aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorChris Zankel <chris@zankel.net>2014-04-09 17:49:29 -0400
committerChris Zankel <chris@zankel.net>2014-04-09 17:49:29 -0400
commite86c4b67feb2c984314eef71314f26cfe39fef36 (patch)
tree3c5a7e2216cdcd1d06b8729bf94eba6dc74a7959 /fs
parentb3fdfc1b4b641d372e35ced98814289bc60bc5d1 (diff)
parent9c602629e34bad88a464e08de08118e80beee0d8 (diff)
Merge tag 'xtensa-for-next-20140406' of git://github.com/jcmvbkbc/linux-xtensa into for_next
Merging from Max' tree. Xtensa fixes for 3.15: - add missing cache manipulation symbol exports to fix build failures; - sort extable at build time; - clean up sysmem code, support memmap boot parameter; - add highmem support for cores with non-aliasing caches; - add DTS for xtfpga on KC705. Signed-off-by: Chris Zankel <chris@zankel.net>
Diffstat (limited to 'fs')
-rw-r--r--fs/anon_inodes.c34
-rw-r--r--fs/bio-integrity.c5
-rw-r--r--fs/ceph/acl.c11
-rw-r--r--fs/ceph/dir.c23
-rw-r--r--fs/ceph/file.c1
-rw-r--r--fs/ceph/super.c32
-rw-r--r--fs/ceph/super.h7
-rw-r--r--fs/ceph/xattr.c54
-rw-r--r--fs/cifs/cifsacl.c33
-rw-r--r--fs/cifs/cifsglob.h4
-rw-r--r--fs/cifs/cifsproto.h6
-rw-r--r--fs/cifs/dir.c2
-rw-r--r--fs/cifs/file.c63
-rw-r--r--fs/cifs/inode.c2
-rw-r--r--fs/cifs/smb1ops.c1
-rw-r--r--fs/cifs/smb2glob.h3
-rw-r--r--fs/cifs/smb2ops.c14
-rw-r--r--fs/cifs/smb2pdu.c4
-rw-r--r--fs/cifs/transport.c29
-rw-r--r--fs/dcache.c4
-rw-r--r--fs/ext4/ext4.h2
-rw-r--r--fs/ext4/extents.c1
-rw-r--r--fs/ext4/inode.c15
-rw-r--r--fs/ext4/ioctl.c3
-rw-r--r--fs/ext4/resize.c34
-rw-r--r--fs/ext4/super.c20
-rw-r--r--fs/file.c45
-rw-r--r--fs/file_table.c1
-rw-r--r--fs/fs-writeback.c33
-rw-r--r--fs/fscache/object-list.c5
-rw-r--r--fs/fscache/object.c3
-rw-r--r--fs/hfsplus/catalog.c41
-rw-r--r--fs/hfsplus/hfsplus_fs.h1
-rw-r--r--fs/hfsplus/hfsplus_raw.h6
-rw-r--r--fs/hfsplus/inode.c9
-rw-r--r--fs/hfsplus/options.c2
-rw-r--r--fs/jbd2/transaction.c6
-rw-r--r--fs/jfs/acl.c2
-rw-r--r--fs/kernfs/mount.c8
-rw-r--r--fs/mount.h4
-rw-r--r--fs/namei.c31
-rw-r--r--fs/namespace.c177
-rw-r--r--fs/nfs/delegation.c11
-rw-r--r--fs/nfs/inode.c14
-rw-r--r--fs/nfs/internal.h12
-rw-r--r--fs/nfs/nfs3proc.c1
-rw-r--r--fs/nfs/nfs4client.c7
-rw-r--r--fs/nfs/nfs4filelayout.c10
-rw-r--r--fs/nfs/nfs4namespace.c12
-rw-r--r--fs/nfs/nfs4proc.c24
-rw-r--r--fs/nfs/nfs4state.c19
-rw-r--r--fs/nfsd/vfs.c1
-rw-r--r--fs/notify/dnotify/dnotify.c2
-rw-r--r--fs/notify/fanotify/fanotify.c8
-rw-r--r--fs/notify/fanotify/fanotify_user.c13
-rw-r--r--fs/notify/fsnotify.c2
-rw-r--r--fs/notify/group.c8
-rw-r--r--fs/notify/inotify/inotify.h2
-rw-r--r--fs/notify/inotify/inotify_fsnotify.c3
-rw-r--r--fs/notify/inotify/inotify_user.c14
-rw-r--r--fs/notify/notification.c20
-rw-r--r--fs/ocfs2/file.c8
-rw-r--r--fs/ocfs2/quota_global.c27
-rw-r--r--fs/ocfs2/quota_local.c4
-rw-r--r--fs/ocfs2/stackglue.c4
-rw-r--r--fs/open.c4
-rw-r--r--fs/pnode.c26
-rw-r--r--fs/pnode.h4
-rw-r--r--fs/proc/base.c1
-rw-r--r--fs/proc/page.c5
-rw-r--r--fs/quota/dquot.c14
-rw-r--r--fs/read_write.c44
-rw-r--r--fs/reiserfs/do_balan.c895
-rw-r--r--fs/sync.c15
-rw-r--r--fs/sysfs/mount.c5
-rw-r--r--fs/udf/file.c14
-rw-r--r--fs/udf/inode.c1
-rw-r--r--fs/xfs/xfs_iops.c16
-rw-r--r--fs/xfs/xfs_log_cil.c19
-rw-r--r--fs/xfs/xfs_mount.c24
-rw-r--r--fs/xfs/xfs_sb.c10
-rw-r--r--fs/xfs/xfs_super.c2
82 files changed, 991 insertions, 1115 deletions
diff --git a/fs/anon_inodes.c b/fs/anon_inodes.c
index 24084732b1d0..80ef38c73e5a 100644
--- a/fs/anon_inodes.c
+++ b/fs/anon_inodes.c
@@ -41,19 +41,8 @@ static const struct dentry_operations anon_inodefs_dentry_operations = {
41static struct dentry *anon_inodefs_mount(struct file_system_type *fs_type, 41static struct dentry *anon_inodefs_mount(struct file_system_type *fs_type,
42 int flags, const char *dev_name, void *data) 42 int flags, const char *dev_name, void *data)
43{ 43{
44 struct dentry *root; 44 return mount_pseudo(fs_type, "anon_inode:", NULL,
45 root = mount_pseudo(fs_type, "anon_inode:", NULL,
46 &anon_inodefs_dentry_operations, ANON_INODE_FS_MAGIC); 45 &anon_inodefs_dentry_operations, ANON_INODE_FS_MAGIC);
47 if (!IS_ERR(root)) {
48 struct super_block *s = root->d_sb;
49 anon_inode_inode = alloc_anon_inode(s);
50 if (IS_ERR(anon_inode_inode)) {
51 dput(root);
52 deactivate_locked_super(s);
53 root = ERR_CAST(anon_inode_inode);
54 }
55 }
56 return root;
57} 46}
58 47
59static struct file_system_type anon_inode_fs_type = { 48static struct file_system_type anon_inode_fs_type = {
@@ -175,22 +164,15 @@ EXPORT_SYMBOL_GPL(anon_inode_getfd);
175 164
176static int __init anon_inode_init(void) 165static int __init anon_inode_init(void)
177{ 166{
178 int error;
179
180 error = register_filesystem(&anon_inode_fs_type);
181 if (error)
182 goto err_exit;
183 anon_inode_mnt = kern_mount(&anon_inode_fs_type); 167 anon_inode_mnt = kern_mount(&anon_inode_fs_type);
184 if (IS_ERR(anon_inode_mnt)) { 168 if (IS_ERR(anon_inode_mnt))
185 error = PTR_ERR(anon_inode_mnt); 169 panic("anon_inode_init() kernel mount failed (%ld)\n", PTR_ERR(anon_inode_mnt));
186 goto err_unregister_filesystem;
187 }
188 return 0;
189 170
190err_unregister_filesystem: 171 anon_inode_inode = alloc_anon_inode(anon_inode_mnt->mnt_sb);
191 unregister_filesystem(&anon_inode_fs_type); 172 if (IS_ERR(anon_inode_inode))
192err_exit: 173 panic("anon_inode_init() inode allocation failed (%ld)\n", PTR_ERR(anon_inode_inode));
193 panic(KERN_ERR "anon_inode_init() failed (%d)\n", error); 174
175 return 0;
194} 176}
195 177
196fs_initcall(anon_inode_init); 178fs_initcall(anon_inode_init);
diff --git a/fs/bio-integrity.c b/fs/bio-integrity.c
index 0129b78a6908..4f70f383132c 100644
--- a/fs/bio-integrity.c
+++ b/fs/bio-integrity.c
@@ -458,11 +458,10 @@ static int bio_integrity_verify(struct bio *bio)
458 struct blk_integrity_exchg bix; 458 struct blk_integrity_exchg bix;
459 struct bio_vec *bv; 459 struct bio_vec *bv;
460 sector_t sector = bio->bi_integrity->bip_iter.bi_sector; 460 sector_t sector = bio->bi_integrity->bip_iter.bi_sector;
461 unsigned int sectors, total, ret; 461 unsigned int sectors, ret = 0;
462 void *prot_buf = bio->bi_integrity->bip_buf; 462 void *prot_buf = bio->bi_integrity->bip_buf;
463 int i; 463 int i;
464 464
465 ret = total = 0;
466 bix.disk_name = bio->bi_bdev->bd_disk->disk_name; 465 bix.disk_name = bio->bi_bdev->bd_disk->disk_name;
467 bix.sector_size = bi->sector_size; 466 bix.sector_size = bi->sector_size;
468 467
@@ -484,8 +483,6 @@ static int bio_integrity_verify(struct bio *bio)
484 sectors = bv->bv_len / bi->sector_size; 483 sectors = bv->bv_len / bi->sector_size;
485 sector += sectors; 484 sector += sectors;
486 prot_buf += sectors * bi->tuple_size; 485 prot_buf += sectors * bi->tuple_size;
487 total += sectors * bi->tuple_size;
488 BUG_ON(total > bio->bi_integrity->bip_iter.bi_size);
489 486
490 kunmap_atomic(kaddr); 487 kunmap_atomic(kaddr);
491 } 488 }
diff --git a/fs/ceph/acl.c b/fs/ceph/acl.c
index 4c2d452c4bfc..21887d63dad5 100644
--- a/fs/ceph/acl.c
+++ b/fs/ceph/acl.c
@@ -54,11 +54,6 @@ static inline struct posix_acl *ceph_get_cached_acl(struct inode *inode,
54 return acl; 54 return acl;
55} 55}
56 56
57void ceph_forget_all_cached_acls(struct inode *inode)
58{
59 forget_all_cached_acls(inode);
60}
61
62struct posix_acl *ceph_get_acl(struct inode *inode, int type) 57struct posix_acl *ceph_get_acl(struct inode *inode, int type)
63{ 58{
64 int size; 59 int size;
@@ -160,11 +155,7 @@ int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type)
160 goto out_dput; 155 goto out_dput;
161 } 156 }
162 157
163 if (value) 158 ret = __ceph_setxattr(dentry, name, value, size, 0);
164 ret = __ceph_setxattr(dentry, name, value, size, 0);
165 else
166 ret = __ceph_removexattr(dentry, name);
167
168 if (ret) { 159 if (ret) {
169 if (new_mode != old_mode) { 160 if (new_mode != old_mode) {
170 newattrs.ia_mode = old_mode; 161 newattrs.ia_mode = old_mode;
diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 6da4df84ba30..45eda6d7a40c 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -100,6 +100,14 @@ static unsigned fpos_off(loff_t p)
100 return p & 0xffffffff; 100 return p & 0xffffffff;
101} 101}
102 102
103static int fpos_cmp(loff_t l, loff_t r)
104{
105 int v = ceph_frag_compare(fpos_frag(l), fpos_frag(r));
106 if (v)
107 return v;
108 return (int)(fpos_off(l) - fpos_off(r));
109}
110
103/* 111/*
104 * When possible, we try to satisfy a readdir by peeking at the 112 * When possible, we try to satisfy a readdir by peeking at the
105 * dcache. We make this work by carefully ordering dentries on 113 * dcache. We make this work by carefully ordering dentries on
@@ -156,7 +164,7 @@ more:
156 if (!d_unhashed(dentry) && dentry->d_inode && 164 if (!d_unhashed(dentry) && dentry->d_inode &&
157 ceph_snap(dentry->d_inode) != CEPH_SNAPDIR && 165 ceph_snap(dentry->d_inode) != CEPH_SNAPDIR &&
158 ceph_ino(dentry->d_inode) != CEPH_INO_CEPH && 166 ceph_ino(dentry->d_inode) != CEPH_INO_CEPH &&
159 ctx->pos <= di->offset) 167 fpos_cmp(ctx->pos, di->offset) <= 0)
160 break; 168 break;
161 dout(" skipping %p %.*s at %llu (%llu)%s%s\n", dentry, 169 dout(" skipping %p %.*s at %llu (%llu)%s%s\n", dentry,
162 dentry->d_name.len, dentry->d_name.name, di->offset, 170 dentry->d_name.len, dentry->d_name.name, di->offset,
@@ -695,9 +703,8 @@ static int ceph_mknod(struct inode *dir, struct dentry *dentry,
695 ceph_mdsc_put_request(req); 703 ceph_mdsc_put_request(req);
696 704
697 if (!err) 705 if (!err)
698 err = ceph_init_acl(dentry, dentry->d_inode, dir); 706 ceph_init_acl(dentry, dentry->d_inode, dir);
699 707 else
700 if (err)
701 d_drop(dentry); 708 d_drop(dentry);
702 return err; 709 return err;
703} 710}
@@ -735,7 +742,9 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry,
735 if (!err && !req->r_reply_info.head->is_dentry) 742 if (!err && !req->r_reply_info.head->is_dentry)
736 err = ceph_handle_notrace_create(dir, dentry); 743 err = ceph_handle_notrace_create(dir, dentry);
737 ceph_mdsc_put_request(req); 744 ceph_mdsc_put_request(req);
738 if (err) 745 if (!err)
746 ceph_init_acl(dentry, dentry->d_inode, dir);
747 else
739 d_drop(dentry); 748 d_drop(dentry);
740 return err; 749 return err;
741} 750}
@@ -776,7 +785,9 @@ static int ceph_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
776 err = ceph_handle_notrace_create(dir, dentry); 785 err = ceph_handle_notrace_create(dir, dentry);
777 ceph_mdsc_put_request(req); 786 ceph_mdsc_put_request(req);
778out: 787out:
779 if (err < 0) 788 if (!err)
789 ceph_init_acl(dentry, dentry->d_inode, dir);
790 else
780 d_drop(dentry); 791 d_drop(dentry);
781 return err; 792 return err;
782} 793}
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index dfd2ce3419f8..09c7afe32e49 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -286,6 +286,7 @@ int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
286 } else { 286 } else {
287 dout("atomic_open finish_open on dn %p\n", dn); 287 dout("atomic_open finish_open on dn %p\n", dn);
288 if (req->r_op == CEPH_MDS_OP_CREATE && req->r_reply_info.has_create_ino) { 288 if (req->r_op == CEPH_MDS_OP_CREATE && req->r_reply_info.has_create_ino) {
289 ceph_init_acl(dentry, dentry->d_inode, dir);
289 *opened |= FILE_CREATED; 290 *opened |= FILE_CREATED;
290 } 291 }
291 err = finish_open(file, dentry, ceph_open, opened); 292 err = finish_open(file, dentry, ceph_open, opened);
diff --git a/fs/ceph/super.c b/fs/ceph/super.c
index 2df963f1cf5a..10a4ccbf38da 100644
--- a/fs/ceph/super.c
+++ b/fs/ceph/super.c
@@ -144,7 +144,11 @@ enum {
144 Opt_ino32, 144 Opt_ino32,
145 Opt_noino32, 145 Opt_noino32,
146 Opt_fscache, 146 Opt_fscache,
147 Opt_nofscache 147 Opt_nofscache,
148#ifdef CONFIG_CEPH_FS_POSIX_ACL
149 Opt_acl,
150#endif
151 Opt_noacl
148}; 152};
149 153
150static match_table_t fsopt_tokens = { 154static match_table_t fsopt_tokens = {
@@ -172,6 +176,10 @@ static match_table_t fsopt_tokens = {
172 {Opt_noino32, "noino32"}, 176 {Opt_noino32, "noino32"},
173 {Opt_fscache, "fsc"}, 177 {Opt_fscache, "fsc"},
174 {Opt_nofscache, "nofsc"}, 178 {Opt_nofscache, "nofsc"},
179#ifdef CONFIG_CEPH_FS_POSIX_ACL
180 {Opt_acl, "acl"},
181#endif
182 {Opt_noacl, "noacl"},
175 {-1, NULL} 183 {-1, NULL}
176}; 184};
177 185
@@ -271,6 +279,14 @@ static int parse_fsopt_token(char *c, void *private)
271 case Opt_nofscache: 279 case Opt_nofscache:
272 fsopt->flags &= ~CEPH_MOUNT_OPT_FSCACHE; 280 fsopt->flags &= ~CEPH_MOUNT_OPT_FSCACHE;
273 break; 281 break;
282#ifdef CONFIG_CEPH_FS_POSIX_ACL
283 case Opt_acl:
284 fsopt->sb_flags |= MS_POSIXACL;
285 break;
286#endif
287 case Opt_noacl:
288 fsopt->sb_flags &= ~MS_POSIXACL;
289 break;
274 default: 290 default:
275 BUG_ON(token); 291 BUG_ON(token);
276 } 292 }
@@ -438,6 +454,13 @@ static int ceph_show_options(struct seq_file *m, struct dentry *root)
438 else 454 else
439 seq_puts(m, ",nofsc"); 455 seq_puts(m, ",nofsc");
440 456
457#ifdef CONFIG_CEPH_FS_POSIX_ACL
458 if (fsopt->sb_flags & MS_POSIXACL)
459 seq_puts(m, ",acl");
460 else
461 seq_puts(m, ",noacl");
462#endif
463
441 if (fsopt->wsize) 464 if (fsopt->wsize)
442 seq_printf(m, ",wsize=%d", fsopt->wsize); 465 seq_printf(m, ",wsize=%d", fsopt->wsize);
443 if (fsopt->rsize != CEPH_RSIZE_DEFAULT) 466 if (fsopt->rsize != CEPH_RSIZE_DEFAULT)
@@ -819,9 +842,6 @@ static int ceph_set_super(struct super_block *s, void *data)
819 842
820 s->s_flags = fsc->mount_options->sb_flags; 843 s->s_flags = fsc->mount_options->sb_flags;
821 s->s_maxbytes = 1ULL << 40; /* temp value until we get mdsmap */ 844 s->s_maxbytes = 1ULL << 40; /* temp value until we get mdsmap */
822#ifdef CONFIG_CEPH_FS_POSIX_ACL
823 s->s_flags |= MS_POSIXACL;
824#endif
825 845
826 s->s_xattr = ceph_xattr_handlers; 846 s->s_xattr = ceph_xattr_handlers;
827 s->s_fs_info = fsc; 847 s->s_fs_info = fsc;
@@ -911,6 +931,10 @@ static struct dentry *ceph_mount(struct file_system_type *fs_type,
911 struct ceph_options *opt = NULL; 931 struct ceph_options *opt = NULL;
912 932
913 dout("ceph_mount\n"); 933 dout("ceph_mount\n");
934
935#ifdef CONFIG_CEPH_FS_POSIX_ACL
936 flags |= MS_POSIXACL;
937#endif
914 err = parse_mount_options(&fsopt, &opt, flags, data, dev_name, &path); 938 err = parse_mount_options(&fsopt, &opt, flags, data, dev_name, &path);
915 if (err < 0) { 939 if (err < 0) {
916 res = ERR_PTR(err); 940 res = ERR_PTR(err);
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index 19793b56d0a7..d8801a95b685 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -13,6 +13,7 @@
13#include <linux/wait.h> 13#include <linux/wait.h>
14#include <linux/writeback.h> 14#include <linux/writeback.h>
15#include <linux/slab.h> 15#include <linux/slab.h>
16#include <linux/posix_acl.h>
16 17
17#include <linux/ceph/libceph.h> 18#include <linux/ceph/libceph.h>
18 19
@@ -743,7 +744,11 @@ extern const struct xattr_handler *ceph_xattr_handlers[];
743struct posix_acl *ceph_get_acl(struct inode *, int); 744struct posix_acl *ceph_get_acl(struct inode *, int);
744int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); 745int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type);
745int ceph_init_acl(struct dentry *, struct inode *, struct inode *); 746int ceph_init_acl(struct dentry *, struct inode *, struct inode *);
746void ceph_forget_all_cached_acls(struct inode *inode); 747
748static inline void ceph_forget_all_cached_acls(struct inode *inode)
749{
750 forget_all_cached_acls(inode);
751}
747 752
748#else 753#else
749 754
diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
index 898b6565ad3e..a55ec37378c6 100644
--- a/fs/ceph/xattr.c
+++ b/fs/ceph/xattr.c
@@ -12,6 +12,9 @@
12#define XATTR_CEPH_PREFIX "ceph." 12#define XATTR_CEPH_PREFIX "ceph."
13#define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1) 13#define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1)
14 14
15static int __remove_xattr(struct ceph_inode_info *ci,
16 struct ceph_inode_xattr *xattr);
17
15/* 18/*
16 * List of handlers for synthetic system.* attributes. Other 19 * List of handlers for synthetic system.* attributes. Other
17 * attributes are handled directly. 20 * attributes are handled directly.
@@ -319,8 +322,7 @@ static struct ceph_vxattr *ceph_match_vxattr(struct inode *inode,
319static int __set_xattr(struct ceph_inode_info *ci, 322static int __set_xattr(struct ceph_inode_info *ci,
320 const char *name, int name_len, 323 const char *name, int name_len,
321 const char *val, int val_len, 324 const char *val, int val_len,
322 int dirty, 325 int flags, int update_xattr,
323 int should_free_name, int should_free_val,
324 struct ceph_inode_xattr **newxattr) 326 struct ceph_inode_xattr **newxattr)
325{ 327{
326 struct rb_node **p; 328 struct rb_node **p;
@@ -349,12 +351,31 @@ static int __set_xattr(struct ceph_inode_info *ci,
349 xattr = NULL; 351 xattr = NULL;
350 } 352 }
351 353
354 if (update_xattr) {
355 int err = 0;
356 if (xattr && (flags & XATTR_CREATE))
357 err = -EEXIST;
358 else if (!xattr && (flags & XATTR_REPLACE))
359 err = -ENODATA;
360 if (err) {
361 kfree(name);
362 kfree(val);
363 return err;
364 }
365 if (update_xattr < 0) {
366 if (xattr)
367 __remove_xattr(ci, xattr);
368 kfree(name);
369 return 0;
370 }
371 }
372
352 if (!xattr) { 373 if (!xattr) {
353 new = 1; 374 new = 1;
354 xattr = *newxattr; 375 xattr = *newxattr;
355 xattr->name = name; 376 xattr->name = name;
356 xattr->name_len = name_len; 377 xattr->name_len = name_len;
357 xattr->should_free_name = should_free_name; 378 xattr->should_free_name = update_xattr;
358 379
359 ci->i_xattrs.count++; 380 ci->i_xattrs.count++;
360 dout("__set_xattr count=%d\n", ci->i_xattrs.count); 381 dout("__set_xattr count=%d\n", ci->i_xattrs.count);
@@ -364,7 +385,7 @@ static int __set_xattr(struct ceph_inode_info *ci,
364 if (xattr->should_free_val) 385 if (xattr->should_free_val)
365 kfree((void *)xattr->val); 386 kfree((void *)xattr->val);
366 387
367 if (should_free_name) { 388 if (update_xattr) {
368 kfree((void *)name); 389 kfree((void *)name);
369 name = xattr->name; 390 name = xattr->name;
370 } 391 }
@@ -379,8 +400,8 @@ static int __set_xattr(struct ceph_inode_info *ci,
379 xattr->val = ""; 400 xattr->val = "";
380 401
381 xattr->val_len = val_len; 402 xattr->val_len = val_len;
382 xattr->dirty = dirty; 403 xattr->dirty = update_xattr;
383 xattr->should_free_val = (val && should_free_val); 404 xattr->should_free_val = (val && update_xattr);
384 405
385 if (new) { 406 if (new) {
386 rb_link_node(&xattr->node, parent, p); 407 rb_link_node(&xattr->node, parent, p);
@@ -442,7 +463,7 @@ static int __remove_xattr(struct ceph_inode_info *ci,
442 struct ceph_inode_xattr *xattr) 463 struct ceph_inode_xattr *xattr)
443{ 464{
444 if (!xattr) 465 if (!xattr)
445 return -EOPNOTSUPP; 466 return -ENODATA;
446 467
447 rb_erase(&xattr->node, &ci->i_xattrs.index); 468 rb_erase(&xattr->node, &ci->i_xattrs.index);
448 469
@@ -588,7 +609,7 @@ start:
588 p += len; 609 p += len;
589 610
590 err = __set_xattr(ci, name, namelen, val, len, 611 err = __set_xattr(ci, name, namelen, val, len,
591 0, 0, 0, &xattrs[numattr]); 612 0, 0, &xattrs[numattr]);
592 613
593 if (err < 0) 614 if (err < 0)
594 goto bad; 615 goto bad;
@@ -850,6 +871,9 @@ static int ceph_sync_setxattr(struct dentry *dentry, const char *name,
850 871
851 dout("setxattr value=%.*s\n", (int)size, value); 872 dout("setxattr value=%.*s\n", (int)size, value);
852 873
874 if (!value)
875 flags |= CEPH_XATTR_REMOVE;
876
853 /* do request */ 877 /* do request */
854 req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETXATTR, 878 req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETXATTR,
855 USE_AUTH_MDS); 879 USE_AUTH_MDS);
@@ -892,7 +916,7 @@ int __ceph_setxattr(struct dentry *dentry, const char *name,
892 struct ceph_inode_info *ci = ceph_inode(inode); 916 struct ceph_inode_info *ci = ceph_inode(inode);
893 int issued; 917 int issued;
894 int err; 918 int err;
895 int dirty; 919 int dirty = 0;
896 int name_len = strlen(name); 920 int name_len = strlen(name);
897 int val_len = size; 921 int val_len = size;
898 char *newname = NULL; 922 char *newname = NULL;
@@ -953,12 +977,14 @@ retry:
953 goto retry; 977 goto retry;
954 } 978 }
955 979
956 err = __set_xattr(ci, newname, name_len, newval, 980 err = __set_xattr(ci, newname, name_len, newval, val_len,
957 val_len, 1, 1, 1, &xattr); 981 flags, value ? 1 : -1, &xattr);
958 982
959 dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_XATTR_EXCL); 983 if (!err) {
960 ci->i_xattrs.dirty = true; 984 dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_XATTR_EXCL);
961 inode->i_ctime = CURRENT_TIME; 985 ci->i_xattrs.dirty = true;
986 inode->i_ctime = CURRENT_TIME;
987 }
962 988
963 spin_unlock(&ci->i_ceph_lock); 989 spin_unlock(&ci->i_ceph_lock);
964 if (dirty) 990 if (dirty)
diff --git a/fs/cifs/cifsacl.c b/fs/cifs/cifsacl.c
index c819b0bd491a..7ff866dbb89e 100644
--- a/fs/cifs/cifsacl.c
+++ b/fs/cifs/cifsacl.c
@@ -865,8 +865,8 @@ static int build_sec_desc(struct cifs_ntsd *pntsd, struct cifs_ntsd *pnntsd,
865 return rc; 865 return rc;
866} 866}
867 867
868static struct cifs_ntsd *get_cifs_acl_by_fid(struct cifs_sb_info *cifs_sb, 868struct cifs_ntsd *get_cifs_acl_by_fid(struct cifs_sb_info *cifs_sb,
869 __u16 fid, u32 *pacllen) 869 const struct cifs_fid *cifsfid, u32 *pacllen)
870{ 870{
871 struct cifs_ntsd *pntsd = NULL; 871 struct cifs_ntsd *pntsd = NULL;
872 unsigned int xid; 872 unsigned int xid;
@@ -877,7 +877,8 @@ static struct cifs_ntsd *get_cifs_acl_by_fid(struct cifs_sb_info *cifs_sb,
877 return ERR_CAST(tlink); 877 return ERR_CAST(tlink);
878 878
879 xid = get_xid(); 879 xid = get_xid();
880 rc = CIFSSMBGetCIFSACL(xid, tlink_tcon(tlink), fid, &pntsd, pacllen); 880 rc = CIFSSMBGetCIFSACL(xid, tlink_tcon(tlink), cifsfid->netfid, &pntsd,
881 pacllen);
881 free_xid(xid); 882 free_xid(xid);
882 883
883 cifs_put_tlink(tlink); 884 cifs_put_tlink(tlink);
@@ -946,7 +947,7 @@ struct cifs_ntsd *get_cifs_acl(struct cifs_sb_info *cifs_sb,
946 if (!open_file) 947 if (!open_file)
947 return get_cifs_acl_by_path(cifs_sb, path, pacllen); 948 return get_cifs_acl_by_path(cifs_sb, path, pacllen);
948 949
949 pntsd = get_cifs_acl_by_fid(cifs_sb, open_file->fid.netfid, pacllen); 950 pntsd = get_cifs_acl_by_fid(cifs_sb, &open_file->fid, pacllen);
950 cifsFileInfo_put(open_file); 951 cifsFileInfo_put(open_file);
951 return pntsd; 952 return pntsd;
952} 953}
@@ -1006,19 +1007,31 @@ out:
1006/* Translate the CIFS ACL (simlar to NTFS ACL) for a file into mode bits */ 1007/* Translate the CIFS ACL (simlar to NTFS ACL) for a file into mode bits */
1007int 1008int
1008cifs_acl_to_fattr(struct cifs_sb_info *cifs_sb, struct cifs_fattr *fattr, 1009cifs_acl_to_fattr(struct cifs_sb_info *cifs_sb, struct cifs_fattr *fattr,
1009 struct inode *inode, const char *path, const __u16 *pfid) 1010 struct inode *inode, const char *path,
1011 const struct cifs_fid *pfid)
1010{ 1012{
1011 struct cifs_ntsd *pntsd = NULL; 1013 struct cifs_ntsd *pntsd = NULL;
1012 u32 acllen = 0; 1014 u32 acllen = 0;
1013 int rc = 0; 1015 int rc = 0;
1016 struct tcon_link *tlink = cifs_sb_tlink(cifs_sb);
1017 struct cifs_tcon *tcon;
1014 1018
1015 cifs_dbg(NOISY, "converting ACL to mode for %s\n", path); 1019 cifs_dbg(NOISY, "converting ACL to mode for %s\n", path);
1016 1020
1017 if (pfid) 1021 if (IS_ERR(tlink))
1018 pntsd = get_cifs_acl_by_fid(cifs_sb, *pfid, &acllen); 1022 return PTR_ERR(tlink);
1019 else 1023 tcon = tlink_tcon(tlink);
1020 pntsd = get_cifs_acl(cifs_sb, inode, path, &acllen);
1021 1024
1025 if (pfid && (tcon->ses->server->ops->get_acl_by_fid))
1026 pntsd = tcon->ses->server->ops->get_acl_by_fid(cifs_sb, pfid,
1027 &acllen);
1028 else if (tcon->ses->server->ops->get_acl)
1029 pntsd = tcon->ses->server->ops->get_acl(cifs_sb, inode, path,
1030 &acllen);
1031 else {
1032 cifs_put_tlink(tlink);
1033 return -EOPNOTSUPP;
1034 }
1022 /* if we can retrieve the ACL, now parse Access Control Entries, ACEs */ 1035 /* if we can retrieve the ACL, now parse Access Control Entries, ACEs */
1023 if (IS_ERR(pntsd)) { 1036 if (IS_ERR(pntsd)) {
1024 rc = PTR_ERR(pntsd); 1037 rc = PTR_ERR(pntsd);
@@ -1030,6 +1043,8 @@ cifs_acl_to_fattr(struct cifs_sb_info *cifs_sb, struct cifs_fattr *fattr,
1030 cifs_dbg(VFS, "parse sec desc failed rc = %d\n", rc); 1043 cifs_dbg(VFS, "parse sec desc failed rc = %d\n", rc);
1031 } 1044 }
1032 1045
1046 cifs_put_tlink(tlink);
1047
1033 return rc; 1048 return rc;
1034} 1049}
1035 1050
diff --git a/fs/cifs/cifsglob.h b/fs/cifs/cifsglob.h
index 86dc28c7aa5c..c0f3718b77a8 100644
--- a/fs/cifs/cifsglob.h
+++ b/fs/cifs/cifsglob.h
@@ -398,6 +398,8 @@ struct smb_version_operations {
398 const struct nls_table *, int); 398 const struct nls_table *, int);
399 struct cifs_ntsd * (*get_acl)(struct cifs_sb_info *, struct inode *, 399 struct cifs_ntsd * (*get_acl)(struct cifs_sb_info *, struct inode *,
400 const char *, u32 *); 400 const char *, u32 *);
401 struct cifs_ntsd * (*get_acl_by_fid)(struct cifs_sb_info *,
402 const struct cifs_fid *, u32 *);
401 int (*set_acl)(struct cifs_ntsd *, __u32, struct inode *, const char *, 403 int (*set_acl)(struct cifs_ntsd *, __u32, struct inode *, const char *,
402 int); 404 int);
403}; 405};
@@ -511,7 +513,7 @@ struct cifs_mnt_data {
511static inline unsigned int 513static inline unsigned int
512get_rfc1002_length(void *buf) 514get_rfc1002_length(void *buf)
513{ 515{
514 return be32_to_cpu(*((__be32 *)buf)); 516 return be32_to_cpu(*((__be32 *)buf)) & 0xffffff;
515} 517}
516 518
517static inline void 519static inline void
diff --git a/fs/cifs/cifsproto.h b/fs/cifs/cifsproto.h
index d00e09dfc452..acc4ee8ed075 100644
--- a/fs/cifs/cifsproto.h
+++ b/fs/cifs/cifsproto.h
@@ -151,7 +151,7 @@ extern struct inode *cifs_iget(struct super_block *sb,
151 151
152extern int cifs_get_inode_info(struct inode **inode, const char *full_path, 152extern int cifs_get_inode_info(struct inode **inode, const char *full_path,
153 FILE_ALL_INFO *data, struct super_block *sb, 153 FILE_ALL_INFO *data, struct super_block *sb,
154 int xid, const __u16 *fid); 154 int xid, const struct cifs_fid *fid);
155extern int cifs_get_inode_info_unix(struct inode **pinode, 155extern int cifs_get_inode_info_unix(struct inode **pinode,
156 const unsigned char *search_path, 156 const unsigned char *search_path,
157 struct super_block *sb, unsigned int xid); 157 struct super_block *sb, unsigned int xid);
@@ -162,11 +162,13 @@ extern int cifs_rename_pending_delete(const char *full_path,
162 const unsigned int xid); 162 const unsigned int xid);
163extern int cifs_acl_to_fattr(struct cifs_sb_info *cifs_sb, 163extern int cifs_acl_to_fattr(struct cifs_sb_info *cifs_sb,
164 struct cifs_fattr *fattr, struct inode *inode, 164 struct cifs_fattr *fattr, struct inode *inode,
165 const char *path, const __u16 *pfid); 165 const char *path, const struct cifs_fid *pfid);
166extern int id_mode_to_cifs_acl(struct inode *inode, const char *path, __u64, 166extern int id_mode_to_cifs_acl(struct inode *inode, const char *path, __u64,
167 kuid_t, kgid_t); 167 kuid_t, kgid_t);
168extern struct cifs_ntsd *get_cifs_acl(struct cifs_sb_info *, struct inode *, 168extern struct cifs_ntsd *get_cifs_acl(struct cifs_sb_info *, struct inode *,
169 const char *, u32 *); 169 const char *, u32 *);
170extern struct cifs_ntsd *get_cifs_acl_by_fid(struct cifs_sb_info *,
171 const struct cifs_fid *, u32 *);
170extern int set_cifs_acl(struct cifs_ntsd *, __u32, struct inode *, 172extern int set_cifs_acl(struct cifs_ntsd *, __u32, struct inode *,
171 const char *, int); 173 const char *, int);
172 174
diff --git a/fs/cifs/dir.c b/fs/cifs/dir.c
index d3a6796caa5a..3db0c5fd9a11 100644
--- a/fs/cifs/dir.c
+++ b/fs/cifs/dir.c
@@ -378,7 +378,7 @@ cifs_create_get_file_info:
378 xid); 378 xid);
379 else { 379 else {
380 rc = cifs_get_inode_info(&newinode, full_path, buf, inode->i_sb, 380 rc = cifs_get_inode_info(&newinode, full_path, buf, inode->i_sb,
381 xid, &fid->netfid); 381 xid, fid);
382 if (newinode) { 382 if (newinode) {
383 if (server->ops->set_lease_key) 383 if (server->ops->set_lease_key)
384 server->ops->set_lease_key(newinode, fid); 384 server->ops->set_lease_key(newinode, fid);
diff --git a/fs/cifs/file.c b/fs/cifs/file.c
index 755584684f6c..834fce759d80 100644
--- a/fs/cifs/file.c
+++ b/fs/cifs/file.c
@@ -244,7 +244,7 @@ cifs_nt_open(char *full_path, struct inode *inode, struct cifs_sb_info *cifs_sb,
244 xid); 244 xid);
245 else 245 else
246 rc = cifs_get_inode_info(&inode, full_path, buf, inode->i_sb, 246 rc = cifs_get_inode_info(&inode, full_path, buf, inode->i_sb,
247 xid, &fid->netfid); 247 xid, fid);
248 248
249out: 249out:
250 kfree(buf); 250 kfree(buf);
@@ -2389,7 +2389,7 @@ cifs_iovec_write(struct file *file, const struct iovec *iov,
2389 unsigned long nr_segs, loff_t *poffset) 2389 unsigned long nr_segs, loff_t *poffset)
2390{ 2390{
2391 unsigned long nr_pages, i; 2391 unsigned long nr_pages, i;
2392 size_t copied, len, cur_len; 2392 size_t bytes, copied, len, cur_len;
2393 ssize_t total_written = 0; 2393 ssize_t total_written = 0;
2394 loff_t offset; 2394 loff_t offset;
2395 struct iov_iter it; 2395 struct iov_iter it;
@@ -2444,14 +2444,45 @@ cifs_iovec_write(struct file *file, const struct iovec *iov,
2444 2444
2445 save_len = cur_len; 2445 save_len = cur_len;
2446 for (i = 0; i < nr_pages; i++) { 2446 for (i = 0; i < nr_pages; i++) {
2447 copied = min_t(const size_t, cur_len, PAGE_SIZE); 2447 bytes = min_t(const size_t, cur_len, PAGE_SIZE);
2448 copied = iov_iter_copy_from_user(wdata->pages[i], &it, 2448 copied = iov_iter_copy_from_user(wdata->pages[i], &it,
2449 0, copied); 2449 0, bytes);
2450 cur_len -= copied; 2450 cur_len -= copied;
2451 iov_iter_advance(&it, copied); 2451 iov_iter_advance(&it, copied);
2452 /*
2453 * If we didn't copy as much as we expected, then that
2454 * may mean we trod into an unmapped area. Stop copying
2455 * at that point. On the next pass through the big
2456 * loop, we'll likely end up getting a zero-length
2457 * write and bailing out of it.
2458 */
2459 if (copied < bytes)
2460 break;
2452 } 2461 }
2453 cur_len = save_len - cur_len; 2462 cur_len = save_len - cur_len;
2454 2463
2464 /*
2465 * If we have no data to send, then that probably means that
2466 * the copy above failed altogether. That's most likely because
2467 * the address in the iovec was bogus. Set the rc to -EFAULT,
2468 * free anything we allocated and bail out.
2469 */
2470 if (!cur_len) {
2471 for (i = 0; i < nr_pages; i++)
2472 put_page(wdata->pages[i]);
2473 kfree(wdata);
2474 rc = -EFAULT;
2475 break;
2476 }
2477
2478 /*
2479 * i + 1 now represents the number of pages we actually used in
2480 * the copy phase above. Bring nr_pages down to that, and free
2481 * any pages that we didn't use.
2482 */
2483 for ( ; nr_pages > i + 1; nr_pages--)
2484 put_page(wdata->pages[nr_pages - 1]);
2485
2455 wdata->sync_mode = WB_SYNC_ALL; 2486 wdata->sync_mode = WB_SYNC_ALL;
2456 wdata->nr_pages = nr_pages; 2487 wdata->nr_pages = nr_pages;
2457 wdata->offset = (__u64)offset; 2488 wdata->offset = (__u64)offset;
@@ -2548,31 +2579,19 @@ cifs_writev(struct kiocb *iocb, const struct iovec *iov,
2548 struct cifsInodeInfo *cinode = CIFS_I(inode); 2579 struct cifsInodeInfo *cinode = CIFS_I(inode);
2549 struct TCP_Server_Info *server = tlink_tcon(cfile->tlink)->ses->server; 2580 struct TCP_Server_Info *server = tlink_tcon(cfile->tlink)->ses->server;
2550 ssize_t rc = -EACCES; 2581 ssize_t rc = -EACCES;
2582 loff_t lock_pos = pos;
2551 2583
2552 BUG_ON(iocb->ki_pos != pos); 2584 if (file->f_flags & O_APPEND)
2553 2585 lock_pos = i_size_read(inode);
2554 /* 2586 /*
2555 * We need to hold the sem to be sure nobody modifies lock list 2587 * We need to hold the sem to be sure nobody modifies lock list
2556 * with a brlock that prevents writing. 2588 * with a brlock that prevents writing.
2557 */ 2589 */
2558 down_read(&cinode->lock_sem); 2590 down_read(&cinode->lock_sem);
2559 if (!cifs_find_lock_conflict(cfile, pos, iov_length(iov, nr_segs), 2591 if (!cifs_find_lock_conflict(cfile, lock_pos, iov_length(iov, nr_segs),
2560 server->vals->exclusive_lock_type, NULL, 2592 server->vals->exclusive_lock_type, NULL,
2561 CIFS_WRITE_OP)) { 2593 CIFS_WRITE_OP))
2562 mutex_lock(&inode->i_mutex); 2594 rc = generic_file_aio_write(iocb, iov, nr_segs, pos);
2563 rc = __generic_file_aio_write(iocb, iov, nr_segs,
2564 &iocb->ki_pos);
2565 mutex_unlock(&inode->i_mutex);
2566 }
2567
2568 if (rc > 0) {
2569 ssize_t err;
2570
2571 err = generic_write_sync(file, iocb->ki_pos - rc, rc);
2572 if (err < 0)
2573 rc = err;
2574 }
2575
2576 up_read(&cinode->lock_sem); 2595 up_read(&cinode->lock_sem);
2577 return rc; 2596 return rc;
2578} 2597}
diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
index be58b8fcdb3c..aadc2b68678b 100644
--- a/fs/cifs/inode.c
+++ b/fs/cifs/inode.c
@@ -677,7 +677,7 @@ cgfi_exit:
677int 677int
678cifs_get_inode_info(struct inode **inode, const char *full_path, 678cifs_get_inode_info(struct inode **inode, const char *full_path,
679 FILE_ALL_INFO *data, struct super_block *sb, int xid, 679 FILE_ALL_INFO *data, struct super_block *sb, int xid,
680 const __u16 *fid) 680 const struct cifs_fid *fid)
681{ 681{
682 bool validinum = false; 682 bool validinum = false;
683 __u16 srchflgs; 683 __u16 srchflgs;
diff --git a/fs/cifs/smb1ops.c b/fs/cifs/smb1ops.c
index bfd66d84831e..526fb89f9230 100644
--- a/fs/cifs/smb1ops.c
+++ b/fs/cifs/smb1ops.c
@@ -1073,6 +1073,7 @@ struct smb_version_operations smb1_operations = {
1073#endif /* CIFS_XATTR */ 1073#endif /* CIFS_XATTR */
1074#ifdef CONFIG_CIFS_ACL 1074#ifdef CONFIG_CIFS_ACL
1075 .get_acl = get_cifs_acl, 1075 .get_acl = get_cifs_acl,
1076 .get_acl_by_fid = get_cifs_acl_by_fid,
1076 .set_acl = set_cifs_acl, 1077 .set_acl = set_cifs_acl,
1077#endif /* CIFS_ACL */ 1078#endif /* CIFS_ACL */
1078}; 1079};
diff --git a/fs/cifs/smb2glob.h b/fs/cifs/smb2glob.h
index c38350851b08..bc0bb9c34f72 100644
--- a/fs/cifs/smb2glob.h
+++ b/fs/cifs/smb2glob.h
@@ -57,4 +57,7 @@
57#define SMB2_CMACAES_SIZE (16) 57#define SMB2_CMACAES_SIZE (16)
58#define SMB3_SIGNKEY_SIZE (16) 58#define SMB3_SIGNKEY_SIZE (16)
59 59
60/* Maximum buffer size value we can send with 1 credit */
61#define SMB2_MAX_BUFFER_SIZE 65536
62
60#endif /* _SMB2_GLOB_H */ 63#endif /* _SMB2_GLOB_H */
diff --git a/fs/cifs/smb2ops.c b/fs/cifs/smb2ops.c
index 757da3e54d3d..192f51a12cf1 100644
--- a/fs/cifs/smb2ops.c
+++ b/fs/cifs/smb2ops.c
@@ -182,11 +182,8 @@ smb2_negotiate_wsize(struct cifs_tcon *tcon, struct smb_vol *volume_info)
182 /* start with specified wsize, or default */ 182 /* start with specified wsize, or default */
183 wsize = volume_info->wsize ? volume_info->wsize : CIFS_DEFAULT_IOSIZE; 183 wsize = volume_info->wsize ? volume_info->wsize : CIFS_DEFAULT_IOSIZE;
184 wsize = min_t(unsigned int, wsize, server->max_write); 184 wsize = min_t(unsigned int, wsize, server->max_write);
185 /* 185 /* set it to the maximum buffer size value we can send with 1 credit */
186 * limit write size to 2 ** 16, because we don't support multicredit 186 wsize = min_t(unsigned int, wsize, SMB2_MAX_BUFFER_SIZE);
187 * requests now.
188 */
189 wsize = min_t(unsigned int, wsize, 2 << 15);
190 187
191 return wsize; 188 return wsize;
192} 189}
@@ -200,11 +197,8 @@ smb2_negotiate_rsize(struct cifs_tcon *tcon, struct smb_vol *volume_info)
200 /* start with specified rsize, or default */ 197 /* start with specified rsize, or default */
201 rsize = volume_info->rsize ? volume_info->rsize : CIFS_DEFAULT_IOSIZE; 198 rsize = volume_info->rsize ? volume_info->rsize : CIFS_DEFAULT_IOSIZE;
202 rsize = min_t(unsigned int, rsize, server->max_read); 199 rsize = min_t(unsigned int, rsize, server->max_read);
203 /* 200 /* set it to the maximum buffer size value we can send with 1 credit */
204 * limit write size to 2 ** 16, because we don't support multicredit 201 rsize = min_t(unsigned int, rsize, SMB2_MAX_BUFFER_SIZE);
205 * requests now.
206 */
207 rsize = min_t(unsigned int, rsize, 2 << 15);
208 202
209 return rsize; 203 return rsize;
210} 204}
diff --git a/fs/cifs/smb2pdu.c b/fs/cifs/smb2pdu.c
index a3f7a9c3cc69..860344701067 100644
--- a/fs/cifs/smb2pdu.c
+++ b/fs/cifs/smb2pdu.c
@@ -413,7 +413,9 @@ SMB2_negotiate(const unsigned int xid, struct cifs_ses *ses)
413 413
414 /* SMB2 only has an extended negflavor */ 414 /* SMB2 only has an extended negflavor */
415 server->negflavor = CIFS_NEGFLAVOR_EXTENDED; 415 server->negflavor = CIFS_NEGFLAVOR_EXTENDED;
416 server->maxBuf = le32_to_cpu(rsp->MaxTransactSize); 416 /* set it to the maximum buffer size value we can send with 1 credit */
417 server->maxBuf = min_t(unsigned int, le32_to_cpu(rsp->MaxTransactSize),
418 SMB2_MAX_BUFFER_SIZE);
417 server->max_read = le32_to_cpu(rsp->MaxReadSize); 419 server->max_read = le32_to_cpu(rsp->MaxReadSize);
418 server->max_write = le32_to_cpu(rsp->MaxWriteSize); 420 server->max_write = le32_to_cpu(rsp->MaxWriteSize);
419 /* BB Do we need to validate the SecurityMode? */ 421 /* BB Do we need to validate the SecurityMode? */
diff --git a/fs/cifs/transport.c b/fs/cifs/transport.c
index b37570952846..18cd5650a5fc 100644
--- a/fs/cifs/transport.c
+++ b/fs/cifs/transport.c
@@ -270,6 +270,26 @@ cifs_rqst_page_to_kvec(struct smb_rqst *rqst, unsigned int idx,
270 iov->iov_len = rqst->rq_pagesz; 270 iov->iov_len = rqst->rq_pagesz;
271} 271}
272 272
273static unsigned long
274rqst_len(struct smb_rqst *rqst)
275{
276 unsigned int i;
277 struct kvec *iov = rqst->rq_iov;
278 unsigned long buflen = 0;
279
280 /* total up iov array first */
281 for (i = 0; i < rqst->rq_nvec; i++)
282 buflen += iov[i].iov_len;
283
284 /* add in the page array if there is one */
285 if (rqst->rq_npages) {
286 buflen += rqst->rq_pagesz * (rqst->rq_npages - 1);
287 buflen += rqst->rq_tailsz;
288 }
289
290 return buflen;
291}
292
273static int 293static int
274smb_send_rqst(struct TCP_Server_Info *server, struct smb_rqst *rqst) 294smb_send_rqst(struct TCP_Server_Info *server, struct smb_rqst *rqst)
275{ 295{
@@ -277,6 +297,7 @@ smb_send_rqst(struct TCP_Server_Info *server, struct smb_rqst *rqst)
277 struct kvec *iov = rqst->rq_iov; 297 struct kvec *iov = rqst->rq_iov;
278 int n_vec = rqst->rq_nvec; 298 int n_vec = rqst->rq_nvec;
279 unsigned int smb_buf_length = get_rfc1002_length(iov[0].iov_base); 299 unsigned int smb_buf_length = get_rfc1002_length(iov[0].iov_base);
300 unsigned long send_length;
280 unsigned int i; 301 unsigned int i;
281 size_t total_len = 0, sent; 302 size_t total_len = 0, sent;
282 struct socket *ssocket = server->ssocket; 303 struct socket *ssocket = server->ssocket;
@@ -285,6 +306,14 @@ smb_send_rqst(struct TCP_Server_Info *server, struct smb_rqst *rqst)
285 if (ssocket == NULL) 306 if (ssocket == NULL)
286 return -ENOTSOCK; 307 return -ENOTSOCK;
287 308
309 /* sanity check send length */
310 send_length = rqst_len(rqst);
311 if (send_length != smb_buf_length + 4) {
312 WARN(1, "Send length mismatch(send_length=%lu smb_buf_length=%u)\n",
313 send_length, smb_buf_length);
314 return -EIO;
315 }
316
288 cifs_dbg(FYI, "Sending smb: smb_len=%u\n", smb_buf_length); 317 cifs_dbg(FYI, "Sending smb: smb_len=%u\n", smb_buf_length);
289 dump_smb(iov[0].iov_base, iov[0].iov_len); 318 dump_smb(iov[0].iov_base, iov[0].iov_len);
290 319
diff --git a/fs/dcache.c b/fs/dcache.c
index 265e0ce9769c..ca02c13a84aa 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -2833,9 +2833,9 @@ static int prepend_name(char **buffer, int *buflen, struct qstr *name)
2833 u32 dlen = ACCESS_ONCE(name->len); 2833 u32 dlen = ACCESS_ONCE(name->len);
2834 char *p; 2834 char *p;
2835 2835
2836 if (*buflen < dlen + 1)
2837 return -ENAMETOOLONG;
2838 *buflen -= dlen + 1; 2836 *buflen -= dlen + 1;
2837 if (*buflen < 0)
2838 return -ENAMETOOLONG;
2839 p = *buffer -= dlen + 1; 2839 p = *buffer -= dlen + 1;
2840 *p++ = '/'; 2840 *p++ = '/';
2841 while (dlen--) { 2841 while (dlen--) {
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index ece55565b9cd..d3a534fdc5ff 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -771,6 +771,8 @@ do { \
771 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 771 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
772 (einode)->xtime.tv_sec = \ 772 (einode)->xtime.tv_sec = \
773 (signed)le32_to_cpu((raw_inode)->xtime); \ 773 (signed)le32_to_cpu((raw_inode)->xtime); \
774 else \
775 (einode)->xtime.tv_sec = 0; \
774 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 776 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
775 ext4_decode_extra_time(&(einode)->xtime, \ 777 ext4_decode_extra_time(&(einode)->xtime, \
776 raw_inode->xtime ## _extra); \ 778 raw_inode->xtime ## _extra); \
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 10cff4736b11..74bc2d549c58 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -3906,6 +3906,7 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
3906 } else 3906 } else
3907 err = ret; 3907 err = ret;
3908 map->m_flags |= EXT4_MAP_MAPPED; 3908 map->m_flags |= EXT4_MAP_MAPPED;
3909 map->m_pblk = newblock;
3909 if (allocated > map->m_len) 3910 if (allocated > map->m_len)
3910 allocated = map->m_len; 3911 allocated = map->m_len;
3911 map->m_len = allocated; 3912 map->m_len = allocated;
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 6e39895a91b8..24bfd7ff3049 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -38,6 +38,7 @@
38#include <linux/slab.h> 38#include <linux/slab.h>
39#include <linux/ratelimit.h> 39#include <linux/ratelimit.h>
40#include <linux/aio.h> 40#include <linux/aio.h>
41#include <linux/bitops.h>
41 42
42#include "ext4_jbd2.h" 43#include "ext4_jbd2.h"
43#include "xattr.h" 44#include "xattr.h"
@@ -3921,18 +3922,20 @@ int ext4_get_inode_loc(struct inode *inode, struct ext4_iloc *iloc)
3921void ext4_set_inode_flags(struct inode *inode) 3922void ext4_set_inode_flags(struct inode *inode)
3922{ 3923{
3923 unsigned int flags = EXT4_I(inode)->i_flags; 3924 unsigned int flags = EXT4_I(inode)->i_flags;
3925 unsigned int new_fl = 0;
3924 3926
3925 inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
3926 if (flags & EXT4_SYNC_FL) 3927 if (flags & EXT4_SYNC_FL)
3927 inode->i_flags |= S_SYNC; 3928 new_fl |= S_SYNC;
3928 if (flags & EXT4_APPEND_FL) 3929 if (flags & EXT4_APPEND_FL)
3929 inode->i_flags |= S_APPEND; 3930 new_fl |= S_APPEND;
3930 if (flags & EXT4_IMMUTABLE_FL) 3931 if (flags & EXT4_IMMUTABLE_FL)
3931 inode->i_flags |= S_IMMUTABLE; 3932 new_fl |= S_IMMUTABLE;
3932 if (flags & EXT4_NOATIME_FL) 3933 if (flags & EXT4_NOATIME_FL)
3933 inode->i_flags |= S_NOATIME; 3934 new_fl |= S_NOATIME;
3934 if (flags & EXT4_DIRSYNC_FL) 3935 if (flags & EXT4_DIRSYNC_FL)
3935 inode->i_flags |= S_DIRSYNC; 3936 new_fl |= S_DIRSYNC;
3937 set_mask_bits(&inode->i_flags,
3938 S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC, new_fl);
3936} 3939}
3937 3940
3938/* Propagate flags from i_flags to EXT4_I(inode)->i_flags */ 3941/* Propagate flags from i_flags to EXT4_I(inode)->i_flags */
diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
index 6bea80614d77..a2a837f00407 100644
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -140,7 +140,7 @@ static long swap_inode_boot_loader(struct super_block *sb,
140 handle = ext4_journal_start(inode_bl, EXT4_HT_MOVE_EXTENTS, 2); 140 handle = ext4_journal_start(inode_bl, EXT4_HT_MOVE_EXTENTS, 2);
141 if (IS_ERR(handle)) { 141 if (IS_ERR(handle)) {
142 err = -EINVAL; 142 err = -EINVAL;
143 goto swap_boot_out; 143 goto journal_err_out;
144 } 144 }
145 145
146 /* Protect extent tree against block allocations via delalloc */ 146 /* Protect extent tree against block allocations via delalloc */
@@ -198,6 +198,7 @@ static long swap_inode_boot_loader(struct super_block *sb,
198 198
199 ext4_double_up_write_data_sem(inode, inode_bl); 199 ext4_double_up_write_data_sem(inode, inode_bl);
200 200
201journal_err_out:
201 ext4_inode_resume_unlocked_dio(inode); 202 ext4_inode_resume_unlocked_dio(inode);
202 ext4_inode_resume_unlocked_dio(inode_bl); 203 ext4_inode_resume_unlocked_dio(inode_bl);
203 204
diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c
index c5adbb318a90..f3b84cd9de56 100644
--- a/fs/ext4/resize.c
+++ b/fs/ext4/resize.c
@@ -243,6 +243,7 @@ static int ext4_alloc_group_tables(struct super_block *sb,
243 ext4_group_t group; 243 ext4_group_t group;
244 ext4_group_t last_group; 244 ext4_group_t last_group;
245 unsigned overhead; 245 unsigned overhead;
246 __u16 uninit_mask = (flexbg_size > 1) ? ~EXT4_BG_BLOCK_UNINIT : ~0;
246 247
247 BUG_ON(flex_gd->count == 0 || group_data == NULL); 248 BUG_ON(flex_gd->count == 0 || group_data == NULL);
248 249
@@ -266,7 +267,7 @@ next_group:
266 src_group++; 267 src_group++;
267 for (; src_group <= last_group; src_group++) { 268 for (; src_group <= last_group; src_group++) {
268 overhead = ext4_group_overhead_blocks(sb, src_group); 269 overhead = ext4_group_overhead_blocks(sb, src_group);
269 if (overhead != 0) 270 if (overhead == 0)
270 last_blk += group_data[src_group - group].blocks_count; 271 last_blk += group_data[src_group - group].blocks_count;
271 else 272 else
272 break; 273 break;
@@ -280,8 +281,7 @@ next_group:
280 group = ext4_get_group_number(sb, start_blk - 1); 281 group = ext4_get_group_number(sb, start_blk - 1);
281 group -= group_data[0].group; 282 group -= group_data[0].group;
282 group_data[group].free_blocks_count--; 283 group_data[group].free_blocks_count--;
283 if (flexbg_size > 1) 284 flex_gd->bg_flags[group] &= uninit_mask;
284 flex_gd->bg_flags[group] &= ~EXT4_BG_BLOCK_UNINIT;
285 } 285 }
286 286
287 /* Allocate inode bitmaps */ 287 /* Allocate inode bitmaps */
@@ -292,22 +292,30 @@ next_group:
292 group = ext4_get_group_number(sb, start_blk - 1); 292 group = ext4_get_group_number(sb, start_blk - 1);
293 group -= group_data[0].group; 293 group -= group_data[0].group;
294 group_data[group].free_blocks_count--; 294 group_data[group].free_blocks_count--;
295 if (flexbg_size > 1) 295 flex_gd->bg_flags[group] &= uninit_mask;
296 flex_gd->bg_flags[group] &= ~EXT4_BG_BLOCK_UNINIT;
297 } 296 }
298 297
299 /* Allocate inode tables */ 298 /* Allocate inode tables */
300 for (; it_index < flex_gd->count; it_index++) { 299 for (; it_index < flex_gd->count; it_index++) {
301 if (start_blk + EXT4_SB(sb)->s_itb_per_group > last_blk) 300 unsigned int itb = EXT4_SB(sb)->s_itb_per_group;
301 ext4_fsblk_t next_group_start;
302
303 if (start_blk + itb > last_blk)
302 goto next_group; 304 goto next_group;
303 group_data[it_index].inode_table = start_blk; 305 group_data[it_index].inode_table = start_blk;
304 group = ext4_get_group_number(sb, start_blk - 1); 306 group = ext4_get_group_number(sb, start_blk);
307 next_group_start = ext4_group_first_block_no(sb, group + 1);
305 group -= group_data[0].group; 308 group -= group_data[0].group;
306 group_data[group].free_blocks_count -=
307 EXT4_SB(sb)->s_itb_per_group;
308 if (flexbg_size > 1)
309 flex_gd->bg_flags[group] &= ~EXT4_BG_BLOCK_UNINIT;
310 309
310 if (start_blk + itb > next_group_start) {
311 flex_gd->bg_flags[group + 1] &= uninit_mask;
312 overhead = start_blk + itb - next_group_start;
313 group_data[group + 1].free_blocks_count -= overhead;
314 itb -= overhead;
315 }
316
317 group_data[group].free_blocks_count -= itb;
318 flex_gd->bg_flags[group] &= uninit_mask;
311 start_blk += EXT4_SB(sb)->s_itb_per_group; 319 start_blk += EXT4_SB(sb)->s_itb_per_group;
312 } 320 }
313 321
@@ -401,7 +409,7 @@ static int set_flexbg_block_bitmap(struct super_block *sb, handle_t *handle,
401 start = ext4_group_first_block_no(sb, group); 409 start = ext4_group_first_block_no(sb, group);
402 group -= flex_gd->groups[0].group; 410 group -= flex_gd->groups[0].group;
403 411
404 count2 = sb->s_blocksize * 8 - (block - start); 412 count2 = EXT4_BLOCKS_PER_GROUP(sb) - (block - start);
405 if (count2 > count) 413 if (count2 > count)
406 count2 = count; 414 count2 = count;
407 415
@@ -620,7 +628,7 @@ handle_ib:
620 if (err) 628 if (err)
621 goto out; 629 goto out;
622 count = group_table_count[j]; 630 count = group_table_count[j];
623 start = group_data[i].block_bitmap; 631 start = (&group_data[i].block_bitmap)[j];
624 block = start; 632 block = start;
625 } 633 }
626 634
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 1f7784de05b6..710fed2377d4 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -3695,16 +3695,22 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
3695 for (i = 0; i < 4; i++) 3695 for (i = 0; i < 4; i++)
3696 sbi->s_hash_seed[i] = le32_to_cpu(es->s_hash_seed[i]); 3696 sbi->s_hash_seed[i] = le32_to_cpu(es->s_hash_seed[i]);
3697 sbi->s_def_hash_version = es->s_def_hash_version; 3697 sbi->s_def_hash_version = es->s_def_hash_version;
3698 i = le32_to_cpu(es->s_flags); 3698 if (EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_DIR_INDEX)) {
3699 if (i & EXT2_FLAGS_UNSIGNED_HASH) 3699 i = le32_to_cpu(es->s_flags);
3700 sbi->s_hash_unsigned = 3; 3700 if (i & EXT2_FLAGS_UNSIGNED_HASH)
3701 else if ((i & EXT2_FLAGS_SIGNED_HASH) == 0) { 3701 sbi->s_hash_unsigned = 3;
3702 else if ((i & EXT2_FLAGS_SIGNED_HASH) == 0) {
3702#ifdef __CHAR_UNSIGNED__ 3703#ifdef __CHAR_UNSIGNED__
3703 es->s_flags |= cpu_to_le32(EXT2_FLAGS_UNSIGNED_HASH); 3704 if (!(sb->s_flags & MS_RDONLY))
3704 sbi->s_hash_unsigned = 3; 3705 es->s_flags |=
3706 cpu_to_le32(EXT2_FLAGS_UNSIGNED_HASH);
3707 sbi->s_hash_unsigned = 3;
3705#else 3708#else
3706 es->s_flags |= cpu_to_le32(EXT2_FLAGS_SIGNED_HASH); 3709 if (!(sb->s_flags & MS_RDONLY))
3710 es->s_flags |=
3711 cpu_to_le32(EXT2_FLAGS_SIGNED_HASH);
3707#endif 3712#endif
3713 }
3708 } 3714 }
3709 3715
3710 /* Handle clustersize */ 3716 /* Handle clustersize */
diff --git a/fs/file.c b/fs/file.c
index db25c2bdfe46..eb56a13dab3e 100644
--- a/fs/file.c
+++ b/fs/file.c
@@ -683,35 +683,54 @@ EXPORT_SYMBOL(fget_raw);
683 * The fput_needed flag returned by fget_light should be passed to the 683 * The fput_needed flag returned by fget_light should be passed to the
684 * corresponding fput_light. 684 * corresponding fput_light.
685 */ 685 */
686struct file *__fget_light(unsigned int fd, fmode_t mask, int *fput_needed) 686static unsigned long __fget_light(unsigned int fd, fmode_t mask)
687{ 687{
688 struct files_struct *files = current->files; 688 struct files_struct *files = current->files;
689 struct file *file; 689 struct file *file;
690 690
691 *fput_needed = 0;
692 if (atomic_read(&files->count) == 1) { 691 if (atomic_read(&files->count) == 1) {
693 file = __fcheck_files(files, fd); 692 file = __fcheck_files(files, fd);
694 if (file && (file->f_mode & mask)) 693 if (!file || unlikely(file->f_mode & mask))
695 file = NULL; 694 return 0;
695 return (unsigned long)file;
696 } else { 696 } else {
697 file = __fget(fd, mask); 697 file = __fget(fd, mask);
698 if (file) 698 if (!file)
699 *fput_needed = 1; 699 return 0;
700 return FDPUT_FPUT | (unsigned long)file;
700 } 701 }
701
702 return file;
703} 702}
704struct file *fget_light(unsigned int fd, int *fput_needed) 703unsigned long __fdget(unsigned int fd)
705{ 704{
706 return __fget_light(fd, FMODE_PATH, fput_needed); 705 return __fget_light(fd, FMODE_PATH);
707} 706}
708EXPORT_SYMBOL(fget_light); 707EXPORT_SYMBOL(__fdget);
709 708
710struct file *fget_raw_light(unsigned int fd, int *fput_needed) 709unsigned long __fdget_raw(unsigned int fd)
711{ 710{
712 return __fget_light(fd, 0, fput_needed); 711 return __fget_light(fd, 0);
713} 712}
714 713
714unsigned long __fdget_pos(unsigned int fd)
715{
716 unsigned long v = __fdget(fd);
717 struct file *file = (struct file *)(v & ~3);
718
719 if (file && (file->f_mode & FMODE_ATOMIC_POS)) {
720 if (file_count(file) > 1) {
721 v |= FDPUT_POS_UNLOCK;
722 mutex_lock(&file->f_pos_lock);
723 }
724 }
725 return v;
726}
727
728/*
729 * We only lock f_pos if we have threads or if the file might be
730 * shared with another process. In both cases we'll have an elevated
731 * file count (done either by fdget() or by fork()).
732 */
733
715void set_close_on_exec(unsigned int fd, int flag) 734void set_close_on_exec(unsigned int fd, int flag)
716{ 735{
717 struct files_struct *files = current->files; 736 struct files_struct *files = current->files;
diff --git a/fs/file_table.c b/fs/file_table.c
index 5fff9030be34..5b24008ea4f6 100644
--- a/fs/file_table.c
+++ b/fs/file_table.c
@@ -135,6 +135,7 @@ struct file *get_empty_filp(void)
135 atomic_long_set(&f->f_count, 1); 135 atomic_long_set(&f->f_count, 1);
136 rwlock_init(&f->f_owner.lock); 136 rwlock_init(&f->f_owner.lock);
137 spin_lock_init(&f->f_lock); 137 spin_lock_init(&f->f_lock);
138 mutex_init(&f->f_pos_lock);
138 eventpoll_init_file(f); 139 eventpoll_init_file(f);
139 /* f->f_version: 0 */ 140 /* f->f_version: 0 */
140 return f; 141 return f;
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index e0259a163f98..d754e3cf99a8 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -40,18 +40,13 @@
40struct wb_writeback_work { 40struct wb_writeback_work {
41 long nr_pages; 41 long nr_pages;
42 struct super_block *sb; 42 struct super_block *sb;
43 /* 43 unsigned long *older_than_this;
44 * Write only inodes dirtied before this time. Don't forget to set
45 * older_than_this_is_set when you set this.
46 */
47 unsigned long older_than_this;
48 enum writeback_sync_modes sync_mode; 44 enum writeback_sync_modes sync_mode;
49 unsigned int tagged_writepages:1; 45 unsigned int tagged_writepages:1;
50 unsigned int for_kupdate:1; 46 unsigned int for_kupdate:1;
51 unsigned int range_cyclic:1; 47 unsigned int range_cyclic:1;
52 unsigned int for_background:1; 48 unsigned int for_background:1;
53 unsigned int for_sync:1; /* sync(2) WB_SYNC_ALL writeback */ 49 unsigned int for_sync:1; /* sync(2) WB_SYNC_ALL writeback */
54 unsigned int older_than_this_is_set:1;
55 enum wb_reason reason; /* why was writeback initiated? */ 50 enum wb_reason reason; /* why was writeback initiated? */
56 51
57 struct list_head list; /* pending work list */ 52 struct list_head list; /* pending work list */
@@ -252,10 +247,10 @@ static int move_expired_inodes(struct list_head *delaying_queue,
252 int do_sb_sort = 0; 247 int do_sb_sort = 0;
253 int moved = 0; 248 int moved = 0;
254 249
255 WARN_ON_ONCE(!work->older_than_this_is_set);
256 while (!list_empty(delaying_queue)) { 250 while (!list_empty(delaying_queue)) {
257 inode = wb_inode(delaying_queue->prev); 251 inode = wb_inode(delaying_queue->prev);
258 if (inode_dirtied_after(inode, work->older_than_this)) 252 if (work->older_than_this &&
253 inode_dirtied_after(inode, *work->older_than_this))
259 break; 254 break;
260 list_move(&inode->i_wb_list, &tmp); 255 list_move(&inode->i_wb_list, &tmp);
261 moved++; 256 moved++;
@@ -742,8 +737,6 @@ static long writeback_inodes_wb(struct bdi_writeback *wb, long nr_pages,
742 .sync_mode = WB_SYNC_NONE, 737 .sync_mode = WB_SYNC_NONE,
743 .range_cyclic = 1, 738 .range_cyclic = 1,
744 .reason = reason, 739 .reason = reason,
745 .older_than_this = jiffies,
746 .older_than_this_is_set = 1,
747 }; 740 };
748 741
749 spin_lock(&wb->list_lock); 742 spin_lock(&wb->list_lock);
@@ -802,13 +795,12 @@ static long wb_writeback(struct bdi_writeback *wb,
802{ 795{
803 unsigned long wb_start = jiffies; 796 unsigned long wb_start = jiffies;
804 long nr_pages = work->nr_pages; 797 long nr_pages = work->nr_pages;
798 unsigned long oldest_jif;
805 struct inode *inode; 799 struct inode *inode;
806 long progress; 800 long progress;
807 801
808 if (!work->older_than_this_is_set) { 802 oldest_jif = jiffies;
809 work->older_than_this = jiffies; 803 work->older_than_this = &oldest_jif;
810 work->older_than_this_is_set = 1;
811 }
812 804
813 spin_lock(&wb->list_lock); 805 spin_lock(&wb->list_lock);
814 for (;;) { 806 for (;;) {
@@ -842,10 +834,10 @@ static long wb_writeback(struct bdi_writeback *wb,
842 * safe. 834 * safe.
843 */ 835 */
844 if (work->for_kupdate) { 836 if (work->for_kupdate) {
845 work->older_than_this = jiffies - 837 oldest_jif = jiffies -
846 msecs_to_jiffies(dirty_expire_interval * 10); 838 msecs_to_jiffies(dirty_expire_interval * 10);
847 } else if (work->for_background) 839 } else if (work->for_background)
848 work->older_than_this = jiffies; 840 oldest_jif = jiffies;
849 841
850 trace_writeback_start(wb->bdi, work); 842 trace_writeback_start(wb->bdi, work);
851 if (list_empty(&wb->b_io)) 843 if (list_empty(&wb->b_io))
@@ -1357,21 +1349,18 @@ EXPORT_SYMBOL(try_to_writeback_inodes_sb);
1357 1349
1358/** 1350/**
1359 * sync_inodes_sb - sync sb inode pages 1351 * sync_inodes_sb - sync sb inode pages
1360 * @sb: the superblock 1352 * @sb: the superblock
1361 * @older_than_this: timestamp
1362 * 1353 *
1363 * This function writes and waits on any dirty inode belonging to this 1354 * This function writes and waits on any dirty inode belonging to this
1364 * superblock that has been dirtied before given timestamp. 1355 * super_block.
1365 */ 1356 */
1366void sync_inodes_sb(struct super_block *sb, unsigned long older_than_this) 1357void sync_inodes_sb(struct super_block *sb)
1367{ 1358{
1368 DECLARE_COMPLETION_ONSTACK(done); 1359 DECLARE_COMPLETION_ONSTACK(done);
1369 struct wb_writeback_work work = { 1360 struct wb_writeback_work work = {
1370 .sb = sb, 1361 .sb = sb,
1371 .sync_mode = WB_SYNC_ALL, 1362 .sync_mode = WB_SYNC_ALL,
1372 .nr_pages = LONG_MAX, 1363 .nr_pages = LONG_MAX,
1373 .older_than_this = older_than_this,
1374 .older_than_this_is_set = 1,
1375 .range_cyclic = 0, 1364 .range_cyclic = 0,
1376 .done = &done, 1365 .done = &done,
1377 .reason = WB_REASON_SYNC, 1366 .reason = WB_REASON_SYNC,
diff --git a/fs/fscache/object-list.c b/fs/fscache/object-list.c
index e1959efad64f..b5ebc2d7d80d 100644
--- a/fs/fscache/object-list.c
+++ b/fs/fscache/object-list.c
@@ -50,6 +50,8 @@ void fscache_objlist_add(struct fscache_object *obj)
50 struct fscache_object *xobj; 50 struct fscache_object *xobj;
51 struct rb_node **p = &fscache_object_list.rb_node, *parent = NULL; 51 struct rb_node **p = &fscache_object_list.rb_node, *parent = NULL;
52 52
53 ASSERT(RB_EMPTY_NODE(&obj->objlist_link));
54
53 write_lock(&fscache_object_list_lock); 55 write_lock(&fscache_object_list_lock);
54 56
55 while (*p) { 57 while (*p) {
@@ -75,6 +77,9 @@ void fscache_objlist_add(struct fscache_object *obj)
75 */ 77 */
76void fscache_objlist_remove(struct fscache_object *obj) 78void fscache_objlist_remove(struct fscache_object *obj)
77{ 79{
80 if (RB_EMPTY_NODE(&obj->objlist_link))
81 return;
82
78 write_lock(&fscache_object_list_lock); 83 write_lock(&fscache_object_list_lock);
79 84
80 BUG_ON(RB_EMPTY_ROOT(&fscache_object_list)); 85 BUG_ON(RB_EMPTY_ROOT(&fscache_object_list));
diff --git a/fs/fscache/object.c b/fs/fscache/object.c
index 53d35c504240..d3b4539f1651 100644
--- a/fs/fscache/object.c
+++ b/fs/fscache/object.c
@@ -314,6 +314,9 @@ void fscache_object_init(struct fscache_object *object,
314 object->cache = cache; 314 object->cache = cache;
315 object->cookie = cookie; 315 object->cookie = cookie;
316 object->parent = NULL; 316 object->parent = NULL;
317#ifdef CONFIG_FSCACHE_OBJECT_LIST
318 RB_CLEAR_NODE(&object->objlist_link);
319#endif
317 320
318 object->oob_event_mask = 0; 321 object->oob_event_mask = 0;
319 for (t = object->oob_table; t->events; t++) 322 for (t = object->oob_table; t->events; t++)
diff --git a/fs/hfsplus/catalog.c b/fs/hfsplus/catalog.c
index 968ce411db53..32602c667b4a 100644
--- a/fs/hfsplus/catalog.c
+++ b/fs/hfsplus/catalog.c
@@ -103,6 +103,8 @@ static int hfsplus_cat_build_record(hfsplus_cat_entry *entry,
103 folder = &entry->folder; 103 folder = &entry->folder;
104 memset(folder, 0, sizeof(*folder)); 104 memset(folder, 0, sizeof(*folder));
105 folder->type = cpu_to_be16(HFSPLUS_FOLDER); 105 folder->type = cpu_to_be16(HFSPLUS_FOLDER);
106 if (test_bit(HFSPLUS_SB_HFSX, &sbi->flags))
107 folder->flags |= cpu_to_be16(HFSPLUS_HAS_FOLDER_COUNT);
106 folder->id = cpu_to_be32(inode->i_ino); 108 folder->id = cpu_to_be32(inode->i_ino);
107 HFSPLUS_I(inode)->create_date = 109 HFSPLUS_I(inode)->create_date =
108 folder->create_date = 110 folder->create_date =
@@ -203,6 +205,36 @@ int hfsplus_find_cat(struct super_block *sb, u32 cnid,
203 return hfs_brec_find(fd, hfs_find_rec_by_key); 205 return hfs_brec_find(fd, hfs_find_rec_by_key);
204} 206}
205 207
208static void hfsplus_subfolders_inc(struct inode *dir)
209{
210 struct hfsplus_sb_info *sbi = HFSPLUS_SB(dir->i_sb);
211
212 if (test_bit(HFSPLUS_SB_HFSX, &sbi->flags)) {
213 /*
214 * Increment subfolder count. Note, the value is only meaningful
215 * for folders with HFSPLUS_HAS_FOLDER_COUNT flag set.
216 */
217 HFSPLUS_I(dir)->subfolders++;
218 }
219}
220
221static void hfsplus_subfolders_dec(struct inode *dir)
222{
223 struct hfsplus_sb_info *sbi = HFSPLUS_SB(dir->i_sb);
224
225 if (test_bit(HFSPLUS_SB_HFSX, &sbi->flags)) {
226 /*
227 * Decrement subfolder count. Note, the value is only meaningful
228 * for folders with HFSPLUS_HAS_FOLDER_COUNT flag set.
229 *
230 * Check for zero. Some subfolders may have been created
231 * by an implementation ignorant of this counter.
232 */
233 if (HFSPLUS_I(dir)->subfolders)
234 HFSPLUS_I(dir)->subfolders--;
235 }
236}
237
206int hfsplus_create_cat(u32 cnid, struct inode *dir, 238int hfsplus_create_cat(u32 cnid, struct inode *dir,
207 struct qstr *str, struct inode *inode) 239 struct qstr *str, struct inode *inode)
208{ 240{
@@ -247,6 +279,8 @@ int hfsplus_create_cat(u32 cnid, struct inode *dir,
247 goto err1; 279 goto err1;
248 280
249 dir->i_size++; 281 dir->i_size++;
282 if (S_ISDIR(inode->i_mode))
283 hfsplus_subfolders_inc(dir);
250 dir->i_mtime = dir->i_ctime = CURRENT_TIME_SEC; 284 dir->i_mtime = dir->i_ctime = CURRENT_TIME_SEC;
251 hfsplus_mark_inode_dirty(dir, HFSPLUS_I_CAT_DIRTY); 285 hfsplus_mark_inode_dirty(dir, HFSPLUS_I_CAT_DIRTY);
252 286
@@ -336,6 +370,8 @@ int hfsplus_delete_cat(u32 cnid, struct inode *dir, struct qstr *str)
336 goto out; 370 goto out;
337 371
338 dir->i_size--; 372 dir->i_size--;
373 if (type == HFSPLUS_FOLDER)
374 hfsplus_subfolders_dec(dir);
339 dir->i_mtime = dir->i_ctime = CURRENT_TIME_SEC; 375 dir->i_mtime = dir->i_ctime = CURRENT_TIME_SEC;
340 hfsplus_mark_inode_dirty(dir, HFSPLUS_I_CAT_DIRTY); 376 hfsplus_mark_inode_dirty(dir, HFSPLUS_I_CAT_DIRTY);
341 377
@@ -380,6 +416,7 @@ int hfsplus_rename_cat(u32 cnid,
380 416
381 hfs_bnode_read(src_fd.bnode, &entry, src_fd.entryoffset, 417 hfs_bnode_read(src_fd.bnode, &entry, src_fd.entryoffset,
382 src_fd.entrylength); 418 src_fd.entrylength);
419 type = be16_to_cpu(entry.type);
383 420
384 /* create new dir entry with the data from the old entry */ 421 /* create new dir entry with the data from the old entry */
385 hfsplus_cat_build_key(sb, dst_fd.search_key, dst_dir->i_ino, dst_name); 422 hfsplus_cat_build_key(sb, dst_fd.search_key, dst_dir->i_ino, dst_name);
@@ -394,6 +431,8 @@ int hfsplus_rename_cat(u32 cnid,
394 if (err) 431 if (err)
395 goto out; 432 goto out;
396 dst_dir->i_size++; 433 dst_dir->i_size++;
434 if (type == HFSPLUS_FOLDER)
435 hfsplus_subfolders_inc(dst_dir);
397 dst_dir->i_mtime = dst_dir->i_ctime = CURRENT_TIME_SEC; 436 dst_dir->i_mtime = dst_dir->i_ctime = CURRENT_TIME_SEC;
398 437
399 /* finally remove the old entry */ 438 /* finally remove the old entry */
@@ -405,6 +444,8 @@ int hfsplus_rename_cat(u32 cnid,
405 if (err) 444 if (err)
406 goto out; 445 goto out;
407 src_dir->i_size--; 446 src_dir->i_size--;
447 if (type == HFSPLUS_FOLDER)
448 hfsplus_subfolders_dec(src_dir);
408 src_dir->i_mtime = src_dir->i_ctime = CURRENT_TIME_SEC; 449 src_dir->i_mtime = src_dir->i_ctime = CURRENT_TIME_SEC;
409 450
410 /* remove old thread entry */ 451 /* remove old thread entry */
diff --git a/fs/hfsplus/hfsplus_fs.h b/fs/hfsplus/hfsplus_fs.h
index 08846425b67f..62d571eb69ba 100644
--- a/fs/hfsplus/hfsplus_fs.h
+++ b/fs/hfsplus/hfsplus_fs.h
@@ -242,6 +242,7 @@ struct hfsplus_inode_info {
242 */ 242 */
243 sector_t fs_blocks; 243 sector_t fs_blocks;
244 u8 userflags; /* BSD user file flags */ 244 u8 userflags; /* BSD user file flags */
245 u32 subfolders; /* Subfolder count (HFSX only) */
245 struct list_head open_dir_list; 246 struct list_head open_dir_list;
246 loff_t phys_size; 247 loff_t phys_size;
247 248
diff --git a/fs/hfsplus/hfsplus_raw.h b/fs/hfsplus/hfsplus_raw.h
index 8ffb3a8ffe75..5a126828d85e 100644
--- a/fs/hfsplus/hfsplus_raw.h
+++ b/fs/hfsplus/hfsplus_raw.h
@@ -261,7 +261,7 @@ struct hfsplus_cat_folder {
261 struct DInfo user_info; 261 struct DInfo user_info;
262 struct DXInfo finder_info; 262 struct DXInfo finder_info;
263 __be32 text_encoding; 263 __be32 text_encoding;
264 u32 reserved; 264 __be32 subfolders; /* Subfolder count in HFSX. Reserved in HFS+. */
265} __packed; 265} __packed;
266 266
267/* HFS file info (stolen from hfs.h) */ 267/* HFS file info (stolen from hfs.h) */
@@ -301,11 +301,13 @@ struct hfsplus_cat_file {
301 struct hfsplus_fork_raw rsrc_fork; 301 struct hfsplus_fork_raw rsrc_fork;
302} __packed; 302} __packed;
303 303
304/* File attribute bits */ 304/* File and folder flag bits */
305#define HFSPLUS_FILE_LOCKED 0x0001 305#define HFSPLUS_FILE_LOCKED 0x0001
306#define HFSPLUS_FILE_THREAD_EXISTS 0x0002 306#define HFSPLUS_FILE_THREAD_EXISTS 0x0002
307#define HFSPLUS_XATTR_EXISTS 0x0004 307#define HFSPLUS_XATTR_EXISTS 0x0004
308#define HFSPLUS_ACL_EXISTS 0x0008 308#define HFSPLUS_ACL_EXISTS 0x0008
309#define HFSPLUS_HAS_FOLDER_COUNT 0x0010 /* Folder has subfolder count
310 * (HFSX only) */
309 311
310/* HFS+ catalog thread (part of a cat_entry) */ 312/* HFS+ catalog thread (part of a cat_entry) */
311struct hfsplus_cat_thread { 313struct hfsplus_cat_thread {
diff --git a/fs/hfsplus/inode.c b/fs/hfsplus/inode.c
index fa929f325f87..a4f45bd88a63 100644
--- a/fs/hfsplus/inode.c
+++ b/fs/hfsplus/inode.c
@@ -375,6 +375,7 @@ struct inode *hfsplus_new_inode(struct super_block *sb, umode_t mode)
375 hip->extent_state = 0; 375 hip->extent_state = 0;
376 hip->flags = 0; 376 hip->flags = 0;
377 hip->userflags = 0; 377 hip->userflags = 0;
378 hip->subfolders = 0;
378 memset(hip->first_extents, 0, sizeof(hfsplus_extent_rec)); 379 memset(hip->first_extents, 0, sizeof(hfsplus_extent_rec));
379 memset(hip->cached_extents, 0, sizeof(hfsplus_extent_rec)); 380 memset(hip->cached_extents, 0, sizeof(hfsplus_extent_rec));
380 hip->alloc_blocks = 0; 381 hip->alloc_blocks = 0;
@@ -494,6 +495,10 @@ int hfsplus_cat_read_inode(struct inode *inode, struct hfs_find_data *fd)
494 inode->i_ctime = hfsp_mt2ut(folder->attribute_mod_date); 495 inode->i_ctime = hfsp_mt2ut(folder->attribute_mod_date);
495 HFSPLUS_I(inode)->create_date = folder->create_date; 496 HFSPLUS_I(inode)->create_date = folder->create_date;
496 HFSPLUS_I(inode)->fs_blocks = 0; 497 HFSPLUS_I(inode)->fs_blocks = 0;
498 if (folder->flags & cpu_to_be16(HFSPLUS_HAS_FOLDER_COUNT)) {
499 HFSPLUS_I(inode)->subfolders =
500 be32_to_cpu(folder->subfolders);
501 }
497 inode->i_op = &hfsplus_dir_inode_operations; 502 inode->i_op = &hfsplus_dir_inode_operations;
498 inode->i_fop = &hfsplus_dir_operations; 503 inode->i_fop = &hfsplus_dir_operations;
499 } else if (type == HFSPLUS_FILE) { 504 } else if (type == HFSPLUS_FILE) {
@@ -566,6 +571,10 @@ int hfsplus_cat_write_inode(struct inode *inode)
566 folder->content_mod_date = hfsp_ut2mt(inode->i_mtime); 571 folder->content_mod_date = hfsp_ut2mt(inode->i_mtime);
567 folder->attribute_mod_date = hfsp_ut2mt(inode->i_ctime); 572 folder->attribute_mod_date = hfsp_ut2mt(inode->i_ctime);
568 folder->valence = cpu_to_be32(inode->i_size - 2); 573 folder->valence = cpu_to_be32(inode->i_size - 2);
574 if (folder->flags & cpu_to_be16(HFSPLUS_HAS_FOLDER_COUNT)) {
575 folder->subfolders =
576 cpu_to_be32(HFSPLUS_I(inode)->subfolders);
577 }
569 hfs_bnode_write(fd.bnode, &entry, fd.entryoffset, 578 hfs_bnode_write(fd.bnode, &entry, fd.entryoffset,
570 sizeof(struct hfsplus_cat_folder)); 579 sizeof(struct hfsplus_cat_folder));
571 } else if (HFSPLUS_IS_RSRC(inode)) { 580 } else if (HFSPLUS_IS_RSRC(inode)) {
diff --git a/fs/hfsplus/options.c b/fs/hfsplus/options.c
index 968eab5bc1f5..68537e8b7a09 100644
--- a/fs/hfsplus/options.c
+++ b/fs/hfsplus/options.c
@@ -75,7 +75,7 @@ int hfsplus_parse_options_remount(char *input, int *force)
75 int token; 75 int token;
76 76
77 if (!input) 77 if (!input)
78 return 0; 78 return 1;
79 79
80 while ((p = strsep(&input, ",")) != NULL) { 80 while ((p = strsep(&input, ",")) != NULL) {
81 if (!*p) 81 if (!*p)
diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index 8360674c85bc..60bb365f54a5 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -514,11 +514,13 @@ int jbd2_journal_start_reserved(handle_t *handle, unsigned int type,
514 * similarly constrained call sites 514 * similarly constrained call sites
515 */ 515 */
516 ret = start_this_handle(journal, handle, GFP_NOFS); 516 ret = start_this_handle(journal, handle, GFP_NOFS);
517 if (ret < 0) 517 if (ret < 0) {
518 jbd2_journal_free_reserved(handle); 518 jbd2_journal_free_reserved(handle);
519 return ret;
520 }
519 handle->h_type = type; 521 handle->h_type = type;
520 handle->h_line_no = line_no; 522 handle->h_line_no = line_no;
521 return ret; 523 return 0;
522} 524}
523EXPORT_SYMBOL(jbd2_journal_start_reserved); 525EXPORT_SYMBOL(jbd2_journal_start_reserved);
524 526
diff --git a/fs/jfs/acl.c b/fs/jfs/acl.c
index e973b85d6afd..5a8ea16eedbc 100644
--- a/fs/jfs/acl.c
+++ b/fs/jfs/acl.c
@@ -86,6 +86,8 @@ static int __jfs_set_acl(tid_t tid, struct inode *inode, int type,
86 rc = posix_acl_equiv_mode(acl, &inode->i_mode); 86 rc = posix_acl_equiv_mode(acl, &inode->i_mode);
87 if (rc < 0) 87 if (rc < 0)
88 return rc; 88 return rc;
89 inode->i_ctime = CURRENT_TIME;
90 mark_inode_dirty(inode);
89 if (rc == 0) 91 if (rc == 0)
90 acl = NULL; 92 acl = NULL;
91 break; 93 break;
diff --git a/fs/kernfs/mount.c b/fs/kernfs/mount.c
index 0d6ce895a9ee..0f4152defe7b 100644
--- a/fs/kernfs/mount.c
+++ b/fs/kernfs/mount.c
@@ -94,6 +94,7 @@ const void *kernfs_super_ns(struct super_block *sb)
94 * @fs_type: file_system_type of the fs being mounted 94 * @fs_type: file_system_type of the fs being mounted
95 * @flags: mount flags specified for the mount 95 * @flags: mount flags specified for the mount
96 * @root: kernfs_root of the hierarchy being mounted 96 * @root: kernfs_root of the hierarchy being mounted
97 * @new_sb_created: tell the caller if we allocated a new superblock
97 * @ns: optional namespace tag of the mount 98 * @ns: optional namespace tag of the mount
98 * 99 *
99 * This is to be called from each kernfs user's file_system_type->mount() 100 * This is to be called from each kernfs user's file_system_type->mount()
@@ -104,7 +105,8 @@ const void *kernfs_super_ns(struct super_block *sb)
104 * The return value can be passed to the vfs layer verbatim. 105 * The return value can be passed to the vfs layer verbatim.
105 */ 106 */
106struct dentry *kernfs_mount_ns(struct file_system_type *fs_type, int flags, 107struct dentry *kernfs_mount_ns(struct file_system_type *fs_type, int flags,
107 struct kernfs_root *root, const void *ns) 108 struct kernfs_root *root, bool *new_sb_created,
109 const void *ns)
108{ 110{
109 struct super_block *sb; 111 struct super_block *sb;
110 struct kernfs_super_info *info; 112 struct kernfs_super_info *info;
@@ -122,6 +124,10 @@ struct dentry *kernfs_mount_ns(struct file_system_type *fs_type, int flags,
122 kfree(info); 124 kfree(info);
123 if (IS_ERR(sb)) 125 if (IS_ERR(sb))
124 return ERR_CAST(sb); 126 return ERR_CAST(sb);
127
128 if (new_sb_created)
129 *new_sb_created = !sb->s_root;
130
125 if (!sb->s_root) { 131 if (!sb->s_root) {
126 error = kernfs_fill_super(sb); 132 error = kernfs_fill_super(sb);
127 if (error) { 133 if (error) {
diff --git a/fs/mount.h b/fs/mount.h
index a17458ca6f29..b29e42f05f34 100644
--- a/fs/mount.h
+++ b/fs/mount.h
@@ -19,13 +19,13 @@ struct mnt_pcp {
19}; 19};
20 20
21struct mountpoint { 21struct mountpoint {
22 struct list_head m_hash; 22 struct hlist_node m_hash;
23 struct dentry *m_dentry; 23 struct dentry *m_dentry;
24 int m_count; 24 int m_count;
25}; 25};
26 26
27struct mount { 27struct mount {
28 struct list_head mnt_hash; 28 struct hlist_node mnt_hash;
29 struct mount *mnt_parent; 29 struct mount *mnt_parent;
30 struct dentry *mnt_mountpoint; 30 struct dentry *mnt_mountpoint;
31 struct vfsmount mnt; 31 struct vfsmount mnt;
diff --git a/fs/namei.c b/fs/namei.c
index 385f7817bfcc..4b491b431990 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -1109,7 +1109,7 @@ static bool __follow_mount_rcu(struct nameidata *nd, struct path *path,
1109 return false; 1109 return false;
1110 1110
1111 if (!d_mountpoint(path->dentry)) 1111 if (!d_mountpoint(path->dentry))
1112 break; 1112 return true;
1113 1113
1114 mounted = __lookup_mnt(path->mnt, path->dentry); 1114 mounted = __lookup_mnt(path->mnt, path->dentry);
1115 if (!mounted) 1115 if (!mounted)
@@ -1125,20 +1125,7 @@ static bool __follow_mount_rcu(struct nameidata *nd, struct path *path,
1125 */ 1125 */
1126 *inode = path->dentry->d_inode; 1126 *inode = path->dentry->d_inode;
1127 } 1127 }
1128 return true; 1128 return read_seqretry(&mount_lock, nd->m_seq);
1129}
1130
1131static void follow_mount_rcu(struct nameidata *nd)
1132{
1133 while (d_mountpoint(nd->path.dentry)) {
1134 struct mount *mounted;
1135 mounted = __lookup_mnt(nd->path.mnt, nd->path.dentry);
1136 if (!mounted)
1137 break;
1138 nd->path.mnt = &mounted->mnt;
1139 nd->path.dentry = mounted->mnt.mnt_root;
1140 nd->seq = read_seqcount_begin(&nd->path.dentry->d_seq);
1141 }
1142} 1129}
1143 1130
1144static int follow_dotdot_rcu(struct nameidata *nd) 1131static int follow_dotdot_rcu(struct nameidata *nd)
@@ -1166,7 +1153,17 @@ static int follow_dotdot_rcu(struct nameidata *nd)
1166 break; 1153 break;
1167 nd->seq = read_seqcount_begin(&nd->path.dentry->d_seq); 1154 nd->seq = read_seqcount_begin(&nd->path.dentry->d_seq);
1168 } 1155 }
1169 follow_mount_rcu(nd); 1156 while (d_mountpoint(nd->path.dentry)) {
1157 struct mount *mounted;
1158 mounted = __lookup_mnt(nd->path.mnt, nd->path.dentry);
1159 if (!mounted)
1160 break;
1161 nd->path.mnt = &mounted->mnt;
1162 nd->path.dentry = mounted->mnt.mnt_root;
1163 nd->seq = read_seqcount_begin(&nd->path.dentry->d_seq);
1164 if (!read_seqretry(&mount_lock, nd->m_seq))
1165 goto failed;
1166 }
1170 nd->inode = nd->path.dentry->d_inode; 1167 nd->inode = nd->path.dentry->d_inode;
1171 return 0; 1168 return 0;
1172 1169
@@ -1884,7 +1881,7 @@ static int path_init(int dfd, const char *name, unsigned int flags,
1884 1881
1885 nd->path = f.file->f_path; 1882 nd->path = f.file->f_path;
1886 if (flags & LOOKUP_RCU) { 1883 if (flags & LOOKUP_RCU) {
1887 if (f.need_put) 1884 if (f.flags & FDPUT_FPUT)
1888 *fp = f.file; 1885 *fp = f.file;
1889 nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq); 1886 nd->seq = __read_seqcount_begin(&nd->path.dentry->d_seq);
1890 rcu_read_lock(); 1887 rcu_read_lock();
diff --git a/fs/namespace.c b/fs/namespace.c
index 22e536705c45..2ffc5a2905d4 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -23,11 +23,34 @@
23#include <linux/uaccess.h> 23#include <linux/uaccess.h>
24#include <linux/proc_ns.h> 24#include <linux/proc_ns.h>
25#include <linux/magic.h> 25#include <linux/magic.h>
26#include <linux/bootmem.h>
26#include "pnode.h" 27#include "pnode.h"
27#include "internal.h" 28#include "internal.h"
28 29
29#define HASH_SHIFT ilog2(PAGE_SIZE / sizeof(struct list_head)) 30static unsigned int m_hash_mask __read_mostly;
30#define HASH_SIZE (1UL << HASH_SHIFT) 31static unsigned int m_hash_shift __read_mostly;
32static unsigned int mp_hash_mask __read_mostly;
33static unsigned int mp_hash_shift __read_mostly;
34
35static __initdata unsigned long mhash_entries;
36static int __init set_mhash_entries(char *str)
37{
38 if (!str)
39 return 0;
40 mhash_entries = simple_strtoul(str, &str, 0);
41 return 1;
42}
43__setup("mhash_entries=", set_mhash_entries);
44
45static __initdata unsigned long mphash_entries;
46static int __init set_mphash_entries(char *str)
47{
48 if (!str)
49 return 0;
50 mphash_entries = simple_strtoul(str, &str, 0);
51 return 1;
52}
53__setup("mphash_entries=", set_mphash_entries);
31 54
32static int event; 55static int event;
33static DEFINE_IDA(mnt_id_ida); 56static DEFINE_IDA(mnt_id_ida);
@@ -36,8 +59,8 @@ static DEFINE_SPINLOCK(mnt_id_lock);
36static int mnt_id_start = 0; 59static int mnt_id_start = 0;
37static int mnt_group_start = 1; 60static int mnt_group_start = 1;
38 61
39static struct list_head *mount_hashtable __read_mostly; 62static struct hlist_head *mount_hashtable __read_mostly;
40static struct list_head *mountpoint_hashtable __read_mostly; 63static struct hlist_head *mountpoint_hashtable __read_mostly;
41static struct kmem_cache *mnt_cache __read_mostly; 64static struct kmem_cache *mnt_cache __read_mostly;
42static DECLARE_RWSEM(namespace_sem); 65static DECLARE_RWSEM(namespace_sem);
43 66
@@ -55,12 +78,19 @@ EXPORT_SYMBOL_GPL(fs_kobj);
55 */ 78 */
56__cacheline_aligned_in_smp DEFINE_SEQLOCK(mount_lock); 79__cacheline_aligned_in_smp DEFINE_SEQLOCK(mount_lock);
57 80
58static inline unsigned long hash(struct vfsmount *mnt, struct dentry *dentry) 81static inline struct hlist_head *m_hash(struct vfsmount *mnt, struct dentry *dentry)
59{ 82{
60 unsigned long tmp = ((unsigned long)mnt / L1_CACHE_BYTES); 83 unsigned long tmp = ((unsigned long)mnt / L1_CACHE_BYTES);
61 tmp += ((unsigned long)dentry / L1_CACHE_BYTES); 84 tmp += ((unsigned long)dentry / L1_CACHE_BYTES);
62 tmp = tmp + (tmp >> HASH_SHIFT); 85 tmp = tmp + (tmp >> m_hash_shift);
63 return tmp & (HASH_SIZE - 1); 86 return &mount_hashtable[tmp & m_hash_mask];
87}
88
89static inline struct hlist_head *mp_hash(struct dentry *dentry)
90{
91 unsigned long tmp = ((unsigned long)dentry / L1_CACHE_BYTES);
92 tmp = tmp + (tmp >> mp_hash_shift);
93 return &mountpoint_hashtable[tmp & mp_hash_mask];
64} 94}
65 95
66/* 96/*
@@ -187,7 +217,7 @@ static struct mount *alloc_vfsmnt(const char *name)
187 mnt->mnt_writers = 0; 217 mnt->mnt_writers = 0;
188#endif 218#endif
189 219
190 INIT_LIST_HEAD(&mnt->mnt_hash); 220 INIT_HLIST_NODE(&mnt->mnt_hash);
191 INIT_LIST_HEAD(&mnt->mnt_child); 221 INIT_LIST_HEAD(&mnt->mnt_child);
192 INIT_LIST_HEAD(&mnt->mnt_mounts); 222 INIT_LIST_HEAD(&mnt->mnt_mounts);
193 INIT_LIST_HEAD(&mnt->mnt_list); 223 INIT_LIST_HEAD(&mnt->mnt_list);
@@ -575,10 +605,10 @@ bool legitimize_mnt(struct vfsmount *bastard, unsigned seq)
575 */ 605 */
576struct mount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry) 606struct mount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry)
577{ 607{
578 struct list_head *head = mount_hashtable + hash(mnt, dentry); 608 struct hlist_head *head = m_hash(mnt, dentry);
579 struct mount *p; 609 struct mount *p;
580 610
581 list_for_each_entry_rcu(p, head, mnt_hash) 611 hlist_for_each_entry_rcu(p, head, mnt_hash)
582 if (&p->mnt_parent->mnt == mnt && p->mnt_mountpoint == dentry) 612 if (&p->mnt_parent->mnt == mnt && p->mnt_mountpoint == dentry)
583 return p; 613 return p;
584 return NULL; 614 return NULL;
@@ -590,13 +620,17 @@ struct mount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry)
590 */ 620 */
591struct mount *__lookup_mnt_last(struct vfsmount *mnt, struct dentry *dentry) 621struct mount *__lookup_mnt_last(struct vfsmount *mnt, struct dentry *dentry)
592{ 622{
593 struct list_head *head = mount_hashtable + hash(mnt, dentry); 623 struct mount *p, *res;
594 struct mount *p; 624 res = p = __lookup_mnt(mnt, dentry);
595 625 if (!p)
596 list_for_each_entry_reverse(p, head, mnt_hash) 626 goto out;
597 if (&p->mnt_parent->mnt == mnt && p->mnt_mountpoint == dentry) 627 hlist_for_each_entry_continue(p, mnt_hash) {
598 return p; 628 if (&p->mnt_parent->mnt != mnt || p->mnt_mountpoint != dentry)
599 return NULL; 629 break;
630 res = p;
631 }
632out:
633 return res;
600} 634}
601 635
602/* 636/*
@@ -633,11 +667,11 @@ struct vfsmount *lookup_mnt(struct path *path)
633 667
634static struct mountpoint *new_mountpoint(struct dentry *dentry) 668static struct mountpoint *new_mountpoint(struct dentry *dentry)
635{ 669{
636 struct list_head *chain = mountpoint_hashtable + hash(NULL, dentry); 670 struct hlist_head *chain = mp_hash(dentry);
637 struct mountpoint *mp; 671 struct mountpoint *mp;
638 int ret; 672 int ret;
639 673
640 list_for_each_entry(mp, chain, m_hash) { 674 hlist_for_each_entry(mp, chain, m_hash) {
641 if (mp->m_dentry == dentry) { 675 if (mp->m_dentry == dentry) {
642 /* might be worth a WARN_ON() */ 676 /* might be worth a WARN_ON() */
643 if (d_unlinked(dentry)) 677 if (d_unlinked(dentry))
@@ -659,7 +693,7 @@ static struct mountpoint *new_mountpoint(struct dentry *dentry)
659 693
660 mp->m_dentry = dentry; 694 mp->m_dentry = dentry;
661 mp->m_count = 1; 695 mp->m_count = 1;
662 list_add(&mp->m_hash, chain); 696 hlist_add_head(&mp->m_hash, chain);
663 return mp; 697 return mp;
664} 698}
665 699
@@ -670,7 +704,7 @@ static void put_mountpoint(struct mountpoint *mp)
670 spin_lock(&dentry->d_lock); 704 spin_lock(&dentry->d_lock);
671 dentry->d_flags &= ~DCACHE_MOUNTED; 705 dentry->d_flags &= ~DCACHE_MOUNTED;
672 spin_unlock(&dentry->d_lock); 706 spin_unlock(&dentry->d_lock);
673 list_del(&mp->m_hash); 707 hlist_del(&mp->m_hash);
674 kfree(mp); 708 kfree(mp);
675 } 709 }
676} 710}
@@ -712,7 +746,7 @@ static void detach_mnt(struct mount *mnt, struct path *old_path)
712 mnt->mnt_parent = mnt; 746 mnt->mnt_parent = mnt;
713 mnt->mnt_mountpoint = mnt->mnt.mnt_root; 747 mnt->mnt_mountpoint = mnt->mnt.mnt_root;
714 list_del_init(&mnt->mnt_child); 748 list_del_init(&mnt->mnt_child);
715 list_del_init(&mnt->mnt_hash); 749 hlist_del_init_rcu(&mnt->mnt_hash);
716 put_mountpoint(mnt->mnt_mp); 750 put_mountpoint(mnt->mnt_mp);
717 mnt->mnt_mp = NULL; 751 mnt->mnt_mp = NULL;
718} 752}
@@ -739,15 +773,14 @@ static void attach_mnt(struct mount *mnt,
739 struct mountpoint *mp) 773 struct mountpoint *mp)
740{ 774{
741 mnt_set_mountpoint(parent, mp, mnt); 775 mnt_set_mountpoint(parent, mp, mnt);
742 list_add_tail(&mnt->mnt_hash, mount_hashtable + 776 hlist_add_head_rcu(&mnt->mnt_hash, m_hash(&parent->mnt, mp->m_dentry));
743 hash(&parent->mnt, mp->m_dentry));
744 list_add_tail(&mnt->mnt_child, &parent->mnt_mounts); 777 list_add_tail(&mnt->mnt_child, &parent->mnt_mounts);
745} 778}
746 779
747/* 780/*
748 * vfsmount lock must be held for write 781 * vfsmount lock must be held for write
749 */ 782 */
750static void commit_tree(struct mount *mnt) 783static void commit_tree(struct mount *mnt, struct mount *shadows)
751{ 784{
752 struct mount *parent = mnt->mnt_parent; 785 struct mount *parent = mnt->mnt_parent;
753 struct mount *m; 786 struct mount *m;
@@ -762,8 +795,11 @@ static void commit_tree(struct mount *mnt)
762 795
763 list_splice(&head, n->list.prev); 796 list_splice(&head, n->list.prev);
764 797
765 list_add_tail(&mnt->mnt_hash, mount_hashtable + 798 if (shadows)
766 hash(&parent->mnt, mnt->mnt_mountpoint)); 799 hlist_add_after_rcu(&shadows->mnt_hash, &mnt->mnt_hash);
800 else
801 hlist_add_head_rcu(&mnt->mnt_hash,
802 m_hash(&parent->mnt, mnt->mnt_mountpoint));
767 list_add_tail(&mnt->mnt_child, &parent->mnt_mounts); 803 list_add_tail(&mnt->mnt_child, &parent->mnt_mounts);
768 touch_mnt_namespace(n); 804 touch_mnt_namespace(n);
769} 805}
@@ -1153,26 +1189,28 @@ int may_umount(struct vfsmount *mnt)
1153 1189
1154EXPORT_SYMBOL(may_umount); 1190EXPORT_SYMBOL(may_umount);
1155 1191
1156static LIST_HEAD(unmounted); /* protected by namespace_sem */ 1192static HLIST_HEAD(unmounted); /* protected by namespace_sem */
1157 1193
1158static void namespace_unlock(void) 1194static void namespace_unlock(void)
1159{ 1195{
1160 struct mount *mnt; 1196 struct mount *mnt;
1161 LIST_HEAD(head); 1197 struct hlist_head head = unmounted;
1162 1198
1163 if (likely(list_empty(&unmounted))) { 1199 if (likely(hlist_empty(&head))) {
1164 up_write(&namespace_sem); 1200 up_write(&namespace_sem);
1165 return; 1201 return;
1166 } 1202 }
1167 1203
1168 list_splice_init(&unmounted, &head); 1204 head.first->pprev = &head.first;
1205 INIT_HLIST_HEAD(&unmounted);
1206
1169 up_write(&namespace_sem); 1207 up_write(&namespace_sem);
1170 1208
1171 synchronize_rcu(); 1209 synchronize_rcu();
1172 1210
1173 while (!list_empty(&head)) { 1211 while (!hlist_empty(&head)) {
1174 mnt = list_first_entry(&head, struct mount, mnt_hash); 1212 mnt = hlist_entry(head.first, struct mount, mnt_hash);
1175 list_del_init(&mnt->mnt_hash); 1213 hlist_del_init(&mnt->mnt_hash);
1176 if (mnt->mnt_ex_mountpoint.mnt) 1214 if (mnt->mnt_ex_mountpoint.mnt)
1177 path_put(&mnt->mnt_ex_mountpoint); 1215 path_put(&mnt->mnt_ex_mountpoint);
1178 mntput(&mnt->mnt); 1216 mntput(&mnt->mnt);
@@ -1193,16 +1231,19 @@ static inline void namespace_lock(void)
1193 */ 1231 */
1194void umount_tree(struct mount *mnt, int how) 1232void umount_tree(struct mount *mnt, int how)
1195{ 1233{
1196 LIST_HEAD(tmp_list); 1234 HLIST_HEAD(tmp_list);
1197 struct mount *p; 1235 struct mount *p;
1236 struct mount *last = NULL;
1198 1237
1199 for (p = mnt; p; p = next_mnt(p, mnt)) 1238 for (p = mnt; p; p = next_mnt(p, mnt)) {
1200 list_move(&p->mnt_hash, &tmp_list); 1239 hlist_del_init_rcu(&p->mnt_hash);
1240 hlist_add_head(&p->mnt_hash, &tmp_list);
1241 }
1201 1242
1202 if (how) 1243 if (how)
1203 propagate_umount(&tmp_list); 1244 propagate_umount(&tmp_list);
1204 1245
1205 list_for_each_entry(p, &tmp_list, mnt_hash) { 1246 hlist_for_each_entry(p, &tmp_list, mnt_hash) {
1206 list_del_init(&p->mnt_expire); 1247 list_del_init(&p->mnt_expire);
1207 list_del_init(&p->mnt_list); 1248 list_del_init(&p->mnt_list);
1208 __touch_mnt_namespace(p->mnt_ns); 1249 __touch_mnt_namespace(p->mnt_ns);
@@ -1220,8 +1261,13 @@ void umount_tree(struct mount *mnt, int how)
1220 p->mnt_mp = NULL; 1261 p->mnt_mp = NULL;
1221 } 1262 }
1222 change_mnt_propagation(p, MS_PRIVATE); 1263 change_mnt_propagation(p, MS_PRIVATE);
1264 last = p;
1265 }
1266 if (last) {
1267 last->mnt_hash.next = unmounted.first;
1268 unmounted.first = tmp_list.first;
1269 unmounted.first->pprev = &unmounted.first;
1223 } 1270 }
1224 list_splice(&tmp_list, &unmounted);
1225} 1271}
1226 1272
1227static void shrink_submounts(struct mount *mnt); 1273static void shrink_submounts(struct mount *mnt);
@@ -1605,24 +1651,23 @@ static int attach_recursive_mnt(struct mount *source_mnt,
1605 struct mountpoint *dest_mp, 1651 struct mountpoint *dest_mp,
1606 struct path *parent_path) 1652 struct path *parent_path)
1607{ 1653{
1608 LIST_HEAD(tree_list); 1654 HLIST_HEAD(tree_list);
1609 struct mount *child, *p; 1655 struct mount *child, *p;
1656 struct hlist_node *n;
1610 int err; 1657 int err;
1611 1658
1612 if (IS_MNT_SHARED(dest_mnt)) { 1659 if (IS_MNT_SHARED(dest_mnt)) {
1613 err = invent_group_ids(source_mnt, true); 1660 err = invent_group_ids(source_mnt, true);
1614 if (err) 1661 if (err)
1615 goto out; 1662 goto out;
1616 } 1663 err = propagate_mnt(dest_mnt, dest_mp, source_mnt, &tree_list);
1617 err = propagate_mnt(dest_mnt, dest_mp, source_mnt, &tree_list); 1664 if (err)
1618 if (err) 1665 goto out_cleanup_ids;
1619 goto out_cleanup_ids; 1666 lock_mount_hash();
1620
1621 lock_mount_hash();
1622
1623 if (IS_MNT_SHARED(dest_mnt)) {
1624 for (p = source_mnt; p; p = next_mnt(p, source_mnt)) 1667 for (p = source_mnt; p; p = next_mnt(p, source_mnt))
1625 set_mnt_shared(p); 1668 set_mnt_shared(p);
1669 } else {
1670 lock_mount_hash();
1626 } 1671 }
1627 if (parent_path) { 1672 if (parent_path) {
1628 detach_mnt(source_mnt, parent_path); 1673 detach_mnt(source_mnt, parent_path);
@@ -1630,20 +1675,22 @@ static int attach_recursive_mnt(struct mount *source_mnt,
1630 touch_mnt_namespace(source_mnt->mnt_ns); 1675 touch_mnt_namespace(source_mnt->mnt_ns);
1631 } else { 1676 } else {
1632 mnt_set_mountpoint(dest_mnt, dest_mp, source_mnt); 1677 mnt_set_mountpoint(dest_mnt, dest_mp, source_mnt);
1633 commit_tree(source_mnt); 1678 commit_tree(source_mnt, NULL);
1634 } 1679 }
1635 1680
1636 list_for_each_entry_safe(child, p, &tree_list, mnt_hash) { 1681 hlist_for_each_entry_safe(child, n, &tree_list, mnt_hash) {
1637 list_del_init(&child->mnt_hash); 1682 struct mount *q;
1638 commit_tree(child); 1683 hlist_del_init(&child->mnt_hash);
1684 q = __lookup_mnt_last(&child->mnt_parent->mnt,
1685 child->mnt_mountpoint);
1686 commit_tree(child, q);
1639 } 1687 }
1640 unlock_mount_hash(); 1688 unlock_mount_hash();
1641 1689
1642 return 0; 1690 return 0;
1643 1691
1644 out_cleanup_ids: 1692 out_cleanup_ids:
1645 if (IS_MNT_SHARED(dest_mnt)) 1693 cleanup_group_ids(source_mnt, NULL);
1646 cleanup_group_ids(source_mnt, NULL);
1647 out: 1694 out:
1648 return err; 1695 return err;
1649} 1696}
@@ -2777,18 +2824,24 @@ void __init mnt_init(void)
2777 mnt_cache = kmem_cache_create("mnt_cache", sizeof(struct mount), 2824 mnt_cache = kmem_cache_create("mnt_cache", sizeof(struct mount),
2778 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, NULL); 2825 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, NULL);
2779 2826
2780 mount_hashtable = (struct list_head *)__get_free_page(GFP_ATOMIC); 2827 mount_hashtable = alloc_large_system_hash("Mount-cache",
2781 mountpoint_hashtable = (struct list_head *)__get_free_page(GFP_ATOMIC); 2828 sizeof(struct hlist_head),
2829 mhash_entries, 19,
2830 0,
2831 &m_hash_shift, &m_hash_mask, 0, 0);
2832 mountpoint_hashtable = alloc_large_system_hash("Mountpoint-cache",
2833 sizeof(struct hlist_head),
2834 mphash_entries, 19,
2835 0,
2836 &mp_hash_shift, &mp_hash_mask, 0, 0);
2782 2837
2783 if (!mount_hashtable || !mountpoint_hashtable) 2838 if (!mount_hashtable || !mountpoint_hashtable)
2784 panic("Failed to allocate mount hash table\n"); 2839 panic("Failed to allocate mount hash table\n");
2785 2840
2786 printk(KERN_INFO "Mount-cache hash table entries: %lu\n", HASH_SIZE); 2841 for (u = 0; u <= m_hash_mask; u++)
2787 2842 INIT_HLIST_HEAD(&mount_hashtable[u]);
2788 for (u = 0; u < HASH_SIZE; u++) 2843 for (u = 0; u <= mp_hash_mask; u++)
2789 INIT_LIST_HEAD(&mount_hashtable[u]); 2844 INIT_HLIST_HEAD(&mountpoint_hashtable[u]);
2790 for (u = 0; u < HASH_SIZE; u++)
2791 INIT_LIST_HEAD(&mountpoint_hashtable[u]);
2792 2845
2793 kernfs_init(); 2846 kernfs_init();
2794 2847
diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c
index ef792f29f831..5d8ccecf5f5c 100644
--- a/fs/nfs/delegation.c
+++ b/fs/nfs/delegation.c
@@ -659,16 +659,19 @@ int nfs_async_inode_return_delegation(struct inode *inode,
659 659
660 rcu_read_lock(); 660 rcu_read_lock();
661 delegation = rcu_dereference(NFS_I(inode)->delegation); 661 delegation = rcu_dereference(NFS_I(inode)->delegation);
662 if (delegation == NULL)
663 goto out_enoent;
662 664
663 if (!clp->cl_mvops->match_stateid(&delegation->stateid, stateid)) { 665 if (!clp->cl_mvops->match_stateid(&delegation->stateid, stateid))
664 rcu_read_unlock(); 666 goto out_enoent;
665 return -ENOENT;
666 }
667 nfs_mark_return_delegation(server, delegation); 667 nfs_mark_return_delegation(server, delegation);
668 rcu_read_unlock(); 668 rcu_read_unlock();
669 669
670 nfs_delegation_run_state_manager(clp); 670 nfs_delegation_run_state_manager(clp);
671 return 0; 671 return 0;
672out_enoent:
673 rcu_read_unlock();
674 return -ENOENT;
672} 675}
673 676
674static struct inode * 677static struct inode *
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 28a0a3cbd3b7..360114ae8b82 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -164,17 +164,16 @@ static void nfs_zap_caches_locked(struct inode *inode)
164 if (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)) { 164 if (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)) {
165 nfs_fscache_invalidate(inode); 165 nfs_fscache_invalidate(inode);
166 nfsi->cache_validity |= NFS_INO_INVALID_ATTR 166 nfsi->cache_validity |= NFS_INO_INVALID_ATTR
167 | NFS_INO_INVALID_LABEL
168 | NFS_INO_INVALID_DATA 167 | NFS_INO_INVALID_DATA
169 | NFS_INO_INVALID_ACCESS 168 | NFS_INO_INVALID_ACCESS
170 | NFS_INO_INVALID_ACL 169 | NFS_INO_INVALID_ACL
171 | NFS_INO_REVAL_PAGECACHE; 170 | NFS_INO_REVAL_PAGECACHE;
172 } else 171 } else
173 nfsi->cache_validity |= NFS_INO_INVALID_ATTR 172 nfsi->cache_validity |= NFS_INO_INVALID_ATTR
174 | NFS_INO_INVALID_LABEL
175 | NFS_INO_INVALID_ACCESS 173 | NFS_INO_INVALID_ACCESS
176 | NFS_INO_INVALID_ACL 174 | NFS_INO_INVALID_ACL
177 | NFS_INO_REVAL_PAGECACHE; 175 | NFS_INO_REVAL_PAGECACHE;
176 nfs_zap_label_cache_locked(nfsi);
178} 177}
179 178
180void nfs_zap_caches(struct inode *inode) 179void nfs_zap_caches(struct inode *inode)
@@ -266,6 +265,13 @@ nfs_init_locked(struct inode *inode, void *opaque)
266} 265}
267 266
268#ifdef CONFIG_NFS_V4_SECURITY_LABEL 267#ifdef CONFIG_NFS_V4_SECURITY_LABEL
268static void nfs_clear_label_invalid(struct inode *inode)
269{
270 spin_lock(&inode->i_lock);
271 NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_LABEL;
272 spin_unlock(&inode->i_lock);
273}
274
269void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr, 275void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr,
270 struct nfs4_label *label) 276 struct nfs4_label *label)
271{ 277{
@@ -283,6 +289,7 @@ void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr,
283 __func__, 289 __func__,
284 (char *)label->label, 290 (char *)label->label,
285 label->len, error); 291 label->len, error);
292 nfs_clear_label_invalid(inode);
286 } 293 }
287} 294}
288 295
@@ -1648,7 +1655,7 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
1648 inode->i_blocks = fattr->du.nfs2.blocks; 1655 inode->i_blocks = fattr->du.nfs2.blocks;
1649 1656
1650 /* Update attrtimeo value if we're out of the unstable period */ 1657 /* Update attrtimeo value if we're out of the unstable period */
1651 if (invalid & (NFS_INO_INVALID_ATTR|NFS_INO_INVALID_LABEL)) { 1658 if (invalid & NFS_INO_INVALID_ATTR) {
1652 nfs_inc_stats(inode, NFSIOS_ATTRINVALIDATE); 1659 nfs_inc_stats(inode, NFSIOS_ATTRINVALIDATE);
1653 nfsi->attrtimeo = NFS_MINATTRTIMEO(inode); 1660 nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
1654 nfsi->attrtimeo_timestamp = now; 1661 nfsi->attrtimeo_timestamp = now;
@@ -1661,7 +1668,6 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
1661 } 1668 }
1662 } 1669 }
1663 invalid &= ~NFS_INO_INVALID_ATTR; 1670 invalid &= ~NFS_INO_INVALID_ATTR;
1664 invalid &= ~NFS_INO_INVALID_LABEL;
1665 /* Don't invalidate the data if we were to blame */ 1671 /* Don't invalidate the data if we were to blame */
1666 if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) 1672 if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode)
1667 || S_ISLNK(inode->i_mode))) 1673 || S_ISLNK(inode->i_mode)))
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index 8b5cc04a8611..b46cf5a67329 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -176,7 +176,8 @@ extern struct nfs_server *nfs4_create_server(
176extern struct nfs_server *nfs4_create_referral_server(struct nfs_clone_mount *, 176extern struct nfs_server *nfs4_create_referral_server(struct nfs_clone_mount *,
177 struct nfs_fh *); 177 struct nfs_fh *);
178extern int nfs4_update_server(struct nfs_server *server, const char *hostname, 178extern int nfs4_update_server(struct nfs_server *server, const char *hostname,
179 struct sockaddr *sap, size_t salen); 179 struct sockaddr *sap, size_t salen,
180 struct net *net);
180extern void nfs_free_server(struct nfs_server *server); 181extern void nfs_free_server(struct nfs_server *server);
181extern struct nfs_server *nfs_clone_server(struct nfs_server *, 182extern struct nfs_server *nfs_clone_server(struct nfs_server *,
182 struct nfs_fh *, 183 struct nfs_fh *,
@@ -279,9 +280,18 @@ static inline void nfs4_label_free(struct nfs4_label *label)
279 } 280 }
280 return; 281 return;
281} 282}
283
284static inline void nfs_zap_label_cache_locked(struct nfs_inode *nfsi)
285{
286 if (nfs_server_capable(&nfsi->vfs_inode, NFS_CAP_SECURITY_LABEL))
287 nfsi->cache_validity |= NFS_INO_INVALID_LABEL;
288}
282#else 289#else
283static inline struct nfs4_label *nfs4_label_alloc(struct nfs_server *server, gfp_t flags) { return NULL; } 290static inline struct nfs4_label *nfs4_label_alloc(struct nfs_server *server, gfp_t flags) { return NULL; }
284static inline void nfs4_label_free(void *label) {} 291static inline void nfs4_label_free(void *label) {}
292static inline void nfs_zap_label_cache_locked(struct nfs_inode *nfsi)
293{
294}
285#endif /* CONFIG_NFS_V4_SECURITY_LABEL */ 295#endif /* CONFIG_NFS_V4_SECURITY_LABEL */
286 296
287/* proc.c */ 297/* proc.c */
diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c
index aa9bc973f36a..a462ef0fb5d6 100644
--- a/fs/nfs/nfs3proc.c
+++ b/fs/nfs/nfs3proc.c
@@ -18,6 +18,7 @@
18#include <linux/lockd/bind.h> 18#include <linux/lockd/bind.h>
19#include <linux/nfs_mount.h> 19#include <linux/nfs_mount.h>
20#include <linux/freezer.h> 20#include <linux/freezer.h>
21#include <linux/xattr.h>
21 22
22#include "iostat.h" 23#include "iostat.h"
23#include "internal.h" 24#include "internal.h"
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index 860ad26a5590..0e46d3d1b6cc 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -1135,6 +1135,7 @@ static int nfs_probe_destination(struct nfs_server *server)
1135 * @hostname: new end-point's hostname 1135 * @hostname: new end-point's hostname
1136 * @sap: new end-point's socket address 1136 * @sap: new end-point's socket address
1137 * @salen: size of "sap" 1137 * @salen: size of "sap"
1138 * @net: net namespace
1138 * 1139 *
1139 * The nfs_server must be quiescent before this function is invoked. 1140 * The nfs_server must be quiescent before this function is invoked.
1140 * Either its session is drained (NFSv4.1+), or its transport is 1141 * Either its session is drained (NFSv4.1+), or its transport is
@@ -1143,13 +1144,13 @@ static int nfs_probe_destination(struct nfs_server *server)
1143 * Returns zero on success, or a negative errno value. 1144 * Returns zero on success, or a negative errno value.
1144 */ 1145 */
1145int nfs4_update_server(struct nfs_server *server, const char *hostname, 1146int nfs4_update_server(struct nfs_server *server, const char *hostname,
1146 struct sockaddr *sap, size_t salen) 1147 struct sockaddr *sap, size_t salen, struct net *net)
1147{ 1148{
1148 struct nfs_client *clp = server->nfs_client; 1149 struct nfs_client *clp = server->nfs_client;
1149 struct rpc_clnt *clnt = server->client; 1150 struct rpc_clnt *clnt = server->client;
1150 struct xprt_create xargs = { 1151 struct xprt_create xargs = {
1151 .ident = clp->cl_proto, 1152 .ident = clp->cl_proto,
1152 .net = &init_net, 1153 .net = net,
1153 .dstaddr = sap, 1154 .dstaddr = sap,
1154 .addrlen = salen, 1155 .addrlen = salen,
1155 .servername = hostname, 1156 .servername = hostname,
@@ -1189,7 +1190,7 @@ int nfs4_update_server(struct nfs_server *server, const char *hostname,
1189 error = nfs4_set_client(server, hostname, sap, salen, buf, 1190 error = nfs4_set_client(server, hostname, sap, salen, buf,
1190 clp->cl_rpcclient->cl_auth->au_flavor, 1191 clp->cl_rpcclient->cl_auth->au_flavor,
1191 clp->cl_proto, clnt->cl_timeout, 1192 clp->cl_proto, clnt->cl_timeout,
1192 clp->cl_minorversion, clp->cl_net); 1193 clp->cl_minorversion, net);
1193 nfs_put_client(clp); 1194 nfs_put_client(clp);
1194 if (error != 0) { 1195 if (error != 0) {
1195 nfs_server_insert_lists(server); 1196 nfs_server_insert_lists(server);
diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c
index 12c8132ad408..b9a35c05b60f 100644
--- a/fs/nfs/nfs4filelayout.c
+++ b/fs/nfs/nfs4filelayout.c
@@ -324,8 +324,9 @@ static void filelayout_read_prepare(struct rpc_task *task, void *data)
324 &rdata->res.seq_res, 324 &rdata->res.seq_res,
325 task)) 325 task))
326 return; 326 return;
327 nfs4_set_rw_stateid(&rdata->args.stateid, rdata->args.context, 327 if (nfs4_set_rw_stateid(&rdata->args.stateid, rdata->args.context,
328 rdata->args.lock_context, FMODE_READ); 328 rdata->args.lock_context, FMODE_READ) == -EIO)
329 rpc_exit(task, -EIO); /* lost lock, terminate I/O */
329} 330}
330 331
331static void filelayout_read_call_done(struct rpc_task *task, void *data) 332static void filelayout_read_call_done(struct rpc_task *task, void *data)
@@ -435,8 +436,9 @@ static void filelayout_write_prepare(struct rpc_task *task, void *data)
435 &wdata->res.seq_res, 436 &wdata->res.seq_res,
436 task)) 437 task))
437 return; 438 return;
438 nfs4_set_rw_stateid(&wdata->args.stateid, wdata->args.context, 439 if (nfs4_set_rw_stateid(&wdata->args.stateid, wdata->args.context,
439 wdata->args.lock_context, FMODE_WRITE); 440 wdata->args.lock_context, FMODE_WRITE) == -EIO)
441 rpc_exit(task, -EIO); /* lost lock, terminate I/O */
440} 442}
441 443
442static void filelayout_write_call_done(struct rpc_task *task, void *data) 444static void filelayout_write_call_done(struct rpc_task *task, void *data)
diff --git a/fs/nfs/nfs4namespace.c b/fs/nfs/nfs4namespace.c
index 4e7f05d3e9db..3d5dbf80d46a 100644
--- a/fs/nfs/nfs4namespace.c
+++ b/fs/nfs/nfs4namespace.c
@@ -121,9 +121,8 @@ static int nfs4_validate_fspath(struct dentry *dentry,
121} 121}
122 122
123static size_t nfs_parse_server_name(char *string, size_t len, 123static size_t nfs_parse_server_name(char *string, size_t len,
124 struct sockaddr *sa, size_t salen, struct nfs_server *server) 124 struct sockaddr *sa, size_t salen, struct net *net)
125{ 125{
126 struct net *net = rpc_net_ns(server->client);
127 ssize_t ret; 126 ssize_t ret;
128 127
129 ret = rpc_pton(net, string, len, sa, salen); 128 ret = rpc_pton(net, string, len, sa, salen);
@@ -223,6 +222,7 @@ static struct vfsmount *try_location(struct nfs_clone_mount *mountdata,
223 const struct nfs4_fs_location *location) 222 const struct nfs4_fs_location *location)
224{ 223{
225 const size_t addr_bufsize = sizeof(struct sockaddr_storage); 224 const size_t addr_bufsize = sizeof(struct sockaddr_storage);
225 struct net *net = rpc_net_ns(NFS_SB(mountdata->sb)->client);
226 struct vfsmount *mnt = ERR_PTR(-ENOENT); 226 struct vfsmount *mnt = ERR_PTR(-ENOENT);
227 char *mnt_path; 227 char *mnt_path;
228 unsigned int maxbuflen; 228 unsigned int maxbuflen;
@@ -248,8 +248,7 @@ static struct vfsmount *try_location(struct nfs_clone_mount *mountdata,
248 continue; 248 continue;
249 249
250 mountdata->addrlen = nfs_parse_server_name(buf->data, buf->len, 250 mountdata->addrlen = nfs_parse_server_name(buf->data, buf->len,
251 mountdata->addr, addr_bufsize, 251 mountdata->addr, addr_bufsize, net);
252 NFS_SB(mountdata->sb));
253 if (mountdata->addrlen == 0) 252 if (mountdata->addrlen == 0)
254 continue; 253 continue;
255 254
@@ -419,6 +418,7 @@ static int nfs4_try_replacing_one_location(struct nfs_server *server,
419 const struct nfs4_fs_location *location) 418 const struct nfs4_fs_location *location)
420{ 419{
421 const size_t addr_bufsize = sizeof(struct sockaddr_storage); 420 const size_t addr_bufsize = sizeof(struct sockaddr_storage);
421 struct net *net = rpc_net_ns(server->client);
422 struct sockaddr *sap; 422 struct sockaddr *sap;
423 unsigned int s; 423 unsigned int s;
424 size_t salen; 424 size_t salen;
@@ -440,7 +440,7 @@ static int nfs4_try_replacing_one_location(struct nfs_server *server,
440 continue; 440 continue;
441 441
442 salen = nfs_parse_server_name(buf->data, buf->len, 442 salen = nfs_parse_server_name(buf->data, buf->len,
443 sap, addr_bufsize, server); 443 sap, addr_bufsize, net);
444 if (salen == 0) 444 if (salen == 0)
445 continue; 445 continue;
446 rpc_set_port(sap, NFS_PORT); 446 rpc_set_port(sap, NFS_PORT);
@@ -450,7 +450,7 @@ static int nfs4_try_replacing_one_location(struct nfs_server *server,
450 if (hostname == NULL) 450 if (hostname == NULL)
451 break; 451 break;
452 452
453 error = nfs4_update_server(server, hostname, sap, salen); 453 error = nfs4_update_server(server, hostname, sap, salen, net);
454 kfree(hostname); 454 kfree(hostname);
455 if (error == 0) 455 if (error == 0)
456 break; 456 break;
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 2da6a698b8f7..450bfedbe2f4 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -2398,13 +2398,16 @@ static int _nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
2398 2398
2399 if (nfs4_copy_delegation_stateid(&arg.stateid, inode, fmode)) { 2399 if (nfs4_copy_delegation_stateid(&arg.stateid, inode, fmode)) {
2400 /* Use that stateid */ 2400 /* Use that stateid */
2401 } else if (truncate && state != NULL && nfs4_valid_open_stateid(state)) { 2401 } else if (truncate && state != NULL) {
2402 struct nfs_lockowner lockowner = { 2402 struct nfs_lockowner lockowner = {
2403 .l_owner = current->files, 2403 .l_owner = current->files,
2404 .l_pid = current->tgid, 2404 .l_pid = current->tgid,
2405 }; 2405 };
2406 nfs4_select_rw_stateid(&arg.stateid, state, FMODE_WRITE, 2406 if (!nfs4_valid_open_stateid(state))
2407 &lockowner); 2407 return -EBADF;
2408 if (nfs4_select_rw_stateid(&arg.stateid, state, FMODE_WRITE,
2409 &lockowner) == -EIO)
2410 return -EBADF;
2408 } else 2411 } else
2409 nfs4_stateid_copy(&arg.stateid, &zero_stateid); 2412 nfs4_stateid_copy(&arg.stateid, &zero_stateid);
2410 2413
@@ -4011,8 +4014,9 @@ static bool nfs4_stateid_is_current(nfs4_stateid *stateid,
4011{ 4014{
4012 nfs4_stateid current_stateid; 4015 nfs4_stateid current_stateid;
4013 4016
4014 if (nfs4_set_rw_stateid(&current_stateid, ctx, l_ctx, fmode)) 4017 /* If the current stateid represents a lost lock, then exit */
4015 return false; 4018 if (nfs4_set_rw_stateid(&current_stateid, ctx, l_ctx, fmode) == -EIO)
4019 return true;
4016 return nfs4_stateid_match(stateid, &current_stateid); 4020 return nfs4_stateid_match(stateid, &current_stateid);
4017} 4021}
4018 4022
@@ -5828,8 +5832,7 @@ struct nfs_release_lockowner_data {
5828 struct nfs4_lock_state *lsp; 5832 struct nfs4_lock_state *lsp;
5829 struct nfs_server *server; 5833 struct nfs_server *server;
5830 struct nfs_release_lockowner_args args; 5834 struct nfs_release_lockowner_args args;
5831 struct nfs4_sequence_args seq_args; 5835 struct nfs_release_lockowner_res res;
5832 struct nfs4_sequence_res seq_res;
5833 unsigned long timestamp; 5836 unsigned long timestamp;
5834}; 5837};
5835 5838
@@ -5837,7 +5840,7 @@ static void nfs4_release_lockowner_prepare(struct rpc_task *task, void *calldata
5837{ 5840{
5838 struct nfs_release_lockowner_data *data = calldata; 5841 struct nfs_release_lockowner_data *data = calldata;
5839 nfs40_setup_sequence(data->server, 5842 nfs40_setup_sequence(data->server,
5840 &data->seq_args, &data->seq_res, task); 5843 &data->args.seq_args, &data->res.seq_res, task);
5841 data->timestamp = jiffies; 5844 data->timestamp = jiffies;
5842} 5845}
5843 5846
@@ -5846,7 +5849,7 @@ static void nfs4_release_lockowner_done(struct rpc_task *task, void *calldata)
5846 struct nfs_release_lockowner_data *data = calldata; 5849 struct nfs_release_lockowner_data *data = calldata;
5847 struct nfs_server *server = data->server; 5850 struct nfs_server *server = data->server;
5848 5851
5849 nfs40_sequence_done(task, &data->seq_res); 5852 nfs40_sequence_done(task, &data->res.seq_res);
5850 5853
5851 switch (task->tk_status) { 5854 switch (task->tk_status) {
5852 case 0: 5855 case 0:
@@ -5887,7 +5890,6 @@ static int nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_st
5887 data = kmalloc(sizeof(*data), GFP_NOFS); 5890 data = kmalloc(sizeof(*data), GFP_NOFS);
5888 if (!data) 5891 if (!data)
5889 return -ENOMEM; 5892 return -ENOMEM;
5890 nfs4_init_sequence(&data->seq_args, &data->seq_res, 0);
5891 data->lsp = lsp; 5893 data->lsp = lsp;
5892 data->server = server; 5894 data->server = server;
5893 data->args.lock_owner.clientid = server->nfs_client->cl_clientid; 5895 data->args.lock_owner.clientid = server->nfs_client->cl_clientid;
@@ -5895,6 +5897,8 @@ static int nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_st
5895 data->args.lock_owner.s_dev = server->s_dev; 5897 data->args.lock_owner.s_dev = server->s_dev;
5896 5898
5897 msg.rpc_argp = &data->args; 5899 msg.rpc_argp = &data->args;
5900 msg.rpc_resp = &data->res;
5901 nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
5898 rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, data); 5902 rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, data);
5899 return 0; 5903 return 0;
5900} 5904}
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index e5be72518bd7..0deb32105ccf 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -974,9 +974,6 @@ static int nfs4_copy_lock_stateid(nfs4_stateid *dst,
974 else if (lsp != NULL && test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0) { 974 else if (lsp != NULL && test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0) {
975 nfs4_stateid_copy(dst, &lsp->ls_stateid); 975 nfs4_stateid_copy(dst, &lsp->ls_stateid);
976 ret = 0; 976 ret = 0;
977 smp_rmb();
978 if (!list_empty(&lsp->ls_seqid.list))
979 ret = -EWOULDBLOCK;
980 } 977 }
981 spin_unlock(&state->state_lock); 978 spin_unlock(&state->state_lock);
982 nfs4_put_lock_state(lsp); 979 nfs4_put_lock_state(lsp);
@@ -984,10 +981,9 @@ out:
984 return ret; 981 return ret;
985} 982}
986 983
987static int nfs4_copy_open_stateid(nfs4_stateid *dst, struct nfs4_state *state) 984static void nfs4_copy_open_stateid(nfs4_stateid *dst, struct nfs4_state *state)
988{ 985{
989 const nfs4_stateid *src; 986 const nfs4_stateid *src;
990 int ret;
991 int seq; 987 int seq;
992 988
993 do { 989 do {
@@ -996,12 +992,7 @@ static int nfs4_copy_open_stateid(nfs4_stateid *dst, struct nfs4_state *state)
996 if (test_bit(NFS_OPEN_STATE, &state->flags)) 992 if (test_bit(NFS_OPEN_STATE, &state->flags))
997 src = &state->open_stateid; 993 src = &state->open_stateid;
998 nfs4_stateid_copy(dst, src); 994 nfs4_stateid_copy(dst, src);
999 ret = 0;
1000 smp_rmb();
1001 if (!list_empty(&state->owner->so_seqid.list))
1002 ret = -EWOULDBLOCK;
1003 } while (read_seqretry(&state->seqlock, seq)); 995 } while (read_seqretry(&state->seqlock, seq));
1004 return ret;
1005} 996}
1006 997
1007/* 998/*
@@ -1015,15 +1006,19 @@ int nfs4_select_rw_stateid(nfs4_stateid *dst, struct nfs4_state *state,
1015 if (ret == -EIO) 1006 if (ret == -EIO)
1016 /* A lost lock - don't even consider delegations */ 1007 /* A lost lock - don't even consider delegations */
1017 goto out; 1008 goto out;
1018 if (nfs4_copy_delegation_stateid(dst, state->inode, fmode)) 1009 /* returns true if delegation stateid found and copied */
1010 if (nfs4_copy_delegation_stateid(dst, state->inode, fmode)) {
1011 ret = 0;
1019 goto out; 1012 goto out;
1013 }
1020 if (ret != -ENOENT) 1014 if (ret != -ENOENT)
1021 /* nfs4_copy_delegation_stateid() didn't over-write 1015 /* nfs4_copy_delegation_stateid() didn't over-write
1022 * dst, so it still has the lock stateid which we now 1016 * dst, so it still has the lock stateid which we now
1023 * choose to use. 1017 * choose to use.
1024 */ 1018 */
1025 goto out; 1019 goto out;
1026 ret = nfs4_copy_open_stateid(dst, state); 1020 nfs4_copy_open_stateid(dst, state);
1021 ret = 0;
1027out: 1022out:
1028 if (nfs_server_capable(state->inode, NFS_CAP_STATEID_NFSV41)) 1023 if (nfs_server_capable(state->inode, NFS_CAP_STATEID_NFSV41))
1029 dst->seqid = 0; 1024 dst->seqid = 0;
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 017d3cb5e99b..6d7be3f80356 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -449,6 +449,7 @@ nfsd_setattr(struct svc_rqst *rqstp, struct svc_fh *fhp, struct iattr *iap,
449 fh_lock(fhp); 449 fh_lock(fhp);
450 host_err = notify_change(dentry, iap, NULL); 450 host_err = notify_change(dentry, iap, NULL);
451 fh_unlock(fhp); 451 fh_unlock(fhp);
452 err = nfserrno(host_err);
452 453
453out_put_write_access: 454out_put_write_access:
454 if (size_change) 455 if (size_change)
diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c
index 0b9ff4395e6a..abc8cbcfe90e 100644
--- a/fs/notify/dnotify/dnotify.c
+++ b/fs/notify/dnotify/dnotify.c
@@ -86,7 +86,7 @@ static int dnotify_handle_event(struct fsnotify_group *group,
86 struct fsnotify_mark *inode_mark, 86 struct fsnotify_mark *inode_mark,
87 struct fsnotify_mark *vfsmount_mark, 87 struct fsnotify_mark *vfsmount_mark,
88 u32 mask, void *data, int data_type, 88 u32 mask, void *data, int data_type,
89 const unsigned char *file_name) 89 const unsigned char *file_name, u32 cookie)
90{ 90{
91 struct dnotify_mark *dn_mark; 91 struct dnotify_mark *dn_mark;
92 struct dnotify_struct *dn; 92 struct dnotify_struct *dn;
diff --git a/fs/notify/fanotify/fanotify.c b/fs/notify/fanotify/fanotify.c
index 0e792f5e3147..dc638f786d5c 100644
--- a/fs/notify/fanotify/fanotify.c
+++ b/fs/notify/fanotify/fanotify.c
@@ -147,7 +147,7 @@ static int fanotify_handle_event(struct fsnotify_group *group,
147 struct fsnotify_mark *inode_mark, 147 struct fsnotify_mark *inode_mark,
148 struct fsnotify_mark *fanotify_mark, 148 struct fsnotify_mark *fanotify_mark,
149 u32 mask, void *data, int data_type, 149 u32 mask, void *data, int data_type,
150 const unsigned char *file_name) 150 const unsigned char *file_name, u32 cookie)
151{ 151{
152 int ret = 0; 152 int ret = 0;
153 struct fanotify_event_info *event; 153 struct fanotify_event_info *event;
@@ -192,10 +192,12 @@ static int fanotify_handle_event(struct fsnotify_group *group,
192 192
193 ret = fsnotify_add_notify_event(group, fsn_event, fanotify_merge); 193 ret = fsnotify_add_notify_event(group, fsn_event, fanotify_merge);
194 if (ret) { 194 if (ret) {
195 BUG_ON(mask & FAN_ALL_PERM_EVENTS); 195 /* Permission events shouldn't be merged */
196 BUG_ON(ret == 1 && mask & FAN_ALL_PERM_EVENTS);
196 /* Our event wasn't used in the end. Free it. */ 197 /* Our event wasn't used in the end. Free it. */
197 fsnotify_destroy_event(group, fsn_event); 198 fsnotify_destroy_event(group, fsn_event);
198 ret = 0; 199
200 return 0;
199 } 201 }
200 202
201#ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS 203#ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c
index b6175fa11bf8..287a22c04149 100644
--- a/fs/notify/fanotify/fanotify_user.c
+++ b/fs/notify/fanotify/fanotify_user.c
@@ -698,6 +698,7 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
698 struct fsnotify_group *group; 698 struct fsnotify_group *group;
699 int f_flags, fd; 699 int f_flags, fd;
700 struct user_struct *user; 700 struct user_struct *user;
701 struct fanotify_event_info *oevent;
701 702
702 pr_debug("%s: flags=%d event_f_flags=%d\n", 703 pr_debug("%s: flags=%d event_f_flags=%d\n",
703 __func__, flags, event_f_flags); 704 __func__, flags, event_f_flags);
@@ -730,8 +731,20 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
730 group->fanotify_data.user = user; 731 group->fanotify_data.user = user;
731 atomic_inc(&user->fanotify_listeners); 732 atomic_inc(&user->fanotify_listeners);
732 733
734 oevent = kmem_cache_alloc(fanotify_event_cachep, GFP_KERNEL);
735 if (unlikely(!oevent)) {
736 fd = -ENOMEM;
737 goto out_destroy_group;
738 }
739 group->overflow_event = &oevent->fse;
740 fsnotify_init_event(group->overflow_event, NULL, FS_Q_OVERFLOW);
741 oevent->tgid = get_pid(task_tgid(current));
742 oevent->path.mnt = NULL;
743 oevent->path.dentry = NULL;
744
733 group->fanotify_data.f_flags = event_f_flags; 745 group->fanotify_data.f_flags = event_f_flags;
734#ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS 746#ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
747 oevent->response = 0;
735 mutex_init(&group->fanotify_data.access_mutex); 748 mutex_init(&group->fanotify_data.access_mutex);
736 init_waitqueue_head(&group->fanotify_data.access_waitq); 749 init_waitqueue_head(&group->fanotify_data.access_waitq);
737 INIT_LIST_HEAD(&group->fanotify_data.access_list); 750 INIT_LIST_HEAD(&group->fanotify_data.access_list);
diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c
index 1d4e1ea2f37c..9d3e9c50066a 100644
--- a/fs/notify/fsnotify.c
+++ b/fs/notify/fsnotify.c
@@ -179,7 +179,7 @@ static int send_to_group(struct inode *to_tell,
179 179
180 return group->ops->handle_event(group, to_tell, inode_mark, 180 return group->ops->handle_event(group, to_tell, inode_mark,
181 vfsmount_mark, mask, data, data_is, 181 vfsmount_mark, mask, data, data_is,
182 file_name); 182 file_name, cookie);
183} 183}
184 184
185/* 185/*
diff --git a/fs/notify/group.c b/fs/notify/group.c
index ee674fe2cec7..ad1995980456 100644
--- a/fs/notify/group.c
+++ b/fs/notify/group.c
@@ -55,6 +55,13 @@ void fsnotify_destroy_group(struct fsnotify_group *group)
55 /* clear the notification queue of all events */ 55 /* clear the notification queue of all events */
56 fsnotify_flush_notify(group); 56 fsnotify_flush_notify(group);
57 57
58 /*
59 * Destroy overflow event (we cannot use fsnotify_destroy_event() as
60 * that deliberately ignores overflow events.
61 */
62 if (group->overflow_event)
63 group->ops->free_event(group->overflow_event);
64
58 fsnotify_put_group(group); 65 fsnotify_put_group(group);
59} 66}
60 67
@@ -99,7 +106,6 @@ struct fsnotify_group *fsnotify_alloc_group(const struct fsnotify_ops *ops)
99 INIT_LIST_HEAD(&group->marks_list); 106 INIT_LIST_HEAD(&group->marks_list);
100 107
101 group->ops = ops; 108 group->ops = ops;
102 fsnotify_init_event(&group->overflow_event, NULL, FS_Q_OVERFLOW);
103 109
104 return group; 110 return group;
105} 111}
diff --git a/fs/notify/inotify/inotify.h b/fs/notify/inotify/inotify.h
index 485eef3f4407..ed855ef6f077 100644
--- a/fs/notify/inotify/inotify.h
+++ b/fs/notify/inotify/inotify.h
@@ -27,6 +27,6 @@ extern int inotify_handle_event(struct fsnotify_group *group,
27 struct fsnotify_mark *inode_mark, 27 struct fsnotify_mark *inode_mark,
28 struct fsnotify_mark *vfsmount_mark, 28 struct fsnotify_mark *vfsmount_mark,
29 u32 mask, void *data, int data_type, 29 u32 mask, void *data, int data_type,
30 const unsigned char *file_name); 30 const unsigned char *file_name, u32 cookie);
31 31
32extern const struct fsnotify_ops inotify_fsnotify_ops; 32extern const struct fsnotify_ops inotify_fsnotify_ops;
diff --git a/fs/notify/inotify/inotify_fsnotify.c b/fs/notify/inotify/inotify_fsnotify.c
index d5ee56348bb8..43ab1e1a07a2 100644
--- a/fs/notify/inotify/inotify_fsnotify.c
+++ b/fs/notify/inotify/inotify_fsnotify.c
@@ -67,7 +67,7 @@ int inotify_handle_event(struct fsnotify_group *group,
67 struct fsnotify_mark *inode_mark, 67 struct fsnotify_mark *inode_mark,
68 struct fsnotify_mark *vfsmount_mark, 68 struct fsnotify_mark *vfsmount_mark,
69 u32 mask, void *data, int data_type, 69 u32 mask, void *data, int data_type,
70 const unsigned char *file_name) 70 const unsigned char *file_name, u32 cookie)
71{ 71{
72 struct inotify_inode_mark *i_mark; 72 struct inotify_inode_mark *i_mark;
73 struct inotify_event_info *event; 73 struct inotify_event_info *event;
@@ -103,6 +103,7 @@ int inotify_handle_event(struct fsnotify_group *group,
103 fsn_event = &event->fse; 103 fsn_event = &event->fse;
104 fsnotify_init_event(fsn_event, inode, mask); 104 fsnotify_init_event(fsn_event, inode, mask);
105 event->wd = i_mark->wd; 105 event->wd = i_mark->wd;
106 event->sync_cookie = cookie;
106 event->name_len = len; 107 event->name_len = len;
107 if (len) 108 if (len)
108 strcpy(event->name, file_name); 109 strcpy(event->name, file_name);
diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index 497395c8274b..78a2ca3966c3 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -495,7 +495,7 @@ void inotify_ignored_and_remove_idr(struct fsnotify_mark *fsn_mark,
495 495
496 /* Queue ignore event for the watch */ 496 /* Queue ignore event for the watch */
497 inotify_handle_event(group, NULL, fsn_mark, NULL, FS_IN_IGNORED, 497 inotify_handle_event(group, NULL, fsn_mark, NULL, FS_IN_IGNORED,
498 NULL, FSNOTIFY_EVENT_NONE, NULL); 498 NULL, FSNOTIFY_EVENT_NONE, NULL, 0);
499 499
500 i_mark = container_of(fsn_mark, struct inotify_inode_mark, fsn_mark); 500 i_mark = container_of(fsn_mark, struct inotify_inode_mark, fsn_mark);
501 /* remove this mark from the idr */ 501 /* remove this mark from the idr */
@@ -633,11 +633,23 @@ static int inotify_update_watch(struct fsnotify_group *group, struct inode *inod
633static struct fsnotify_group *inotify_new_group(unsigned int max_events) 633static struct fsnotify_group *inotify_new_group(unsigned int max_events)
634{ 634{
635 struct fsnotify_group *group; 635 struct fsnotify_group *group;
636 struct inotify_event_info *oevent;
636 637
637 group = fsnotify_alloc_group(&inotify_fsnotify_ops); 638 group = fsnotify_alloc_group(&inotify_fsnotify_ops);
638 if (IS_ERR(group)) 639 if (IS_ERR(group))
639 return group; 640 return group;
640 641
642 oevent = kmalloc(sizeof(struct inotify_event_info), GFP_KERNEL);
643 if (unlikely(!oevent)) {
644 fsnotify_destroy_group(group);
645 return ERR_PTR(-ENOMEM);
646 }
647 group->overflow_event = &oevent->fse;
648 fsnotify_init_event(group->overflow_event, NULL, FS_Q_OVERFLOW);
649 oevent->wd = -1;
650 oevent->sync_cookie = 0;
651 oevent->name_len = 0;
652
641 group->max_events = max_events; 653 group->max_events = max_events;
642 654
643 spin_lock_init(&group->inotify_data.idr_lock); 655 spin_lock_init(&group->inotify_data.idr_lock);
diff --git a/fs/notify/notification.c b/fs/notify/notification.c
index 18b3c4427dca..1e58402171a5 100644
--- a/fs/notify/notification.c
+++ b/fs/notify/notification.c
@@ -80,7 +80,8 @@ void fsnotify_destroy_event(struct fsnotify_group *group,
80/* 80/*
81 * Add an event to the group notification queue. The group can later pull this 81 * Add an event to the group notification queue. The group can later pull this
82 * event off the queue to deal with. The function returns 0 if the event was 82 * event off the queue to deal with. The function returns 0 if the event was
83 * added to the queue, 1 if the event was merged with some other queued event. 83 * added to the queue, 1 if the event was merged with some other queued event,
84 * 2 if the queue of events has overflown.
84 */ 85 */
85int fsnotify_add_notify_event(struct fsnotify_group *group, 86int fsnotify_add_notify_event(struct fsnotify_group *group,
86 struct fsnotify_event *event, 87 struct fsnotify_event *event,
@@ -95,10 +96,14 @@ int fsnotify_add_notify_event(struct fsnotify_group *group,
95 mutex_lock(&group->notification_mutex); 96 mutex_lock(&group->notification_mutex);
96 97
97 if (group->q_len >= group->max_events) { 98 if (group->q_len >= group->max_events) {
99 ret = 2;
98 /* Queue overflow event only if it isn't already queued */ 100 /* Queue overflow event only if it isn't already queued */
99 if (list_empty(&group->overflow_event.list)) 101 if (!list_empty(&group->overflow_event->list)) {
100 event = &group->overflow_event; 102 mutex_unlock(&group->notification_mutex);
101 ret = 1; 103 return ret;
104 }
105 event = group->overflow_event;
106 goto queue;
102 } 107 }
103 108
104 if (!list_empty(list) && merge) { 109 if (!list_empty(list) && merge) {
@@ -109,6 +114,7 @@ int fsnotify_add_notify_event(struct fsnotify_group *group,
109 } 114 }
110 } 115 }
111 116
117queue:
112 group->q_len++; 118 group->q_len++;
113 list_add_tail(&event->list, list); 119 list_add_tail(&event->list, list);
114 mutex_unlock(&group->notification_mutex); 120 mutex_unlock(&group->notification_mutex);
@@ -132,7 +138,11 @@ struct fsnotify_event *fsnotify_remove_notify_event(struct fsnotify_group *group
132 138
133 event = list_first_entry(&group->notification_list, 139 event = list_first_entry(&group->notification_list,
134 struct fsnotify_event, list); 140 struct fsnotify_event, list);
135 list_del(&event->list); 141 /*
142 * We need to init list head for the case of overflow event so that
143 * check in fsnotify_add_notify_events() works
144 */
145 list_del_init(&event->list);
136 group->q_len--; 146 group->q_len--;
137 147
138 return event; 148 return event;
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 8450262bcf2a..51632c40e896 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -2393,8 +2393,8 @@ out_dio:
2393 2393
2394 if (((file->f_flags & O_DSYNC) && !direct_io) || IS_SYNC(inode) || 2394 if (((file->f_flags & O_DSYNC) && !direct_io) || IS_SYNC(inode) ||
2395 ((file->f_flags & O_DIRECT) && !direct_io)) { 2395 ((file->f_flags & O_DIRECT) && !direct_io)) {
2396 ret = filemap_fdatawrite_range(file->f_mapping, pos, 2396 ret = filemap_fdatawrite_range(file->f_mapping, *ppos,
2397 pos + count - 1); 2397 *ppos + count - 1);
2398 if (ret < 0) 2398 if (ret < 0)
2399 written = ret; 2399 written = ret;
2400 2400
@@ -2407,8 +2407,8 @@ out_dio:
2407 } 2407 }
2408 2408
2409 if (!ret) 2409 if (!ret)
2410 ret = filemap_fdatawait_range(file->f_mapping, pos, 2410 ret = filemap_fdatawait_range(file->f_mapping, *ppos,
2411 pos + count - 1); 2411 *ppos + count - 1);
2412 } 2412 }
2413 2413
2414 /* 2414 /*
diff --git a/fs/ocfs2/quota_global.c b/fs/ocfs2/quota_global.c
index aaa50611ec66..d7b5108789e2 100644
--- a/fs/ocfs2/quota_global.c
+++ b/fs/ocfs2/quota_global.c
@@ -717,6 +717,12 @@ static int ocfs2_release_dquot(struct dquot *dquot)
717 */ 717 */
718 if (status < 0) 718 if (status < 0)
719 mlog_errno(status); 719 mlog_errno(status);
720 /*
721 * Clear dq_off so that we search for the structure in quota file next
722 * time we acquire it. The structure might be deleted and reallocated
723 * elsewhere by another node while our dquot structure is on freelist.
724 */
725 dquot->dq_off = 0;
720 clear_bit(DQ_ACTIVE_B, &dquot->dq_flags); 726 clear_bit(DQ_ACTIVE_B, &dquot->dq_flags);
721out_trans: 727out_trans:
722 ocfs2_commit_trans(osb, handle); 728 ocfs2_commit_trans(osb, handle);
@@ -756,16 +762,17 @@ static int ocfs2_acquire_dquot(struct dquot *dquot)
756 status = ocfs2_lock_global_qf(info, 1); 762 status = ocfs2_lock_global_qf(info, 1);
757 if (status < 0) 763 if (status < 0)
758 goto out; 764 goto out;
759 if (!test_bit(DQ_READ_B, &dquot->dq_flags)) { 765 status = ocfs2_qinfo_lock(info, 0);
760 status = ocfs2_qinfo_lock(info, 0); 766 if (status < 0)
761 if (status < 0) 767 goto out_dq;
762 goto out_dq; 768 /*
763 status = qtree_read_dquot(&info->dqi_gi, dquot); 769 * We always want to read dquot structure from disk because we don't
764 ocfs2_qinfo_unlock(info, 0); 770 * know what happened with it while it was on freelist.
765 if (status < 0) 771 */
766 goto out_dq; 772 status = qtree_read_dquot(&info->dqi_gi, dquot);
767 } 773 ocfs2_qinfo_unlock(info, 0);
768 set_bit(DQ_READ_B, &dquot->dq_flags); 774 if (status < 0)
775 goto out_dq;
769 776
770 OCFS2_DQUOT(dquot)->dq_use_count++; 777 OCFS2_DQUOT(dquot)->dq_use_count++;
771 OCFS2_DQUOT(dquot)->dq_origspace = dquot->dq_dqb.dqb_curspace; 778 OCFS2_DQUOT(dquot)->dq_origspace = dquot->dq_dqb.dqb_curspace;
diff --git a/fs/ocfs2/quota_local.c b/fs/ocfs2/quota_local.c
index 2e4344be3b96..2001862bf2b1 100644
--- a/fs/ocfs2/quota_local.c
+++ b/fs/ocfs2/quota_local.c
@@ -1303,10 +1303,6 @@ int ocfs2_local_release_dquot(handle_t *handle, struct dquot *dquot)
1303 ocfs2_journal_dirty(handle, od->dq_chunk->qc_headerbh); 1303 ocfs2_journal_dirty(handle, od->dq_chunk->qc_headerbh);
1304 1304
1305out: 1305out:
1306 /* Clear the read bit so that next time someone uses this
1307 * dquot he reads fresh info from disk and allocates local
1308 * dquot structure */
1309 clear_bit(DQ_READ_B, &dquot->dq_flags);
1310 return status; 1306 return status;
1311} 1307}
1312 1308
diff --git a/fs/ocfs2/stackglue.c b/fs/ocfs2/stackglue.c
index 1324e6600e57..ca5ce14cbddc 100644
--- a/fs/ocfs2/stackglue.c
+++ b/fs/ocfs2/stackglue.c
@@ -346,7 +346,9 @@ int ocfs2_cluster_connect(const char *stack_name,
346 346
347 strlcpy(new_conn->cc_name, group, GROUP_NAME_MAX + 1); 347 strlcpy(new_conn->cc_name, group, GROUP_NAME_MAX + 1);
348 new_conn->cc_namelen = grouplen; 348 new_conn->cc_namelen = grouplen;
349 strlcpy(new_conn->cc_cluster_name, cluster_name, CLUSTER_NAME_MAX + 1); 349 if (cluster_name_len)
350 strlcpy(new_conn->cc_cluster_name, cluster_name,
351 CLUSTER_NAME_MAX + 1);
350 new_conn->cc_cluster_name_len = cluster_name_len; 352 new_conn->cc_cluster_name_len = cluster_name_len;
351 new_conn->cc_recovery_handler = recovery_handler; 353 new_conn->cc_recovery_handler = recovery_handler;
352 new_conn->cc_recovery_data = recovery_data; 354 new_conn->cc_recovery_data = recovery_data;
diff --git a/fs/open.c b/fs/open.c
index 4b3e1edf2fe4..b9ed8b25c108 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -705,6 +705,10 @@ static int do_dentry_open(struct file *f,
705 return 0; 705 return 0;
706 } 706 }
707 707
708 /* POSIX.1-2008/SUSv4 Section XSI 2.9.7 */
709 if (S_ISREG(inode->i_mode))
710 f->f_mode |= FMODE_ATOMIC_POS;
711
708 f->f_op = fops_get(inode->i_fop); 712 f->f_op = fops_get(inode->i_fop);
709 if (unlikely(WARN_ON(!f->f_op))) { 713 if (unlikely(WARN_ON(!f->f_op))) {
710 error = -ENODEV; 714 error = -ENODEV;
diff --git a/fs/pnode.c b/fs/pnode.c
index c7221bb19801..88396df725b4 100644
--- a/fs/pnode.c
+++ b/fs/pnode.c
@@ -220,14 +220,14 @@ static struct mount *get_source(struct mount *dest,
220 * @tree_list : list of heads of trees to be attached. 220 * @tree_list : list of heads of trees to be attached.
221 */ 221 */
222int propagate_mnt(struct mount *dest_mnt, struct mountpoint *dest_mp, 222int propagate_mnt(struct mount *dest_mnt, struct mountpoint *dest_mp,
223 struct mount *source_mnt, struct list_head *tree_list) 223 struct mount *source_mnt, struct hlist_head *tree_list)
224{ 224{
225 struct user_namespace *user_ns = current->nsproxy->mnt_ns->user_ns; 225 struct user_namespace *user_ns = current->nsproxy->mnt_ns->user_ns;
226 struct mount *m, *child; 226 struct mount *m, *child;
227 int ret = 0; 227 int ret = 0;
228 struct mount *prev_dest_mnt = dest_mnt; 228 struct mount *prev_dest_mnt = dest_mnt;
229 struct mount *prev_src_mnt = source_mnt; 229 struct mount *prev_src_mnt = source_mnt;
230 LIST_HEAD(tmp_list); 230 HLIST_HEAD(tmp_list);
231 231
232 for (m = propagation_next(dest_mnt, dest_mnt); m; 232 for (m = propagation_next(dest_mnt, dest_mnt); m;
233 m = propagation_next(m, dest_mnt)) { 233 m = propagation_next(m, dest_mnt)) {
@@ -246,27 +246,29 @@ int propagate_mnt(struct mount *dest_mnt, struct mountpoint *dest_mp,
246 child = copy_tree(source, source->mnt.mnt_root, type); 246 child = copy_tree(source, source->mnt.mnt_root, type);
247 if (IS_ERR(child)) { 247 if (IS_ERR(child)) {
248 ret = PTR_ERR(child); 248 ret = PTR_ERR(child);
249 list_splice(tree_list, tmp_list.prev); 249 tmp_list = *tree_list;
250 tmp_list.first->pprev = &tmp_list.first;
251 INIT_HLIST_HEAD(tree_list);
250 goto out; 252 goto out;
251 } 253 }
252 254
253 if (is_subdir(dest_mp->m_dentry, m->mnt.mnt_root)) { 255 if (is_subdir(dest_mp->m_dentry, m->mnt.mnt_root)) {
254 mnt_set_mountpoint(m, dest_mp, child); 256 mnt_set_mountpoint(m, dest_mp, child);
255 list_add_tail(&child->mnt_hash, tree_list); 257 hlist_add_head(&child->mnt_hash, tree_list);
256 } else { 258 } else {
257 /* 259 /*
258 * This can happen if the parent mount was bind mounted 260 * This can happen if the parent mount was bind mounted
259 * on some subdirectory of a shared/slave mount. 261 * on some subdirectory of a shared/slave mount.
260 */ 262 */
261 list_add_tail(&child->mnt_hash, &tmp_list); 263 hlist_add_head(&child->mnt_hash, &tmp_list);
262 } 264 }
263 prev_dest_mnt = m; 265 prev_dest_mnt = m;
264 prev_src_mnt = child; 266 prev_src_mnt = child;
265 } 267 }
266out: 268out:
267 lock_mount_hash(); 269 lock_mount_hash();
268 while (!list_empty(&tmp_list)) { 270 while (!hlist_empty(&tmp_list)) {
269 child = list_first_entry(&tmp_list, struct mount, mnt_hash); 271 child = hlist_entry(tmp_list.first, struct mount, mnt_hash);
270 umount_tree(child, 0); 272 umount_tree(child, 0);
271 } 273 }
272 unlock_mount_hash(); 274 unlock_mount_hash();
@@ -338,8 +340,10 @@ static void __propagate_umount(struct mount *mnt)
338 * umount the child only if the child has no 340 * umount the child only if the child has no
339 * other children 341 * other children
340 */ 342 */
341 if (child && list_empty(&child->mnt_mounts)) 343 if (child && list_empty(&child->mnt_mounts)) {
342 list_move_tail(&child->mnt_hash, &mnt->mnt_hash); 344 hlist_del_init_rcu(&child->mnt_hash);
345 hlist_add_before_rcu(&child->mnt_hash, &mnt->mnt_hash);
346 }
343 } 347 }
344} 348}
345 349
@@ -350,11 +354,11 @@ static void __propagate_umount(struct mount *mnt)
350 * 354 *
351 * vfsmount lock must be held for write 355 * vfsmount lock must be held for write
352 */ 356 */
353int propagate_umount(struct list_head *list) 357int propagate_umount(struct hlist_head *list)
354{ 358{
355 struct mount *mnt; 359 struct mount *mnt;
356 360
357 list_for_each_entry(mnt, list, mnt_hash) 361 hlist_for_each_entry(mnt, list, mnt_hash)
358 __propagate_umount(mnt); 362 __propagate_umount(mnt);
359 return 0; 363 return 0;
360} 364}
diff --git a/fs/pnode.h b/fs/pnode.h
index 59e7eda1851e..fc28a27fa892 100644
--- a/fs/pnode.h
+++ b/fs/pnode.h
@@ -36,8 +36,8 @@ static inline void set_mnt_shared(struct mount *mnt)
36 36
37void change_mnt_propagation(struct mount *, int); 37void change_mnt_propagation(struct mount *, int);
38int propagate_mnt(struct mount *, struct mountpoint *, struct mount *, 38int propagate_mnt(struct mount *, struct mountpoint *, struct mount *,
39 struct list_head *); 39 struct hlist_head *);
40int propagate_umount(struct list_head *); 40int propagate_umount(struct hlist_head *);
41int propagate_mount_busy(struct mount *, int); 41int propagate_mount_busy(struct mount *, int);
42void mnt_release_group_id(struct mount *); 42void mnt_release_group_id(struct mount *);
43int get_dominating_id(struct mount *mnt, const struct path *root); 43int get_dominating_id(struct mount *mnt, const struct path *root);
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 51507065263b..b9760628e1fd 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -1824,6 +1824,7 @@ static int proc_map_files_get_link(struct dentry *dentry, struct path *path)
1824 if (rc) 1824 if (rc)
1825 goto out_mmput; 1825 goto out_mmput;
1826 1826
1827 rc = -ENOENT;
1827 down_read(&mm->mmap_sem); 1828 down_read(&mm->mmap_sem);
1828 vma = find_exact_vma(mm, vm_start, vm_end); 1829 vma = find_exact_vma(mm, vm_start, vm_end);
1829 if (vma && vma->vm_file) { 1830 if (vma && vma->vm_file) {
diff --git a/fs/proc/page.c b/fs/proc/page.c
index 02174a610315..e647c55275d9 100644
--- a/fs/proc/page.c
+++ b/fs/proc/page.c
@@ -121,9 +121,8 @@ u64 stable_page_flags(struct page *page)
121 * just checks PG_head/PG_tail, so we need to check PageLRU/PageAnon 121 * just checks PG_head/PG_tail, so we need to check PageLRU/PageAnon
122 * to make sure a given page is a thp, not a non-huge compound page. 122 * to make sure a given page is a thp, not a non-huge compound page.
123 */ 123 */
124 else if (PageTransCompound(page) && 124 else if (PageTransCompound(page) && (PageLRU(compound_head(page)) ||
125 (PageLRU(compound_trans_head(page)) || 125 PageAnon(compound_head(page))))
126 PageAnon(compound_trans_head(page))))
127 u |= 1 << KPF_THP; 126 u |= 1 << KPF_THP;
128 127
129 /* 128 /*
diff --git a/fs/quota/dquot.c b/fs/quota/dquot.c
index 831d49a4111f..cfc8dcc16043 100644
--- a/fs/quota/dquot.c
+++ b/fs/quota/dquot.c
@@ -581,9 +581,17 @@ int dquot_scan_active(struct super_block *sb,
581 dqstats_inc(DQST_LOOKUPS); 581 dqstats_inc(DQST_LOOKUPS);
582 dqput(old_dquot); 582 dqput(old_dquot);
583 old_dquot = dquot; 583 old_dquot = dquot;
584 ret = fn(dquot, priv); 584 /*
585 if (ret < 0) 585 * ->release_dquot() can be racing with us. Our reference
586 goto out; 586 * protects us from new calls to it so just wait for any
587 * outstanding call and recheck the DQ_ACTIVE_B after that.
588 */
589 wait_on_dquot(dquot);
590 if (test_bit(DQ_ACTIVE_B, &dquot->dq_flags)) {
591 ret = fn(dquot, priv);
592 if (ret < 0)
593 goto out;
594 }
587 spin_lock(&dq_list_lock); 595 spin_lock(&dq_list_lock);
588 /* We are safe to continue now because our dquot could not 596 /* We are safe to continue now because our dquot could not
589 * be moved out of the inuse list while we hold the reference */ 597 * be moved out of the inuse list while we hold the reference */
diff --git a/fs/read_write.c b/fs/read_write.c
index edc5746a902a..28cc9c810744 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -264,10 +264,22 @@ loff_t vfs_llseek(struct file *file, loff_t offset, int whence)
264} 264}
265EXPORT_SYMBOL(vfs_llseek); 265EXPORT_SYMBOL(vfs_llseek);
266 266
267static inline struct fd fdget_pos(int fd)
268{
269 return __to_fd(__fdget_pos(fd));
270}
271
272static inline void fdput_pos(struct fd f)
273{
274 if (f.flags & FDPUT_POS_UNLOCK)
275 mutex_unlock(&f.file->f_pos_lock);
276 fdput(f);
277}
278
267SYSCALL_DEFINE3(lseek, unsigned int, fd, off_t, offset, unsigned int, whence) 279SYSCALL_DEFINE3(lseek, unsigned int, fd, off_t, offset, unsigned int, whence)
268{ 280{
269 off_t retval; 281 off_t retval;
270 struct fd f = fdget(fd); 282 struct fd f = fdget_pos(fd);
271 if (!f.file) 283 if (!f.file)
272 return -EBADF; 284 return -EBADF;
273 285
@@ -278,7 +290,7 @@ SYSCALL_DEFINE3(lseek, unsigned int, fd, off_t, offset, unsigned int, whence)
278 if (res != (loff_t)retval) 290 if (res != (loff_t)retval)
279 retval = -EOVERFLOW; /* LFS: should only happen on 32 bit platforms */ 291 retval = -EOVERFLOW; /* LFS: should only happen on 32 bit platforms */
280 } 292 }
281 fdput(f); 293 fdput_pos(f);
282 return retval; 294 return retval;
283} 295}
284 296
@@ -295,7 +307,7 @@ SYSCALL_DEFINE5(llseek, unsigned int, fd, unsigned long, offset_high,
295 unsigned int, whence) 307 unsigned int, whence)
296{ 308{
297 int retval; 309 int retval;
298 struct fd f = fdget(fd); 310 struct fd f = fdget_pos(fd);
299 loff_t offset; 311 loff_t offset;
300 312
301 if (!f.file) 313 if (!f.file)
@@ -315,7 +327,7 @@ SYSCALL_DEFINE5(llseek, unsigned int, fd, unsigned long, offset_high,
315 retval = 0; 327 retval = 0;
316 } 328 }
317out_putf: 329out_putf:
318 fdput(f); 330 fdput_pos(f);
319 return retval; 331 return retval;
320} 332}
321#endif 333#endif
@@ -498,7 +510,7 @@ static inline void file_pos_write(struct file *file, loff_t pos)
498 510
499SYSCALL_DEFINE3(read, unsigned int, fd, char __user *, buf, size_t, count) 511SYSCALL_DEFINE3(read, unsigned int, fd, char __user *, buf, size_t, count)
500{ 512{
501 struct fd f = fdget(fd); 513 struct fd f = fdget_pos(fd);
502 ssize_t ret = -EBADF; 514 ssize_t ret = -EBADF;
503 515
504 if (f.file) { 516 if (f.file) {
@@ -506,7 +518,7 @@ SYSCALL_DEFINE3(read, unsigned int, fd, char __user *, buf, size_t, count)
506 ret = vfs_read(f.file, buf, count, &pos); 518 ret = vfs_read(f.file, buf, count, &pos);
507 if (ret >= 0) 519 if (ret >= 0)
508 file_pos_write(f.file, pos); 520 file_pos_write(f.file, pos);
509 fdput(f); 521 fdput_pos(f);
510 } 522 }
511 return ret; 523 return ret;
512} 524}
@@ -514,7 +526,7 @@ SYSCALL_DEFINE3(read, unsigned int, fd, char __user *, buf, size_t, count)
514SYSCALL_DEFINE3(write, unsigned int, fd, const char __user *, buf, 526SYSCALL_DEFINE3(write, unsigned int, fd, const char __user *, buf,
515 size_t, count) 527 size_t, count)
516{ 528{
517 struct fd f = fdget(fd); 529 struct fd f = fdget_pos(fd);
518 ssize_t ret = -EBADF; 530 ssize_t ret = -EBADF;
519 531
520 if (f.file) { 532 if (f.file) {
@@ -522,7 +534,7 @@ SYSCALL_DEFINE3(write, unsigned int, fd, const char __user *, buf,
522 ret = vfs_write(f.file, buf, count, &pos); 534 ret = vfs_write(f.file, buf, count, &pos);
523 if (ret >= 0) 535 if (ret >= 0)
524 file_pos_write(f.file, pos); 536 file_pos_write(f.file, pos);
525 fdput(f); 537 fdput_pos(f);
526 } 538 }
527 539
528 return ret; 540 return ret;
@@ -797,7 +809,7 @@ EXPORT_SYMBOL(vfs_writev);
797SYSCALL_DEFINE3(readv, unsigned long, fd, const struct iovec __user *, vec, 809SYSCALL_DEFINE3(readv, unsigned long, fd, const struct iovec __user *, vec,
798 unsigned long, vlen) 810 unsigned long, vlen)
799{ 811{
800 struct fd f = fdget(fd); 812 struct fd f = fdget_pos(fd);
801 ssize_t ret = -EBADF; 813 ssize_t ret = -EBADF;
802 814
803 if (f.file) { 815 if (f.file) {
@@ -805,7 +817,7 @@ SYSCALL_DEFINE3(readv, unsigned long, fd, const struct iovec __user *, vec,
805 ret = vfs_readv(f.file, vec, vlen, &pos); 817 ret = vfs_readv(f.file, vec, vlen, &pos);
806 if (ret >= 0) 818 if (ret >= 0)
807 file_pos_write(f.file, pos); 819 file_pos_write(f.file, pos);
808 fdput(f); 820 fdput_pos(f);
809 } 821 }
810 822
811 if (ret > 0) 823 if (ret > 0)
@@ -817,7 +829,7 @@ SYSCALL_DEFINE3(readv, unsigned long, fd, const struct iovec __user *, vec,
817SYSCALL_DEFINE3(writev, unsigned long, fd, const struct iovec __user *, vec, 829SYSCALL_DEFINE3(writev, unsigned long, fd, const struct iovec __user *, vec,
818 unsigned long, vlen) 830 unsigned long, vlen)
819{ 831{
820 struct fd f = fdget(fd); 832 struct fd f = fdget_pos(fd);
821 ssize_t ret = -EBADF; 833 ssize_t ret = -EBADF;
822 834
823 if (f.file) { 835 if (f.file) {
@@ -825,7 +837,7 @@ SYSCALL_DEFINE3(writev, unsigned long, fd, const struct iovec __user *, vec,
825 ret = vfs_writev(f.file, vec, vlen, &pos); 837 ret = vfs_writev(f.file, vec, vlen, &pos);
826 if (ret >= 0) 838 if (ret >= 0)
827 file_pos_write(f.file, pos); 839 file_pos_write(f.file, pos);
828 fdput(f); 840 fdput_pos(f);
829 } 841 }
830 842
831 if (ret > 0) 843 if (ret > 0)
@@ -968,7 +980,7 @@ COMPAT_SYSCALL_DEFINE3(readv, compat_ulong_t, fd,
968 const struct compat_iovec __user *,vec, 980 const struct compat_iovec __user *,vec,
969 compat_ulong_t, vlen) 981 compat_ulong_t, vlen)
970{ 982{
971 struct fd f = fdget(fd); 983 struct fd f = fdget_pos(fd);
972 ssize_t ret; 984 ssize_t ret;
973 loff_t pos; 985 loff_t pos;
974 986
@@ -978,7 +990,7 @@ COMPAT_SYSCALL_DEFINE3(readv, compat_ulong_t, fd,
978 ret = compat_readv(f.file, vec, vlen, &pos); 990 ret = compat_readv(f.file, vec, vlen, &pos);
979 if (ret >= 0) 991 if (ret >= 0)
980 f.file->f_pos = pos; 992 f.file->f_pos = pos;
981 fdput(f); 993 fdput_pos(f);
982 return ret; 994 return ret;
983} 995}
984 996
@@ -1035,7 +1047,7 @@ COMPAT_SYSCALL_DEFINE3(writev, compat_ulong_t, fd,
1035 const struct compat_iovec __user *, vec, 1047 const struct compat_iovec __user *, vec,
1036 compat_ulong_t, vlen) 1048 compat_ulong_t, vlen)
1037{ 1049{
1038 struct fd f = fdget(fd); 1050 struct fd f = fdget_pos(fd);
1039 ssize_t ret; 1051 ssize_t ret;
1040 loff_t pos; 1052 loff_t pos;
1041 1053
@@ -1045,7 +1057,7 @@ COMPAT_SYSCALL_DEFINE3(writev, compat_ulong_t, fd,
1045 ret = compat_writev(f.file, vec, vlen, &pos); 1057 ret = compat_writev(f.file, vec, vlen, &pos);
1046 if (ret >= 0) 1058 if (ret >= 0)
1047 f.file->f_pos = pos; 1059 f.file->f_pos = pos;
1048 fdput(f); 1060 fdput_pos(f);
1049 return ret; 1061 return ret;
1050} 1062}
1051 1063
diff --git a/fs/reiserfs/do_balan.c b/fs/reiserfs/do_balan.c
index 2b7882b508db..9a3c68cf6026 100644
--- a/fs/reiserfs/do_balan.c
+++ b/fs/reiserfs/do_balan.c
@@ -324,23 +324,17 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
324 switch (flag) { 324 switch (flag) {
325 case M_INSERT: /* insert item into L[0] */ 325 case M_INSERT: /* insert item into L[0] */
326 326
327 if (item_pos == tb->lnum[0] - 1 327 if (item_pos == tb->lnum[0] - 1 && tb->lbytes != -1) {
328 && tb->lbytes != -1) {
329 /* part of new item falls into L[0] */ 328 /* part of new item falls into L[0] */
330 int new_item_len; 329 int new_item_len;
331 int version; 330 int version;
332 331
333 ret_val = 332 ret_val = leaf_shift_left(tb, tb->lnum[0] - 1, -1);
334 leaf_shift_left(tb, tb->lnum[0] - 1,
335 -1);
336 333
337 /* Calculate item length to insert to S[0] */ 334 /* Calculate item length to insert to S[0] */
338 new_item_len = 335 new_item_len = ih_item_len(ih) - tb->lbytes;
339 ih_item_len(ih) - tb->lbytes;
340 /* Calculate and check item length to insert to L[0] */ 336 /* Calculate and check item length to insert to L[0] */
341 put_ih_item_len(ih, 337 put_ih_item_len(ih, ih_item_len(ih) - new_item_len);
342 ih_item_len(ih) -
343 new_item_len);
344 338
345 RFALSE(ih_item_len(ih) <= 0, 339 RFALSE(ih_item_len(ih) <= 0,
346 "PAP-12080: there is nothing to insert into L[0]: ih_item_len=%d", 340 "PAP-12080: there is nothing to insert into L[0]: ih_item_len=%d",
@@ -349,30 +343,18 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
349 /* Insert new item into L[0] */ 343 /* Insert new item into L[0] */
350 buffer_info_init_left(tb, &bi); 344 buffer_info_init_left(tb, &bi);
351 leaf_insert_into_buf(&bi, 345 leaf_insert_into_buf(&bi,
352 n + item_pos - 346 n + item_pos - ret_val, ih, body,
353 ret_val, ih, body, 347 zeros_num > ih_item_len(ih) ? ih_item_len(ih) : zeros_num);
354 zeros_num >
355 ih_item_len(ih) ?
356 ih_item_len(ih) :
357 zeros_num);
358 348
359 version = ih_version(ih); 349 version = ih_version(ih);
360 350
361 /* Calculate key component, item length and body to insert into S[0] */ 351 /* Calculate key component, item length and body to insert into S[0] */
362 set_le_ih_k_offset(ih, 352 set_le_ih_k_offset(ih, le_ih_k_offset(ih) +
363 le_ih_k_offset(ih) + 353 (tb-> lbytes << (is_indirect_le_ih(ih) ? tb->tb_sb-> s_blocksize_bits - UNFM_P_SHIFT : 0)));
364 (tb->
365 lbytes <<
366 (is_indirect_le_ih
367 (ih) ? tb->tb_sb->
368 s_blocksize_bits -
369 UNFM_P_SHIFT :
370 0)));
371 354
372 put_ih_item_len(ih, new_item_len); 355 put_ih_item_len(ih, new_item_len);
373 if (tb->lbytes > zeros_num) { 356 if (tb->lbytes > zeros_num) {
374 body += 357 body += (tb->lbytes - zeros_num);
375 (tb->lbytes - zeros_num);
376 zeros_num = 0; 358 zeros_num = 0;
377 } else 359 } else
378 zeros_num -= tb->lbytes; 360 zeros_num -= tb->lbytes;
@@ -383,15 +365,10 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
383 } else { 365 } else {
384 /* new item in whole falls into L[0] */ 366 /* new item in whole falls into L[0] */
385 /* Shift lnum[0]-1 items to L[0] */ 367 /* Shift lnum[0]-1 items to L[0] */
386 ret_val = 368 ret_val = leaf_shift_left(tb, tb->lnum[0] - 1, tb->lbytes);
387 leaf_shift_left(tb, tb->lnum[0] - 1,
388 tb->lbytes);
389 /* Insert new item into L[0] */ 369 /* Insert new item into L[0] */
390 buffer_info_init_left(tb, &bi); 370 buffer_info_init_left(tb, &bi);
391 leaf_insert_into_buf(&bi, 371 leaf_insert_into_buf(&bi, n + item_pos - ret_val, ih, body, zeros_num);
392 n + item_pos -
393 ret_val, ih, body,
394 zeros_num);
395 tb->insert_size[0] = 0; 372 tb->insert_size[0] = 0;
396 zeros_num = 0; 373 zeros_num = 0;
397 } 374 }
@@ -399,264 +376,117 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
399 376
400 case M_PASTE: /* append item in L[0] */ 377 case M_PASTE: /* append item in L[0] */
401 378
402 if (item_pos == tb->lnum[0] - 1 379 if (item_pos == tb->lnum[0] - 1 && tb->lbytes != -1) {
403 && tb->lbytes != -1) {
404 /* we must shift the part of the appended item */ 380 /* we must shift the part of the appended item */
405 if (is_direntry_le_ih 381 if (is_direntry_le_ih(B_N_PITEM_HEAD(tbS0, item_pos))) {
406 (B_N_PITEM_HEAD(tbS0, item_pos))) {
407 382
408 RFALSE(zeros_num, 383 RFALSE(zeros_num,
409 "PAP-12090: invalid parameter in case of a directory"); 384 "PAP-12090: invalid parameter in case of a directory");
410 /* directory item */ 385 /* directory item */
411 if (tb->lbytes > pos_in_item) { 386 if (tb->lbytes > pos_in_item) {
412 /* new directory entry falls into L[0] */ 387 /* new directory entry falls into L[0] */
413 struct item_head 388 struct item_head *pasted;
414 *pasted; 389 int l_pos_in_item = pos_in_item;
415 int l_pos_in_item =
416 pos_in_item;
417 390
418 /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 entries from given directory item */ 391 /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 entries from given directory item */
419 ret_val = 392 ret_val = leaf_shift_left(tb, tb->lnum[0], tb->lbytes-1);
420 leaf_shift_left(tb, 393 if (ret_val && !item_pos) {
421 tb-> 394 pasted = B_N_PITEM_HEAD(tb->L[0], B_NR_ITEMS(tb->L[0]) - 1);
422 lnum 395 l_pos_in_item += I_ENTRY_COUNT(pasted) - (tb->lbytes -1);
423 [0],
424 tb->
425 lbytes
426 -
427 1);
428 if (ret_val
429 && !item_pos) {
430 pasted =
431 B_N_PITEM_HEAD
432 (tb->L[0],
433 B_NR_ITEMS
434 (tb->
435 L[0]) -
436 1);
437 l_pos_in_item +=
438 I_ENTRY_COUNT
439 (pasted) -
440 (tb->
441 lbytes -
442 1);
443 } 396 }
444 397
445 /* Append given directory entry to directory item */ 398 /* Append given directory entry to directory item */
446 buffer_info_init_left(tb, &bi); 399 buffer_info_init_left(tb, &bi);
447 leaf_paste_in_buffer 400 leaf_paste_in_buffer(&bi, n + item_pos - ret_val, l_pos_in_item, tb->insert_size[0], body, zeros_num);
448 (&bi,
449 n + item_pos -
450 ret_val,
451 l_pos_in_item,
452 tb->insert_size[0],
453 body, zeros_num);
454 401
455 /* previous string prepared space for pasting new entry, following string pastes this entry */ 402 /* previous string prepared space for pasting new entry, following string pastes this entry */
456 403
457 /* when we have merge directory item, pos_in_item has been changed too */ 404 /* when we have merge directory item, pos_in_item has been changed too */
458 405
459 /* paste new directory entry. 1 is entry number */ 406 /* paste new directory entry. 1 is entry number */
460 leaf_paste_entries(&bi, 407 leaf_paste_entries(&bi, n + item_pos - ret_val, l_pos_in_item,
461 n + 408 1, (struct reiserfs_de_head *) body,
462 item_pos 409 body + DEH_SIZE, tb->insert_size[0]);
463 -
464 ret_val,
465 l_pos_in_item,
466 1,
467 (struct
468 reiserfs_de_head
469 *)
470 body,
471 body
472 +
473 DEH_SIZE,
474 tb->
475 insert_size
476 [0]
477 );
478 tb->insert_size[0] = 0; 410 tb->insert_size[0] = 0;
479 } else { 411 } else {
480 /* new directory item doesn't fall into L[0] */ 412 /* new directory item doesn't fall into L[0] */
481 /* Shift lnum[0]-1 items in whole. Shift lbytes directory entries from directory item number lnum[0] */ 413 /* Shift lnum[0]-1 items in whole. Shift lbytes directory entries from directory item number lnum[0] */
482 leaf_shift_left(tb, 414 leaf_shift_left(tb, tb->lnum[0], tb->lbytes);
483 tb->
484 lnum[0],
485 tb->
486 lbytes);
487 } 415 }
488 /* Calculate new position to append in item body */ 416 /* Calculate new position to append in item body */
489 pos_in_item -= tb->lbytes; 417 pos_in_item -= tb->lbytes;
490 } else { 418 } else {
491 /* regular object */ 419 /* regular object */
492 RFALSE(tb->lbytes <= 0, 420 RFALSE(tb->lbytes <= 0, "PAP-12095: there is nothing to shift to L[0]. lbytes=%d", tb->lbytes);
493 "PAP-12095: there is nothing to shift to L[0]. lbytes=%d", 421 RFALSE(pos_in_item != ih_item_len(B_N_PITEM_HEAD(tbS0, item_pos)),
494 tb->lbytes);
495 RFALSE(pos_in_item !=
496 ih_item_len
497 (B_N_PITEM_HEAD
498 (tbS0, item_pos)),
499 "PAP-12100: incorrect position to paste: item_len=%d, pos_in_item=%d", 422 "PAP-12100: incorrect position to paste: item_len=%d, pos_in_item=%d",
500 ih_item_len 423 ih_item_len(B_N_PITEM_HEAD(tbS0, item_pos)),pos_in_item);
501 (B_N_PITEM_HEAD
502 (tbS0, item_pos)),
503 pos_in_item);
504 424
505 if (tb->lbytes >= pos_in_item) { 425 if (tb->lbytes >= pos_in_item) {
506 /* appended item will be in L[0] in whole */ 426 /* appended item will be in L[0] in whole */
507 int l_n; 427 int l_n;
508 428
509 /* this bytes number must be appended to the last item of L[h] */ 429 /* this bytes number must be appended to the last item of L[h] */
510 l_n = 430 l_n = tb->lbytes - pos_in_item;
511 tb->lbytes -
512 pos_in_item;
513 431
514 /* Calculate new insert_size[0] */ 432 /* Calculate new insert_size[0] */
515 tb->insert_size[0] -= 433 tb->insert_size[0] -= l_n;
516 l_n;
517 434
518 RFALSE(tb-> 435 RFALSE(tb->insert_size[0] <= 0,
519 insert_size[0] <=
520 0,
521 "PAP-12105: there is nothing to paste into L[0]. insert_size=%d", 436 "PAP-12105: there is nothing to paste into L[0]. insert_size=%d",
522 tb-> 437 tb->insert_size[0]);
523 insert_size[0]); 438 ret_val = leaf_shift_left(tb, tb->lnum[0], ih_item_len
524 ret_val = 439 (B_N_PITEM_HEAD(tbS0, item_pos)));
525 leaf_shift_left(tb,
526 tb->
527 lnum
528 [0],
529 ih_item_len
530 (B_N_PITEM_HEAD
531 (tbS0,
532 item_pos)));
533 /* Append to body of item in L[0] */ 440 /* Append to body of item in L[0] */
534 buffer_info_init_left(tb, &bi); 441 buffer_info_init_left(tb, &bi);
535 leaf_paste_in_buffer 442 leaf_paste_in_buffer
536 (&bi, 443 (&bi, n + item_pos - ret_val, ih_item_len
537 n + item_pos - 444 (B_N_PITEM_HEAD(tb->L[0], n + item_pos - ret_val)),
538 ret_val, 445 l_n, body,
539 ih_item_len 446 zeros_num > l_n ? l_n : zeros_num);
540 (B_N_PITEM_HEAD
541 (tb->L[0],
542 n + item_pos -
543 ret_val)), l_n,
544 body,
545 zeros_num >
546 l_n ? l_n :
547 zeros_num);
548 /* 0-th item in S0 can be only of DIRECT type when l_n != 0 */ 447 /* 0-th item in S0 can be only of DIRECT type when l_n != 0 */
549 { 448 {
550 int version; 449 int version;
551 int temp_l = 450 int temp_l = l_n;
552 l_n; 451
553 452 RFALSE(ih_item_len(B_N_PITEM_HEAD(tbS0, 0)),
554 RFALSE
555 (ih_item_len
556 (B_N_PITEM_HEAD
557 (tbS0,
558 0)),
559 "PAP-12106: item length must be 0"); 453 "PAP-12106: item length must be 0");
560 RFALSE 454 RFALSE(comp_short_le_keys(B_N_PKEY(tbS0, 0), B_N_PKEY
561 (comp_short_le_keys 455 (tb->L[0], n + item_pos - ret_val)),
562 (B_N_PKEY
563 (tbS0, 0),
564 B_N_PKEY
565 (tb->L[0],
566 n +
567 item_pos
568 -
569 ret_val)),
570 "PAP-12107: items must be of the same file"); 456 "PAP-12107: items must be of the same file");
571 if (is_indirect_le_ih(B_N_PITEM_HEAD(tb->L[0], n + item_pos - ret_val))) { 457 if (is_indirect_le_ih(B_N_PITEM_HEAD(tb->L[0], n + item_pos - ret_val))) {
572 temp_l = 458 temp_l = l_n << (tb->tb_sb-> s_blocksize_bits - UNFM_P_SHIFT);
573 l_n
574 <<
575 (tb->
576 tb_sb->
577 s_blocksize_bits
578 -
579 UNFM_P_SHIFT);
580 } 459 }
581 /* update key of first item in S0 */ 460 /* update key of first item in S0 */
582 version = 461 version = ih_version(B_N_PITEM_HEAD(tbS0, 0));
583 ih_version 462 set_le_key_k_offset(version, B_N_PKEY(tbS0, 0),
584 (B_N_PITEM_HEAD 463 le_key_k_offset(version,B_N_PKEY(tbS0, 0)) + temp_l);
585 (tbS0, 0));
586 set_le_key_k_offset
587 (version,
588 B_N_PKEY
589 (tbS0, 0),
590 le_key_k_offset
591 (version,
592 B_N_PKEY
593 (tbS0,
594 0)) +
595 temp_l);
596 /* update left delimiting key */ 464 /* update left delimiting key */
597 set_le_key_k_offset 465 set_le_key_k_offset(version, B_N_PDELIM_KEY(tb->CFL[0], tb->lkey[0]),
598 (version, 466 le_key_k_offset(version, B_N_PDELIM_KEY(tb->CFL[0], tb->lkey[0])) + temp_l);
599 B_N_PDELIM_KEY
600 (tb->
601 CFL[0],
602 tb->
603 lkey[0]),
604 le_key_k_offset
605 (version,
606 B_N_PDELIM_KEY
607 (tb->
608 CFL[0],
609 tb->
610 lkey[0]))
611 + temp_l);
612 } 467 }
613 468
614 /* Calculate new body, position in item and insert_size[0] */ 469 /* Calculate new body, position in item and insert_size[0] */
615 if (l_n > zeros_num) { 470 if (l_n > zeros_num) {
616 body += 471 body += (l_n - zeros_num);
617 (l_n -
618 zeros_num);
619 zeros_num = 0; 472 zeros_num = 0;
620 } else 473 } else
621 zeros_num -= 474 zeros_num -= l_n;
622 l_n;
623 pos_in_item = 0; 475 pos_in_item = 0;
624 476
625 RFALSE 477 RFALSE(comp_short_le_keys(B_N_PKEY(tbS0, 0), B_N_PKEY(tb->L[0], B_NR_ITEMS(tb->L[0]) - 1))
626 (comp_short_le_keys 478 || !op_is_left_mergeable(B_N_PKEY(tbS0, 0), tbS0->b_size)
627 (B_N_PKEY(tbS0, 0), 479 || !op_is_left_mergeable(B_N_PDELIM_KEY(tb->CFL[0], tb->lkey[0]), tbS0->b_size),
628 B_N_PKEY(tb->L[0],
629 B_NR_ITEMS
630 (tb->
631 L[0]) -
632 1))
633 ||
634 !op_is_left_mergeable
635 (B_N_PKEY(tbS0, 0),
636 tbS0->b_size)
637 ||
638 !op_is_left_mergeable
639 (B_N_PDELIM_KEY
640 (tb->CFL[0],
641 tb->lkey[0]),
642 tbS0->b_size),
643 "PAP-12120: item must be merge-able with left neighboring item"); 480 "PAP-12120: item must be merge-able with left neighboring item");
644 } else { /* only part of the appended item will be in L[0] */ 481 } else { /* only part of the appended item will be in L[0] */
645 482
646 /* Calculate position in item for append in S[0] */ 483 /* Calculate position in item for append in S[0] */
647 pos_in_item -= 484 pos_in_item -= tb->lbytes;
648 tb->lbytes;
649 485
650 RFALSE(pos_in_item <= 0, 486 RFALSE(pos_in_item <= 0, "PAP-12125: no place for paste. pos_in_item=%d", pos_in_item);
651 "PAP-12125: no place for paste. pos_in_item=%d",
652 pos_in_item);
653 487
654 /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 byte from item number lnum[0] */ 488 /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 byte from item number lnum[0] */
655 leaf_shift_left(tb, 489 leaf_shift_left(tb, tb->lnum[0], tb->lbytes);
656 tb->
657 lnum[0],
658 tb->
659 lbytes);
660 } 490 }
661 } 491 }
662 } else { /* appended item will be in L[0] in whole */ 492 } else { /* appended item will be in L[0] in whole */
@@ -665,52 +495,30 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
665 495
666 if (!item_pos && op_is_left_mergeable(B_N_PKEY(tbS0, 0), tbS0->b_size)) { /* if we paste into first item of S[0] and it is left mergable */ 496 if (!item_pos && op_is_left_mergeable(B_N_PKEY(tbS0, 0), tbS0->b_size)) { /* if we paste into first item of S[0] and it is left mergable */
667 /* then increment pos_in_item by the size of the last item in L[0] */ 497 /* then increment pos_in_item by the size of the last item in L[0] */
668 pasted = 498 pasted = B_N_PITEM_HEAD(tb->L[0], n - 1);
669 B_N_PITEM_HEAD(tb->L[0],
670 n - 1);
671 if (is_direntry_le_ih(pasted)) 499 if (is_direntry_le_ih(pasted))
672 pos_in_item += 500 pos_in_item += ih_entry_count(pasted);
673 ih_entry_count
674 (pasted);
675 else 501 else
676 pos_in_item += 502 pos_in_item += ih_item_len(pasted);
677 ih_item_len(pasted);
678 } 503 }
679 504
680 /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 byte from item number lnum[0] */ 505 /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 byte from item number lnum[0] */
681 ret_val = 506 ret_val = leaf_shift_left(tb, tb->lnum[0], tb->lbytes);
682 leaf_shift_left(tb, tb->lnum[0],
683 tb->lbytes);
684 /* Append to body of item in L[0] */ 507 /* Append to body of item in L[0] */
685 buffer_info_init_left(tb, &bi); 508 buffer_info_init_left(tb, &bi);
686 leaf_paste_in_buffer(&bi, 509 leaf_paste_in_buffer(&bi, n + item_pos - ret_val,
687 n + item_pos -
688 ret_val,
689 pos_in_item, 510 pos_in_item,
690 tb->insert_size[0], 511 tb->insert_size[0],
691 body, zeros_num); 512 body, zeros_num);
692 513
693 /* if appended item is directory, paste entry */ 514 /* if appended item is directory, paste entry */
694 pasted = 515 pasted = B_N_PITEM_HEAD(tb->L[0], n + item_pos - ret_val);
695 B_N_PITEM_HEAD(tb->L[0],
696 n + item_pos -
697 ret_val);
698 if (is_direntry_le_ih(pasted)) 516 if (is_direntry_le_ih(pasted))
699 leaf_paste_entries(&bi, 517 leaf_paste_entries(&bi, n + item_pos - ret_val,
700 n + 518 pos_in_item, 1,
701 item_pos - 519 (struct reiserfs_de_head *) body,
702 ret_val, 520 body + DEH_SIZE,
703 pos_in_item, 521 tb->insert_size[0]);
704 1,
705 (struct
706 reiserfs_de_head
707 *)body,
708 body +
709 DEH_SIZE,
710 tb->
711 insert_size
712 [0]
713 );
714 /* if appended item is indirect item, put unformatted node into un list */ 522 /* if appended item is indirect item, put unformatted node into un list */
715 if (is_indirect_le_ih(pasted)) 523 if (is_indirect_le_ih(pasted))
716 set_ih_free_space(pasted, 0); 524 set_ih_free_space(pasted, 0);
@@ -722,13 +530,7 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
722 reiserfs_panic(tb->tb_sb, "PAP-12130", 530 reiserfs_panic(tb->tb_sb, "PAP-12130",
723 "lnum > 0: unexpected mode: " 531 "lnum > 0: unexpected mode: "
724 " %s(%d)", 532 " %s(%d)",
725 (flag == 533 (flag == M_DELETE) ? "DELETE" : ((flag == M_CUT) ? "CUT" : "UNKNOWN"), flag);
726 M_DELETE) ? "DELETE" : ((flag ==
727 M_CUT)
728 ? "CUT"
729 :
730 "UNKNOWN"),
731 flag);
732 } 534 }
733 } else { 535 } else {
734 /* new item doesn't fall into L[0] */ 536 /* new item doesn't fall into L[0] */
@@ -748,14 +550,12 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
748 case M_INSERT: /* insert item */ 550 case M_INSERT: /* insert item */
749 if (n - tb->rnum[0] < item_pos) { /* new item or its part falls to R[0] */ 551 if (n - tb->rnum[0] < item_pos) { /* new item or its part falls to R[0] */
750 if (item_pos == n - tb->rnum[0] + 1 && tb->rbytes != -1) { /* part of new item falls into R[0] */ 552 if (item_pos == n - tb->rnum[0] + 1 && tb->rbytes != -1) { /* part of new item falls into R[0] */
751 loff_t old_key_comp, old_len, 553 loff_t old_key_comp, old_len, r_zeros_number;
752 r_zeros_number;
753 const char *r_body; 554 const char *r_body;
754 int version; 555 int version;
755 loff_t offset; 556 loff_t offset;
756 557
757 leaf_shift_right(tb, tb->rnum[0] - 1, 558 leaf_shift_right(tb, tb->rnum[0] - 1, -1);
758 -1);
759 559
760 version = ih_version(ih); 560 version = ih_version(ih);
761 /* Remember key component and item length */ 561 /* Remember key component and item length */
@@ -763,29 +563,17 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
763 old_len = ih_item_len(ih); 563 old_len = ih_item_len(ih);
764 564
765 /* Calculate key component and item length to insert into R[0] */ 565 /* Calculate key component and item length to insert into R[0] */
766 offset = 566 offset = le_ih_k_offset(ih) + ((old_len - tb->rbytes) << (is_indirect_le_ih(ih) ? tb->tb_sb->s_blocksize_bits - UNFM_P_SHIFT : 0));
767 le_ih_k_offset(ih) +
768 ((old_len -
769 tb->
770 rbytes) << (is_indirect_le_ih(ih)
771 ? tb->tb_sb->
772 s_blocksize_bits -
773 UNFM_P_SHIFT : 0));
774 set_le_ih_k_offset(ih, offset); 567 set_le_ih_k_offset(ih, offset);
775 put_ih_item_len(ih, tb->rbytes); 568 put_ih_item_len(ih, tb->rbytes);
776 /* Insert part of the item into R[0] */ 569 /* Insert part of the item into R[0] */
777 buffer_info_init_right(tb, &bi); 570 buffer_info_init_right(tb, &bi);
778 if ((old_len - tb->rbytes) > zeros_num) { 571 if ((old_len - tb->rbytes) > zeros_num) {
779 r_zeros_number = 0; 572 r_zeros_number = 0;
780 r_body = 573 r_body = body + (old_len - tb->rbytes) - zeros_num;
781 body + (old_len -
782 tb->rbytes) -
783 zeros_num;
784 } else { 574 } else {
785 r_body = body; 575 r_body = body;
786 r_zeros_number = 576 r_zeros_number = zeros_num - (old_len - tb->rbytes);
787 zeros_num - (old_len -
788 tb->rbytes);
789 zeros_num -= r_zeros_number; 577 zeros_num -= r_zeros_number;
790 } 578 }
791 579
@@ -798,25 +586,18 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
798 586
799 /* Calculate key component and item length to insert into S[0] */ 587 /* Calculate key component and item length to insert into S[0] */
800 set_le_ih_k_offset(ih, old_key_comp); 588 set_le_ih_k_offset(ih, old_key_comp);
801 put_ih_item_len(ih, 589 put_ih_item_len(ih, old_len - tb->rbytes);
802 old_len - tb->rbytes);
803 590
804 tb->insert_size[0] -= tb->rbytes; 591 tb->insert_size[0] -= tb->rbytes;
805 592
806 } else { /* whole new item falls into R[0] */ 593 } else { /* whole new item falls into R[0] */
807 594
808 /* Shift rnum[0]-1 items to R[0] */ 595 /* Shift rnum[0]-1 items to R[0] */
809 ret_val = 596 ret_val = leaf_shift_right(tb, tb->rnum[0] - 1, tb->rbytes);
810 leaf_shift_right(tb,
811 tb->rnum[0] - 1,
812 tb->rbytes);
813 /* Insert new item into R[0] */ 597 /* Insert new item into R[0] */
814 buffer_info_init_right(tb, &bi); 598 buffer_info_init_right(tb, &bi);
815 leaf_insert_into_buf(&bi, 599 leaf_insert_into_buf(&bi, item_pos - n + tb->rnum[0] - 1,
816 item_pos - n + 600 ih, body, zeros_num);
817 tb->rnum[0] - 1,
818 ih, body,
819 zeros_num);
820 601
821 if (item_pos - n + tb->rnum[0] - 1 == 0) { 602 if (item_pos - n + tb->rnum[0] - 1 == 0) {
822 replace_key(tb, tb->CFR[0], 603 replace_key(tb, tb->CFR[0],
@@ -841,200 +622,97 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
841 622
842 RFALSE(zeros_num, 623 RFALSE(zeros_num,
843 "PAP-12145: invalid parameter in case of a directory"); 624 "PAP-12145: invalid parameter in case of a directory");
844 entry_count = 625 entry_count = I_ENTRY_COUNT(B_N_PITEM_HEAD
845 I_ENTRY_COUNT(B_N_PITEM_HEAD 626 (tbS0, item_pos));
846 (tbS0,
847 item_pos));
848 if (entry_count - tb->rbytes < 627 if (entry_count - tb->rbytes <
849 pos_in_item) 628 pos_in_item)
850 /* new directory entry falls into R[0] */ 629 /* new directory entry falls into R[0] */
851 { 630 {
852 int paste_entry_position; 631 int paste_entry_position;
853 632
854 RFALSE(tb->rbytes - 1 >= 633 RFALSE(tb->rbytes - 1 >= entry_count || !tb-> insert_size[0],
855 entry_count
856 || !tb->
857 insert_size[0],
858 "PAP-12150: no enough of entries to shift to R[0]: rbytes=%d, entry_count=%d", 634 "PAP-12150: no enough of entries to shift to R[0]: rbytes=%d, entry_count=%d",
859 tb->rbytes, 635 tb->rbytes, entry_count);
860 entry_count);
861 /* Shift rnum[0]-1 items in whole. Shift rbytes-1 directory entries from directory item number rnum[0] */ 636 /* Shift rnum[0]-1 items in whole. Shift rbytes-1 directory entries from directory item number rnum[0] */
862 leaf_shift_right(tb, 637 leaf_shift_right(tb, tb->rnum[0], tb->rbytes - 1);
863 tb->
864 rnum
865 [0],
866 tb->
867 rbytes
868 - 1);
869 /* Paste given directory entry to directory item */ 638 /* Paste given directory entry to directory item */
870 paste_entry_position = 639 paste_entry_position = pos_in_item - entry_count + tb->rbytes - 1;
871 pos_in_item -
872 entry_count +
873 tb->rbytes - 1;
874 buffer_info_init_right(tb, &bi); 640 buffer_info_init_right(tb, &bi);
875 leaf_paste_in_buffer 641 leaf_paste_in_buffer(&bi, 0, paste_entry_position, tb->insert_size[0], body, zeros_num);
876 (&bi, 0,
877 paste_entry_position,
878 tb->insert_size[0],
879 body, zeros_num);
880 /* paste entry */ 642 /* paste entry */
881 leaf_paste_entries(&bi, 643 leaf_paste_entries(&bi, 0, paste_entry_position, 1,
882 0, 644 (struct reiserfs_de_head *) body,
883 paste_entry_position, 645 body + DEH_SIZE, tb->insert_size[0]);
884 1, 646
885 (struct 647 if (paste_entry_position == 0) {
886 reiserfs_de_head
887 *)
888 body,
889 body
890 +
891 DEH_SIZE,
892 tb->
893 insert_size
894 [0]
895 );
896
897 if (paste_entry_position
898 == 0) {
899 /* change delimiting keys */ 648 /* change delimiting keys */
900 replace_key(tb, 649 replace_key(tb, tb->CFR[0], tb->rkey[0], tb->R[0],0);
901 tb->
902 CFR
903 [0],
904 tb->
905 rkey
906 [0],
907 tb->
908 R
909 [0],
910 0);
911 } 650 }
912 651
913 tb->insert_size[0] = 0; 652 tb->insert_size[0] = 0;
914 pos_in_item++; 653 pos_in_item++;
915 } else { /* new directory entry doesn't fall into R[0] */ 654 } else { /* new directory entry doesn't fall into R[0] */
916 655
917 leaf_shift_right(tb, 656 leaf_shift_right(tb, tb->rnum[0], tb->rbytes);
918 tb->
919 rnum
920 [0],
921 tb->
922 rbytes);
923 } 657 }
924 } else { /* regular object */ 658 } else { /* regular object */
925 659
926 int n_shift, n_rem, 660 int n_shift, n_rem, r_zeros_number;
927 r_zeros_number;
928 const char *r_body; 661 const char *r_body;
929 662
930 /* Calculate number of bytes which must be shifted from appended item */ 663 /* Calculate number of bytes which must be shifted from appended item */
931 if ((n_shift = 664 if ((n_shift = tb->rbytes - tb->insert_size[0]) < 0)
932 tb->rbytes -
933 tb->insert_size[0]) < 0)
934 n_shift = 0; 665 n_shift = 0;
935 666
936 RFALSE(pos_in_item != 667 RFALSE(pos_in_item != ih_item_len
937 ih_item_len 668 (B_N_PITEM_HEAD(tbS0, item_pos)),
938 (B_N_PITEM_HEAD
939 (tbS0, item_pos)),
940 "PAP-12155: invalid position to paste. ih_item_len=%d, pos_in_item=%d", 669 "PAP-12155: invalid position to paste. ih_item_len=%d, pos_in_item=%d",
941 pos_in_item, 670 pos_in_item, ih_item_len
942 ih_item_len 671 (B_N_PITEM_HEAD(tbS0, item_pos)));
943 (B_N_PITEM_HEAD 672
944 (tbS0, item_pos))); 673 leaf_shift_right(tb, tb->rnum[0], n_shift);
945
946 leaf_shift_right(tb,
947 tb->rnum[0],
948 n_shift);
949 /* Calculate number of bytes which must remain in body after appending to R[0] */ 674 /* Calculate number of bytes which must remain in body after appending to R[0] */
950 if ((n_rem = 675 if ((n_rem = tb->insert_size[0] - tb->rbytes) < 0)
951 tb->insert_size[0] -
952 tb->rbytes) < 0)
953 n_rem = 0; 676 n_rem = 0;
954 677
955 { 678 {
956 int version; 679 int version;
957 unsigned long temp_rem = 680 unsigned long temp_rem = n_rem;
958 n_rem; 681
959 682 version = ih_version(B_N_PITEM_HEAD(tb->R[0], 0));
960 version = 683 if (is_indirect_le_key(version, B_N_PKEY(tb->R[0], 0))) {
961 ih_version 684 temp_rem = n_rem << (tb->tb_sb->s_blocksize_bits - UNFM_P_SHIFT);
962 (B_N_PITEM_HEAD
963 (tb->R[0], 0));
964 if (is_indirect_le_key
965 (version,
966 B_N_PKEY(tb->R[0],
967 0))) {
968 temp_rem =
969 n_rem <<
970 (tb->tb_sb->
971 s_blocksize_bits
972 -
973 UNFM_P_SHIFT);
974 } 685 }
975 set_le_key_k_offset 686 set_le_key_k_offset(version, B_N_PKEY(tb->R[0], 0),
976 (version, 687 le_key_k_offset(version, B_N_PKEY(tb->R[0], 0)) + temp_rem);
977 B_N_PKEY(tb->R[0], 688 set_le_key_k_offset(version, B_N_PDELIM_KEY(tb->CFR[0], tb->rkey[0]),
978 0), 689 le_key_k_offset(version, B_N_PDELIM_KEY(tb->CFR[0], tb->rkey[0])) + temp_rem);
979 le_key_k_offset
980 (version,
981 B_N_PKEY(tb->R[0],
982 0)) +
983 temp_rem);
984 set_le_key_k_offset
985 (version,
986 B_N_PDELIM_KEY(tb->
987 CFR
988 [0],
989 tb->
990 rkey
991 [0]),
992 le_key_k_offset
993 (version,
994 B_N_PDELIM_KEY
995 (tb->CFR[0],
996 tb->rkey[0])) +
997 temp_rem);
998 } 690 }
999/* k_offset (B_N_PKEY(tb->R[0],0)) += n_rem; 691/* k_offset (B_N_PKEY(tb->R[0],0)) += n_rem;
1000 k_offset (B_N_PDELIM_KEY(tb->CFR[0],tb->rkey[0])) += n_rem;*/ 692 k_offset (B_N_PDELIM_KEY(tb->CFR[0],tb->rkey[0])) += n_rem;*/
1001 do_balance_mark_internal_dirty 693 do_balance_mark_internal_dirty(tb, tb->CFR[0], 0);
1002 (tb, tb->CFR[0], 0);
1003 694
1004 /* Append part of body into R[0] */ 695 /* Append part of body into R[0] */
1005 buffer_info_init_right(tb, &bi); 696 buffer_info_init_right(tb, &bi);
1006 if (n_rem > zeros_num) { 697 if (n_rem > zeros_num) {
1007 r_zeros_number = 0; 698 r_zeros_number = 0;
1008 r_body = 699 r_body = body + n_rem - zeros_num;
1009 body + n_rem -
1010 zeros_num;
1011 } else { 700 } else {
1012 r_body = body; 701 r_body = body;
1013 r_zeros_number = 702 r_zeros_number = zeros_num - n_rem;
1014 zeros_num - n_rem; 703 zeros_num -= r_zeros_number;
1015 zeros_num -=
1016 r_zeros_number;
1017 } 704 }
1018 705
1019 leaf_paste_in_buffer(&bi, 0, 706 leaf_paste_in_buffer(&bi, 0, n_shift,
1020 n_shift, 707 tb->insert_size[0] - n_rem,
1021 tb-> 708 r_body, r_zeros_number);
1022 insert_size 709
1023 [0] - 710 if (is_indirect_le_ih(B_N_PITEM_HEAD(tb->R[0], 0))) {
1024 n_rem,
1025 r_body,
1026 r_zeros_number);
1027
1028 if (is_indirect_le_ih
1029 (B_N_PITEM_HEAD
1030 (tb->R[0], 0))) {
1031#if 0 711#if 0
1032 RFALSE(n_rem, 712 RFALSE(n_rem,
1033 "PAP-12160: paste more than one unformatted node pointer"); 713 "PAP-12160: paste more than one unformatted node pointer");
1034#endif 714#endif
1035 set_ih_free_space 715 set_ih_free_space(B_N_PITEM_HEAD(tb->R[0], 0), 0);
1036 (B_N_PITEM_HEAD
1037 (tb->R[0], 0), 0);
1038 } 716 }
1039 tb->insert_size[0] = n_rem; 717 tb->insert_size[0] = n_rem;
1040 if (!n_rem) 718 if (!n_rem)
@@ -1044,58 +722,28 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1044 722
1045 struct item_head *pasted; 723 struct item_head *pasted;
1046 724
1047 ret_val = 725 ret_val = leaf_shift_right(tb, tb->rnum[0], tb->rbytes);
1048 leaf_shift_right(tb, tb->rnum[0],
1049 tb->rbytes);
1050 /* append item in R[0] */ 726 /* append item in R[0] */
1051 if (pos_in_item >= 0) { 727 if (pos_in_item >= 0) {
1052 buffer_info_init_right(tb, &bi); 728 buffer_info_init_right(tb, &bi);
1053 leaf_paste_in_buffer(&bi, 729 leaf_paste_in_buffer(&bi, item_pos - n + tb->rnum[0], pos_in_item,
1054 item_pos - 730 tb->insert_size[0], body, zeros_num);
1055 n +
1056 tb->
1057 rnum[0],
1058 pos_in_item,
1059 tb->
1060 insert_size
1061 [0], body,
1062 zeros_num);
1063 } 731 }
1064 732
1065 /* paste new entry, if item is directory item */ 733 /* paste new entry, if item is directory item */
1066 pasted = 734 pasted = B_N_PITEM_HEAD(tb->R[0], item_pos - n + tb->rnum[0]);
1067 B_N_PITEM_HEAD(tb->R[0], 735 if (is_direntry_le_ih(pasted) && pos_in_item >= 0) {
1068 item_pos - n + 736 leaf_paste_entries(&bi, item_pos - n + tb->rnum[0],
1069 tb->rnum[0]); 737 pos_in_item, 1,
1070 if (is_direntry_le_ih(pasted) 738 (struct reiserfs_de_head *) body,
1071 && pos_in_item >= 0) { 739 body + DEH_SIZE, tb->insert_size[0]);
1072 leaf_paste_entries(&bi,
1073 item_pos -
1074 n +
1075 tb->rnum[0],
1076 pos_in_item,
1077 1,
1078 (struct
1079 reiserfs_de_head
1080 *)body,
1081 body +
1082 DEH_SIZE,
1083 tb->
1084 insert_size
1085 [0]
1086 );
1087 if (!pos_in_item) { 740 if (!pos_in_item) {
1088 741
1089 RFALSE(item_pos - n + 742 RFALSE(item_pos - n + tb->rnum[0],
1090 tb->rnum[0],
1091 "PAP-12165: directory item must be first item of node when pasting is in 0th position"); 743 "PAP-12165: directory item must be first item of node when pasting is in 0th position");
1092 744
1093 /* update delimiting keys */ 745 /* update delimiting keys */
1094 replace_key(tb, 746 replace_key(tb, tb->CFR[0], tb->rkey[0], tb->R[0], 0);
1095 tb->CFR[0],
1096 tb->rkey[0],
1097 tb->R[0],
1098 0);
1099 } 747 }
1100 } 748 }
1101 749
@@ -1111,22 +759,16 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1111 default: /* cases d and t */ 759 default: /* cases d and t */
1112 reiserfs_panic(tb->tb_sb, "PAP-12175", 760 reiserfs_panic(tb->tb_sb, "PAP-12175",
1113 "rnum > 0: unexpected mode: %s(%d)", 761 "rnum > 0: unexpected mode: %s(%d)",
1114 (flag == 762 (flag == M_DELETE) ? "DELETE" : ((flag == M_CUT) ? "CUT" : "UNKNOWN"), flag);
1115 M_DELETE) ? "DELETE" : ((flag ==
1116 M_CUT) ? "CUT"
1117 : "UNKNOWN"),
1118 flag);
1119 } 763 }
1120 764
1121 } 765 }
1122 766
1123 /* tb->rnum[0] > 0 */ 767 /* tb->rnum[0] > 0 */
1124 RFALSE(tb->blknum[0] > 3, 768 RFALSE(tb->blknum[0] > 3,
1125 "PAP-12180: blknum can not be %d. It must be <= 3", 769 "PAP-12180: blknum can not be %d. It must be <= 3", tb->blknum[0]);
1126 tb->blknum[0]);
1127 RFALSE(tb->blknum[0] < 0, 770 RFALSE(tb->blknum[0] < 0,
1128 "PAP-12185: blknum can not be %d. It must be >= 0", 771 "PAP-12185: blknum can not be %d. It must be >= 0", tb->blknum[0]);
1129 tb->blknum[0]);
1130 772
1131 /* if while adding to a node we discover that it is possible to split 773 /* if while adding to a node we discover that it is possible to split
1132 it in two, and merge the left part into the left neighbor and the 774 it in two, and merge the left part into the left neighbor and the
@@ -1177,8 +819,7 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1177 819
1178 if (n - snum[i] < item_pos) { /* new item or it's part falls to first new node S_new[i] */ 820 if (n - snum[i] < item_pos) { /* new item or it's part falls to first new node S_new[i] */
1179 if (item_pos == n - snum[i] + 1 && sbytes[i] != -1) { /* part of new item falls into S_new[i] */ 821 if (item_pos == n - snum[i] + 1 && sbytes[i] != -1) { /* part of new item falls into S_new[i] */
1180 int old_key_comp, old_len, 822 int old_key_comp, old_len, r_zeros_number;
1181 r_zeros_number;
1182 const char *r_body; 823 const char *r_body;
1183 int version; 824 int version;
1184 825
@@ -1192,15 +833,8 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1192 old_len = ih_item_len(ih); 833 old_len = ih_item_len(ih);
1193 834
1194 /* Calculate key component and item length to insert into S_new[i] */ 835 /* Calculate key component and item length to insert into S_new[i] */
1195 set_le_ih_k_offset(ih, 836 set_le_ih_k_offset(ih, le_ih_k_offset(ih) +
1196 le_ih_k_offset(ih) + 837 ((old_len - sbytes[i]) << (is_indirect_le_ih(ih) ? tb->tb_sb-> s_blocksize_bits - UNFM_P_SHIFT : 0)));
1197 ((old_len -
1198 sbytes[i]) <<
1199 (is_indirect_le_ih
1200 (ih) ? tb->tb_sb->
1201 s_blocksize_bits -
1202 UNFM_P_SHIFT :
1203 0)));
1204 838
1205 put_ih_item_len(ih, sbytes[i]); 839 put_ih_item_len(ih, sbytes[i]);
1206 840
@@ -1209,39 +843,29 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1209 843
1210 if ((old_len - sbytes[i]) > zeros_num) { 844 if ((old_len - sbytes[i]) > zeros_num) {
1211 r_zeros_number = 0; 845 r_zeros_number = 0;
1212 r_body = 846 r_body = body + (old_len - sbytes[i]) - zeros_num;
1213 body + (old_len -
1214 sbytes[i]) -
1215 zeros_num;
1216 } else { 847 } else {
1217 r_body = body; 848 r_body = body;
1218 r_zeros_number = 849 r_zeros_number = zeros_num - (old_len - sbytes[i]);
1219 zeros_num - (old_len -
1220 sbytes[i]);
1221 zeros_num -= r_zeros_number; 850 zeros_num -= r_zeros_number;
1222 } 851 }
1223 852
1224 leaf_insert_into_buf(&bi, 0, ih, r_body, 853 leaf_insert_into_buf(&bi, 0, ih, r_body, r_zeros_number);
1225 r_zeros_number);
1226 854
1227 /* Calculate key component and item length to insert into S[i] */ 855 /* Calculate key component and item length to insert into S[i] */
1228 set_le_ih_k_offset(ih, old_key_comp); 856 set_le_ih_k_offset(ih, old_key_comp);
1229 put_ih_item_len(ih, 857 put_ih_item_len(ih, old_len - sbytes[i]);
1230 old_len - sbytes[i]);
1231 tb->insert_size[0] -= sbytes[i]; 858 tb->insert_size[0] -= sbytes[i];
1232 } else { /* whole new item falls into S_new[i] */ 859 } else { /* whole new item falls into S_new[i] */
1233 860
1234 /* Shift snum[0] - 1 items to S_new[i] (sbytes[i] of split item) */ 861 /* Shift snum[0] - 1 items to S_new[i] (sbytes[i] of split item) */
1235 leaf_move_items(LEAF_FROM_S_TO_SNEW, tb, 862 leaf_move_items(LEAF_FROM_S_TO_SNEW, tb,
1236 snum[i] - 1, sbytes[i], 863 snum[i] - 1, sbytes[i], S_new[i]);
1237 S_new[i]);
1238 864
1239 /* Insert new item into S_new[i] */ 865 /* Insert new item into S_new[i] */
1240 buffer_info_init_bh(tb, &bi, S_new[i]); 866 buffer_info_init_bh(tb, &bi, S_new[i]);
1241 leaf_insert_into_buf(&bi, 867 leaf_insert_into_buf(&bi, item_pos - n + snum[i] - 1,
1242 item_pos - n + 868 ih, body, zeros_num);
1243 snum[i] - 1, ih,
1244 body, zeros_num);
1245 869
1246 zeros_num = tb->insert_size[0] = 0; 870 zeros_num = tb->insert_size[0] = 0;
1247 } 871 }
@@ -1268,150 +892,73 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1268 892
1269 int entry_count; 893 int entry_count;
1270 894
1271 entry_count = 895 entry_count = ih_entry_count(aux_ih);
1272 ih_entry_count(aux_ih);
1273 896
1274 if (entry_count - sbytes[i] < 897 if (entry_count - sbytes[i] < pos_in_item && pos_in_item <= entry_count) {
1275 pos_in_item
1276 && pos_in_item <=
1277 entry_count) {
1278 /* new directory entry falls into S_new[i] */ 898 /* new directory entry falls into S_new[i] */
1279 899
1280 RFALSE(!tb-> 900 RFALSE(!tb->insert_size[0], "PAP-12215: insert_size is already 0");
1281 insert_size[0], 901 RFALSE(sbytes[i] - 1 >= entry_count,
1282 "PAP-12215: insert_size is already 0");
1283 RFALSE(sbytes[i] - 1 >=
1284 entry_count,
1285 "PAP-12220: there are no so much entries (%d), only %d", 902 "PAP-12220: there are no so much entries (%d), only %d",
1286 sbytes[i] - 1, 903 sbytes[i] - 1, entry_count);
1287 entry_count);
1288 904
1289 /* Shift snum[i]-1 items in whole. Shift sbytes[i] directory entries from directory item number snum[i] */ 905 /* Shift snum[i]-1 items in whole. Shift sbytes[i] directory entries from directory item number snum[i] */
1290 leaf_move_items 906 leaf_move_items(LEAF_FROM_S_TO_SNEW, tb, snum[i], sbytes[i] - 1, S_new[i]);
1291 (LEAF_FROM_S_TO_SNEW,
1292 tb, snum[i],
1293 sbytes[i] - 1,
1294 S_new[i]);
1295 /* Paste given directory entry to directory item */ 907 /* Paste given directory entry to directory item */
1296 buffer_info_init_bh(tb, &bi, S_new[i]); 908 buffer_info_init_bh(tb, &bi, S_new[i]);
1297 leaf_paste_in_buffer 909 leaf_paste_in_buffer(&bi, 0, pos_in_item - entry_count + sbytes[i] - 1,
1298 (&bi, 0, 910 tb->insert_size[0], body, zeros_num);
1299 pos_in_item -
1300 entry_count +
1301 sbytes[i] - 1,
1302 tb->insert_size[0],
1303 body, zeros_num);
1304 /* paste new directory entry */ 911 /* paste new directory entry */
1305 leaf_paste_entries(&bi, 912 leaf_paste_entries(&bi, 0, pos_in_item - entry_count + sbytes[i] - 1, 1,
1306 0, 913 (struct reiserfs_de_head *) body,
1307 pos_in_item 914 body + DEH_SIZE, tb->insert_size[0]);
1308 -
1309 entry_count
1310 +
1311 sbytes
1312 [i] -
1313 1, 1,
1314 (struct
1315 reiserfs_de_head
1316 *)
1317 body,
1318 body
1319 +
1320 DEH_SIZE,
1321 tb->
1322 insert_size
1323 [0]
1324 );
1325 tb->insert_size[0] = 0; 915 tb->insert_size[0] = 0;
1326 pos_in_item++; 916 pos_in_item++;
1327 } else { /* new directory entry doesn't fall into S_new[i] */ 917 } else { /* new directory entry doesn't fall into S_new[i] */
1328 leaf_move_items 918 leaf_move_items(LEAF_FROM_S_TO_SNEW,tb, snum[i], sbytes[i], S_new[i]);
1329 (LEAF_FROM_S_TO_SNEW,
1330 tb, snum[i],
1331 sbytes[i],
1332 S_new[i]);
1333 } 919 }
1334 } else { /* regular object */ 920 } else { /* regular object */
1335 921
1336 int n_shift, n_rem, 922 int n_shift, n_rem, r_zeros_number;
1337 r_zeros_number;
1338 const char *r_body; 923 const char *r_body;
1339 924
1340 RFALSE(pos_in_item != 925 RFALSE(pos_in_item != ih_item_len(B_N_PITEM_HEAD(tbS0, item_pos)) || tb->insert_size[0] <= 0,
1341 ih_item_len
1342 (B_N_PITEM_HEAD
1343 (tbS0, item_pos))
1344 || tb->insert_size[0] <=
1345 0,
1346 "PAP-12225: item too short or insert_size <= 0"); 926 "PAP-12225: item too short or insert_size <= 0");
1347 927
1348 /* Calculate number of bytes which must be shifted from appended item */ 928 /* Calculate number of bytes which must be shifted from appended item */
1349 n_shift = 929 n_shift = sbytes[i] - tb->insert_size[0];
1350 sbytes[i] -
1351 tb->insert_size[0];
1352 if (n_shift < 0) 930 if (n_shift < 0)
1353 n_shift = 0; 931 n_shift = 0;
1354 leaf_move_items 932 leaf_move_items(LEAF_FROM_S_TO_SNEW, tb, snum[i], n_shift, S_new[i]);
1355 (LEAF_FROM_S_TO_SNEW, tb,
1356 snum[i], n_shift,
1357 S_new[i]);
1358 933
1359 /* Calculate number of bytes which must remain in body after append to S_new[i] */ 934 /* Calculate number of bytes which must remain in body after append to S_new[i] */
1360 n_rem = 935 n_rem = tb->insert_size[0] - sbytes[i];
1361 tb->insert_size[0] -
1362 sbytes[i];
1363 if (n_rem < 0) 936 if (n_rem < 0)
1364 n_rem = 0; 937 n_rem = 0;
1365 /* Append part of body into S_new[0] */ 938 /* Append part of body into S_new[0] */
1366 buffer_info_init_bh(tb, &bi, S_new[i]); 939 buffer_info_init_bh(tb, &bi, S_new[i]);
1367 if (n_rem > zeros_num) { 940 if (n_rem > zeros_num) {
1368 r_zeros_number = 0; 941 r_zeros_number = 0;
1369 r_body = 942 r_body = body + n_rem - zeros_num;
1370 body + n_rem -
1371 zeros_num;
1372 } else { 943 } else {
1373 r_body = body; 944 r_body = body;
1374 r_zeros_number = 945 r_zeros_number = zeros_num - n_rem;
1375 zeros_num - n_rem; 946 zeros_num -= r_zeros_number;
1376 zeros_num -=
1377 r_zeros_number;
1378 } 947 }
1379 948
1380 leaf_paste_in_buffer(&bi, 0, 949 leaf_paste_in_buffer(&bi, 0, n_shift,
1381 n_shift, 950 tb->insert_size[0] - n_rem,
1382 tb-> 951 r_body, r_zeros_number);
1383 insert_size
1384 [0] -
1385 n_rem,
1386 r_body,
1387 r_zeros_number);
1388 { 952 {
1389 struct item_head *tmp; 953 struct item_head *tmp;
1390 954
1391 tmp = 955 tmp = B_N_PITEM_HEAD(S_new[i], 0);
1392 B_N_PITEM_HEAD(S_new
1393 [i],
1394 0);
1395 if (is_indirect_le_ih 956 if (is_indirect_le_ih
1396 (tmp)) { 957 (tmp)) {
1397 set_ih_free_space 958 set_ih_free_space(tmp, 0);
1398 (tmp, 0); 959 set_le_ih_k_offset(tmp, le_ih_k_offset(tmp) + (n_rem << (tb->tb_sb->s_blocksize_bits - UNFM_P_SHIFT)));
1399 set_le_ih_k_offset
1400 (tmp,
1401 le_ih_k_offset
1402 (tmp) +
1403 (n_rem <<
1404 (tb->
1405 tb_sb->
1406 s_blocksize_bits
1407 -
1408 UNFM_P_SHIFT)));
1409 } else { 960 } else {
1410 set_le_ih_k_offset 961 set_le_ih_k_offset(tmp, le_ih_k_offset(tmp) + n_rem);
1411 (tmp,
1412 le_ih_k_offset
1413 (tmp) +
1414 n_rem);
1415 } 962 }
1416 } 963 }
1417 964
@@ -1426,8 +973,7 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1426 struct item_head *pasted; 973 struct item_head *pasted;
1427 974
1428#ifdef CONFIG_REISERFS_CHECK 975#ifdef CONFIG_REISERFS_CHECK
1429 struct item_head *ih_check = 976 struct item_head *ih_check = B_N_PITEM_HEAD(tbS0, item_pos);
1430 B_N_PITEM_HEAD(tbS0, item_pos);
1431 977
1432 if (!is_direntry_le_ih(ih_check) 978 if (!is_direntry_le_ih(ih_check)
1433 && (pos_in_item != ih_item_len(ih_check) 979 && (pos_in_item != ih_item_len(ih_check)
@@ -1439,8 +985,7 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1439 "to ih_item_len"); 985 "to ih_item_len");
1440#endif /* CONFIG_REISERFS_CHECK */ 986#endif /* CONFIG_REISERFS_CHECK */
1441 987
1442 leaf_mi = 988 leaf_mi = leaf_move_items(LEAF_FROM_S_TO_SNEW,
1443 leaf_move_items(LEAF_FROM_S_TO_SNEW,
1444 tb, snum[i], 989 tb, snum[i],
1445 sbytes[i], 990 sbytes[i],
1446 S_new[i]); 991 S_new[i]);
@@ -1452,30 +997,19 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1452 /* paste into item */ 997 /* paste into item */
1453 buffer_info_init_bh(tb, &bi, S_new[i]); 998 buffer_info_init_bh(tb, &bi, S_new[i]);
1454 leaf_paste_in_buffer(&bi, 999 leaf_paste_in_buffer(&bi,
1455 item_pos - n + 1000 item_pos - n + snum[i],
1456 snum[i],
1457 pos_in_item, 1001 pos_in_item,
1458 tb->insert_size[0], 1002 tb->insert_size[0],
1459 body, zeros_num); 1003 body, zeros_num);
1460 1004
1461 pasted = 1005 pasted = B_N_PITEM_HEAD(S_new[i], item_pos - n + snum[i]);
1462 B_N_PITEM_HEAD(S_new[i],
1463 item_pos - n +
1464 snum[i]);
1465 if (is_direntry_le_ih(pasted)) { 1006 if (is_direntry_le_ih(pasted)) {
1466 leaf_paste_entries(&bi, 1007 leaf_paste_entries(&bi,
1467 item_pos - 1008 item_pos - n + snum[i],
1468 n + snum[i], 1009 pos_in_item, 1,
1469 pos_in_item, 1010 (struct reiserfs_de_head *)body,
1470 1, 1011 body + DEH_SIZE,
1471 (struct 1012 tb->insert_size[0]
1472 reiserfs_de_head
1473 *)body,
1474 body +
1475 DEH_SIZE,
1476 tb->
1477 insert_size
1478 [0]
1479 ); 1013 );
1480 } 1014 }
1481 1015
@@ -1495,11 +1029,7 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1495 default: /* cases d and t */ 1029 default: /* cases d and t */
1496 reiserfs_panic(tb->tb_sb, "PAP-12245", 1030 reiserfs_panic(tb->tb_sb, "PAP-12245",
1497 "blknum > 2: unexpected mode: %s(%d)", 1031 "blknum > 2: unexpected mode: %s(%d)",
1498 (flag == 1032 (flag == M_DELETE) ? "DELETE" : ((flag == M_CUT) ? "CUT" : "UNKNOWN"), flag);
1499 M_DELETE) ? "DELETE" : ((flag ==
1500 M_CUT) ? "CUT"
1501 : "UNKNOWN"),
1502 flag);
1503 } 1033 }
1504 1034
1505 memcpy(insert_key + i, B_N_PKEY(S_new[i], 0), KEY_SIZE); 1035 memcpy(insert_key + i, B_N_PKEY(S_new[i], 0), KEY_SIZE);
@@ -1524,9 +1054,7 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1524 /* If we insert the first key change the delimiting key */ 1054 /* If we insert the first key change the delimiting key */
1525 if (item_pos == 0) { 1055 if (item_pos == 0) {
1526 if (tb->CFL[0]) /* can be 0 in reiserfsck */ 1056 if (tb->CFL[0]) /* can be 0 in reiserfsck */
1527 replace_key(tb, tb->CFL[0], tb->lkey[0], 1057 replace_key(tb, tb->CFL[0], tb->lkey[0], tbS0, 0);
1528 tbS0, 0);
1529
1530 } 1058 }
1531 break; 1059 break;
1532 1060
@@ -1536,53 +1064,27 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1536 pasted = B_N_PITEM_HEAD(tbS0, item_pos); 1064 pasted = B_N_PITEM_HEAD(tbS0, item_pos);
1537 /* when directory, may be new entry already pasted */ 1065 /* when directory, may be new entry already pasted */
1538 if (is_direntry_le_ih(pasted)) { 1066 if (is_direntry_le_ih(pasted)) {
1539 if (pos_in_item >= 0 && 1067 if (pos_in_item >= 0 && pos_in_item <= ih_entry_count(pasted)) {
1540 pos_in_item <=
1541 ih_entry_count(pasted)) {
1542 1068
1543 RFALSE(!tb->insert_size[0], 1069 RFALSE(!tb->insert_size[0],
1544 "PAP-12260: insert_size is 0 already"); 1070 "PAP-12260: insert_size is 0 already");
1545 1071
1546 /* prepare space */ 1072 /* prepare space */
1547 buffer_info_init_tbS0(tb, &bi); 1073 buffer_info_init_tbS0(tb, &bi);
1548 leaf_paste_in_buffer(&bi, 1074 leaf_paste_in_buffer(&bi, item_pos, pos_in_item,
1549 item_pos, 1075 tb->insert_size[0], body,
1550 pos_in_item,
1551 tb->
1552 insert_size
1553 [0], body,
1554 zeros_num); 1076 zeros_num);
1555 1077
1556 /* paste entry */ 1078 /* paste entry */
1557 leaf_paste_entries(&bi, 1079 leaf_paste_entries(&bi, item_pos, pos_in_item, 1,
1558 item_pos, 1080 (struct reiserfs_de_head *)body,
1559 pos_in_item, 1081 body + DEH_SIZE,
1560 1, 1082 tb->insert_size[0]);
1561 (struct
1562 reiserfs_de_head
1563 *)body,
1564 body +
1565 DEH_SIZE,
1566 tb->
1567 insert_size
1568 [0]
1569 );
1570 if (!item_pos && !pos_in_item) { 1083 if (!item_pos && !pos_in_item) {
1571 RFALSE(!tb->CFL[0] 1084 RFALSE(!tb->CFL[0] || !tb->L[0],
1572 || !tb->L[0],
1573 "PAP-12270: CFL[0]/L[0] must be specified"); 1085 "PAP-12270: CFL[0]/L[0] must be specified");
1574 if (tb->CFL[0]) { 1086 if (tb->CFL[0])
1575 replace_key(tb, 1087 replace_key(tb, tb->CFL[0], tb->lkey[0], tbS0, 0);
1576 tb->
1577 CFL
1578 [0],
1579 tb->
1580 lkey
1581 [0],
1582 tbS0,
1583 0);
1584
1585 }
1586 } 1088 }
1587 tb->insert_size[0] = 0; 1089 tb->insert_size[0] = 0;
1588 } 1090 }
@@ -1593,13 +1095,8 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1593 "PAP-12275: insert size must not be %d", 1095 "PAP-12275: insert size must not be %d",
1594 tb->insert_size[0]); 1096 tb->insert_size[0]);
1595 buffer_info_init_tbS0(tb, &bi); 1097 buffer_info_init_tbS0(tb, &bi);
1596 leaf_paste_in_buffer(&bi, 1098 leaf_paste_in_buffer(&bi, item_pos, pos_in_item,
1597 item_pos, 1099 tb->insert_size[0], body, zeros_num);
1598 pos_in_item,
1599 tb->
1600 insert_size
1601 [0], body,
1602 zeros_num);
1603 1100
1604 if (is_indirect_le_ih(pasted)) { 1101 if (is_indirect_le_ih(pasted)) {
1605#if 0 1102#if 0
@@ -1611,8 +1108,7 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1611 tb-> 1108 tb->
1612 insert_size[0]); 1109 insert_size[0]);
1613#endif 1110#endif
1614 set_ih_free_space 1111 set_ih_free_space(pasted, 0);
1615 (pasted, 0);
1616 } 1112 }
1617 tb->insert_size[0] = 0; 1113 tb->insert_size[0] = 0;
1618 } 1114 }
@@ -1620,8 +1116,7 @@ static int balance_leaf(struct tree_balance *tb, struct item_head *ih, /* item h
1620 else { 1116 else {
1621 if (tb->insert_size[0]) { 1117 if (tb->insert_size[0]) {
1622 print_cur_tb("12285"); 1118 print_cur_tb("12285");
1623 reiserfs_panic(tb-> 1119 reiserfs_panic(tb->tb_sb,
1624 tb_sb,
1625 "PAP-12285", 1120 "PAP-12285",
1626 "insert_size " 1121 "insert_size "
1627 "must be 0 " 1122 "must be 0 "
diff --git a/fs/sync.c b/fs/sync.c
index e8ba024a055b..b28d1dd10e8b 100644
--- a/fs/sync.c
+++ b/fs/sync.c
@@ -27,11 +27,10 @@
27 * wait == 1 case since in that case write_inode() functions do 27 * wait == 1 case since in that case write_inode() functions do
28 * sync_dirty_buffer() and thus effectively write one block at a time. 28 * sync_dirty_buffer() and thus effectively write one block at a time.
29 */ 29 */
30static int __sync_filesystem(struct super_block *sb, int wait, 30static int __sync_filesystem(struct super_block *sb, int wait)
31 unsigned long start)
32{ 31{
33 if (wait) 32 if (wait)
34 sync_inodes_sb(sb, start); 33 sync_inodes_sb(sb);
35 else 34 else
36 writeback_inodes_sb(sb, WB_REASON_SYNC); 35 writeback_inodes_sb(sb, WB_REASON_SYNC);
37 36
@@ -48,7 +47,6 @@ static int __sync_filesystem(struct super_block *sb, int wait,
48int sync_filesystem(struct super_block *sb) 47int sync_filesystem(struct super_block *sb)
49{ 48{
50 int ret; 49 int ret;
51 unsigned long start = jiffies;
52 50
53 /* 51 /*
54 * We need to be protected against the filesystem going from 52 * We need to be protected against the filesystem going from
@@ -62,17 +60,17 @@ int sync_filesystem(struct super_block *sb)
62 if (sb->s_flags & MS_RDONLY) 60 if (sb->s_flags & MS_RDONLY)
63 return 0; 61 return 0;
64 62
65 ret = __sync_filesystem(sb, 0, start); 63 ret = __sync_filesystem(sb, 0);
66 if (ret < 0) 64 if (ret < 0)
67 return ret; 65 return ret;
68 return __sync_filesystem(sb, 1, start); 66 return __sync_filesystem(sb, 1);
69} 67}
70EXPORT_SYMBOL_GPL(sync_filesystem); 68EXPORT_SYMBOL_GPL(sync_filesystem);
71 69
72static void sync_inodes_one_sb(struct super_block *sb, void *arg) 70static void sync_inodes_one_sb(struct super_block *sb, void *arg)
73{ 71{
74 if (!(sb->s_flags & MS_RDONLY)) 72 if (!(sb->s_flags & MS_RDONLY))
75 sync_inodes_sb(sb, *((unsigned long *)arg)); 73 sync_inodes_sb(sb);
76} 74}
77 75
78static void sync_fs_one_sb(struct super_block *sb, void *arg) 76static void sync_fs_one_sb(struct super_block *sb, void *arg)
@@ -104,10 +102,9 @@ static void fdatawait_one_bdev(struct block_device *bdev, void *arg)
104SYSCALL_DEFINE0(sync) 102SYSCALL_DEFINE0(sync)
105{ 103{
106 int nowait = 0, wait = 1; 104 int nowait = 0, wait = 1;
107 unsigned long start = jiffies;
108 105
109 wakeup_flusher_threads(0, WB_REASON_SYNC); 106 wakeup_flusher_threads(0, WB_REASON_SYNC);
110 iterate_supers(sync_inodes_one_sb, &start); 107 iterate_supers(sync_inodes_one_sb, NULL);
111 iterate_supers(sync_fs_one_sb, &nowait); 108 iterate_supers(sync_fs_one_sb, &nowait);
112 iterate_supers(sync_fs_one_sb, &wait); 109 iterate_supers(sync_fs_one_sb, &wait);
113 iterate_bdevs(fdatawrite_one_bdev, NULL); 110 iterate_bdevs(fdatawrite_one_bdev, NULL);
diff --git a/fs/sysfs/mount.c b/fs/sysfs/mount.c
index 6211230814fd..3eaf5c6622eb 100644
--- a/fs/sysfs/mount.c
+++ b/fs/sysfs/mount.c
@@ -27,6 +27,7 @@ static struct dentry *sysfs_mount(struct file_system_type *fs_type,
27{ 27{
28 struct dentry *root; 28 struct dentry *root;
29 void *ns; 29 void *ns;
30 bool new_sb;
30 31
31 if (!(flags & MS_KERNMOUNT)) { 32 if (!(flags & MS_KERNMOUNT)) {
32 if (!capable(CAP_SYS_ADMIN) && !fs_fully_visible(fs_type)) 33 if (!capable(CAP_SYS_ADMIN) && !fs_fully_visible(fs_type))
@@ -37,8 +38,8 @@ static struct dentry *sysfs_mount(struct file_system_type *fs_type,
37 } 38 }
38 39
39 ns = kobj_ns_grab_current(KOBJ_NS_TYPE_NET); 40 ns = kobj_ns_grab_current(KOBJ_NS_TYPE_NET);
40 root = kernfs_mount_ns(fs_type, flags, sysfs_root, ns); 41 root = kernfs_mount_ns(fs_type, flags, sysfs_root, &new_sb, ns);
41 if (IS_ERR(root)) 42 if (IS_ERR(root) || !new_sb)
42 kobj_ns_drop(KOBJ_NS_TYPE_NET, ns); 43 kobj_ns_drop(KOBJ_NS_TYPE_NET, ns);
43 return root; 44 return root;
44} 45}
diff --git a/fs/udf/file.c b/fs/udf/file.c
index c02a27a19c6d..1037637957c7 100644
--- a/fs/udf/file.c
+++ b/fs/udf/file.c
@@ -144,6 +144,7 @@ static ssize_t udf_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
144 size_t count = iocb->ki_nbytes; 144 size_t count = iocb->ki_nbytes;
145 struct udf_inode_info *iinfo = UDF_I(inode); 145 struct udf_inode_info *iinfo = UDF_I(inode);
146 146
147 mutex_lock(&inode->i_mutex);
147 down_write(&iinfo->i_data_sem); 148 down_write(&iinfo->i_data_sem);
148 if (iinfo->i_alloc_type == ICBTAG_FLAG_AD_IN_ICB) { 149 if (iinfo->i_alloc_type == ICBTAG_FLAG_AD_IN_ICB) {
149 if (file->f_flags & O_APPEND) 150 if (file->f_flags & O_APPEND)
@@ -156,6 +157,7 @@ static ssize_t udf_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
156 pos + count)) { 157 pos + count)) {
157 err = udf_expand_file_adinicb(inode); 158 err = udf_expand_file_adinicb(inode);
158 if (err) { 159 if (err) {
160 mutex_unlock(&inode->i_mutex);
159 udf_debug("udf_expand_adinicb: err=%d\n", err); 161 udf_debug("udf_expand_adinicb: err=%d\n", err);
160 return err; 162 return err;
161 } 163 }
@@ -169,9 +171,17 @@ static ssize_t udf_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
169 } else 171 } else
170 up_write(&iinfo->i_data_sem); 172 up_write(&iinfo->i_data_sem);
171 173
172 retval = generic_file_aio_write(iocb, iov, nr_segs, ppos); 174 retval = __generic_file_aio_write(iocb, iov, nr_segs, &iocb->ki_pos);
173 if (retval > 0) 175 mutex_unlock(&inode->i_mutex);
176
177 if (retval > 0) {
178 ssize_t err;
179
174 mark_inode_dirty(inode); 180 mark_inode_dirty(inode);
181 err = generic_write_sync(file, iocb->ki_pos - retval, retval);
182 if (err < 0)
183 retval = err;
184 }
175 185
176 return retval; 186 return retval;
177} 187}
diff --git a/fs/udf/inode.c b/fs/udf/inode.c
index 062b7925bca0..982ce05c87ed 100644
--- a/fs/udf/inode.c
+++ b/fs/udf/inode.c
@@ -265,6 +265,7 @@ int udf_expand_file_adinicb(struct inode *inode)
265 .nr_to_write = 1, 265 .nr_to_write = 1,
266 }; 266 };
267 267
268 WARN_ON_ONCE(!mutex_is_locked(&inode->i_mutex));
268 if (!iinfo->i_lenAlloc) { 269 if (!iinfo->i_lenAlloc) {
269 if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_USE_SHORT_AD)) 270 if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_USE_SHORT_AD))
270 iinfo->i_alloc_type = ICBTAG_FLAG_AD_SHORT; 271 iinfo->i_alloc_type = ICBTAG_FLAG_AD_SHORT;
diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
index f35d5c953ff9..9ddfb8190ca1 100644
--- a/fs/xfs/xfs_iops.c
+++ b/fs/xfs/xfs_iops.c
@@ -705,7 +705,6 @@ xfs_setattr_size(
705{ 705{
706 struct xfs_mount *mp = ip->i_mount; 706 struct xfs_mount *mp = ip->i_mount;
707 struct inode *inode = VFS_I(ip); 707 struct inode *inode = VFS_I(ip);
708 int mask = iattr->ia_valid;
709 xfs_off_t oldsize, newsize; 708 xfs_off_t oldsize, newsize;
710 struct xfs_trans *tp; 709 struct xfs_trans *tp;
711 int error; 710 int error;
@@ -726,8 +725,8 @@ xfs_setattr_size(
726 725
727 ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL)); 726 ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
728 ASSERT(S_ISREG(ip->i_d.di_mode)); 727 ASSERT(S_ISREG(ip->i_d.di_mode));
729 ASSERT((mask & (ATTR_UID|ATTR_GID|ATTR_ATIME|ATTR_ATIME_SET| 728 ASSERT((iattr->ia_valid & (ATTR_UID|ATTR_GID|ATTR_ATIME|ATTR_ATIME_SET|
730 ATTR_MTIME_SET|ATTR_KILL_PRIV|ATTR_TIMES_SET)) == 0); 729 ATTR_MTIME_SET|ATTR_KILL_PRIV|ATTR_TIMES_SET)) == 0);
731 730
732 oldsize = inode->i_size; 731 oldsize = inode->i_size;
733 newsize = iattr->ia_size; 732 newsize = iattr->ia_size;
@@ -736,7 +735,7 @@ xfs_setattr_size(
736 * Short circuit the truncate case for zero length files. 735 * Short circuit the truncate case for zero length files.
737 */ 736 */
738 if (newsize == 0 && oldsize == 0 && ip->i_d.di_nextents == 0) { 737 if (newsize == 0 && oldsize == 0 && ip->i_d.di_nextents == 0) {
739 if (!(mask & (ATTR_CTIME|ATTR_MTIME))) 738 if (!(iattr->ia_valid & (ATTR_CTIME|ATTR_MTIME)))
740 return 0; 739 return 0;
741 740
742 /* 741 /*
@@ -824,10 +823,11 @@ xfs_setattr_size(
824 * these flags set. For all other operations the VFS set these flags 823 * these flags set. For all other operations the VFS set these flags
825 * explicitly if it wants a timestamp update. 824 * explicitly if it wants a timestamp update.
826 */ 825 */
827 if (newsize != oldsize && (!(mask & (ATTR_CTIME | ATTR_MTIME)))) { 826 if (newsize != oldsize &&
827 !(iattr->ia_valid & (ATTR_CTIME | ATTR_MTIME))) {
828 iattr->ia_ctime = iattr->ia_mtime = 828 iattr->ia_ctime = iattr->ia_mtime =
829 current_fs_time(inode->i_sb); 829 current_fs_time(inode->i_sb);
830 mask |= ATTR_CTIME | ATTR_MTIME; 830 iattr->ia_valid |= ATTR_CTIME | ATTR_MTIME;
831 } 831 }
832 832
833 /* 833 /*
@@ -863,9 +863,9 @@ xfs_setattr_size(
863 xfs_inode_clear_eofblocks_tag(ip); 863 xfs_inode_clear_eofblocks_tag(ip);
864 } 864 }
865 865
866 if (mask & ATTR_MODE) 866 if (iattr->ia_valid & ATTR_MODE)
867 xfs_setattr_mode(ip, iattr); 867 xfs_setattr_mode(ip, iattr);
868 if (mask & (ATTR_ATIME|ATTR_CTIME|ATTR_MTIME)) 868 if (iattr->ia_valid & (ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
869 xfs_setattr_time(ip, iattr); 869 xfs_setattr_time(ip, iattr);
870 870
871 xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE); 871 xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
diff --git a/fs/xfs/xfs_log_cil.c b/fs/xfs/xfs_log_cil.c
index cdebd832c3db..4ef6fdbced78 100644
--- a/fs/xfs/xfs_log_cil.c
+++ b/fs/xfs/xfs_log_cil.c
@@ -205,16 +205,25 @@ xlog_cil_insert_format_items(
205 /* 205 /*
206 * We 64-bit align the length of each iovec so that the start 206 * We 64-bit align the length of each iovec so that the start
207 * of the next one is naturally aligned. We'll need to 207 * of the next one is naturally aligned. We'll need to
208 * account for that slack space here. 208 * account for that slack space here. Then round nbytes up
209 * to 64-bit alignment so that the initial buffer alignment is
210 * easy to calculate and verify.
209 */ 211 */
210 nbytes += niovecs * sizeof(uint64_t); 212 nbytes += niovecs * sizeof(uint64_t);
213 nbytes = round_up(nbytes, sizeof(uint64_t));
211 214
212 /* grab the old item if it exists for reservation accounting */ 215 /* grab the old item if it exists for reservation accounting */
213 old_lv = lip->li_lv; 216 old_lv = lip->li_lv;
214 217
215 /* calc buffer size */ 218 /*
216 buf_size = sizeof(struct xfs_log_vec) + nbytes + 219 * The data buffer needs to start 64-bit aligned, so round up
217 niovecs * sizeof(struct xfs_log_iovec); 220 * that space to ensure we can align it appropriately and not
221 * overrun the buffer.
222 */
223 buf_size = nbytes +
224 round_up((sizeof(struct xfs_log_vec) +
225 niovecs * sizeof(struct xfs_log_iovec)),
226 sizeof(uint64_t));
218 227
219 /* compare to existing item size */ 228 /* compare to existing item size */
220 if (lip->li_lv && buf_size <= lip->li_lv->lv_size) { 229 if (lip->li_lv && buf_size <= lip->li_lv->lv_size) {
@@ -251,6 +260,8 @@ xlog_cil_insert_format_items(
251 /* The allocated data region lies beyond the iovec region */ 260 /* The allocated data region lies beyond the iovec region */
252 lv->lv_buf_len = 0; 261 lv->lv_buf_len = 0;
253 lv->lv_buf = (char *)lv + buf_size - nbytes; 262 lv->lv_buf = (char *)lv + buf_size - nbytes;
263 ASSERT(IS_ALIGNED((unsigned long)lv->lv_buf, sizeof(uint64_t)));
264
254 lip->li_ops->iop_format(lip, lv); 265 lip->li_ops->iop_format(lip, lv);
255insert: 266insert:
256 ASSERT(lv->lv_buf_len <= nbytes); 267 ASSERT(lv->lv_buf_len <= nbytes);
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 02df7b408a26..f96c05669a9e 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -282,22 +282,29 @@ xfs_readsb(
282 struct xfs_sb *sbp = &mp->m_sb; 282 struct xfs_sb *sbp = &mp->m_sb;
283 int error; 283 int error;
284 int loud = !(flags & XFS_MFSI_QUIET); 284 int loud = !(flags & XFS_MFSI_QUIET);
285 const struct xfs_buf_ops *buf_ops;
285 286
286 ASSERT(mp->m_sb_bp == NULL); 287 ASSERT(mp->m_sb_bp == NULL);
287 ASSERT(mp->m_ddev_targp != NULL); 288 ASSERT(mp->m_ddev_targp != NULL);
288 289
289 /* 290 /*
291 * For the initial read, we must guess at the sector
292 * size based on the block device. It's enough to
293 * get the sb_sectsize out of the superblock and
294 * then reread with the proper length.
295 * We don't verify it yet, because it may not be complete.
296 */
297 sector_size = xfs_getsize_buftarg(mp->m_ddev_targp);
298 buf_ops = NULL;
299
300 /*
290 * Allocate a (locked) buffer to hold the superblock. 301 * Allocate a (locked) buffer to hold the superblock.
291 * This will be kept around at all times to optimize 302 * This will be kept around at all times to optimize
292 * access to the superblock. 303 * access to the superblock.
293 */ 304 */
294 sector_size = xfs_getsize_buftarg(mp->m_ddev_targp);
295
296reread: 305reread:
297 bp = xfs_buf_read_uncached(mp->m_ddev_targp, XFS_SB_DADDR, 306 bp = xfs_buf_read_uncached(mp->m_ddev_targp, XFS_SB_DADDR,
298 BTOBB(sector_size), 0, 307 BTOBB(sector_size), 0, buf_ops);
299 loud ? &xfs_sb_buf_ops
300 : &xfs_sb_quiet_buf_ops);
301 if (!bp) { 308 if (!bp) {
302 if (loud) 309 if (loud)
303 xfs_warn(mp, "SB buffer read failed"); 310 xfs_warn(mp, "SB buffer read failed");
@@ -328,12 +335,13 @@ reread:
328 } 335 }
329 336
330 /* 337 /*
331 * If device sector size is smaller than the superblock size, 338 * Re-read the superblock so the buffer is correctly sized,
332 * re-read the superblock so the buffer is correctly sized. 339 * and properly verified.
333 */ 340 */
334 if (sector_size < sbp->sb_sectsize) { 341 if (buf_ops == NULL) {
335 xfs_buf_relse(bp); 342 xfs_buf_relse(bp);
336 sector_size = sbp->sb_sectsize; 343 sector_size = sbp->sb_sectsize;
344 buf_ops = loud ? &xfs_sb_buf_ops : &xfs_sb_quiet_buf_ops;
337 goto reread; 345 goto reread;
338 } 346 }
339 347
diff --git a/fs/xfs/xfs_sb.c b/fs/xfs/xfs_sb.c
index b7c9aea77f8f..1e116794bb66 100644
--- a/fs/xfs/xfs_sb.c
+++ b/fs/xfs/xfs_sb.c
@@ -295,8 +295,7 @@ xfs_mount_validate_sb(
295 sbp->sb_dblocks == 0 || 295 sbp->sb_dblocks == 0 ||
296 sbp->sb_dblocks > XFS_MAX_DBLOCKS(sbp) || 296 sbp->sb_dblocks > XFS_MAX_DBLOCKS(sbp) ||
297 sbp->sb_dblocks < XFS_MIN_DBLOCKS(sbp))) { 297 sbp->sb_dblocks < XFS_MIN_DBLOCKS(sbp))) {
298 XFS_CORRUPTION_ERROR("SB sanity check failed", 298 xfs_notice(mp, "SB sanity check failed");
299 XFS_ERRLEVEL_LOW, mp, sbp);
300 return XFS_ERROR(EFSCORRUPTED); 299 return XFS_ERROR(EFSCORRUPTED);
301 } 300 }
302 301
@@ -611,10 +610,10 @@ xfs_sb_read_verify(
611 XFS_SB_VERSION_5) || 610 XFS_SB_VERSION_5) ||
612 dsb->sb_crc != 0)) { 611 dsb->sb_crc != 0)) {
613 612
614 if (!xfs_verify_cksum(bp->b_addr, be16_to_cpu(dsb->sb_sectsize), 613 if (!xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
615 offsetof(struct xfs_sb, sb_crc))) { 614 offsetof(struct xfs_sb, sb_crc))) {
616 /* Only fail bad secondaries on a known V5 filesystem */ 615 /* Only fail bad secondaries on a known V5 filesystem */
617 if (bp->b_bn != XFS_SB_DADDR && 616 if (bp->b_bn == XFS_SB_DADDR ||
618 xfs_sb_version_hascrc(&mp->m_sb)) { 617 xfs_sb_version_hascrc(&mp->m_sb)) {
619 error = EFSCORRUPTED; 618 error = EFSCORRUPTED;
620 goto out_error; 619 goto out_error;
@@ -625,7 +624,7 @@ xfs_sb_read_verify(
625 624
626out_error: 625out_error:
627 if (error) { 626 if (error) {
628 if (error != EWRONGFS) 627 if (error == EFSCORRUPTED)
629 XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, 628 XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
630 mp, bp->b_addr); 629 mp, bp->b_addr);
631 xfs_buf_ioerror(bp, error); 630 xfs_buf_ioerror(bp, error);
@@ -644,7 +643,6 @@ xfs_sb_quiet_read_verify(
644{ 643{
645 struct xfs_dsb *dsb = XFS_BUF_TO_SBP(bp); 644 struct xfs_dsb *dsb = XFS_BUF_TO_SBP(bp);
646 645
647
648 if (dsb->sb_magicnum == cpu_to_be32(XFS_SB_MAGIC)) { 646 if (dsb->sb_magicnum == cpu_to_be32(XFS_SB_MAGIC)) {
649 /* XFS filesystem, verify noisily! */ 647 /* XFS filesystem, verify noisily! */
650 xfs_sb_read_verify(bp); 648 xfs_sb_read_verify(bp);
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index f317488263dd..d971f4932b5d 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -913,7 +913,7 @@ xfs_flush_inodes(
913 struct super_block *sb = mp->m_super; 913 struct super_block *sb = mp->m_super;
914 914
915 if (down_read_trylock(&sb->s_umount)) { 915 if (down_read_trylock(&sb->s_umount)) {
916 sync_inodes_sb(sb, jiffies); 916 sync_inodes_sb(sb);
917 up_read(&sb->s_umount); 917 up_read(&sb->s_umount);
918 } 918 }
919} 919}