aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
Diffstat (limited to 'fs')
-rw-r--r--fs/btrfs/ctree.c10
-rw-r--r--fs/btrfs/disk-io.c5
-rw-r--r--fs/btrfs/extent-tree.c55
-rw-r--r--fs/btrfs/extent_io.h2
-rw-r--r--fs/btrfs/free-space-cache.c163
-rw-r--r--fs/btrfs/inode.c4
-rw-r--r--fs/btrfs/ioctl.c23
-rw-r--r--fs/btrfs/scrub.c69
-rw-r--r--fs/btrfs/super.c2
-rw-r--r--fs/btrfs/transaction.c7
-rw-r--r--fs/btrfs/volumes.c8
-rw-r--r--fs/ceph/addr.c2
-rw-r--r--fs/ceph/caps.c10
-rw-r--r--fs/ceph/dir.c11
-rw-r--r--fs/ceph/export.c4
-rw-r--r--fs/ceph/file.c35
-rw-r--r--fs/ceph/inode.c18
-rw-r--r--fs/ceph/ioctl.c6
-rw-r--r--fs/ceph/locks.c29
-rw-r--r--fs/ceph/snap.c2
-rw-r--r--fs/ceph/xattr.c6
-rw-r--r--fs/cifs/Kconfig3
-rw-r--r--fs/cifs/cifsencrypt.c2
-rw-r--r--fs/cifs/connect.c6
-rw-r--r--fs/dcookies.c3
-rw-r--r--fs/exec.c5
-rw-r--r--fs/fat/file.c2
-rw-r--r--fs/fuse/inode.c2
-rw-r--r--fs/gfs2/glock.c9
-rw-r--r--fs/jfs/jfs_logmgr.c2
-rw-r--r--fs/namei.c11
-rw-r--r--fs/nilfs2/btree.c39
-rw-r--r--fs/nilfs2/segment.c2
-rw-r--r--fs/ocfs2/super.c2
-rw-r--r--fs/super.c2
35 files changed, 362 insertions, 199 deletions
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index d84089349c82..2e667868e0d2 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -1228,6 +1228,7 @@ static void reada_for_search(struct btrfs_root *root,
1228 u32 nr; 1228 u32 nr;
1229 u32 blocksize; 1229 u32 blocksize;
1230 u32 nscan = 0; 1230 u32 nscan = 0;
1231 bool map = true;
1231 1232
1232 if (level != 1) 1233 if (level != 1)
1233 return; 1234 return;
@@ -1249,8 +1250,11 @@ static void reada_for_search(struct btrfs_root *root,
1249 1250
1250 nritems = btrfs_header_nritems(node); 1251 nritems = btrfs_header_nritems(node);
1251 nr = slot; 1252 nr = slot;
1253 if (node->map_token || path->skip_locking)
1254 map = false;
1255
1252 while (1) { 1256 while (1) {
1253 if (!node->map_token) { 1257 if (map && !node->map_token) {
1254 unsigned long offset = btrfs_node_key_ptr_offset(nr); 1258 unsigned long offset = btrfs_node_key_ptr_offset(nr);
1255 map_private_extent_buffer(node, offset, 1259 map_private_extent_buffer(node, offset,
1256 sizeof(struct btrfs_key_ptr), 1260 sizeof(struct btrfs_key_ptr),
@@ -1277,7 +1281,7 @@ static void reada_for_search(struct btrfs_root *root,
1277 if ((search <= target && target - search <= 65536) || 1281 if ((search <= target && target - search <= 65536) ||
1278 (search > target && search - target <= 65536)) { 1282 (search > target && search - target <= 65536)) {
1279 gen = btrfs_node_ptr_generation(node, nr); 1283 gen = btrfs_node_ptr_generation(node, nr);
1280 if (node->map_token) { 1284 if (map && node->map_token) {
1281 unmap_extent_buffer(node, node->map_token, 1285 unmap_extent_buffer(node, node->map_token,
1282 KM_USER1); 1286 KM_USER1);
1283 node->map_token = NULL; 1287 node->map_token = NULL;
@@ -1289,7 +1293,7 @@ static void reada_for_search(struct btrfs_root *root,
1289 if ((nread > 65536 || nscan > 32)) 1293 if ((nread > 65536 || nscan > 32))
1290 break; 1294 break;
1291 } 1295 }
1292 if (node->map_token) { 1296 if (map && node->map_token) {
1293 unmap_extent_buffer(node, node->map_token, KM_USER1); 1297 unmap_extent_buffer(node, node->map_token, KM_USER1);
1294 node->map_token = NULL; 1298 node->map_token = NULL;
1295 } 1299 }
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index a203d363184d..9f68c6898653 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1668,8 +1668,6 @@ struct btrfs_root *open_ctree(struct super_block *sb,
1668 init_waitqueue_head(&fs_info->scrub_pause_wait); 1668 init_waitqueue_head(&fs_info->scrub_pause_wait);
1669 init_rwsem(&fs_info->scrub_super_lock); 1669 init_rwsem(&fs_info->scrub_super_lock);
1670 fs_info->scrub_workers_refcnt = 0; 1670 fs_info->scrub_workers_refcnt = 0;
1671 btrfs_init_workers(&fs_info->scrub_workers, "scrub",
1672 fs_info->thread_pool_size, &fs_info->generic_worker);
1673 1671
1674 sb->s_blocksize = 4096; 1672 sb->s_blocksize = 4096;
1675 sb->s_blocksize_bits = blksize_bits(4096); 1673 sb->s_blocksize_bits = blksize_bits(4096);
@@ -2911,9 +2909,8 @@ static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
2911 2909
2912 INIT_LIST_HEAD(&splice); 2910 INIT_LIST_HEAD(&splice);
2913 2911
2914 list_splice_init(&root->fs_info->delalloc_inodes, &splice);
2915
2916 spin_lock(&root->fs_info->delalloc_lock); 2912 spin_lock(&root->fs_info->delalloc_lock);
2913 list_splice_init(&root->fs_info->delalloc_inodes, &splice);
2917 2914
2918 while (!list_empty(&splice)) { 2915 while (!list_empty(&splice)) {
2919 btrfs_inode = list_entry(splice.next, struct btrfs_inode, 2916 btrfs_inode = list_entry(splice.next, struct btrfs_inode,
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 5b9b6b6df242..b42efc2ded51 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3089,6 +3089,13 @@ alloc:
3089 } 3089 }
3090 goto again; 3090 goto again;
3091 } 3091 }
3092
3093 /*
3094 * If we have less pinned bytes than we want to allocate then
3095 * don't bother committing the transaction, it won't help us.
3096 */
3097 if (data_sinfo->bytes_pinned < bytes)
3098 committed = 1;
3092 spin_unlock(&data_sinfo->lock); 3099 spin_unlock(&data_sinfo->lock);
3093 3100
3094 /* commit the current transaction and try again */ 3101 /* commit the current transaction and try again */
@@ -5211,9 +5218,7 @@ loop:
5211 * LOOP_NO_EMPTY_SIZE, set empty_size and empty_cluster to 0 and try 5218 * LOOP_NO_EMPTY_SIZE, set empty_size and empty_cluster to 0 and try
5212 * again 5219 * again
5213 */ 5220 */
5214 if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE && 5221 if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE) {
5215 (found_uncached_bg || empty_size || empty_cluster ||
5216 allowed_chunk_alloc)) {
5217 index = 0; 5222 index = 0;
5218 if (loop == LOOP_FIND_IDEAL && found_uncached_bg) { 5223 if (loop == LOOP_FIND_IDEAL && found_uncached_bg) {
5219 found_uncached_bg = false; 5224 found_uncached_bg = false;
@@ -5253,32 +5258,36 @@ loop:
5253 goto search; 5258 goto search;
5254 } 5259 }
5255 5260
5256 if (loop < LOOP_CACHING_WAIT) { 5261 loop++;
5257 loop++;
5258 goto search;
5259 }
5260 5262
5261 if (loop == LOOP_ALLOC_CHUNK) { 5263 if (loop == LOOP_ALLOC_CHUNK) {
5262 empty_size = 0; 5264 if (allowed_chunk_alloc) {
5263 empty_cluster = 0; 5265 ret = do_chunk_alloc(trans, root, num_bytes +
5264 } 5266 2 * 1024 * 1024, data,
5267 CHUNK_ALLOC_LIMITED);
5268 allowed_chunk_alloc = 0;
5269 if (ret == 1)
5270 done_chunk_alloc = 1;
5271 } else if (!done_chunk_alloc &&
5272 space_info->force_alloc ==
5273 CHUNK_ALLOC_NO_FORCE) {
5274 space_info->force_alloc = CHUNK_ALLOC_LIMITED;
5275 }
5265 5276
5266 if (allowed_chunk_alloc) { 5277 /*
5267 ret = do_chunk_alloc(trans, root, num_bytes + 5278 * We didn't allocate a chunk, go ahead and drop the
5268 2 * 1024 * 1024, data, 5279 * empty size and loop again.
5269 CHUNK_ALLOC_LIMITED); 5280 */
5270 allowed_chunk_alloc = 0; 5281 if (!done_chunk_alloc)
5271 done_chunk_alloc = 1; 5282 loop = LOOP_NO_EMPTY_SIZE;
5272 } else if (!done_chunk_alloc &&
5273 space_info->force_alloc == CHUNK_ALLOC_NO_FORCE) {
5274 space_info->force_alloc = CHUNK_ALLOC_LIMITED;
5275 } 5283 }
5276 5284
5277 if (loop < LOOP_NO_EMPTY_SIZE) { 5285 if (loop == LOOP_NO_EMPTY_SIZE) {
5278 loop++; 5286 empty_size = 0;
5279 goto search; 5287 empty_cluster = 0;
5280 } 5288 }
5281 ret = -ENOSPC; 5289
5290 goto search;
5282 } else if (!ins->objectid) { 5291 } else if (!ins->objectid) {
5283 ret = -ENOSPC; 5292 ret = -ENOSPC;
5284 } else if (ins->objectid) { 5293 } else if (ins->objectid) {
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 4e8445a4757c..a11a92ee2d30 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -126,9 +126,9 @@ struct extent_buffer {
126 unsigned long map_len; 126 unsigned long map_len;
127 struct page *first_page; 127 struct page *first_page;
128 unsigned long bflags; 128 unsigned long bflags;
129 atomic_t refs;
130 struct list_head leak_list; 129 struct list_head leak_list;
131 struct rcu_head rcu_head; 130 struct rcu_head rcu_head;
131 atomic_t refs;
132 132
133 /* the spinlock is used to protect most operations */ 133 /* the spinlock is used to protect most operations */
134 spinlock_t lock; 134 spinlock_t lock;
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index ad144736a5fd..9f985a429877 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -250,7 +250,7 @@ int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
250 pgoff_t index = 0; 250 pgoff_t index = 0;
251 unsigned long first_page_offset; 251 unsigned long first_page_offset;
252 int num_checksums; 252 int num_checksums;
253 int ret = 0, ret2; 253 int ret = 0;
254 254
255 INIT_LIST_HEAD(&bitmaps); 255 INIT_LIST_HEAD(&bitmaps);
256 256
@@ -421,11 +421,10 @@ int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
421 goto free_cache; 421 goto free_cache;
422 } 422 }
423 spin_lock(&ctl->tree_lock); 423 spin_lock(&ctl->tree_lock);
424 ret2 = link_free_space(ctl, e); 424 ret = link_free_space(ctl, e);
425 ctl->total_bitmaps++; 425 ctl->total_bitmaps++;
426 ctl->op->recalc_thresholds(ctl); 426 ctl->op->recalc_thresholds(ctl);
427 spin_unlock(&ctl->tree_lock); 427 spin_unlock(&ctl->tree_lock);
428 list_add_tail(&e->list, &bitmaps);
429 if (ret) { 428 if (ret) {
430 printk(KERN_ERR "Duplicate entries in " 429 printk(KERN_ERR "Duplicate entries in "
431 "free space cache, dumping\n"); 430 "free space cache, dumping\n");
@@ -434,6 +433,7 @@ int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
434 page_cache_release(page); 433 page_cache_release(page);
435 goto free_cache; 434 goto free_cache;
436 } 435 }
436 list_add_tail(&e->list, &bitmaps);
437 } 437 }
438 438
439 num_entries--; 439 num_entries--;
@@ -1417,6 +1417,23 @@ again:
1417 return 0; 1417 return 0;
1418} 1418}
1419 1419
1420static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
1421 struct btrfs_free_space *info, u64 offset,
1422 u64 bytes)
1423{
1424 u64 bytes_to_set = 0;
1425 u64 end;
1426
1427 end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
1428
1429 bytes_to_set = min(end - offset, bytes);
1430
1431 bitmap_set_bits(ctl, info, offset, bytes_to_set);
1432
1433 return bytes_to_set;
1434
1435}
1436
1420static bool use_bitmap(struct btrfs_free_space_ctl *ctl, 1437static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
1421 struct btrfs_free_space *info) 1438 struct btrfs_free_space *info)
1422{ 1439{
@@ -1453,12 +1470,18 @@ static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
1453 return true; 1470 return true;
1454} 1471}
1455 1472
1473static struct btrfs_free_space_op free_space_op = {
1474 .recalc_thresholds = recalculate_thresholds,
1475 .use_bitmap = use_bitmap,
1476};
1477
1456static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl, 1478static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
1457 struct btrfs_free_space *info) 1479 struct btrfs_free_space *info)
1458{ 1480{
1459 struct btrfs_free_space *bitmap_info; 1481 struct btrfs_free_space *bitmap_info;
1482 struct btrfs_block_group_cache *block_group = NULL;
1460 int added = 0; 1483 int added = 0;
1461 u64 bytes, offset, end; 1484 u64 bytes, offset, bytes_added;
1462 int ret; 1485 int ret;
1463 1486
1464 bytes = info->bytes; 1487 bytes = info->bytes;
@@ -1467,7 +1490,49 @@ static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
1467 if (!ctl->op->use_bitmap(ctl, info)) 1490 if (!ctl->op->use_bitmap(ctl, info))
1468 return 0; 1491 return 0;
1469 1492
1493 if (ctl->op == &free_space_op)
1494 block_group = ctl->private;
1470again: 1495again:
1496 /*
1497 * Since we link bitmaps right into the cluster we need to see if we
1498 * have a cluster here, and if so and it has our bitmap we need to add
1499 * the free space to that bitmap.
1500 */
1501 if (block_group && !list_empty(&block_group->cluster_list)) {
1502 struct btrfs_free_cluster *cluster;
1503 struct rb_node *node;
1504 struct btrfs_free_space *entry;
1505
1506 cluster = list_entry(block_group->cluster_list.next,
1507 struct btrfs_free_cluster,
1508 block_group_list);
1509 spin_lock(&cluster->lock);
1510 node = rb_first(&cluster->root);
1511 if (!node) {
1512 spin_unlock(&cluster->lock);
1513 goto no_cluster_bitmap;
1514 }
1515
1516 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1517 if (!entry->bitmap) {
1518 spin_unlock(&cluster->lock);
1519 goto no_cluster_bitmap;
1520 }
1521
1522 if (entry->offset == offset_to_bitmap(ctl, offset)) {
1523 bytes_added = add_bytes_to_bitmap(ctl, entry,
1524 offset, bytes);
1525 bytes -= bytes_added;
1526 offset += bytes_added;
1527 }
1528 spin_unlock(&cluster->lock);
1529 if (!bytes) {
1530 ret = 1;
1531 goto out;
1532 }
1533 }
1534
1535no_cluster_bitmap:
1471 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset), 1536 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
1472 1, 0); 1537 1, 0);
1473 if (!bitmap_info) { 1538 if (!bitmap_info) {
@@ -1475,19 +1540,10 @@ again:
1475 goto new_bitmap; 1540 goto new_bitmap;
1476 } 1541 }
1477 1542
1478 end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit); 1543 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
1479 1544 bytes -= bytes_added;
1480 if (offset >= bitmap_info->offset && offset + bytes > end) { 1545 offset += bytes_added;
1481 bitmap_set_bits(ctl, bitmap_info, offset, end - offset); 1546 added = 0;
1482 bytes -= end - offset;
1483 offset = end;
1484 added = 0;
1485 } else if (offset >= bitmap_info->offset && offset + bytes <= end) {
1486 bitmap_set_bits(ctl, bitmap_info, offset, bytes);
1487 bytes = 0;
1488 } else {
1489 BUG();
1490 }
1491 1547
1492 if (!bytes) { 1548 if (!bytes) {
1493 ret = 1; 1549 ret = 1;
@@ -1766,11 +1822,6 @@ void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group,
1766 "\n", count); 1822 "\n", count);
1767} 1823}
1768 1824
1769static struct btrfs_free_space_op free_space_op = {
1770 .recalc_thresholds = recalculate_thresholds,
1771 .use_bitmap = use_bitmap,
1772};
1773
1774void btrfs_init_free_space_ctl(struct btrfs_block_group_cache *block_group) 1825void btrfs_init_free_space_ctl(struct btrfs_block_group_cache *block_group)
1775{ 1826{
1776 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl; 1827 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
@@ -2142,9 +2193,11 @@ again:
2142/* 2193/*
2143 * This searches the block group for just extents to fill the cluster with. 2194 * This searches the block group for just extents to fill the cluster with.
2144 */ 2195 */
2145static int setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group, 2196static noinline int
2146 struct btrfs_free_cluster *cluster, 2197setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2147 u64 offset, u64 bytes, u64 min_bytes) 2198 struct btrfs_free_cluster *cluster,
2199 struct list_head *bitmaps, u64 offset, u64 bytes,
2200 u64 min_bytes)
2148{ 2201{
2149 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl; 2202 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2150 struct btrfs_free_space *first = NULL; 2203 struct btrfs_free_space *first = NULL;
@@ -2166,6 +2219,8 @@ static int setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2166 * extent entry. 2219 * extent entry.
2167 */ 2220 */
2168 while (entry->bitmap) { 2221 while (entry->bitmap) {
2222 if (list_empty(&entry->list))
2223 list_add_tail(&entry->list, bitmaps);
2169 node = rb_next(&entry->offset_index); 2224 node = rb_next(&entry->offset_index);
2170 if (!node) 2225 if (!node)
2171 return -ENOSPC; 2226 return -ENOSPC;
@@ -2185,8 +2240,12 @@ static int setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2185 return -ENOSPC; 2240 return -ENOSPC;
2186 entry = rb_entry(node, struct btrfs_free_space, offset_index); 2241 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2187 2242
2188 if (entry->bitmap) 2243 if (entry->bitmap) {
2244 if (list_empty(&entry->list))
2245 list_add_tail(&entry->list, bitmaps);
2189 continue; 2246 continue;
2247 }
2248
2190 /* 2249 /*
2191 * we haven't filled the empty size and the window is 2250 * we haven't filled the empty size and the window is
2192 * very large. reset and try again 2251 * very large. reset and try again
@@ -2238,9 +2297,11 @@ static int setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2238 * This specifically looks for bitmaps that may work in the cluster, we assume 2297 * This specifically looks for bitmaps that may work in the cluster, we assume
2239 * that we have already failed to find extents that will work. 2298 * that we have already failed to find extents that will work.
2240 */ 2299 */
2241static int setup_cluster_bitmap(struct btrfs_block_group_cache *block_group, 2300static noinline int
2242 struct btrfs_free_cluster *cluster, 2301setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2243 u64 offset, u64 bytes, u64 min_bytes) 2302 struct btrfs_free_cluster *cluster,
2303 struct list_head *bitmaps, u64 offset, u64 bytes,
2304 u64 min_bytes)
2244{ 2305{
2245 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl; 2306 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2246 struct btrfs_free_space *entry; 2307 struct btrfs_free_space *entry;
@@ -2250,10 +2311,39 @@ static int setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2250 if (ctl->total_bitmaps == 0) 2311 if (ctl->total_bitmaps == 0)
2251 return -ENOSPC; 2312 return -ENOSPC;
2252 2313
2314 /*
2315 * First check our cached list of bitmaps and see if there is an entry
2316 * here that will work.
2317 */
2318 list_for_each_entry(entry, bitmaps, list) {
2319 if (entry->bytes < min_bytes)
2320 continue;
2321 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
2322 bytes, min_bytes);
2323 if (!ret)
2324 return 0;
2325 }
2326
2327 /*
2328 * If we do have entries on our list and we are here then we didn't find
2329 * anything, so go ahead and get the next entry after the last entry in
2330 * this list and start the search from there.
2331 */
2332 if (!list_empty(bitmaps)) {
2333 entry = list_entry(bitmaps->prev, struct btrfs_free_space,
2334 list);
2335 node = rb_next(&entry->offset_index);
2336 if (!node)
2337 return -ENOSPC;
2338 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2339 goto search;
2340 }
2341
2253 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, offset), 0, 1); 2342 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, offset), 0, 1);
2254 if (!entry) 2343 if (!entry)
2255 return -ENOSPC; 2344 return -ENOSPC;
2256 2345
2346search:
2257 node = &entry->offset_index; 2347 node = &entry->offset_index;
2258 do { 2348 do {
2259 entry = rb_entry(node, struct btrfs_free_space, offset_index); 2349 entry = rb_entry(node, struct btrfs_free_space, offset_index);
@@ -2284,6 +2374,8 @@ int btrfs_find_space_cluster(struct btrfs_trans_handle *trans,
2284 u64 offset, u64 bytes, u64 empty_size) 2374 u64 offset, u64 bytes, u64 empty_size)
2285{ 2375{
2286 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl; 2376 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2377 struct list_head bitmaps;
2378 struct btrfs_free_space *entry, *tmp;
2287 u64 min_bytes; 2379 u64 min_bytes;
2288 int ret; 2380 int ret;
2289 2381
@@ -2322,11 +2414,16 @@ int btrfs_find_space_cluster(struct btrfs_trans_handle *trans,
2322 goto out; 2414 goto out;
2323 } 2415 }
2324 2416
2325 ret = setup_cluster_no_bitmap(block_group, cluster, offset, bytes, 2417 INIT_LIST_HEAD(&bitmaps);
2326 min_bytes); 2418 ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
2419 bytes, min_bytes);
2327 if (ret) 2420 if (ret)
2328 ret = setup_cluster_bitmap(block_group, cluster, offset, 2421 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
2329 bytes, min_bytes); 2422 offset, bytes, min_bytes);
2423
2424 /* Clear our temporary list */
2425 list_for_each_entry_safe(entry, tmp, &bitmaps, list)
2426 list_del_init(&entry->list);
2330 2427
2331 if (!ret) { 2428 if (!ret) {
2332 atomic_inc(&block_group->count); 2429 atomic_inc(&block_group->count);
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index ebf95f7a44d6..751ddf8fc58a 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -1986,7 +1986,7 @@ static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end,
1986 } 1986 }
1987 1987
1988 if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM) 1988 if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)
1989 return 0; 1989 goto good;
1990 1990
1991 if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID && 1991 if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID &&
1992 test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1, NULL)) { 1992 test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1, NULL)) {
@@ -3646,7 +3646,7 @@ void btrfs_evict_inode(struct inode *inode)
3646 btrfs_i_size_write(inode, 0); 3646 btrfs_i_size_write(inode, 0);
3647 3647
3648 while (1) { 3648 while (1) {
3649 trans = btrfs_start_transaction(root, 0); 3649 trans = btrfs_join_transaction(root);
3650 BUG_ON(IS_ERR(trans)); 3650 BUG_ON(IS_ERR(trans));
3651 trans->block_rsv = root->orphan_block_rsv; 3651 trans->block_rsv = root->orphan_block_rsv;
3652 3652
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index ac37040e426a..b793d112d1f6 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -2054,29 +2054,34 @@ static long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg)
2054 2054
2055static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg) 2055static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg)
2056{ 2056{
2057 struct btrfs_ioctl_fs_info_args fi_args; 2057 struct btrfs_ioctl_fs_info_args *fi_args;
2058 struct btrfs_device *device; 2058 struct btrfs_device *device;
2059 struct btrfs_device *next; 2059 struct btrfs_device *next;
2060 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices; 2060 struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
2061 int ret = 0;
2061 2062
2062 if (!capable(CAP_SYS_ADMIN)) 2063 if (!capable(CAP_SYS_ADMIN))
2063 return -EPERM; 2064 return -EPERM;
2064 2065
2065 fi_args.num_devices = fs_devices->num_devices; 2066 fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
2066 fi_args.max_id = 0; 2067 if (!fi_args)
2067 memcpy(&fi_args.fsid, root->fs_info->fsid, sizeof(fi_args.fsid)); 2068 return -ENOMEM;
2069
2070 fi_args->num_devices = fs_devices->num_devices;
2071 memcpy(&fi_args->fsid, root->fs_info->fsid, sizeof(fi_args->fsid));
2068 2072
2069 mutex_lock(&fs_devices->device_list_mutex); 2073 mutex_lock(&fs_devices->device_list_mutex);
2070 list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) { 2074 list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
2071 if (device->devid > fi_args.max_id) 2075 if (device->devid > fi_args->max_id)
2072 fi_args.max_id = device->devid; 2076 fi_args->max_id = device->devid;
2073 } 2077 }
2074 mutex_unlock(&fs_devices->device_list_mutex); 2078 mutex_unlock(&fs_devices->device_list_mutex);
2075 2079
2076 if (copy_to_user(arg, &fi_args, sizeof(fi_args))) 2080 if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
2077 return -EFAULT; 2081 ret = -EFAULT;
2078 2082
2079 return 0; 2083 kfree(fi_args);
2084 return ret;
2080} 2085}
2081 2086
2082static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg) 2087static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg)
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index df50fd1eca8f..a8d03d5efb5d 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -16,13 +16,7 @@
16 * Boston, MA 021110-1307, USA. 16 * Boston, MA 021110-1307, USA.
17 */ 17 */
18 18
19#include <linux/sched.h>
20#include <linux/pagemap.h>
21#include <linux/writeback.h>
22#include <linux/blkdev.h> 19#include <linux/blkdev.h>
23#include <linux/rbtree.h>
24#include <linux/slab.h>
25#include <linux/workqueue.h>
26#include "ctree.h" 20#include "ctree.h"
27#include "volumes.h" 21#include "volumes.h"
28#include "disk-io.h" 22#include "disk-io.h"
@@ -804,18 +798,12 @@ static noinline_for_stack int scrub_stripe(struct scrub_dev *sdev,
804 798
805 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); 799 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
806 if (ret < 0) 800 if (ret < 0)
807 goto out; 801 goto out_noplug;
808
809 l = path->nodes[0];
810 slot = path->slots[0];
811 btrfs_item_key_to_cpu(l, &key, slot);
812 if (key.objectid != logical) {
813 ret = btrfs_previous_item(root, path, 0,
814 BTRFS_EXTENT_ITEM_KEY);
815 if (ret < 0)
816 goto out;
817 }
818 802
803 /*
804 * we might miss half an extent here, but that doesn't matter,
805 * as it's only the prefetch
806 */
819 while (1) { 807 while (1) {
820 l = path->nodes[0]; 808 l = path->nodes[0];
821 slot = path->slots[0]; 809 slot = path->slots[0];
@@ -824,7 +812,7 @@ static noinline_for_stack int scrub_stripe(struct scrub_dev *sdev,
824 if (ret == 0) 812 if (ret == 0)
825 continue; 813 continue;
826 if (ret < 0) 814 if (ret < 0)
827 goto out; 815 goto out_noplug;
828 816
829 break; 817 break;
830 } 818 }
@@ -906,15 +894,20 @@ again:
906 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); 894 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
907 if (ret < 0) 895 if (ret < 0)
908 goto out; 896 goto out;
909 897 if (ret > 0) {
910 l = path->nodes[0];
911 slot = path->slots[0];
912 btrfs_item_key_to_cpu(l, &key, slot);
913 if (key.objectid != logical) {
914 ret = btrfs_previous_item(root, path, 0, 898 ret = btrfs_previous_item(root, path, 0,
915 BTRFS_EXTENT_ITEM_KEY); 899 BTRFS_EXTENT_ITEM_KEY);
916 if (ret < 0) 900 if (ret < 0)
917 goto out; 901 goto out;
902 if (ret > 0) {
903 /* there's no smaller item, so stick with the
904 * larger one */
905 btrfs_release_path(path);
906 ret = btrfs_search_slot(NULL, root, &key,
907 path, 0, 0);
908 if (ret < 0)
909 goto out;
910 }
918 } 911 }
919 912
920 while (1) { 913 while (1) {
@@ -989,6 +982,7 @@ next:
989 982
990out: 983out:
991 blk_finish_plug(&plug); 984 blk_finish_plug(&plug);
985out_noplug:
992 btrfs_free_path(path); 986 btrfs_free_path(path);
993 return ret < 0 ? ret : 0; 987 return ret < 0 ? ret : 0;
994} 988}
@@ -1064,8 +1058,15 @@ int scrub_enumerate_chunks(struct scrub_dev *sdev, u64 start, u64 end)
1064 while (1) { 1058 while (1) {
1065 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); 1059 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1066 if (ret < 0) 1060 if (ret < 0)
1067 goto out; 1061 break;
1068 ret = 0; 1062 if (ret > 0) {
1063 if (path->slots[0] >=
1064 btrfs_header_nritems(path->nodes[0])) {
1065 ret = btrfs_next_leaf(root, path);
1066 if (ret)
1067 break;
1068 }
1069 }
1069 1070
1070 l = path->nodes[0]; 1071 l = path->nodes[0];
1071 slot = path->slots[0]; 1072 slot = path->slots[0];
@@ -1075,7 +1076,7 @@ int scrub_enumerate_chunks(struct scrub_dev *sdev, u64 start, u64 end)
1075 if (found_key.objectid != sdev->dev->devid) 1076 if (found_key.objectid != sdev->dev->devid)
1076 break; 1077 break;
1077 1078
1078 if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY) 1079 if (btrfs_key_type(&found_key) != BTRFS_DEV_EXTENT_KEY)
1079 break; 1080 break;
1080 1081
1081 if (found_key.offset >= end) 1082 if (found_key.offset >= end)
@@ -1104,7 +1105,7 @@ int scrub_enumerate_chunks(struct scrub_dev *sdev, u64 start, u64 end)
1104 cache = btrfs_lookup_block_group(fs_info, chunk_offset); 1105 cache = btrfs_lookup_block_group(fs_info, chunk_offset);
1105 if (!cache) { 1106 if (!cache) {
1106 ret = -ENOENT; 1107 ret = -ENOENT;
1107 goto out; 1108 break;
1108 } 1109 }
1109 ret = scrub_chunk(sdev, chunk_tree, chunk_objectid, 1110 ret = scrub_chunk(sdev, chunk_tree, chunk_objectid,
1110 chunk_offset, length); 1111 chunk_offset, length);
@@ -1116,9 +1117,13 @@ int scrub_enumerate_chunks(struct scrub_dev *sdev, u64 start, u64 end)
1116 btrfs_release_path(path); 1117 btrfs_release_path(path);
1117 } 1118 }
1118 1119
1119out:
1120 btrfs_free_path(path); 1120 btrfs_free_path(path);
1121 return ret; 1121
1122 /*
1123 * ret can still be 1 from search_slot or next_leaf,
1124 * that's not an error
1125 */
1126 return ret < 0 ? ret : 0;
1122} 1127}
1123 1128
1124static noinline_for_stack int scrub_supers(struct scrub_dev *sdev) 1129static noinline_for_stack int scrub_supers(struct scrub_dev *sdev)
@@ -1155,8 +1160,12 @@ static noinline_for_stack int scrub_workers_get(struct btrfs_root *root)
1155 struct btrfs_fs_info *fs_info = root->fs_info; 1160 struct btrfs_fs_info *fs_info = root->fs_info;
1156 1161
1157 mutex_lock(&fs_info->scrub_lock); 1162 mutex_lock(&fs_info->scrub_lock);
1158 if (fs_info->scrub_workers_refcnt == 0) 1163 if (fs_info->scrub_workers_refcnt == 0) {
1164 btrfs_init_workers(&fs_info->scrub_workers, "scrub",
1165 fs_info->thread_pool_size, &fs_info->generic_worker);
1166 fs_info->scrub_workers.idle_thresh = 4;
1159 btrfs_start_workers(&fs_info->scrub_workers, 1); 1167 btrfs_start_workers(&fs_info->scrub_workers, 1);
1168 }
1160 ++fs_info->scrub_workers_refcnt; 1169 ++fs_info->scrub_workers_refcnt;
1161 mutex_unlock(&fs_info->scrub_lock); 1170 mutex_unlock(&fs_info->scrub_lock);
1162 1171
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 117e74e3604b..0bb4ebbb71b7 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -825,7 +825,7 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
825 } else { 825 } else {
826 char b[BDEVNAME_SIZE]; 826 char b[BDEVNAME_SIZE];
827 827
828 s->s_flags = flags; 828 s->s_flags = flags | MS_NOSEC;
829 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id)); 829 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
830 error = btrfs_fill_super(s, fs_devices, data, 830 error = btrfs_fill_super(s, fs_devices, data,
831 flags & MS_SILENT ? 1 : 0); 831 flags & MS_SILENT ? 1 : 0);
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index dd719662340e..2b3590b9fe98 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -349,7 +349,7 @@ int btrfs_wait_for_commit(struct btrfs_root *root, u64 transid)
349 list) { 349 list) {
350 if (t->in_commit) { 350 if (t->in_commit) {
351 if (t->commit_done) 351 if (t->commit_done)
352 goto out; 352 break;
353 cur_trans = t; 353 cur_trans = t;
354 atomic_inc(&cur_trans->use_count); 354 atomic_inc(&cur_trans->use_count);
355 break; 355 break;
@@ -1118,8 +1118,11 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
1118 wait_current_trans_commit_start_and_unblock(root, cur_trans); 1118 wait_current_trans_commit_start_and_unblock(root, cur_trans);
1119 else 1119 else
1120 wait_current_trans_commit_start(root, cur_trans); 1120 wait_current_trans_commit_start(root, cur_trans);
1121 put_transaction(cur_trans);
1122 1121
1122 if (current->journal_info == trans)
1123 current->journal_info = NULL;
1124
1125 put_transaction(cur_trans);
1123 return 0; 1126 return 0;
1124} 1127}
1125 1128
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index da541dfca2e3..1efa56e18f9b 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -689,12 +689,8 @@ int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
689 transid = btrfs_super_generation(disk_super); 689 transid = btrfs_super_generation(disk_super);
690 if (disk_super->label[0]) 690 if (disk_super->label[0])
691 printk(KERN_INFO "device label %s ", disk_super->label); 691 printk(KERN_INFO "device label %s ", disk_super->label);
692 else { 692 else
693 /* FIXME, make a readl uuid parser */ 693 printk(KERN_INFO "device fsid %pU ", disk_super->fsid);
694 printk(KERN_INFO "device fsid %llx-%llx ",
695 *(unsigned long long *)disk_super->fsid,
696 *(unsigned long long *)(disk_super->fsid + 8));
697 }
698 printk(KERN_CONT "devid %llu transid %llu %s\n", 694 printk(KERN_CONT "devid %llu transid %llu %s\n",
699 (unsigned long long)devid, (unsigned long long)transid, path); 695 (unsigned long long)devid, (unsigned long long)transid, path);
700 ret = device_list_add(path, disk_super, devid, fs_devices_ret); 696 ret = device_list_add(path, disk_super, devid, fs_devices_ret);
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index 33da49dc3cc6..5a3953db8118 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -453,7 +453,7 @@ static int ceph_writepage(struct page *page, struct writeback_control *wbc)
453 int err; 453 int err;
454 struct inode *inode = page->mapping->host; 454 struct inode *inode = page->mapping->host;
455 BUG_ON(!inode); 455 BUG_ON(!inode);
456 igrab(inode); 456 ihold(inode);
457 err = writepage_nounlock(page, wbc); 457 err = writepage_nounlock(page, wbc);
458 unlock_page(page); 458 unlock_page(page);
459 iput(inode); 459 iput(inode);
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index 1f72b00447c4..f605753c8fe9 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -2940,14 +2940,12 @@ void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc)
2940 while (!list_empty(&mdsc->cap_dirty)) { 2940 while (!list_empty(&mdsc->cap_dirty)) {
2941 ci = list_first_entry(&mdsc->cap_dirty, struct ceph_inode_info, 2941 ci = list_first_entry(&mdsc->cap_dirty, struct ceph_inode_info,
2942 i_dirty_item); 2942 i_dirty_item);
2943 inode = igrab(&ci->vfs_inode); 2943 inode = &ci->vfs_inode;
2944 ihold(inode);
2944 dout("flush_dirty_caps %p\n", inode); 2945 dout("flush_dirty_caps %p\n", inode);
2945 spin_unlock(&mdsc->cap_dirty_lock); 2946 spin_unlock(&mdsc->cap_dirty_lock);
2946 if (inode) { 2947 ceph_check_caps(ci, CHECK_CAPS_NODELAY|CHECK_CAPS_FLUSH, NULL);
2947 ceph_check_caps(ci, CHECK_CAPS_NODELAY|CHECK_CAPS_FLUSH, 2948 iput(inode);
2948 NULL);
2949 iput(inode);
2950 }
2951 spin_lock(&mdsc->cap_dirty_lock); 2949 spin_lock(&mdsc->cap_dirty_lock);
2952 } 2950 }
2953 spin_unlock(&mdsc->cap_dirty_lock); 2951 spin_unlock(&mdsc->cap_dirty_lock);
diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 33729e822bb9..ef8f08c343e8 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -308,7 +308,8 @@ more:
308 req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS); 308 req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
309 if (IS_ERR(req)) 309 if (IS_ERR(req))
310 return PTR_ERR(req); 310 return PTR_ERR(req);
311 req->r_inode = igrab(inode); 311 req->r_inode = inode;
312 ihold(inode);
312 req->r_dentry = dget(filp->f_dentry); 313 req->r_dentry = dget(filp->f_dentry);
313 /* hints to request -> mds selection code */ 314 /* hints to request -> mds selection code */
314 req->r_direct_mode = USE_AUTH_MDS; 315 req->r_direct_mode = USE_AUTH_MDS;
@@ -787,10 +788,12 @@ static int ceph_link(struct dentry *old_dentry, struct inode *dir,
787 req->r_dentry_drop = CEPH_CAP_FILE_SHARED; 788 req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
788 req->r_dentry_unless = CEPH_CAP_FILE_EXCL; 789 req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
789 err = ceph_mdsc_do_request(mdsc, dir, req); 790 err = ceph_mdsc_do_request(mdsc, dir, req);
790 if (err) 791 if (err) {
791 d_drop(dentry); 792 d_drop(dentry);
792 else if (!req->r_reply_info.head->is_dentry) 793 } else if (!req->r_reply_info.head->is_dentry) {
793 d_instantiate(dentry, igrab(old_dentry->d_inode)); 794 ihold(old_dentry->d_inode);
795 d_instantiate(dentry, old_dentry->d_inode);
796 }
794 ceph_mdsc_put_request(req); 797 ceph_mdsc_put_request(req);
795 return err; 798 return err;
796} 799}
diff --git a/fs/ceph/export.c b/fs/ceph/export.c
index a610d3d67488..f67b687550de 100644
--- a/fs/ceph/export.c
+++ b/fs/ceph/export.c
@@ -109,7 +109,7 @@ static struct dentry *__fh_to_dentry(struct super_block *sb,
109 err = ceph_mdsc_do_request(mdsc, NULL, req); 109 err = ceph_mdsc_do_request(mdsc, NULL, req);
110 inode = req->r_target_inode; 110 inode = req->r_target_inode;
111 if (inode) 111 if (inode)
112 igrab(inode); 112 ihold(inode);
113 ceph_mdsc_put_request(req); 113 ceph_mdsc_put_request(req);
114 if (!inode) 114 if (!inode)
115 return ERR_PTR(-ESTALE); 115 return ERR_PTR(-ESTALE);
@@ -167,7 +167,7 @@ static struct dentry *__cfh_to_dentry(struct super_block *sb,
167 err = ceph_mdsc_do_request(mdsc, NULL, req); 167 err = ceph_mdsc_do_request(mdsc, NULL, req);
168 inode = req->r_target_inode; 168 inode = req->r_target_inode;
169 if (inode) 169 if (inode)
170 igrab(inode); 170 ihold(inode);
171 ceph_mdsc_put_request(req); 171 ceph_mdsc_put_request(req);
172 if (!inode) 172 if (!inode)
173 return ERR_PTR(err ? err : -ESTALE); 173 return ERR_PTR(err ? err : -ESTALE);
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 203252d88d9f..9542f07d0b93 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -191,7 +191,8 @@ int ceph_open(struct inode *inode, struct file *file)
191 err = PTR_ERR(req); 191 err = PTR_ERR(req);
192 goto out; 192 goto out;
193 } 193 }
194 req->r_inode = igrab(inode); 194 req->r_inode = inode;
195 ihold(inode);
195 req->r_num_caps = 1; 196 req->r_num_caps = 1;
196 err = ceph_mdsc_do_request(mdsc, parent_inode, req); 197 err = ceph_mdsc_do_request(mdsc, parent_inode, req);
197 if (!err) 198 if (!err)
@@ -282,7 +283,7 @@ int ceph_release(struct inode *inode, struct file *file)
282static int striped_read(struct inode *inode, 283static int striped_read(struct inode *inode,
283 u64 off, u64 len, 284 u64 off, u64 len,
284 struct page **pages, int num_pages, 285 struct page **pages, int num_pages,
285 int *checkeof, bool align_to_pages, 286 int *checkeof, bool o_direct,
286 unsigned long buf_align) 287 unsigned long buf_align)
287{ 288{
288 struct ceph_fs_client *fsc = ceph_inode_to_client(inode); 289 struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
@@ -307,7 +308,7 @@ static int striped_read(struct inode *inode,
307 io_align = off & ~PAGE_MASK; 308 io_align = off & ~PAGE_MASK;
308 309
309more: 310more:
310 if (align_to_pages) 311 if (o_direct)
311 page_align = (pos - io_align + buf_align) & ~PAGE_MASK; 312 page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
312 else 313 else
313 page_align = pos & ~PAGE_MASK; 314 page_align = pos & ~PAGE_MASK;
@@ -317,10 +318,10 @@ more:
317 ci->i_truncate_seq, 318 ci->i_truncate_seq,
318 ci->i_truncate_size, 319 ci->i_truncate_size,
319 page_pos, pages_left, page_align); 320 page_pos, pages_left, page_align);
320 hit_stripe = this_len < left;
321 was_short = ret >= 0 && ret < this_len;
322 if (ret == -ENOENT) 321 if (ret == -ENOENT)
323 ret = 0; 322 ret = 0;
323 hit_stripe = this_len < left;
324 was_short = ret >= 0 && ret < this_len;
324 dout("striped_read %llu~%u (read %u) got %d%s%s\n", pos, left, read, 325 dout("striped_read %llu~%u (read %u) got %d%s%s\n", pos, left, read,
325 ret, hit_stripe ? " HITSTRIPE" : "", was_short ? " SHORT" : ""); 326 ret, hit_stripe ? " HITSTRIPE" : "", was_short ? " SHORT" : "");
326 327
@@ -345,20 +346,22 @@ more:
345 } 346 }
346 347
347 if (was_short) { 348 if (was_short) {
348 /* was original extent fully inside i_size? */ 349 /* did we bounce off eof? */
349 if (pos + left <= inode->i_size) { 350 if (pos + left > inode->i_size)
350 dout("zero tail\n"); 351 *checkeof = 1;
351 ceph_zero_page_vector_range(page_off + read, len - read, 352
353 /* zero trailing bytes (inside i_size) */
354 if (left > 0 && pos < inode->i_size) {
355 if (pos + left > inode->i_size)
356 left = inode->i_size - pos;
357
358 dout("zero tail %d\n", left);
359 ceph_zero_page_vector_range(page_off + read, left,
352 pages); 360 pages);
353 read = len; 361 read += left;
354 goto out;
355 } 362 }
356
357 /* check i_size */
358 *checkeof = 1;
359 } 363 }
360 364
361out:
362 if (ret >= 0) 365 if (ret >= 0)
363 ret = read; 366 ret = read;
364 dout("striped_read returns %d\n", ret); 367 dout("striped_read returns %d\n", ret);
@@ -658,7 +661,7 @@ out:
658 661
659 /* hit EOF or hole? */ 662 /* hit EOF or hole? */
660 if (statret == 0 && *ppos < inode->i_size) { 663 if (statret == 0 && *ppos < inode->i_size) {
661 dout("aio_read sync_read hit hole, reading more\n"); 664 dout("aio_read sync_read hit hole, ppos %lld < size %lld, reading more\n", *ppos, inode->i_size);
662 read += ret; 665 read += ret;
663 base += ret; 666 base += ret;
664 len -= ret; 667 len -= ret;
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 70b6a4839c38..d8858e96ab18 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -1101,10 +1101,10 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req,
1101 goto done; 1101 goto done;
1102 } 1102 }
1103 req->r_dentry = dn; /* may have spliced */ 1103 req->r_dentry = dn; /* may have spliced */
1104 igrab(in); 1104 ihold(in);
1105 } else if (ceph_ino(in) == vino.ino && 1105 } else if (ceph_ino(in) == vino.ino &&
1106 ceph_snap(in) == vino.snap) { 1106 ceph_snap(in) == vino.snap) {
1107 igrab(in); 1107 ihold(in);
1108 } else { 1108 } else {
1109 dout(" %p links to %p %llx.%llx, not %llx.%llx\n", 1109 dout(" %p links to %p %llx.%llx, not %llx.%llx\n",
1110 dn, in, ceph_ino(in), ceph_snap(in), 1110 dn, in, ceph_ino(in), ceph_snap(in),
@@ -1144,7 +1144,7 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req,
1144 goto done; 1144 goto done;
1145 } 1145 }
1146 req->r_dentry = dn; /* may have spliced */ 1146 req->r_dentry = dn; /* may have spliced */
1147 igrab(in); 1147 ihold(in);
1148 rinfo->head->is_dentry = 1; /* fool notrace handlers */ 1148 rinfo->head->is_dentry = 1; /* fool notrace handlers */
1149 } 1149 }
1150 1150
@@ -1328,7 +1328,7 @@ void ceph_queue_writeback(struct inode *inode)
1328 if (queue_work(ceph_inode_to_client(inode)->wb_wq, 1328 if (queue_work(ceph_inode_to_client(inode)->wb_wq,
1329 &ceph_inode(inode)->i_wb_work)) { 1329 &ceph_inode(inode)->i_wb_work)) {
1330 dout("ceph_queue_writeback %p\n", inode); 1330 dout("ceph_queue_writeback %p\n", inode);
1331 igrab(inode); 1331 ihold(inode);
1332 } else { 1332 } else {
1333 dout("ceph_queue_writeback %p failed\n", inode); 1333 dout("ceph_queue_writeback %p failed\n", inode);
1334 } 1334 }
@@ -1353,7 +1353,7 @@ void ceph_queue_invalidate(struct inode *inode)
1353 if (queue_work(ceph_inode_to_client(inode)->pg_inv_wq, 1353 if (queue_work(ceph_inode_to_client(inode)->pg_inv_wq,
1354 &ceph_inode(inode)->i_pg_inv_work)) { 1354 &ceph_inode(inode)->i_pg_inv_work)) {
1355 dout("ceph_queue_invalidate %p\n", inode); 1355 dout("ceph_queue_invalidate %p\n", inode);
1356 igrab(inode); 1356 ihold(inode);
1357 } else { 1357 } else {
1358 dout("ceph_queue_invalidate %p failed\n", inode); 1358 dout("ceph_queue_invalidate %p failed\n", inode);
1359 } 1359 }
@@ -1477,7 +1477,7 @@ void ceph_queue_vmtruncate(struct inode *inode)
1477 if (queue_work(ceph_sb_to_client(inode->i_sb)->trunc_wq, 1477 if (queue_work(ceph_sb_to_client(inode->i_sb)->trunc_wq,
1478 &ci->i_vmtruncate_work)) { 1478 &ci->i_vmtruncate_work)) {
1479 dout("ceph_queue_vmtruncate %p\n", inode); 1479 dout("ceph_queue_vmtruncate %p\n", inode);
1480 igrab(inode); 1480 ihold(inode);
1481 } else { 1481 } else {
1482 dout("ceph_queue_vmtruncate %p failed, pending=%d\n", 1482 dout("ceph_queue_vmtruncate %p failed, pending=%d\n",
1483 inode, ci->i_truncate_pending); 1483 inode, ci->i_truncate_pending);
@@ -1738,7 +1738,8 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr)
1738 __mark_inode_dirty(inode, inode_dirty_flags); 1738 __mark_inode_dirty(inode, inode_dirty_flags);
1739 1739
1740 if (mask) { 1740 if (mask) {
1741 req->r_inode = igrab(inode); 1741 req->r_inode = inode;
1742 ihold(inode);
1742 req->r_inode_drop = release; 1743 req->r_inode_drop = release;
1743 req->r_args.setattr.mask = cpu_to_le32(mask); 1744 req->r_args.setattr.mask = cpu_to_le32(mask);
1744 req->r_num_caps = 1; 1745 req->r_num_caps = 1;
@@ -1779,7 +1780,8 @@ int ceph_do_getattr(struct inode *inode, int mask)
1779 req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETATTR, USE_ANY_MDS); 1780 req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETATTR, USE_ANY_MDS);
1780 if (IS_ERR(req)) 1781 if (IS_ERR(req))
1781 return PTR_ERR(req); 1782 return PTR_ERR(req);
1782 req->r_inode = igrab(inode); 1783 req->r_inode = inode;
1784 ihold(inode);
1783 req->r_num_caps = 1; 1785 req->r_num_caps = 1;
1784 req->r_args.getattr.mask = cpu_to_le32(mask); 1786 req->r_args.getattr.mask = cpu_to_le32(mask);
1785 err = ceph_mdsc_do_request(mdsc, NULL, req); 1787 err = ceph_mdsc_do_request(mdsc, NULL, req);
diff --git a/fs/ceph/ioctl.c b/fs/ceph/ioctl.c
index 8888c9ba68db..ef0b5f48e13a 100644
--- a/fs/ceph/ioctl.c
+++ b/fs/ceph/ioctl.c
@@ -73,7 +73,8 @@ static long ceph_ioctl_set_layout(struct file *file, void __user *arg)
73 USE_AUTH_MDS); 73 USE_AUTH_MDS);
74 if (IS_ERR(req)) 74 if (IS_ERR(req))
75 return PTR_ERR(req); 75 return PTR_ERR(req);
76 req->r_inode = igrab(inode); 76 req->r_inode = inode;
77 ihold(inode);
77 req->r_inode_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL; 78 req->r_inode_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL;
78 79
79 req->r_args.setlayout.layout.fl_stripe_unit = 80 req->r_args.setlayout.layout.fl_stripe_unit =
@@ -135,7 +136,8 @@ static long ceph_ioctl_set_layout_policy (struct file *file, void __user *arg)
135 136
136 if (IS_ERR(req)) 137 if (IS_ERR(req))
137 return PTR_ERR(req); 138 return PTR_ERR(req);
138 req->r_inode = igrab(inode); 139 req->r_inode = inode;
140 ihold(inode);
139 141
140 req->r_args.setlayout.layout.fl_stripe_unit = 142 req->r_args.setlayout.layout.fl_stripe_unit =
141 cpu_to_le32(l.stripe_unit); 143 cpu_to_le32(l.stripe_unit);
diff --git a/fs/ceph/locks.c b/fs/ceph/locks.c
index 476b329867d4..80576d05d687 100644
--- a/fs/ceph/locks.c
+++ b/fs/ceph/locks.c
@@ -23,7 +23,8 @@ static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
23 req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS); 23 req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS);
24 if (IS_ERR(req)) 24 if (IS_ERR(req))
25 return PTR_ERR(req); 25 return PTR_ERR(req);
26 req->r_inode = igrab(inode); 26 req->r_inode = inode;
27 ihold(inode);
27 28
28 /* mds requires start and length rather than start and end */ 29 /* mds requires start and length rather than start and end */
29 if (LLONG_MAX == fl->fl_end) 30 if (LLONG_MAX == fl->fl_end)
@@ -32,11 +33,10 @@ static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
32 length = fl->fl_end - fl->fl_start + 1; 33 length = fl->fl_end - fl->fl_start + 1;
33 34
34 dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, " 35 dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
35 "length: %llu, wait: %d, type`: %d", (int)lock_type, 36 "length: %llu, wait: %d, type: %d", (int)lock_type,
36 (int)operation, (u64)fl->fl_pid, fl->fl_start, 37 (int)operation, (u64)fl->fl_pid, fl->fl_start,
37 length, wait, fl->fl_type); 38 length, wait, fl->fl_type);
38 39
39
40 req->r_args.filelock_change.rule = lock_type; 40 req->r_args.filelock_change.rule = lock_type;
41 req->r_args.filelock_change.type = cmd; 41 req->r_args.filelock_change.type = cmd;
42 req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid); 42 req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
@@ -70,7 +70,7 @@ static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
70 } 70 }
71 ceph_mdsc_put_request(req); 71 ceph_mdsc_put_request(req);
72 dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, " 72 dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
73 "length: %llu, wait: %d, type`: %d, err code %d", (int)lock_type, 73 "length: %llu, wait: %d, type: %d, err code %d", (int)lock_type,
74 (int)operation, (u64)fl->fl_pid, fl->fl_start, 74 (int)operation, (u64)fl->fl_pid, fl->fl_start,
75 length, wait, fl->fl_type, err); 75 length, wait, fl->fl_type, err);
76 return err; 76 return err;
@@ -109,16 +109,20 @@ int ceph_lock(struct file *file, int cmd, struct file_lock *fl)
109 dout("mds locked, locking locally"); 109 dout("mds locked, locking locally");
110 err = posix_lock_file(file, fl, NULL); 110 err = posix_lock_file(file, fl, NULL);
111 if (err && (CEPH_MDS_OP_SETFILELOCK == op)) { 111 if (err && (CEPH_MDS_OP_SETFILELOCK == op)) {
112 /* undo! This should only happen if the kernel detects 112 /* undo! This should only happen if
113 * local deadlock. */ 113 * the kernel detects local
114 * deadlock. */
114 ceph_lock_message(CEPH_LOCK_FCNTL, op, file, 115 ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
115 CEPH_LOCK_UNLOCK, 0, fl); 116 CEPH_LOCK_UNLOCK, 0, fl);
116 dout("got %d on posix_lock_file, undid lock", err); 117 dout("got %d on posix_lock_file, undid lock",
118 err);
117 } 119 }
118 } 120 }
119 121
120 } else { 122 } else if (err == -ERESTARTSYS) {
121 dout("mds returned error code %d", err); 123 dout("undoing lock\n");
124 ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
125 CEPH_LOCK_UNLOCK, 0, fl);
122 } 126 }
123 return err; 127 return err;
124} 128}
@@ -155,8 +159,11 @@ int ceph_flock(struct file *file, int cmd, struct file_lock *fl)
155 file, CEPH_LOCK_UNLOCK, 0, fl); 159 file, CEPH_LOCK_UNLOCK, 0, fl);
156 dout("got %d on flock_lock_file_wait, undid lock", err); 160 dout("got %d on flock_lock_file_wait, undid lock", err);
157 } 161 }
158 } else { 162 } else if (err == -ERESTARTSYS) {
159 dout("mds error code %d", err); 163 dout("undoing lock\n");
164 ceph_lock_message(CEPH_LOCK_FLOCK,
165 CEPH_MDS_OP_SETFILELOCK,
166 file, CEPH_LOCK_UNLOCK, 0, fl);
160 } 167 }
161 return err; 168 return err;
162} 169}
diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c
index 24067d68a554..54b14de2e729 100644
--- a/fs/ceph/snap.c
+++ b/fs/ceph/snap.c
@@ -722,7 +722,7 @@ static void flush_snaps(struct ceph_mds_client *mdsc)
722 ci = list_first_entry(&mdsc->snap_flush_list, 722 ci = list_first_entry(&mdsc->snap_flush_list,
723 struct ceph_inode_info, i_snap_flush_item); 723 struct ceph_inode_info, i_snap_flush_item);
724 inode = &ci->vfs_inode; 724 inode = &ci->vfs_inode;
725 igrab(inode); 725 ihold(inode);
726 spin_unlock(&mdsc->snap_flush_lock); 726 spin_unlock(&mdsc->snap_flush_lock);
727 spin_lock(&inode->i_lock); 727 spin_lock(&inode->i_lock);
728 __ceph_flush_snaps(ci, &session, 0); 728 __ceph_flush_snaps(ci, &session, 0);
diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
index f2b628696180..f42d730f1b66 100644
--- a/fs/ceph/xattr.c
+++ b/fs/ceph/xattr.c
@@ -665,7 +665,8 @@ static int ceph_sync_setxattr(struct dentry *dentry, const char *name,
665 err = PTR_ERR(req); 665 err = PTR_ERR(req);
666 goto out; 666 goto out;
667 } 667 }
668 req->r_inode = igrab(inode); 668 req->r_inode = inode;
669 ihold(inode);
669 req->r_inode_drop = CEPH_CAP_XATTR_SHARED; 670 req->r_inode_drop = CEPH_CAP_XATTR_SHARED;
670 req->r_num_caps = 1; 671 req->r_num_caps = 1;
671 req->r_args.setxattr.flags = cpu_to_le32(flags); 672 req->r_args.setxattr.flags = cpu_to_le32(flags);
@@ -795,7 +796,8 @@ static int ceph_send_removexattr(struct dentry *dentry, const char *name)
795 USE_AUTH_MDS); 796 USE_AUTH_MDS);
796 if (IS_ERR(req)) 797 if (IS_ERR(req))
797 return PTR_ERR(req); 798 return PTR_ERR(req);
798 req->r_inode = igrab(inode); 799 req->r_inode = inode;
800 ihold(inode);
799 req->r_inode_drop = CEPH_CAP_XATTR_SHARED; 801 req->r_inode_drop = CEPH_CAP_XATTR_SHARED;
800 req->r_num_caps = 1; 802 req->r_num_caps = 1;
801 req->r_path2 = kstrdup(name, GFP_NOFS); 803 req->r_path2 = kstrdup(name, GFP_NOFS);
diff --git a/fs/cifs/Kconfig b/fs/cifs/Kconfig
index 1cd4c3a1862d..53ed1ad2c112 100644
--- a/fs/cifs/Kconfig
+++ b/fs/cifs/Kconfig
@@ -7,6 +7,7 @@ config CIFS
7 select CRYPTO_MD5 7 select CRYPTO_MD5
8 select CRYPTO_HMAC 8 select CRYPTO_HMAC
9 select CRYPTO_ARC4 9 select CRYPTO_ARC4
10 select CRYPTO_ECB
10 select CRYPTO_DES 11 select CRYPTO_DES
11 help 12 help
12 This is the client VFS module for the Common Internet File System 13 This is the client VFS module for the Common Internet File System
@@ -148,7 +149,7 @@ config CIFS_FSCACHE
148 149
149config CIFS_ACL 150config CIFS_ACL
150 bool "Provide CIFS ACL support (EXPERIMENTAL)" 151 bool "Provide CIFS ACL support (EXPERIMENTAL)"
151 depends on EXPERIMENTAL && CIFS_XATTR 152 depends on EXPERIMENTAL && CIFS_XATTR && KEYS
152 help 153 help
153 Allows to fetch CIFS/NTFS ACL from the server. The DACL blob 154 Allows to fetch CIFS/NTFS ACL from the server. The DACL blob
154 is handed over to the application/caller. 155 is handed over to the application/caller.
diff --git a/fs/cifs/cifsencrypt.c b/fs/cifs/cifsencrypt.c
index dfbd9f1f373d..5a0ee7f2af06 100644
--- a/fs/cifs/cifsencrypt.c
+++ b/fs/cifs/cifsencrypt.c
@@ -184,7 +184,7 @@ int cifs_verify_signature(struct smb_hdr *cifs_pdu,
184 if (cifs_pdu == NULL || server == NULL) 184 if (cifs_pdu == NULL || server == NULL)
185 return -EINVAL; 185 return -EINVAL;
186 186
187 if (cifs_pdu->Command == SMB_COM_NEGOTIATE) 187 if (!server->session_estab)
188 return 0; 188 return 0;
189 189
190 if (cifs_pdu->Command == SMB_COM_LOCKING_ANDX) { 190 if (cifs_pdu->Command == SMB_COM_LOCKING_ANDX) {
diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index 6d88b82537c3..bb659eb73810 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -784,7 +784,7 @@ cifs_parse_mount_options(const char *mountdata, const char *devname,
784 struct smb_vol *vol) 784 struct smb_vol *vol)
785{ 785{
786 char *value, *data, *end; 786 char *value, *data, *end;
787 char *mountdata_copy, *options; 787 char *mountdata_copy = NULL, *options;
788 unsigned int temp_len, i, j; 788 unsigned int temp_len, i, j;
789 char separator[2]; 789 char separator[2];
790 short int override_uid = -1; 790 short int override_uid = -1;
@@ -1391,7 +1391,7 @@ cifs_parse_mount_options(const char *mountdata, const char *devname,
1391 "/proc/fs/cifs/LookupCacheEnabled to 0\n"); 1391 "/proc/fs/cifs/LookupCacheEnabled to 0\n");
1392 } else if (strnicmp(data, "fsc", 3) == 0) { 1392 } else if (strnicmp(data, "fsc", 3) == 0) {
1393#ifndef CONFIG_CIFS_FSCACHE 1393#ifndef CONFIG_CIFS_FSCACHE
1394 cERROR(1, "FS-Cache support needs CONFIG_CIFS_FSCACHE" 1394 cERROR(1, "FS-Cache support needs CONFIG_CIFS_FSCACHE "
1395 "kernel config option set"); 1395 "kernel config option set");
1396 goto cifs_parse_mount_err; 1396 goto cifs_parse_mount_err;
1397#endif 1397#endif
@@ -1976,7 +1976,7 @@ cifs_get_smb_ses(struct TCP_Server_Info *server, struct smb_vol *volume_info)
1976 warned_on_ntlm = true; 1976 warned_on_ntlm = true;
1977 cERROR(1, "default security mechanism requested. The default " 1977 cERROR(1, "default security mechanism requested. The default "
1978 "security mechanism will be upgraded from ntlm to " 1978 "security mechanism will be upgraded from ntlm to "
1979 "ntlmv2 in kernel release 2.6.41"); 1979 "ntlmv2 in kernel release 3.1");
1980 } 1980 }
1981 ses->overrideSecFlg = volume_info->secFlg; 1981 ses->overrideSecFlg = volume_info->secFlg;
1982 1982
diff --git a/fs/dcookies.c b/fs/dcookies.c
index a21cabdbd87b..dda0dc702d1b 100644
--- a/fs/dcookies.c
+++ b/fs/dcookies.c
@@ -178,6 +178,8 @@ SYSCALL_DEFINE(lookup_dcookie)(u64 cookie64, char __user * buf, size_t len)
178 /* FIXME: (deleted) ? */ 178 /* FIXME: (deleted) ? */
179 path = d_path(&dcs->path, kbuf, PAGE_SIZE); 179 path = d_path(&dcs->path, kbuf, PAGE_SIZE);
180 180
181 mutex_unlock(&dcookie_mutex);
182
181 if (IS_ERR(path)) { 183 if (IS_ERR(path)) {
182 err = PTR_ERR(path); 184 err = PTR_ERR(path);
183 goto out_free; 185 goto out_free;
@@ -194,6 +196,7 @@ SYSCALL_DEFINE(lookup_dcookie)(u64 cookie64, char __user * buf, size_t len)
194 196
195out_free: 197out_free:
196 kfree(kbuf); 198 kfree(kbuf);
199 return err;
197out: 200out:
198 mutex_unlock(&dcookie_mutex); 201 mutex_unlock(&dcookie_mutex);
199 return err; 202 return err;
diff --git a/fs/exec.c b/fs/exec.c
index ea5f748906a8..97e0d52d72fd 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -1093,6 +1093,7 @@ int flush_old_exec(struct linux_binprm * bprm)
1093 1093
1094 bprm->mm = NULL; /* We're using it now */ 1094 bprm->mm = NULL; /* We're using it now */
1095 1095
1096 set_fs(USER_DS);
1096 current->flags &= ~(PF_RANDOMIZE | PF_KTHREAD); 1097 current->flags &= ~(PF_RANDOMIZE | PF_KTHREAD);
1097 flush_thread(); 1098 flush_thread();
1098 current->personality &= ~bprm->per_clear; 1099 current->personality &= ~bprm->per_clear;
@@ -1357,10 +1358,6 @@ int search_binary_handler(struct linux_binprm *bprm,struct pt_regs *regs)
1357 if (retval) 1358 if (retval)
1358 return retval; 1359 return retval;
1359 1360
1360 /* kernel module loader fixup */
1361 /* so we don't try to load run modprobe in kernel space. */
1362 set_fs(USER_DS);
1363
1364 retval = audit_bprm(bprm); 1361 retval = audit_bprm(bprm);
1365 if (retval) 1362 if (retval)
1366 return retval; 1363 return retval;
diff --git a/fs/fat/file.c b/fs/fat/file.c
index 7257752b6d5d..7018e1d8902d 100644
--- a/fs/fat/file.c
+++ b/fs/fat/file.c
@@ -102,7 +102,7 @@ static int fat_ioctl_set_attributes(struct file *file, u32 __user *user_attr)
102 if (attr & ATTR_SYS) 102 if (attr & ATTR_SYS)
103 inode->i_flags |= S_IMMUTABLE; 103 inode->i_flags |= S_IMMUTABLE;
104 else 104 else
105 inode->i_flags &= S_IMMUTABLE; 105 inode->i_flags &= ~S_IMMUTABLE;
106 } 106 }
107 107
108 fat_save_attrs(inode, attr); 108 fat_save_attrs(inode, attr);
diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index cc6ec4b2f0ff..38f84cd48b67 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -921,6 +921,8 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
921 if (sb->s_flags & MS_MANDLOCK) 921 if (sb->s_flags & MS_MANDLOCK)
922 goto err; 922 goto err;
923 923
924 sb->s_flags &= ~MS_NOSEC;
925
924 if (!parse_fuse_opt((char *) data, &d, is_bdev)) 926 if (!parse_fuse_opt((char *) data, &d, is_bdev))
925 goto err; 927 goto err;
926 928
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 2792a790e50b..1c1336e7b3b2 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -663,14 +663,19 @@ static void glock_work_func(struct work_struct *work)
663 drop_ref = 1; 663 drop_ref = 1;
664 } 664 }
665 spin_lock(&gl->gl_spin); 665 spin_lock(&gl->gl_spin);
666 if (test_and_clear_bit(GLF_PENDING_DEMOTE, &gl->gl_flags) && 666 if (test_bit(GLF_PENDING_DEMOTE, &gl->gl_flags) &&
667 gl->gl_state != LM_ST_UNLOCKED && 667 gl->gl_state != LM_ST_UNLOCKED &&
668 gl->gl_demote_state != LM_ST_EXCLUSIVE) { 668 gl->gl_demote_state != LM_ST_EXCLUSIVE) {
669 unsigned long holdtime, now = jiffies; 669 unsigned long holdtime, now = jiffies;
670
670 holdtime = gl->gl_tchange + gl->gl_ops->go_min_hold_time; 671 holdtime = gl->gl_tchange + gl->gl_ops->go_min_hold_time;
671 if (time_before(now, holdtime)) 672 if (time_before(now, holdtime))
672 delay = holdtime - now; 673 delay = holdtime - now;
673 set_bit(delay ? GLF_PENDING_DEMOTE : GLF_DEMOTE, &gl->gl_flags); 674
675 if (!delay) {
676 clear_bit(GLF_PENDING_DEMOTE, &gl->gl_flags);
677 set_bit(GLF_DEMOTE, &gl->gl_flags);
678 }
674 } 679 }
675 run_queue(gl, 0); 680 run_queue(gl, 0);
676 spin_unlock(&gl->gl_spin); 681 spin_unlock(&gl->gl_spin);
diff --git a/fs/jfs/jfs_logmgr.c b/fs/jfs/jfs_logmgr.c
index 278e3fb40b71..583636f745e5 100644
--- a/fs/jfs/jfs_logmgr.c
+++ b/fs/jfs/jfs_logmgr.c
@@ -1123,7 +1123,7 @@ int lmLogOpen(struct super_block *sb)
1123 bdev = blkdev_get_by_dev(sbi->logdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL, 1123 bdev = blkdev_get_by_dev(sbi->logdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL,
1124 log); 1124 log);
1125 if (IS_ERR(bdev)) { 1125 if (IS_ERR(bdev)) {
1126 rc = -PTR_ERR(bdev); 1126 rc = PTR_ERR(bdev);
1127 goto free; 1127 goto free;
1128 } 1128 }
1129 1129
diff --git a/fs/namei.c b/fs/namei.c
index e2e4e8d032ee..9802345df5e7 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -2624,6 +2624,10 @@ static long do_rmdir(int dfd, const char __user *pathname)
2624 error = PTR_ERR(dentry); 2624 error = PTR_ERR(dentry);
2625 if (IS_ERR(dentry)) 2625 if (IS_ERR(dentry))
2626 goto exit2; 2626 goto exit2;
2627 if (!dentry->d_inode) {
2628 error = -ENOENT;
2629 goto exit3;
2630 }
2627 error = mnt_want_write(nd.path.mnt); 2631 error = mnt_want_write(nd.path.mnt);
2628 if (error) 2632 if (error)
2629 goto exit3; 2633 goto exit3;
@@ -2709,11 +2713,10 @@ static long do_unlinkat(int dfd, const char __user *pathname)
2709 error = PTR_ERR(dentry); 2713 error = PTR_ERR(dentry);
2710 if (!IS_ERR(dentry)) { 2714 if (!IS_ERR(dentry)) {
2711 /* Why not before? Because we want correct error value */ 2715 /* Why not before? Because we want correct error value */
2712 if (nd.last.name[nd.last.len])
2713 goto slashes;
2714 inode = dentry->d_inode; 2716 inode = dentry->d_inode;
2715 if (inode) 2717 if (nd.last.name[nd.last.len] || !inode)
2716 ihold(inode); 2718 goto slashes;
2719 ihold(inode);
2717 error = mnt_want_write(nd.path.mnt); 2720 error = mnt_want_write(nd.path.mnt);
2718 if (error) 2721 if (error)
2719 goto exit2; 2722 goto exit2;
diff --git a/fs/nilfs2/btree.c b/fs/nilfs2/btree.c
index 7eafe468a29c..b2e3ff347620 100644
--- a/fs/nilfs2/btree.c
+++ b/fs/nilfs2/btree.c
@@ -1346,6 +1346,11 @@ static void nilfs_btree_shrink(struct nilfs_bmap *btree,
1346 path[level].bp_bh = NULL; 1346 path[level].bp_bh = NULL;
1347} 1347}
1348 1348
1349static void nilfs_btree_nop(struct nilfs_bmap *btree,
1350 struct nilfs_btree_path *path,
1351 int level, __u64 *keyp, __u64 *ptrp)
1352{
1353}
1349 1354
1350static int nilfs_btree_prepare_delete(struct nilfs_bmap *btree, 1355static int nilfs_btree_prepare_delete(struct nilfs_bmap *btree,
1351 struct nilfs_btree_path *path, 1356 struct nilfs_btree_path *path,
@@ -1356,20 +1361,19 @@ static int nilfs_btree_prepare_delete(struct nilfs_bmap *btree,
1356 struct buffer_head *bh; 1361 struct buffer_head *bh;
1357 struct nilfs_btree_node *node, *parent, *sib; 1362 struct nilfs_btree_node *node, *parent, *sib;
1358 __u64 sibptr; 1363 __u64 sibptr;
1359 int pindex, level, ncmin, ncmax, ncblk, ret; 1364 int pindex, dindex, level, ncmin, ncmax, ncblk, ret;
1360 1365
1361 ret = 0; 1366 ret = 0;
1362 stats->bs_nblocks = 0; 1367 stats->bs_nblocks = 0;
1363 ncmin = NILFS_BTREE_NODE_NCHILDREN_MIN(nilfs_btree_node_size(btree)); 1368 ncmin = NILFS_BTREE_NODE_NCHILDREN_MIN(nilfs_btree_node_size(btree));
1364 ncblk = nilfs_btree_nchildren_per_block(btree); 1369 ncblk = nilfs_btree_nchildren_per_block(btree);
1365 1370
1366 for (level = NILFS_BTREE_LEVEL_NODE_MIN; 1371 for (level = NILFS_BTREE_LEVEL_NODE_MIN, dindex = path[level].bp_index;
1367 level < nilfs_btree_height(btree) - 1; 1372 level < nilfs_btree_height(btree) - 1;
1368 level++) { 1373 level++) {
1369 node = nilfs_btree_get_nonroot_node(path, level); 1374 node = nilfs_btree_get_nonroot_node(path, level);
1370 path[level].bp_oldreq.bpr_ptr = 1375 path[level].bp_oldreq.bpr_ptr =
1371 nilfs_btree_node_get_ptr(node, path[level].bp_index, 1376 nilfs_btree_node_get_ptr(node, dindex, ncblk);
1372 ncblk);
1373 ret = nilfs_bmap_prepare_end_ptr(btree, 1377 ret = nilfs_bmap_prepare_end_ptr(btree,
1374 &path[level].bp_oldreq, dat); 1378 &path[level].bp_oldreq, dat);
1375 if (ret < 0) 1379 if (ret < 0)
@@ -1383,6 +1387,7 @@ static int nilfs_btree_prepare_delete(struct nilfs_bmap *btree,
1383 1387
1384 parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax); 1388 parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax);
1385 pindex = path[level + 1].bp_index; 1389 pindex = path[level + 1].bp_index;
1390 dindex = pindex;
1386 1391
1387 if (pindex > 0) { 1392 if (pindex > 0) {
1388 /* left sibling */ 1393 /* left sibling */
@@ -1421,6 +1426,14 @@ static int nilfs_btree_prepare_delete(struct nilfs_bmap *btree,
1421 path[level].bp_sib_bh = bh; 1426 path[level].bp_sib_bh = bh;
1422 path[level].bp_op = nilfs_btree_concat_right; 1427 path[level].bp_op = nilfs_btree_concat_right;
1423 stats->bs_nblocks++; 1428 stats->bs_nblocks++;
1429 /*
1430 * When merging right sibling node
1431 * into the current node, pointer to
1432 * the right sibling node must be
1433 * terminated instead. The adjustment
1434 * below is required for that.
1435 */
1436 dindex = pindex + 1;
1424 /* continue; */ 1437 /* continue; */
1425 } 1438 }
1426 } else { 1439 } else {
@@ -1431,29 +1444,31 @@ static int nilfs_btree_prepare_delete(struct nilfs_bmap *btree,
1431 NILFS_BTREE_ROOT_NCHILDREN_MAX) { 1444 NILFS_BTREE_ROOT_NCHILDREN_MAX) {
1432 path[level].bp_op = nilfs_btree_shrink; 1445 path[level].bp_op = nilfs_btree_shrink;
1433 stats->bs_nblocks += 2; 1446 stats->bs_nblocks += 2;
1447 level++;
1448 path[level].bp_op = nilfs_btree_nop;
1449 goto shrink_root_child;
1434 } else { 1450 } else {
1435 path[level].bp_op = nilfs_btree_do_delete; 1451 path[level].bp_op = nilfs_btree_do_delete;
1436 stats->bs_nblocks++; 1452 stats->bs_nblocks++;
1453 goto out;
1437 } 1454 }
1438
1439 goto out;
1440
1441 } 1455 }
1442 } 1456 }
1443 1457
1458 /* child of the root node is deleted */
1459 path[level].bp_op = nilfs_btree_do_delete;
1460 stats->bs_nblocks++;
1461
1462shrink_root_child:
1444 node = nilfs_btree_get_root(btree); 1463 node = nilfs_btree_get_root(btree);
1445 path[level].bp_oldreq.bpr_ptr = 1464 path[level].bp_oldreq.bpr_ptr =
1446 nilfs_btree_node_get_ptr(node, path[level].bp_index, 1465 nilfs_btree_node_get_ptr(node, dindex,
1447 NILFS_BTREE_ROOT_NCHILDREN_MAX); 1466 NILFS_BTREE_ROOT_NCHILDREN_MAX);
1448 1467
1449 ret = nilfs_bmap_prepare_end_ptr(btree, &path[level].bp_oldreq, dat); 1468 ret = nilfs_bmap_prepare_end_ptr(btree, &path[level].bp_oldreq, dat);
1450 if (ret < 0) 1469 if (ret < 0)
1451 goto err_out_child_node; 1470 goto err_out_child_node;
1452 1471
1453 /* child of the root node is deleted */
1454 path[level].bp_op = nilfs_btree_do_delete;
1455 stats->bs_nblocks++;
1456
1457 /* success */ 1472 /* success */
1458 out: 1473 out:
1459 *levelp = level; 1474 *levelp = level;
diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
index 141646e88fb5..bb24ab6c282f 100644
--- a/fs/nilfs2/segment.c
+++ b/fs/nilfs2/segment.c
@@ -2573,7 +2573,7 @@ static struct nilfs_sc_info *nilfs_segctor_new(struct super_block *sb,
2573 sci->sc_watermark = NILFS_SC_DEFAULT_WATERMARK; 2573 sci->sc_watermark = NILFS_SC_DEFAULT_WATERMARK;
2574 2574
2575 if (nilfs->ns_interval) 2575 if (nilfs->ns_interval)
2576 sci->sc_interval = nilfs->ns_interval; 2576 sci->sc_interval = HZ * nilfs->ns_interval;
2577 if (nilfs->ns_watermark) 2577 if (nilfs->ns_watermark)
2578 sci->sc_watermark = nilfs->ns_watermark; 2578 sci->sc_watermark = nilfs->ns_watermark;
2579 return sci; 2579 return sci;
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index cdbaf5e97308..56f61027236b 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -1072,7 +1072,7 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent)
1072 1072
1073 sb->s_magic = OCFS2_SUPER_MAGIC; 1073 sb->s_magic = OCFS2_SUPER_MAGIC;
1074 1074
1075 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) | 1075 sb->s_flags = (sb->s_flags & ~(MS_POSIXACL | MS_NOSEC)) |
1076 ((osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0); 1076 ((osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
1077 1077
1078 /* Hard readonly mode only if: bdev_read_only, MS_RDONLY, 1078 /* Hard readonly mode only if: bdev_read_only, MS_RDONLY,
diff --git a/fs/super.c b/fs/super.c
index c75593953c52..ab3d672db0de 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -822,7 +822,7 @@ struct dentry *mount_bdev(struct file_system_type *fs_type,
822 } else { 822 } else {
823 char b[BDEVNAME_SIZE]; 823 char b[BDEVNAME_SIZE];
824 824
825 s->s_flags = flags; 825 s->s_flags = flags | MS_NOSEC;
826 s->s_mode = mode; 826 s->s_mode = mode;
827 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id)); 827 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
828 sb_set_blocksize(s, block_size(bdev)); 828 sb_set_blocksize(s, block_size(bdev));