aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorJames Bottomley <jejb@mulgrave.il.steeleye.com>2006-08-27 22:59:59 -0400
committerJames Bottomley <jejb@mulgrave.il.steeleye.com>2006-08-27 22:59:59 -0400
commit8ce7a9c159c8c4eb480f0a65c6af753dbf9a1a70 (patch)
treebe59573c0af3617d0cd8a7d61f0ed119e58b1156 /fs
parentd2afb3ae04e36dbc6e9eb2d8bd54406ff7b6b3bd (diff)
parent01da5fd83d6b2c5e36b77539f6cbdd8f49849225 (diff)
Merge ../linux-2.6
Diffstat (limited to 'fs')
-rw-r--r--fs/9p/conv.c6
-rw-r--r--fs/9p/vfs_inode.c6
-rw-r--r--fs/adfs/super.c2
-rw-r--r--fs/befs/linuxvfs.c11
-rw-r--r--fs/block_dev.c114
-rw-r--r--fs/buffer.c7
-rw-r--r--fs/coda/file.c4
-rw-r--r--fs/efs/symlink.c3
-rw-r--r--fs/eventpoll.c4
-rw-r--r--fs/exec.c10
-rw-r--r--fs/ext2/super.c2
-rw-r--r--fs/ext3/balloc.c6
-rw-r--r--fs/ext3/inode.c19
-rw-r--r--fs/ext3/namei.c15
-rw-r--r--fs/freevxfs/vxfs_lookup.c2
-rw-r--r--fs/fuse/control.c4
-rw-r--r--fs/fuse/dir.c47
-rw-r--r--fs/fuse/file.c10
-rw-r--r--fs/fuse/fuse_i.h2
-rw-r--r--fs/fuse/inode.c2
-rw-r--r--fs/inotify_user.c2
-rw-r--r--fs/ioprio.c30
-rw-r--r--fs/jbd/commit.c6
-rw-r--r--fs/jbd/journal.c92
-rw-r--r--fs/jbd/transaction.c9
-rw-r--r--fs/jfs/inode.c16
-rw-r--r--fs/jfs/jfs_inode.h1
-rw-r--r--fs/jfs/super.c118
-rw-r--r--fs/lockd/svclock.c12
-rw-r--r--fs/lockd/svcsubs.c15
-rw-r--r--fs/locks.c6
-rw-r--r--fs/minix/inode.c13
-rw-r--r--fs/namei.c19
-rw-r--r--fs/nfs/file.c8
-rw-r--r--fs/nfs/idmap.c4
-rw-r--r--fs/nfs/namespace.c4
-rw-r--r--fs/nfs/nfs4proc.c29
-rw-r--r--fs/nfs/nfs4xdr.c21
-rw-r--r--fs/nfs/read.c25
-rw-r--r--fs/nfs/write.c2
-rw-r--r--fs/nfsd/nfsfh.c20
-rw-r--r--fs/ocfs2/dlm/dlmmaster.c1
-rw-r--r--fs/ocfs2/dlm/dlmunlock.c43
-rw-r--r--fs/ocfs2/localalloc.c8
-rw-r--r--fs/ocfs2/ocfs2.h2
-rw-r--r--fs/ocfs2/suballoc.c261
-rw-r--r--fs/ocfs2/suballoc.h2
-rw-r--r--fs/ocfs2/super.c8
-rw-r--r--fs/partitions/Kconfig2
-rw-r--r--fs/partitions/sun.c2
-rw-r--r--fs/proc/proc_misc.c2
-rw-r--r--fs/reiserfs/file.c2
-rw-r--r--fs/reiserfs/inode.c26
-rw-r--r--fs/reiserfs/ioctl.c2
-rw-r--r--fs/reiserfs/xattr.c2
-rw-r--r--fs/udf/ialloc.c11
-rw-r--r--fs/udf/super.c9
-rw-r--r--fs/udf/truncate.c64
-rw-r--r--fs/ufs/balloc.c2
-rw-r--r--fs/ufs/inode.c35
-rw-r--r--fs/ufs/namei.c3
-rw-r--r--fs/ufs/truncate.c77
-rw-r--r--fs/ufs/util.c17
-rw-r--r--fs/xfs/linux-2.6/xfs_buf.h4
-rw-r--r--fs/xfs/linux-2.6/xfs_super.c7
-rw-r--r--fs/xfs/quota/xfs_qm_bhv.c19
-rw-r--r--fs/xfs/xfs_alloc.c103
-rw-r--r--fs/xfs/xfs_inode.c17
-rw-r--r--fs/xfs/xfs_log.c12
-rw-r--r--fs/xfs/xfs_vfsops.c2
70 files changed, 1015 insertions, 458 deletions
diff --git a/fs/9p/conv.c b/fs/9p/conv.c
index 1e898144eb7c..56d88c1a09c5 100644
--- a/fs/9p/conv.c
+++ b/fs/9p/conv.c
@@ -673,8 +673,10 @@ struct v9fs_fcall *v9fs_create_tcreate(u32 fid, char *name, u32 perm, u8 mode,
673 struct cbuf *bufp = &buffer; 673 struct cbuf *bufp = &buffer;
674 674
675 size = 4 + 2 + strlen(name) + 4 + 1; /* fid[4] name[s] perm[4] mode[1] */ 675 size = 4 + 2 + strlen(name) + 4 + 1; /* fid[4] name[s] perm[4] mode[1] */
676 if (extended && extension!=NULL) 676 if (extended) {
677 size += 2 + strlen(extension); /* extension[s] */ 677 size += 2 + /* extension[s] */
678 (extension == NULL ? 0 : strlen(extension));
679 }
678 680
679 fc = v9fs_create_common(bufp, size, TCREATE); 681 fc = v9fs_create_common(bufp, size, TCREATE);
680 if (IS_ERR(fc)) 682 if (IS_ERR(fc))
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 2f580a197b8d..eae50c9d6dc4 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -434,11 +434,11 @@ static int v9fs_remove(struct inode *dir, struct dentry *file, int rmdir)
434 result = v9fs_t_remove(v9ses, fid, &fcall); 434 result = v9fs_t_remove(v9ses, fid, &fcall);
435 if (result < 0) { 435 if (result < 0) {
436 PRINT_FCALL_ERROR("remove fails", fcall); 436 PRINT_FCALL_ERROR("remove fails", fcall);
437 } else {
438 v9fs_put_idpool(fid, &v9ses->fidpool);
439 v9fs_fid_destroy(v9fid);
440 } 437 }
441 438
439 v9fs_put_idpool(fid, &v9ses->fidpool);
440 v9fs_fid_destroy(v9fid);
441
442 kfree(fcall); 442 kfree(fcall);
443 return result; 443 return result;
444} 444}
diff --git a/fs/adfs/super.c b/fs/adfs/super.c
index ba1c88af49fe..82011019494c 100644
--- a/fs/adfs/super.c
+++ b/fs/adfs/super.c
@@ -308,7 +308,7 @@ static struct adfs_discmap *adfs_read_map(struct super_block *sb, struct adfs_di
308 if (adfs_checkmap(sb, dm)) 308 if (adfs_checkmap(sb, dm))
309 return dm; 309 return dm;
310 310
311 adfs_error(sb, NULL, "map corrupted"); 311 adfs_error(sb, "map corrupted");
312 312
313error_free: 313error_free:
314 while (--zone >= 0) 314 while (--zone >= 0)
diff --git a/fs/befs/linuxvfs.c b/fs/befs/linuxvfs.c
index fcaeead9696b..50cfca5c7efd 100644
--- a/fs/befs/linuxvfs.c
+++ b/fs/befs/linuxvfs.c
@@ -512,7 +512,11 @@ befs_utf2nls(struct super_block *sb, const char *in,
512 wchar_t uni; 512 wchar_t uni;
513 int unilen, utflen; 513 int unilen, utflen;
514 char *result; 514 char *result;
515 int maxlen = in_len; /* The utf8->nls conversion can't make more chars */ 515 /* The utf8->nls conversion won't make the final nls string bigger
516 * than the utf one, but if the string is pure ascii they'll have the
517 * same width and an extra char is needed to save the additional \0
518 */
519 int maxlen = in_len + 1;
516 520
517 befs_debug(sb, "---> utf2nls()"); 521 befs_debug(sb, "---> utf2nls()");
518 522
@@ -588,7 +592,10 @@ befs_nls2utf(struct super_block *sb, const char *in,
588 wchar_t uni; 592 wchar_t uni;
589 int unilen, utflen; 593 int unilen, utflen;
590 char *result; 594 char *result;
591 int maxlen = 3 * in_len; 595 /* There're nls characters that will translate to 3-chars-wide UTF-8
596 * characters, a additional byte is needed to save the final \0
597 * in special cases */
598 int maxlen = (3 * in_len) + 1;
592 599
593 befs_debug(sb, "---> nls2utf()\n"); 600 befs_debug(sb, "---> nls2utf()\n");
594 601
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 37534573960b..045f98854f14 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -884,6 +884,61 @@ void bd_set_size(struct block_device *bdev, loff_t size)
884} 884}
885EXPORT_SYMBOL(bd_set_size); 885EXPORT_SYMBOL(bd_set_size);
886 886
887static int __blkdev_put(struct block_device *bdev, unsigned int subclass)
888{
889 int ret = 0;
890 struct inode *bd_inode = bdev->bd_inode;
891 struct gendisk *disk = bdev->bd_disk;
892
893 mutex_lock_nested(&bdev->bd_mutex, subclass);
894 lock_kernel();
895 if (!--bdev->bd_openers) {
896 sync_blockdev(bdev);
897 kill_bdev(bdev);
898 }
899 if (bdev->bd_contains == bdev) {
900 if (disk->fops->release)
901 ret = disk->fops->release(bd_inode, NULL);
902 } else {
903 mutex_lock_nested(&bdev->bd_contains->bd_mutex,
904 subclass + 1);
905 bdev->bd_contains->bd_part_count--;
906 mutex_unlock(&bdev->bd_contains->bd_mutex);
907 }
908 if (!bdev->bd_openers) {
909 struct module *owner = disk->fops->owner;
910
911 put_disk(disk);
912 module_put(owner);
913
914 if (bdev->bd_contains != bdev) {
915 kobject_put(&bdev->bd_part->kobj);
916 bdev->bd_part = NULL;
917 }
918 bdev->bd_disk = NULL;
919 bdev->bd_inode->i_data.backing_dev_info = &default_backing_dev_info;
920 if (bdev != bdev->bd_contains)
921 __blkdev_put(bdev->bd_contains, subclass + 1);
922 bdev->bd_contains = NULL;
923 }
924 unlock_kernel();
925 mutex_unlock(&bdev->bd_mutex);
926 bdput(bdev);
927 return ret;
928}
929
930int blkdev_put(struct block_device *bdev)
931{
932 return __blkdev_put(bdev, BD_MUTEX_NORMAL);
933}
934EXPORT_SYMBOL(blkdev_put);
935
936int blkdev_put_partition(struct block_device *bdev)
937{
938 return __blkdev_put(bdev, BD_MUTEX_PARTITION);
939}
940EXPORT_SYMBOL(blkdev_put_partition);
941
887static int 942static int
888blkdev_get_whole(struct block_device *bdev, mode_t mode, unsigned flags); 943blkdev_get_whole(struct block_device *bdev, mode_t mode, unsigned flags);
889 944
@@ -980,7 +1035,7 @@ out_first:
980 bdev->bd_disk = NULL; 1035 bdev->bd_disk = NULL;
981 bdev->bd_inode->i_data.backing_dev_info = &default_backing_dev_info; 1036 bdev->bd_inode->i_data.backing_dev_info = &default_backing_dev_info;
982 if (bdev != bdev->bd_contains) 1037 if (bdev != bdev->bd_contains)
983 blkdev_put(bdev->bd_contains); 1038 __blkdev_put(bdev->bd_contains, BD_MUTEX_WHOLE);
984 bdev->bd_contains = NULL; 1039 bdev->bd_contains = NULL;
985 put_disk(disk); 1040 put_disk(disk);
986 module_put(owner); 1041 module_put(owner);
@@ -1079,63 +1134,6 @@ static int blkdev_open(struct inode * inode, struct file * filp)
1079 return res; 1134 return res;
1080} 1135}
1081 1136
1082static int __blkdev_put(struct block_device *bdev, unsigned int subclass)
1083{
1084 int ret = 0;
1085 struct inode *bd_inode = bdev->bd_inode;
1086 struct gendisk *disk = bdev->bd_disk;
1087
1088 mutex_lock_nested(&bdev->bd_mutex, subclass);
1089 lock_kernel();
1090 if (!--bdev->bd_openers) {
1091 sync_blockdev(bdev);
1092 kill_bdev(bdev);
1093 }
1094 if (bdev->bd_contains == bdev) {
1095 if (disk->fops->release)
1096 ret = disk->fops->release(bd_inode, NULL);
1097 } else {
1098 mutex_lock_nested(&bdev->bd_contains->bd_mutex,
1099 subclass + 1);
1100 bdev->bd_contains->bd_part_count--;
1101 mutex_unlock(&bdev->bd_contains->bd_mutex);
1102 }
1103 if (!bdev->bd_openers) {
1104 struct module *owner = disk->fops->owner;
1105
1106 put_disk(disk);
1107 module_put(owner);
1108
1109 if (bdev->bd_contains != bdev) {
1110 kobject_put(&bdev->bd_part->kobj);
1111 bdev->bd_part = NULL;
1112 }
1113 bdev->bd_disk = NULL;
1114 bdev->bd_inode->i_data.backing_dev_info = &default_backing_dev_info;
1115 if (bdev != bdev->bd_contains)
1116 __blkdev_put(bdev->bd_contains, subclass + 1);
1117 bdev->bd_contains = NULL;
1118 }
1119 unlock_kernel();
1120 mutex_unlock(&bdev->bd_mutex);
1121 bdput(bdev);
1122 return ret;
1123}
1124
1125int blkdev_put(struct block_device *bdev)
1126{
1127 return __blkdev_put(bdev, BD_MUTEX_NORMAL);
1128}
1129
1130EXPORT_SYMBOL(blkdev_put);
1131
1132int blkdev_put_partition(struct block_device *bdev)
1133{
1134 return __blkdev_put(bdev, BD_MUTEX_PARTITION);
1135}
1136
1137EXPORT_SYMBOL(blkdev_put_partition);
1138
1139static int blkdev_close(struct inode * inode, struct file * filp) 1137static int blkdev_close(struct inode * inode, struct file * filp)
1140{ 1138{
1141 struct block_device *bdev = I_BDEV(filp->f_mapping->host); 1139 struct block_device *bdev = I_BDEV(filp->f_mapping->host);
diff --git a/fs/buffer.c b/fs/buffer.c
index 3660dcb97591..71649ef9b658 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -470,13 +470,18 @@ out:
470 pass does the actual I/O. */ 470 pass does the actual I/O. */
471void invalidate_bdev(struct block_device *bdev, int destroy_dirty_buffers) 471void invalidate_bdev(struct block_device *bdev, int destroy_dirty_buffers)
472{ 472{
473 struct address_space *mapping = bdev->bd_inode->i_mapping;
474
475 if (mapping->nrpages == 0)
476 return;
477
473 invalidate_bh_lrus(); 478 invalidate_bh_lrus();
474 /* 479 /*
475 * FIXME: what about destroy_dirty_buffers? 480 * FIXME: what about destroy_dirty_buffers?
476 * We really want to use invalidate_inode_pages2() for 481 * We really want to use invalidate_inode_pages2() for
477 * that, but not until that's cleaned up. 482 * that, but not until that's cleaned up.
478 */ 483 */
479 invalidate_inode_pages(bdev->bd_inode->i_mapping); 484 invalidate_inode_pages(mapping);
480} 485}
481 486
482/* 487/*
diff --git a/fs/coda/file.c b/fs/coda/file.c
index cc66c681bd11..dbfbcfa5b3c0 100644
--- a/fs/coda/file.c
+++ b/fs/coda/file.c
@@ -136,10 +136,8 @@ int coda_open(struct inode *coda_inode, struct file *coda_file)
136 coda_vfs_stat.open++; 136 coda_vfs_stat.open++;
137 137
138 cfi = kmalloc(sizeof(struct coda_file_info), GFP_KERNEL); 138 cfi = kmalloc(sizeof(struct coda_file_info), GFP_KERNEL);
139 if (!cfi) { 139 if (!cfi)
140 unlock_kernel();
141 return -ENOMEM; 140 return -ENOMEM;
142 }
143 141
144 lock_kernel(); 142 lock_kernel();
145 143
diff --git a/fs/efs/symlink.c b/fs/efs/symlink.c
index e249cf733a6b..1d30d2ff440f 100644
--- a/fs/efs/symlink.c
+++ b/fs/efs/symlink.c
@@ -22,7 +22,7 @@ static int efs_symlink_readpage(struct file *file, struct page *page)
22 22
23 err = -ENAMETOOLONG; 23 err = -ENAMETOOLONG;
24 if (size > 2 * EFS_BLOCKSIZE) 24 if (size > 2 * EFS_BLOCKSIZE)
25 goto fail; 25 goto fail_notlocked;
26 26
27 lock_kernel(); 27 lock_kernel();
28 /* read first 512 bytes of link target */ 28 /* read first 512 bytes of link target */
@@ -47,6 +47,7 @@ static int efs_symlink_readpage(struct file *file, struct page *page)
47 return 0; 47 return 0;
48fail: 48fail:
49 unlock_kernel(); 49 unlock_kernel();
50fail_notlocked:
50 SetPageError(page); 51 SetPageError(page);
51 kunmap(page); 52 kunmap(page);
52 unlock_page(page); 53 unlock_page(page);
diff --git a/fs/eventpoll.c b/fs/eventpoll.c
index 19ffb043abbc..3a3567433b92 100644
--- a/fs/eventpoll.c
+++ b/fs/eventpoll.c
@@ -1168,7 +1168,7 @@ static int ep_unlink(struct eventpoll *ep, struct epitem *epi)
1168eexit_1: 1168eexit_1:
1169 1169
1170 DNPRINTK(3, (KERN_INFO "[%p] eventpoll: ep_unlink(%p, %p) = %d\n", 1170 DNPRINTK(3, (KERN_INFO "[%p] eventpoll: ep_unlink(%p, %p) = %d\n",
1171 current, ep, epi->file, error)); 1171 current, ep, epi->ffd.file, error));
1172 1172
1173 return error; 1173 return error;
1174} 1174}
@@ -1236,7 +1236,7 @@ static int ep_poll_callback(wait_queue_t *wait, unsigned mode, int sync, void *k
1236 struct eventpoll *ep = epi->ep; 1236 struct eventpoll *ep = epi->ep;
1237 1237
1238 DNPRINTK(3, (KERN_INFO "[%p] eventpoll: poll_callback(%p) epi=%p ep=%p\n", 1238 DNPRINTK(3, (KERN_INFO "[%p] eventpoll: poll_callback(%p) epi=%p ep=%p\n",
1239 current, epi->file, epi, ep)); 1239 current, epi->ffd.file, epi, ep));
1240 1240
1241 write_lock_irqsave(&ep->lock, flags); 1241 write_lock_irqsave(&ep->lock, flags);
1242 1242
diff --git a/fs/exec.c b/fs/exec.c
index 8344ba73a2a6..54135df2a966 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -486,8 +486,6 @@ struct file *open_exec(const char *name)
486 if (!(nd.mnt->mnt_flags & MNT_NOEXEC) && 486 if (!(nd.mnt->mnt_flags & MNT_NOEXEC) &&
487 S_ISREG(inode->i_mode)) { 487 S_ISREG(inode->i_mode)) {
488 int err = vfs_permission(&nd, MAY_EXEC); 488 int err = vfs_permission(&nd, MAY_EXEC);
489 if (!err && !(inode->i_mode & 0111))
490 err = -EACCES;
491 file = ERR_PTR(err); 489 file = ERR_PTR(err);
492 if (!err) { 490 if (!err) {
493 file = nameidata_to_filp(&nd, O_RDONLY); 491 file = nameidata_to_filp(&nd, O_RDONLY);
@@ -753,7 +751,7 @@ no_thread_group:
753 751
754 write_lock_irq(&tasklist_lock); 752 write_lock_irq(&tasklist_lock);
755 spin_lock(&oldsighand->siglock); 753 spin_lock(&oldsighand->siglock);
756 spin_lock(&newsighand->siglock); 754 spin_lock_nested(&newsighand->siglock, SINGLE_DEPTH_NESTING);
757 755
758 rcu_assign_pointer(current->sighand, newsighand); 756 rcu_assign_pointer(current->sighand, newsighand);
759 recalc_sigpending(); 757 recalc_sigpending();
@@ -922,12 +920,6 @@ int prepare_binprm(struct linux_binprm *bprm)
922 int retval; 920 int retval;
923 921
924 mode = inode->i_mode; 922 mode = inode->i_mode;
925 /*
926 * Check execute perms again - if the caller has CAP_DAC_OVERRIDE,
927 * generic_permission lets a non-executable through
928 */
929 if (!(mode & 0111)) /* with at least _one_ execute bit set */
930 return -EACCES;
931 if (bprm->file->f_op == NULL) 923 if (bprm->file->f_op == NULL)
932 return -EACCES; 924 return -EACCES;
933 925
diff --git a/fs/ext2/super.c b/fs/ext2/super.c
index f2702cda9779..681dea8f9532 100644
--- a/fs/ext2/super.c
+++ b/fs/ext2/super.c
@@ -775,7 +775,7 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent)
775 if (EXT2_INODE_SIZE(sb) == 0) 775 if (EXT2_INODE_SIZE(sb) == 0)
776 goto cantfind_ext2; 776 goto cantfind_ext2;
777 sbi->s_inodes_per_block = sb->s_blocksize / EXT2_INODE_SIZE(sb); 777 sbi->s_inodes_per_block = sb->s_blocksize / EXT2_INODE_SIZE(sb);
778 if (sbi->s_inodes_per_block == 0) 778 if (sbi->s_inodes_per_block == 0 || sbi->s_inodes_per_group == 0)
779 goto cantfind_ext2; 779 goto cantfind_ext2;
780 sbi->s_itb_per_group = sbi->s_inodes_per_group / 780 sbi->s_itb_per_group = sbi->s_inodes_per_group /
781 sbi->s_inodes_per_block; 781 sbi->s_inodes_per_block;
diff --git a/fs/ext3/balloc.c b/fs/ext3/balloc.c
index a504a40d6d29..063d994bda0b 100644
--- a/fs/ext3/balloc.c
+++ b/fs/ext3/balloc.c
@@ -1269,12 +1269,12 @@ ext3_fsblk_t ext3_new_blocks(handle_t *handle, struct inode *inode,
1269 goal = le32_to_cpu(es->s_first_data_block); 1269 goal = le32_to_cpu(es->s_first_data_block);
1270 group_no = (goal - le32_to_cpu(es->s_first_data_block)) / 1270 group_no = (goal - le32_to_cpu(es->s_first_data_block)) /
1271 EXT3_BLOCKS_PER_GROUP(sb); 1271 EXT3_BLOCKS_PER_GROUP(sb);
1272 goal_group = group_no;
1273retry_alloc:
1272 gdp = ext3_get_group_desc(sb, group_no, &gdp_bh); 1274 gdp = ext3_get_group_desc(sb, group_no, &gdp_bh);
1273 if (!gdp) 1275 if (!gdp)
1274 goto io_error; 1276 goto io_error;
1275 1277
1276 goal_group = group_no;
1277retry:
1278 free_blocks = le16_to_cpu(gdp->bg_free_blocks_count); 1278 free_blocks = le16_to_cpu(gdp->bg_free_blocks_count);
1279 /* 1279 /*
1280 * if there is not enough free blocks to make a new resevation 1280 * if there is not enough free blocks to make a new resevation
@@ -1349,7 +1349,7 @@ retry:
1349 if (my_rsv) { 1349 if (my_rsv) {
1350 my_rsv = NULL; 1350 my_rsv = NULL;
1351 group_no = goal_group; 1351 group_no = goal_group;
1352 goto retry; 1352 goto retry_alloc;
1353 } 1353 }
1354 /* No space left on the device */ 1354 /* No space left on the device */
1355 *errp = -ENOSPC; 1355 *errp = -ENOSPC;
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index f804d5e9d60c..c5ee9f0691e3 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -1158,7 +1158,7 @@ retry:
1158 ret = PTR_ERR(handle); 1158 ret = PTR_ERR(handle);
1159 goto out; 1159 goto out;
1160 } 1160 }
1161 if (test_opt(inode->i_sb, NOBH)) 1161 if (test_opt(inode->i_sb, NOBH) && ext3_should_writeback_data(inode))
1162 ret = nobh_prepare_write(page, from, to, ext3_get_block); 1162 ret = nobh_prepare_write(page, from, to, ext3_get_block);
1163 else 1163 else
1164 ret = block_prepare_write(page, from, to, ext3_get_block); 1164 ret = block_prepare_write(page, from, to, ext3_get_block);
@@ -1244,7 +1244,7 @@ static int ext3_writeback_commit_write(struct file *file, struct page *page,
1244 if (new_i_size > EXT3_I(inode)->i_disksize) 1244 if (new_i_size > EXT3_I(inode)->i_disksize)
1245 EXT3_I(inode)->i_disksize = new_i_size; 1245 EXT3_I(inode)->i_disksize = new_i_size;
1246 1246
1247 if (test_opt(inode->i_sb, NOBH)) 1247 if (test_opt(inode->i_sb, NOBH) && ext3_should_writeback_data(inode))
1248 ret = nobh_commit_write(file, page, from, to); 1248 ret = nobh_commit_write(file, page, from, to);
1249 else 1249 else
1250 ret = generic_commit_write(file, page, from, to); 1250 ret = generic_commit_write(file, page, from, to);
@@ -1494,7 +1494,7 @@ static int ext3_writeback_writepage(struct page *page,
1494 goto out_fail; 1494 goto out_fail;
1495 } 1495 }
1496 1496
1497 if (test_opt(inode->i_sb, NOBH)) 1497 if (test_opt(inode->i_sb, NOBH) && ext3_should_writeback_data(inode))
1498 ret = nobh_writepage(page, ext3_get_block, wbc); 1498 ret = nobh_writepage(page, ext3_get_block, wbc);
1499 else 1499 else
1500 ret = block_write_full_page(page, ext3_get_block, wbc); 1500 ret = block_write_full_page(page, ext3_get_block, wbc);
@@ -2402,14 +2402,15 @@ static ext3_fsblk_t ext3_get_inode_block(struct super_block *sb,
2402 struct buffer_head *bh; 2402 struct buffer_head *bh;
2403 struct ext3_group_desc * gdp; 2403 struct ext3_group_desc * gdp;
2404 2404
2405 2405 if (!ext3_valid_inum(sb, ino)) {
2406 if ((ino != EXT3_ROOT_INO && ino != EXT3_JOURNAL_INO && 2406 /*
2407 ino != EXT3_RESIZE_INO && ino < EXT3_FIRST_INO(sb)) || 2407 * This error is already checked for in namei.c unless we are
2408 ino > le32_to_cpu(EXT3_SB(sb)->s_es->s_inodes_count)) { 2408 * looking at an NFS filehandle, in which case no error
2409 ext3_error(sb, "ext3_get_inode_block", 2409 * report is needed
2410 "bad inode number: %lu", ino); 2410 */
2411 return 0; 2411 return 0;
2412 } 2412 }
2413
2413 block_group = (ino - 1) / EXT3_INODES_PER_GROUP(sb); 2414 block_group = (ino - 1) / EXT3_INODES_PER_GROUP(sb);
2414 if (block_group >= EXT3_SB(sb)->s_groups_count) { 2415 if (block_group >= EXT3_SB(sb)->s_groups_count) {
2415 ext3_error(sb,"ext3_get_inode_block","group >= groups count"); 2416 ext3_error(sb,"ext3_get_inode_block","group >= groups count");
diff --git a/fs/ext3/namei.c b/fs/ext3/namei.c
index d9176dba3698..2aa7101b27cd 100644
--- a/fs/ext3/namei.c
+++ b/fs/ext3/namei.c
@@ -1000,7 +1000,12 @@ static struct dentry *ext3_lookup(struct inode * dir, struct dentry *dentry, str
1000 if (bh) { 1000 if (bh) {
1001 unsigned long ino = le32_to_cpu(de->inode); 1001 unsigned long ino = le32_to_cpu(de->inode);
1002 brelse (bh); 1002 brelse (bh);
1003 inode = iget(dir->i_sb, ino); 1003 if (!ext3_valid_inum(dir->i_sb, ino)) {
1004 ext3_error(dir->i_sb, "ext3_lookup",
1005 "bad inode number: %lu", ino);
1006 inode = NULL;
1007 } else
1008 inode = iget(dir->i_sb, ino);
1004 1009
1005 if (!inode) 1010 if (!inode)
1006 return ERR_PTR(-EACCES); 1011 return ERR_PTR(-EACCES);
@@ -1028,7 +1033,13 @@ struct dentry *ext3_get_parent(struct dentry *child)
1028 return ERR_PTR(-ENOENT); 1033 return ERR_PTR(-ENOENT);
1029 ino = le32_to_cpu(de->inode); 1034 ino = le32_to_cpu(de->inode);
1030 brelse(bh); 1035 brelse(bh);
1031 inode = iget(child->d_inode->i_sb, ino); 1036
1037 if (!ext3_valid_inum(child->d_inode->i_sb, ino)) {
1038 ext3_error(child->d_inode->i_sb, "ext3_get_parent",
1039 "bad inode number: %lu", ino);
1040 inode = NULL;
1041 } else
1042 inode = iget(child->d_inode->i_sb, ino);
1032 1043
1033 if (!inode) 1044 if (!inode)
1034 return ERR_PTR(-EACCES); 1045 return ERR_PTR(-EACCES);
diff --git a/fs/freevxfs/vxfs_lookup.c b/fs/freevxfs/vxfs_lookup.c
index 29cce456c7ce..43886fa00a2a 100644
--- a/fs/freevxfs/vxfs_lookup.c
+++ b/fs/freevxfs/vxfs_lookup.c
@@ -246,6 +246,8 @@ vxfs_readdir(struct file *fp, void *retp, filldir_t filler)
246 u_long page, npages, block, pblocks, nblocks, offset; 246 u_long page, npages, block, pblocks, nblocks, offset;
247 loff_t pos; 247 loff_t pos;
248 248
249 lock_kernel();
250
249 switch ((long)fp->f_pos) { 251 switch ((long)fp->f_pos) {
250 case 0: 252 case 0:
251 if (filler(retp, ".", 1, fp->f_pos, ip->i_ino, DT_DIR) < 0) 253 if (filler(retp, ".", 1, fp->f_pos, ip->i_ino, DT_DIR) < 0)
diff --git a/fs/fuse/control.c b/fs/fuse/control.c
index a3bce3a77253..46fe60b2da23 100644
--- a/fs/fuse/control.c
+++ b/fs/fuse/control.c
@@ -105,7 +105,7 @@ static struct dentry *fuse_ctl_add_dentry(struct dentry *parent,
105 105
106/* 106/*
107 * Add a connection to the control filesystem (if it exists). Caller 107 * Add a connection to the control filesystem (if it exists). Caller
108 * must host fuse_mutex 108 * must hold fuse_mutex
109 */ 109 */
110int fuse_ctl_add_conn(struct fuse_conn *fc) 110int fuse_ctl_add_conn(struct fuse_conn *fc)
111{ 111{
@@ -139,7 +139,7 @@ int fuse_ctl_add_conn(struct fuse_conn *fc)
139 139
140/* 140/*
141 * Remove a connection from the control filesystem (if it exists). 141 * Remove a connection from the control filesystem (if it exists).
142 * Caller must host fuse_mutex 142 * Caller must hold fuse_mutex
143 */ 143 */
144void fuse_ctl_remove_conn(struct fuse_conn *fc) 144void fuse_ctl_remove_conn(struct fuse_conn *fc)
145{ 145{
diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c
index 72a74cde6de8..409ce6a7cca4 100644
--- a/fs/fuse/dir.c
+++ b/fs/fuse/dir.c
@@ -14,6 +14,33 @@
14#include <linux/sched.h> 14#include <linux/sched.h>
15#include <linux/namei.h> 15#include <linux/namei.h>
16 16
17#if BITS_PER_LONG >= 64
18static inline void fuse_dentry_settime(struct dentry *entry, u64 time)
19{
20 entry->d_time = time;
21}
22
23static inline u64 fuse_dentry_time(struct dentry *entry)
24{
25 return entry->d_time;
26}
27#else
28/*
29 * On 32 bit archs store the high 32 bits of time in d_fsdata
30 */
31static void fuse_dentry_settime(struct dentry *entry, u64 time)
32{
33 entry->d_time = time;
34 entry->d_fsdata = (void *) (unsigned long) (time >> 32);
35}
36
37static u64 fuse_dentry_time(struct dentry *entry)
38{
39 return (u64) entry->d_time +
40 ((u64) (unsigned long) entry->d_fsdata << 32);
41}
42#endif
43
17/* 44/*
18 * FUSE caches dentries and attributes with separate timeout. The 45 * FUSE caches dentries and attributes with separate timeout. The
19 * time in jiffies until the dentry/attributes are valid is stored in 46 * time in jiffies until the dentry/attributes are valid is stored in
@@ -23,10 +50,13 @@
23/* 50/*
24 * Calculate the time in jiffies until a dentry/attributes are valid 51 * Calculate the time in jiffies until a dentry/attributes are valid
25 */ 52 */
26static unsigned long time_to_jiffies(unsigned long sec, unsigned long nsec) 53static u64 time_to_jiffies(unsigned long sec, unsigned long nsec)
27{ 54{
28 struct timespec ts = {sec, nsec}; 55 if (sec || nsec) {
29 return jiffies + timespec_to_jiffies(&ts); 56 struct timespec ts = {sec, nsec};
57 return get_jiffies_64() + timespec_to_jiffies(&ts);
58 } else
59 return 0;
30} 60}
31 61
32/* 62/*
@@ -35,7 +65,8 @@ static unsigned long time_to_jiffies(unsigned long sec, unsigned long nsec)
35 */ 65 */
36static void fuse_change_timeout(struct dentry *entry, struct fuse_entry_out *o) 66static void fuse_change_timeout(struct dentry *entry, struct fuse_entry_out *o)
37{ 67{
38 entry->d_time = time_to_jiffies(o->entry_valid, o->entry_valid_nsec); 68 fuse_dentry_settime(entry,
69 time_to_jiffies(o->entry_valid, o->entry_valid_nsec));
39 if (entry->d_inode) 70 if (entry->d_inode)
40 get_fuse_inode(entry->d_inode)->i_time = 71 get_fuse_inode(entry->d_inode)->i_time =
41 time_to_jiffies(o->attr_valid, o->attr_valid_nsec); 72 time_to_jiffies(o->attr_valid, o->attr_valid_nsec);
@@ -47,7 +78,7 @@ static void fuse_change_timeout(struct dentry *entry, struct fuse_entry_out *o)
47 */ 78 */
48void fuse_invalidate_attr(struct inode *inode) 79void fuse_invalidate_attr(struct inode *inode)
49{ 80{
50 get_fuse_inode(inode)->i_time = jiffies - 1; 81 get_fuse_inode(inode)->i_time = 0;
51} 82}
52 83
53/* 84/*
@@ -60,7 +91,7 @@ void fuse_invalidate_attr(struct inode *inode)
60 */ 91 */
61static void fuse_invalidate_entry_cache(struct dentry *entry) 92static void fuse_invalidate_entry_cache(struct dentry *entry)
62{ 93{
63 entry->d_time = jiffies - 1; 94 fuse_dentry_settime(entry, 0);
64} 95}
65 96
66/* 97/*
@@ -102,7 +133,7 @@ static int fuse_dentry_revalidate(struct dentry *entry, struct nameidata *nd)
102 133
103 if (inode && is_bad_inode(inode)) 134 if (inode && is_bad_inode(inode))
104 return 0; 135 return 0;
105 else if (time_after(jiffies, entry->d_time)) { 136 else if (fuse_dentry_time(entry) < get_jiffies_64()) {
106 int err; 137 int err;
107 struct fuse_entry_out outarg; 138 struct fuse_entry_out outarg;
108 struct fuse_conn *fc; 139 struct fuse_conn *fc;
@@ -666,7 +697,7 @@ static int fuse_revalidate(struct dentry *entry)
666 if (!fuse_allow_task(fc, current)) 697 if (!fuse_allow_task(fc, current))
667 return -EACCES; 698 return -EACCES;
668 if (get_node_id(inode) != FUSE_ROOT_ID && 699 if (get_node_id(inode) != FUSE_ROOT_ID &&
669 time_before_eq(jiffies, fi->i_time)) 700 fi->i_time >= get_jiffies_64())
670 return 0; 701 return 0;
671 702
672 return fuse_do_getattr(inode); 703 return fuse_do_getattr(inode);
diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index 63614ed16336..5c4fcd1dbf59 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -395,14 +395,16 @@ static int fuse_readpages(struct file *file, struct address_space *mapping,
395 struct fuse_readpages_data data; 395 struct fuse_readpages_data data;
396 int err; 396 int err;
397 397
398 err = -EIO;
398 if (is_bad_inode(inode)) 399 if (is_bad_inode(inode))
399 return -EIO; 400 goto clean_pages_up;
400 401
401 data.file = file; 402 data.file = file;
402 data.inode = inode; 403 data.inode = inode;
403 data.req = fuse_get_req(fc); 404 data.req = fuse_get_req(fc);
405 err = PTR_ERR(data.req);
404 if (IS_ERR(data.req)) 406 if (IS_ERR(data.req))
405 return PTR_ERR(data.req); 407 goto clean_pages_up;
406 408
407 err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data); 409 err = read_cache_pages(mapping, pages, fuse_readpages_fill, &data);
408 if (!err) { 410 if (!err) {
@@ -412,6 +414,10 @@ static int fuse_readpages(struct file *file, struct address_space *mapping,
412 fuse_put_request(fc, data.req); 414 fuse_put_request(fc, data.req);
413 } 415 }
414 return err; 416 return err;
417
418clean_pages_up:
419 put_pages_list(pages);
420 return err;
415} 421}
416 422
417static size_t fuse_send_write(struct fuse_req *req, struct file *file, 423static size_t fuse_send_write(struct fuse_req *req, struct file *file,
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index 0dbf96621841..69c7750d55b8 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -59,7 +59,7 @@ struct fuse_inode {
59 struct fuse_req *forget_req; 59 struct fuse_req *forget_req;
60 60
61 /** Time in jiffies until the file attributes are valid */ 61 /** Time in jiffies until the file attributes are valid */
62 unsigned long i_time; 62 u64 i_time;
63}; 63};
64 64
65/** FUSE specific file data */ 65/** FUSE specific file data */
diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index dcaaabd3b9c4..7d25092262ae 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -51,7 +51,7 @@ static struct inode *fuse_alloc_inode(struct super_block *sb)
51 return NULL; 51 return NULL;
52 52
53 fi = get_fuse_inode(inode); 53 fi = get_fuse_inode(inode);
54 fi->i_time = jiffies - 1; 54 fi->i_time = 0;
55 fi->nodeid = 0; 55 fi->nodeid = 0;
56 fi->nlookup = 0; 56 fi->nlookup = 0;
57 fi->forget_req = fuse_request_alloc(); 57 fi->forget_req = fuse_request_alloc();
diff --git a/fs/inotify_user.c b/fs/inotify_user.c
index f2386442adee..017cb0f134d6 100644
--- a/fs/inotify_user.c
+++ b/fs/inotify_user.c
@@ -187,7 +187,7 @@ static struct inotify_kernel_event * kernel_event(s32 wd, u32 mask, u32 cookie,
187{ 187{
188 struct inotify_kernel_event *kevent; 188 struct inotify_kernel_event *kevent;
189 189
190 kevent = kmem_cache_alloc(event_cachep, GFP_KERNEL); 190 kevent = kmem_cache_alloc(event_cachep, GFP_NOFS);
191 if (unlikely(!kevent)) 191 if (unlikely(!kevent))
192 return NULL; 192 return NULL;
193 193
diff --git a/fs/ioprio.c b/fs/ioprio.c
index 93aa5715f224..78b1deae3fa2 100644
--- a/fs/ioprio.c
+++ b/fs/ioprio.c
@@ -44,6 +44,9 @@ static int set_task_ioprio(struct task_struct *task, int ioprio)
44 task->ioprio = ioprio; 44 task->ioprio = ioprio;
45 45
46 ioc = task->io_context; 46 ioc = task->io_context;
47 /* see wmb() in current_io_context() */
48 smp_read_barrier_depends();
49
47 if (ioc && ioc->set_ioprio) 50 if (ioc && ioc->set_ioprio)
48 ioc->set_ioprio(ioc, ioprio); 51 ioc->set_ioprio(ioc, ioprio);
49 52
@@ -111,9 +114,9 @@ asmlinkage long sys_ioprio_set(int which, int who, int ioprio)
111 continue; 114 continue;
112 ret = set_task_ioprio(p, ioprio); 115 ret = set_task_ioprio(p, ioprio);
113 if (ret) 116 if (ret)
114 break; 117 goto free_uid;
115 } while_each_thread(g, p); 118 } while_each_thread(g, p);
116 119free_uid:
117 if (who) 120 if (who)
118 free_uid(user); 121 free_uid(user);
119 break; 122 break;
@@ -137,6 +140,29 @@ out:
137 return ret; 140 return ret;
138} 141}
139 142
143int ioprio_best(unsigned short aprio, unsigned short bprio)
144{
145 unsigned short aclass = IOPRIO_PRIO_CLASS(aprio);
146 unsigned short bclass = IOPRIO_PRIO_CLASS(bprio);
147
148 if (!ioprio_valid(aprio))
149 return bprio;
150 if (!ioprio_valid(bprio))
151 return aprio;
152
153 if (aclass == IOPRIO_CLASS_NONE)
154 aclass = IOPRIO_CLASS_BE;
155 if (bclass == IOPRIO_CLASS_NONE)
156 bclass = IOPRIO_CLASS_BE;
157
158 if (aclass == bclass)
159 return min(aprio, bprio);
160 if (aclass > bclass)
161 return bprio;
162 else
163 return aprio;
164}
165
140asmlinkage long sys_ioprio_get(int which, int who) 166asmlinkage long sys_ioprio_get(int which, int who)
141{ 167{
142 struct task_struct *g, *p; 168 struct task_struct *g, *p;
diff --git a/fs/jbd/commit.c b/fs/jbd/commit.c
index 0971814c38b8..42da60784311 100644
--- a/fs/jbd/commit.c
+++ b/fs/jbd/commit.c
@@ -261,7 +261,7 @@ void journal_commit_transaction(journal_t *journal)
261 struct buffer_head *bh = jh2bh(jh); 261 struct buffer_head *bh = jh2bh(jh);
262 262
263 jbd_lock_bh_state(bh); 263 jbd_lock_bh_state(bh);
264 kfree(jh->b_committed_data); 264 jbd_slab_free(jh->b_committed_data, bh->b_size);
265 jh->b_committed_data = NULL; 265 jh->b_committed_data = NULL;
266 jbd_unlock_bh_state(bh); 266 jbd_unlock_bh_state(bh);
267 } 267 }
@@ -745,14 +745,14 @@ restart_loop:
745 * Otherwise, we can just throw away the frozen data now. 745 * Otherwise, we can just throw away the frozen data now.
746 */ 746 */
747 if (jh->b_committed_data) { 747 if (jh->b_committed_data) {
748 kfree(jh->b_committed_data); 748 jbd_slab_free(jh->b_committed_data, bh->b_size);
749 jh->b_committed_data = NULL; 749 jh->b_committed_data = NULL;
750 if (jh->b_frozen_data) { 750 if (jh->b_frozen_data) {
751 jh->b_committed_data = jh->b_frozen_data; 751 jh->b_committed_data = jh->b_frozen_data;
752 jh->b_frozen_data = NULL; 752 jh->b_frozen_data = NULL;
753 } 753 }
754 } else if (jh->b_frozen_data) { 754 } else if (jh->b_frozen_data) {
755 kfree(jh->b_frozen_data); 755 jbd_slab_free(jh->b_frozen_data, bh->b_size);
756 jh->b_frozen_data = NULL; 756 jh->b_frozen_data = NULL;
757 } 757 }
758 758
diff --git a/fs/jbd/journal.c b/fs/jbd/journal.c
index 8c9b28dff119..f66724ce443a 100644
--- a/fs/jbd/journal.c
+++ b/fs/jbd/journal.c
@@ -84,6 +84,7 @@ EXPORT_SYMBOL(journal_force_commit);
84 84
85static int journal_convert_superblock_v1(journal_t *, journal_superblock_t *); 85static int journal_convert_superblock_v1(journal_t *, journal_superblock_t *);
86static void __journal_abort_soft (journal_t *journal, int errno); 86static void __journal_abort_soft (journal_t *journal, int errno);
87static int journal_create_jbd_slab(size_t slab_size);
87 88
88/* 89/*
89 * Helper function used to manage commit timeouts 90 * Helper function used to manage commit timeouts
@@ -328,10 +329,10 @@ repeat:
328 char *tmp; 329 char *tmp;
329 330
330 jbd_unlock_bh_state(bh_in); 331 jbd_unlock_bh_state(bh_in);
331 tmp = jbd_rep_kmalloc(bh_in->b_size, GFP_NOFS); 332 tmp = jbd_slab_alloc(bh_in->b_size, GFP_NOFS);
332 jbd_lock_bh_state(bh_in); 333 jbd_lock_bh_state(bh_in);
333 if (jh_in->b_frozen_data) { 334 if (jh_in->b_frozen_data) {
334 kfree(tmp); 335 jbd_slab_free(tmp, bh_in->b_size);
335 goto repeat; 336 goto repeat;
336 } 337 }
337 338
@@ -1069,17 +1070,17 @@ static int load_superblock(journal_t *journal)
1069int journal_load(journal_t *journal) 1070int journal_load(journal_t *journal)
1070{ 1071{
1071 int err; 1072 int err;
1073 journal_superblock_t *sb;
1072 1074
1073 err = load_superblock(journal); 1075 err = load_superblock(journal);
1074 if (err) 1076 if (err)
1075 return err; 1077 return err;
1076 1078
1079 sb = journal->j_superblock;
1077 /* If this is a V2 superblock, then we have to check the 1080 /* If this is a V2 superblock, then we have to check the
1078 * features flags on it. */ 1081 * features flags on it. */
1079 1082
1080 if (journal->j_format_version >= 2) { 1083 if (journal->j_format_version >= 2) {
1081 journal_superblock_t *sb = journal->j_superblock;
1082
1083 if ((sb->s_feature_ro_compat & 1084 if ((sb->s_feature_ro_compat &
1084 ~cpu_to_be32(JFS_KNOWN_ROCOMPAT_FEATURES)) || 1085 ~cpu_to_be32(JFS_KNOWN_ROCOMPAT_FEATURES)) ||
1085 (sb->s_feature_incompat & 1086 (sb->s_feature_incompat &
@@ -1090,6 +1091,13 @@ int journal_load(journal_t *journal)
1090 } 1091 }
1091 } 1092 }
1092 1093
1094 /*
1095 * Create a slab for this blocksize
1096 */
1097 err = journal_create_jbd_slab(cpu_to_be32(sb->s_blocksize));
1098 if (err)
1099 return err;
1100
1093 /* Let the recovery code check whether it needs to recover any 1101 /* Let the recovery code check whether it needs to recover any
1094 * data from the journal. */ 1102 * data from the journal. */
1095 if (journal_recover(journal)) 1103 if (journal_recover(journal))
@@ -1612,6 +1620,77 @@ void * __jbd_kmalloc (const char *where, size_t size, gfp_t flags, int retry)
1612} 1620}
1613 1621
1614/* 1622/*
1623 * jbd slab management: create 1k, 2k, 4k, 8k slabs as needed
1624 * and allocate frozen and commit buffers from these slabs.
1625 *
1626 * Reason for doing this is to avoid, SLAB_DEBUG - since it could
1627 * cause bh to cross page boundary.
1628 */
1629
1630#define JBD_MAX_SLABS 5
1631#define JBD_SLAB_INDEX(size) (size >> 11)
1632
1633static kmem_cache_t *jbd_slab[JBD_MAX_SLABS];
1634static const char *jbd_slab_names[JBD_MAX_SLABS] = {
1635 "jbd_1k", "jbd_2k", "jbd_4k", NULL, "jbd_8k"
1636};
1637
1638static void journal_destroy_jbd_slabs(void)
1639{
1640 int i;
1641
1642 for (i = 0; i < JBD_MAX_SLABS; i++) {
1643 if (jbd_slab[i])
1644 kmem_cache_destroy(jbd_slab[i]);
1645 jbd_slab[i] = NULL;
1646 }
1647}
1648
1649static int journal_create_jbd_slab(size_t slab_size)
1650{
1651 int i = JBD_SLAB_INDEX(slab_size);
1652
1653 BUG_ON(i >= JBD_MAX_SLABS);
1654
1655 /*
1656 * Check if we already have a slab created for this size
1657 */
1658 if (jbd_slab[i])
1659 return 0;
1660
1661 /*
1662 * Create a slab and force alignment to be same as slabsize -
1663 * this will make sure that allocations won't cross the page
1664 * boundary.
1665 */
1666 jbd_slab[i] = kmem_cache_create(jbd_slab_names[i],
1667 slab_size, slab_size, 0, NULL, NULL);
1668 if (!jbd_slab[i]) {
1669 printk(KERN_EMERG "JBD: no memory for jbd_slab cache\n");
1670 return -ENOMEM;
1671 }
1672 return 0;
1673}
1674
1675void * jbd_slab_alloc(size_t size, gfp_t flags)
1676{
1677 int idx;
1678
1679 idx = JBD_SLAB_INDEX(size);
1680 BUG_ON(jbd_slab[idx] == NULL);
1681 return kmem_cache_alloc(jbd_slab[idx], flags | __GFP_NOFAIL);
1682}
1683
1684void jbd_slab_free(void *ptr, size_t size)
1685{
1686 int idx;
1687
1688 idx = JBD_SLAB_INDEX(size);
1689 BUG_ON(jbd_slab[idx] == NULL);
1690 kmem_cache_free(jbd_slab[idx], ptr);
1691}
1692
1693/*
1615 * Journal_head storage management 1694 * Journal_head storage management
1616 */ 1695 */
1617static kmem_cache_t *journal_head_cache; 1696static kmem_cache_t *journal_head_cache;
@@ -1799,13 +1878,13 @@ static void __journal_remove_journal_head(struct buffer_head *bh)
1799 printk(KERN_WARNING "%s: freeing " 1878 printk(KERN_WARNING "%s: freeing "
1800 "b_frozen_data\n", 1879 "b_frozen_data\n",
1801 __FUNCTION__); 1880 __FUNCTION__);
1802 kfree(jh->b_frozen_data); 1881 jbd_slab_free(jh->b_frozen_data, bh->b_size);
1803 } 1882 }
1804 if (jh->b_committed_data) { 1883 if (jh->b_committed_data) {
1805 printk(KERN_WARNING "%s: freeing " 1884 printk(KERN_WARNING "%s: freeing "
1806 "b_committed_data\n", 1885 "b_committed_data\n",
1807 __FUNCTION__); 1886 __FUNCTION__);
1808 kfree(jh->b_committed_data); 1887 jbd_slab_free(jh->b_committed_data, bh->b_size);
1809 } 1888 }
1810 bh->b_private = NULL; 1889 bh->b_private = NULL;
1811 jh->b_bh = NULL; /* debug, really */ 1890 jh->b_bh = NULL; /* debug, really */
@@ -1961,6 +2040,7 @@ static void journal_destroy_caches(void)
1961 journal_destroy_revoke_caches(); 2040 journal_destroy_revoke_caches();
1962 journal_destroy_journal_head_cache(); 2041 journal_destroy_journal_head_cache();
1963 journal_destroy_handle_cache(); 2042 journal_destroy_handle_cache();
2043 journal_destroy_jbd_slabs();
1964} 2044}
1965 2045
1966static int __init journal_init(void) 2046static int __init journal_init(void)
diff --git a/fs/jbd/transaction.c b/fs/jbd/transaction.c
index 508b2ea91f43..de2e4cbbf79a 100644
--- a/fs/jbd/transaction.c
+++ b/fs/jbd/transaction.c
@@ -666,8 +666,9 @@ repeat:
666 if (!frozen_buffer) { 666 if (!frozen_buffer) {
667 JBUFFER_TRACE(jh, "allocate memory for buffer"); 667 JBUFFER_TRACE(jh, "allocate memory for buffer");
668 jbd_unlock_bh_state(bh); 668 jbd_unlock_bh_state(bh);
669 frozen_buffer = jbd_kmalloc(jh2bh(jh)->b_size, 669 frozen_buffer =
670 GFP_NOFS); 670 jbd_slab_alloc(jh2bh(jh)->b_size,
671 GFP_NOFS);
671 if (!frozen_buffer) { 672 if (!frozen_buffer) {
672 printk(KERN_EMERG 673 printk(KERN_EMERG
673 "%s: OOM for frozen_buffer\n", 674 "%s: OOM for frozen_buffer\n",
@@ -879,7 +880,7 @@ int journal_get_undo_access(handle_t *handle, struct buffer_head *bh)
879 880
880repeat: 881repeat:
881 if (!jh->b_committed_data) { 882 if (!jh->b_committed_data) {
882 committed_data = jbd_kmalloc(jh2bh(jh)->b_size, GFP_NOFS); 883 committed_data = jbd_slab_alloc(jh2bh(jh)->b_size, GFP_NOFS);
883 if (!committed_data) { 884 if (!committed_data) {
884 printk(KERN_EMERG "%s: No memory for committed data\n", 885 printk(KERN_EMERG "%s: No memory for committed data\n",
885 __FUNCTION__); 886 __FUNCTION__);
@@ -906,7 +907,7 @@ repeat:
906out: 907out:
907 journal_put_journal_head(jh); 908 journal_put_journal_head(jh);
908 if (unlikely(committed_data)) 909 if (unlikely(committed_data))
909 kfree(committed_data); 910 jbd_slab_free(committed_data, bh->b_size);
910 return err; 911 return err;
911} 912}
912 913
diff --git a/fs/jfs/inode.c b/fs/jfs/inode.c
index 43e3f566aad6..a223cf4faa9b 100644
--- a/fs/jfs/inode.c
+++ b/fs/jfs/inode.c
@@ -168,16 +168,15 @@ void jfs_dirty_inode(struct inode *inode)
168 set_cflag(COMMIT_Dirty, inode); 168 set_cflag(COMMIT_Dirty, inode);
169} 169}
170 170
171static int 171int jfs_get_block(struct inode *ip, sector_t lblock,
172jfs_get_blocks(struct inode *ip, sector_t lblock, unsigned long max_blocks, 172 struct buffer_head *bh_result, int create)
173 struct buffer_head *bh_result, int create)
174{ 173{
175 s64 lblock64 = lblock; 174 s64 lblock64 = lblock;
176 int rc = 0; 175 int rc = 0;
177 xad_t xad; 176 xad_t xad;
178 s64 xaddr; 177 s64 xaddr;
179 int xflag; 178 int xflag;
180 s32 xlen = max_blocks; 179 s32 xlen = bh_result->b_size >> ip->i_blkbits;
181 180
182 /* 181 /*
183 * Take appropriate lock on inode 182 * Take appropriate lock on inode
@@ -188,7 +187,7 @@ jfs_get_blocks(struct inode *ip, sector_t lblock, unsigned long max_blocks,
188 IREAD_LOCK(ip); 187 IREAD_LOCK(ip);
189 188
190 if (((lblock64 << ip->i_sb->s_blocksize_bits) < ip->i_size) && 189 if (((lblock64 << ip->i_sb->s_blocksize_bits) < ip->i_size) &&
191 (!xtLookup(ip, lblock64, max_blocks, &xflag, &xaddr, &xlen, 0)) && 190 (!xtLookup(ip, lblock64, xlen, &xflag, &xaddr, &xlen, 0)) &&
192 xaddr) { 191 xaddr) {
193 if (xflag & XAD_NOTRECORDED) { 192 if (xflag & XAD_NOTRECORDED) {
194 if (!create) 193 if (!create)
@@ -255,13 +254,6 @@ jfs_get_blocks(struct inode *ip, sector_t lblock, unsigned long max_blocks,
255 return rc; 254 return rc;
256} 255}
257 256
258static int jfs_get_block(struct inode *ip, sector_t lblock,
259 struct buffer_head *bh_result, int create)
260{
261 return jfs_get_blocks(ip, lblock, bh_result->b_size >> ip->i_blkbits,
262 bh_result, create);
263}
264
265static int jfs_writepage(struct page *page, struct writeback_control *wbc) 257static int jfs_writepage(struct page *page, struct writeback_control *wbc)
266{ 258{
267 return nobh_writepage(page, jfs_get_block, wbc); 259 return nobh_writepage(page, jfs_get_block, wbc);
diff --git a/fs/jfs/jfs_inode.h b/fs/jfs/jfs_inode.h
index b5c7da6190dc..1fc48df670c8 100644
--- a/fs/jfs/jfs_inode.h
+++ b/fs/jfs/jfs_inode.h
@@ -32,6 +32,7 @@ extern void jfs_truncate_nolock(struct inode *, loff_t);
32extern void jfs_free_zero_link(struct inode *); 32extern void jfs_free_zero_link(struct inode *);
33extern struct dentry *jfs_get_parent(struct dentry *dentry); 33extern struct dentry *jfs_get_parent(struct dentry *dentry);
34extern void jfs_set_inode_flags(struct inode *); 34extern void jfs_set_inode_flags(struct inode *);
35extern int jfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
35 36
36extern const struct address_space_operations jfs_aops; 37extern const struct address_space_operations jfs_aops;
37extern struct inode_operations jfs_dir_inode_operations; 38extern struct inode_operations jfs_dir_inode_operations;
diff --git a/fs/jfs/super.c b/fs/jfs/super.c
index 4f6cfebc82db..143bcd1d5eaa 100644
--- a/fs/jfs/super.c
+++ b/fs/jfs/super.c
@@ -26,6 +26,7 @@
26#include <linux/moduleparam.h> 26#include <linux/moduleparam.h>
27#include <linux/kthread.h> 27#include <linux/kthread.h>
28#include <linux/posix_acl.h> 28#include <linux/posix_acl.h>
29#include <linux/buffer_head.h>
29#include <asm/uaccess.h> 30#include <asm/uaccess.h>
30#include <linux/seq_file.h> 31#include <linux/seq_file.h>
31 32
@@ -298,7 +299,7 @@ static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
298 break; 299 break;
299 } 300 }
300 301
301#if defined(CONFIG_QUOTA) 302#ifdef CONFIG_QUOTA
302 case Opt_quota: 303 case Opt_quota:
303 case Opt_usrquota: 304 case Opt_usrquota:
304 *flag |= JFS_USRQUOTA; 305 *flag |= JFS_USRQUOTA;
@@ -597,7 +598,7 @@ static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
597 if (sbi->flag & JFS_NOINTEGRITY) 598 if (sbi->flag & JFS_NOINTEGRITY)
598 seq_puts(seq, ",nointegrity"); 599 seq_puts(seq, ",nointegrity");
599 600
600#if defined(CONFIG_QUOTA) 601#ifdef CONFIG_QUOTA
601 if (sbi->flag & JFS_USRQUOTA) 602 if (sbi->flag & JFS_USRQUOTA)
602 seq_puts(seq, ",usrquota"); 603 seq_puts(seq, ",usrquota");
603 604
@@ -608,6 +609,113 @@ static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
608 return 0; 609 return 0;
609} 610}
610 611
612#ifdef CONFIG_QUOTA
613
614/* Read data from quotafile - avoid pagecache and such because we cannot afford
615 * acquiring the locks... As quota files are never truncated and quota code
616 * itself serializes the operations (and noone else should touch the files)
617 * we don't have to be afraid of races */
618static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
619 size_t len, loff_t off)
620{
621 struct inode *inode = sb_dqopt(sb)->files[type];
622 sector_t blk = off >> sb->s_blocksize_bits;
623 int err = 0;
624 int offset = off & (sb->s_blocksize - 1);
625 int tocopy;
626 size_t toread;
627 struct buffer_head tmp_bh;
628 struct buffer_head *bh;
629 loff_t i_size = i_size_read(inode);
630
631 if (off > i_size)
632 return 0;
633 if (off+len > i_size)
634 len = i_size-off;
635 toread = len;
636 while (toread > 0) {
637 tocopy = sb->s_blocksize - offset < toread ?
638 sb->s_blocksize - offset : toread;
639
640 tmp_bh.b_state = 0;
641 tmp_bh.b_size = 1 << inode->i_blkbits;
642 err = jfs_get_block(inode, blk, &tmp_bh, 0);
643 if (err)
644 return err;
645 if (!buffer_mapped(&tmp_bh)) /* A hole? */
646 memset(data, 0, tocopy);
647 else {
648 bh = sb_bread(sb, tmp_bh.b_blocknr);
649 if (!bh)
650 return -EIO;
651 memcpy(data, bh->b_data+offset, tocopy);
652 brelse(bh);
653 }
654 offset = 0;
655 toread -= tocopy;
656 data += tocopy;
657 blk++;
658 }
659 return len;
660}
661
662/* Write to quotafile */
663static ssize_t jfs_quota_write(struct super_block *sb, int type,
664 const char *data, size_t len, loff_t off)
665{
666 struct inode *inode = sb_dqopt(sb)->files[type];
667 sector_t blk = off >> sb->s_blocksize_bits;
668 int err = 0;
669 int offset = off & (sb->s_blocksize - 1);
670 int tocopy;
671 size_t towrite = len;
672 struct buffer_head tmp_bh;
673 struct buffer_head *bh;
674
675 mutex_lock(&inode->i_mutex);
676 while (towrite > 0) {
677 tocopy = sb->s_blocksize - offset < towrite ?
678 sb->s_blocksize - offset : towrite;
679
680 tmp_bh.b_state = 0;
681 tmp_bh.b_size = 1 << inode->i_blkbits;
682 err = jfs_get_block(inode, blk, &tmp_bh, 1);
683 if (err)
684 goto out;
685 if (offset || tocopy != sb->s_blocksize)
686 bh = sb_bread(sb, tmp_bh.b_blocknr);
687 else
688 bh = sb_getblk(sb, tmp_bh.b_blocknr);
689 if (!bh) {
690 err = -EIO;
691 goto out;
692 }
693 lock_buffer(bh);
694 memcpy(bh->b_data+offset, data, tocopy);
695 flush_dcache_page(bh->b_page);
696 set_buffer_uptodate(bh);
697 mark_buffer_dirty(bh);
698 unlock_buffer(bh);
699 brelse(bh);
700 offset = 0;
701 towrite -= tocopy;
702 data += tocopy;
703 blk++;
704 }
705out:
706 if (len == towrite)
707 return err;
708 if (inode->i_size < off+len-towrite)
709 i_size_write(inode, off+len-towrite);
710 inode->i_version++;
711 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
712 mark_inode_dirty(inode);
713 mutex_unlock(&inode->i_mutex);
714 return len - towrite;
715}
716
717#endif
718
611static struct super_operations jfs_super_operations = { 719static struct super_operations jfs_super_operations = {
612 .alloc_inode = jfs_alloc_inode, 720 .alloc_inode = jfs_alloc_inode,
613 .destroy_inode = jfs_destroy_inode, 721 .destroy_inode = jfs_destroy_inode,
@@ -621,7 +729,11 @@ static struct super_operations jfs_super_operations = {
621 .unlockfs = jfs_unlockfs, 729 .unlockfs = jfs_unlockfs,
622 .statfs = jfs_statfs, 730 .statfs = jfs_statfs,
623 .remount_fs = jfs_remount, 731 .remount_fs = jfs_remount,
624 .show_options = jfs_show_options 732 .show_options = jfs_show_options,
733#ifdef CONFIG_QUOTA
734 .quota_read = jfs_quota_read,
735 .quota_write = jfs_quota_write,
736#endif
625}; 737};
626 738
627static struct export_operations jfs_export_operations = { 739static struct export_operations jfs_export_operations = {
diff --git a/fs/lockd/svclock.c b/fs/lockd/svclock.c
index baf5ae513481..c9d419703cf3 100644
--- a/fs/lockd/svclock.c
+++ b/fs/lockd/svclock.c
@@ -638,9 +638,6 @@ static void nlmsvc_grant_callback(struct rpc_task *task, void *data)
638 if (task->tk_status < 0) { 638 if (task->tk_status < 0) {
639 /* RPC error: Re-insert for retransmission */ 639 /* RPC error: Re-insert for retransmission */
640 timeout = 10 * HZ; 640 timeout = 10 * HZ;
641 } else if (block->b_done) {
642 /* Block already removed, kill it for real */
643 timeout = 0;
644 } else { 641 } else {
645 /* Call was successful, now wait for client callback */ 642 /* Call was successful, now wait for client callback */
646 timeout = 60 * HZ; 643 timeout = 60 * HZ;
@@ -709,13 +706,10 @@ nlmsvc_retry_blocked(void)
709 break; 706 break;
710 if (time_after(block->b_when,jiffies)) 707 if (time_after(block->b_when,jiffies))
711 break; 708 break;
712 dprintk("nlmsvc_retry_blocked(%p, when=%ld, done=%d)\n", 709 dprintk("nlmsvc_retry_blocked(%p, when=%ld)\n",
713 block, block->b_when, block->b_done); 710 block, block->b_when);
714 kref_get(&block->b_count); 711 kref_get(&block->b_count);
715 if (block->b_done) 712 nlmsvc_grant_blocked(block);
716 nlmsvc_unlink_block(block);
717 else
718 nlmsvc_grant_blocked(block);
719 nlmsvc_release_block(block); 713 nlmsvc_release_block(block);
720 } 714 }
721 715
diff --git a/fs/lockd/svcsubs.c b/fs/lockd/svcsubs.c
index 2a4df9b3779a..01b4db9e5466 100644
--- a/fs/lockd/svcsubs.c
+++ b/fs/lockd/svcsubs.c
@@ -237,19 +237,22 @@ static int
237nlm_traverse_files(struct nlm_host *host, int action) 237nlm_traverse_files(struct nlm_host *host, int action)
238{ 238{
239 struct nlm_file *file, **fp; 239 struct nlm_file *file, **fp;
240 int i; 240 int i, ret = 0;
241 241
242 mutex_lock(&nlm_file_mutex); 242 mutex_lock(&nlm_file_mutex);
243 for (i = 0; i < FILE_NRHASH; i++) { 243 for (i = 0; i < FILE_NRHASH; i++) {
244 fp = nlm_files + i; 244 fp = nlm_files + i;
245 while ((file = *fp) != NULL) { 245 while ((file = *fp) != NULL) {
246 file->f_count++;
247 mutex_unlock(&nlm_file_mutex);
248
246 /* Traverse locks, blocks and shares of this file 249 /* Traverse locks, blocks and shares of this file
247 * and update file->f_locks count */ 250 * and update file->f_locks count */
248 if (nlm_inspect_file(host, file, action)) { 251 if (nlm_inspect_file(host, file, action))
249 mutex_unlock(&nlm_file_mutex); 252 ret = 1;
250 return 1;
251 }
252 253
254 mutex_lock(&nlm_file_mutex);
255 file->f_count--;
253 /* No more references to this file. Let go of it. */ 256 /* No more references to this file. Let go of it. */
254 if (!file->f_blocks && !file->f_locks 257 if (!file->f_blocks && !file->f_locks
255 && !file->f_shares && !file->f_count) { 258 && !file->f_shares && !file->f_count) {
@@ -262,7 +265,7 @@ nlm_traverse_files(struct nlm_host *host, int action)
262 } 265 }
263 } 266 }
264 mutex_unlock(&nlm_file_mutex); 267 mutex_unlock(&nlm_file_mutex);
265 return 0; 268 return ret;
266} 269}
267 270
268/* 271/*
diff --git a/fs/locks.c b/fs/locks.c
index b0b41a64e10b..d7c53392cac1 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -1421,8 +1421,9 @@ static int __setlease(struct file *filp, long arg, struct file_lock **flp)
1421 if (!leases_enable) 1421 if (!leases_enable)
1422 goto out; 1422 goto out;
1423 1423
1424 error = lease_alloc(filp, arg, &fl); 1424 error = -ENOMEM;
1425 if (error) 1425 fl = locks_alloc_lock();
1426 if (fl == NULL)
1426 goto out; 1427 goto out;
1427 1428
1428 locks_copy_lock(fl, lease); 1429 locks_copy_lock(fl, lease);
@@ -1430,6 +1431,7 @@ static int __setlease(struct file *filp, long arg, struct file_lock **flp)
1430 locks_insert_lock(before, fl); 1431 locks_insert_lock(before, fl);
1431 1432
1432 *flp = fl; 1433 *flp = fl;
1434 error = 0;
1433out: 1435out:
1434 return error; 1436 return error;
1435} 1437}
diff --git a/fs/minix/inode.c b/fs/minix/inode.c
index 9ea91c5eeb7b..330ff9fc7cf0 100644
--- a/fs/minix/inode.c
+++ b/fs/minix/inode.c
@@ -204,6 +204,8 @@ static int minix_fill_super(struct super_block *s, void *data, int silent)
204 /* 204 /*
205 * Allocate the buffer map to keep the superblock small. 205 * Allocate the buffer map to keep the superblock small.
206 */ 206 */
207 if (sbi->s_imap_blocks == 0 || sbi->s_zmap_blocks == 0)
208 goto out_illegal_sb;
207 i = (sbi->s_imap_blocks + sbi->s_zmap_blocks) * sizeof(bh); 209 i = (sbi->s_imap_blocks + sbi->s_zmap_blocks) * sizeof(bh);
208 map = kmalloc(i, GFP_KERNEL); 210 map = kmalloc(i, GFP_KERNEL);
209 if (!map) 211 if (!map)
@@ -263,7 +265,7 @@ out_no_root:
263 265
264out_no_bitmap: 266out_no_bitmap:
265 printk("MINIX-fs: bad superblock or unable to read bitmaps\n"); 267 printk("MINIX-fs: bad superblock or unable to read bitmaps\n");
266 out_freemap: 268out_freemap:
267 for (i = 0; i < sbi->s_imap_blocks; i++) 269 for (i = 0; i < sbi->s_imap_blocks; i++)
268 brelse(sbi->s_imap[i]); 270 brelse(sbi->s_imap[i]);
269 for (i = 0; i < sbi->s_zmap_blocks; i++) 271 for (i = 0; i < sbi->s_zmap_blocks; i++)
@@ -276,11 +278,16 @@ out_no_map:
276 printk("MINIX-fs: can't allocate map\n"); 278 printk("MINIX-fs: can't allocate map\n");
277 goto out_release; 279 goto out_release;
278 280
281out_illegal_sb:
282 if (!silent)
283 printk("MINIX-fs: bad superblock\n");
284 goto out_release;
285
279out_no_fs: 286out_no_fs:
280 if (!silent) 287 if (!silent)
281 printk("VFS: Can't find a Minix or Minix V2 filesystem " 288 printk("VFS: Can't find a Minix or Minix V2 filesystem "
282 "on device %s\n", s->s_id); 289 "on device %s\n", s->s_id);
283 out_release: 290out_release:
284 brelse(bh); 291 brelse(bh);
285 goto out; 292 goto out;
286 293
@@ -290,7 +297,7 @@ out_bad_hblock:
290 297
291out_bad_sb: 298out_bad_sb:
292 printk("MINIX-fs: unable to read superblock\n"); 299 printk("MINIX-fs: unable to read superblock\n");
293 out: 300out:
294 s->s_fs_info = NULL; 301 s->s_fs_info = NULL;
295 kfree(sbi); 302 kfree(sbi);
296 return -EINVAL; 303 return -EINVAL;
diff --git a/fs/namei.c b/fs/namei.c
index e01070d7bf58..432d6bc6fab0 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -159,7 +159,7 @@ char * getname(const char __user * filename)
159#ifdef CONFIG_AUDITSYSCALL 159#ifdef CONFIG_AUDITSYSCALL
160void putname(const char *name) 160void putname(const char *name)
161{ 161{
162 if (unlikely(current->audit_context)) 162 if (unlikely(!audit_dummy_context()))
163 audit_putname(name); 163 audit_putname(name);
164 else 164 else
165 __putname(name); 165 __putname(name);
@@ -227,10 +227,10 @@ int generic_permission(struct inode *inode, int mask,
227 227
228int permission(struct inode *inode, int mask, struct nameidata *nd) 228int permission(struct inode *inode, int mask, struct nameidata *nd)
229{ 229{
230 umode_t mode = inode->i_mode;
230 int retval, submask; 231 int retval, submask;
231 232
232 if (mask & MAY_WRITE) { 233 if (mask & MAY_WRITE) {
233 umode_t mode = inode->i_mode;
234 234
235 /* 235 /*
236 * Nobody gets write access to a read-only fs. 236 * Nobody gets write access to a read-only fs.
@@ -247,6 +247,13 @@ int permission(struct inode *inode, int mask, struct nameidata *nd)
247 } 247 }
248 248
249 249
250 /*
251 * MAY_EXEC on regular files requires special handling: We override
252 * filesystem execute permissions if the mode bits aren't set.
253 */
254 if ((mask & MAY_EXEC) && S_ISREG(mode) && !(mode & S_IXUGO))
255 return -EACCES;
256
250 /* Ordinary permission routines do not understand MAY_APPEND. */ 257 /* Ordinary permission routines do not understand MAY_APPEND. */
251 submask = mask & ~MAY_APPEND; 258 submask = mask & ~MAY_APPEND;
252 if (inode->i_op && inode->i_op->permission) 259 if (inode->i_op && inode->i_op->permission)
@@ -1125,7 +1132,7 @@ static int fastcall do_path_lookup(int dfd, const char *name,
1125 retval = link_path_walk(name, nd); 1132 retval = link_path_walk(name, nd);
1126out: 1133out:
1127 if (likely(retval == 0)) { 1134 if (likely(retval == 0)) {
1128 if (unlikely(current->audit_context && nd && nd->dentry && 1135 if (unlikely(!audit_dummy_context() && nd && nd->dentry &&
1129 nd->dentry->d_inode)) 1136 nd->dentry->d_inode))
1130 audit_inode(name, nd->dentry->d_inode); 1137 audit_inode(name, nd->dentry->d_inode);
1131 } 1138 }
@@ -1357,7 +1364,7 @@ static int may_delete(struct inode *dir,struct dentry *victim,int isdir)
1357 return -ENOENT; 1364 return -ENOENT;
1358 1365
1359 BUG_ON(victim->d_parent->d_inode != dir); 1366 BUG_ON(victim->d_parent->d_inode != dir);
1360 audit_inode_child(victim->d_name.name, victim->d_inode, dir->i_ino); 1367 audit_inode_child(victim->d_name.name, victim->d_inode, dir);
1361 1368
1362 error = permission(dir,MAY_WRITE | MAY_EXEC, NULL); 1369 error = permission(dir,MAY_WRITE | MAY_EXEC, NULL);
1363 if (error) 1370 if (error)
@@ -1659,6 +1666,7 @@ do_last:
1659 * It already exists. 1666 * It already exists.
1660 */ 1667 */
1661 mutex_unlock(&dir->d_inode->i_mutex); 1668 mutex_unlock(&dir->d_inode->i_mutex);
1669 audit_inode_update(path.dentry->d_inode);
1662 1670
1663 error = -EEXIST; 1671 error = -EEXIST;
1664 if (flag & O_EXCL) 1672 if (flag & O_EXCL)
@@ -1669,6 +1677,7 @@ do_last:
1669 if (flag & O_NOFOLLOW) 1677 if (flag & O_NOFOLLOW)
1670 goto exit_dput; 1678 goto exit_dput;
1671 } 1679 }
1680
1672 error = -ENOENT; 1681 error = -ENOENT;
1673 if (!path.dentry->d_inode) 1682 if (!path.dentry->d_inode)
1674 goto exit_dput; 1683 goto exit_dput;
@@ -1765,6 +1774,8 @@ struct dentry *lookup_create(struct nameidata *nd, int is_dir)
1765 if (nd->last_type != LAST_NORM) 1774 if (nd->last_type != LAST_NORM)
1766 goto fail; 1775 goto fail;
1767 nd->flags &= ~LOOKUP_PARENT; 1776 nd->flags &= ~LOOKUP_PARENT;
1777 nd->flags |= LOOKUP_CREATE;
1778 nd->intent.open.flags = O_EXCL;
1768 1779
1769 /* 1780 /*
1770 * Do the final lookup. 1781 * Do the final lookup.
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index cc2b874ad5a4..48e892880d5b 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -312,7 +312,13 @@ static void nfs_invalidate_page(struct page *page, unsigned long offset)
312 312
313static int nfs_release_page(struct page *page, gfp_t gfp) 313static int nfs_release_page(struct page *page, gfp_t gfp)
314{ 314{
315 return !nfs_wb_page(page->mapping->host, page); 315 if (gfp & __GFP_FS)
316 return !nfs_wb_page(page->mapping->host, page);
317 else
318 /*
319 * Avoid deadlock on nfs_wait_on_request().
320 */
321 return 0;
316} 322}
317 323
318const struct address_space_operations nfs_file_aops = { 324const struct address_space_operations nfs_file_aops = {
diff --git a/fs/nfs/idmap.c b/fs/nfs/idmap.c
index b81e7ed3c902..07a5dd57646e 100644
--- a/fs/nfs/idmap.c
+++ b/fs/nfs/idmap.c
@@ -130,9 +130,7 @@ nfs_idmap_delete(struct nfs4_client *clp)
130 130
131 if (!idmap) 131 if (!idmap)
132 return; 132 return;
133 dput(idmap->idmap_dentry); 133 rpc_unlink(idmap->idmap_dentry);
134 idmap->idmap_dentry = NULL;
135 rpc_unlink(idmap->idmap_path);
136 clp->cl_idmap = NULL; 134 clp->cl_idmap = NULL;
137 kfree(idmap); 135 kfree(idmap);
138} 136}
diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
index 19b98ca468eb..86b3169c8cac 100644
--- a/fs/nfs/namespace.c
+++ b/fs/nfs/namespace.c
@@ -51,7 +51,7 @@ char *nfs_path(const char *base, const struct dentry *dentry,
51 namelen = dentry->d_name.len; 51 namelen = dentry->d_name.len;
52 buflen -= namelen + 1; 52 buflen -= namelen + 1;
53 if (buflen < 0) 53 if (buflen < 0)
54 goto Elong; 54 goto Elong_unlock;
55 end -= namelen; 55 end -= namelen;
56 memcpy(end, dentry->d_name.name, namelen); 56 memcpy(end, dentry->d_name.name, namelen);
57 *--end = '/'; 57 *--end = '/';
@@ -68,6 +68,8 @@ char *nfs_path(const char *base, const struct dentry *dentry,
68 end -= namelen; 68 end -= namelen;
69 memcpy(end, base, namelen); 69 memcpy(end, base, namelen);
70 return end; 70 return end;
71Elong_unlock:
72 spin_unlock(&dcache_lock);
71Elong: 73Elong:
72 return ERR_PTR(-ENAMETOOLONG); 74 return ERR_PTR(-ENAMETOOLONG);
73} 75}
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index e6ee97f19d81..153898e1331f 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -2668,7 +2668,7 @@ out:
2668 nfs4_set_cached_acl(inode, acl); 2668 nfs4_set_cached_acl(inode, acl);
2669} 2669}
2670 2670
2671static inline ssize_t nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen) 2671static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
2672{ 2672{
2673 struct page *pages[NFS4ACL_MAXPAGES]; 2673 struct page *pages[NFS4ACL_MAXPAGES];
2674 struct nfs_getaclargs args = { 2674 struct nfs_getaclargs args = {
@@ -2721,6 +2721,19 @@ out_free:
2721 return ret; 2721 return ret;
2722} 2722}
2723 2723
2724static ssize_t nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
2725{
2726 struct nfs4_exception exception = { };
2727 ssize_t ret;
2728 do {
2729 ret = __nfs4_get_acl_uncached(inode, buf, buflen);
2730 if (ret >= 0)
2731 break;
2732 ret = nfs4_handle_exception(NFS_SERVER(inode), ret, &exception);
2733 } while (exception.retry);
2734 return ret;
2735}
2736
2724static ssize_t nfs4_proc_get_acl(struct inode *inode, void *buf, size_t buflen) 2737static ssize_t nfs4_proc_get_acl(struct inode *inode, void *buf, size_t buflen)
2725{ 2738{
2726 struct nfs_server *server = NFS_SERVER(inode); 2739 struct nfs_server *server = NFS_SERVER(inode);
@@ -2737,7 +2750,7 @@ static ssize_t nfs4_proc_get_acl(struct inode *inode, void *buf, size_t buflen)
2737 return nfs4_get_acl_uncached(inode, buf, buflen); 2750 return nfs4_get_acl_uncached(inode, buf, buflen);
2738} 2751}
2739 2752
2740static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen) 2753static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
2741{ 2754{
2742 struct nfs_server *server = NFS_SERVER(inode); 2755 struct nfs_server *server = NFS_SERVER(inode);
2743 struct page *pages[NFS4ACL_MAXPAGES]; 2756 struct page *pages[NFS4ACL_MAXPAGES];
@@ -2763,6 +2776,18 @@ static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen
2763 return ret; 2776 return ret;
2764} 2777}
2765 2778
2779static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
2780{
2781 struct nfs4_exception exception = { };
2782 int err;
2783 do {
2784 err = nfs4_handle_exception(NFS_SERVER(inode),
2785 __nfs4_proc_set_acl(inode, buf, buflen),
2786 &exception);
2787 } while (exception.retry);
2788 return err;
2789}
2790
2766static int 2791static int
2767nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server) 2792nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server)
2768{ 2793{
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index 1750d996f49f..730ec8fb31c6 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -3355,7 +3355,7 @@ static int decode_readdir(struct xdr_stream *xdr, struct rpc_rqst *req, struct n
3355 struct kvec *iov = rcvbuf->head; 3355 struct kvec *iov = rcvbuf->head;
3356 unsigned int nr, pglen = rcvbuf->page_len; 3356 unsigned int nr, pglen = rcvbuf->page_len;
3357 uint32_t *end, *entry, *p, *kaddr; 3357 uint32_t *end, *entry, *p, *kaddr;
3358 uint32_t len, attrlen; 3358 uint32_t len, attrlen, xlen;
3359 int hdrlen, recvd, status; 3359 int hdrlen, recvd, status;
3360 3360
3361 status = decode_op_hdr(xdr, OP_READDIR); 3361 status = decode_op_hdr(xdr, OP_READDIR);
@@ -3377,10 +3377,10 @@ static int decode_readdir(struct xdr_stream *xdr, struct rpc_rqst *req, struct n
3377 3377
3378 BUG_ON(pglen + readdir->pgbase > PAGE_CACHE_SIZE); 3378 BUG_ON(pglen + readdir->pgbase > PAGE_CACHE_SIZE);
3379 kaddr = p = (uint32_t *) kmap_atomic(page, KM_USER0); 3379 kaddr = p = (uint32_t *) kmap_atomic(page, KM_USER0);
3380 end = (uint32_t *) ((char *)p + pglen + readdir->pgbase); 3380 end = p + ((pglen + readdir->pgbase) >> 2);
3381 entry = p; 3381 entry = p;
3382 for (nr = 0; *p++; nr++) { 3382 for (nr = 0; *p++; nr++) {
3383 if (p + 3 > end) 3383 if (end - p < 3)
3384 goto short_pkt; 3384 goto short_pkt;
3385 dprintk("cookie = %Lu, ", *((unsigned long long *)p)); 3385 dprintk("cookie = %Lu, ", *((unsigned long long *)p));
3386 p += 2; /* cookie */ 3386 p += 2; /* cookie */
@@ -3389,18 +3389,19 @@ static int decode_readdir(struct xdr_stream *xdr, struct rpc_rqst *req, struct n
3389 printk(KERN_WARNING "NFS: giant filename in readdir (len 0x%x)\n", len); 3389 printk(KERN_WARNING "NFS: giant filename in readdir (len 0x%x)\n", len);
3390 goto err_unmap; 3390 goto err_unmap;
3391 } 3391 }
3392 dprintk("filename = %*s\n", len, (char *)p); 3392 xlen = XDR_QUADLEN(len);
3393 p += XDR_QUADLEN(len); 3393 if (end - p < xlen + 1)
3394 if (p + 1 > end)
3395 goto short_pkt; 3394 goto short_pkt;
3395 dprintk("filename = %*s\n", len, (char *)p);
3396 p += xlen;
3396 len = ntohl(*p++); /* bitmap length */ 3397 len = ntohl(*p++); /* bitmap length */
3397 p += len; 3398 if (end - p < len + 1)
3398 if (p + 1 > end)
3399 goto short_pkt; 3399 goto short_pkt;
3400 p += len;
3400 attrlen = XDR_QUADLEN(ntohl(*p++)); 3401 attrlen = XDR_QUADLEN(ntohl(*p++));
3401 p += attrlen; /* attributes */ 3402 if (end - p < attrlen + 2)
3402 if (p + 2 > end)
3403 goto short_pkt; 3403 goto short_pkt;
3404 p += attrlen; /* attributes */
3404 entry = p; 3405 entry = p;
3405 } 3406 }
3406 if (!nr && (entry[0] != 0 || entry[1] == 0)) 3407 if (!nr && (entry[0] != 0 || entry[1] == 0))
diff --git a/fs/nfs/read.c b/fs/nfs/read.c
index 52bf634260a1..da9cf11c326f 100644
--- a/fs/nfs/read.c
+++ b/fs/nfs/read.c
@@ -63,7 +63,7 @@ struct nfs_read_data *nfs_readdata_alloc(unsigned int pagecount)
63 return p; 63 return p;
64} 64}
65 65
66void nfs_readdata_free(struct nfs_read_data *p) 66static void nfs_readdata_free(struct nfs_read_data *p)
67{ 67{
68 if (p && (p->pagevec != &p->page_array[0])) 68 if (p && (p->pagevec != &p->page_array[0]))
69 kfree(p->pagevec); 69 kfree(p->pagevec);
@@ -116,10 +116,17 @@ static void nfs_readpage_truncate_uninitialised_page(struct nfs_read_data *data)
116 pages = &data->args.pages[base >> PAGE_CACHE_SHIFT]; 116 pages = &data->args.pages[base >> PAGE_CACHE_SHIFT];
117 base &= ~PAGE_CACHE_MASK; 117 base &= ~PAGE_CACHE_MASK;
118 pglen = PAGE_CACHE_SIZE - base; 118 pglen = PAGE_CACHE_SIZE - base;
119 if (pglen < remainder) 119 for (;;) {
120 if (remainder <= pglen) {
121 memclear_highpage_flush(*pages, base, remainder);
122 break;
123 }
120 memclear_highpage_flush(*pages, base, pglen); 124 memclear_highpage_flush(*pages, base, pglen);
121 else 125 pages++;
122 memclear_highpage_flush(*pages, base, remainder); 126 remainder -= pglen;
127 pglen = PAGE_CACHE_SIZE;
128 base = 0;
129 }
123} 130}
124 131
125/* 132/*
@@ -476,6 +483,8 @@ static void nfs_readpage_set_pages_uptodate(struct nfs_read_data *data)
476 unsigned int base = data->args.pgbase; 483 unsigned int base = data->args.pgbase;
477 struct page **pages; 484 struct page **pages;
478 485
486 if (data->res.eof)
487 count = data->args.count;
479 if (unlikely(count == 0)) 488 if (unlikely(count == 0))
480 return; 489 return;
481 pages = &data->args.pages[base >> PAGE_CACHE_SHIFT]; 490 pages = &data->args.pages[base >> PAGE_CACHE_SHIFT];
@@ -483,11 +492,7 @@ static void nfs_readpage_set_pages_uptodate(struct nfs_read_data *data)
483 count += base; 492 count += base;
484 for (;count >= PAGE_CACHE_SIZE; count -= PAGE_CACHE_SIZE, pages++) 493 for (;count >= PAGE_CACHE_SIZE; count -= PAGE_CACHE_SIZE, pages++)
485 SetPageUptodate(*pages); 494 SetPageUptodate(*pages);
486 /* 495 if (count != 0)
487 * Was this an eof or a short read? If the latter, don't mark the page
488 * as uptodate yet.
489 */
490 if (count > 0 && (data->res.eof || data->args.count == data->res.count))
491 SetPageUptodate(*pages); 496 SetPageUptodate(*pages);
492} 497}
493 498
@@ -502,6 +507,8 @@ static void nfs_readpage_set_pages_error(struct nfs_read_data *data)
502 count += base; 507 count += base;
503 for (;count >= PAGE_CACHE_SIZE; count -= PAGE_CACHE_SIZE, pages++) 508 for (;count >= PAGE_CACHE_SIZE; count -= PAGE_CACHE_SIZE, pages++)
504 SetPageError(*pages); 509 SetPageError(*pages);
510 if (count != 0)
511 SetPageError(*pages);
505} 512}
506 513
507/* 514/*
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 86bac6a5008e..50774991f8d5 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -137,7 +137,7 @@ struct nfs_write_data *nfs_writedata_alloc(unsigned int pagecount)
137 return p; 137 return p;
138} 138}
139 139
140void nfs_writedata_free(struct nfs_write_data *p) 140static void nfs_writedata_free(struct nfs_write_data *p)
141{ 141{
142 if (p && (p->pagevec != &p->page_array[0])) 142 if (p && (p->pagevec != &p->page_array[0]))
143 kfree(p->pagevec); 143 kfree(p->pagevec);
diff --git a/fs/nfsd/nfsfh.c b/fs/nfsd/nfsfh.c
index ecc439d2565f..501d83884530 100644
--- a/fs/nfsd/nfsfh.c
+++ b/fs/nfsd/nfsfh.c
@@ -187,6 +187,11 @@ fh_verify(struct svc_rqst *rqstp, struct svc_fh *fhp, int type, int access)
187 goto out; 187 goto out;
188 } 188 }
189 189
190 /* Set user creds for this exportpoint */
191 error = nfserrno(nfsd_setuser(rqstp, exp));
192 if (error)
193 goto out;
194
190 /* 195 /*
191 * Look up the dentry using the NFS file handle. 196 * Look up the dentry using the NFS file handle.
192 */ 197 */
@@ -241,16 +246,17 @@ fh_verify(struct svc_rqst *rqstp, struct svc_fh *fhp, int type, int access)
241 dprintk("nfsd: fh_verify - just checking\n"); 246 dprintk("nfsd: fh_verify - just checking\n");
242 dentry = fhp->fh_dentry; 247 dentry = fhp->fh_dentry;
243 exp = fhp->fh_export; 248 exp = fhp->fh_export;
249 /* Set user creds for this exportpoint; necessary even
250 * in the "just checking" case because this may be a
251 * filehandle that was created by fh_compose, and that
252 * is about to be used in another nfsv4 compound
253 * operation */
254 error = nfserrno(nfsd_setuser(rqstp, exp));
255 if (error)
256 goto out;
244 } 257 }
245 cache_get(&exp->h); 258 cache_get(&exp->h);
246 259
247 /* Set user creds for this exportpoint; necessary even in the "just
248 * checking" case because this may be a filehandle that was created by
249 * fh_compose, and that is about to be used in another nfsv4 compound
250 * operation */
251 error = nfserrno(nfsd_setuser(rqstp, exp));
252 if (error)
253 goto out;
254 260
255 error = nfsd_mode_check(rqstp, dentry->d_inode->i_mode, type); 261 error = nfsd_mode_check(rqstp, dentry->d_inode->i_mode, type);
256 if (error) 262 if (error)
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index 1b8346dd0572..9503240ef0e5 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -2375,7 +2375,6 @@ leave:
2375 mlog(0, "returning %d\n", ret); 2375 mlog(0, "returning %d\n", ret);
2376 return ret; 2376 return ret;
2377} 2377}
2378EXPORT_SYMBOL_GPL(dlm_migrate_lockres);
2379 2378
2380int dlm_lock_basts_flushed(struct dlm_ctxt *dlm, struct dlm_lock *lock) 2379int dlm_lock_basts_flushed(struct dlm_ctxt *dlm, struct dlm_lock *lock)
2381{ 2380{
diff --git a/fs/ocfs2/dlm/dlmunlock.c b/fs/ocfs2/dlm/dlmunlock.c
index b0c3134f4f70..37be4b2e0d4a 100644
--- a/fs/ocfs2/dlm/dlmunlock.c
+++ b/fs/ocfs2/dlm/dlmunlock.c
@@ -155,7 +155,7 @@ static enum dlm_status dlmunlock_common(struct dlm_ctxt *dlm,
155 else 155 else
156 status = dlm_get_unlock_actions(dlm, res, lock, lksb, &actions); 156 status = dlm_get_unlock_actions(dlm, res, lock, lksb, &actions);
157 157
158 if (status != DLM_NORMAL) 158 if (status != DLM_NORMAL && (status != DLM_CANCELGRANT || !master_node))
159 goto leave; 159 goto leave;
160 160
161 /* By now this has been masked out of cancel requests. */ 161 /* By now this has been masked out of cancel requests. */
@@ -183,8 +183,7 @@ static enum dlm_status dlmunlock_common(struct dlm_ctxt *dlm,
183 spin_lock(&lock->spinlock); 183 spin_lock(&lock->spinlock);
184 /* if the master told us the lock was already granted, 184 /* if the master told us the lock was already granted,
185 * let the ast handle all of these actions */ 185 * let the ast handle all of these actions */
186 if (status == DLM_NORMAL && 186 if (status == DLM_CANCELGRANT) {
187 lksb->status == DLM_CANCELGRANT) {
188 actions &= ~(DLM_UNLOCK_REMOVE_LOCK| 187 actions &= ~(DLM_UNLOCK_REMOVE_LOCK|
189 DLM_UNLOCK_REGRANT_LOCK| 188 DLM_UNLOCK_REGRANT_LOCK|
190 DLM_UNLOCK_CLEAR_CONVERT_TYPE); 189 DLM_UNLOCK_CLEAR_CONVERT_TYPE);
@@ -349,14 +348,9 @@ static enum dlm_status dlm_send_remote_unlock_request(struct dlm_ctxt *dlm,
349 vec, veclen, owner, &status); 348 vec, veclen, owner, &status);
350 if (tmpret >= 0) { 349 if (tmpret >= 0) {
351 // successfully sent and received 350 // successfully sent and received
352 if (status == DLM_CANCELGRANT) 351 if (status == DLM_FORWARD)
353 ret = DLM_NORMAL;
354 else if (status == DLM_FORWARD) {
355 mlog(0, "master was in-progress. retry\n"); 352 mlog(0, "master was in-progress. retry\n");
356 ret = DLM_FORWARD; 353 ret = status;
357 } else
358 ret = status;
359 lksb->status = status;
360 } else { 354 } else {
361 mlog_errno(tmpret); 355 mlog_errno(tmpret);
362 if (dlm_is_host_down(tmpret)) { 356 if (dlm_is_host_down(tmpret)) {
@@ -372,7 +366,6 @@ static enum dlm_status dlm_send_remote_unlock_request(struct dlm_ctxt *dlm,
372 /* something bad. this will BUG in ocfs2 */ 366 /* something bad. this will BUG in ocfs2 */
373 ret = dlm_err_to_dlm_status(tmpret); 367 ret = dlm_err_to_dlm_status(tmpret);
374 } 368 }
375 lksb->status = ret;
376 } 369 }
377 370
378 return ret; 371 return ret;
@@ -483,6 +476,10 @@ int dlm_unlock_lock_handler(struct o2net_msg *msg, u32 len, void *data)
483 476
484 /* lock was found on queue */ 477 /* lock was found on queue */
485 lksb = lock->lksb; 478 lksb = lock->lksb;
479 if (flags & (LKM_VALBLK|LKM_PUT_LVB) &&
480 lock->ml.type != LKM_EXMODE)
481 flags &= ~(LKM_VALBLK|LKM_PUT_LVB);
482
486 /* unlockast only called on originating node */ 483 /* unlockast only called on originating node */
487 if (flags & LKM_PUT_LVB) { 484 if (flags & LKM_PUT_LVB) {
488 lksb->flags |= DLM_LKSB_PUT_LVB; 485 lksb->flags |= DLM_LKSB_PUT_LVB;
@@ -507,11 +504,8 @@ not_found:
507 "cookie=%u:%llu\n", 504 "cookie=%u:%llu\n",
508 dlm_get_lock_cookie_node(unlock->cookie), 505 dlm_get_lock_cookie_node(unlock->cookie),
509 dlm_get_lock_cookie_seq(unlock->cookie)); 506 dlm_get_lock_cookie_seq(unlock->cookie));
510 else { 507 else
511 /* send the lksb->status back to the other node */
512 status = lksb->status;
513 dlm_lock_put(lock); 508 dlm_lock_put(lock);
514 }
515 509
516leave: 510leave:
517 if (res) 511 if (res)
@@ -533,26 +527,22 @@ static enum dlm_status dlm_get_cancel_actions(struct dlm_ctxt *dlm,
533 527
534 if (dlm_lock_on_list(&res->blocked, lock)) { 528 if (dlm_lock_on_list(&res->blocked, lock)) {
535 /* cancel this outright */ 529 /* cancel this outright */
536 lksb->status = DLM_NORMAL;
537 status = DLM_NORMAL; 530 status = DLM_NORMAL;
538 *actions = (DLM_UNLOCK_CALL_AST | 531 *actions = (DLM_UNLOCK_CALL_AST |
539 DLM_UNLOCK_REMOVE_LOCK); 532 DLM_UNLOCK_REMOVE_LOCK);
540 } else if (dlm_lock_on_list(&res->converting, lock)) { 533 } else if (dlm_lock_on_list(&res->converting, lock)) {
541 /* cancel the request, put back on granted */ 534 /* cancel the request, put back on granted */
542 lksb->status = DLM_NORMAL;
543 status = DLM_NORMAL; 535 status = DLM_NORMAL;
544 *actions = (DLM_UNLOCK_CALL_AST | 536 *actions = (DLM_UNLOCK_CALL_AST |
545 DLM_UNLOCK_REMOVE_LOCK | 537 DLM_UNLOCK_REMOVE_LOCK |
546 DLM_UNLOCK_REGRANT_LOCK | 538 DLM_UNLOCK_REGRANT_LOCK |
547 DLM_UNLOCK_CLEAR_CONVERT_TYPE); 539 DLM_UNLOCK_CLEAR_CONVERT_TYPE);
548 } else if (dlm_lock_on_list(&res->granted, lock)) { 540 } else if (dlm_lock_on_list(&res->granted, lock)) {
549 /* too late, already granted. DLM_CANCELGRANT */ 541 /* too late, already granted. */
550 lksb->status = DLM_CANCELGRANT; 542 status = DLM_CANCELGRANT;
551 status = DLM_NORMAL;
552 *actions = DLM_UNLOCK_CALL_AST; 543 *actions = DLM_UNLOCK_CALL_AST;
553 } else { 544 } else {
554 mlog(ML_ERROR, "lock to cancel is not on any list!\n"); 545 mlog(ML_ERROR, "lock to cancel is not on any list!\n");
555 lksb->status = DLM_IVLOCKID;
556 status = DLM_IVLOCKID; 546 status = DLM_IVLOCKID;
557 *actions = 0; 547 *actions = 0;
558 } 548 }
@@ -569,13 +559,11 @@ static enum dlm_status dlm_get_unlock_actions(struct dlm_ctxt *dlm,
569 559
570 /* unlock request */ 560 /* unlock request */
571 if (!dlm_lock_on_list(&res->granted, lock)) { 561 if (!dlm_lock_on_list(&res->granted, lock)) {
572 lksb->status = DLM_DENIED;
573 status = DLM_DENIED; 562 status = DLM_DENIED;
574 dlm_error(status); 563 dlm_error(status);
575 *actions = 0; 564 *actions = 0;
576 } else { 565 } else {
577 /* unlock granted lock */ 566 /* unlock granted lock */
578 lksb->status = DLM_NORMAL;
579 status = DLM_NORMAL; 567 status = DLM_NORMAL;
580 *actions = (DLM_UNLOCK_FREE_LOCK | 568 *actions = (DLM_UNLOCK_FREE_LOCK |
581 DLM_UNLOCK_CALL_AST | 569 DLM_UNLOCK_CALL_AST |
@@ -632,6 +620,8 @@ retry:
632 620
633 spin_lock(&res->spinlock); 621 spin_lock(&res->spinlock);
634 is_master = (res->owner == dlm->node_num); 622 is_master = (res->owner == dlm->node_num);
623 if (flags & LKM_VALBLK && lock->ml.type != LKM_EXMODE)
624 flags &= ~LKM_VALBLK;
635 spin_unlock(&res->spinlock); 625 spin_unlock(&res->spinlock);
636 626
637 if (is_master) { 627 if (is_master) {
@@ -665,7 +655,7 @@ retry:
665 } 655 }
666 656
667 if (call_ast) { 657 if (call_ast) {
668 mlog(0, "calling unlockast(%p, %d)\n", data, lksb->status); 658 mlog(0, "calling unlockast(%p, %d)\n", data, status);
669 if (is_master) { 659 if (is_master) {
670 /* it is possible that there is one last bast 660 /* it is possible that there is one last bast
671 * pending. make sure it is flushed, then 661 * pending. make sure it is flushed, then
@@ -677,9 +667,12 @@ retry:
677 wait_event(dlm->ast_wq, 667 wait_event(dlm->ast_wq,
678 dlm_lock_basts_flushed(dlm, lock)); 668 dlm_lock_basts_flushed(dlm, lock));
679 } 669 }
680 (*unlockast)(data, lksb->status); 670 (*unlockast)(data, status);
681 } 671 }
682 672
673 if (status == DLM_CANCELGRANT)
674 status = DLM_NORMAL;
675
683 if (status == DLM_NORMAL) { 676 if (status == DLM_NORMAL) {
684 mlog(0, "kicking the thread\n"); 677 mlog(0, "kicking the thread\n");
685 dlm_kick_thread(dlm, res); 678 dlm_kick_thread(dlm, res);
diff --git a/fs/ocfs2/localalloc.c b/fs/ocfs2/localalloc.c
index 0d1973ea32b0..1f17a4d08287 100644
--- a/fs/ocfs2/localalloc.c
+++ b/fs/ocfs2/localalloc.c
@@ -840,6 +840,12 @@ static int ocfs2_local_alloc_new_window(struct ocfs2_super *osb,
840 840
841 mlog(0, "Allocating %u clusters for a new window.\n", 841 mlog(0, "Allocating %u clusters for a new window.\n",
842 ocfs2_local_alloc_window_bits(osb)); 842 ocfs2_local_alloc_window_bits(osb));
843
844 /* Instruct the allocation code to try the most recently used
845 * cluster group. We'll re-record the group used this pass
846 * below. */
847 ac->ac_last_group = osb->la_last_gd;
848
843 /* we used the generic suballoc reserve function, but we set 849 /* we used the generic suballoc reserve function, but we set
844 * everything up nicely, so there's no reason why we can't use 850 * everything up nicely, so there's no reason why we can't use
845 * the more specific cluster api to claim bits. */ 851 * the more specific cluster api to claim bits. */
@@ -852,6 +858,8 @@ static int ocfs2_local_alloc_new_window(struct ocfs2_super *osb,
852 goto bail; 858 goto bail;
853 } 859 }
854 860
861 osb->la_last_gd = ac->ac_last_group;
862
855 la->la_bm_off = cpu_to_le32(cluster_off); 863 la->la_bm_off = cpu_to_le32(cluster_off);
856 alloc->id1.bitmap1.i_total = cpu_to_le32(cluster_count); 864 alloc->id1.bitmap1.i_total = cpu_to_le32(cluster_count);
857 /* just in case... In the future when we find space ourselves, 865 /* just in case... In the future when we find space ourselves,
diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h
index cd4a6f253d13..0462a7f4e21b 100644
--- a/fs/ocfs2/ocfs2.h
+++ b/fs/ocfs2/ocfs2.h
@@ -197,7 +197,6 @@ struct ocfs2_super
197 struct ocfs2_node_map recovery_map; 197 struct ocfs2_node_map recovery_map;
198 struct ocfs2_node_map umount_map; 198 struct ocfs2_node_map umount_map;
199 199
200 u32 num_clusters;
201 u64 root_blkno; 200 u64 root_blkno;
202 u64 system_dir_blkno; 201 u64 system_dir_blkno;
203 u64 bitmap_blkno; 202 u64 bitmap_blkno;
@@ -237,6 +236,7 @@ struct ocfs2_super
237 236
238 enum ocfs2_local_alloc_state local_alloc_state; 237 enum ocfs2_local_alloc_state local_alloc_state;
239 struct buffer_head *local_alloc_bh; 238 struct buffer_head *local_alloc_bh;
239 u64 la_last_gd;
240 240
241 /* Next two fields are for local node slot recovery during 241 /* Next two fields are for local node slot recovery during
242 * mount. */ 242 * mount. */
diff --git a/fs/ocfs2/suballoc.c b/fs/ocfs2/suballoc.c
index 195523090c87..9d91e66f51a9 100644
--- a/fs/ocfs2/suballoc.c
+++ b/fs/ocfs2/suballoc.c
@@ -70,12 +70,6 @@ static int ocfs2_block_group_search(struct inode *inode,
70 struct buffer_head *group_bh, 70 struct buffer_head *group_bh,
71 u32 bits_wanted, u32 min_bits, 71 u32 bits_wanted, u32 min_bits,
72 u16 *bit_off, u16 *bits_found); 72 u16 *bit_off, u16 *bits_found);
73static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
74 u32 bits_wanted,
75 u32 min_bits,
76 u16 *bit_off,
77 unsigned int *num_bits,
78 u64 *bg_blkno);
79static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb, 73static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
80 struct ocfs2_alloc_context *ac, 74 struct ocfs2_alloc_context *ac,
81 u32 bits_wanted, 75 u32 bits_wanted,
@@ -85,11 +79,6 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
85 u64 *bg_blkno); 79 u64 *bg_blkno);
86static int ocfs2_test_bg_bit_allocatable(struct buffer_head *bg_bh, 80static int ocfs2_test_bg_bit_allocatable(struct buffer_head *bg_bh,
87 int nr); 81 int nr);
88static int ocfs2_block_group_find_clear_bits(struct ocfs2_super *osb,
89 struct buffer_head *bg_bh,
90 unsigned int bits_wanted,
91 u16 *bit_off,
92 u16 *bits_found);
93static inline int ocfs2_block_group_set_bits(struct ocfs2_journal_handle *handle, 82static inline int ocfs2_block_group_set_bits(struct ocfs2_journal_handle *handle,
94 struct inode *alloc_inode, 83 struct inode *alloc_inode,
95 struct ocfs2_group_desc *bg, 84 struct ocfs2_group_desc *bg,
@@ -143,6 +132,64 @@ static u32 ocfs2_bits_per_group(struct ocfs2_chain_list *cl)
143 return (u32)le16_to_cpu(cl->cl_cpg) * (u32)le16_to_cpu(cl->cl_bpc); 132 return (u32)le16_to_cpu(cl->cl_cpg) * (u32)le16_to_cpu(cl->cl_bpc);
144} 133}
145 134
135/* somewhat more expensive than our other checks, so use sparingly. */
136static int ocfs2_check_group_descriptor(struct super_block *sb,
137 struct ocfs2_dinode *di,
138 struct ocfs2_group_desc *gd)
139{
140 unsigned int max_bits;
141
142 if (!OCFS2_IS_VALID_GROUP_DESC(gd)) {
143 OCFS2_RO_ON_INVALID_GROUP_DESC(sb, gd);
144 return -EIO;
145 }
146
147 if (di->i_blkno != gd->bg_parent_dinode) {
148 ocfs2_error(sb, "Group descriptor # %llu has bad parent "
149 "pointer (%llu, expected %llu)",
150 (unsigned long long)le64_to_cpu(gd->bg_blkno),
151 (unsigned long long)le64_to_cpu(gd->bg_parent_dinode),
152 (unsigned long long)le64_to_cpu(di->i_blkno));
153 return -EIO;
154 }
155
156 max_bits = le16_to_cpu(di->id2.i_chain.cl_cpg) * le16_to_cpu(di->id2.i_chain.cl_bpc);
157 if (le16_to_cpu(gd->bg_bits) > max_bits) {
158 ocfs2_error(sb, "Group descriptor # %llu has bit count of %u",
159 (unsigned long long)le64_to_cpu(gd->bg_blkno),
160 le16_to_cpu(gd->bg_bits));
161 return -EIO;
162 }
163
164 if (le16_to_cpu(gd->bg_chain) >=
165 le16_to_cpu(di->id2.i_chain.cl_next_free_rec)) {
166 ocfs2_error(sb, "Group descriptor # %llu has bad chain %u",
167 (unsigned long long)le64_to_cpu(gd->bg_blkno),
168 le16_to_cpu(gd->bg_chain));
169 return -EIO;
170 }
171
172 if (le16_to_cpu(gd->bg_free_bits_count) > le16_to_cpu(gd->bg_bits)) {
173 ocfs2_error(sb, "Group descriptor # %llu has bit count %u but "
174 "claims that %u are free",
175 (unsigned long long)le64_to_cpu(gd->bg_blkno),
176 le16_to_cpu(gd->bg_bits),
177 le16_to_cpu(gd->bg_free_bits_count));
178 return -EIO;
179 }
180
181 if (le16_to_cpu(gd->bg_bits) > (8 * le16_to_cpu(gd->bg_size))) {
182 ocfs2_error(sb, "Group descriptor # %llu has bit count %u but "
183 "max bitmap bits of %u",
184 (unsigned long long)le64_to_cpu(gd->bg_blkno),
185 le16_to_cpu(gd->bg_bits),
186 8 * le16_to_cpu(gd->bg_size));
187 return -EIO;
188 }
189
190 return 0;
191}
192
146static int ocfs2_block_group_fill(struct ocfs2_journal_handle *handle, 193static int ocfs2_block_group_fill(struct ocfs2_journal_handle *handle,
147 struct inode *alloc_inode, 194 struct inode *alloc_inode,
148 struct buffer_head *bg_bh, 195 struct buffer_head *bg_bh,
@@ -663,6 +710,7 @@ static int ocfs2_test_bg_bit_allocatable(struct buffer_head *bg_bh,
663static int ocfs2_block_group_find_clear_bits(struct ocfs2_super *osb, 710static int ocfs2_block_group_find_clear_bits(struct ocfs2_super *osb,
664 struct buffer_head *bg_bh, 711 struct buffer_head *bg_bh,
665 unsigned int bits_wanted, 712 unsigned int bits_wanted,
713 unsigned int total_bits,
666 u16 *bit_off, 714 u16 *bit_off,
667 u16 *bits_found) 715 u16 *bits_found)
668{ 716{
@@ -679,10 +727,8 @@ static int ocfs2_block_group_find_clear_bits(struct ocfs2_super *osb,
679 found = start = best_offset = best_size = 0; 727 found = start = best_offset = best_size = 0;
680 bitmap = bg->bg_bitmap; 728 bitmap = bg->bg_bitmap;
681 729
682 while((offset = ocfs2_find_next_zero_bit(bitmap, 730 while((offset = ocfs2_find_next_zero_bit(bitmap, total_bits, start)) != -1) {
683 le16_to_cpu(bg->bg_bits), 731 if (offset == total_bits)
684 start)) != -1) {
685 if (offset == le16_to_cpu(bg->bg_bits))
686 break; 732 break;
687 733
688 if (!ocfs2_test_bg_bit_allocatable(bg_bh, offset)) { 734 if (!ocfs2_test_bg_bit_allocatable(bg_bh, offset)) {
@@ -911,14 +957,35 @@ static int ocfs2_cluster_group_search(struct inode *inode,
911{ 957{
912 int search = -ENOSPC; 958 int search = -ENOSPC;
913 int ret; 959 int ret;
914 struct ocfs2_group_desc *bg = (struct ocfs2_group_desc *) group_bh->b_data; 960 struct ocfs2_group_desc *gd = (struct ocfs2_group_desc *) group_bh->b_data;
915 u16 tmp_off, tmp_found; 961 u16 tmp_off, tmp_found;
962 unsigned int max_bits, gd_cluster_off;
916 963
917 BUG_ON(!ocfs2_is_cluster_bitmap(inode)); 964 BUG_ON(!ocfs2_is_cluster_bitmap(inode));
918 965
919 if (bg->bg_free_bits_count) { 966 if (gd->bg_free_bits_count) {
967 max_bits = le16_to_cpu(gd->bg_bits);
968
969 /* Tail groups in cluster bitmaps which aren't cpg
970 * aligned are prone to partial extention by a failed
971 * fs resize. If the file system resize never got to
972 * update the dinode cluster count, then we don't want
973 * to trust any clusters past it, regardless of what
974 * the group descriptor says. */
975 gd_cluster_off = ocfs2_blocks_to_clusters(inode->i_sb,
976 le64_to_cpu(gd->bg_blkno));
977 if ((gd_cluster_off + max_bits) >
978 OCFS2_I(inode)->ip_clusters) {
979 max_bits = OCFS2_I(inode)->ip_clusters - gd_cluster_off;
980 mlog(0, "Desc %llu, bg_bits %u, clusters %u, use %u\n",
981 (unsigned long long)le64_to_cpu(gd->bg_blkno),
982 le16_to_cpu(gd->bg_bits),
983 OCFS2_I(inode)->ip_clusters, max_bits);
984 }
985
920 ret = ocfs2_block_group_find_clear_bits(OCFS2_SB(inode->i_sb), 986 ret = ocfs2_block_group_find_clear_bits(OCFS2_SB(inode->i_sb),
921 group_bh, bits_wanted, 987 group_bh, bits_wanted,
988 max_bits,
922 &tmp_off, &tmp_found); 989 &tmp_off, &tmp_found);
923 if (ret) 990 if (ret)
924 return ret; 991 return ret;
@@ -951,17 +1018,109 @@ static int ocfs2_block_group_search(struct inode *inode,
951 if (bg->bg_free_bits_count) 1018 if (bg->bg_free_bits_count)
952 ret = ocfs2_block_group_find_clear_bits(OCFS2_SB(inode->i_sb), 1019 ret = ocfs2_block_group_find_clear_bits(OCFS2_SB(inode->i_sb),
953 group_bh, bits_wanted, 1020 group_bh, bits_wanted,
1021 le16_to_cpu(bg->bg_bits),
954 bit_off, bits_found); 1022 bit_off, bits_found);
955 1023
956 return ret; 1024 return ret;
957} 1025}
958 1026
1027static int ocfs2_alloc_dinode_update_counts(struct inode *inode,
1028 struct ocfs2_journal_handle *handle,
1029 struct buffer_head *di_bh,
1030 u32 num_bits,
1031 u16 chain)
1032{
1033 int ret;
1034 u32 tmp_used;
1035 struct ocfs2_dinode *di = (struct ocfs2_dinode *) di_bh->b_data;
1036 struct ocfs2_chain_list *cl = (struct ocfs2_chain_list *) &di->id2.i_chain;
1037
1038 ret = ocfs2_journal_access(handle, inode, di_bh,
1039 OCFS2_JOURNAL_ACCESS_WRITE);
1040 if (ret < 0) {
1041 mlog_errno(ret);
1042 goto out;
1043 }
1044
1045 tmp_used = le32_to_cpu(di->id1.bitmap1.i_used);
1046 di->id1.bitmap1.i_used = cpu_to_le32(num_bits + tmp_used);
1047 le32_add_cpu(&cl->cl_recs[chain].c_free, -num_bits);
1048
1049 ret = ocfs2_journal_dirty(handle, di_bh);
1050 if (ret < 0)
1051 mlog_errno(ret);
1052
1053out:
1054 return ret;
1055}
1056
1057static int ocfs2_search_one_group(struct ocfs2_alloc_context *ac,
1058 u32 bits_wanted,
1059 u32 min_bits,
1060 u16 *bit_off,
1061 unsigned int *num_bits,
1062 u64 gd_blkno,
1063 u16 *bits_left)
1064{
1065 int ret;
1066 u16 found;
1067 struct buffer_head *group_bh = NULL;
1068 struct ocfs2_group_desc *gd;
1069 struct inode *alloc_inode = ac->ac_inode;
1070 struct ocfs2_journal_handle *handle = ac->ac_handle;
1071
1072 ret = ocfs2_read_block(OCFS2_SB(alloc_inode->i_sb), gd_blkno,
1073 &group_bh, OCFS2_BH_CACHED, alloc_inode);
1074 if (ret < 0) {
1075 mlog_errno(ret);
1076 return ret;
1077 }
1078
1079 gd = (struct ocfs2_group_desc *) group_bh->b_data;
1080 if (!OCFS2_IS_VALID_GROUP_DESC(gd)) {
1081 OCFS2_RO_ON_INVALID_GROUP_DESC(alloc_inode->i_sb, gd);
1082 ret = -EIO;
1083 goto out;
1084 }
1085
1086 ret = ac->ac_group_search(alloc_inode, group_bh, bits_wanted, min_bits,
1087 bit_off, &found);
1088 if (ret < 0) {
1089 if (ret != -ENOSPC)
1090 mlog_errno(ret);
1091 goto out;
1092 }
1093
1094 *num_bits = found;
1095
1096 ret = ocfs2_alloc_dinode_update_counts(alloc_inode, handle, ac->ac_bh,
1097 *num_bits,
1098 le16_to_cpu(gd->bg_chain));
1099 if (ret < 0) {
1100 mlog_errno(ret);
1101 goto out;
1102 }
1103
1104 ret = ocfs2_block_group_set_bits(handle, alloc_inode, gd, group_bh,
1105 *bit_off, *num_bits);
1106 if (ret < 0)
1107 mlog_errno(ret);
1108
1109 *bits_left = le16_to_cpu(gd->bg_free_bits_count);
1110
1111out:
1112 brelse(group_bh);
1113
1114 return ret;
1115}
1116
959static int ocfs2_search_chain(struct ocfs2_alloc_context *ac, 1117static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
960 u32 bits_wanted, 1118 u32 bits_wanted,
961 u32 min_bits, 1119 u32 min_bits,
962 u16 *bit_off, 1120 u16 *bit_off,
963 unsigned int *num_bits, 1121 unsigned int *num_bits,
964 u64 *bg_blkno) 1122 u64 *bg_blkno,
1123 u16 *bits_left)
965{ 1124{
966 int status; 1125 int status;
967 u16 chain, tmp_bits; 1126 u16 chain, tmp_bits;
@@ -988,9 +1147,9 @@ static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
988 goto bail; 1147 goto bail;
989 } 1148 }
990 bg = (struct ocfs2_group_desc *) group_bh->b_data; 1149 bg = (struct ocfs2_group_desc *) group_bh->b_data;
991 if (!OCFS2_IS_VALID_GROUP_DESC(bg)) { 1150 status = ocfs2_check_group_descriptor(alloc_inode->i_sb, fe, bg);
992 OCFS2_RO_ON_INVALID_GROUP_DESC(alloc_inode->i_sb, bg); 1151 if (status) {
993 status = -EIO; 1152 mlog_errno(status);
994 goto bail; 1153 goto bail;
995 } 1154 }
996 1155
@@ -1018,9 +1177,9 @@ static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
1018 goto bail; 1177 goto bail;
1019 } 1178 }
1020 bg = (struct ocfs2_group_desc *) group_bh->b_data; 1179 bg = (struct ocfs2_group_desc *) group_bh->b_data;
1021 if (!OCFS2_IS_VALID_GROUP_DESC(bg)) { 1180 status = ocfs2_check_group_descriptor(alloc_inode->i_sb, fe, bg);
1022 OCFS2_RO_ON_INVALID_GROUP_DESC(alloc_inode->i_sb, bg); 1181 if (status) {
1023 status = -EIO; 1182 mlog_errno(status);
1024 goto bail; 1183 goto bail;
1025 } 1184 }
1026 } 1185 }
@@ -1099,6 +1258,7 @@ static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
1099 (unsigned long long)fe->i_blkno); 1258 (unsigned long long)fe->i_blkno);
1100 1259
1101 *bg_blkno = le64_to_cpu(bg->bg_blkno); 1260 *bg_blkno = le64_to_cpu(bg->bg_blkno);
1261 *bits_left = le16_to_cpu(bg->bg_free_bits_count);
1102bail: 1262bail:
1103 if (group_bh) 1263 if (group_bh)
1104 brelse(group_bh); 1264 brelse(group_bh);
@@ -1120,6 +1280,8 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
1120{ 1280{
1121 int status; 1281 int status;
1122 u16 victim, i; 1282 u16 victim, i;
1283 u16 bits_left = 0;
1284 u64 hint_blkno = ac->ac_last_group;
1123 struct ocfs2_chain_list *cl; 1285 struct ocfs2_chain_list *cl;
1124 struct ocfs2_dinode *fe; 1286 struct ocfs2_dinode *fe;
1125 1287
@@ -1146,6 +1308,28 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
1146 goto bail; 1308 goto bail;
1147 } 1309 }
1148 1310
1311 if (hint_blkno) {
1312 /* Attempt to short-circuit the usual search mechanism
1313 * by jumping straight to the most recently used
1314 * allocation group. This helps us mantain some
1315 * contiguousness across allocations. */
1316 status = ocfs2_search_one_group(ac, bits_wanted, min_bits,
1317 bit_off, num_bits,
1318 hint_blkno, &bits_left);
1319 if (!status) {
1320 /* Be careful to update *bg_blkno here as the
1321 * caller is expecting it to be filled in, and
1322 * ocfs2_search_one_group() won't do that for
1323 * us. */
1324 *bg_blkno = hint_blkno;
1325 goto set_hint;
1326 }
1327 if (status < 0 && status != -ENOSPC) {
1328 mlog_errno(status);
1329 goto bail;
1330 }
1331 }
1332
1149 cl = (struct ocfs2_chain_list *) &fe->id2.i_chain; 1333 cl = (struct ocfs2_chain_list *) &fe->id2.i_chain;
1150 1334
1151 victim = ocfs2_find_victim_chain(cl); 1335 victim = ocfs2_find_victim_chain(cl);
@@ -1153,9 +1337,9 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
1153 ac->ac_allow_chain_relink = 1; 1337 ac->ac_allow_chain_relink = 1;
1154 1338
1155 status = ocfs2_search_chain(ac, bits_wanted, min_bits, bit_off, 1339 status = ocfs2_search_chain(ac, bits_wanted, min_bits, bit_off,
1156 num_bits, bg_blkno); 1340 num_bits, bg_blkno, &bits_left);
1157 if (!status) 1341 if (!status)
1158 goto bail; 1342 goto set_hint;
1159 if (status < 0 && status != -ENOSPC) { 1343 if (status < 0 && status != -ENOSPC) {
1160 mlog_errno(status); 1344 mlog_errno(status);
1161 goto bail; 1345 goto bail;
@@ -1177,8 +1361,8 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
1177 1361
1178 ac->ac_chain = i; 1362 ac->ac_chain = i;
1179 status = ocfs2_search_chain(ac, bits_wanted, min_bits, 1363 status = ocfs2_search_chain(ac, bits_wanted, min_bits,
1180 bit_off, num_bits, 1364 bit_off, num_bits, bg_blkno,
1181 bg_blkno); 1365 &bits_left);
1182 if (!status) 1366 if (!status)
1183 break; 1367 break;
1184 if (status < 0 && status != -ENOSPC) { 1368 if (status < 0 && status != -ENOSPC) {
@@ -1186,8 +1370,19 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
1186 goto bail; 1370 goto bail;
1187 } 1371 }
1188 } 1372 }
1189bail:
1190 1373
1374set_hint:
1375 if (status != -ENOSPC) {
1376 /* If the next search of this group is not likely to
1377 * yield a suitable extent, then we reset the last
1378 * group hint so as to not waste a disk read */
1379 if (bits_left < min_bits)
1380 ac->ac_last_group = 0;
1381 else
1382 ac->ac_last_group = *bg_blkno;
1383 }
1384
1385bail:
1191 mlog_exit(status); 1386 mlog_exit(status);
1192 return status; 1387 return status;
1193} 1388}
@@ -1341,7 +1536,7 @@ int ocfs2_claim_clusters(struct ocfs2_super *osb,
1341{ 1536{
1342 int status; 1537 int status;
1343 unsigned int bits_wanted = ac->ac_bits_wanted - ac->ac_bits_given; 1538 unsigned int bits_wanted = ac->ac_bits_wanted - ac->ac_bits_given;
1344 u64 bg_blkno; 1539 u64 bg_blkno = 0;
1345 u16 bg_bit_off; 1540 u16 bg_bit_off;
1346 1541
1347 mlog_entry_void(); 1542 mlog_entry_void();
@@ -1494,9 +1689,9 @@ static int ocfs2_free_suballoc_bits(struct ocfs2_journal_handle *handle,
1494 } 1689 }
1495 1690
1496 group = (struct ocfs2_group_desc *) group_bh->b_data; 1691 group = (struct ocfs2_group_desc *) group_bh->b_data;
1497 if (!OCFS2_IS_VALID_GROUP_DESC(group)) { 1692 status = ocfs2_check_group_descriptor(alloc_inode->i_sb, fe, group);
1498 OCFS2_RO_ON_INVALID_GROUP_DESC(alloc_inode->i_sb, group); 1693 if (status) {
1499 status = -EIO; 1694 mlog_errno(status);
1500 goto bail; 1695 goto bail;
1501 } 1696 }
1502 BUG_ON((count + start_bit) > le16_to_cpu(group->bg_bits)); 1697 BUG_ON((count + start_bit) > le16_to_cpu(group->bg_bits));
diff --git a/fs/ocfs2/suballoc.h b/fs/ocfs2/suballoc.h
index a76c82a7ceac..c787838d1052 100644
--- a/fs/ocfs2/suballoc.h
+++ b/fs/ocfs2/suballoc.h
@@ -49,6 +49,8 @@ struct ocfs2_alloc_context {
49 u16 ac_chain; 49 u16 ac_chain;
50 int ac_allow_chain_relink; 50 int ac_allow_chain_relink;
51 group_search_t *ac_group_search; 51 group_search_t *ac_group_search;
52
53 u64 ac_last_group;
52}; 54};
53 55
54void ocfs2_free_alloc_context(struct ocfs2_alloc_context *ac); 56void ocfs2_free_alloc_context(struct ocfs2_alloc_context *ac);
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 382706a67ffd..d17e33e66a1e 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -1442,8 +1442,13 @@ static int ocfs2_initialize_super(struct super_block *sb,
1442 1442
1443 osb->bitmap_blkno = OCFS2_I(inode)->ip_blkno; 1443 osb->bitmap_blkno = OCFS2_I(inode)->ip_blkno;
1444 1444
1445 /* We don't have a cluster lock on the bitmap here because
1446 * we're only interested in static information and the extra
1447 * complexity at mount time isn't worht it. Don't pass the
1448 * inode in to the read function though as we don't want it to
1449 * be put in the cache. */
1445 status = ocfs2_read_block(osb, osb->bitmap_blkno, &bitmap_bh, 0, 1450 status = ocfs2_read_block(osb, osb->bitmap_blkno, &bitmap_bh, 0,
1446 inode); 1451 NULL);
1447 iput(inode); 1452 iput(inode);
1448 if (status < 0) { 1453 if (status < 0) {
1449 mlog_errno(status); 1454 mlog_errno(status);
@@ -1452,7 +1457,6 @@ static int ocfs2_initialize_super(struct super_block *sb,
1452 1457
1453 di = (struct ocfs2_dinode *) bitmap_bh->b_data; 1458 di = (struct ocfs2_dinode *) bitmap_bh->b_data;
1454 osb->bitmap_cpg = le16_to_cpu(di->id2.i_chain.cl_cpg); 1459 osb->bitmap_cpg = le16_to_cpu(di->id2.i_chain.cl_cpg);
1455 osb->num_clusters = le32_to_cpu(di->id1.bitmap1.i_total);
1456 brelse(bitmap_bh); 1460 brelse(bitmap_bh);
1457 mlog(0, "cluster bitmap inode: %llu, clusters per group: %u\n", 1461 mlog(0, "cluster bitmap inode: %llu, clusters per group: %u\n",
1458 (unsigned long long)osb->bitmap_blkno, osb->bitmap_cpg); 1462 (unsigned long long)osb->bitmap_blkno, osb->bitmap_cpg);
diff --git a/fs/partitions/Kconfig b/fs/partitions/Kconfig
index c9a478099281..e478f1941831 100644
--- a/fs/partitions/Kconfig
+++ b/fs/partitions/Kconfig
@@ -99,7 +99,7 @@ config IBM_PARTITION
99 99
100config MAC_PARTITION 100config MAC_PARTITION
101 bool "Macintosh partition map support" if PARTITION_ADVANCED 101 bool "Macintosh partition map support" if PARTITION_ADVANCED
102 default y if MAC 102 default y if (MAC || PPC_PMAC)
103 help 103 help
104 Say Y here if you would like to use hard disks under Linux which 104 Say Y here if you would like to use hard disks under Linux which
105 were partitioned on a Macintosh. 105 were partitioned on a Macintosh.
diff --git a/fs/partitions/sun.c b/fs/partitions/sun.c
index abe91ca03edf..0a5927c806ca 100644
--- a/fs/partitions/sun.c
+++ b/fs/partitions/sun.c
@@ -74,7 +74,7 @@ int sun_partition(struct parsed_partitions *state, struct block_device *bdev)
74 spc = be16_to_cpu(label->ntrks) * be16_to_cpu(label->nsect); 74 spc = be16_to_cpu(label->ntrks) * be16_to_cpu(label->nsect);
75 for (i = 0; i < 8; i++, p++) { 75 for (i = 0; i < 8; i++, p++) {
76 unsigned long st_sector; 76 unsigned long st_sector;
77 int num_sectors; 77 unsigned int num_sectors;
78 78
79 st_sector = be32_to_cpu(p->start_cylinder) * spc; 79 st_sector = be32_to_cpu(p->start_cylinder) * spc;
80 num_sectors = be32_to_cpu(p->num_sectors); 80 num_sectors = be32_to_cpu(p->num_sectors);
diff --git a/fs/proc/proc_misc.c b/fs/proc/proc_misc.c
index 9f2cfc30f9cf..942156225447 100644
--- a/fs/proc/proc_misc.c
+++ b/fs/proc/proc_misc.c
@@ -169,7 +169,7 @@ static int meminfo_read_proc(char *page, char **start, off_t off,
169 "Mapped: %8lu kB\n" 169 "Mapped: %8lu kB\n"
170 "Slab: %8lu kB\n" 170 "Slab: %8lu kB\n"
171 "PageTables: %8lu kB\n" 171 "PageTables: %8lu kB\n"
172 "NFS Unstable: %8lu kB\n" 172 "NFS_Unstable: %8lu kB\n"
173 "Bounce: %8lu kB\n" 173 "Bounce: %8lu kB\n"
174 "CommitLimit: %8lu kB\n" 174 "CommitLimit: %8lu kB\n"
175 "Committed_AS: %8lu kB\n" 175 "Committed_AS: %8lu kB\n"
diff --git a/fs/reiserfs/file.c b/fs/reiserfs/file.c
index f318b58510fd..1627edd50810 100644
--- a/fs/reiserfs/file.c
+++ b/fs/reiserfs/file.c
@@ -48,8 +48,8 @@ static int reiserfs_file_release(struct inode *inode, struct file *filp)
48 return 0; 48 return 0;
49 } 49 }
50 50
51 reiserfs_write_lock(inode->i_sb);
52 mutex_lock(&inode->i_mutex); 51 mutex_lock(&inode->i_mutex);
52 reiserfs_write_lock(inode->i_sb);
53 /* freeing preallocation only involves relogging blocks that 53 /* freeing preallocation only involves relogging blocks that
54 * are already in the current transaction. preallocation gets 54 * are already in the current transaction. preallocation gets
55 * freed at the end of each transaction, so it is impossible for 55 * freed at the end of each transaction, so it is impossible for
diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c
index 12dfdcfbee3d..52f1e2136546 100644
--- a/fs/reiserfs/inode.c
+++ b/fs/reiserfs/inode.c
@@ -39,14 +39,10 @@ void reiserfs_delete_inode(struct inode *inode)
39 39
40 /* The = 0 happens when we abort creating a new inode for some reason like lack of space.. */ 40 /* The = 0 happens when we abort creating a new inode for some reason like lack of space.. */
41 if (!(inode->i_state & I_NEW) && INODE_PKEY(inode)->k_objectid != 0) { /* also handles bad_inode case */ 41 if (!(inode->i_state & I_NEW) && INODE_PKEY(inode)->k_objectid != 0) { /* also handles bad_inode case */
42 mutex_lock(&inode->i_mutex);
43
44 reiserfs_delete_xattrs(inode); 42 reiserfs_delete_xattrs(inode);
45 43
46 if (journal_begin(&th, inode->i_sb, jbegin_count)) { 44 if (journal_begin(&th, inode->i_sb, jbegin_count))
47 mutex_unlock(&inode->i_mutex);
48 goto out; 45 goto out;
49 }
50 reiserfs_update_inode_transaction(inode); 46 reiserfs_update_inode_transaction(inode);
51 47
52 err = reiserfs_delete_object(&th, inode); 48 err = reiserfs_delete_object(&th, inode);
@@ -57,12 +53,8 @@ void reiserfs_delete_inode(struct inode *inode)
57 if (!err) 53 if (!err)
58 DQUOT_FREE_INODE(inode); 54 DQUOT_FREE_INODE(inode);
59 55
60 if (journal_end(&th, inode->i_sb, jbegin_count)) { 56 if (journal_end(&th, inode->i_sb, jbegin_count))
61 mutex_unlock(&inode->i_mutex);
62 goto out; 57 goto out;
63 }
64
65 mutex_unlock(&inode->i_mutex);
66 58
67 /* check return value from reiserfs_delete_object after 59 /* check return value from reiserfs_delete_object after
68 * ending the transaction 60 * ending the transaction
@@ -2348,6 +2340,7 @@ static int reiserfs_write_full_page(struct page *page,
2348 unsigned long end_index = inode->i_size >> PAGE_CACHE_SHIFT; 2340 unsigned long end_index = inode->i_size >> PAGE_CACHE_SHIFT;
2349 int error = 0; 2341 int error = 0;
2350 unsigned long block; 2342 unsigned long block;
2343 sector_t last_block;
2351 struct buffer_head *head, *bh; 2344 struct buffer_head *head, *bh;
2352 int partial = 0; 2345 int partial = 0;
2353 int nr = 0; 2346 int nr = 0;
@@ -2395,10 +2388,19 @@ static int reiserfs_write_full_page(struct page *page,
2395 } 2388 }
2396 bh = head; 2389 bh = head;
2397 block = page->index << (PAGE_CACHE_SHIFT - s->s_blocksize_bits); 2390 block = page->index << (PAGE_CACHE_SHIFT - s->s_blocksize_bits);
2391 last_block = (i_size_read(inode) - 1) >> inode->i_blkbits;
2398 /* first map all the buffers, logging any direct items we find */ 2392 /* first map all the buffers, logging any direct items we find */
2399 do { 2393 do {
2400 if ((checked || buffer_dirty(bh)) && (!buffer_mapped(bh) || 2394 if (block > last_block) {
2401 (buffer_mapped(bh) 2395 /*
2396 * This can happen when the block size is less than
2397 * the page size. The corresponding bytes in the page
2398 * were zero filled above
2399 */
2400 clear_buffer_dirty(bh);
2401 set_buffer_uptodate(bh);
2402 } else if ((checked || buffer_dirty(bh)) &&
2403 (!buffer_mapped(bh) || (buffer_mapped(bh)
2402 && bh->b_blocknr == 2404 && bh->b_blocknr ==
2403 0))) { 2405 0))) {
2404 /* not mapped yet, or it points to a direct item, search 2406 /* not mapped yet, or it points to a direct item, search
diff --git a/fs/reiserfs/ioctl.c b/fs/reiserfs/ioctl.c
index 745c88100895..a986b5e1e288 100644
--- a/fs/reiserfs/ioctl.c
+++ b/fs/reiserfs/ioctl.c
@@ -116,12 +116,12 @@ static int reiserfs_unpack(struct inode *inode, struct file *filp)
116 if (REISERFS_I(inode)->i_flags & i_nopack_mask) { 116 if (REISERFS_I(inode)->i_flags & i_nopack_mask) {
117 return 0; 117 return 0;
118 } 118 }
119 reiserfs_write_lock(inode->i_sb);
120 119
121 /* we need to make sure nobody is changing the file size beneath 120 /* we need to make sure nobody is changing the file size beneath
122 ** us 121 ** us
123 */ 122 */
124 mutex_lock(&inode->i_mutex); 123 mutex_lock(&inode->i_mutex);
124 reiserfs_write_lock(inode->i_sb);
125 125
126 write_from = inode->i_size & (blocksize - 1); 126 write_from = inode->i_size & (blocksize - 1);
127 /* if we are on a block boundary, we are already unpacked. */ 127 /* if we are on a block boundary, we are already unpacked. */
diff --git a/fs/reiserfs/xattr.c b/fs/reiserfs/xattr.c
index 39fedaa88a0c..d935fb9394e3 100644
--- a/fs/reiserfs/xattr.c
+++ b/fs/reiserfs/xattr.c
@@ -424,7 +424,7 @@ int xattr_readdir(struct file *file, filldir_t filler, void *buf)
424 int res = -ENOTDIR; 424 int res = -ENOTDIR;
425 if (!file->f_op || !file->f_op->readdir) 425 if (!file->f_op || !file->f_op->readdir)
426 goto out; 426 goto out;
427 mutex_lock(&inode->i_mutex); 427 mutex_lock_nested(&inode->i_mutex, I_MUTEX_XATTR);
428// down(&inode->i_zombie); 428// down(&inode->i_zombie);
429 res = -ENOENT; 429 res = -ENOENT;
430 if (!IS_DEADDIR(inode)) { 430 if (!IS_DEADDIR(inode)) {
diff --git a/fs/udf/ialloc.c b/fs/udf/ialloc.c
index 3873c672cb4c..33323473e3c4 100644
--- a/fs/udf/ialloc.c
+++ b/fs/udf/ialloc.c
@@ -75,6 +75,12 @@ struct inode * udf_new_inode (struct inode *dir, int mode, int * err)
75 } 75 }
76 *err = -ENOSPC; 76 *err = -ENOSPC;
77 77
78 UDF_I_UNIQUE(inode) = 0;
79 UDF_I_LENEXTENTS(inode) = 0;
80 UDF_I_NEXT_ALLOC_BLOCK(inode) = 0;
81 UDF_I_NEXT_ALLOC_GOAL(inode) = 0;
82 UDF_I_STRAT4096(inode) = 0;
83
78 block = udf_new_block(dir->i_sb, NULL, UDF_I_LOCATION(dir).partitionReferenceNum, 84 block = udf_new_block(dir->i_sb, NULL, UDF_I_LOCATION(dir).partitionReferenceNum,
79 start, err); 85 start, err);
80 if (*err) 86 if (*err)
@@ -84,11 +90,6 @@ struct inode * udf_new_inode (struct inode *dir, int mode, int * err)
84 } 90 }
85 91
86 mutex_lock(&sbi->s_alloc_mutex); 92 mutex_lock(&sbi->s_alloc_mutex);
87 UDF_I_UNIQUE(inode) = 0;
88 UDF_I_LENEXTENTS(inode) = 0;
89 UDF_I_NEXT_ALLOC_BLOCK(inode) = 0;
90 UDF_I_NEXT_ALLOC_GOAL(inode) = 0;
91 UDF_I_STRAT4096(inode) = 0;
92 if (UDF_SB_LVIDBH(sb)) 93 if (UDF_SB_LVIDBH(sb))
93 { 94 {
94 struct logicalVolHeaderDesc *lvhd; 95 struct logicalVolHeaderDesc *lvhd;
diff --git a/fs/udf/super.c b/fs/udf/super.c
index 4df822c881b6..fcce1a21a51b 100644
--- a/fs/udf/super.c
+++ b/fs/udf/super.c
@@ -115,6 +115,13 @@ static struct inode *udf_alloc_inode(struct super_block *sb)
115 ei = (struct udf_inode_info *)kmem_cache_alloc(udf_inode_cachep, SLAB_KERNEL); 115 ei = (struct udf_inode_info *)kmem_cache_alloc(udf_inode_cachep, SLAB_KERNEL);
116 if (!ei) 116 if (!ei)
117 return NULL; 117 return NULL;
118
119 ei->i_unique = 0;
120 ei->i_lenExtents = 0;
121 ei->i_next_alloc_block = 0;
122 ei->i_next_alloc_goal = 0;
123 ei->i_strat4096 = 0;
124
118 return &ei->vfs_inode; 125 return &ei->vfs_inode;
119} 126}
120 127
@@ -1652,7 +1659,7 @@ static int udf_fill_super(struct super_block *sb, void *options, int silent)
1652 iput(inode); 1659 iput(inode);
1653 goto error_out; 1660 goto error_out;
1654 } 1661 }
1655 sb->s_maxbytes = MAX_LFS_FILESIZE; 1662 sb->s_maxbytes = 1<<30;
1656 return 0; 1663 return 0;
1657 1664
1658error_out: 1665error_out:
diff --git a/fs/udf/truncate.c b/fs/udf/truncate.c
index e1b0e8cfecb4..0abd66ce36ea 100644
--- a/fs/udf/truncate.c
+++ b/fs/udf/truncate.c
@@ -239,37 +239,51 @@ void udf_truncate_extents(struct inode * inode)
239 { 239 {
240 if (offset) 240 if (offset)
241 { 241 {
242 extoffset -= adsize; 242 /*
243 etype = udf_next_aext(inode, &bloc, &extoffset, &eloc, &elen, &bh, 1); 243 * OK, there is not extent covering inode->i_size and
244 if (etype == (EXT_NOT_RECORDED_NOT_ALLOCATED >> 30)) 244 * no extent above inode->i_size => truncate is
245 { 245 * extending the file by 'offset'.
246 extoffset -= adsize; 246 */
247 elen = EXT_NOT_RECORDED_NOT_ALLOCATED | (elen + offset); 247 if ((!bh && extoffset == udf_file_entry_alloc_offset(inode)) ||
248 udf_write_aext(inode, bloc, &extoffset, eloc, elen, bh, 0); 248 (bh && extoffset == sizeof(struct allocExtDesc))) {
249 /* File has no extents at all! */
250 memset(&eloc, 0x00, sizeof(kernel_lb_addr));
251 elen = EXT_NOT_RECORDED_NOT_ALLOCATED | offset;
252 udf_add_aext(inode, &bloc, &extoffset, eloc, elen, &bh, 1);
249 } 253 }
250 else if (etype == (EXT_NOT_RECORDED_ALLOCATED >> 30)) 254 else {
251 {
252 kernel_lb_addr neloc = { 0, 0 };
253 extoffset -= adsize; 255 extoffset -= adsize;
254 nelen = EXT_NOT_RECORDED_NOT_ALLOCATED | 256 etype = udf_next_aext(inode, &bloc, &extoffset, &eloc, &elen, &bh, 1);
255 ((elen + offset + inode->i_sb->s_blocksize - 1) & 257 if (etype == (EXT_NOT_RECORDED_NOT_ALLOCATED >> 30))
256 ~(inode->i_sb->s_blocksize - 1)); 258 {
257 udf_write_aext(inode, bloc, &extoffset, neloc, nelen, bh, 1); 259 extoffset -= adsize;
258 udf_add_aext(inode, &bloc, &extoffset, eloc, (etype << 30) | elen, &bh, 1); 260 elen = EXT_NOT_RECORDED_NOT_ALLOCATED | (elen + offset);
259 } 261 udf_write_aext(inode, bloc, &extoffset, eloc, elen, bh, 0);
260 else 262 }
261 { 263 else if (etype == (EXT_NOT_RECORDED_ALLOCATED >> 30))
262 if (elen & (inode->i_sb->s_blocksize - 1))
263 { 264 {
265 kernel_lb_addr neloc = { 0, 0 };
264 extoffset -= adsize; 266 extoffset -= adsize;
265 elen = EXT_RECORDED_ALLOCATED | 267 nelen = EXT_NOT_RECORDED_NOT_ALLOCATED |
266 ((elen + inode->i_sb->s_blocksize - 1) & 268 ((elen + offset + inode->i_sb->s_blocksize - 1) &
267 ~(inode->i_sb->s_blocksize - 1)); 269 ~(inode->i_sb->s_blocksize - 1));
268 udf_write_aext(inode, bloc, &extoffset, eloc, elen, bh, 1); 270 udf_write_aext(inode, bloc, &extoffset, neloc, nelen, bh, 1);
271 udf_add_aext(inode, &bloc, &extoffset, eloc, (etype << 30) | elen, &bh, 1);
272 }
273 else
274 {
275 if (elen & (inode->i_sb->s_blocksize - 1))
276 {
277 extoffset -= adsize;
278 elen = EXT_RECORDED_ALLOCATED |
279 ((elen + inode->i_sb->s_blocksize - 1) &
280 ~(inode->i_sb->s_blocksize - 1));
281 udf_write_aext(inode, bloc, &extoffset, eloc, elen, bh, 1);
282 }
283 memset(&eloc, 0x00, sizeof(kernel_lb_addr));
284 elen = EXT_NOT_RECORDED_NOT_ALLOCATED | offset;
285 udf_add_aext(inode, &bloc, &extoffset, eloc, elen, &bh, 1);
269 } 286 }
270 memset(&eloc, 0x00, sizeof(kernel_lb_addr));
271 elen = EXT_NOT_RECORDED_NOT_ALLOCATED | offset;
272 udf_add_aext(inode, &bloc, &extoffset, eloc, elen, &bh, 1);
273 } 287 }
274 } 288 }
275 } 289 }
diff --git a/fs/ufs/balloc.c b/fs/ufs/balloc.c
index b01804baa120..b82381475779 100644
--- a/fs/ufs/balloc.c
+++ b/fs/ufs/balloc.c
@@ -248,7 +248,7 @@ static void ufs_change_blocknr(struct inode *inode, unsigned int baseblk,
248 248
249 if (likely(cur_index != index)) { 249 if (likely(cur_index != index)) {
250 page = ufs_get_locked_page(mapping, index); 250 page = ufs_get_locked_page(mapping, index);
251 if (IS_ERR(page)) 251 if (!page || IS_ERR(page)) /* it was truncated or EIO */
252 continue; 252 continue;
253 } else 253 } else
254 page = locked_page; 254 page = locked_page;
diff --git a/fs/ufs/inode.c b/fs/ufs/inode.c
index e7c8615beb65..30c6e8a9446c 100644
--- a/fs/ufs/inode.c
+++ b/fs/ufs/inode.c
@@ -169,18 +169,20 @@ static void ufs_clear_frag(struct inode *inode, struct buffer_head *bh)
169 169
170static struct buffer_head * 170static struct buffer_head *
171ufs_clear_frags(struct inode *inode, sector_t beg, 171ufs_clear_frags(struct inode *inode, sector_t beg,
172 unsigned int n) 172 unsigned int n, sector_t want)
173{ 173{
174 struct buffer_head *res, *bh; 174 struct buffer_head *res = NULL, *bh;
175 sector_t end = beg + n; 175 sector_t end = beg + n;
176 176
177 res = sb_getblk(inode->i_sb, beg); 177 for (; beg < end; ++beg) {
178 ufs_clear_frag(inode, res);
179 for (++beg; beg < end; ++beg) {
180 bh = sb_getblk(inode->i_sb, beg); 178 bh = sb_getblk(inode->i_sb, beg);
181 ufs_clear_frag(inode, bh); 179 ufs_clear_frag(inode, bh);
182 brelse(bh); 180 if (want != beg)
181 brelse(bh);
182 else
183 res = bh;
183 } 184 }
185 BUG_ON(!res);
184 return res; 186 return res;
185} 187}
186 188
@@ -265,7 +267,9 @@ repeat:
265 lastfrag = ufsi->i_lastfrag; 267 lastfrag = ufsi->i_lastfrag;
266 268
267 } 269 }
268 goal = fs32_to_cpu(sb, ufsi->i_u1.i_data[lastblock]) + uspi->s_fpb; 270 tmp = fs32_to_cpu(sb, ufsi->i_u1.i_data[lastblock]);
271 if (tmp)
272 goal = tmp + uspi->s_fpb;
269 tmp = ufs_new_fragments (inode, p, fragment - blockoff, 273 tmp = ufs_new_fragments (inode, p, fragment - blockoff,
270 goal, required + blockoff, 274 goal, required + blockoff,
271 err, locked_page); 275 err, locked_page);
@@ -277,13 +281,15 @@ repeat:
277 tmp = ufs_new_fragments(inode, p, fragment - (blockoff - lastblockoff), 281 tmp = ufs_new_fragments(inode, p, fragment - (blockoff - lastblockoff),
278 fs32_to_cpu(sb, *p), required + (blockoff - lastblockoff), 282 fs32_to_cpu(sb, *p), required + (blockoff - lastblockoff),
279 err, locked_page); 283 err, locked_page);
280 } 284 } else /* (lastblock > block) */ {
281 /* 285 /*
282 * We will allocate new block before last allocated block 286 * We will allocate new block before last allocated block
283 */ 287 */
284 else /* (lastblock > block) */ { 288 if (block) {
285 if (lastblock && (tmp = fs32_to_cpu(sb, ufsi->i_u1.i_data[lastblock-1]))) 289 tmp = fs32_to_cpu(sb, ufsi->i_u1.i_data[block-1]);
286 goal = tmp + uspi->s_fpb; 290 if (tmp)
291 goal = tmp + uspi->s_fpb;
292 }
287 tmp = ufs_new_fragments(inode, p, fragment - blockoff, 293 tmp = ufs_new_fragments(inode, p, fragment - blockoff,
288 goal, uspi->s_fpb, err, locked_page); 294 goal, uspi->s_fpb, err, locked_page);
289 } 295 }
@@ -296,7 +302,7 @@ repeat:
296 } 302 }
297 303
298 if (!phys) { 304 if (!phys) {
299 result = ufs_clear_frags(inode, tmp + blockoff, required); 305 result = ufs_clear_frags(inode, tmp, required, tmp + blockoff);
300 } else { 306 } else {
301 *phys = tmp + blockoff; 307 *phys = tmp + blockoff;
302 result = NULL; 308 result = NULL;
@@ -383,7 +389,7 @@ repeat:
383 } 389 }
384 } 390 }
385 391
386 if (block && (tmp = fs32_to_cpu(sb, ((__fs32*)bh->b_data)[block-1]) + uspi->s_fpb)) 392 if (block && (tmp = fs32_to_cpu(sb, ((__fs32*)bh->b_data)[block-1])))
387 goal = tmp + uspi->s_fpb; 393 goal = tmp + uspi->s_fpb;
388 else 394 else
389 goal = bh->b_blocknr + uspi->s_fpb; 395 goal = bh->b_blocknr + uspi->s_fpb;
@@ -397,7 +403,8 @@ repeat:
397 403
398 404
399 if (!phys) { 405 if (!phys) {
400 result = ufs_clear_frags(inode, tmp + blockoff, uspi->s_fpb); 406 result = ufs_clear_frags(inode, tmp, uspi->s_fpb,
407 tmp + blockoff);
401 } else { 408 } else {
402 *phys = tmp + blockoff; 409 *phys = tmp + blockoff;
403 *new = 1; 410 *new = 1;
diff --git a/fs/ufs/namei.c b/fs/ufs/namei.c
index abd5f23a426d..d344b411e261 100644
--- a/fs/ufs/namei.c
+++ b/fs/ufs/namei.c
@@ -129,7 +129,7 @@ static int ufs_symlink (struct inode * dir, struct dentry * dentry,
129 struct inode * inode; 129 struct inode * inode;
130 130
131 if (l > sb->s_blocksize) 131 if (l > sb->s_blocksize)
132 goto out; 132 goto out_notlocked;
133 133
134 lock_kernel(); 134 lock_kernel();
135 inode = ufs_new_inode(dir, S_IFLNK | S_IRWXUGO); 135 inode = ufs_new_inode(dir, S_IFLNK | S_IRWXUGO);
@@ -155,6 +155,7 @@ static int ufs_symlink (struct inode * dir, struct dentry * dentry,
155 err = ufs_add_nondir(dentry, inode); 155 err = ufs_add_nondir(dentry, inode);
156out: 156out:
157 unlock_kernel(); 157 unlock_kernel();
158out_notlocked:
158 return err; 159 return err;
159 160
160out_fail: 161out_fail:
diff --git a/fs/ufs/truncate.c b/fs/ufs/truncate.c
index c9b55872079b..ea11d04c41a0 100644
--- a/fs/ufs/truncate.c
+++ b/fs/ufs/truncate.c
@@ -375,17 +375,15 @@ static int ufs_alloc_lastblock(struct inode *inode)
375 int err = 0; 375 int err = 0;
376 struct address_space *mapping = inode->i_mapping; 376 struct address_space *mapping = inode->i_mapping;
377 struct ufs_sb_private_info *uspi = UFS_SB(inode->i_sb)->s_uspi; 377 struct ufs_sb_private_info *uspi = UFS_SB(inode->i_sb)->s_uspi;
378 struct ufs_inode_info *ufsi = UFS_I(inode);
379 unsigned lastfrag, i, end; 378 unsigned lastfrag, i, end;
380 struct page *lastpage; 379 struct page *lastpage;
381 struct buffer_head *bh; 380 struct buffer_head *bh;
382 381
383 lastfrag = (i_size_read(inode) + uspi->s_fsize - 1) >> uspi->s_fshift; 382 lastfrag = (i_size_read(inode) + uspi->s_fsize - 1) >> uspi->s_fshift;
384 383
385 if (!lastfrag) { 384 if (!lastfrag)
386 ufsi->i_lastfrag = 0;
387 goto out; 385 goto out;
388 } 386
389 lastfrag--; 387 lastfrag--;
390 388
391 lastpage = ufs_get_locked_page(mapping, lastfrag >> 389 lastpage = ufs_get_locked_page(mapping, lastfrag >>
@@ -400,25 +398,25 @@ static int ufs_alloc_lastblock(struct inode *inode)
400 for (i = 0; i < end; ++i) 398 for (i = 0; i < end; ++i)
401 bh = bh->b_this_page; 399 bh = bh->b_this_page;
402 400
403 if (!buffer_mapped(bh)) { 401
404 err = ufs_getfrag_block(inode, lastfrag, bh, 1); 402 err = ufs_getfrag_block(inode, lastfrag, bh, 1);
405 403
406 if (unlikely(err)) 404 if (unlikely(err))
407 goto out_unlock; 405 goto out_unlock;
408 406
409 if (buffer_new(bh)) { 407 if (buffer_new(bh)) {
410 clear_buffer_new(bh); 408 clear_buffer_new(bh);
411 unmap_underlying_metadata(bh->b_bdev, 409 unmap_underlying_metadata(bh->b_bdev,
412 bh->b_blocknr); 410 bh->b_blocknr);
413 /* 411 /*
414 * we do not zeroize fragment, because of 412 * we do not zeroize fragment, because of
415 * if it maped to hole, it already contains zeroes 413 * if it maped to hole, it already contains zeroes
416 */ 414 */
417 set_buffer_uptodate(bh); 415 set_buffer_uptodate(bh);
418 mark_buffer_dirty(bh); 416 mark_buffer_dirty(bh);
419 set_page_dirty(lastpage); 417 set_page_dirty(lastpage);
420 }
421 } 418 }
419
422out_unlock: 420out_unlock:
423 ufs_put_locked_page(lastpage); 421 ufs_put_locked_page(lastpage);
424out: 422out:
@@ -440,23 +438,11 @@ int ufs_truncate(struct inode *inode, loff_t old_i_size)
440 if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) 438 if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
441 return -EPERM; 439 return -EPERM;
442 440
443 if (inode->i_size > old_i_size) { 441 err = ufs_alloc_lastblock(inode);
444 /*
445 * if we expand file we should care about
446 * allocation of block for last byte first of all
447 */
448 err = ufs_alloc_lastblock(inode);
449 442
450 if (err) { 443 if (err) {
451 i_size_write(inode, old_i_size); 444 i_size_write(inode, old_i_size);
452 goto out; 445 goto out;
453 }
454 /*
455 * go away, because of we expand file, and we do not
456 * need free blocks, and zeroizes page
457 */
458 lock_kernel();
459 goto almost_end;
460 } 446 }
461 447
462 block_truncate_page(inode->i_mapping, inode->i_size, ufs_getfrag_block); 448 block_truncate_page(inode->i_mapping, inode->i_size, ufs_getfrag_block);
@@ -477,21 +463,8 @@ int ufs_truncate(struct inode *inode, loff_t old_i_size)
477 yield(); 463 yield();
478 } 464 }
479 465
480 if (inode->i_size < old_i_size) {
481 /*
482 * now we should have enough space
483 * to allocate block for last byte
484 */
485 err = ufs_alloc_lastblock(inode);
486 if (err)
487 /*
488 * looks like all the same - we have no space,
489 * but we truncate file already
490 */
491 inode->i_size = (ufsi->i_lastfrag - 1) * uspi->s_fsize;
492 }
493almost_end:
494 inode->i_mtime = inode->i_ctime = CURRENT_TIME_SEC; 466 inode->i_mtime = inode->i_ctime = CURRENT_TIME_SEC;
467 ufsi->i_lastfrag = DIRECT_FRAGMENT;
495 unlock_kernel(); 468 unlock_kernel();
496 mark_inode_dirty(inode); 469 mark_inode_dirty(inode);
497out: 470out:
diff --git a/fs/ufs/util.c b/fs/ufs/util.c
index 337cf2c46d10..22f820a9b15c 100644
--- a/fs/ufs/util.c
+++ b/fs/ufs/util.c
@@ -251,12 +251,12 @@ struct page *ufs_get_locked_page(struct address_space *mapping,
251{ 251{
252 struct page *page; 252 struct page *page;
253 253
254try_again:
255 page = find_lock_page(mapping, index); 254 page = find_lock_page(mapping, index);
256 if (!page) { 255 if (!page) {
257 page = read_cache_page(mapping, index, 256 page = read_cache_page(mapping, index,
258 (filler_t*)mapping->a_ops->readpage, 257 (filler_t*)mapping->a_ops->readpage,
259 NULL); 258 NULL);
259
260 if (IS_ERR(page)) { 260 if (IS_ERR(page)) {
261 printk(KERN_ERR "ufs_change_blocknr: " 261 printk(KERN_ERR "ufs_change_blocknr: "
262 "read_cache_page error: ino %lu, index: %lu\n", 262 "read_cache_page error: ino %lu, index: %lu\n",
@@ -266,6 +266,14 @@ try_again:
266 266
267 lock_page(page); 267 lock_page(page);
268 268
269 if (unlikely(page->mapping == NULL)) {
270 /* Truncate got there first */
271 unlock_page(page);
272 page_cache_release(page);
273 page = NULL;
274 goto out;
275 }
276
269 if (!PageUptodate(page) || PageError(page)) { 277 if (!PageUptodate(page) || PageError(page)) {
270 unlock_page(page); 278 unlock_page(page);
271 page_cache_release(page); 279 page_cache_release(page);
@@ -275,15 +283,8 @@ try_again:
275 mapping->host->i_ino, index); 283 mapping->host->i_ino, index);
276 284
277 page = ERR_PTR(-EIO); 285 page = ERR_PTR(-EIO);
278 goto out;
279 } 286 }
280 } 287 }
281
282 if (unlikely(!page->mapping || !page_has_buffers(page))) {
283 unlock_page(page);
284 page_cache_release(page);
285 goto try_again;/*we really need these buffers*/
286 }
287out: 288out:
288 return page; 289 return page;
289} 290}
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
index ceda3a2859d2..7858703ed84c 100644
--- a/fs/xfs/linux-2.6/xfs_buf.h
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -246,8 +246,8 @@ extern void xfs_buf_trace(xfs_buf_t *, char *, void *, void *);
246#define BUF_BUSY XBF_DONT_BLOCK 246#define BUF_BUSY XBF_DONT_BLOCK
247 247
248#define XFS_BUF_BFLAGS(bp) ((bp)->b_flags) 248#define XFS_BUF_BFLAGS(bp) ((bp)->b_flags)
249#define XFS_BUF_ZEROFLAGS(bp) \ 249#define XFS_BUF_ZEROFLAGS(bp) ((bp)->b_flags &= \
250 ((bp)->b_flags &= ~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI)) 250 ~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI|XBF_ORDERED))
251 251
252#define XFS_BUF_STALE(bp) ((bp)->b_flags |= XFS_B_STALE) 252#define XFS_BUF_STALE(bp) ((bp)->b_flags |= XFS_B_STALE)
253#define XFS_BUF_UNSTALE(bp) ((bp)->b_flags &= ~XFS_B_STALE) 253#define XFS_BUF_UNSTALE(bp) ((bp)->b_flags &= ~XFS_B_STALE)
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 9bdef9d51900..4754f342a5d3 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -314,6 +314,13 @@ xfs_mountfs_check_barriers(xfs_mount_t *mp)
314 return; 314 return;
315 } 315 }
316 316
317 if (xfs_readonly_buftarg(mp->m_ddev_targp)) {
318 xfs_fs_cmn_err(CE_NOTE, mp,
319 "Disabling barriers, underlying device is readonly");
320 mp->m_flags &= ~XFS_MOUNT_BARRIER;
321 return;
322 }
323
317 error = xfs_barrier_test(mp); 324 error = xfs_barrier_test(mp);
318 if (error) { 325 if (error) {
319 xfs_fs_cmn_err(CE_NOTE, mp, 326 xfs_fs_cmn_err(CE_NOTE, mp,
diff --git a/fs/xfs/quota/xfs_qm_bhv.c b/fs/xfs/quota/xfs_qm_bhv.c
index e95e99f7168f..f137856c3261 100644
--- a/fs/xfs/quota/xfs_qm_bhv.c
+++ b/fs/xfs/quota/xfs_qm_bhv.c
@@ -217,17 +217,24 @@ xfs_qm_statvfs(
217 return 0; 217 return 0;
218 dp = &dqp->q_core; 218 dp = &dqp->q_core;
219 219
220 limit = dp->d_blk_softlimit ? dp->d_blk_softlimit : dp->d_blk_hardlimit; 220 limit = dp->d_blk_softlimit ?
221 be64_to_cpu(dp->d_blk_softlimit) :
222 be64_to_cpu(dp->d_blk_hardlimit);
221 if (limit && statp->f_blocks > limit) { 223 if (limit && statp->f_blocks > limit) {
222 statp->f_blocks = limit; 224 statp->f_blocks = limit;
223 statp->f_bfree = (statp->f_blocks > dp->d_bcount) ? 225 statp->f_bfree =
224 (statp->f_blocks - dp->d_bcount) : 0; 226 (statp->f_blocks > be64_to_cpu(dp->d_bcount)) ?
227 (statp->f_blocks - be64_to_cpu(dp->d_bcount)) : 0;
225 } 228 }
226 limit = dp->d_ino_softlimit ? dp->d_ino_softlimit : dp->d_ino_hardlimit; 229
230 limit = dp->d_ino_softlimit ?
231 be64_to_cpu(dp->d_ino_softlimit) :
232 be64_to_cpu(dp->d_ino_hardlimit);
227 if (limit && statp->f_files > limit) { 233 if (limit && statp->f_files > limit) {
228 statp->f_files = limit; 234 statp->f_files = limit;
229 statp->f_ffree = (statp->f_files > dp->d_icount) ? 235 statp->f_ffree =
230 (statp->f_ffree - dp->d_icount) : 0; 236 (statp->f_files > be64_to_cpu(dp->d_icount)) ?
237 (statp->f_ffree - be64_to_cpu(dp->d_icount)) : 0;
231 } 238 }
232 239
233 xfs_qm_dqput(dqp); 240 xfs_qm_dqput(dqp);
diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index eef6763f3a67..d2bbcd882a69 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -1835,40 +1835,47 @@ xfs_alloc_fix_freelist(
1835 &agbp))) 1835 &agbp)))
1836 return error; 1836 return error;
1837 if (!pag->pagf_init) { 1837 if (!pag->pagf_init) {
1838 ASSERT(flags & XFS_ALLOC_FLAG_TRYLOCK);
1839 ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
1838 args->agbp = NULL; 1840 args->agbp = NULL;
1839 return 0; 1841 return 0;
1840 } 1842 }
1841 } else 1843 } else
1842 agbp = NULL; 1844 agbp = NULL;
1843 1845
1844 /* If this is a metadata preferred pag and we are user data 1846 /*
1847 * If this is a metadata preferred pag and we are user data
1845 * then try somewhere else if we are not being asked to 1848 * then try somewhere else if we are not being asked to
1846 * try harder at this point 1849 * try harder at this point
1847 */ 1850 */
1848 if (pag->pagf_metadata && args->userdata && flags) { 1851 if (pag->pagf_metadata && args->userdata &&
1852 (flags & XFS_ALLOC_FLAG_TRYLOCK)) {
1853 ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
1849 args->agbp = NULL; 1854 args->agbp = NULL;
1850 return 0; 1855 return 0;
1851 } 1856 }
1852 1857
1853 need = XFS_MIN_FREELIST_PAG(pag, mp); 1858 if (!(flags & XFS_ALLOC_FLAG_FREEING)) {
1854 delta = need > pag->pagf_flcount ? need - pag->pagf_flcount : 0; 1859 need = XFS_MIN_FREELIST_PAG(pag, mp);
1855 /* 1860 delta = need > pag->pagf_flcount ? need - pag->pagf_flcount : 0;
1856 * If it looks like there isn't a long enough extent, or enough 1861 /*
1857 * total blocks, reject it. 1862 * If it looks like there isn't a long enough extent, or enough
1858 */ 1863 * total blocks, reject it.
1859 longest = (pag->pagf_longest > delta) ? 1864 */
1860 (pag->pagf_longest - delta) : 1865 longest = (pag->pagf_longest > delta) ?
1861 (pag->pagf_flcount > 0 || pag->pagf_longest > 0); 1866 (pag->pagf_longest - delta) :
1862 if (args->minlen + args->alignment + args->minalignslop - 1 > longest || 1867 (pag->pagf_flcount > 0 || pag->pagf_longest > 0);
1863 (!(flags & XFS_ALLOC_FLAG_FREEING) && 1868 if ((args->minlen + args->alignment + args->minalignslop - 1) >
1864 (int)(pag->pagf_freeblks + pag->pagf_flcount - 1869 longest ||
1865 need - args->total) < 1870 ((int)(pag->pagf_freeblks + pag->pagf_flcount -
1866 (int)args->minleft)) { 1871 need - args->total) < (int)args->minleft)) {
1867 if (agbp) 1872 if (agbp)
1868 xfs_trans_brelse(tp, agbp); 1873 xfs_trans_brelse(tp, agbp);
1869 args->agbp = NULL; 1874 args->agbp = NULL;
1870 return 0; 1875 return 0;
1876 }
1871 } 1877 }
1878
1872 /* 1879 /*
1873 * Get the a.g. freespace buffer. 1880 * Get the a.g. freespace buffer.
1874 * Can fail if we're not blocking on locks, and it's held. 1881 * Can fail if we're not blocking on locks, and it's held.
@@ -1878,6 +1885,8 @@ xfs_alloc_fix_freelist(
1878 &agbp))) 1885 &agbp)))
1879 return error; 1886 return error;
1880 if (agbp == NULL) { 1887 if (agbp == NULL) {
1888 ASSERT(flags & XFS_ALLOC_FLAG_TRYLOCK);
1889 ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
1881 args->agbp = NULL; 1890 args->agbp = NULL;
1882 return 0; 1891 return 0;
1883 } 1892 }
@@ -1887,22 +1896,24 @@ xfs_alloc_fix_freelist(
1887 */ 1896 */
1888 agf = XFS_BUF_TO_AGF(agbp); 1897 agf = XFS_BUF_TO_AGF(agbp);
1889 need = XFS_MIN_FREELIST(agf, mp); 1898 need = XFS_MIN_FREELIST(agf, mp);
1890 delta = need > be32_to_cpu(agf->agf_flcount) ?
1891 (need - be32_to_cpu(agf->agf_flcount)) : 0;
1892 /* 1899 /*
1893 * If there isn't enough total or single-extent, reject it. 1900 * If there isn't enough total or single-extent, reject it.
1894 */ 1901 */
1895 longest = be32_to_cpu(agf->agf_longest); 1902 if (!(flags & XFS_ALLOC_FLAG_FREEING)) {
1896 longest = (longest > delta) ? (longest - delta) : 1903 delta = need > be32_to_cpu(agf->agf_flcount) ?
1897 (be32_to_cpu(agf->agf_flcount) > 0 || longest > 0); 1904 (need - be32_to_cpu(agf->agf_flcount)) : 0;
1898 if (args->minlen + args->alignment + args->minalignslop - 1 > longest || 1905 longest = be32_to_cpu(agf->agf_longest);
1899 (!(flags & XFS_ALLOC_FLAG_FREEING) && 1906 longest = (longest > delta) ? (longest - delta) :
1900 (int)(be32_to_cpu(agf->agf_freeblks) + 1907 (be32_to_cpu(agf->agf_flcount) > 0 || longest > 0);
1901 be32_to_cpu(agf->agf_flcount) - need - args->total) < 1908 if ((args->minlen + args->alignment + args->minalignslop - 1) >
1902 (int)args->minleft)) { 1909 longest ||
1903 xfs_trans_brelse(tp, agbp); 1910 ((int)(be32_to_cpu(agf->agf_freeblks) +
1904 args->agbp = NULL; 1911 be32_to_cpu(agf->agf_flcount) - need - args->total) <
1905 return 0; 1912 (int)args->minleft)) {
1913 xfs_trans_brelse(tp, agbp);
1914 args->agbp = NULL;
1915 return 0;
1916 }
1906 } 1917 }
1907 /* 1918 /*
1908 * Make the freelist shorter if it's too long. 1919 * Make the freelist shorter if it's too long.
@@ -1950,12 +1961,11 @@ xfs_alloc_fix_freelist(
1950 * on a completely full ag. 1961 * on a completely full ag.
1951 */ 1962 */
1952 if (targs.agbno == NULLAGBLOCK) { 1963 if (targs.agbno == NULLAGBLOCK) {
1953 if (!(flags & XFS_ALLOC_FLAG_FREEING)) { 1964 if (flags & XFS_ALLOC_FLAG_FREEING)
1954 xfs_trans_brelse(tp, agflbp); 1965 break;
1955 args->agbp = NULL; 1966 xfs_trans_brelse(tp, agflbp);
1956 return 0; 1967 args->agbp = NULL;
1957 } 1968 return 0;
1958 break;
1959 } 1969 }
1960 /* 1970 /*
1961 * Put each allocated block on the list. 1971 * Put each allocated block on the list.
@@ -2442,31 +2452,26 @@ xfs_free_extent(
2442 xfs_fsblock_t bno, /* starting block number of extent */ 2452 xfs_fsblock_t bno, /* starting block number of extent */
2443 xfs_extlen_t len) /* length of extent */ 2453 xfs_extlen_t len) /* length of extent */
2444{ 2454{
2445#ifdef DEBUG 2455 xfs_alloc_arg_t args;
2446 xfs_agf_t *agf; /* a.g. freespace header */
2447#endif
2448 xfs_alloc_arg_t args; /* allocation argument structure */
2449 int error; 2456 int error;
2450 2457
2451 ASSERT(len != 0); 2458 ASSERT(len != 0);
2459 memset(&args, 0, sizeof(xfs_alloc_arg_t));
2452 args.tp = tp; 2460 args.tp = tp;
2453 args.mp = tp->t_mountp; 2461 args.mp = tp->t_mountp;
2454 args.agno = XFS_FSB_TO_AGNO(args.mp, bno); 2462 args.agno = XFS_FSB_TO_AGNO(args.mp, bno);
2455 ASSERT(args.agno < args.mp->m_sb.sb_agcount); 2463 ASSERT(args.agno < args.mp->m_sb.sb_agcount);
2456 args.agbno = XFS_FSB_TO_AGBNO(args.mp, bno); 2464 args.agbno = XFS_FSB_TO_AGBNO(args.mp, bno);
2457 args.alignment = 1;
2458 args.minlen = args.minleft = args.minalignslop = 0;
2459 down_read(&args.mp->m_peraglock); 2465 down_read(&args.mp->m_peraglock);
2460 args.pag = &args.mp->m_perag[args.agno]; 2466 args.pag = &args.mp->m_perag[args.agno];
2461 if ((error = xfs_alloc_fix_freelist(&args, XFS_ALLOC_FLAG_FREEING))) 2467 if ((error = xfs_alloc_fix_freelist(&args, XFS_ALLOC_FLAG_FREEING)))
2462 goto error0; 2468 goto error0;
2463#ifdef DEBUG 2469#ifdef DEBUG
2464 ASSERT(args.agbp != NULL); 2470 ASSERT(args.agbp != NULL);
2465 agf = XFS_BUF_TO_AGF(args.agbp); 2471 ASSERT((args.agbno + len) <=
2466 ASSERT(args.agbno + len <= be32_to_cpu(agf->agf_length)); 2472 be32_to_cpu(XFS_BUF_TO_AGF(args.agbp)->agf_length));
2467#endif 2473#endif
2468 error = xfs_free_ag_extent(tp, args.agbp, args.agno, args.agbno, 2474 error = xfs_free_ag_extent(tp, args.agbp, args.agno, args.agbno, len, 0);
2469 len, 0);
2470error0: 2475error0:
2471 up_read(&args.mp->m_peraglock); 2476 up_read(&args.mp->m_peraglock);
2472 return error; 2477 return error;
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 86c1bf0bba9e..1f8ecff8553a 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -334,10 +334,9 @@ xfs_itobp(
334#if !defined(__KERNEL__) 334#if !defined(__KERNEL__)
335 ni = 0; 335 ni = 0;
336#elif defined(DEBUG) 336#elif defined(DEBUG)
337 ni = (imap_flags & XFS_IMAP_BULKSTAT) ? 0 : 337 ni = BBTOB(imap.im_len) >> mp->m_sb.sb_inodelog;
338 (BBTOB(imap.im_len) >> mp->m_sb.sb_inodelog);
339#else /* usual case */ 338#else /* usual case */
340 ni = (imap_flags & XFS_IMAP_BULKSTAT) ? 0 : 1; 339 ni = 1;
341#endif 340#endif
342 341
343 for (i = 0; i < ni; i++) { 342 for (i = 0; i < ni; i++) {
@@ -348,11 +347,15 @@ xfs_itobp(
348 (i << mp->m_sb.sb_inodelog)); 347 (i << mp->m_sb.sb_inodelog));
349 di_ok = INT_GET(dip->di_core.di_magic, ARCH_CONVERT) == XFS_DINODE_MAGIC && 348 di_ok = INT_GET(dip->di_core.di_magic, ARCH_CONVERT) == XFS_DINODE_MAGIC &&
350 XFS_DINODE_GOOD_VERSION(INT_GET(dip->di_core.di_version, ARCH_CONVERT)); 349 XFS_DINODE_GOOD_VERSION(INT_GET(dip->di_core.di_version, ARCH_CONVERT));
351 if (unlikely(XFS_TEST_ERROR(!di_ok, mp, XFS_ERRTAG_ITOBP_INOTOBP, 350 if (unlikely(XFS_TEST_ERROR(!di_ok, mp,
352 XFS_RANDOM_ITOBP_INOTOBP))) { 351 XFS_ERRTAG_ITOBP_INOTOBP,
352 XFS_RANDOM_ITOBP_INOTOBP))) {
353 if (imap_flags & XFS_IMAP_BULKSTAT) {
354 xfs_trans_brelse(tp, bp);
355 return XFS_ERROR(EINVAL);
356 }
353#ifdef DEBUG 357#ifdef DEBUG
354 if (!(imap_flags & XFS_IMAP_BULKSTAT)) 358 cmn_err(CE_ALERT,
355 cmn_err(CE_ALERT,
356 "Device %s - bad inode magic/vsn " 359 "Device %s - bad inode magic/vsn "
357 "daddr %lld #%d (magic=%x)", 360 "daddr %lld #%d (magic=%x)",
358 XFS_BUFTARG_NAME(mp->m_ddev_targp), 361 XFS_BUFTARG_NAME(mp->m_ddev_targp),
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index e730328636c3..21ac1a67e3e0 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -1413,7 +1413,7 @@ xlog_sync(xlog_t *log,
1413 ops = iclog->ic_header.h_num_logops; 1413 ops = iclog->ic_header.h_num_logops;
1414 INT_SET(iclog->ic_header.h_num_logops, ARCH_CONVERT, ops); 1414 INT_SET(iclog->ic_header.h_num_logops, ARCH_CONVERT, ops);
1415 1415
1416 bp = iclog->ic_bp; 1416 bp = iclog->ic_bp;
1417 ASSERT(XFS_BUF_FSPRIVATE2(bp, unsigned long) == (unsigned long)1); 1417 ASSERT(XFS_BUF_FSPRIVATE2(bp, unsigned long) == (unsigned long)1);
1418 XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)2); 1418 XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)2);
1419 XFS_BUF_SET_ADDR(bp, BLOCK_LSN(INT_GET(iclog->ic_header.h_lsn, ARCH_CONVERT))); 1419 XFS_BUF_SET_ADDR(bp, BLOCK_LSN(INT_GET(iclog->ic_header.h_lsn, ARCH_CONVERT)));
@@ -1430,15 +1430,14 @@ xlog_sync(xlog_t *log,
1430 } 1430 }
1431 XFS_BUF_SET_PTR(bp, (xfs_caddr_t) &(iclog->ic_header), count); 1431 XFS_BUF_SET_PTR(bp, (xfs_caddr_t) &(iclog->ic_header), count);
1432 XFS_BUF_SET_FSPRIVATE(bp, iclog); /* save for later */ 1432 XFS_BUF_SET_FSPRIVATE(bp, iclog); /* save for later */
1433 XFS_BUF_ZEROFLAGS(bp);
1433 XFS_BUF_BUSY(bp); 1434 XFS_BUF_BUSY(bp);
1434 XFS_BUF_ASYNC(bp); 1435 XFS_BUF_ASYNC(bp);
1435 /* 1436 /*
1436 * Do an ordered write for the log block. 1437 * Do an ordered write for the log block.
1437 * 1438 * Its unnecessary to flush the first split block in the log wrap case.
1438 * It may not be needed to flush the first split block in the log wrap
1439 * case, but do it anyways to be safe -AK
1440 */ 1439 */
1441 if (log->l_mp->m_flags & XFS_MOUNT_BARRIER) 1440 if (!split && (log->l_mp->m_flags & XFS_MOUNT_BARRIER))
1442 XFS_BUF_ORDERED(bp); 1441 XFS_BUF_ORDERED(bp);
1443 1442
1444 ASSERT(XFS_BUF_ADDR(bp) <= log->l_logBBsize-1); 1443 ASSERT(XFS_BUF_ADDR(bp) <= log->l_logBBsize-1);
@@ -1460,7 +1459,7 @@ xlog_sync(xlog_t *log,
1460 return error; 1459 return error;
1461 } 1460 }
1462 if (split) { 1461 if (split) {
1463 bp = iclog->ic_log->l_xbuf; 1462 bp = iclog->ic_log->l_xbuf;
1464 ASSERT(XFS_BUF_FSPRIVATE2(bp, unsigned long) == 1463 ASSERT(XFS_BUF_FSPRIVATE2(bp, unsigned long) ==
1465 (unsigned long)1); 1464 (unsigned long)1);
1466 XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)2); 1465 XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)2);
@@ -1468,6 +1467,7 @@ xlog_sync(xlog_t *log,
1468 XFS_BUF_SET_PTR(bp, (xfs_caddr_t)((__psint_t)&(iclog->ic_header)+ 1467 XFS_BUF_SET_PTR(bp, (xfs_caddr_t)((__psint_t)&(iclog->ic_header)+
1469 (__psint_t)count), split); 1468 (__psint_t)count), split);
1470 XFS_BUF_SET_FSPRIVATE(bp, iclog); 1469 XFS_BUF_SET_FSPRIVATE(bp, iclog);
1470 XFS_BUF_ZEROFLAGS(bp);
1471 XFS_BUF_BUSY(bp); 1471 XFS_BUF_BUSY(bp);
1472 XFS_BUF_ASYNC(bp); 1472 XFS_BUF_ASYNC(bp);
1473 if (log->l_mp->m_flags & XFS_MOUNT_BARRIER) 1473 if (log->l_mp->m_flags & XFS_MOUNT_BARRIER)
diff --git a/fs/xfs/xfs_vfsops.c b/fs/xfs/xfs_vfsops.c
index 6c96391f3f1a..b427d220a169 100644
--- a/fs/xfs/xfs_vfsops.c
+++ b/fs/xfs/xfs_vfsops.c
@@ -515,7 +515,7 @@ xfs_mount(
515 if (error) 515 if (error)
516 goto error2; 516 goto error2;
517 517
518 if ((mp->m_flags & XFS_MOUNT_BARRIER) && !(vfsp->vfs_flag & VFS_RDONLY)) 518 if (mp->m_flags & XFS_MOUNT_BARRIER)
519 xfs_mountfs_check_barriers(mp); 519 xfs_mountfs_check_barriers(mp);
520 520
521 error = XFS_IOINIT(vfsp, args, flags); 521 error = XFS_IOINIT(vfsp, args, flags);
generated by cgit v1.2.2 (git 2.25.0) at 2025-01-08 23:12:52 -0500