(handle);
1739 1739
@@ -1769,7 +1769,7 @@ static int ext3_mkdir(struct inode * dir, struct dentry * dentry, int mode)
1769retry: 1769retry:
1770 handle = ext3_journal_start(dir, EXT3_DATA_TRANS_BLOCKS(dir->i_sb) + 1770 handle = ext3_journal_start(dir, EXT3_DATA_TRANS_BLOCKS(dir->i_sb) +
1771 EXT3_INDEX_EXTRA_TRANS_BLOCKS + 3 + 1771 EXT3_INDEX_EXTRA_TRANS_BLOCKS + 3 +
1772 2*EXT3_QUOTA_INIT_BLOCKS(dir->i_sb)); 1772 EXT3_MAXQUOTAS_INIT_BLOCKS(dir->i_sb));
1773 if (IS_ERR(handle)) 1773 if (IS_ERR(handle))
1774 return PTR_ERR(handle); 1774 return PTR_ERR(handle);
1775 1775
@@ -1920,7 +1920,7 @@ int ext3_orphan_add(handle_t *handle, struct inode *inode)
1920 struct ext3_iloc iloc; 1920 struct ext3_iloc iloc;
1921 int err = 0, rc; 1921 int err = 0, rc;
1922 1922
1923 lock_super(sb); 1923 mutex_lock(&EXT3_SB(sb)->s_orphan_lock);
1924 if (!list_empty(&EXT3_I(inode)->i_orphan)) 1924 if (!list_empty(&EXT3_I(inode)->i_orphan))
1925 goto out_unlock; 1925 goto out_unlock;
1926 1926
@@ -1929,9 +1929,13 @@ int ext3_orphan_add(handle_t *handle, struct inode *inode)
1929 1929
1930 /* @@@ FIXME: Observation from aviro: 1930 /* @@@ FIXME: Observation from aviro:
1931 * I think I can trigger J_ASSERT in ext3_orphan_add(). We block 1931 * I think I can trigger J_ASSERT in ext3_orphan_add(). We block
1932 * here (on lock_super()), so race with ext3_link() which might bump 1932 * here (on s_orphan_lock), so race with ext3_link() which might bump
1933 * ->i_nlink. For, say it, character device. Not a regular file, 1933 * ->i_nlink. For, say it, character device. Not a regular file,
1934 * not a directory, not a symlink and ->i_nlink > 0. 1934 * not a directory, not a symlink and ->i_nlink > 0.
1935 *
1936 * tytso, 4/25/2009: I'm not sure how that could happen;
1937 * shouldn't the fs core protect us from these sort of
1938 * unlink()/link() races?
1935 */ 1939 */
1936 J_ASSERT ((S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || 1940 J_ASSERT ((S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1937 S_ISLNK(inode->i_mode)) || inode->i_nlink == 0); 1941 S_ISLNK(inode->i_mode)) || inode->i_nlink == 0);
@@ -1968,7 +1972,7 @@ int ext3_orphan_add(handle_t *handle, struct inode *inode)
1968 jbd_debug(4, "orphan inode %lu will point to %d\n", 1972 jbd_debug(4, "orphan inode %lu will point to %d\n",
1969 inode->i_ino, NEXT_ORPHAN(inode)); 1973 inode->i_ino, NEXT_ORPHAN(inode));
1970out_unlock: 1974out_unlock:
1971 unlock_super(sb); 1975 mutex_unlock(&EXT3_SB(sb)->s_orphan_lock);
1972 ext3_std_error(inode->i_sb, err); 1976 ext3_std_error(inode->i_sb, err);
1973 return err; 1977 return err;
1974} 1978}
@@ -1986,11 +1990,9 @@ int ext3_orphan_del(handle_t *handle, struct inode *inode)
1986 struct ext3_iloc iloc; 1990 struct ext3_iloc iloc;
1987 int err = 0; 1991 int err = 0;
1988 1992
1989 lock_super(inode->i_sb); 1993 mutex_lock(&EXT3_SB(inode->i_sb)->s_orphan_lock);
1990 if (list_empty(&ei->i_orphan)) { 1994 if (list_empty(&ei->i_orphan))
1991 unlock_super(inode->i_sb); 1995 goto out;
1992 return 0;
1993 }
1994 1996
1995 ino_next = NEXT_ORPHAN(inode); 1997 ino_next = NEXT_ORPHAN(inode);
1996 prev = ei->i_orphan.prev; 1998 prev = ei->i_orphan.prev;
@@ -2040,7 +2042,7 @@ int ext3_orphan_del(handle_t *handle, struct inode *inode)
2040out_err: 2042out_err:
2041 ext3_std_error(inode->i_sb, err); 2043 ext3_std_error(inode->i_sb, err);
2042out: 2044out:
2043 unlock_super(inode->i_sb); 2045 mutex_unlock(&EXT3_SB(inode->i_sb)->s_orphan_lock);
2044 return err; 2046 return err;
2045 2047
2046out_brelse: 2048out_brelse:
@@ -2175,7 +2177,7 @@ static int ext3_symlink (struct inode * dir,
2175retry: 2177retry:
2176 handle = ext3_journal_start(dir, EXT3_DATA_TRANS_BLOCKS(dir->i_sb) + 2178 handle = ext3_journal_start(dir, EXT3_DATA_TRANS_BLOCKS(dir->i_sb) +
2177 EXT3_INDEX_EXTRA_TRANS_BLOCKS + 5 + 2179 EXT3_INDEX_EXTRA_TRANS_BLOCKS + 5 +
2178 2*EXT3_QUOTA_INIT_BLOCKS(dir->i_sb)); 2180 EXT3_MAXQUOTAS_INIT_BLOCKS(dir->i_sb));
2179 if (IS_ERR(handle)) 2181 if (IS_ERR(handle))
2180 return PTR_ERR(handle); 2182 return PTR_ERR(handle);
2181 2183
diff --git a/fs/ext3/resize.c b/fs/ext3/resize.c
index 5f83b6179178..54351ac7cef9 100644
--- a/fs/ext3/resize.c
+++ b/fs/ext3/resize.c
@@ -209,7 +209,7 @@ static int setup_new_group_blocks(struct super_block *sb,
209 if (IS_ERR(handle)) 209 if (IS_ERR(handle))
210 return PTR_ERR(handle); 210 return PTR_ERR(handle);
211 211
212 lock_super(sb); 212 mutex_lock(&sbi->s_resize_lock);
213 if (input->group != sbi->s_groups_count) { 213 if (input->group != sbi->s_groups_count) {
214 err = -EBUSY; 214 err = -EBUSY;
215 goto exit_journal; 215 goto exit_journal;
@@ -324,7 +324,7 @@ exit_bh:
324 brelse(bh); 324 brelse(bh);
325 325
326exit_journal: 326exit_journal:
327 unlock_super(sb); 327 mutex_unlock(&sbi->s_resize_lock);
328 if ((err2 = ext3_journal_stop(handle)) && !err) 328 if ((err2 = ext3_journal_stop(handle)) && !err)
329 err = err2; 329 err = err2;
330 330
@@ -662,11 +662,12 @@ exit_free:
662 * important part is that the new block and inode counts are in the backup 662 * important part is that the new block and inode counts are in the backup
663 * superblocks, and the location of the new group metadata in the GDT backups. 663 * superblocks, and the location of the new group metadata in the GDT backups.
664 * 664 *
665 * We do not need lock_super() for this, because these blocks are not 665 * We do not need take the s_resize_lock for this, because these
666 * otherwise touched by the filesystem code when it is mounted. We don't 666 * blocks are not otherwise touched by the filesystem code when it is
667 * need to worry about last changing from sbi->s_groups_count, because the 667 * mounted. We don't need to worry about last changing from
668 * worst that can happen is that we do not copy the full number of backups 668 * sbi->s_groups_count, because the worst that can happen is that we
669 * at this time. The resize which changed s_groups_count will backup again. 669 * do not copy the full number of backups at this time. The resize
670 * which changed s_groups_count will backup again.
670 */ 671 */
671static void update_backups(struct super_block *sb, 672static void update_backups(struct super_block *sb,
672 int blk_off, char *data, int size) 673 int blk_off, char *data, int size)
@@ -825,7 +826,7 @@ int ext3_group_add(struct super_block *sb, struct ext3_new_group_data *input)
825 goto exit_put; 826 goto exit_put;
826 } 827 }
827 828
828 lock_super(sb); 829 mutex_lock(&sbi->s_resize_lock);
829 if (input->group != sbi->s_groups_count) { 830 if (input->group != sbi->s_groups_count) {
830 ext3_warning(sb, __func__, 831 ext3_warning(sb, __func__,
831 "multiple resizers run on filesystem!"); 832 "multiple resizers run on filesystem!");
@@ -856,7 +857,7 @@ int ext3_group_add(struct super_block *sb, struct ext3_new_group_data *input)
856 /* 857 /*
857 * OK, now we've set up the new group. Time to make it active. 858 * OK, now we've set up the new group. Time to make it active.
858 * 859 *
859 * Current kernels don't lock all allocations via lock_super(), 860 * We do not lock all allocations via s_resize_lock
860 * so we have to be safe wrt. concurrent accesses the group 861 * so we have to be safe wrt. concurrent accesses the group
861 * data. So we need to be careful to set all of the relevant 862 * data. So we need to be careful to set all of the relevant
862 * group descriptor data etc. *before* we enable the group. 863 * group descriptor data etc. *before* we enable the group.
@@ -900,12 +901,12 @@ int ext3_group_add(struct super_block *sb, struct ext3_new_group_data *input)
900 * 901 *
901 * The precise rules we use are: 902 * The precise rules we use are:
902 * 903 *
903 * * Writers of s_groups_count *must* hold lock_super 904 * * Writers of s_groups_count *must* hold s_resize_lock
904 * AND 905 * AND
905 * * Writers must perform a smp_wmb() after updating all dependent 906 * * Writers must perform a smp_wmb() after updating all dependent
906 * data and before modifying the groups count 907 * data and before modifying the groups count
907 * 908 *
908 * * Readers must hold lock_super() over the access 909 * * Readers must hold s_resize_lock over the access
909 * OR 910 * OR
910 * * Readers must perform an smp_rmb() after reading the groups count 911 * * Readers must perform an smp_rmb() after reading the groups count
911 * and before reading any dependent data. 912 * and before reading any dependent data.
@@ -936,7 +937,7 @@ int ext3_group_add(struct super_block *sb, struct ext3_new_group_data *input)
936 ext3_journal_dirty_metadata(handle, sbi->s_sbh); 937 ext3_journal_dirty_metadata(handle, sbi->s_sbh);
937 938
938exit_journal: 939exit_journal:
939 unlock_super(sb); 940 mutex_unlock(&sbi->s_resize_lock);
940 if ((err2 = ext3_journal_stop(handle)) && !err) 941 if ((err2 = ext3_journal_stop(handle)) && !err)
941 err = err2; 942 err = err2;
942 if (!err) { 943 if (!err) {
@@ -973,7 +974,7 @@ int ext3_group_extend(struct super_block *sb, struct ext3_super_block *es,
973 974
974 /* We don't need to worry about locking wrt other resizers just 975 /* We don't need to worry about locking wrt other resizers just
975 * yet: we're going to revalidate es->s_blocks_count after 976 * yet: we're going to revalidate es->s_blocks_count after
976 * taking lock_super() below. */ 977 * taking the s_resize_lock below. */
977 o_blocks_count = le32_to_cpu(es->s_blocks_count); 978 o_blocks_count = le32_to_cpu(es->s_blocks_count);
978 o_groups_count = EXT3_SB(sb)->s_groups_count; 979 o_groups_count = EXT3_SB(sb)->s_groups_count;
979 980
@@ -1045,11 +1046,11 @@ int ext3_group_extend(struct super_block *sb, struct ext3_super_block *es,
1045 goto exit_put; 1046 goto exit_put;
1046 } 1047 }
1047 1048
1048 lock_super(sb); 1049 mutex_lock(&EXT3_SB(sb)->s_resize_lock);
1049 if (o_blocks_count != le32_to_cpu(es->s_blocks_count)) { 1050 if (o_blocks_count != le32_to_cpu(es->s_blocks_count)) {
1050 ext3_warning(sb, __func__, 1051 ext3_warning(sb, __func__,
1051 "multiple resizers run on filesystem!"); 1052 "multiple resizers run on filesystem!");
1052 unlock_super(sb); 1053 mutex_unlock(&EXT3_SB(sb)->s_resize_lock);
1053 ext3_journal_stop(handle); 1054 ext3_journal_stop(handle);
1054 err = -EBUSY; 1055 err = -EBUSY;
1055 goto exit_put; 1056 goto exit_put;
@@ -1059,13 +1060,13 @@ int ext3_group_extend(struct super_block *sb, struct ext3_super_block *es,
1059 EXT3_SB(sb)->s_sbh))) { 1060 EXT3_SB(sb)->s_sbh))) {
1060 ext3_warning(sb, __func__, 1061 ext3_warning(sb, __func__,
1061 "error %d on journal write access", err); 1062 "error %d on journal write access", err);
1062 unlock_super(sb); 1063 mutex_unlock(&EXT3_SB(sb)->s_resize_lock);
1063 ext3_journal_stop(handle); 1064 ext3_journal_stop(handle);
1064 goto exit_put; 1065 goto exit_put;
1065 } 1066 }
1066 es->s_blocks_count = cpu_to_le32(o_blocks_count + add); 1067 es->s_blocks_count = cpu_to_le32(o_blocks_count + add);
1067 ext3_journal_dirty_metadata(handle, EXT3_SB(sb)->s_sbh); 1068 ext3_journal_dirty_metadata(handle, EXT3_SB(sb)->s_sbh);
1068 unlock_super(sb); 1069 mutex_unlock(&EXT3_SB(sb)->s_resize_lock);
1069 ext3_debug("freeing blocks %lu through "E3FSBLK"\n", o_blocks_count, 1070 ext3_debug("freeing blocks %lu through "E3FSBLK"\n", o_blocks_count,
1070 o_blocks_count + add); 1071 o_blocks_count + add);
1071 ext3_free_blocks_sb(handle, sb, o_blocks_count, add, &freed_blocks); 1072 ext3_free_blocks_sb(handle, sb, o_blocks_count, add, &freed_blocks);
diff --git a/fs/ext3/super.c b/fs/ext3/super.c
index 7ad1e8c30bd0..afa2b569da10 100644
--- a/fs/ext3/super.c
+++ b/fs/ext3/super.c
@@ -1928,6 +1928,8 @@ static int ext3_fill_super (struct super_block *sb, void *data, int silent)
1928 sb->dq_op = &ext3_quota_operations; 1928 sb->dq_op = &ext3_quota_operations;
1929#endif 1929#endif
1930 INIT_LIST_HEAD(&sbi->s_orphan); /* unlinked but open files */ 1930 INIT_LIST_HEAD(&sbi->s_orphan); /* unlinked but open files */
1931 mutex_init(&sbi->s_orphan_lock);
1932 mutex_init(&sbi->s_resize_lock);
1931 1933
1932 sb->s_root = NULL; 1934 sb->s_root = NULL;
1933 1935
@@ -2014,14 +2016,7 @@ static int ext3_fill_super (struct super_block *sb, void *data, int silent)
2014 } 2016 }
2015 2017
2016 ext3_setup_super (sb, es, sb->s_flags & MS_RDONLY); 2018 ext3_setup_super (sb, es, sb->s_flags & MS_RDONLY);
2017 /* 2019
2018 * akpm: core read_super() calls in here with the superblock locked.
2019 * That deadlocks, because orphan cleanup needs to lock the superblock
2020 * in numerous places. Here we just pop the lock - it's relatively
2021 * harmless, because we are now ready to accept write_super() requests,
2022 * and aviro says that's the only reason for hanging onto the
2023 * superblock lock.
2024 */
2025 EXT3_SB(sb)->s_mount_state |= EXT3_ORPHAN_FS; 2020 EXT3_SB(sb)->s_mount_state |= EXT3_ORPHAN_FS;
2026 ext3_orphan_cleanup(sb, es); 2021 ext3_orphan_cleanup(sb, es);
2027 EXT3_SB(sb)->s_mount_state &= ~EXT3_ORPHAN_FS; 2022 EXT3_SB(sb)->s_mount_state &= ~EXT3_ORPHAN_FS;
@@ -2403,13 +2398,11 @@ static void ext3_mark_recovery_complete(struct super_block * sb,
2403 if (journal_flush(journal) < 0) 2398 if (journal_flush(journal) < 0)
2404 goto out; 2399 goto out;
2405 2400
2406 lock_super(sb);
2407 if (EXT3_HAS_INCOMPAT_FEATURE(sb, EXT3_FEATURE_INCOMPAT_RECOVER) && 2401 if (EXT3_HAS_INCOMPAT_FEATURE(sb, EXT3_FEATURE_INCOMPAT_RECOVER) &&
2408 sb->s_flags & MS_RDONLY) { 2402 sb->s_flags & MS_RDONLY) {
2409 EXT3_CLEAR_INCOMPAT_FEATURE(sb, EXT3_FEATURE_INCOMPAT_RECOVER); 2403 EXT3_CLEAR_INCOMPAT_FEATURE(sb, EXT3_FEATURE_INCOMPAT_RECOVER);
2410 ext3_commit_super(sb, es, 1); 2404 ext3_commit_super(sb, es, 1);
2411 } 2405 }
2412 unlock_super(sb);
2413 2406
2414out: 2407out:
2415 journal_unlock_updates(journal); 2408 journal_unlock_updates(journal);
@@ -2601,13 +2594,7 @@ static int ext3_remount (struct super_block * sb, int * flags, char * data)
2601 (sbi->s_mount_state & EXT3_VALID_FS)) 2594 (sbi->s_mount_state & EXT3_VALID_FS))
2602 es->s_state = cpu_to_le16(sbi->s_mount_state); 2595 es->s_state = cpu_to_le16(sbi->s_mount_state);
2603 2596
2604 /*
2605 * We have to unlock super so that we can wait for
2606 * transactions.
2607 */
2608 unlock_super(sb);
2609 ext3_mark_recovery_complete(sb, es); 2597 ext3_mark_recovery_complete(sb, es);
2610 lock_super(sb);
2611 } else { 2598 } else {
2612 __le32 ret; 2599 __le32 ret;
2613 if ((ret = EXT3_HAS_RO_COMPAT_FEATURE(sb, 2600 if ((ret = EXT3_HAS_RO_COMPAT_FEATURE(sb,
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index ab31e65d46d0..56f9271ee8cc 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -704,6 +704,10 @@ struct ext4_inode_info {
704 __u16 i_extra_isize; 704 __u16 i_extra_isize;
705 705
706 spinlock_t i_block_reservation_lock; 706 spinlock_t i_block_reservation_lock;
707#ifdef CONFIG_QUOTA
708 /* quota space reservation, managed internally by quota code */
709 qsize_t i_reserved_quota;
710#endif
707 711
708 /* completed async DIOs that might need unwritten extents handling */ 712 /* completed async DIOs that might need unwritten extents handling */
709 struct list_head i_aio_dio_complete_list; 713 struct list_head i_aio_dio_complete_list;
@@ -1435,7 +1439,7 @@ extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1435extern int ext4_block_truncate_page(handle_t *handle, 1439extern int ext4_block_truncate_page(handle_t *handle,
1436 struct address_space *mapping, loff_t from); 1440 struct address_space *mapping, loff_t from);
1437extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 1441extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1438extern qsize_t ext4_get_reserved_space(struct inode *inode); 1442extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1439extern int flush_aio_dio_completed_IO(struct inode *inode); 1443extern int flush_aio_dio_completed_IO(struct inode *inode);
1440/* ioctl.c */ 1444/* ioctl.c */
1441extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 1445extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 5352db1a3086..ab807963a614 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1003,17 +1003,12 @@ out:
1003 return err; 1003 return err;
1004} 1004}
1005 1005
1006qsize_t ext4_get_reserved_space(struct inode *inode) 1006#ifdef CONFIG_QUOTA
1007qsize_t *ext4_get_reserved_space(struct inode *inode)
1007{ 1008{
1008 unsigned long long total; 1009 return &EXT4_I(inode)->i_reserved_quota;
1009
1010 spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
1011 total = EXT4_I(inode)->i_reserved_data_blocks +
1012 EXT4_I(inode)->i_reserved_meta_blocks;
1013 spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1014
1015 return (total << inode->i_blkbits);
1016} 1010}
1011#endif
1017/* 1012/*
1018 * Calculate the number of metadata blocks need to reserve 1013 * Calculate the number of metadata blocks need to reserve
1019 * to allocate @blocks for non extent file based file 1014 * to allocate @blocks for non extent file based file
@@ -1051,7 +1046,7 @@ static int ext4_calc_metadata_amount(struct inode *inode, int blocks)
1051static void ext4_da_update_reserve_space(struct inode *inode, int used) 1046static void ext4_da_update_reserve_space(struct inode *inode, int used)
1052{ 1047{
1053 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); 1048 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
1054 int total, mdb, mdb_free; 1049 int total, mdb, mdb_free, mdb_claim = 0;
1055 1050
1056 spin_lock(&EXT4_I(inode)->i_block_reservation_lock); 1051 spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
1057 /* recalculate the number of metablocks still need to be reserved */ 1052 /* recalculate the number of metablocks still need to be reserved */
@@ -1064,7 +1059,9 @@ static void ext4_da_update_reserve_space(struct inode *inode, int used)
1064 1059
1065 if (mdb_free) { 1060 if (mdb_free) {
1066 /* Account for allocated meta_blocks */ 1061 /* Account for allocated meta_blocks */
1067 mdb_free -= EXT4_I(inode)->i_allocated_meta_blocks; 1062 mdb_claim = EXT4_I(inode)->i_allocated_meta_blocks;
1063 BUG_ON(mdb_free < mdb_claim);
1064 mdb_free -= mdb_claim;
1068 1065
1069 /* update fs dirty blocks counter */ 1066 /* update fs dirty blocks counter */
1070 percpu_counter_sub(&sbi->s_dirtyblocks_counter, mdb_free); 1067 percpu_counter_sub(&sbi->s_dirtyblocks_counter, mdb_free);
@@ -1075,8 +1072,11 @@ static void ext4_da_update_reserve_space(struct inode *inode, int used)
1075 /* update per-inode reservations */ 1072 /* update per-inode reservations */
1076 BUG_ON(used > EXT4_I(inode)->i_reserved_data_blocks); 1073 BUG_ON(used > EXT4_I(inode)->i_reserved_data_blocks);
1077 EXT4_I(inode)->i_reserved_data_blocks -= used; 1074 EXT4_I(inode)->i_reserved_data_blocks -= used;
1075 percpu_counter_sub(&sbi->s_dirtyblocks_counter, used + mdb_claim);
1078 spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); 1076 spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1079 1077
1078 vfs_dq_claim_block(inode, used + mdb_claim);
1079
1080 /* 1080 /*
1081 * free those over-booking quota for metadata blocks 1081 * free those over-booking quota for metadata blocks
1082 */ 1082 */
@@ -1816,19 +1816,17 @@ repeat:
1816 1816
1817 md_needed = mdblocks - EXT4_I(inode)->i_reserved_meta_blocks; 1817 md_needed = mdblocks - EXT4_I(inode)->i_reserved_meta_blocks;
1818 total = md_needed + nrblocks; 1818 total = md_needed + nrblocks;
1819 spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1819 1820
1820 /* 1821 /*
1821 * Make quota reservation here to prevent quota overflow 1822 * Make quota reservation here to prevent quota overflow
1822 * later. Real quota accounting is done at pages writeout 1823 * later. Real quota accounting is done at pages writeout
1823 * time. 1824 * time.
1824 */ 1825 */
1825 if (vfs_dq_reserve_block(inode, total)) { 1826 if (vfs_dq_reserve_block(inode, total))
1826 spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1827 return -EDQUOT; 1827 return -EDQUOT;
1828 }
1829 1828
1830 if (ext4_claim_free_blocks(sbi, total)) { 1829 if (ext4_claim_free_blocks(sbi, total)) {
1831 spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1832 vfs_dq_release_reservation_block(inode, total); 1830 vfs_dq_release_reservation_block(inode, total);
1833 if (ext4_should_retry_alloc(inode->i_sb, &retries)) { 1831 if (ext4_should_retry_alloc(inode->i_sb, &retries)) {
1834 yield(); 1832 yield();
@@ -1836,10 +1834,11 @@ repeat:
1836 } 1834 }
1837 return -ENOSPC; 1835 return -ENOSPC;
1838 } 1836 }
1837 spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
1839 EXT4_I(inode)->i_reserved_data_blocks += nrblocks; 1838 EXT4_I(inode)->i_reserved_data_blocks += nrblocks;
1840 EXT4_I(inode)->i_reserved_meta_blocks = mdblocks; 1839 EXT4_I(inode)->i_reserved_meta_blocks += md_needed;
1841
1842 spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); 1840 spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1841
1843 return 0; /* success */ 1842 return 0; /* success */
1844} 1843}
1845 1844
@@ -4794,6 +4793,9 @@ struct inode *ext4_iget(struct super_block *sb, unsigned long ino)
4794 ((__u64)le16_to_cpu(raw_inode->i_file_acl_high)) << 32; 4793 ((__u64)le16_to_cpu(raw_inode->i_file_acl_high)) << 32;
4795 inode->i_size = ext4_isize(raw_inode); 4794 inode->i_size = ext4_isize(raw_inode);
4796 ei->i_disksize = inode->i_size; 4795 ei->i_disksize = inode->i_size;
4796#ifdef CONFIG_QUOTA
4797 ei->i_reserved_quota = 0;
4798#endif
4797 inode->i_generation = le32_to_cpu(raw_inode->i_generation); 4799 inode->i_generation = le32_to_cpu(raw_inode->i_generation);
4798 ei->i_block_group = iloc.block_group; 4800 ei->i_block_group = iloc.block_group;
4799 ei->i_last_alloc_group = ~0; 4801 ei->i_last_alloc_group = ~0;
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index b1fd3daadc9c..d34afad3e137 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -2755,12 +2755,6 @@ ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
2755 if (!(ac->ac_flags & EXT4_MB_DELALLOC_RESERVED)) 2755 if (!(ac->ac_flags & EXT4_MB_DELALLOC_RESERVED))
2756 /* release all the reserved blocks if non delalloc */ 2756 /* release all the reserved blocks if non delalloc */
2757 percpu_counter_sub(&sbi->s_dirtyblocks_counter, reserv_blks); 2757 percpu_counter_sub(&sbi->s_dirtyblocks_counter, reserv_blks);
2758 else {
2759 percpu_counter_sub(&sbi->s_dirtyblocks_counter,
2760 ac->ac_b_ex.fe_len);
2761 /* convert reserved quota blocks to real quota blocks */
2762 vfs_dq_claim_block(ac->ac_inode, ac->ac_b_ex.fe_len);
2763 }
2764 2758
2765 if (sbi->s_log_groups_per_flex) { 2759 if (sbi->s_log_groups_per_flex) {
2766 ext4_group_t flex_group = ext4_flex_group(sbi, 2760 ext4_group_t flex_group = ext4_flex_group(sbi,
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 827bde1f2594..6ed9aa91f27d 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -704,6 +704,9 @@ static struct inode *ext4_alloc_inode(struct super_block *sb)
704 ei->i_allocated_meta_blocks = 0; 704 ei->i_allocated_meta_blocks = 0;
705 ei->i_delalloc_reserved_flag = 0; 705 ei->i_delalloc_reserved_flag = 0;
706 spin_lock_init(&(ei->i_block_reservation_lock)); 706 spin_lock_init(&(ei->i_block_reservation_lock));
707#ifdef CONFIG_QUOTA
708 ei->i_reserved_quota = 0;
709#endif
707 INIT_LIST_HEAD(&ei->i_aio_dio_complete_list); 710 INIT_LIST_HEAD(&ei->i_aio_dio_complete_list);
708 ei->cur_aio_dio = NULL; 711 ei->cur_aio_dio = NULL;
709 ei->i_sync_tid = 0; 712 ei->i_sync_tid = 0;
@@ -1014,7 +1017,9 @@ static const struct dquot_operations ext4_quota_operations = {
1014 .reserve_space = dquot_reserve_space, 1017 .reserve_space = dquot_reserve_space,
1015 .claim_space = dquot_claim_space, 1018 .claim_space = dquot_claim_space,
1016 .release_rsv = dquot_release_reserved_space, 1019 .release_rsv = dquot_release_reserved_space,
1020#ifdef CONFIG_QUOTA
1017 .get_reserved_space = ext4_get_reserved_space, 1021 .get_reserved_space = ext4_get_reserved_space,
1022#endif
1018 .alloc_inode = dquot_alloc_inode, 1023 .alloc_inode = dquot_alloc_inode,
1019 .free_space = dquot_free_space, 1024 .free_space = dquot_free_space,
1020 .free_inode = dquot_free_inode, 1025 .free_inode = dquot_free_inode,
diff --git a/fs/file_table.c b/fs/file_table.c
index 0afacf654398..69652c5bd5f0 100644
--- a/fs/file_table.c
+++ b/fs/file_table.c
@@ -186,10 +186,8 @@ struct file *alloc_file(struct path *path, fmode_t mode,
186 * that we can do debugging checks at __fput() 186 * that we can do debugging checks at __fput()
187 */ 187 */
188 if ((mode & FMODE_WRITE) && !special_file(path->dentry->d_inode->i_mode)) { 188 if ((mode & FMODE_WRITE) && !special_file(path->dentry->d_inode->i_mode)) {
189 int error = 0;
190 file_take_write(file); 189 file_take_write(file);
191 error = mnt_clone_write(path->mnt); 190 WARN_ON(mnt_clone_write(path->mnt));
192 WARN_ON(error);
193 } 191 }
194 ima_counts_get(file); 192 ima_counts_get(file);
195 return file; 193 return file;
diff --git a/fs/internal.h b/fs/internal.h
index f67cd141d9a8..e96a1667d749 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -85,3 +85,10 @@ extern struct file *get_empty_filp(void);
85 * super.c 85 * super.c
86 */ 86 */
87extern int do_remount_sb(struct super_block *, int, void *, int); 87extern int do_remount_sb(struct super_block *, int, void *, int);
88
89/*
90 * open.c
91 */
92struct nameidata;
93extern struct file *nameidata_to_filp(struct nameidata *);
94extern void release_open_intent(struct nameidata *);
diff --git a/fs/jbd/journal.c b/fs/jbd/journal.c
index 4160afad6d00..bd224eec9b07 100644
--- a/fs/jbd/journal.c
+++ b/fs/jbd/journal.c
@@ -1913,7 +1913,7 @@ static void __init jbd_create_debugfs_entry(void)
1913{ 1913{
1914 jbd_debugfs_dir = debugfs_create_dir("jbd", NULL); 1914 jbd_debugfs_dir = debugfs_create_dir("jbd", NULL);
1915 if (jbd_debugfs_dir) 1915 if (jbd_debugfs_dir)
1916 jbd_debug = debugfs_create_u8("jbd-debug", S_IRUGO, 1916 jbd_debug = debugfs_create_u8("jbd-debug", S_IRUGO | S_IWUSR,
1917 jbd_debugfs_dir, 1917 jbd_debugfs_dir,
1918 &journal_enable_debug); 1918 &journal_enable_debug);
1919} 1919}
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
index b7ca3a92a4db..17af879e6e9e 100644
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -2115,7 +2115,8 @@ static void __init jbd2_create_debugfs_entry(void)
2115{ 2115{
2116 jbd2_debugfs_dir = debugfs_create_dir("jbd2", NULL); 2116 jbd2_debugfs_dir = debugfs_create_dir("jbd2", NULL);
2117 if (jbd2_debugfs_dir) 2117 if (jbd2_debugfs_dir)
2118 jbd2_debug = debugfs_create_u8(JBD2_DEBUG_NAME, S_IRUGO, 2118 jbd2_debug = debugfs_create_u8(JBD2_DEBUG_NAME,
2119 S_IRUGO | S_IWUSR,
2119 jbd2_debugfs_dir, 2120 jbd2_debugfs_dir,
2120 &jbd2_journal_enable_debug); 2121 &jbd2_journal_enable_debug);
2121} 2122}
diff --git a/fs/jfs/super.c b/fs/jfs/super.c
index 2234c73fc577..d929a822a74e 100644
--- a/fs/jfs/super.c
+++ b/fs/jfs/super.c
@@ -524,7 +524,7 @@ static int jfs_fill_super(struct super_block *sb, void *data, int silent)
524 * Page cache is indexed by long. 524 * Page cache is indexed by long.
525 * I would use MAX_LFS_FILESIZE, but it's only half as big 525 * I would use MAX_LFS_FILESIZE, but it's only half as big
526 */ 526 */
527 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, sb->s_maxbytes); 527 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, (u64)sb->s_maxbytes);
528#endif 528#endif
529 sb->s_time_gran = 1; 529 sb->s_time_gran = 1;
530 return 0; 530 return 0;
diff --git a/fs/namei.c b/fs/namei.c
index dad4b80257db..68921d9b5302 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -37,8 +37,6 @@
37 37
38#include "internal.h" 38#include "internal.h"
39 39
40#define ACC_MODE(x) ("\000\004\002\006"[(x)&O_ACCMODE])
41
42/* [Feb-1997 T. Schoebel-Theuer] 40/* [Feb-1997 T. Schoebel-Theuer]
43 * Fundamental changes in the pathname lookup mechanisms (namei) 41 * Fundamental changes in the pathname lookup mechanisms (namei)
44 * were necessary because of omirr. The reason is that omirr needs 42 * were necessary because of omirr. The reason is that omirr needs
@@ -1640,6 +1638,7 @@ struct file *do_filp_open(int dfd, const char *pathname,
1640 if (filp == NULL) 1638 if (filp == NULL)
1641 return ERR_PTR(-ENFILE); 1639 return ERR_PTR(-ENFILE);
1642 nd.intent.open.file = filp; 1640 nd.intent.open.file = filp;
1641 filp->f_flags = open_flag;
1643 nd.intent.open.flags = flag; 1642 nd.intent.open.flags = flag;
1644 nd.intent.open.create_mode = 0; 1643 nd.intent.open.create_mode = 0;
1645 error = do_path_lookup(dfd, pathname, 1644 error = do_path_lookup(dfd, pathname,
@@ -1685,6 +1684,7 @@ struct file *do_filp_open(int dfd, const char *pathname,
1685 if (filp == NULL) 1684 if (filp == NULL)
1686 goto exit_parent; 1685 goto exit_parent;
1687 nd.intent.open.file = filp; 1686 nd.intent.open.file = filp;
1687 filp->f_flags = open_flag;
1688 nd.intent.open.flags = flag; 1688 nd.intent.open.flags = flag;
1689 nd.intent.open.create_mode = mode; 1689 nd.intent.open.create_mode = mode;
1690 dir = nd.path.dentry; 1690 dir = nd.path.dentry;
@@ -1725,7 +1725,7 @@ do_last:
1725 mnt_drop_write(nd.path.mnt); 1725 mnt_drop_write(nd.path.mnt);
1726 goto exit; 1726 goto exit;
1727 } 1727 }
1728 filp = nameidata_to_filp(&nd, open_flag); 1728 filp = nameidata_to_filp(&nd);
1729 mnt_drop_write(nd.path.mnt); 1729 mnt_drop_write(nd.path.mnt);
1730 if (nd.root.mnt) 1730 if (nd.root.mnt)
1731 path_put(&nd.root); 1731 path_put(&nd.root);
@@ -1789,7 +1789,7 @@ ok:
1789 mnt_drop_write(nd.path.mnt); 1789 mnt_drop_write(nd.path.mnt);
1790 goto exit; 1790 goto exit;
1791 } 1791 }
1792 filp = nameidata_to_filp(&nd, open_flag); 1792 filp = nameidata_to_filp(&nd);
1793 if (!IS_ERR(filp)) { 1793 if (!IS_ERR(filp)) {
1794 error = ima_path_check(&filp->f_path, filp->f_mode & 1794 error = ima_path_check(&filp->f_path, filp->f_mode &
1795 (MAY_READ | MAY_WRITE | MAY_EXEC)); 1795 (MAY_READ | MAY_WRITE | MAY_EXEC));
diff --git a/fs/nfsd/nfsfh.c b/fs/nfsd/nfsfh.c
index 1c12177b908c..55c8e63af0be 100644
--- a/fs/nfsd/nfsfh.c
+++ b/fs/nfsd/nfsfh.c
@@ -89,7 +89,7 @@ static __be32 nfsd_setuser_and_check_port(struct svc_rqst *rqstp,
89 int flags = nfsexp_flags(rqstp, exp); 89 int flags = nfsexp_flags(rqstp, exp);
90 90
91 /* Check if the request originated from a secure port. */ 91 /* Check if the request originated from a secure port. */
92 if (!rqstp->rq_secure && (flags & NFSEXP_INSECURE_PORT)) { 92 if (!rqstp->rq_secure && !(flags & NFSEXP_INSECURE_PORT)) {
93 RPC_IFDEBUG(char buf[RPC_MAX_ADDRBUFLEN]); 93 RPC_IFDEBUG(char buf[RPC_MAX_ADDRBUFLEN]);
94 dprintk(KERN_WARNING 94 dprintk(KERN_WARNING
95 "nfsd: request from insecure port %s!\n", 95 "nfsd: request from insecure port %s!\n",
diff --git a/fs/ocfs2/Kconfig b/fs/ocfs2/Kconfig
index 701b7a3a872e..0d840669698e 100644
--- a/fs/ocfs2/Kconfig
+++ b/fs/ocfs2/Kconfig
@@ -6,6 +6,7 @@ config OCFS2_FS
6 select CRC32 6 select CRC32
7 select QUOTA 7 select QUOTA
8 select QUOTA_TREE 8 select QUOTA_TREE
9 select FS_POSIX_ACL
9 help 10 help
10 OCFS2 is a general purpose extent based shared disk cluster file 11 OCFS2 is a general purpose extent based shared disk cluster file
11 system with many similarities to ext3. It supports 64 bit inode 12 system with many similarities to ext3. It supports 64 bit inode
@@ -74,12 +75,3 @@ config OCFS2_DEBUG_FS
74 This option will enable expensive consistency checks. Enable 75 This option will enable expensive consistency checks. Enable
75 this option for debugging only as it is likely to decrease 76 this option for debugging only as it is likely to decrease
76 performance of the filesystem. 77 performance of the filesystem.
77
78config OCFS2_FS_POSIX_ACL
79 bool "OCFS2 POSIX Access Control Lists"
80 depends on OCFS2_FS
81 select FS_POSIX_ACL
82 default n
83 help
84 Posix Access Control Lists (ACLs) support permissions for users and
85 groups beyond the owner/group/world scheme.
diff --git a/fs/ocfs2/Makefile b/fs/ocfs2/Makefile
index 31f25ce32c97..600d2d2ade11 100644
--- a/fs/ocfs2/Makefile
+++ b/fs/ocfs2/Makefile
@@ -39,11 +39,8 @@ ocfs2-objs := \
39 ver.o \ 39 ver.o \
40 quota_local.o \ 40 quota_local.o \
41 quota_global.o \ 41 quota_global.o \
42 xattr.o 42 xattr.o \
43 43 acl.o
44ifeq ($(CONFIG_OCFS2_FS_POSIX_ACL),y)
45ocfs2-objs += acl.o
46endif
47 44
48ocfs2_stackglue-objs := stackglue.o 45ocfs2_stackglue-objs := stackglue.o
49ocfs2_stack_o2cb-objs := stack_o2cb.o 46ocfs2_stack_o2cb-objs := stack_o2cb.o
diff --git a/fs/ocfs2/acl.c b/fs/ocfs2/acl.c
index e3e47415d851..0501974bedd0 100644
--- a/fs/ocfs2/acl.c
+++ b/fs/ocfs2/acl.c
@@ -98,15 +98,11 @@ static struct posix_acl *ocfs2_get_acl_nolock(struct inode *inode,
98 int type, 98 int type,
99 struct buffer_head *di_bh) 99 struct buffer_head *di_bh)
100{ 100{
101 struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
102 int name_index; 101 int name_index;
103 char *value = NULL; 102 char *value = NULL;
104 struct posix_acl *acl; 103 struct posix_acl *acl;
105 int retval; 104 int retval;
106 105
107 if (!(osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL))
108 return NULL;
109
110 switch (type) { 106 switch (type) {
111 case ACL_TYPE_ACCESS: 107 case ACL_TYPE_ACCESS:
112 name_index = OCFS2_XATTR_INDEX_POSIX_ACL_ACCESS; 108 name_index = OCFS2_XATTR_INDEX_POSIX_ACL_ACCESS;
diff --git a/fs/ocfs2/acl.h b/fs/ocfs2/acl.h
index 8f6389ed4da5..5c5d31f05853 100644
--- a/fs/ocfs2/acl.h
+++ b/fs/ocfs2/acl.h
@@ -26,8 +26,6 @@ struct ocfs2_acl_entry {
26 __le32 e_id; 26 __le32 e_id;
27}; 27};
28 28
29#ifdef CONFIG_OCFS2_FS_POSIX_ACL
30
31extern int ocfs2_check_acl(struct inode *, int); 29extern int ocfs2_check_acl(struct inode *, int);
32extern int ocfs2_acl_chmod(struct inode *); 30extern int ocfs2_acl_chmod(struct inode *);
33extern int ocfs2_init_acl(handle_t *, struct inode *, struct inode *, 31extern int ocfs2_init_acl(handle_t *, struct inode *, struct inode *,
@@ -35,24 +33,4 @@ extern int ocfs2_init_acl(handle_t *, struct inode *, struct inode *,
35 struct ocfs2_alloc_context *, 33 struct ocfs2_alloc_context *,
36 struct ocfs2_alloc_context *); 34 struct ocfs2_alloc_context *);
37 35
38#else /* CONFIG_OCFS2_FS_POSIX_ACL*/
39
40#define ocfs2_check_acl NULL
41static inline int ocfs2_acl_chmod(struct inode *inode)
42{
43 return 0;
44}
45static inline int ocfs2_init_acl(handle_t *handle,
46 struct inode *inode,
47 struct inode *dir,
48 struct buffer_head *di_bh,
49 struct buffer_head *dir_bh,
50 struct ocfs2_alloc_context *meta_ac,
51 struct ocfs2_alloc_context *data_ac)
52{
53 return 0;
54}
55
56#endif /* CONFIG_OCFS2_FS_POSIX_ACL*/
57
58#endif /* OCFS2_ACL_H */ 36#endif /* OCFS2_ACL_H */
diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c
index fb4e672579b8..d17bdc718f74 100644
--- a/fs/ocfs2/alloc.c
+++ b/fs/ocfs2/alloc.c
@@ -1765,9 +1765,9 @@ set_and_inc:
1765 * 1765 *
1766 * The array index of the subtree root is passed back. 1766 * The array index of the subtree root is passed back.
1767 */ 1767 */
1768static int ocfs2_find_subtree_root(struct ocfs2_extent_tree *et, 1768int ocfs2_find_subtree_root(struct ocfs2_extent_tree *et,
1769 struct ocfs2_path *left, 1769 struct ocfs2_path *left,
1770 struct ocfs2_path *right) 1770 struct ocfs2_path *right)
1771{ 1771{
1772 int i = 0; 1772 int i = 0;
1773 1773
@@ -2872,8 +2872,8 @@ out:
2872 * This looks similar, but is subtly different to 2872 * This looks similar, but is subtly different to
2873 * ocfs2_find_cpos_for_left_leaf(). 2873 * ocfs2_find_cpos_for_left_leaf().
2874 */ 2874 */
2875static int ocfs2_find_cpos_for_right_leaf(struct super_block *sb, 2875int ocfs2_find_cpos_for_right_leaf(struct super_block *sb,
2876 struct ocfs2_path *path, u32 *cpos) 2876 struct ocfs2_path *path, u32 *cpos)
2877{ 2877{
2878 int i, j, ret = 0; 2878 int i, j, ret = 0;
2879 u64 blkno; 2879 u64 blkno;
diff --git a/fs/ocfs2/alloc.h b/fs/ocfs2/alloc.h
index 9c122d574464..1db4359ccb90 100644
--- a/fs/ocfs2/alloc.h
+++ b/fs/ocfs2/alloc.h
@@ -317,4 +317,9 @@ int ocfs2_path_bh_journal_access(handle_t *handle,
317int ocfs2_journal_access_path(struct ocfs2_caching_info *ci, 317int ocfs2_journal_access_path(struct ocfs2_caching_info *ci,
318 handle_t *handle, 318 handle_t *handle,
319 struct ocfs2_path *path); 319 struct ocfs2_path *path);
320int ocfs2_find_cpos_for_right_leaf(struct super_block *sb,
321 struct ocfs2_path *path, u32 *cpos);
322int ocfs2_find_subtree_root(struct ocfs2_extent_tree *et,
323 struct ocfs2_path *left,
324 struct ocfs2_path *right);
320#endif /* OCFS2_ALLOC_H */ 325#endif /* OCFS2_ALLOC_H */
diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c
index c452d116b892..eda5b8bcddd5 100644
--- a/fs/ocfs2/cluster/heartbeat.c
+++ b/fs/ocfs2/cluster/heartbeat.c
@@ -176,7 +176,8 @@ static void o2hb_write_timeout(struct work_struct *work)
176 176
177static void o2hb_arm_write_timeout(struct o2hb_region *reg) 177static void o2hb_arm_write_timeout(struct o2hb_region *reg)
178{ 178{
179 mlog(0, "Queue write timeout for %u ms\n", O2HB_MAX_WRITE_TIMEOUT_MS); 179 mlog(ML_HEARTBEAT, "Queue write timeout for %u ms\n",
180 O2HB_MAX_WRITE_TIMEOUT_MS);
180 181
181 cancel_delayed_work(&reg->hr_write_timeout_work); 182 cancel_delayed_work(&reg->hr_write_timeout_work);
182 reg->hr_last_timeout_start = jiffies; 183 reg->hr_last_timeout_start = jiffies;
@@ -874,7 +875,8 @@ static int o2hb_thread(void *data)
874 do_gettimeofday(&after_hb); 875 do_gettimeofday(&after_hb);
875 elapsed_msec = o2hb_elapsed_msecs(&before_hb, &after_hb); 876 elapsed_msec = o2hb_elapsed_msecs(&before_hb, &after_hb);
876 877
877 mlog(0, "start = %lu.%lu, end = %lu.%lu, msec = %u\n", 878 mlog(ML_HEARTBEAT,
879 "start = %lu.%lu, end = %lu.%lu, msec = %u\n",
878 before_hb.tv_sec, (unsigned long) before_hb.tv_usec, 880 before_hb.tv_sec, (unsigned long) before_hb.tv_usec,
879 after_hb.tv_sec, (unsigned long) after_hb.tv_usec, 881 after_hb.tv_sec, (unsigned long) after_hb.tv_usec,
880 elapsed_msec); 882 elapsed_msec);
diff --git a/fs/ocfs2/cluster/nodemanager.c b/fs/ocfs2/cluster/nodemanager.c
index 7ee6188bc79a..c81142e3ef84 100644
--- a/fs/ocfs2/cluster/nodemanager.c
+++ b/fs/ocfs2/cluster/nodemanager.c
@@ -35,6 +35,10 @@
35 * cluster references throughout where nodes are looked up */ 35 * cluster references throughout where nodes are looked up */
36struct o2nm_cluster *o2nm_single_cluster = NULL; 36struct o2nm_cluster *o2nm_single_cluster = NULL;
37 37
38char *o2nm_fence_method_desc[O2NM_FENCE_METHODS] = {
39 "reset", /* O2NM_FENCE_RESET */
40 "panic", /* O2NM_FENCE_PANIC */
41};
38 42
39struct o2nm_node *o2nm_get_node_by_num(u8 node_num) 43struct o2nm_node *o2nm_get_node_by_num(u8 node_num)
40{ 44{
@@ -579,6 +583,43 @@ static ssize_t o2nm_cluster_attr_reconnect_delay_ms_write(
579 return o2nm_cluster_attr_write(page, count, 583 return o2nm_cluster_attr_write(page, count,
580 &cluster->cl_reconnect_delay_ms); 584 &cluster->cl_reconnect_delay_ms);
581} 585}
586
587static ssize_t o2nm_cluster_attr_fence_method_read(
588 struct o2nm_cluster *cluster, char *page)
589{
590 ssize_t ret = 0;
591
592 if (cluster)
593 ret = sprintf(page, "%s\n",
594 o2nm_fence_method_desc[cluster->cl_fence_method]);
595 return ret;
596}
597
598static ssize_t o2nm_cluster_attr_fence_method_write(
599 struct o2nm_cluster *cluster, const char *page, size_t count)
600{
601 unsigned int i;
602
603 if (page[count - 1] != '\n')
604 goto bail;
605
606 for (i = 0; i < O2NM_FENCE_METHODS; ++i) {
607 if (count != strlen(o2nm_fence_method_desc[i]) + 1)
608 continue;
609 if (strncasecmp(page, o2nm_fence_method_desc[i], count - 1))
610 continue;
611 if (cluster->cl_fence_method != i) {
612 printk(KERN_INFO "ocfs2: Changing fence method to %s\n",
613 o2nm_fence_method_desc[i]);
614 cluster->cl_fence_method = i;
615 }
616 return count;
617 }
618
619bail:
620 return -EINVAL;
621}
622
582static struct o2nm_cluster_attribute o2nm_cluster_attr_idle_timeout_ms = { 623static struct o2nm_cluster_attribute o2nm_cluster_attr_idle_timeout_ms = {
583 .attr = { .ca_owner = THIS_MODULE, 624 .attr = { .ca_owner = THIS_MODULE,
584 .ca_name = "idle_timeout_ms", 625 .ca_name = "idle_timeout_ms",
@@ -603,10 +644,19 @@ static struct o2nm_cluster_attribute o2nm_cluster_attr_reconnect_delay_ms = {
603 .store = o2nm_cluster_attr_reconnect_delay_ms_write, 644 .store = o2nm_cluster_attr_reconnect_delay_ms_write,
604}; 645};
605 646
647static struct o2nm_cluster_attribute o2nm_cluster_attr_fence_method = {
648 .attr = { .ca_owner = THIS_MODULE,
649 .ca_name = "fence_method",
650 .ca_mode = S_IRUGO | S_IWUSR },
651 .show = o2nm_cluster_attr_fence_method_read,
652 .store = o2nm_cluster_attr_fence_method_write,
653};
654
606static struct configfs_attribute *o2nm_cluster_attrs[] = { 655static struct configfs_attribute *o2nm_cluster_attrs[] = {
607 &o2nm_cluster_attr_idle_timeout_ms.attr, 656 &o2nm_cluster_attr_idle_timeout_ms.attr,
608 &o2nm_cluster_attr_keepalive_delay_ms.attr, 657 &o2nm_cluster_attr_keepalive_delay_ms.attr,
609 &o2nm_cluster_attr_reconnect_delay_ms.attr, 658 &o2nm_cluster_attr_reconnect_delay_ms.attr,
659 &o2nm_cluster_attr_fence_method.attr,
610 NULL, 660 NULL,
611}; 661};
612static ssize_t o2nm_cluster_show(struct config_item *item, 662static ssize_t o2nm_cluster_show(struct config_item *item,
@@ -778,6 +828,7 @@ static struct config_group *o2nm_cluster_group_make_group(struct config_group *g
778 cluster->cl_reconnect_delay_ms = O2NET_RECONNECT_DELAY_MS_DEFAULT; 828 cluster->cl_reconnect_delay_ms = O2NET_RECONNECT_DELAY_MS_DEFAULT;
779 cluster->cl_idle_timeout_ms = O2NET_IDLE_TIMEOUT_MS_DEFAULT; 829 cluster->cl_idle_timeout_ms = O2NET_IDLE_TIMEOUT_MS_DEFAULT;
780 cluster->cl_keepalive_delay_ms = O2NET_KEEPALIVE_DELAY_MS_DEFAULT; 830 cluster->cl_keepalive_delay_ms = O2NET_KEEPALIVE_DELAY_MS_DEFAULT;
831 cluster->cl_fence_method = O2NM_FENCE_RESET;
781 832
782 ret = &cluster->cl_group; 833 ret = &cluster->cl_group;
783 o2nm_single_cluster = cluster; 834 o2nm_single_cluster = cluster;
diff --git a/fs/ocfs2/cluster/nodemanager.h b/fs/ocfs2/cluster/nodemanager.h
index c992ea0da4ad..09ea2d388bbb 100644
--- a/fs/ocfs2/cluster/nodemanager.h
+++ b/fs/ocfs2/cluster/nodemanager.h
@@ -33,6 +33,12 @@
33#include <linux/configfs.h> 33#include <linux/configfs.h>
34#include <linux/rbtree.h> 34#include <linux/rbtree.h>
35 35
36enum o2nm_fence_method {
37 O2NM_FENCE_RESET = 0,
38 O2NM_FENCE_PANIC,
39 O2NM_FENCE_METHODS, /* Number of fence methods */
40};
41
36struct o2nm_node { 42struct o2nm_node {
37 spinlock_t nd_lock; 43 spinlock_t nd_lock;
38 struct config_item nd_item; 44 struct config_item nd_item;
@@ -58,6 +64,7 @@ struct o2nm_cluster {
58 unsigned int cl_idle_timeout_ms; 64 unsigned int cl_idle_timeout_ms;
59 unsigned int cl_keepalive_delay_ms; 65 unsigned int cl_keepalive_delay_ms;
60 unsigned int cl_reconnect_delay_ms; 66 unsigned int cl_reconnect_delay_ms;
67 enum o2nm_fence_method cl_fence_method;
61 68
62 /* this bitmap is part of a hack for disk bitmap.. will go eventually. - zab */ 69 /* this bitmap is part of a hack for disk bitmap.. will go eventually. - zab */
63 unsigned long cl_nodes_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; 70 unsigned long cl_nodes_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)];
diff --git a/fs/ocfs2/cluster/quorum.c b/fs/ocfs2/cluster/quorum.c
index bbacf7da48a4..639024033fce 100644
--- a/fs/ocfs2/cluster/quorum.c
+++ b/fs/ocfs2/cluster/quorum.c
@@ -74,8 +74,20 @@ static void o2quo_fence_self(void)
74 * threads can still schedule, etc, etc */ 74 * threads can still schedule, etc, etc */
75 o2hb_stop_all_regions(); 75 o2hb_stop_all_regions();
76 76
77 printk("ocfs2 is very sorry to be fencing this system by restarting\n"); 77 switch (o2nm_single_cluster->cl_fence_method) {
78 emergency_restart(); 78 case O2NM_FENCE_PANIC:
79 panic("*** ocfs2 is very sorry to be fencing this system by "
80 "panicing ***\n");
81 break;
82 default:
83 WARN_ON(o2nm_single_cluster->cl_fence_method >=
84 O2NM_FENCE_METHODS);
85 case O2NM_FENCE_RESET:
86 printk(KERN_ERR "*** ocfs2 is very sorry to be fencing this "
87 "system by restarting ***\n");
88 emergency_restart();
89 break;
90 };
79} 91}
80 92
81/* Indicate that a timeout occured on a hearbeat region write. The 93/* Indicate that a timeout occured on a hearbeat region write. The
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index d9fa3d22e17c..2f9e4e19a4f2 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -2589,6 +2589,14 @@ retry:
2589 "begin reco msg (%d)\n", dlm->name, nodenum, ret); 2589 "begin reco msg (%d)\n", dlm->name, nodenum, ret);
2590 ret = 0; 2590 ret = 0;
2591 } 2591 }
2592 if (ret == -EAGAIN) {
2593 mlog(0, "%s: trying to start recovery of node "
2594 "%u, but node %u is waiting for last recovery "
2595 "to complete, backoff for a bit\n", dlm->name,
2596 dead_node, nodenum);
2597 msleep(100);
2598 goto retry;
2599 }
2592 if (ret < 0) { 2600 if (ret < 0) {
2593 struct dlm_lock_resource *res; 2601 struct dlm_lock_resource *res;
2594 /* this is now a serious problem, possibly ENOMEM 2602 /* this is now a serious problem, possibly ENOMEM
@@ -2608,14 +2616,6 @@ retry:
2608 * another ENOMEM */ 2616 * another ENOMEM */
2609 msleep(100); 2617 msleep(100);
2610 goto retry; 2618 goto retry;
2611 } else if (ret == EAGAIN) {
2612 mlog(0, "%s: trying to start recovery of node "
2613 "%u, but node %u is waiting for last recovery "
2614 "to complete, backoff for a bit\n", dlm->name,
2615 dead_node, nodenum);
2616 /* TODO Look into replacing msleep with cond_resched() */
2617 msleep(100);
2618 goto retry;
2619 } 2619 }
2620 } 2620 }
2621 2621
@@ -2639,7 +2639,7 @@ int dlm_begin_reco_handler(struct o2net_msg *msg, u32 len, void *data,
2639 dlm->name, br->node_idx, br->dead_node, 2639 dlm->name, br->node_idx, br->dead_node,
2640 dlm->reco.dead_node, dlm->reco.new_master); 2640 dlm->reco.dead_node, dlm->reco.new_master);
2641 spin_unlock(&dlm->spinlock); 2641 spin_unlock(&dlm->spinlock);
2642 return EAGAIN; 2642 return -EAGAIN;
2643 } 2643 }
2644 spin_unlock(&dlm->spinlock); 2644 spin_unlock(&dlm->spinlock);
2645 2645
diff --git a/fs/ocfs2/extent_map.c b/fs/ocfs2/extent_map.c
index 843db64e9d4a..d35a27f4523e 100644
--- a/fs/ocfs2/extent_map.c
+++ b/fs/ocfs2/extent_map.c
@@ -37,6 +37,7 @@
37#include "extent_map.h" 37#include "extent_map.h"
38#include "inode.h" 38#include "inode.h"
39#include "super.h" 39#include "super.h"
40#include "symlink.h"
40 41
41#include "buffer_head_io.h" 42#include "buffer_head_io.h"
42 43
@@ -703,6 +704,12 @@ out:
703 return ret; 704 return ret;
704} 705}
705 706
707/*
708 * The ocfs2_fiemap_inline() may be a little bit misleading, since
709 * it not only handles the fiemap for inlined files, but also deals
710 * with the fast symlink, cause they have no difference for extent
711 * mapping per se.
712 */
706static int ocfs2_fiemap_inline(struct inode *inode, struct buffer_head *di_bh, 713static int ocfs2_fiemap_inline(struct inode *inode, struct buffer_head *di_bh,
707 struct fiemap_extent_info *fieinfo, 714 struct fiemap_extent_info *fieinfo,
708 u64 map_start) 715 u64 map_start)
@@ -715,11 +722,18 @@ static int ocfs2_fiemap_inline(struct inode *inode, struct buffer_head *di_bh,
715 struct ocfs2_inode_info *oi = OCFS2_I(inode); 722 struct ocfs2_inode_info *oi = OCFS2_I(inode);
716 723
717 di = (struct ocfs2_dinode *)di_bh->b_data; 724 di = (struct ocfs2_dinode *)di_bh->b_data;
718 id_count = le16_to_cpu(di->id2.i_data.id_count); 725 if (ocfs2_inode_is_fast_symlink(inode))
726 id_count = ocfs2_fast_symlink_chars(inode->i_sb);
727 else
728 id_count = le16_to_cpu(di->id2.i_data.id_count);
719 729
720 if (map_start < id_count) { 730 if (map_start < id_count) {
721 phys = oi->ip_blkno << inode->i_sb->s_blocksize_bits; 731 phys = oi->ip_blkno << inode->i_sb->s_blocksize_bits;
722 phys += offsetof(struct ocfs2_dinode, id2.i_data.id_data); 732 if (ocfs2_inode_is_fast_symlink(inode))
733 phys += offsetof(struct ocfs2_dinode, id2.i_symlink);
734 else
735 phys += offsetof(struct ocfs2_dinode,
736 id2.i_data.id_data);
723 737
724 ret = fiemap_fill_next_extent(fieinfo, 0, phys, id_count, 738 ret = fiemap_fill_next_extent(fieinfo, 0, phys, id_count,
725 flags); 739 flags);
@@ -756,9 +770,10 @@ int ocfs2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
756 down_read(&OCFS2_I(inode)->ip_alloc_sem); 770 down_read(&OCFS2_I(inode)->ip_alloc_sem);
757 771
758 /* 772 /*
759 * Handle inline-data separately. 773 * Handle inline-data and fast symlink separately.
760 */ 774 */
761 if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) { 775 if ((OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) ||
776 ocfs2_inode_is_fast_symlink(inode)) {
762 ret = ocfs2_fiemap_inline(inode, di_bh, fieinfo, map_start); 777 ret = ocfs2_fiemap_inline(inode, di_bh, fieinfo, map_start);
763 goto out_unlock; 778 goto out_unlock;
764 } 779 }
@@ -786,6 +801,8 @@ int ocfs2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
786 fe_flags = 0; 801 fe_flags = 0;
787 if (rec.e_flags & OCFS2_EXT_UNWRITTEN) 802 if (rec.e_flags & OCFS2_EXT_UNWRITTEN)
788 fe_flags |= FIEMAP_EXTENT_UNWRITTEN; 803 fe_flags |= FIEMAP_EXTENT_UNWRITTEN;
804 if (rec.e_flags & OCFS2_EXT_REFCOUNTED)
805 fe_flags |= FIEMAP_EXTENT_SHARED;
789 if (is_last) 806 if (is_last)
790 fe_flags |= FIEMAP_EXTENT_LAST; 807 fe_flags |= FIEMAP_EXTENT_LAST;
791 len_bytes = (u64)le16_to_cpu(rec.e_leaf_clusters) << osb->s_clustersize_bits; 808 len_bytes = (u64)le16_to_cpu(rec.e_leaf_clusters) << osb->s_clustersize_bits;
diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index f010b22b1c44..50fb26a6a5f5 100644
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -2108,6 +2108,7 @@ int ocfs2_create_inode_in_orphan(struct inode *dir,
2108 } 2108 }
2109 did_quota_inode = 1; 2109 did_quota_inode = 1;
2110 2110
2111 inode->i_nlink = 0;
2111 /* do the real work now. */ 2112 /* do the real work now. */
2112 status = ocfs2_mknod_locked(osb, dir, inode, 2113 status = ocfs2_mknod_locked(osb, dir, inode,
2113 0, &new_di_bh, parent_di_bh, handle, 2114 0, &new_di_bh, parent_di_bh, handle,
@@ -2136,6 +2137,7 @@ int ocfs2_create_inode_in_orphan(struct inode *dir,
2136 if (status < 0) 2137 if (status < 0)
2137 mlog_errno(status); 2138 mlog_errno(status);
2138 2139
2140 insert_inode_hash(inode);
2139leave: 2141leave:
2140 if (status < 0 && did_quota_inode) 2142 if (status < 0 && did_quota_inode)
2141 vfs_dq_free_inode(inode); 2143 vfs_dq_free_inode(inode);
@@ -2267,6 +2269,8 @@ int ocfs2_mv_orphaned_inode_to_new(struct inode *dir,
2267 di = (struct ocfs2_dinode *)di_bh->b_data; 2269 di = (struct ocfs2_dinode *)di_bh->b_data;
2268 le32_add_cpu(&di->i_flags, -OCFS2_ORPHANED_FL); 2270 le32_add_cpu(&di->i_flags, -OCFS2_ORPHANED_FL);
2269 di->i_orphaned_slot = 0; 2271 di->i_orphaned_slot = 0;
2272 inode->i_nlink = 1;
2273 ocfs2_set_links_count(di, inode->i_nlink);
2270 ocfs2_journal_dirty(handle, di_bh); 2274 ocfs2_journal_dirty(handle, di_bh);
2271 2275
2272 status = ocfs2_add_entry(handle, dentry, inode, 2276 status = ocfs2_add_entry(handle, dentry, inode,
@@ -2284,7 +2288,6 @@ int ocfs2_mv_orphaned_inode_to_new(struct inode *dir,
2284 goto out_commit; 2288 goto out_commit;
2285 } 2289 }
2286 2290
2287 insert_inode_hash(inode);
2288 dentry->d_op = &ocfs2_dentry_ops; 2291 dentry->d_op = &ocfs2_dentry_ops;
2289 d_instantiate(dentry, inode); 2292 d_instantiate(dentry, inode);
2290 status = 0; 2293 status = 0;
@@ -2326,4 +2329,5 @@ const struct inode_operations ocfs2_dir_iops = {
2326 .getxattr = generic_getxattr, 2329 .getxattr = generic_getxattr,
2327 .listxattr = ocfs2_listxattr, 2330 .listxattr = ocfs2_listxattr,
2328 .removexattr = generic_removexattr, 2331 .removexattr = generic_removexattr,
2332 .fiemap = ocfs2_fiemap,
2329}; 2333};
diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h
index d963d8638709..9362eea7424b 100644
--- a/fs/ocfs2/ocfs2.h
+++ b/fs/ocfs2/ocfs2.h
@@ -245,9 +245,11 @@ enum ocfs2_mount_options
245 OCFS2_MOUNT_LOCALFLOCKS = 1 << 5, /* No cluster aware user file locks */ 245 OCFS2_MOUNT_LOCALFLOCKS = 1 << 5, /* No cluster aware user file locks */
246 OCFS2_MOUNT_NOUSERXATTR = 1 << 6, /* No user xattr */ 246 OCFS2_MOUNT_NOUSERXATTR = 1 << 6, /* No user xattr */
247 OCFS2_MOUNT_INODE64 = 1 << 7, /* Allow inode numbers > 2^32 */ 247 OCFS2_MOUNT_INODE64 = 1 << 7, /* Allow inode numbers > 2^32 */
248 OCFS2_MOUNT_POSIX_ACL = 1 << 8, /* POSIX access control lists */ 248 OCFS2_MOUNT_POSIX_ACL = 1 << 8, /* Force POSIX access control lists */
249 OCFS2_MOUNT_USRQUOTA = 1 << 9, /* We support user quotas */ 249 OCFS2_MOUNT_NO_POSIX_ACL = 1 << 9, /* Disable POSIX access
250 OCFS2_MOUNT_GRPQUOTA = 1 << 10, /* We support group quotas */ 250 control lists */
251 OCFS2_MOUNT_USRQUOTA = 1 << 10, /* We support user quotas */
252 OCFS2_MOUNT_GRPQUOTA = 1 << 11, /* We support group quotas */
251}; 253};
252 254
253#define OCFS2_OSB_SOFT_RO 0x0001 255#define OCFS2_OSB_SOFT_RO 0x0001
diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h
index e9431e4a5e7c..1a1a679e51b5 100644
--- a/fs/ocfs2/ocfs2_fs.h
+++ b/fs/ocfs2/ocfs2_fs.h
@@ -1202,7 +1202,7 @@ struct ocfs2_local_disk_dqinfo {
1202/* Header of one chunk of a quota file */ 1202/* Header of one chunk of a quota file */
1203struct ocfs2_local_disk_chunk { 1203struct ocfs2_local_disk_chunk {
1204 __le32 dqc_free; /* Number of free entries in the bitmap */ 1204 __le32 dqc_free; /* Number of free entries in the bitmap */
1205 u8 dqc_bitmap[0]; /* Bitmap of entries in the corresponding 1205 __u8 dqc_bitmap[0]; /* Bitmap of entries in the corresponding
1206 * chunk of quota file */ 1206 * chunk of quota file */
1207}; 1207};
1208 1208
diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
index 30967e3f5e43..74db2be75dd6 100644
--- a/fs/ocfs2/refcounttree.c
+++ b/fs/ocfs2/refcounttree.c
@@ -276,7 +276,7 @@ static void ocfs2_erase_refcount_tree_from_list(struct ocfs2_super *osb,
276 spin_unlock(&osb->osb_lock); 276 spin_unlock(&osb->osb_lock);
277} 277}
278 278
279void ocfs2_kref_remove_refcount_tree(struct kref *kref) 279static void ocfs2_kref_remove_refcount_tree(struct kref *kref)
280{ 280{
281 struct ocfs2_refcount_tree *tree = 281 struct ocfs2_refcount_tree *tree =
282 container_of(kref, struct ocfs2_refcount_tree, rf_getcnt); 282 container_of(kref, struct ocfs2_refcount_tree, rf_getcnt);
@@ -524,23 +524,6 @@ out:
524 return ret; 524 return ret;
525} 525}
526 526
527int ocfs2_lock_refcount_tree_by_inode(struct inode *inode, int rw,
528 struct ocfs2_refcount_tree **ret_tree,
529 struct buffer_head **ref_bh)
530{
531 int ret;
532 u64 ref_blkno;
533
534 ret = ocfs2_get_refcount_block(inode, &ref_blkno);
535 if (ret) {
536 mlog_errno(ret);
537 return ret;
538 }
539
540 return ocfs2_lock_refcount_tree(OCFS2_SB(inode->i_sb), ref_blkno,
541 rw, ret_tree, ref_bh);
542}
543
544void ocfs2_unlock_refcount_tree(struct ocfs2_super *osb, 527void ocfs2_unlock_refcount_tree(struct ocfs2_super *osb,
545 struct ocfs2_refcount_tree *tree, int rw) 528 struct ocfs2_refcount_tree *tree, int rw)
546{ 529{
@@ -969,6 +952,103 @@ out:
969} 952}
970 953
971/* 954/*
955 * Find the end range for a leaf refcount block indicated by
956 * el->l_recs[index].e_blkno.
957 */
958static int ocfs2_get_refcount_cpos_end(struct ocfs2_caching_info *ci,
959 struct buffer_head *ref_root_bh,
960 struct ocfs2_extent_block *eb,
961 struct ocfs2_extent_list *el,
962 int index, u32 *cpos_end)
963{
964 int ret, i, subtree_root;
965 u32 cpos;
966 u64 blkno;
967 struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
968 struct ocfs2_path *left_path = NULL, *right_path = NULL;
969 struct ocfs2_extent_tree et;
970 struct ocfs2_extent_list *tmp_el;
971
972 if (index < le16_to_cpu(el->l_next_free_rec) - 1) {
973 /*
974 * We have a extent rec after index, so just use the e_cpos
975 * of the next extent rec.
976 */
977 *cpos_end = le32_to_cpu(el->l_recs[index+1].e_cpos);
978 return 0;
979 }
980
981 if (!eb || (eb && !eb->h_next_leaf_blk)) {
982 /*
983 * We are the last extent rec, so any high cpos should
984 * be stored in this leaf refcount block.
985 */
986 *cpos_end = UINT_MAX;
987 return 0;
988 }
989
990 /*
991 * If the extent block isn't the last one, we have to find
992 * the subtree root between this extent block and the next
993 * leaf extent block and get the corresponding e_cpos from
994 * the subroot. Otherwise we may corrupt the b-tree.
995 */
996 ocfs2_init_refcount_extent_tree(&et, ci, ref_root_bh);
997
998 left_path = ocfs2_new_path_from_et(&et);
999 if (!left_path) {
1000 ret = -ENOMEM;
1001 mlog_errno(ret);
1002 goto out;
1003 }
1004
1005 cpos = le32_to_cpu(eb->h_list.l_recs[index].e_cpos);
1006 ret = ocfs2_find_path(ci, left_path, cpos);
1007 if (ret) {
1008 mlog_errno(ret);
1009 goto out;
1010 }
1011
1012 right_path = ocfs2_new_path_from_path(left_path);
1013 if (!right_path) {
1014 ret = -ENOMEM;
1015 mlog_errno(ret);
1016 goto out;
1017 }
1018
1019 ret = ocfs2_find_cpos_for_right_leaf(sb, left_path, &cpos);
1020 if (ret) {
1021 mlog_errno(ret);
1022 goto out;
1023 }
1024
1025 ret = ocfs2_find_path(ci, right_path, cpos);
1026 if (ret) {
1027 mlog_errno(ret);
1028 goto out;
1029 }
1030
1031 subtree_root = ocfs2_find_subtree_root(&et, left_path,
1032 right_path);
1033
1034 tmp_el = left_path->p_node[subtree_root].el;
1035 blkno = left_path->p_node[subtree_root+1].bh->b_blocknr;
1036 for (i = 0; i < le32_to_cpu(tmp_el->l_next_free_rec); i++) {
1037 if (le64_to_cpu(tmp_el->l_recs[i].e_blkno) == blkno) {
1038 *cpos_end = le32_to_cpu(tmp_el->l_recs[i+1].e_cpos);
1039 break;
1040 }
1041 }
1042
1043 BUG_ON(i == le32_to_cpu(tmp_el->l_next_free_rec));
1044
1045out:
1046 ocfs2_free_path(left_path);
1047 ocfs2_free_path(right_path);
1048 return ret;
1049}
1050
1051/*
972 * Given a cpos and len, try to find the refcount record which contains cpos. 1052 * Given a cpos and len, try to find the refcount record which contains cpos.
973 * 1. If cpos can be found in one refcount record, return the record. 1053 * 1. If cpos can be found in one refcount record, return the record.
974 * 2. If cpos can't be found, return a fake record which start from cpos 1054 * 2. If cpos can't be found, return a fake record which start from cpos
@@ -983,10 +1063,10 @@ static int ocfs2_get_refcount_rec(struct ocfs2_caching_info *ci,
983 struct buffer_head **ret_bh) 1063 struct buffer_head **ret_bh)
984{ 1064{
985 int ret = 0, i, found; 1065 int ret = 0, i, found;
986 u32 low_cpos; 1066 u32 low_cpos, uninitialized_var(cpos_end);
987 struct ocfs2_extent_list *el; 1067 struct ocfs2_extent_list *el;
988 struct ocfs2_extent_rec *tmp, *rec = NULL; 1068 struct ocfs2_extent_rec *rec = NULL;
989 struct ocfs2_extent_block *eb; 1069 struct ocfs2_extent_block *eb = NULL;
990 struct buffer_head *eb_bh = NULL, *ref_leaf_bh = NULL; 1070 struct buffer_head *eb_bh = NULL, *ref_leaf_bh = NULL;
991 struct super_block *sb = ocfs2_metadata_cache_get_super(ci); 1071 struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
992 struct ocfs2_refcount_block *rb = 1072 struct ocfs2_refcount_block *rb =
@@ -1034,12 +1114,16 @@ static int ocfs2_get_refcount_rec(struct ocfs2_caching_info *ci,
1034 } 1114 }
1035 } 1115 }
1036 1116
1037 /* adjust len when we have ocfs2_extent_rec after it. */ 1117 if (found) {
1038 if (found && i < le16_to_cpu(el->l_next_free_rec) - 1) { 1118 ret = ocfs2_get_refcount_cpos_end(ci, ref_root_bh,
1039 tmp = &el->l_recs[i+1]; 1119 eb, el, i, &cpos_end);
1120 if (ret) {
1121 mlog_errno(ret);
1122 goto out;
1123 }
1040 1124
1041 if (le32_to_cpu(tmp->e_cpos) < cpos + len) 1125 if (cpos_end < low_cpos + len)
1042 len = le32_to_cpu(tmp->e_cpos) - cpos; 1126 len = cpos_end - low_cpos;
1043 } 1127 }
1044 1128
1045 ret = ocfs2_read_refcount_block(ci, le64_to_cpu(rec->e_blkno), 1129 ret = ocfs2_read_refcount_block(ci, le64_to_cpu(rec->e_blkno),
@@ -1418,7 +1502,7 @@ static int ocfs2_divide_leaf_refcount_block(struct buffer_head *ref_leaf_bh,
1418 1502
1419 /* change old and new rl_used accordingly. */ 1503 /* change old and new rl_used accordingly. */
1420 le16_add_cpu(&rl->rl_used, -num_moved); 1504 le16_add_cpu(&rl->rl_used, -num_moved);
1421 new_rl->rl_used = cpu_to_le32(num_moved); 1505 new_rl->rl_used = cpu_to_le16(num_moved);
1422 1506
1423 sort(&rl->rl_recs, le16_to_cpu(rl->rl_used), 1507 sort(&rl->rl_recs, le16_to_cpu(rl->rl_used),
1424 sizeof(struct ocfs2_refcount_rec), 1508 sizeof(struct ocfs2_refcount_rec),
@@ -1797,7 +1881,8 @@ static int ocfs2_split_refcount_rec(handle_t *handle,
1797 recs_need++; 1881 recs_need++;
1798 1882
1799 /* If the leaf block don't have enough record, expand it. */ 1883 /* If the leaf block don't have enough record, expand it. */
1800 if (le16_to_cpu(rf_list->rl_used) + recs_need > rf_list->rl_count) { 1884 if (le16_to_cpu(rf_list->rl_used) + recs_need >
1885 le16_to_cpu(rf_list->rl_count)) {
1801 struct ocfs2_refcount_rec tmp_rec; 1886 struct ocfs2_refcount_rec tmp_rec;
1802 u64 cpos = le64_to_cpu(orig_rec->r_cpos); 1887 u64 cpos = le64_to_cpu(orig_rec->r_cpos);
1803 len = le32_to_cpu(orig_rec->r_clusters); 1888 len = le32_to_cpu(orig_rec->r_clusters);
@@ -1859,7 +1944,7 @@ static int ocfs2_split_refcount_rec(handle_t *handle,
1859 memcpy(tail_rec, orig_rec, sizeof(struct ocfs2_refcount_rec)); 1944 memcpy(tail_rec, orig_rec, sizeof(struct ocfs2_refcount_rec));
1860 le64_add_cpu(&tail_rec->r_cpos, 1945 le64_add_cpu(&tail_rec->r_cpos,
1861 le32_to_cpu(tail_rec->r_clusters) - len); 1946 le32_to_cpu(tail_rec->r_clusters) - len);
1862 tail_rec->r_clusters = le32_to_cpu(len); 1947 tail_rec->r_clusters = cpu_to_le32(len);
1863 } 1948 }
1864 1949
1865 /* 1950 /*
@@ -3840,8 +3925,7 @@ static int ocfs2_add_refcounted_extent(struct inode *inode,
3840 } 3925 }
3841 3926
3842 ret = ocfs2_insert_extent(handle, et, cpos, 3927 ret = ocfs2_insert_extent(handle, et, cpos,
3843 cpu_to_le64(ocfs2_clusters_to_blocks(inode->i_sb, 3928 ocfs2_clusters_to_blocks(inode->i_sb, p_cluster),
3844 p_cluster)),
3845 num_clusters, ext_flags, meta_ac); 3929 num_clusters, ext_flags, meta_ac);
3846 if (ret) { 3930 if (ret) {
3847 mlog_errno(ret); 3931 mlog_errno(ret);
@@ -4253,8 +4337,8 @@ static int ocfs2_user_path_parent(const char __user *path,
4253 * @new_dentry: target dentry 4337 * @new_dentry: target dentry
4254 * @preserve: if true, preserve all file attributes 4338 * @preserve: if true, preserve all file attributes
4255 */ 4339 */
4256int ocfs2_vfs_reflink(struct dentry *old_dentry, struct inode *dir, 4340static int ocfs2_vfs_reflink(struct dentry *old_dentry, struct inode *dir,
4257 struct dentry *new_dentry, bool preserve) 4341 struct dentry *new_dentry, bool preserve)
4258{ 4342{
4259 struct inode *inode = old_dentry->d_inode; 4343 struct inode *inode = old_dentry->d_inode;
4260 int error; 4344 int error;
diff --git a/fs/ocfs2/stack_user.c b/fs/ocfs2/stack_user.c
index ff4c798a5635..da78a2a334fd 100644
--- a/fs/ocfs2/stack_user.c
+++ b/fs/ocfs2/stack_user.c
@@ -814,7 +814,7 @@ static int fs_protocol_compare(struct ocfs2_protocol_version *existing,
814static int user_cluster_connect(struct ocfs2_cluster_connection *conn) 814static int user_cluster_connect(struct ocfs2_cluster_connection *conn)
815{ 815{
816 dlm_lockspace_t *fsdlm; 816 dlm_lockspace_t *fsdlm;
817 struct ocfs2_live_connection *control; 817 struct ocfs2_live_connection *uninitialized_var(control);
818 int rc = 0; 818 int rc = 0;
819 819
820 BUG_ON(conn == NULL); 820 BUG_ON(conn == NULL);
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 14f47d2bfe02..26069917a9f5 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -100,6 +100,8 @@ struct mount_options
100static int ocfs2_parse_options(struct super_block *sb, char *options, 100static int ocfs2_parse_options(struct super_block *sb, char *options,
101 struct mount_options *mopt, 101 struct mount_options *mopt,
102 int is_remount); 102 int is_remount);
103static int ocfs2_check_set_options(struct super_block *sb,
104 struct mount_options *options);
103static int ocfs2_show_options(struct seq_file *s, struct vfsmount *mnt); 105static int ocfs2_show_options(struct seq_file *s, struct vfsmount *mnt);
104static void ocfs2_put_super(struct super_block *sb); 106static void ocfs2_put_super(struct super_block *sb);
105static int ocfs2_mount_volume(struct super_block *sb); 107static int ocfs2_mount_volume(struct super_block *sb);
@@ -600,7 +602,8 @@ static int ocfs2_remount(struct super_block *sb, int *flags, char *data)
600 602
601 lock_kernel(); 603 lock_kernel();
602 604
603 if (!ocfs2_parse_options(sb, data, &parsed_options, 1)) { 605 if (!ocfs2_parse_options(sb, data, &parsed_options, 1) ||
606 !ocfs2_check_set_options(sb, &parsed_options)) {
604 ret = -EINVAL; 607 ret = -EINVAL;
605 goto out; 608 goto out;
606 } 609 }
@@ -691,8 +694,6 @@ unlock_osb:
691 if (!ret) { 694 if (!ret) {
692 /* Only save off the new mount options in case of a successful 695 /* Only save off the new mount options in case of a successful
693 * remount. */ 696 * remount. */
694 if (!(osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_XATTR))
695 parsed_options.mount_opt &= ~OCFS2_MOUNT_POSIX_ACL;
696 osb->s_mount_opt = parsed_options.mount_opt; 697 osb->s_mount_opt = parsed_options.mount_opt;
697 osb->s_atime_quantum = parsed_options.atime_quantum; 698 osb->s_atime_quantum = parsed_options.atime_quantum;
698 osb->preferred_slot = parsed_options.slot; 699 osb->preferred_slot = parsed_options.slot;
@@ -701,6 +702,10 @@ unlock_osb:
701 702
702 if (!ocfs2_is_hard_readonly(osb)) 703 if (!ocfs2_is_hard_readonly(osb))
703 ocfs2_set_journal_params(osb); 704 ocfs2_set_journal_params(osb);
705
706 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
707 ((osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL) ?
708 MS_POSIXACL : 0);
704 } 709 }
705out: 710out:
706 unlock_kernel(); 711 unlock_kernel();
@@ -1011,31 +1016,16 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent)
1011 brelse(bh); 1016 brelse(bh);
1012 bh = NULL; 1017 bh = NULL;
1013 1018
1014 if (!(osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_XATTR)) 1019 if (!ocfs2_check_set_options(sb, &parsed_options)) {
1015 parsed_options.mount_opt &= ~OCFS2_MOUNT_POSIX_ACL; 1020 status = -EINVAL;
1016 1021 goto read_super_error;
1022 }
1017 osb->s_mount_opt = parsed_options.mount_opt; 1023 osb->s_mount_opt = parsed_options.mount_opt;
1018 osb->s_atime_quantum = parsed_options.atime_quantum; 1024 osb->s_atime_quantum = parsed_options.atime_quantum;
1019 osb->preferred_slot = parsed_options.slot; 1025 osb->preferred_slot = parsed_options.slot;
1020 osb->osb_commit_interval = parsed_options.commit_interval; 1026 osb->osb_commit_interval = parsed_options.commit_interval;
1021 osb->local_alloc_default_bits = ocfs2_megabytes_to_clusters(sb, parsed_options.localalloc_opt); 1027 osb->local_alloc_default_bits = ocfs2_megabytes_to_clusters(sb, parsed_options.localalloc_opt);
1022 osb->local_alloc_bits = osb->local_alloc_default_bits; 1028 osb->local_alloc_bits = osb->local_alloc_default_bits;
1023 if (osb->s_mount_opt & OCFS2_MOUNT_USRQUOTA &&
1024 !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
1025 OCFS2_FEATURE_RO_COMPAT_USRQUOTA)) {
1026 status = -EINVAL;
1027 mlog(ML_ERROR, "User quotas were requested, but this "
1028 "filesystem does not have the feature enabled.\n");
1029 goto read_super_error;
1030 }
1031 if (osb->s_mount_opt & OCFS2_MOUNT_GRPQUOTA &&
1032 !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
1033 OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)) {
1034 status = -EINVAL;
1035 mlog(ML_ERROR, "Group quotas were requested, but this "
1036 "filesystem does not have the feature enabled.\n");
1037 goto read_super_error;
1038 }
1039 1029
1040 status = ocfs2_verify_userspace_stack(osb, &parsed_options); 1030 status = ocfs2_verify_userspace_stack(osb, &parsed_options);
1041 if (status) 1031 if (status)
@@ -1245,6 +1235,40 @@ static struct file_system_type ocfs2_fs_type = {
1245 .next = NULL 1235 .next = NULL
1246}; 1236};
1247 1237
1238static int ocfs2_check_set_options(struct super_block *sb,
1239 struct mount_options *options)
1240{
1241 if (options->mount_opt & OCFS2_MOUNT_USRQUOTA &&
1242 !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
1243 OCFS2_FEATURE_RO_COMPAT_USRQUOTA)) {
1244 mlog(ML_ERROR, "User quotas were requested, but this "
1245 "filesystem does not have the feature enabled.\n");
1246 return 0;
1247 }
1248 if (options->mount_opt & OCFS2_MOUNT_GRPQUOTA &&
1249 !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
1250 OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)) {
1251 mlog(ML_ERROR, "Group quotas were requested, but this "
1252 "filesystem does not have the feature enabled.\n");
1253 return 0;
1254 }
1255 if (options->mount_opt & OCFS2_MOUNT_POSIX_ACL &&
1256 !OCFS2_HAS_INCOMPAT_FEATURE(sb, OCFS2_FEATURE_INCOMPAT_XATTR)) {
1257 mlog(ML_ERROR, "ACL support requested but extended attributes "
1258 "feature is not enabled\n");
1259 return 0;
1260 }
1261 /* No ACL setting specified? Use XATTR feature... */
1262 if (!(options->mount_opt & (OCFS2_MOUNT_POSIX_ACL |
1263 OCFS2_MOUNT_NO_POSIX_ACL))) {
1264 if (OCFS2_HAS_INCOMPAT_FEATURE(sb, OCFS2_FEATURE_INCOMPAT_XATTR))
1265 options->mount_opt |= OCFS2_MOUNT_POSIX_ACL;
1266 else
1267 options->mount_opt |= OCFS2_MOUNT_NO_POSIX_ACL;
1268 }
1269 return 1;
1270}
1271
1248static int ocfs2_parse_options(struct super_block *sb, 1272static int ocfs2_parse_options(struct super_block *sb,
1249 char *options, 1273 char *options,
1250 struct mount_options *mopt, 1274 struct mount_options *mopt,
@@ -1392,40 +1416,19 @@ static int ocfs2_parse_options(struct super_block *sb,
1392 mopt->mount_opt |= OCFS2_MOUNT_INODE64; 1416 mopt->mount_opt |= OCFS2_MOUNT_INODE64;
1393 break; 1417 break;
1394 case Opt_usrquota: 1418 case Opt_usrquota:
1395 /* We check only on remount, otherwise features
1396 * aren't yet initialized. */
1397 if (is_remount && !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
1398 OCFS2_FEATURE_RO_COMPAT_USRQUOTA)) {
1399 mlog(ML_ERROR, "User quota requested but "
1400 "filesystem feature is not set\n");
1401 status = 0;
1402 goto bail;
1403 }
1404 mopt->mount_opt |= OCFS2_MOUNT_USRQUOTA; 1419 mopt->mount_opt |= OCFS2_MOUNT_USRQUOTA;
1405 break; 1420 break;
1406 case Opt_grpquota: 1421 case Opt_grpquota:
1407 if (is_remount && !OCFS2_HAS_RO_COMPAT_FEATURE(sb,
1408 OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)) {
1409 mlog(ML_ERROR, "Group quota requested but "
1410 "filesystem feature is not set\n");
1411 status = 0;
1412 goto bail;
1413 }
1414 mopt->mount_opt |= OCFS2_MOUNT_GRPQUOTA; 1422 mopt->mount_opt |= OCFS2_MOUNT_GRPQUOTA;
1415 break; 1423 break;
1416#ifdef CONFIG_OCFS2_FS_POSIX_ACL
1417 case Opt_acl: 1424 case Opt_acl:
1418 mopt->mount_opt |= OCFS2_MOUNT_POSIX_ACL; 1425 mopt->mount_opt |= OCFS2_MOUNT_POSIX_ACL;
1426 mopt->mount_opt &= ~OCFS2_MOUNT_NO_POSIX_ACL;
1419 break; 1427 break;
1420 case Opt_noacl: 1428 case Opt_noacl:
1429 mopt->mount_opt |= OCFS2_MOUNT_NO_POSIX_ACL;
1421 mopt->mount_opt &= ~OCFS2_MOUNT_POSIX_ACL; 1430 mopt->mount_opt &= ~OCFS2_MOUNT_POSIX_ACL;
1422 break; 1431 break;
1423#else
1424 case Opt_acl:
1425 case Opt_noacl:
1426 printk(KERN_INFO "ocfs2 (no)acl options not supported\n");
1427 break;
1428#endif
1429 default: 1432 default:
1430 mlog(ML_ERROR, 1433 mlog(ML_ERROR,
1431 "Unrecognized mount option \"%s\" " 1434 "Unrecognized mount option \"%s\" "
@@ -1502,12 +1505,10 @@ static int ocfs2_show_options(struct seq_file *s, struct vfsmount *mnt)
1502 if (opts & OCFS2_MOUNT_INODE64) 1505 if (opts & OCFS2_MOUNT_INODE64)
1503 seq_printf(s, ",inode64"); 1506 seq_printf(s, ",inode64");
1504 1507
1505#ifdef CONFIG_OCFS2_FS_POSIX_ACL
1506 if (opts & OCFS2_MOUNT_POSIX_ACL) 1508 if (opts & OCFS2_MOUNT_POSIX_ACL)
1507 seq_printf(s, ",acl"); 1509 seq_printf(s, ",acl");
1508 else 1510 else
1509 seq_printf(s, ",noacl"); 1511 seq_printf(s, ",noacl");
1510#endif
1511 1512
1512 return 0; 1513 return 0;
1513} 1514}
diff --git a/fs/ocfs2/symlink.c b/fs/ocfs2/symlink.c
index e3421030a69f..49b133ccbf11 100644
--- a/fs/ocfs2/symlink.c
+++ b/fs/ocfs2/symlink.c
@@ -163,6 +163,7 @@ const struct inode_operations ocfs2_symlink_inode_operations = {
163 .getxattr = generic_getxattr, 163 .getxattr = generic_getxattr,
164 .listxattr = ocfs2_listxattr, 164 .listxattr = ocfs2_listxattr,
165 .removexattr = generic_removexattr, 165 .removexattr = generic_removexattr,
166 .fiemap = ocfs2_fiemap,
166}; 167};
167const struct inode_operations ocfs2_fast_symlink_inode_operations = { 168const struct inode_operations ocfs2_fast_symlink_inode_operations = {
168 .readlink = ocfs2_readlink, 169 .readlink = ocfs2_readlink,
@@ -174,4 +175,5 @@ const struct inode_operations ocfs2_fast_symlink_inode_operations = {
174 .getxattr = generic_getxattr, 175 .getxattr = generic_getxattr,
175 .listxattr = ocfs2_listxattr, 176 .listxattr = ocfs2_listxattr,
176 .removexattr = generic_removexattr, 177 .removexattr = generic_removexattr,
178 .fiemap = ocfs2_fiemap,
177}; 179};
diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
index 43c114831c0d..8fc6fb071c6d 100644
--- a/fs/ocfs2/xattr.c
+++ b/fs/ocfs2/xattr.c
@@ -98,10 +98,8 @@ static struct ocfs2_xattr_def_value_root def_xv = {
98 98
99struct xattr_handler *ocfs2_xattr_handlers[] = { 99struct xattr_handler *ocfs2_xattr_handlers[] = {
100 &ocfs2_xattr_user_handler, 100 &ocfs2_xattr_user_handler,
101#ifdef CONFIG_OCFS2_FS_POSIX_ACL
102 &ocfs2_xattr_acl_access_handler, 101 &ocfs2_xattr_acl_access_handler,
103 &ocfs2_xattr_acl_default_handler, 102 &ocfs2_xattr_acl_default_handler,
104#endif
105 &ocfs2_xattr_trusted_handler, 103 &ocfs2_xattr_trusted_handler,
106 &ocfs2_xattr_security_handler, 104 &ocfs2_xattr_security_handler,
107 NULL 105 NULL
@@ -109,12 +107,10 @@ struct xattr_handler *ocfs2_xattr_handlers[] = {
109 107
110static struct xattr_handler *ocfs2_xattr_handler_map[OCFS2_XATTR_MAX] = { 108static struct xattr_handler *ocfs2_xattr_handler_map[OCFS2_XATTR_MAX] = {
111 [OCFS2_XATTR_INDEX_USER] = &ocfs2_xattr_user_handler, 109 [OCFS2_XATTR_INDEX_USER] = &ocfs2_xattr_user_handler,
112#ifdef CONFIG_OCFS2_FS_POSIX_ACL
113 [OCFS2_XATTR_INDEX_POSIX_ACL_ACCESS] 110 [OCFS2_XATTR_INDEX_POSIX_ACL_ACCESS]
114 = &ocfs2_xattr_acl_access_handler, 111 = &ocfs2_xattr_acl_access_handler,
115 [OCFS2_XATTR_INDEX_POSIX_ACL_DEFAULT] 112 [OCFS2_XATTR_INDEX_POSIX_ACL_DEFAULT]
116 = &ocfs2_xattr_acl_default_handler, 113 = &ocfs2_xattr_acl_default_handler,
117#endif
118 [OCFS2_XATTR_INDEX_TRUSTED] = &ocfs2_xattr_trusted_handler, 114 [OCFS2_XATTR_INDEX_TRUSTED] = &ocfs2_xattr_trusted_handler,
119 [OCFS2_XATTR_INDEX_SECURITY] = &ocfs2_xattr_security_handler, 115 [OCFS2_XATTR_INDEX_SECURITY] = &ocfs2_xattr_security_handler,
120}; 116};
@@ -6064,7 +6060,7 @@ static int ocfs2_value_metas_in_xattr_header(struct super_block *sb,
6064 * to the extent block, so just calculate a maximum record num. 6060 * to the extent block, so just calculate a maximum record num.
6065 */ 6061 */
6066 if (!xv->xr_list.l_tree_depth) 6062 if (!xv->xr_list.l_tree_depth)
6067 *num_recs += xv->xr_list.l_next_free_rec; 6063 *num_recs += le16_to_cpu(xv->xr_list.l_next_free_rec);
6068 else 6064 else
6069 *num_recs += ocfs2_clusters_for_bytes(sb, 6065 *num_recs += ocfs2_clusters_for_bytes(sb,
6070 XATTR_SIZE_MAX); 6066 XATTR_SIZE_MAX);
diff --git a/fs/ocfs2/xattr.h b/fs/ocfs2/xattr.h
index 08e36389f56d..abd72a47f520 100644
--- a/fs/ocfs2/xattr.h
+++ b/fs/ocfs2/xattr.h
@@ -40,10 +40,8 @@ struct ocfs2_security_xattr_info {
40extern struct xattr_handler ocfs2_xattr_user_handler; 40extern struct xattr_handler ocfs2_xattr_user_handler;
41extern struct xattr_handler ocfs2_xattr_trusted_handler; 41extern struct xattr_handler ocfs2_xattr_trusted_handler;
42extern struct xattr_handler ocfs2_xattr_security_handler; 42extern struct xattr_handler ocfs2_xattr_security_handler;
43#ifdef CONFIG_OCFS2_FS_POSIX_ACL
44extern struct xattr_handler ocfs2_xattr_acl_access_handler; 43extern struct xattr_handler ocfs2_xattr_acl_access_handler;
45extern struct xattr_handler ocfs2_xattr_acl_default_handler; 44extern struct xattr_handler ocfs2_xattr_acl_default_handler;
46#endif
47extern struct xattr_handler *ocfs2_xattr_handlers[]; 45extern struct xattr_handler *ocfs2_xattr_handlers[];
48 46
49ssize_t ocfs2_listxattr(struct dentry *, char *, size_t); 47ssize_t ocfs2_listxattr(struct dentry *, char *, size_t);
diff --git a/fs/open.c b/fs/open.c
index ca69241796bd..040cef72bc00 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -821,15 +821,14 @@ static inline int __get_file_write_access(struct inode *inode,
821} 821}
822 822
823static struct file *__dentry_open(struct dentry *dentry, struct vfsmount *mnt, 823static struct file *__dentry_open(struct dentry *dentry, struct vfsmount *mnt,
824 int flags, struct file *f, 824 struct file *f,
825 int (*open)(struct inode *, struct file *), 825 int (*open)(struct inode *, struct file *),
826 const struct cred *cred) 826 const struct cred *cred)
827{ 827{
828 struct inode *inode; 828 struct inode *inode;
829 int error; 829 int error;
830 830
831 f->f_flags = flags; 831 f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
832 f->f_mode = (__force fmode_t)((flags+1) & O_ACCMODE) | FMODE_LSEEK |
833 FMODE_PREAD | FMODE_PWRITE; 832 FMODE_PREAD | FMODE_PWRITE;
834 inode = dentry->d_inode; 833 inode = dentry->d_inode;
835 if (f->f_mode & FMODE_WRITE) { 834 if (f->f_mode & FMODE_WRITE) {
@@ -930,7 +929,6 @@ struct file *lookup_instantiate_filp(struct nameidata *nd, struct dentry *dentry
930 if (IS_ERR(dentry)) 929 if (IS_ERR(dentry))
931 goto out_err; 930 goto out_err;
932 nd->intent.open.file = __dentry_open(dget(dentry), mntget(nd->path.mnt), 931 nd->intent.open.file = __dentry_open(dget(dentry), mntget(nd->path.mnt),
933 nd->intent.open.flags - 1,
934 nd->intent.open.file, 932 nd->intent.open.file,
935 open, cred); 933 open, cred);
936out: 934out:
@@ -949,7 +947,7 @@ EXPORT_SYMBOL_GPL(lookup_instantiate_filp);
949 * 947 *
950 * Note that this function destroys the original nameidata 948 * Note that this function destroys the original nameidata
951 */ 949 */
952struct file *nameidata_to_filp(struct nameidata *nd, int flags) 950struct file *nameidata_to_filp(struct nameidata *nd)
953{ 951{
954 const struct cred *cred = current_cred(); 952 const struct cred *cred = current_cred();
955 struct file *filp; 953 struct file *filp;
@@ -958,7 +956,7 @@ struct file *nameidata_to_filp(struct nameidata *nd, int flags)
958 filp = nd->intent.open.file; 956 filp = nd->intent.open.file;
959 /* Has the filesystem initialised the file for us? */ 957 /* Has the filesystem initialised the file for us? */
960 if (filp->f_path.dentry == NULL) 958 if (filp->f_path.dentry == NULL)
961 filp = __dentry_open(nd->path.dentry, nd->path.mnt, flags, filp, 959 filp = __dentry_open(nd->path.dentry, nd->path.mnt, filp,
962 NULL, cred); 960 NULL, cred);
963 else 961 else
964 path_put(&nd->path); 962 path_put(&nd->path);
@@ -997,7 +995,8 @@ struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags,
997 return ERR_PTR(error); 995 return ERR_PTR(error);
998 } 996 }
999 997
1000 return __dentry_open(dentry, mnt, flags, f, NULL, cred); 998 f->f_flags = flags;
999 return __dentry_open(dentry, mnt, f, NULL, cred);
1001} 1000}
1002EXPORT_SYMBOL(dentry_open); 1001EXPORT_SYMBOL(dentry_open);
1003 1002
diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c
index cd6bb9a33c13..dea86abdf2e7 100644
--- a/fs/quota/dquot.c
+++ b/fs/quota/dquot.c
@@ -323,6 +323,30 @@ int dquot_mark_dquot_dirty(struct dquot *dquot)
323} 323}
324EXPORT_SYMBOL(dquot_mark_dquot_dirty); 324EXPORT_SYMBOL(dquot_mark_dquot_dirty);
325 325
326/* Dirtify all the dquots - this can block when journalling */
327static inline int mark_all_dquot_dirty(struct dquot * const *dquot)
328{
329 int ret, err, cnt;
330
331 ret = err = 0;
332 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
333 if (dquot[cnt])
334 /* Even in case of error we have to continue */
335 ret = mark_dquot_dirty(dquot[cnt]);
336 if (!err)
337 err = ret;
338 }
339 return err;
340}
341
342static inline void dqput_all(struct dquot **dquot)
343{
344 unsigned int cnt;
345
346 for (cnt = 0; cnt < MAXQUOTAS; cnt++)
347 dqput(dquot[cnt]);
348}
349
326/* This function needs dq_list_lock */ 350/* This function needs dq_list_lock */
327static inline int clear_dquot_dirty(struct dquot *dquot) 351static inline int clear_dquot_dirty(struct dquot *dquot)
328{ 352{
@@ -1268,8 +1292,7 @@ int dquot_initialize(struct inode *inode, int type)
1268out_err: 1292out_err:
1269 up_write(&sb_dqopt(sb)->dqptr_sem); 1293 up_write(&sb_dqopt(sb)->dqptr_sem);
1270 /* Drop unused references */ 1294 /* Drop unused references */
1271 for (cnt = 0; cnt < MAXQUOTAS; cnt++) 1295 dqput_all(got);
1272 dqput(got[cnt]);
1273 return ret; 1296 return ret;
1274} 1297}
1275EXPORT_SYMBOL(dquot_initialize); 1298EXPORT_SYMBOL(dquot_initialize);
@@ -1288,9 +1311,7 @@ int dquot_drop(struct inode *inode)
1288 inode->i_dquot[cnt] = NULL; 1311 inode->i_dquot[cnt] = NULL;
1289 } 1312 }
1290 up_write(&sb_dqopt(inode->i_sb)->dqptr_sem); 1313 up_write(&sb_dqopt(inode->i_sb)->dqptr_sem);
1291 1314 dqput_all(put);
1292 for (cnt = 0; cnt < MAXQUOTAS; cnt++)
1293 dqput(put[cnt]);
1294 return 0; 1315 return 0;
1295} 1316}
1296EXPORT_SYMBOL(dquot_drop); 1317EXPORT_SYMBOL(dquot_drop);
@@ -1319,6 +1340,67 @@ void vfs_dq_drop(struct inode *inode)
1319EXPORT_SYMBOL(vfs_dq_drop); 1340EXPORT_SYMBOL(vfs_dq_drop);
1320 1341
1321/* 1342/*
1343 * inode_reserved_space is managed internally by quota, and protected by
1344 * i_lock similar to i_blocks+i_bytes.
1345 */
1346static qsize_t *inode_reserved_space(struct inode * inode)
1347{
1348 /* Filesystem must explicitly define it's own method in order to use
1349 * quota reservation interface */
1350 BUG_ON(!inode->i_sb->dq_op->get_reserved_space);
1351 return inode->i_sb->dq_op->get_reserved_space(inode);
1352}
1353
1354static void inode_add_rsv_space(struct inode *inode, qsize_t number)
1355{
1356 spin_lock(&inode->i_lock);
1357 *inode_reserved_space(inode) += number;
1358 spin_unlock(&inode->i_lock);
1359}
1360
1361
1362static void inode_claim_rsv_space(struct inode *inode, qsize_t number)
1363{
1364 spin_lock(&inode->i_lock);
1365 *inode_reserved_space(inode) -= number;
1366 __inode_add_bytes(inode, number);
1367 spin_unlock(&inode->i_lock);
1368}
1369
1370static void inode_sub_rsv_space(struct inode *inode, qsize_t number)
1371{
1372 spin_lock(&inode->i_lock);
1373 *inode_reserved_space(inode) -= number;
1374 spin_unlock(&inode->i_lock);
1375}
1376
1377static qsize_t inode_get_rsv_space(struct inode *inode)
1378{
1379 qsize_t ret;
1380 spin_lock(&inode->i_lock);
1381 ret = *inode_reserved_space(inode);
1382 spin_unlock(&inode->i_lock);
1383 return ret;
1384}
1385
1386static void inode_incr_space(struct inode *inode, qsize_t number,
1387 int reserve)
1388{
1389 if (reserve)
1390 inode_add_rsv_space(inode, number);
1391 else
1392 inode_add_bytes(inode, number);
1393}
1394
1395static void inode_decr_space(struct inode *inode, qsize_t number, int reserve)
1396{
1397 if (reserve)
1398 inode_sub_rsv_space(inode, number);
1399 else
1400 inode_sub_bytes(inode, number);
1401}
1402
1403/*
1322 * Following four functions update i_blocks+i_bytes fields and 1404 * Following four functions update i_blocks+i_bytes fields and
1323 * quota information (together with appropriate checks) 1405 * quota information (together with appropriate checks)
1324 * NOTE: We absolutely rely on the fact that caller dirties 1406 * NOTE: We absolutely rely on the fact that caller dirties
@@ -1336,6 +1418,21 @@ int __dquot_alloc_space(struct inode *inode, qsize_t number,
1336 int cnt, ret = QUOTA_OK; 1418 int cnt, ret = QUOTA_OK;
1337 char warntype[MAXQUOTAS]; 1419 char warntype[MAXQUOTAS];
1338 1420
1421 /*
1422 * First test before acquiring mutex - solves deadlocks when we
1423 * re-enter the quota code and are already holding the mutex
1424 */
1425 if (IS_NOQUOTA(inode)) {
1426 inode_incr_space(inode, number, reserve);
1427 goto out;
1428 }
1429
1430 down_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1431 if (IS_NOQUOTA(inode)) {
1432 inode_incr_space(inode, number, reserve);
1433 goto out_unlock;
1434 }
1435
1339 for (cnt = 0; cnt < MAXQUOTAS; cnt++) 1436 for (cnt = 0; cnt < MAXQUOTAS; cnt++)
1340 warntype[cnt] = QUOTA_NL_NOWARN; 1437 warntype[cnt] = QUOTA_NL_NOWARN;
1341 1438
@@ -1346,7 +1443,8 @@ int __dquot_alloc_space(struct inode *inode, qsize_t number,
1346 if (check_bdq(inode->i_dquot[cnt], number, warn, warntype+cnt) 1443 if (check_bdq(inode->i_dquot[cnt], number, warn, warntype+cnt)
1347 == NO_QUOTA) { 1444 == NO_QUOTA) {
1348 ret = NO_QUOTA; 1445 ret = NO_QUOTA;
1349 goto out_unlock; 1446 spin_unlock(&dq_data_lock);
1447 goto out_flush_warn;
1350 } 1448 }
1351 } 1449 }
1352 for (cnt = 0; cnt < MAXQUOTAS; cnt++) { 1450 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
@@ -1357,64 +1455,29 @@ int __dquot_alloc_space(struct inode *inode, qsize_t number,
1357 else 1455 else
1358 dquot_incr_space(inode->i_dquot[cnt], number); 1456 dquot_incr_space(inode->i_dquot[cnt], number);
1359 } 1457 }
1360 if (!reserve) 1458 inode_incr_space(inode, number, reserve);
1361 inode_add_bytes(inode, number);
1362out_unlock:
1363 spin_unlock(&dq_data_lock); 1459 spin_unlock(&dq_data_lock);
1460
1461 if (reserve)
1462 goto out_flush_warn;
1463 mark_all_dquot_dirty(inode->i_dquot);
1464out_flush_warn:
1364 flush_warnings(inode->i_dquot, warntype); 1465 flush_warnings(inode->i_dquot, warntype);
1466out_unlock:
1467 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1468out:
1365 return ret; 1469 return ret;
1366} 1470}
1367 1471
1368int dquot_alloc_space(struct inode *inode, qsize_t number, int warn) 1472int dquot_alloc_space(struct inode *inode, qsize_t number, int warn)
1369{ 1473{
1370 int cnt, ret = QUOTA_OK; 1474 return __dquot_alloc_space(inode, number, warn, 0);
1371
1372 /*
1373 * First test before acquiring mutex - solves deadlocks when we
1374 * re-enter the quota code and are already holding the mutex
1375 */
1376 if (IS_NOQUOTA(inode)) {
1377 inode_add_bytes(inode, number);
1378 goto out;
1379 }
1380
1381 down_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1382 if (IS_NOQUOTA(inode)) {
1383 inode_add_bytes(inode, number);
1384 goto out_unlock;
1385 }
1386
1387 ret = __dquot_alloc_space(inode, number, warn, 0);
1388 if (ret == NO_QUOTA)
1389 goto out_unlock;
1390
1391 /* Dirtify all the dquots - this can block when journalling */
1392 for (cnt = 0; cnt < MAXQUOTAS; cnt++)
1393 if (inode->i_dquot[cnt])
1394 mark_dquot_dirty(inode->i_dquot[cnt]);
1395out_unlock:
1396 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1397out:
1398 return ret;
1399} 1475}
1400EXPORT_SYMBOL(dquot_alloc_space); 1476EXPORT_SYMBOL(dquot_alloc_space);
1401 1477
1402int dquot_reserve_space(struct inode *inode, qsize_t number, int warn) 1478int dquot_reserve_space(struct inode *inode, qsize_t number, int warn)
1403{ 1479{
1404 int ret = QUOTA_OK; 1480 return __dquot_alloc_space(inode, number, warn, 1);
1405
1406 if (IS_NOQUOTA(inode))
1407 goto out;
1408
1409 down_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1410 if (IS_NOQUOTA(inode))
1411 goto out_unlock;
1412
1413 ret = __dquot_alloc_space(inode, number, warn, 1);
1414out_unlock:
1415 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1416out:
1417 return ret;
1418} 1481}
1419EXPORT_SYMBOL(dquot_reserve_space); 1482EXPORT_SYMBOL(dquot_reserve_space);
1420 1483
@@ -1455,10 +1518,7 @@ int dquot_alloc_inode(const struct inode *inode, qsize_t number)
1455warn_put_all: 1518warn_put_all:
1456 spin_unlock(&dq_data_lock); 1519 spin_unlock(&dq_data_lock);
1457 if (ret == QUOTA_OK) 1520 if (ret == QUOTA_OK)
1458 /* Dirtify all the dquots - this can block when journalling */ 1521 mark_all_dquot_dirty(inode->i_dquot);
1459 for (cnt = 0; cnt < MAXQUOTAS; cnt++)
1460 if (inode->i_dquot[cnt])
1461 mark_dquot_dirty(inode->i_dquot[cnt]);
1462 flush_warnings(inode->i_dquot, warntype); 1522 flush_warnings(inode->i_dquot, warntype);
1463 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem); 1523 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1464 return ret; 1524 return ret;
@@ -1471,14 +1531,14 @@ int dquot_claim_space(struct inode *inode, qsize_t number)
1471 int ret = QUOTA_OK; 1531 int ret = QUOTA_OK;
1472 1532
1473 if (IS_NOQUOTA(inode)) { 1533 if (IS_NOQUOTA(inode)) {
1474 inode_add_bytes(inode, number); 1534 inode_claim_rsv_space(inode, number);
1475 goto out; 1535 goto out;
1476 } 1536 }
1477 1537
1478 down_read(&sb_dqopt(inode->i_sb)->dqptr_sem); 1538 down_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1479 if (IS_NOQUOTA(inode)) { 1539 if (IS_NOQUOTA(inode)) {
1480 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem); 1540 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1481 inode_add_bytes(inode, number); 1541 inode_claim_rsv_space(inode, number);
1482 goto out; 1542 goto out;
1483 } 1543 }
1484 1544
@@ -1490,12 +1550,9 @@ int dquot_claim_space(struct inode *inode, qsize_t number)
1490 number); 1550 number);
1491 } 1551 }
1492 /* Update inode bytes */ 1552 /* Update inode bytes */
1493 inode_add_bytes(inode, number); 1553 inode_claim_rsv_space(inode, number);
1494 spin_unlock(&dq_data_lock); 1554 spin_unlock(&dq_data_lock);
1495 /* Dirtify all the dquots - this can block when journalling */ 1555 mark_all_dquot_dirty(inode->i_dquot);
1496 for (cnt = 0; cnt < MAXQUOTAS; cnt++)
1497 if (inode->i_dquot[cnt])
1498 mark_dquot_dirty(inode->i_dquot[cnt]);
1499 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem); 1556 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1500out: 1557out:
1501 return ret; 1558 return ret;
@@ -1503,38 +1560,9 @@ out:
1503EXPORT_SYMBOL(dquot_claim_space); 1560EXPORT_SYMBOL(dquot_claim_space);
1504 1561
1505/* 1562/*
1506 * Release reserved quota space
1507 */
1508void dquot_release_reserved_space(struct inode *inode, qsize_t number)
1509{
1510 int cnt;
1511
1512 if (IS_NOQUOTA(inode))
1513 goto out;
1514
1515 down_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1516 if (IS_NOQUOTA(inode))
1517 goto out_unlock;
1518
1519 spin_lock(&dq_data_lock);
1520 /* Release reserved dquots */
1521 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
1522 if (inode->i_dquot[cnt])
1523 dquot_free_reserved_space(inode->i_dquot[cnt], number);
1524 }
1525 spin_unlock(&dq_data_lock);
1526
1527out_unlock:
1528 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1529out:
1530 return;
1531}
1532EXPORT_SYMBOL(dquot_release_reserved_space);
1533
1534/*
1535 * This operation can block, but only after everything is updated 1563 * This operation can block, but only after everything is updated
1536 */ 1564 */
1537int dquot_free_space(struct inode *inode, qsize_t number) 1565int __dquot_free_space(struct inode *inode, qsize_t number, int reserve)
1538{ 1566{
1539 unsigned int cnt; 1567 unsigned int cnt;
1540 char warntype[MAXQUOTAS]; 1568 char warntype[MAXQUOTAS];
@@ -1543,7 +1571,7 @@ int dquot_free_space(struct inode *inode, qsize_t number)
1543 * re-enter the quota code and are already holding the mutex */ 1571 * re-enter the quota code and are already holding the mutex */
1544 if (IS_NOQUOTA(inode)) { 1572 if (IS_NOQUOTA(inode)) {
1545out_sub: 1573out_sub:
1546 inode_sub_bytes(inode, number); 1574 inode_decr_space(inode, number, reserve);
1547 return QUOTA_OK; 1575 return QUOTA_OK;
1548 } 1576 }
1549 1577
@@ -1558,21 +1586,40 @@ out_sub:
1558 if (!inode->i_dquot[cnt]) 1586 if (!inode->i_dquot[cnt])
1559 continue; 1587 continue;
1560 warntype[cnt] = info_bdq_free(inode->i_dquot[cnt], number); 1588 warntype[cnt] = info_bdq_free(inode->i_dquot[cnt], number);
1561 dquot_decr_space(inode->i_dquot[cnt], number); 1589 if (reserve)
1590 dquot_free_reserved_space(inode->i_dquot[cnt], number);
1591 else
1592 dquot_decr_space(inode->i_dquot[cnt], number);
1562 } 1593 }
1563 inode_sub_bytes(inode, number); 1594 inode_decr_space(inode, number, reserve);
1564 spin_unlock(&dq_data_lock); 1595 spin_unlock(&dq_data_lock);
1565 /* Dirtify all the dquots - this can block when journalling */ 1596
1566 for (cnt = 0; cnt < MAXQUOTAS; cnt++) 1597 if (reserve)
1567 if (inode->i_dquot[cnt]) 1598 goto out_unlock;
1568 mark_dquot_dirty(inode->i_dquot[cnt]); 1599 mark_all_dquot_dirty(inode->i_dquot);
1600out_unlock:
1569 flush_warnings(inode->i_dquot, warntype); 1601 flush_warnings(inode->i_dquot, warntype);
1570 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem); 1602 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1571 return QUOTA_OK; 1603 return QUOTA_OK;
1572} 1604}
1605
1606int dquot_free_space(struct inode *inode, qsize_t number)
1607{
1608 return __dquot_free_space(inode, number, 0);
1609}
1573EXPORT_SYMBOL(dquot_free_space); 1610EXPORT_SYMBOL(dquot_free_space);
1574 1611
1575/* 1612/*
1613 * Release reserved quota space
1614 */
1615void dquot_release_reserved_space(struct inode *inode, qsize_t number)
1616{
1617 __dquot_free_space(inode, number, 1);
1618
1619}
1620EXPORT_SYMBOL(dquot_release_reserved_space);
1621
1622/*
1576 * This operation can block, but only after everything is updated 1623 * This operation can block, but only after everything is updated
1577 */ 1624 */
1578int dquot_free_inode(const struct inode *inode, qsize_t number) 1625int dquot_free_inode(const struct inode *inode, qsize_t number)
@@ -1599,10 +1646,7 @@ int dquot_free_inode(const struct inode *inode, qsize_t number)
1599 dquot_decr_inodes(inode->i_dquot[cnt], number); 1646 dquot_decr_inodes(inode->i_dquot[cnt], number);
1600 } 1647 }
1601 spin_unlock(&dq_data_lock); 1648 spin_unlock(&dq_data_lock);
1602 /* Dirtify all the dquots - this can block when journalling */ 1649 mark_all_dquot_dirty(inode->i_dquot);
1603 for (cnt = 0; cnt < MAXQUOTAS; cnt++)
1604 if (inode->i_dquot[cnt])
1605 mark_dquot_dirty(inode->i_dquot[cnt]);
1606 flush_warnings(inode->i_dquot, warntype); 1650 flush_warnings(inode->i_dquot, warntype);
1607 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem); 1651 up_read(&sb_dqopt(inode->i_sb)->dqptr_sem);
1608 return QUOTA_OK; 1652 return QUOTA_OK;
@@ -1610,19 +1654,6 @@ int dquot_free_inode(const struct inode *inode, qsize_t number)
1610EXPORT_SYMBOL(dquot_free_inode); 1654EXPORT_SYMBOL(dquot_free_inode);
1611 1655
1612/* 1656/*
1613 * call back function, get reserved quota space from underlying fs
1614 */
1615qsize_t dquot_get_reserved_space(struct inode *inode)
1616{
1617 qsize_t reserved_space = 0;
1618
1619 if (sb_any_quota_active(inode->i_sb) &&
1620 inode->i_sb->dq_op->get_reserved_space)
1621 reserved_space = inode->i_sb->dq_op->get_reserved_space(inode);
1622 return reserved_space;
1623}
1624
1625/*
1626 * Transfer the number of inode and blocks from one diskquota to an other. 1657 * Transfer the number of inode and blocks from one diskquota to an other.
1627 * 1658 *
1628 * This operation can block, but only after everything is updated 1659 * This operation can block, but only after everything is updated
@@ -1665,7 +1696,7 @@ int dquot_transfer(struct inode *inode, struct iattr *iattr)
1665 } 1696 }
1666 spin_lock(&dq_data_lock); 1697 spin_lock(&dq_data_lock);
1667 cur_space = inode_get_bytes(inode); 1698 cur_space = inode_get_bytes(inode);
1668 rsv_space = dquot_get_reserved_space(inode); 1699 rsv_space = inode_get_rsv_space(inode);
1669 space = cur_space + rsv_space; 1700 space = cur_space + rsv_space;
1670 /* Build the transfer_from list and check the limits */ 1701 /* Build the transfer_from list and check the limits */
1671 for (cnt = 0; cnt < MAXQUOTAS; cnt++) { 1702 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
@@ -1709,25 +1740,18 @@ int dquot_transfer(struct inode *inode, struct iattr *iattr)
1709 spin_unlock(&dq_data_lock); 1740 spin_unlock(&dq_data_lock);
1710 up_write(&sb_dqopt(inode->i_sb)->dqptr_sem); 1741 up_write(&sb_dqopt(inode->i_sb)->dqptr_sem);
1711 1742
1712 /* Dirtify all the dquots - this can block when journalling */ 1743 mark_all_dquot_dirty(transfer_from);
1713 for (cnt = 0; cnt < MAXQUOTAS; cnt++) { 1744 mark_all_dquot_dirty(transfer_to);
1714 if (transfer_from[cnt]) 1745 /* The reference we got is transferred to the inode */
1715 mark_dquot_dirty(transfer_from[cnt]); 1746 for (cnt = 0; cnt < MAXQUOTAS; cnt++)
1716 if (transfer_to[cnt]) { 1747 transfer_to[cnt] = NULL;
1717 mark_dquot_dirty(transfer_to[cnt]);
1718 /* The reference we got is transferred to the inode */
1719 transfer_to[cnt] = NULL;
1720 }
1721 }
1722warn_put_all: 1748warn_put_all:
1723 flush_warnings(transfer_to, warntype_to); 1749 flush_warnings(transfer_to, warntype_to);
1724 flush_warnings(transfer_from, warntype_from_inodes); 1750 flush_warnings(transfer_from, warntype_from_inodes);
1725 flush_warnings(transfer_from, warntype_from_space); 1751 flush_warnings(transfer_from, warntype_from_space);
1726put_all: 1752put_all:
1727 for (cnt = 0; cnt < MAXQUOTAS; cnt++) { 1753 dqput_all(transfer_from);
1728 dqput(transfer_from[cnt]); 1754 dqput_all(transfer_to);
1729 dqput(transfer_to[cnt]);
1730 }
1731 return ret; 1755 return ret;
1732over_quota: 1756over_quota:
1733 spin_unlock(&dq_data_lock); 1757 spin_unlock(&dq_data_lock);
diff --git a/fs/quota/quota_v2.c b/fs/quota/quota_v2.c
index 3dfc23e02135..e3da02f4986f 100644
--- a/fs/quota/quota_v2.c
+++ b/fs/quota/quota_v2.c
@@ -97,8 +97,11 @@ static int v2_read_file_info(struct super_block *sb, int type)
97 unsigned int version; 97 unsigned int version;
98 98
99 if (!v2_read_header(sb, type, &dqhead)) 99 if (!v2_read_header(sb, type, &dqhead))
100 return 0; 100 return -1;
101 version = le32_to_cpu(dqhead.dqh_version); 101 version = le32_to_cpu(dqhead.dqh_version);
102 if ((info->dqi_fmt_id == QFMT_VFS_V0 && version != 0) ||
103 (info->dqi_fmt_id == QFMT_VFS_V1 && version != 1))
104 return -1;
102 105
103 size = sb->s_op->quota_read(sb, type, (char *)&dinfo, 106 size = sb->s_op->quota_read(sb, type, (char *)&dinfo,
104 sizeof(struct v2_disk_dqinfo), V2_DQINFOOFF); 107 sizeof(struct v2_disk_dqinfo), V2_DQINFOOFF);
@@ -120,8 +123,8 @@ static int v2_read_file_info(struct super_block *sb, int type)
120 info->dqi_maxilimit = 0xffffffff; 123 info->dqi_maxilimit = 0xffffffff;
121 } else { 124 } else {
122 /* used space is stored as unsigned 64-bit value */ 125 /* used space is stored as unsigned 64-bit value */
123 info->dqi_maxblimit = 0xffffffffffffffff; /* 2^64-1 */ 126 info->dqi_maxblimit = 0xffffffffffffffffULL; /* 2^64-1 */
124 info->dqi_maxilimit = 0xffffffffffffffff; 127 info->dqi_maxilimit = 0xffffffffffffffffULL;
125 } 128 }
126 info->dqi_bgrace = le32_to_cpu(dinfo.dqi_bgrace); 129 info->dqi_bgrace = le32_to_cpu(dinfo.dqi_bgrace);
127 info->dqi_igrace = le32_to_cpu(dinfo.dqi_igrace); 130 info->dqi_igrace = le32_to_cpu(dinfo.dqi_igrace);
diff --git a/fs/signalfd.c b/fs/signalfd.c
index b07565c94386..1dabe4ee02fe 100644
--- a/fs/signalfd.c
+++ b/fs/signalfd.c
@@ -236,7 +236,7 @@ SYSCALL_DEFINE4(signalfd4, int, ufd, sigset_t __user *, user_mask,
236 * anon_inode_getfd() will install the fd. 236 * anon_inode_getfd() will install the fd.
237 */ 237 */
238 ufd = anon_inode_getfd("[signalfd]", &signalfd_fops, ctx, 238 ufd = anon_inode_getfd("[signalfd]", &signalfd_fops, ctx,
239 flags & (O_CLOEXEC | O_NONBLOCK)); 239 O_RDWR | (flags & (O_CLOEXEC | O_NONBLOCK)));
240 if (ufd < 0) 240 if (ufd < 0)
241 kfree(ctx); 241 kfree(ctx);
242 } else { 242 } else {
diff --git a/fs/stat.c b/fs/stat.c
index 075694e31d8b..c4ecd52c5737 100644
--- a/fs/stat.c
+++ b/fs/stat.c
@@ -401,9 +401,9 @@ SYSCALL_DEFINE4(fstatat64, int, dfd, char __user *, filename,
401} 401}
402#endif /* __ARCH_WANT_STAT64 */ 402#endif /* __ARCH_WANT_STAT64 */
403 403
404void inode_add_bytes(struct inode *inode, loff_t bytes) 404/* Caller is here responsible for sufficient locking (ie. inode->i_lock) */
405void __inode_add_bytes(struct inode *inode, loff_t bytes)
405{ 406{
406 spin_lock(&inode->i_lock);
407 inode->i_blocks += bytes >> 9; 407 inode->i_blocks += bytes >> 9;
408 bytes &= 511; 408 bytes &= 511;
409 inode->i_bytes += bytes; 409 inode->i_bytes += bytes;
@@ -411,6 +411,12 @@ void inode_add_bytes(struct inode *inode, loff_t bytes)
411 inode->i_blocks++; 411 inode->i_blocks++;
412 inode->i_bytes -= 512; 412 inode->i_bytes -= 512;
413 } 413 }
414}
415
416void inode_add_bytes(struct inode *inode, loff_t bytes)
417{
418 spin_lock(&inode->i_lock);
419 __inode_add_bytes(inode, bytes);
414 spin_unlock(&inode->i_lock); 420 spin_unlock(&inode->i_lock);
415} 421}
416 422
diff --git a/fs/super.c b/fs/super.c
index 19eb70b374bc..aff046b0fe78 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -901,8 +901,9 @@ int get_sb_single(struct file_system_type *fs_type,
901 return error; 901 return error;
902 } 902 }
903 s->s_flags |= MS_ACTIVE; 903 s->s_flags |= MS_ACTIVE;
904 } else {
905 do_remount_sb(s, flags, data, 0);
904 } 906 }
905 do_remount_sb(s, flags, data, 0);
906 simple_set_mnt(mnt, s); 907 simple_set_mnt(mnt, s);
907 return 0; 908 return 0;
908} 909}
diff --git a/fs/sysfs/bin.c b/fs/sysfs/bin.c
index 60c702bc10ae..a0a500af24a1 100644
--- a/fs/sysfs/bin.c
+++ b/fs/sysfs/bin.c
@@ -483,7 +483,8 @@ void unmap_bin_file(struct sysfs_dirent *attr_sd)
483 * @attr: attribute descriptor. 483 * @attr: attribute descriptor.
484 */ 484 */
485 485
486int sysfs_create_bin_file(struct kobject * kobj, struct bin_attribute * attr) 486int sysfs_create_bin_file(struct kobject *kobj,
487 const struct bin_attribute *attr)
487{ 488{
488 BUG_ON(!kobj || !kobj->sd || !attr); 489 BUG_ON(!kobj || !kobj->sd || !attr);
489 490
@@ -497,7 +498,8 @@ int sysfs_create_bin_file(struct kobject * kobj, struct bin_attribute * attr)
497 * @attr: attribute descriptor. 498 * @attr: attribute descriptor.
498 */ 499 */
499 500
500void sysfs_remove_bin_file(struct kobject * kobj, struct bin_attribute * attr) 501void sysfs_remove_bin_file(struct kobject *kobj,
502 const struct bin_attribute *attr)
501{ 503{
502 sysfs_hash_and_remove(kobj->sd, attr->attr.name); 504 sysfs_hash_and_remove(kobj->sd, attr->attr.name);
503} 505}
diff --git a/fs/timerfd.c b/fs/timerfd.c
index b042bd7034b1..1bfc95ad5f71 100644
--- a/fs/timerfd.c
+++ b/fs/timerfd.c
@@ -200,7 +200,7 @@ SYSCALL_DEFINE2(timerfd_create, int, clockid, int, flags)
200 hrtimer_init(&ctx->tmr, clockid, HRTIMER_MODE_ABS); 200 hrtimer_init(&ctx->tmr, clockid, HRTIMER_MODE_ABS);
201 201
202 ufd = anon_inode_getfd("[timerfd]", &timerfd_fops, ctx, 202 ufd = anon_inode_getfd("[timerfd]", &timerfd_fops, ctx,
203 flags & TFD_SHARED_FCNTL_FLAGS); 203 O_RDWR | (flags & TFD_SHARED_FCNTL_FLAGS));
204 if (ufd < 0) 204 if (ufd < 0)
205 kfree(ctx); 205 kfree(ctx);
206 206
generated by cgit v1.2.2 (git 2.25.0) at 2025-10-12 11:17:11 -0400