aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorThomas Gleixner <tglx@linutronix.de>2016-01-12 05:01:12 -0500
committerThomas Gleixner <tglx@linutronix.de>2016-01-12 05:01:12 -0500
commit1f16f116b01c110db20ab808562c8b8bc3ee3d6e (patch)
tree44db563f64cf5f8d62af8f99a61e2b248c44ea3a /fs
parent03724ac3d48f8f0e3caf1d30fa134f8fd96c94e2 (diff)
parentf9eccf24615672896dc13251410c3f2f33a14f95 (diff)
Merge branches 'clockevents/4.4-fixes' and 'clockevents/4.5-fixes' of http://git.linaro.org/people/daniel.lezcano/linux into timers/urgent
Pull in fixes from Daniel Lezcano: - Fix the vt8500 timer leading to a system lock up when dealing with too small delta (Roman Volkov) - Select the CLKSRC_MMIO when the fsl_ftm_timer is enabled with COMPILE_TEST (Daniel Lezcano) - Prevent to compile timers using the 'iomem' API when the architecture has not HAS_IOMEM set (Richard Weinberger)
Diffstat (limited to 'fs')
-rw-r--r--fs/9p/vfs_inode.c4
-rw-r--r--fs/block_dev.c9
-rw-r--r--fs/btrfs/backref.c2
-rw-r--r--fs/btrfs/ctree.h4
-rw-r--r--fs/btrfs/extent-tree.c133
-rw-r--r--fs/btrfs/file.c28
-rw-r--r--fs/btrfs/free-space-cache.c10
-rw-r--r--fs/btrfs/inode.c24
-rw-r--r--fs/btrfs/qgroup.c5
-rw-r--r--fs/btrfs/scrub.c62
-rw-r--r--fs/btrfs/tests/free-space-tests.c4
-rw-r--r--fs/btrfs/transaction.c33
-rw-r--r--fs/btrfs/transaction.h6
-rw-r--r--fs/btrfs/volumes.c16
-rw-r--r--fs/btrfs/volumes.h2
-rw-r--r--fs/cifs/inode.c6
-rw-r--r--fs/direct-io.c11
-rw-r--r--fs/dlm/lowcomms.c4
-rw-r--r--fs/exofs/inode.c5
-rw-r--r--fs/ext4/crypto.c2
-rw-r--r--fs/ext4/ext4.h51
-rw-r--r--fs/ext4/symlink.c2
-rw-r--r--fs/ext4/sysfs.c2
-rw-r--r--fs/fuse/cuse.c2
-rw-r--r--fs/fuse/file.c2
-rw-r--r--fs/jbd2/transaction.c12
-rw-r--r--fs/namei.c1
-rw-r--r--fs/nfs/inode.c17
-rw-r--r--fs/nfs/internal.h2
-rw-r--r--fs/nfs/nfs42proc.c3
-rw-r--r--fs/nfs/nfs4client.c2
-rw-r--r--fs/nfs/nfs4file.c59
-rw-r--r--fs/nfs/nfs4proc.c2
-rw-r--r--fs/nfs/nfs4xdr.c1
-rw-r--r--fs/nfs/objlayout/objio_osd.c5
-rw-r--r--fs/nfs/pagelist.c2
-rw-r--r--fs/nfs/pnfs.c60
-rw-r--r--fs/nfsd/nfs4layouts.c2
-rw-r--r--fs/ocfs2/namei.c4
-rw-r--r--fs/overlayfs/copy_up.c23
-rw-r--r--fs/overlayfs/inode.c19
-rw-r--r--fs/overlayfs/overlayfs.h3
-rw-r--r--fs/proc/base.c1
-rw-r--r--fs/splice.c8
-rw-r--r--fs/sysv/inode.c11
45 files changed, 439 insertions, 227 deletions
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 699941e90667..511078586fa1 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -451,9 +451,9 @@ void v9fs_evict_inode(struct inode *inode)
451{ 451{
452 struct v9fs_inode *v9inode = V9FS_I(inode); 452 struct v9fs_inode *v9inode = V9FS_I(inode);
453 453
454 truncate_inode_pages_final(inode->i_mapping); 454 truncate_inode_pages_final(&inode->i_data);
455 clear_inode(inode); 455 clear_inode(inode);
456 filemap_fdatawrite(inode->i_mapping); 456 filemap_fdatawrite(&inode->i_data);
457 457
458 v9fs_cache_inode_put_cookie(inode); 458 v9fs_cache_inode_put_cookie(inode);
459 /* clunk the fid stashed in writeback_fid */ 459 /* clunk the fid stashed in writeback_fid */
diff --git a/fs/block_dev.c b/fs/block_dev.c
index c25639e907bd..44d4a1e9244e 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -1523,11 +1523,14 @@ static void __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part)
1523 WARN_ON_ONCE(bdev->bd_holders); 1523 WARN_ON_ONCE(bdev->bd_holders);
1524 sync_blockdev(bdev); 1524 sync_blockdev(bdev);
1525 kill_bdev(bdev); 1525 kill_bdev(bdev);
1526
1527 bdev_write_inode(bdev);
1526 /* 1528 /*
1527 * ->release can cause the queue to disappear, so flush all 1529 * Detaching bdev inode from its wb in __destroy_inode()
1528 * dirty data before. 1530 * is too late: the queue which embeds its bdi (along with
1531 * root wb) can be gone as soon as we put_disk() below.
1529 */ 1532 */
1530 bdev_write_inode(bdev); 1533 inode_detach_wb(bdev->bd_inode);
1531 } 1534 }
1532 if (bdev->bd_contains == bdev) { 1535 if (bdev->bd_contains == bdev) {
1533 if (disk->fops->release) 1536 if (disk->fops->release)
diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index 6dcdb2ec9211..d453d62ab0c6 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -355,7 +355,7 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
355 355
356 index = srcu_read_lock(&fs_info->subvol_srcu); 356 index = srcu_read_lock(&fs_info->subvol_srcu);
357 357
358 root = btrfs_read_fs_root_no_name(fs_info, &root_key); 358 root = btrfs_get_fs_root(fs_info, &root_key, false);
359 if (IS_ERR(root)) { 359 if (IS_ERR(root)) {
360 srcu_read_unlock(&fs_info->subvol_srcu, index); 360 srcu_read_unlock(&fs_info->subvol_srcu, index);
361 ret = PTR_ERR(root); 361 ret = PTR_ERR(root);
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 8c58191249cc..35489e7129a7 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -3416,6 +3416,7 @@ int btrfs_cross_ref_exist(struct btrfs_trans_handle *trans,
3416struct btrfs_block_group_cache *btrfs_lookup_block_group( 3416struct btrfs_block_group_cache *btrfs_lookup_block_group(
3417 struct btrfs_fs_info *info, 3417 struct btrfs_fs_info *info,
3418 u64 bytenr); 3418 u64 bytenr);
3419void btrfs_get_block_group(struct btrfs_block_group_cache *cache);
3419void btrfs_put_block_group(struct btrfs_block_group_cache *cache); 3420void btrfs_put_block_group(struct btrfs_block_group_cache *cache);
3420int get_block_group_index(struct btrfs_block_group_cache *cache); 3421int get_block_group_index(struct btrfs_block_group_cache *cache);
3421struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans, 3422struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
@@ -3479,6 +3480,9 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
3479 struct btrfs_root *root, u64 bytes_used, 3480 struct btrfs_root *root, u64 bytes_used,
3480 u64 type, u64 chunk_objectid, u64 chunk_offset, 3481 u64 type, u64 chunk_objectid, u64 chunk_offset,
3481 u64 size); 3482 u64 size);
3483struct btrfs_trans_handle *btrfs_start_trans_remove_block_group(
3484 struct btrfs_fs_info *fs_info,
3485 const u64 chunk_offset);
3482int btrfs_remove_block_group(struct btrfs_trans_handle *trans, 3486int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
3483 struct btrfs_root *root, u64 group_start, 3487 struct btrfs_root *root, u64 group_start,
3484 struct extent_map *em); 3488 struct extent_map *em);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index acf3ed11cfb6..c4661db2b72a 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -124,7 +124,7 @@ static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
124 return (cache->flags & bits) == bits; 124 return (cache->flags & bits) == bits;
125} 125}
126 126
127static void btrfs_get_block_group(struct btrfs_block_group_cache *cache) 127void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
128{ 128{
129 atomic_inc(&cache->count); 129 atomic_inc(&cache->count);
130} 130}
@@ -5915,19 +5915,6 @@ static int update_block_group(struct btrfs_trans_handle *trans,
5915 set_extent_dirty(info->pinned_extents, 5915 set_extent_dirty(info->pinned_extents,
5916 bytenr, bytenr + num_bytes - 1, 5916 bytenr, bytenr + num_bytes - 1,
5917 GFP_NOFS | __GFP_NOFAIL); 5917 GFP_NOFS | __GFP_NOFAIL);
5918 /*
5919 * No longer have used bytes in this block group, queue
5920 * it for deletion.
5921 */
5922 if (old_val == 0) {
5923 spin_lock(&info->unused_bgs_lock);
5924 if (list_empty(&cache->bg_list)) {
5925 btrfs_get_block_group(cache);
5926 list_add_tail(&cache->bg_list,
5927 &info->unused_bgs);
5928 }
5929 spin_unlock(&info->unused_bgs_lock);
5930 }
5931 } 5918 }
5932 5919
5933 spin_lock(&trans->transaction->dirty_bgs_lock); 5920 spin_lock(&trans->transaction->dirty_bgs_lock);
@@ -5939,6 +5926,22 @@ static int update_block_group(struct btrfs_trans_handle *trans,
5939 } 5926 }
5940 spin_unlock(&trans->transaction->dirty_bgs_lock); 5927 spin_unlock(&trans->transaction->dirty_bgs_lock);
5941 5928
5929 /*
5930 * No longer have used bytes in this block group, queue it for
5931 * deletion. We do this after adding the block group to the
5932 * dirty list to avoid races between cleaner kthread and space
5933 * cache writeout.
5934 */
5935 if (!alloc && old_val == 0) {
5936 spin_lock(&info->unused_bgs_lock);
5937 if (list_empty(&cache->bg_list)) {
5938 btrfs_get_block_group(cache);
5939 list_add_tail(&cache->bg_list,
5940 &info->unused_bgs);
5941 }
5942 spin_unlock(&info->unused_bgs_lock);
5943 }
5944
5942 btrfs_put_block_group(cache); 5945 btrfs_put_block_group(cache);
5943 total -= num_bytes; 5946 total -= num_bytes;
5944 bytenr += num_bytes; 5947 bytenr += num_bytes;
@@ -8105,21 +8108,47 @@ reada:
8105} 8108}
8106 8109
8107/* 8110/*
8108 * TODO: Modify related function to add related node/leaf to dirty_extent_root, 8111 * These may not be seen by the usual inc/dec ref code so we have to
8109 * for later qgroup accounting. 8112 * add them here.
8110 *
8111 * Current, this function does nothing.
8112 */ 8113 */
8114static int record_one_subtree_extent(struct btrfs_trans_handle *trans,
8115 struct btrfs_root *root, u64 bytenr,
8116 u64 num_bytes)
8117{
8118 struct btrfs_qgroup_extent_record *qrecord;
8119 struct btrfs_delayed_ref_root *delayed_refs;
8120
8121 qrecord = kmalloc(sizeof(*qrecord), GFP_NOFS);
8122 if (!qrecord)
8123 return -ENOMEM;
8124
8125 qrecord->bytenr = bytenr;
8126 qrecord->num_bytes = num_bytes;
8127 qrecord->old_roots = NULL;
8128
8129 delayed_refs = &trans->transaction->delayed_refs;
8130 spin_lock(&delayed_refs->lock);
8131 if (btrfs_qgroup_insert_dirty_extent(delayed_refs, qrecord))
8132 kfree(qrecord);
8133 spin_unlock(&delayed_refs->lock);
8134
8135 return 0;
8136}
8137
8113static int account_leaf_items(struct btrfs_trans_handle *trans, 8138static int account_leaf_items(struct btrfs_trans_handle *trans,
8114 struct btrfs_root *root, 8139 struct btrfs_root *root,
8115 struct extent_buffer *eb) 8140 struct extent_buffer *eb)
8116{ 8141{
8117 int nr = btrfs_header_nritems(eb); 8142 int nr = btrfs_header_nritems(eb);
8118 int i, extent_type; 8143 int i, extent_type, ret;
8119 struct btrfs_key key; 8144 struct btrfs_key key;
8120 struct btrfs_file_extent_item *fi; 8145 struct btrfs_file_extent_item *fi;
8121 u64 bytenr, num_bytes; 8146 u64 bytenr, num_bytes;
8122 8147
8148 /* We can be called directly from walk_up_proc() */
8149 if (!root->fs_info->quota_enabled)
8150 return 0;
8151
8123 for (i = 0; i < nr; i++) { 8152 for (i = 0; i < nr; i++) {
8124 btrfs_item_key_to_cpu(eb, &key, i); 8153 btrfs_item_key_to_cpu(eb, &key, i);
8125 8154
@@ -8138,6 +8167,10 @@ static int account_leaf_items(struct btrfs_trans_handle *trans,
8138 continue; 8167 continue;
8139 8168
8140 num_bytes = btrfs_file_extent_disk_num_bytes(eb, fi); 8169 num_bytes = btrfs_file_extent_disk_num_bytes(eb, fi);
8170
8171 ret = record_one_subtree_extent(trans, root, bytenr, num_bytes);
8172 if (ret)
8173 return ret;
8141 } 8174 }
8142 return 0; 8175 return 0;
8143} 8176}
@@ -8206,8 +8239,6 @@ static int adjust_slots_upwards(struct btrfs_root *root,
8206 8239
8207/* 8240/*
8208 * root_eb is the subtree root and is locked before this function is called. 8241 * root_eb is the subtree root and is locked before this function is called.
8209 * TODO: Modify this function to mark all (including complete shared node)
8210 * to dirty_extent_root to allow it get accounted in qgroup.
8211 */ 8242 */
8212static int account_shared_subtree(struct btrfs_trans_handle *trans, 8243static int account_shared_subtree(struct btrfs_trans_handle *trans,
8213 struct btrfs_root *root, 8244 struct btrfs_root *root,
@@ -8285,6 +8316,11 @@ walk_down:
8285 btrfs_tree_read_lock(eb); 8316 btrfs_tree_read_lock(eb);
8286 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK); 8317 btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
8287 path->locks[level] = BTRFS_READ_LOCK_BLOCKING; 8318 path->locks[level] = BTRFS_READ_LOCK_BLOCKING;
8319
8320 ret = record_one_subtree_extent(trans, root, child_bytenr,
8321 root->nodesize);
8322 if (ret)
8323 goto out;
8288 } 8324 }
8289 8325
8290 if (level == 0) { 8326 if (level == 0) {
@@ -10256,6 +10292,47 @@ out:
10256 return ret; 10292 return ret;
10257} 10293}
10258 10294
10295struct btrfs_trans_handle *
10296btrfs_start_trans_remove_block_group(struct btrfs_fs_info *fs_info,
10297 const u64 chunk_offset)
10298{
10299 struct extent_map_tree *em_tree = &fs_info->mapping_tree.map_tree;
10300 struct extent_map *em;
10301 struct map_lookup *map;
10302 unsigned int num_items;
10303
10304 read_lock(&em_tree->lock);
10305 em = lookup_extent_mapping(em_tree, chunk_offset, 1);
10306 read_unlock(&em_tree->lock);
10307 ASSERT(em && em->start == chunk_offset);
10308
10309 /*
10310 * We need to reserve 3 + N units from the metadata space info in order
10311 * to remove a block group (done at btrfs_remove_chunk() and at
10312 * btrfs_remove_block_group()), which are used for:
10313 *
10314 * 1 unit for adding the free space inode's orphan (located in the tree
10315 * of tree roots).
10316 * 1 unit for deleting the block group item (located in the extent
10317 * tree).
10318 * 1 unit for deleting the free space item (located in tree of tree
10319 * roots).
10320 * N units for deleting N device extent items corresponding to each
10321 * stripe (located in the device tree).
10322 *
10323 * In order to remove a block group we also need to reserve units in the
10324 * system space info in order to update the chunk tree (update one or
10325 * more device items and remove one chunk item), but this is done at
10326 * btrfs_remove_chunk() through a call to check_system_chunk().
10327 */
10328 map = (struct map_lookup *)em->bdev;
10329 num_items = 3 + map->num_stripes;
10330 free_extent_map(em);
10331
10332 return btrfs_start_transaction_fallback_global_rsv(fs_info->extent_root,
10333 num_items, 1);
10334}
10335
10259/* 10336/*
10260 * Process the unused_bgs list and remove any that don't have any allocated 10337 * Process the unused_bgs list and remove any that don't have any allocated
10261 * space inside of them. 10338 * space inside of them.
@@ -10322,8 +10399,8 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
10322 * Want to do this before we do anything else so we can recover 10399 * Want to do this before we do anything else so we can recover
10323 * properly if we fail to join the transaction. 10400 * properly if we fail to join the transaction.
10324 */ 10401 */
10325 /* 1 for btrfs_orphan_reserve_metadata() */ 10402 trans = btrfs_start_trans_remove_block_group(fs_info,
10326 trans = btrfs_start_transaction(root, 1); 10403 block_group->key.objectid);
10327 if (IS_ERR(trans)) { 10404 if (IS_ERR(trans)) {
10328 btrfs_dec_block_group_ro(root, block_group); 10405 btrfs_dec_block_group_ro(root, block_group);
10329 ret = PTR_ERR(trans); 10406 ret = PTR_ERR(trans);
@@ -10403,11 +10480,15 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
10403 * until transaction commit to do the actual discard. 10480 * until transaction commit to do the actual discard.
10404 */ 10481 */
10405 if (trimming) { 10482 if (trimming) {
10406 WARN_ON(!list_empty(&block_group->bg_list)); 10483 spin_lock(&fs_info->unused_bgs_lock);
10407 spin_lock(&trans->transaction->deleted_bgs_lock); 10484 /*
10485 * A concurrent scrub might have added us to the list
10486 * fs_info->unused_bgs, so use a list_move operation
10487 * to add the block group to the deleted_bgs list.
10488 */
10408 list_move(&block_group->bg_list, 10489 list_move(&block_group->bg_list,
10409 &trans->transaction->deleted_bgs); 10490 &trans->transaction->deleted_bgs);
10410 spin_unlock(&trans->transaction->deleted_bgs_lock); 10491 spin_unlock(&fs_info->unused_bgs_lock);
10411 btrfs_get_block_group(block_group); 10492 btrfs_get_block_group(block_group);
10412 } 10493 }
10413end_trans: 10494end_trans:
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 977e715f0bf2..0f09526aa7d9 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1291,7 +1291,8 @@ out:
1291 * on error we return an unlocked page and the error value 1291 * on error we return an unlocked page and the error value
1292 * on success we return a locked page and 0 1292 * on success we return a locked page and 0
1293 */ 1293 */
1294static int prepare_uptodate_page(struct page *page, u64 pos, 1294static int prepare_uptodate_page(struct inode *inode,
1295 struct page *page, u64 pos,
1295 bool force_uptodate) 1296 bool force_uptodate)
1296{ 1297{
1297 int ret = 0; 1298 int ret = 0;
@@ -1306,6 +1307,10 @@ static int prepare_uptodate_page(struct page *page, u64 pos,
1306 unlock_page(page); 1307 unlock_page(page);
1307 return -EIO; 1308 return -EIO;
1308 } 1309 }
1310 if (page->mapping != inode->i_mapping) {
1311 unlock_page(page);
1312 return -EAGAIN;
1313 }
1309 } 1314 }
1310 return 0; 1315 return 0;
1311} 1316}
@@ -1324,6 +1329,7 @@ static noinline int prepare_pages(struct inode *inode, struct page **pages,
1324 int faili; 1329 int faili;
1325 1330
1326 for (i = 0; i < num_pages; i++) { 1331 for (i = 0; i < num_pages; i++) {
1332again:
1327 pages[i] = find_or_create_page(inode->i_mapping, index + i, 1333 pages[i] = find_or_create_page(inode->i_mapping, index + i,
1328 mask | __GFP_WRITE); 1334 mask | __GFP_WRITE);
1329 if (!pages[i]) { 1335 if (!pages[i]) {
@@ -1333,13 +1339,17 @@ static noinline int prepare_pages(struct inode *inode, struct page **pages,
1333 } 1339 }
1334 1340
1335 if (i == 0) 1341 if (i == 0)
1336 err = prepare_uptodate_page(pages[i], pos, 1342 err = prepare_uptodate_page(inode, pages[i], pos,
1337 force_uptodate); 1343 force_uptodate);
1338 if (i == num_pages - 1) 1344 if (!err && i == num_pages - 1)
1339 err = prepare_uptodate_page(pages[i], 1345 err = prepare_uptodate_page(inode, pages[i],
1340 pos + write_bytes, false); 1346 pos + write_bytes, false);
1341 if (err) { 1347 if (err) {
1342 page_cache_release(pages[i]); 1348 page_cache_release(pages[i]);
1349 if (err == -EAGAIN) {
1350 err = 0;
1351 goto again;
1352 }
1343 faili = i - 1; 1353 faili = i - 1;
1344 goto fail; 1354 goto fail;
1345 } 1355 }
@@ -1882,8 +1892,13 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
1882 struct btrfs_log_ctx ctx; 1892 struct btrfs_log_ctx ctx;
1883 int ret = 0; 1893 int ret = 0;
1884 bool full_sync = 0; 1894 bool full_sync = 0;
1885 const u64 len = end - start + 1; 1895 u64 len;
1886 1896
1897 /*
1898 * The range length can be represented by u64, we have to do the typecasts
1899 * to avoid signed overflow if it's [0, LLONG_MAX] eg. from fsync()
1900 */
1901 len = (u64)end - (u64)start + 1;
1887 trace_btrfs_sync_file(file, datasync); 1902 trace_btrfs_sync_file(file, datasync);
1888 1903
1889 /* 1904 /*
@@ -2071,8 +2086,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
2071 } 2086 }
2072 } 2087 }
2073 if (!full_sync) { 2088 if (!full_sync) {
2074 ret = btrfs_wait_ordered_range(inode, start, 2089 ret = btrfs_wait_ordered_range(inode, start, len);
2075 end - start + 1);
2076 if (ret) { 2090 if (ret) {
2077 btrfs_end_transaction(trans, root); 2091 btrfs_end_transaction(trans, root);
2078 goto out; 2092 goto out;
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 85a1f8621b51..cfe99bec49de 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -891,7 +891,7 @@ out:
891 spin_unlock(&block_group->lock); 891 spin_unlock(&block_group->lock);
892 ret = 0; 892 ret = 0;
893 893
894 btrfs_warn(fs_info, "failed to load free space cache for block group %llu, rebuild it now", 894 btrfs_warn(fs_info, "failed to load free space cache for block group %llu, rebuilding it now",
895 block_group->key.objectid); 895 block_group->key.objectid);
896 } 896 }
897 897
@@ -2972,7 +2972,7 @@ setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2972 u64 cont1_bytes, u64 min_bytes) 2972 u64 cont1_bytes, u64 min_bytes)
2973{ 2973{
2974 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl; 2974 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2975 struct btrfs_free_space *entry; 2975 struct btrfs_free_space *entry = NULL;
2976 int ret = -ENOSPC; 2976 int ret = -ENOSPC;
2977 u64 bitmap_offset = offset_to_bitmap(ctl, offset); 2977 u64 bitmap_offset = offset_to_bitmap(ctl, offset);
2978 2978
@@ -2983,8 +2983,10 @@ setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2983 * The bitmap that covers offset won't be in the list unless offset 2983 * The bitmap that covers offset won't be in the list unless offset
2984 * is just its start offset. 2984 * is just its start offset.
2985 */ 2985 */
2986 entry = list_first_entry(bitmaps, struct btrfs_free_space, list); 2986 if (!list_empty(bitmaps))
2987 if (entry->offset != bitmap_offset) { 2987 entry = list_first_entry(bitmaps, struct btrfs_free_space, list);
2988
2989 if (!entry || entry->offset != bitmap_offset) {
2988 entry = tree_search_offset(ctl, bitmap_offset, 1, 0); 2990 entry = tree_search_offset(ctl, bitmap_offset, 1, 0);
2989 if (entry && list_empty(&entry->list)) 2991 if (entry && list_empty(&entry->list))
2990 list_add(&entry->list, bitmaps); 2992 list_add(&entry->list, bitmaps);
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 994490d5fa64..a70c5790f8f5 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4046,9 +4046,7 @@ int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
4046 */ 4046 */
4047static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir) 4047static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir)
4048{ 4048{
4049 struct btrfs_trans_handle *trans;
4050 struct btrfs_root *root = BTRFS_I(dir)->root; 4049 struct btrfs_root *root = BTRFS_I(dir)->root;
4051 int ret;
4052 4050
4053 /* 4051 /*
4054 * 1 for the possible orphan item 4052 * 1 for the possible orphan item
@@ -4057,27 +4055,7 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir)
4057 * 1 for the inode ref 4055 * 1 for the inode ref
4058 * 1 for the inode 4056 * 1 for the inode
4059 */ 4057 */
4060 trans = btrfs_start_transaction(root, 5); 4058 return btrfs_start_transaction_fallback_global_rsv(root, 5, 5);
4061 if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC)
4062 return trans;
4063
4064 if (PTR_ERR(trans) == -ENOSPC) {
4065 u64 num_bytes = btrfs_calc_trans_metadata_size(root, 5);
4066
4067 trans = btrfs_start_transaction(root, 0);
4068 if (IS_ERR(trans))
4069 return trans;
4070 ret = btrfs_cond_migrate_bytes(root->fs_info,
4071 &root->fs_info->trans_block_rsv,
4072 num_bytes, 5);
4073 if (ret) {
4074 btrfs_end_transaction(trans, root);
4075 return ERR_PTR(ret);
4076 }
4077 trans->block_rsv = &root->fs_info->trans_block_rsv;
4078 trans->bytes_reserved = num_bytes;
4079 }
4080 return trans;
4081} 4059}
4082 4060
4083static int btrfs_unlink(struct inode *dir, struct dentry *dentry) 4061static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index 93e12c18ffd7..5279fdae7142 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -993,9 +993,10 @@ int btrfs_quota_disable(struct btrfs_trans_handle *trans,
993 mutex_lock(&fs_info->qgroup_ioctl_lock); 993 mutex_lock(&fs_info->qgroup_ioctl_lock);
994 if (!fs_info->quota_root) 994 if (!fs_info->quota_root)
995 goto out; 995 goto out;
996 spin_lock(&fs_info->qgroup_lock);
997 fs_info->quota_enabled = 0; 996 fs_info->quota_enabled = 0;
998 fs_info->pending_quota_state = 0; 997 fs_info->pending_quota_state = 0;
998 btrfs_qgroup_wait_for_completion(fs_info);
999 spin_lock(&fs_info->qgroup_lock);
999 quota_root = fs_info->quota_root; 1000 quota_root = fs_info->quota_root;
1000 fs_info->quota_root = NULL; 1001 fs_info->quota_root = NULL;
1001 fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_ON; 1002 fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_ON;
@@ -1461,6 +1462,8 @@ struct btrfs_qgroup_extent_record
1461 struct btrfs_qgroup_extent_record *entry; 1462 struct btrfs_qgroup_extent_record *entry;
1462 u64 bytenr = record->bytenr; 1463 u64 bytenr = record->bytenr;
1463 1464
1465 assert_spin_locked(&delayed_refs->lock);
1466
1464 while (*p) { 1467 while (*p) {
1465 parent_node = *p; 1468 parent_node = *p;
1466 entry = rb_entry(parent_node, struct btrfs_qgroup_extent_record, 1469 entry = rb_entry(parent_node, struct btrfs_qgroup_extent_record,
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 2907a77fb1f6..b091d94ceef6 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -3432,7 +3432,9 @@ out:
3432static noinline_for_stack int scrub_chunk(struct scrub_ctx *sctx, 3432static noinline_for_stack int scrub_chunk(struct scrub_ctx *sctx,
3433 struct btrfs_device *scrub_dev, 3433 struct btrfs_device *scrub_dev,
3434 u64 chunk_offset, u64 length, 3434 u64 chunk_offset, u64 length,
3435 u64 dev_offset, int is_dev_replace) 3435 u64 dev_offset,
3436 struct btrfs_block_group_cache *cache,
3437 int is_dev_replace)
3436{ 3438{
3437 struct btrfs_mapping_tree *map_tree = 3439 struct btrfs_mapping_tree *map_tree =
3438 &sctx->dev_root->fs_info->mapping_tree; 3440 &sctx->dev_root->fs_info->mapping_tree;
@@ -3445,8 +3447,18 @@ static noinline_for_stack int scrub_chunk(struct scrub_ctx *sctx,
3445 em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1); 3447 em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
3446 read_unlock(&map_tree->map_tree.lock); 3448 read_unlock(&map_tree->map_tree.lock);
3447 3449
3448 if (!em) 3450 if (!em) {
3449 return -EINVAL; 3451 /*
3452 * Might have been an unused block group deleted by the cleaner
3453 * kthread or relocation.
3454 */
3455 spin_lock(&cache->lock);
3456 if (!cache->removed)
3457 ret = -EINVAL;
3458 spin_unlock(&cache->lock);
3459
3460 return ret;
3461 }
3450 3462
3451 map = (struct map_lookup *)em->bdev; 3463 map = (struct map_lookup *)em->bdev;
3452 if (em->start != chunk_offset) 3464 if (em->start != chunk_offset)
@@ -3483,6 +3495,7 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
3483 u64 length; 3495 u64 length;
3484 u64 chunk_offset; 3496 u64 chunk_offset;
3485 int ret = 0; 3497 int ret = 0;
3498 int ro_set;
3486 int slot; 3499 int slot;
3487 struct extent_buffer *l; 3500 struct extent_buffer *l;
3488 struct btrfs_key key; 3501 struct btrfs_key key;
@@ -3568,7 +3581,21 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
3568 scrub_pause_on(fs_info); 3581 scrub_pause_on(fs_info);
3569 ret = btrfs_inc_block_group_ro(root, cache); 3582 ret = btrfs_inc_block_group_ro(root, cache);
3570 scrub_pause_off(fs_info); 3583 scrub_pause_off(fs_info);
3571 if (ret) { 3584
3585 if (ret == 0) {
3586 ro_set = 1;
3587 } else if (ret == -ENOSPC) {
3588 /*
3589 * btrfs_inc_block_group_ro return -ENOSPC when it
3590 * failed in creating new chunk for metadata.
3591 * It is not a problem for scrub/replace, because
3592 * metadata are always cowed, and our scrub paused
3593 * commit_transactions.
3594 */
3595 ro_set = 0;
3596 } else {
3597 btrfs_warn(fs_info, "failed setting block group ro, ret=%d\n",
3598 ret);
3572 btrfs_put_block_group(cache); 3599 btrfs_put_block_group(cache);
3573 break; 3600 break;
3574 } 3601 }
@@ -3577,7 +3604,7 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
3577 dev_replace->cursor_left = found_key.offset; 3604 dev_replace->cursor_left = found_key.offset;
3578 dev_replace->item_needs_writeback = 1; 3605 dev_replace->item_needs_writeback = 1;
3579 ret = scrub_chunk(sctx, scrub_dev, chunk_offset, length, 3606 ret = scrub_chunk(sctx, scrub_dev, chunk_offset, length,
3580 found_key.offset, is_dev_replace); 3607 found_key.offset, cache, is_dev_replace);
3581 3608
3582 /* 3609 /*
3583 * flush, submit all pending read and write bios, afterwards 3610 * flush, submit all pending read and write bios, afterwards
@@ -3611,7 +3638,30 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
3611 3638
3612 scrub_pause_off(fs_info); 3639 scrub_pause_off(fs_info);
3613 3640
3614 btrfs_dec_block_group_ro(root, cache); 3641 if (ro_set)
3642 btrfs_dec_block_group_ro(root, cache);
3643
3644 /*
3645 * We might have prevented the cleaner kthread from deleting
3646 * this block group if it was already unused because we raced
3647 * and set it to RO mode first. So add it back to the unused
3648 * list, otherwise it might not ever be deleted unless a manual
3649 * balance is triggered or it becomes used and unused again.
3650 */
3651 spin_lock(&cache->lock);
3652 if (!cache->removed && !cache->ro && cache->reserved == 0 &&
3653 btrfs_block_group_used(&cache->item) == 0) {
3654 spin_unlock(&cache->lock);
3655 spin_lock(&fs_info->unused_bgs_lock);
3656 if (list_empty(&cache->bg_list)) {
3657 btrfs_get_block_group(cache);
3658 list_add_tail(&cache->bg_list,
3659 &fs_info->unused_bgs);
3660 }
3661 spin_unlock(&fs_info->unused_bgs_lock);
3662 } else {
3663 spin_unlock(&cache->lock);
3664 }
3615 3665
3616 btrfs_put_block_group(cache); 3666 btrfs_put_block_group(cache);
3617 if (ret) 3667 if (ret)
diff --git a/fs/btrfs/tests/free-space-tests.c b/fs/btrfs/tests/free-space-tests.c
index c8c3d70c31ff..8b72b005bfb9 100644
--- a/fs/btrfs/tests/free-space-tests.c
+++ b/fs/btrfs/tests/free-space-tests.c
@@ -898,8 +898,10 @@ int btrfs_test_free_space_cache(void)
898 } 898 }
899 899
900 root = btrfs_alloc_dummy_root(); 900 root = btrfs_alloc_dummy_root();
901 if (!root) 901 if (IS_ERR(root)) {
902 ret = PTR_ERR(root);
902 goto out; 903 goto out;
904 }
903 905
904 root->fs_info = btrfs_alloc_dummy_fs_info(); 906 root->fs_info = btrfs_alloc_dummy_fs_info();
905 if (!root->fs_info) 907 if (!root->fs_info)
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 418c6a2ad7d8..be8eae80ff65 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -274,7 +274,6 @@ loop:
274 cur_trans->num_dirty_bgs = 0; 274 cur_trans->num_dirty_bgs = 0;
275 spin_lock_init(&cur_trans->dirty_bgs_lock); 275 spin_lock_init(&cur_trans->dirty_bgs_lock);
276 INIT_LIST_HEAD(&cur_trans->deleted_bgs); 276 INIT_LIST_HEAD(&cur_trans->deleted_bgs);
277 spin_lock_init(&cur_trans->deleted_bgs_lock);
278 spin_lock_init(&cur_trans->dropped_roots_lock); 277 spin_lock_init(&cur_trans->dropped_roots_lock);
279 list_add_tail(&cur_trans->list, &fs_info->trans_list); 278 list_add_tail(&cur_trans->list, &fs_info->trans_list);
280 extent_io_tree_init(&cur_trans->dirty_pages, 279 extent_io_tree_init(&cur_trans->dirty_pages,
@@ -592,6 +591,38 @@ struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
592 return start_transaction(root, num_items, TRANS_START, 591 return start_transaction(root, num_items, TRANS_START,
593 BTRFS_RESERVE_FLUSH_ALL); 592 BTRFS_RESERVE_FLUSH_ALL);
594} 593}
594struct btrfs_trans_handle *btrfs_start_transaction_fallback_global_rsv(
595 struct btrfs_root *root,
596 unsigned int num_items,
597 int min_factor)
598{
599 struct btrfs_trans_handle *trans;
600 u64 num_bytes;
601 int ret;
602
603 trans = btrfs_start_transaction(root, num_items);
604 if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC)
605 return trans;
606
607 trans = btrfs_start_transaction(root, 0);
608 if (IS_ERR(trans))
609 return trans;
610
611 num_bytes = btrfs_calc_trans_metadata_size(root, num_items);
612 ret = btrfs_cond_migrate_bytes(root->fs_info,
613 &root->fs_info->trans_block_rsv,
614 num_bytes,
615 min_factor);
616 if (ret) {
617 btrfs_end_transaction(trans, root);
618 return ERR_PTR(ret);
619 }
620
621 trans->block_rsv = &root->fs_info->trans_block_rsv;
622 trans->bytes_reserved = num_bytes;
623
624 return trans;
625}
595 626
596struct btrfs_trans_handle *btrfs_start_transaction_lflush( 627struct btrfs_trans_handle *btrfs_start_transaction_lflush(
597 struct btrfs_root *root, 628 struct btrfs_root *root,
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index b05b2f64d913..64c8221b6165 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -77,8 +77,8 @@ struct btrfs_transaction {
77 */ 77 */
78 struct mutex cache_write_mutex; 78 struct mutex cache_write_mutex;
79 spinlock_t dirty_bgs_lock; 79 spinlock_t dirty_bgs_lock;
80 /* Protected by spin lock fs_info->unused_bgs_lock. */
80 struct list_head deleted_bgs; 81 struct list_head deleted_bgs;
81 spinlock_t deleted_bgs_lock;
82 spinlock_t dropped_roots_lock; 82 spinlock_t dropped_roots_lock;
83 struct btrfs_delayed_ref_root delayed_refs; 83 struct btrfs_delayed_ref_root delayed_refs;
84 int aborted; 84 int aborted;
@@ -185,6 +185,10 @@ int btrfs_end_transaction(struct btrfs_trans_handle *trans,
185 struct btrfs_root *root); 185 struct btrfs_root *root);
186struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root, 186struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
187 unsigned int num_items); 187 unsigned int num_items);
188struct btrfs_trans_handle *btrfs_start_transaction_fallback_global_rsv(
189 struct btrfs_root *root,
190 unsigned int num_items,
191 int min_factor);
188struct btrfs_trans_handle *btrfs_start_transaction_lflush( 192struct btrfs_trans_handle *btrfs_start_transaction_lflush(
189 struct btrfs_root *root, 193 struct btrfs_root *root,
190 unsigned int num_items); 194 unsigned int num_items);
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index a6df8fdc1312..a23399e8e3ab 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1973,8 +1973,7 @@ void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
1973 if (srcdev->writeable) { 1973 if (srcdev->writeable) {
1974 fs_devices->rw_devices--; 1974 fs_devices->rw_devices--;
1975 /* zero out the old super if it is writable */ 1975 /* zero out the old super if it is writable */
1976 btrfs_scratch_superblocks(srcdev->bdev, 1976 btrfs_scratch_superblocks(srcdev->bdev, srcdev->name->str);
1977 rcu_str_deref(srcdev->name));
1978 } 1977 }
1979 1978
1980 if (srcdev->bdev) 1979 if (srcdev->bdev)
@@ -2024,8 +2023,7 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
2024 btrfs_sysfs_rm_device_link(fs_info->fs_devices, tgtdev); 2023 btrfs_sysfs_rm_device_link(fs_info->fs_devices, tgtdev);
2025 2024
2026 if (tgtdev->bdev) { 2025 if (tgtdev->bdev) {
2027 btrfs_scratch_superblocks(tgtdev->bdev, 2026 btrfs_scratch_superblocks(tgtdev->bdev, tgtdev->name->str);
2028 rcu_str_deref(tgtdev->name));
2029 fs_info->fs_devices->open_devices--; 2027 fs_info->fs_devices->open_devices--;
2030 } 2028 }
2031 fs_info->fs_devices->num_devices--; 2029 fs_info->fs_devices->num_devices--;
@@ -2853,7 +2851,8 @@ static int btrfs_relocate_chunk(struct btrfs_root *root, u64 chunk_offset)
2853 if (ret) 2851 if (ret)
2854 return ret; 2852 return ret;
2855 2853
2856 trans = btrfs_start_transaction(root, 0); 2854 trans = btrfs_start_trans_remove_block_group(root->fs_info,
2855 chunk_offset);
2857 if (IS_ERR(trans)) { 2856 if (IS_ERR(trans)) {
2858 ret = PTR_ERR(trans); 2857 ret = PTR_ERR(trans);
2859 btrfs_std_error(root->fs_info, ret, NULL); 2858 btrfs_std_error(root->fs_info, ret, NULL);
@@ -3123,7 +3122,7 @@ static int chunk_profiles_filter(u64 chunk_type,
3123 return 1; 3122 return 1;
3124} 3123}
3125 3124
3126static int chunk_usage_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset, 3125static int chunk_usage_range_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset,
3127 struct btrfs_balance_args *bargs) 3126 struct btrfs_balance_args *bargs)
3128{ 3127{
3129 struct btrfs_block_group_cache *cache; 3128 struct btrfs_block_group_cache *cache;
@@ -3156,7 +3155,7 @@ static int chunk_usage_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset,
3156 return ret; 3155 return ret;
3157} 3156}
3158 3157
3159static int chunk_usage_range_filter(struct btrfs_fs_info *fs_info, 3158static int chunk_usage_filter(struct btrfs_fs_info *fs_info,
3160 u64 chunk_offset, struct btrfs_balance_args *bargs) 3159 u64 chunk_offset, struct btrfs_balance_args *bargs)
3161{ 3160{
3162 struct btrfs_block_group_cache *cache; 3161 struct btrfs_block_group_cache *cache;
@@ -3549,12 +3548,11 @@ again:
3549 3548
3550 ret = btrfs_force_chunk_alloc(trans, chunk_root, 3549 ret = btrfs_force_chunk_alloc(trans, chunk_root,
3551 BTRFS_BLOCK_GROUP_DATA); 3550 BTRFS_BLOCK_GROUP_DATA);
3551 btrfs_end_transaction(trans, chunk_root);
3552 if (ret < 0) { 3552 if (ret < 0) {
3553 mutex_unlock(&fs_info->delete_unused_bgs_mutex); 3553 mutex_unlock(&fs_info->delete_unused_bgs_mutex);
3554 goto error; 3554 goto error;
3555 } 3555 }
3556
3557 btrfs_end_transaction(trans, chunk_root);
3558 chunk_reserved = 1; 3556 chunk_reserved = 1;
3559 } 3557 }
3560 3558
diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index ec5712372732..d5c84f6b1353 100644
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -382,7 +382,7 @@ struct map_lookup {
382#define BTRFS_BALANCE_ARGS_LIMIT (1ULL << 5) 382#define BTRFS_BALANCE_ARGS_LIMIT (1ULL << 5)
383#define BTRFS_BALANCE_ARGS_LIMIT_RANGE (1ULL << 6) 383#define BTRFS_BALANCE_ARGS_LIMIT_RANGE (1ULL << 6)
384#define BTRFS_BALANCE_ARGS_STRIPES_RANGE (1ULL << 7) 384#define BTRFS_BALANCE_ARGS_STRIPES_RANGE (1ULL << 7)
385#define BTRFS_BALANCE_ARGS_USAGE_RANGE (1ULL << 8) 385#define BTRFS_BALANCE_ARGS_USAGE_RANGE (1ULL << 10)
386 386
387#define BTRFS_BALANCE_ARGS_MASK \ 387#define BTRFS_BALANCE_ARGS_MASK \
388 (BTRFS_BALANCE_ARGS_PROFILES | \ 388 (BTRFS_BALANCE_ARGS_PROFILES | \
diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
index 6b66dd5d1540..a329f5ba35aa 100644
--- a/fs/cifs/inode.c
+++ b/fs/cifs/inode.c
@@ -1831,11 +1831,11 @@ cifs_invalidate_mapping(struct inode *inode)
1831 * @word: long word containing the bit lock 1831 * @word: long word containing the bit lock
1832 */ 1832 */
1833static int 1833static int
1834cifs_wait_bit_killable(struct wait_bit_key *key) 1834cifs_wait_bit_killable(struct wait_bit_key *key, int mode)
1835{ 1835{
1836 if (fatal_signal_pending(current))
1837 return -ERESTARTSYS;
1838 freezable_schedule_unsafe(); 1836 freezable_schedule_unsafe();
1837 if (signal_pending_state(mode, current))
1838 return -ERESTARTSYS;
1839 return 0; 1839 return 0;
1840} 1840}
1841 1841
diff --git a/fs/direct-io.c b/fs/direct-io.c
index cb5337d8c273..602e8441bc0f 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -1169,6 +1169,16 @@ do_blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
1169 } 1169 }
1170 } 1170 }
1171 1171
1172 /* Once we sampled i_size check for reads beyond EOF */
1173 dio->i_size = i_size_read(inode);
1174 if (iov_iter_rw(iter) == READ && offset >= dio->i_size) {
1175 if (dio->flags & DIO_LOCKING)
1176 mutex_unlock(&inode->i_mutex);
1177 kmem_cache_free(dio_cache, dio);
1178 retval = 0;
1179 goto out;
1180 }
1181
1172 /* 1182 /*
1173 * For file extending writes updating i_size before data writeouts 1183 * For file extending writes updating i_size before data writeouts
1174 * complete can expose uninitialized blocks in dumb filesystems. 1184 * complete can expose uninitialized blocks in dumb filesystems.
@@ -1222,7 +1232,6 @@ do_blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
1222 sdio.next_block_for_io = -1; 1232 sdio.next_block_for_io = -1;
1223 1233
1224 dio->iocb = iocb; 1234 dio->iocb = iocb;
1225 dio->i_size = i_size_read(inode);
1226 1235
1227 spin_lock_init(&dio->bio_lock); 1236 spin_lock_init(&dio->bio_lock);
1228 dio->refcount = 1; 1237 dio->refcount = 1;
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 87e9d796cf7d..3a37bd3f9637 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -421,7 +421,7 @@ static void lowcomms_write_space(struct sock *sk)
421 421
422 if (test_and_clear_bit(CF_APP_LIMITED, &con->flags)) { 422 if (test_and_clear_bit(CF_APP_LIMITED, &con->flags)) {
423 con->sock->sk->sk_write_pending--; 423 con->sock->sk->sk_write_pending--;
424 clear_bit(SOCK_ASYNC_NOSPACE, &con->sock->flags); 424 clear_bit(SOCKWQ_ASYNC_NOSPACE, &con->sock->flags);
425 } 425 }
426 426
427 if (!test_and_set_bit(CF_WRITE_PENDING, &con->flags)) 427 if (!test_and_set_bit(CF_WRITE_PENDING, &con->flags))
@@ -1448,7 +1448,7 @@ static void send_to_sock(struct connection *con)
1448 msg_flags); 1448 msg_flags);
1449 if (ret == -EAGAIN || ret == 0) { 1449 if (ret == -EAGAIN || ret == 0) {
1450 if (ret == -EAGAIN && 1450 if (ret == -EAGAIN &&
1451 test_bit(SOCK_ASYNC_NOSPACE, &con->sock->flags) && 1451 test_bit(SOCKWQ_ASYNC_NOSPACE, &con->sock->flags) &&
1452 !test_and_set_bit(CF_APP_LIMITED, &con->flags)) { 1452 !test_and_set_bit(CF_APP_LIMITED, &con->flags)) {
1453 /* Notify TCP that we're limited by the 1453 /* Notify TCP that we're limited by the
1454 * application window size. 1454 * application window size.
diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c
index 73c64daa0f55..60f03b78914e 100644
--- a/fs/exofs/inode.c
+++ b/fs/exofs/inode.c
@@ -592,10 +592,7 @@ static struct page *__r4w_get_page(void *priv, u64 offset, bool *uptodate)
592 } 592 }
593 unlock_page(page); 593 unlock_page(page);
594 } 594 }
595 if (PageDirty(page) || PageWriteback(page)) 595 *uptodate = PageUptodate(page);
596 *uptodate = true;
597 else
598 *uptodate = PageUptodate(page);
599 EXOFS_DBGMSG2("index=0x%lx uptodate=%d\n", index, *uptodate); 596 EXOFS_DBGMSG2("index=0x%lx uptodate=%d\n", index, *uptodate);
600 return page; 597 return page;
601 } else { 598 } else {
diff --git a/fs/ext4/crypto.c b/fs/ext4/crypto.c
index af06830bfc00..1a0835073663 100644
--- a/fs/ext4/crypto.c
+++ b/fs/ext4/crypto.c
@@ -389,7 +389,7 @@ int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex)
389 struct ext4_crypto_ctx *ctx; 389 struct ext4_crypto_ctx *ctx;
390 struct page *ciphertext_page = NULL; 390 struct page *ciphertext_page = NULL;
391 struct bio *bio; 391 struct bio *bio;
392 ext4_lblk_t lblk = ex->ee_block; 392 ext4_lblk_t lblk = le32_to_cpu(ex->ee_block);
393 ext4_fsblk_t pblk = ext4_ext_pblock(ex); 393 ext4_fsblk_t pblk = ext4_ext_pblock(ex);
394 unsigned int len = ext4_ext_get_actual_len(ex); 394 unsigned int len = ext4_ext_get_actual_len(ex);
395 int ret, err = 0; 395 int ret, err = 0;
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 750063f7a50c..cc7ca4e87144 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -26,6 +26,7 @@
26#include <linux/seqlock.h> 26#include <linux/seqlock.h>
27#include <linux/mutex.h> 27#include <linux/mutex.h>
28#include <linux/timer.h> 28#include <linux/timer.h>
29#include <linux/version.h>
29#include <linux/wait.h> 30#include <linux/wait.h>
30#include <linux/blockgroup_lock.h> 31#include <linux/blockgroup_lock.h>
31#include <linux/percpu_counter.h> 32#include <linux/percpu_counter.h>
@@ -727,19 +728,55 @@ struct move_extent {
727 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 728 <= (EXT4_GOOD_OLD_INODE_SIZE + \
728 (einode)->i_extra_isize)) \ 729 (einode)->i_extra_isize)) \
729 730
731/*
732 * We use an encoding that preserves the times for extra epoch "00":
733 *
734 * extra msb of adjust for signed
735 * epoch 32-bit 32-bit tv_sec to
736 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
737 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
738 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
739 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
740 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
741 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
742 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
743 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
744 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
745 *
746 * Note that previous versions of the kernel on 64-bit systems would
747 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
748 * 1970. e2fsck will correct this, assuming that it is run on the
749 * affected filesystem before 2242.
750 */
751
730static inline __le32 ext4_encode_extra_time(struct timespec *time) 752static inline __le32 ext4_encode_extra_time(struct timespec *time)
731{ 753{
732 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 754 u32 extra = sizeof(time->tv_sec) > 4 ?
733 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 755 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
734 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 756 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
735} 757}
736 758
737static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 759static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
738{ 760{
739 if (sizeof(time->tv_sec) > 4) 761 if (unlikely(sizeof(time->tv_sec) > 4 &&
740 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 762 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
741 << 32; 763#if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
742 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 764 /* Handle legacy encoding of pre-1970 dates with epoch
765 * bits 1,1. We assume that by kernel version 4.20,
766 * everyone will have run fsck over the affected
767 * filesystems to correct the problem. (This
768 * backwards compatibility may be removed before this
769 * time, at the discretion of the ext4 developers.)
770 */
771 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
772 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
773 extra_bits = 0;
774 time->tv_sec += extra_bits << 32;
775#else
776 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
777#endif
778 }
779 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
743} 780}
744 781
745#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 782#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
diff --git a/fs/ext4/symlink.c b/fs/ext4/symlink.c
index abe2401ce405..e8e7af62ac95 100644
--- a/fs/ext4/symlink.c
+++ b/fs/ext4/symlink.c
@@ -52,7 +52,7 @@ static const char *ext4_encrypted_follow_link(struct dentry *dentry, void **cook
52 /* Symlink is encrypted */ 52 /* Symlink is encrypted */
53 sd = (struct ext4_encrypted_symlink_data *)caddr; 53 sd = (struct ext4_encrypted_symlink_data *)caddr;
54 cstr.name = sd->encrypted_path; 54 cstr.name = sd->encrypted_path;
55 cstr.len = le32_to_cpu(sd->len); 55 cstr.len = le16_to_cpu(sd->len);
56 if ((cstr.len + 56 if ((cstr.len +
57 sizeof(struct ext4_encrypted_symlink_data) - 1) > 57 sizeof(struct ext4_encrypted_symlink_data) - 1) >
58 max_size) { 58 max_size) {
diff --git a/fs/ext4/sysfs.c b/fs/ext4/sysfs.c
index 1b57c72f4a00..1420a3c614af 100644
--- a/fs/ext4/sysfs.c
+++ b/fs/ext4/sysfs.c
@@ -358,7 +358,7 @@ static int name##_open(struct inode *inode, struct file *file) \
358 return single_open(file, ext4_seq_##name##_show, PDE_DATA(inode)); \ 358 return single_open(file, ext4_seq_##name##_show, PDE_DATA(inode)); \
359} \ 359} \
360\ 360\
361const struct file_operations ext4_seq_##name##_fops = { \ 361static const struct file_operations ext4_seq_##name##_fops = { \
362 .owner = THIS_MODULE, \ 362 .owner = THIS_MODULE, \
363 .open = name##_open, \ 363 .open = name##_open, \
364 .read = seq_read, \ 364 .read = seq_read, \
diff --git a/fs/fuse/cuse.c b/fs/fuse/cuse.c
index eae2c11268bc..8e3ee1936c7e 100644
--- a/fs/fuse/cuse.c
+++ b/fs/fuse/cuse.c
@@ -549,6 +549,8 @@ static int cuse_channel_release(struct inode *inode, struct file *file)
549 unregister_chrdev_region(cc->cdev->dev, 1); 549 unregister_chrdev_region(cc->cdev->dev, 1);
550 cdev_del(cc->cdev); 550 cdev_del(cc->cdev);
551 } 551 }
552 /* Base reference is now owned by "fud" */
553 fuse_conn_put(&cc->fc);
552 554
553 rc = fuse_dev_release(inode, file); /* puts the base reference */ 555 rc = fuse_dev_release(inode, file); /* puts the base reference */
554 556
diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index e0faf8f2c868..570ca4053c80 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -1049,6 +1049,7 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
1049 tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes); 1049 tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
1050 flush_dcache_page(page); 1050 flush_dcache_page(page);
1051 1051
1052 iov_iter_advance(ii, tmp);
1052 if (!tmp) { 1053 if (!tmp) {
1053 unlock_page(page); 1054 unlock_page(page);
1054 page_cache_release(page); 1055 page_cache_release(page);
@@ -1061,7 +1062,6 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
1061 req->page_descs[req->num_pages].length = tmp; 1062 req->page_descs[req->num_pages].length = tmp;
1062 req->num_pages++; 1063 req->num_pages++;
1063 1064
1064 iov_iter_advance(ii, tmp);
1065 count += tmp; 1065 count += tmp;
1066 pos += tmp; 1066 pos += tmp;
1067 offset += tmp; 1067 offset += tmp;
diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index 89463eee6791..ca181e81c765 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -1009,7 +1009,8 @@ out:
1009} 1009}
1010 1010
1011/* Fast check whether buffer is already attached to the required transaction */ 1011/* Fast check whether buffer is already attached to the required transaction */
1012static bool jbd2_write_access_granted(handle_t *handle, struct buffer_head *bh) 1012static bool jbd2_write_access_granted(handle_t *handle, struct buffer_head *bh,
1013 bool undo)
1013{ 1014{
1014 struct journal_head *jh; 1015 struct journal_head *jh;
1015 bool ret = false; 1016 bool ret = false;
@@ -1036,6 +1037,9 @@ static bool jbd2_write_access_granted(handle_t *handle, struct buffer_head *bh)
1036 jh = READ_ONCE(bh->b_private); 1037 jh = READ_ONCE(bh->b_private);
1037 if (!jh) 1038 if (!jh)
1038 goto out; 1039 goto out;
1040 /* For undo access buffer must have data copied */
1041 if (undo && !jh->b_committed_data)
1042 goto out;
1039 if (jh->b_transaction != handle->h_transaction && 1043 if (jh->b_transaction != handle->h_transaction &&
1040 jh->b_next_transaction != handle->h_transaction) 1044 jh->b_next_transaction != handle->h_transaction)
1041 goto out; 1045 goto out;
@@ -1073,7 +1077,7 @@ int jbd2_journal_get_write_access(handle_t *handle, struct buffer_head *bh)
1073 struct journal_head *jh; 1077 struct journal_head *jh;
1074 int rc; 1078 int rc;
1075 1079
1076 if (jbd2_write_access_granted(handle, bh)) 1080 if (jbd2_write_access_granted(handle, bh, false))
1077 return 0; 1081 return 0;
1078 1082
1079 jh = jbd2_journal_add_journal_head(bh); 1083 jh = jbd2_journal_add_journal_head(bh);
@@ -1210,7 +1214,7 @@ int jbd2_journal_get_undo_access(handle_t *handle, struct buffer_head *bh)
1210 char *committed_data = NULL; 1214 char *committed_data = NULL;
1211 1215
1212 JBUFFER_TRACE(jh, "entry"); 1216 JBUFFER_TRACE(jh, "entry");
1213 if (jbd2_write_access_granted(handle, bh)) 1217 if (jbd2_write_access_granted(handle, bh, true))
1214 return 0; 1218 return 0;
1215 1219
1216 jh = jbd2_journal_add_journal_head(bh); 1220 jh = jbd2_journal_add_journal_head(bh);
@@ -2152,6 +2156,7 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh,
2152 2156
2153 if (!buffer_dirty(bh)) { 2157 if (!buffer_dirty(bh)) {
2154 /* bdflush has written it. We can drop it now */ 2158 /* bdflush has written it. We can drop it now */
2159 __jbd2_journal_remove_checkpoint(jh);
2155 goto zap_buffer; 2160 goto zap_buffer;
2156 } 2161 }
2157 2162
@@ -2181,6 +2186,7 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh,
2181 /* The orphan record's transaction has 2186 /* The orphan record's transaction has
2182 * committed. We can cleanse this buffer */ 2187 * committed. We can cleanse this buffer */
2183 clear_buffer_jbddirty(bh); 2188 clear_buffer_jbddirty(bh);
2189 __jbd2_journal_remove_checkpoint(jh);
2184 goto zap_buffer; 2190 goto zap_buffer;
2185 } 2191 }
2186 } 2192 }
diff --git a/fs/namei.c b/fs/namei.c
index d84d7c7515fc..0c3974cd3ecd 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -1996,7 +1996,6 @@ static const char *path_init(struct nameidata *nd, unsigned flags)
1996 nd->last_type = LAST_ROOT; /* if there are only slashes... */ 1996 nd->last_type = LAST_ROOT; /* if there are only slashes... */
1997 nd->flags = flags | LOOKUP_JUMPED | LOOKUP_PARENT; 1997 nd->flags = flags | LOOKUP_JUMPED | LOOKUP_PARENT;
1998 nd->depth = 0; 1998 nd->depth = 0;
1999 nd->total_link_count = 0;
2000 if (flags & LOOKUP_ROOT) { 1999 if (flags & LOOKUP_ROOT) {
2001 struct dentry *root = nd->root.dentry; 2000 struct dentry *root = nd->root.dentry;
2002 struct inode *inode = root->d_inode; 2001 struct inode *inode = root->d_inode;
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 326d9e10d833..c7e8b87da5b2 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -75,11 +75,11 @@ nfs_fattr_to_ino_t(struct nfs_fattr *fattr)
75 * nfs_wait_bit_killable - helper for functions that are sleeping on bit locks 75 * nfs_wait_bit_killable - helper for functions that are sleeping on bit locks
76 * @word: long word containing the bit lock 76 * @word: long word containing the bit lock
77 */ 77 */
78int nfs_wait_bit_killable(struct wait_bit_key *key) 78int nfs_wait_bit_killable(struct wait_bit_key *key, int mode)
79{ 79{
80 if (fatal_signal_pending(current))
81 return -ERESTARTSYS;
82 freezable_schedule_unsafe(); 80 freezable_schedule_unsafe();
81 if (signal_pending_state(mode, current))
82 return -ERESTARTSYS;
83 return 0; 83 return 0;
84} 84}
85EXPORT_SYMBOL_GPL(nfs_wait_bit_killable); 85EXPORT_SYMBOL_GPL(nfs_wait_bit_killable);
@@ -618,7 +618,10 @@ void nfs_setattr_update_inode(struct inode *inode, struct iattr *attr,
618 nfs_inc_stats(inode, NFSIOS_SETATTRTRUNC); 618 nfs_inc_stats(inode, NFSIOS_SETATTRTRUNC);
619 nfs_vmtruncate(inode, attr->ia_size); 619 nfs_vmtruncate(inode, attr->ia_size);
620 } 620 }
621 nfs_update_inode(inode, fattr); 621 if (fattr->valid)
622 nfs_update_inode(inode, fattr);
623 else
624 NFS_I(inode)->cache_validity |= NFS_INO_INVALID_ATTR;
622 spin_unlock(&inode->i_lock); 625 spin_unlock(&inode->i_lock);
623} 626}
624EXPORT_SYMBOL_GPL(nfs_setattr_update_inode); 627EXPORT_SYMBOL_GPL(nfs_setattr_update_inode);
@@ -1824,7 +1827,11 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
1824 if ((long)fattr->gencount - (long)nfsi->attr_gencount > 0) 1827 if ((long)fattr->gencount - (long)nfsi->attr_gencount > 0)
1825 nfsi->attr_gencount = fattr->gencount; 1828 nfsi->attr_gencount = fattr->gencount;
1826 } 1829 }
1827 invalid &= ~NFS_INO_INVALID_ATTR; 1830
1831 /* Don't declare attrcache up to date if there were no attrs! */
1832 if (fattr->valid != 0)
1833 invalid &= ~NFS_INO_INVALID_ATTR;
1834
1828 /* Don't invalidate the data if we were to blame */ 1835 /* Don't invalidate the data if we were to blame */
1829 if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) 1836 if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode)
1830 || S_ISLNK(inode->i_mode))) 1837 || S_ISLNK(inode->i_mode)))
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index 56cfde26fb9c..9dea85f7f918 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -379,7 +379,7 @@ extern int nfs_drop_inode(struct inode *);
379extern void nfs_clear_inode(struct inode *); 379extern void nfs_clear_inode(struct inode *);
380extern void nfs_evict_inode(struct inode *); 380extern void nfs_evict_inode(struct inode *);
381void nfs_zap_acl_cache(struct inode *inode); 381void nfs_zap_acl_cache(struct inode *inode);
382extern int nfs_wait_bit_killable(struct wait_bit_key *key); 382extern int nfs_wait_bit_killable(struct wait_bit_key *key, int mode);
383 383
384/* super.c */ 384/* super.c */
385extern const struct super_operations nfs_sops; 385extern const struct super_operations nfs_sops;
diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 3e92a3cde15d..6b1ce9825430 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -14,7 +14,7 @@
14#include "pnfs.h" 14#include "pnfs.h"
15#include "internal.h" 15#include "internal.h"
16 16
17#define NFSDBG_FACILITY NFSDBG_PNFS 17#define NFSDBG_FACILITY NFSDBG_PROC
18 18
19static int nfs42_set_rw_stateid(nfs4_stateid *dst, struct file *file, 19static int nfs42_set_rw_stateid(nfs4_stateid *dst, struct file *file,
20 fmode_t fmode) 20 fmode_t fmode)
@@ -284,6 +284,7 @@ static int _nfs42_proc_clone(struct rpc_message *msg, struct file *src_f,
284 .dst_fh = NFS_FH(dst_inode), 284 .dst_fh = NFS_FH(dst_inode),
285 .src_offset = src_offset, 285 .src_offset = src_offset,
286 .dst_offset = dst_offset, 286 .dst_offset = dst_offset,
287 .count = count,
287 .dst_bitmask = server->cache_consistency_bitmask, 288 .dst_bitmask = server->cache_consistency_bitmask,
288 }; 289 };
289 struct nfs42_clone_res res = { 290 struct nfs42_clone_res res = {
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index 223bedda64ae..10410e8b5853 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -33,7 +33,7 @@ static int nfs_get_cb_ident_idr(struct nfs_client *clp, int minorversion)
33 return ret; 33 return ret;
34 idr_preload(GFP_KERNEL); 34 idr_preload(GFP_KERNEL);
35 spin_lock(&nn->nfs_client_lock); 35 spin_lock(&nn->nfs_client_lock);
36 ret = idr_alloc(&nn->cb_ident_idr, clp, 0, 0, GFP_NOWAIT); 36 ret = idr_alloc(&nn->cb_ident_idr, clp, 1, 0, GFP_NOWAIT);
37 if (ret >= 0) 37 if (ret >= 0)
38 clp->cl_cb_ident = ret; 38 clp->cl_cb_ident = ret;
39 spin_unlock(&nn->nfs_client_lock); 39 spin_unlock(&nn->nfs_client_lock);
diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
index 4aa571956cd6..db9b5fea5b3e 100644
--- a/fs/nfs/nfs4file.c
+++ b/fs/nfs/nfs4file.c
@@ -7,6 +7,7 @@
7#include <linux/file.h> 7#include <linux/file.h>
8#include <linux/falloc.h> 8#include <linux/falloc.h>
9#include <linux/nfs_fs.h> 9#include <linux/nfs_fs.h>
10#include <uapi/linux/btrfs.h> /* BTRFS_IOC_CLONE/BTRFS_IOC_CLONE_RANGE */
10#include "delegation.h" 11#include "delegation.h"
11#include "internal.h" 12#include "internal.h"
12#include "iostat.h" 13#include "iostat.h"
@@ -203,6 +204,7 @@ nfs42_ioctl_clone(struct file *dst_file, unsigned long srcfd,
203 struct fd src_file; 204 struct fd src_file;
204 struct inode *src_inode; 205 struct inode *src_inode;
205 unsigned int bs = server->clone_blksize; 206 unsigned int bs = server->clone_blksize;
207 bool same_inode = false;
206 int ret; 208 int ret;
207 209
208 /* dst file must be opened for writing */ 210 /* dst file must be opened for writing */
@@ -221,10 +223,8 @@ nfs42_ioctl_clone(struct file *dst_file, unsigned long srcfd,
221 223
222 src_inode = file_inode(src_file.file); 224 src_inode = file_inode(src_file.file);
223 225
224 /* src and dst must be different files */
225 ret = -EINVAL;
226 if (src_inode == dst_inode) 226 if (src_inode == dst_inode)
227 goto out_fput; 227 same_inode = true;
228 228
229 /* src file must be opened for reading */ 229 /* src file must be opened for reading */
230 if (!(src_file.file->f_mode & FMODE_READ)) 230 if (!(src_file.file->f_mode & FMODE_READ))
@@ -249,8 +249,16 @@ nfs42_ioctl_clone(struct file *dst_file, unsigned long srcfd,
249 goto out_fput; 249 goto out_fput;
250 } 250 }
251 251
252 /* verify if ranges are overlapped within the same file */
253 if (same_inode) {
254 if (dst_off + count > src_off && dst_off < src_off + count)
255 goto out_fput;
256 }
257
252 /* XXX: do we lock at all? what if server needs CB_RECALL_LAYOUT? */ 258 /* XXX: do we lock at all? what if server needs CB_RECALL_LAYOUT? */
253 if (dst_inode < src_inode) { 259 if (same_inode) {
260 mutex_lock(&src_inode->i_mutex);
261 } else if (dst_inode < src_inode) {
254 mutex_lock_nested(&dst_inode->i_mutex, I_MUTEX_PARENT); 262 mutex_lock_nested(&dst_inode->i_mutex, I_MUTEX_PARENT);
255 mutex_lock_nested(&src_inode->i_mutex, I_MUTEX_CHILD); 263 mutex_lock_nested(&src_inode->i_mutex, I_MUTEX_CHILD);
256 } else { 264 } else {
@@ -275,7 +283,9 @@ nfs42_ioctl_clone(struct file *dst_file, unsigned long srcfd,
275 truncate_inode_pages_range(&dst_inode->i_data, dst_off, dst_off + count - 1); 283 truncate_inode_pages_range(&dst_inode->i_data, dst_off, dst_off + count - 1);
276 284
277out_unlock: 285out_unlock:
278 if (dst_inode < src_inode) { 286 if (same_inode) {
287 mutex_unlock(&src_inode->i_mutex);
288 } else if (dst_inode < src_inode) {
279 mutex_unlock(&src_inode->i_mutex); 289 mutex_unlock(&src_inode->i_mutex);
280 mutex_unlock(&dst_inode->i_mutex); 290 mutex_unlock(&dst_inode->i_mutex);
281 } else { 291 } else {
@@ -291,46 +301,31 @@ out_drop_write:
291 301
292static long nfs42_ioctl_clone_range(struct file *dst_file, void __user *argp) 302static long nfs42_ioctl_clone_range(struct file *dst_file, void __user *argp)
293{ 303{
294 struct nfs_ioctl_clone_range_args args; 304 struct btrfs_ioctl_clone_range_args args;
295 305
296 if (copy_from_user(&args, argp, sizeof(args))) 306 if (copy_from_user(&args, argp, sizeof(args)))
297 return -EFAULT; 307 return -EFAULT;
298 308
299 return nfs42_ioctl_clone(dst_file, args.src_fd, args.src_off, args.dst_off, args.count); 309 return nfs42_ioctl_clone(dst_file, args.src_fd, args.src_offset,
300} 310 args.dest_offset, args.src_length);
301#else
302static long nfs42_ioctl_clone(struct file *dst_file, unsigned long srcfd,
303 u64 src_off, u64 dst_off, u64 count)
304{
305 return -ENOTTY;
306}
307
308static long nfs42_ioctl_clone_range(struct file *dst_file, void __user *argp)
309{
310 return -ENOTTY;
311} 311}
312#endif /* CONFIG_NFS_V4_2 */
313 312
314long nfs4_ioctl(struct file *file, unsigned int cmd, unsigned long arg) 313long nfs4_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
315{ 314{
316 void __user *argp = (void __user *)arg; 315 void __user *argp = (void __user *)arg;
317 316
318 switch (cmd) { 317 switch (cmd) {
319 case NFS_IOC_CLONE: 318 case BTRFS_IOC_CLONE:
320 return nfs42_ioctl_clone(file, arg, 0, 0, 0); 319 return nfs42_ioctl_clone(file, arg, 0, 0, 0);
321 case NFS_IOC_CLONE_RANGE: 320 case BTRFS_IOC_CLONE_RANGE:
322 return nfs42_ioctl_clone_range(file, argp); 321 return nfs42_ioctl_clone_range(file, argp);
323 } 322 }
324 323
325 return -ENOTTY; 324 return -ENOTTY;
326} 325}
326#endif /* CONFIG_NFS_V4_2 */
327 327
328const struct file_operations nfs4_file_operations = { 328const struct file_operations nfs4_file_operations = {
329#ifdef CONFIG_NFS_V4_2
330 .llseek = nfs4_file_llseek,
331#else
332 .llseek = nfs_file_llseek,
333#endif
334 .read_iter = nfs_file_read, 329 .read_iter = nfs_file_read,
335 .write_iter = nfs_file_write, 330 .write_iter = nfs_file_write,
336 .mmap = nfs_file_mmap, 331 .mmap = nfs_file_mmap,
@@ -342,14 +337,14 @@ const struct file_operations nfs4_file_operations = {
342 .flock = nfs_flock, 337 .flock = nfs_flock,
343 .splice_read = nfs_file_splice_read, 338 .splice_read = nfs_file_splice_read,
344 .splice_write = iter_file_splice_write, 339 .splice_write = iter_file_splice_write,
345#ifdef CONFIG_NFS_V4_2
346 .fallocate = nfs42_fallocate,
347#endif /* CONFIG_NFS_V4_2 */
348 .check_flags = nfs_check_flags, 340 .check_flags = nfs_check_flags,
349 .setlease = simple_nosetlease, 341 .setlease = simple_nosetlease,
350#ifdef CONFIG_COMPAT 342#ifdef CONFIG_NFS_V4_2
343 .llseek = nfs4_file_llseek,
344 .fallocate = nfs42_fallocate,
351 .unlocked_ioctl = nfs4_ioctl, 345 .unlocked_ioctl = nfs4_ioctl,
352#else
353 .compat_ioctl = nfs4_ioctl, 346 .compat_ioctl = nfs4_ioctl,
354#endif /* CONFIG_COMPAT */ 347#else
348 .llseek = nfs_file_llseek,
349#endif
355}; 350};
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 765a03559363..89818036f035 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -7866,7 +7866,7 @@ static void nfs4_layoutget_done(struct rpc_task *task, void *calldata)
7866 spin_unlock(&inode->i_lock); 7866 spin_unlock(&inode->i_lock);
7867 goto out_restart; 7867 goto out_restart;
7868 } 7868 }
7869 if (nfs4_async_handle_error(task, server, state, NULL) == -EAGAIN) 7869 if (nfs4_async_handle_error(task, server, state, &lgp->timeout) == -EAGAIN)
7870 goto out_restart; 7870 goto out_restart;
7871out: 7871out:
7872 dprintk("<-- %s\n", __func__); 7872 dprintk("<-- %s\n", __func__);
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index dfed4f5c8fcc..4e4441216804 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -3615,6 +3615,7 @@ static int decode_attr_fs_locations(struct xdr_stream *xdr, uint32_t *bitmap, st
3615 status = 0; 3615 status = 0;
3616 if (unlikely(!(bitmap[0] & FATTR4_WORD0_FS_LOCATIONS))) 3616 if (unlikely(!(bitmap[0] & FATTR4_WORD0_FS_LOCATIONS)))
3617 goto out; 3617 goto out;
3618 bitmap[0] &= ~FATTR4_WORD0_FS_LOCATIONS;
3618 status = -EIO; 3619 status = -EIO;
3619 /* Ignore borken servers that return unrequested attrs */ 3620 /* Ignore borken servers that return unrequested attrs */
3620 if (unlikely(res == NULL)) 3621 if (unlikely(res == NULL))
diff --git a/fs/nfs/objlayout/objio_osd.c b/fs/nfs/objlayout/objio_osd.c
index 5c0c6b58157f..9aebffb40505 100644
--- a/fs/nfs/objlayout/objio_osd.c
+++ b/fs/nfs/objlayout/objio_osd.c
@@ -476,10 +476,7 @@ static struct page *__r4w_get_page(void *priv, u64 offset, bool *uptodate)
476 } 476 }
477 unlock_page(page); 477 unlock_page(page);
478 } 478 }
479 if (PageDirty(page) || PageWriteback(page)) 479 *uptodate = PageUptodate(page);
480 *uptodate = true;
481 else
482 *uptodate = PageUptodate(page);
483 dprintk("%s: index=0x%lx uptodate=%d\n", __func__, index, *uptodate); 480 dprintk("%s: index=0x%lx uptodate=%d\n", __func__, index, *uptodate);
484 return page; 481 return page;
485} 482}
diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c
index fe3ddd20ff89..452a011ba0d8 100644
--- a/fs/nfs/pagelist.c
+++ b/fs/nfs/pagelist.c
@@ -129,7 +129,7 @@ __nfs_iocounter_wait(struct nfs_io_counter *c)
129 set_bit(NFS_IO_INPROGRESS, &c->flags); 129 set_bit(NFS_IO_INPROGRESS, &c->flags);
130 if (atomic_read(&c->io_count) == 0) 130 if (atomic_read(&c->io_count) == 0)
131 break; 131 break;
132 ret = nfs_wait_bit_killable(&q.key); 132 ret = nfs_wait_bit_killable(&q.key, TASK_KILLABLE);
133 } while (atomic_read(&c->io_count) != 0 && !ret); 133 } while (atomic_read(&c->io_count) != 0 && !ret);
134 finish_wait(wq, &q.wait); 134 finish_wait(wq, &q.wait);
135 return ret; 135 return ret;
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 93496c059837..bec0384499f7 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -872,33 +872,38 @@ send_layoutget(struct pnfs_layout_hdr *lo,
872 872
873 dprintk("--> %s\n", __func__); 873 dprintk("--> %s\n", __func__);
874 874
875 lgp = kzalloc(sizeof(*lgp), gfp_flags); 875 /*
876 if (lgp == NULL) 876 * Synchronously retrieve layout information from server and
877 return NULL; 877 * store in lseg. If we race with a concurrent seqid morphing
878 * op, then re-send the LAYOUTGET.
879 */
880 do {
881 lgp = kzalloc(sizeof(*lgp), gfp_flags);
882 if (lgp == NULL)
883 return NULL;
884
885 i_size = i_size_read(ino);
886
887 lgp->args.minlength = PAGE_CACHE_SIZE;
888 if (lgp->args.minlength > range->length)
889 lgp->args.minlength = range->length;
890 if (range->iomode == IOMODE_READ) {
891 if (range->offset >= i_size)
892 lgp->args.minlength = 0;
893 else if (i_size - range->offset < lgp->args.minlength)
894 lgp->args.minlength = i_size - range->offset;
895 }
896 lgp->args.maxcount = PNFS_LAYOUT_MAXSIZE;
897 lgp->args.range = *range;
898 lgp->args.type = server->pnfs_curr_ld->id;
899 lgp->args.inode = ino;
900 lgp->args.ctx = get_nfs_open_context(ctx);
901 lgp->gfp_flags = gfp_flags;
902 lgp->cred = lo->plh_lc_cred;
878 903
879 i_size = i_size_read(ino); 904 lseg = nfs4_proc_layoutget(lgp, gfp_flags);
905 } while (lseg == ERR_PTR(-EAGAIN));
880 906
881 lgp->args.minlength = PAGE_CACHE_SIZE;
882 if (lgp->args.minlength > range->length)
883 lgp->args.minlength = range->length;
884 if (range->iomode == IOMODE_READ) {
885 if (range->offset >= i_size)
886 lgp->args.minlength = 0;
887 else if (i_size - range->offset < lgp->args.minlength)
888 lgp->args.minlength = i_size - range->offset;
889 }
890 lgp->args.maxcount = PNFS_LAYOUT_MAXSIZE;
891 lgp->args.range = *range;
892 lgp->args.type = server->pnfs_curr_ld->id;
893 lgp->args.inode = ino;
894 lgp->args.ctx = get_nfs_open_context(ctx);
895 lgp->gfp_flags = gfp_flags;
896 lgp->cred = lo->plh_lc_cred;
897
898 /* Synchronously retrieve layout information from server and
899 * store in lseg.
900 */
901 lseg = nfs4_proc_layoutget(lgp, gfp_flags);
902 if (IS_ERR(lseg)) { 907 if (IS_ERR(lseg)) {
903 switch (PTR_ERR(lseg)) { 908 switch (PTR_ERR(lseg)) {
904 case -ENOMEM: 909 case -ENOMEM:
@@ -1461,11 +1466,11 @@ static bool pnfs_within_mdsthreshold(struct nfs_open_context *ctx,
1461} 1466}
1462 1467
1463/* stop waiting if someone clears NFS_LAYOUT_RETRY_LAYOUTGET bit. */ 1468/* stop waiting if someone clears NFS_LAYOUT_RETRY_LAYOUTGET bit. */
1464static int pnfs_layoutget_retry_bit_wait(struct wait_bit_key *key) 1469static int pnfs_layoutget_retry_bit_wait(struct wait_bit_key *key, int mode)
1465{ 1470{
1466 if (!test_bit(NFS_LAYOUT_RETRY_LAYOUTGET, key->flags)) 1471 if (!test_bit(NFS_LAYOUT_RETRY_LAYOUTGET, key->flags))
1467 return 1; 1472 return 1;
1468 return nfs_wait_bit_killable(key); 1473 return nfs_wait_bit_killable(key, mode);
1469} 1474}
1470 1475
1471static bool pnfs_prepare_to_retry_layoutget(struct pnfs_layout_hdr *lo) 1476static bool pnfs_prepare_to_retry_layoutget(struct pnfs_layout_hdr *lo)
@@ -1687,6 +1692,7 @@ pnfs_layout_process(struct nfs4_layoutget *lgp)
1687 /* existing state ID, make sure the sequence number matches. */ 1692 /* existing state ID, make sure the sequence number matches. */
1688 if (pnfs_layout_stateid_blocked(lo, &res->stateid)) { 1693 if (pnfs_layout_stateid_blocked(lo, &res->stateid)) {
1689 dprintk("%s forget reply due to sequence\n", __func__); 1694 dprintk("%s forget reply due to sequence\n", __func__);
1695 status = -EAGAIN;
1690 goto out_forget_reply; 1696 goto out_forget_reply;
1691 } 1697 }
1692 pnfs_set_layout_stateid(lo, &res->stateid, false); 1698 pnfs_set_layout_stateid(lo, &res->stateid, false);
diff --git a/fs/nfsd/nfs4layouts.c b/fs/nfsd/nfs4layouts.c
index 9ffef06b30d5..c9d6c715c0fb 100644
--- a/fs/nfsd/nfs4layouts.c
+++ b/fs/nfsd/nfs4layouts.c
@@ -616,6 +616,7 @@ nfsd4_cb_layout_prepare(struct nfsd4_callback *cb)
616 616
617 mutex_lock(&ls->ls_mutex); 617 mutex_lock(&ls->ls_mutex);
618 nfs4_inc_and_copy_stateid(&ls->ls_recall_sid, &ls->ls_stid); 618 nfs4_inc_and_copy_stateid(&ls->ls_recall_sid, &ls->ls_stid);
619 mutex_unlock(&ls->ls_mutex);
619} 620}
620 621
621static int 622static int
@@ -659,7 +660,6 @@ nfsd4_cb_layout_release(struct nfsd4_callback *cb)
659 660
660 trace_layout_recall_release(&ls->ls_stid.sc_stateid); 661 trace_layout_recall_release(&ls->ls_stid.sc_stateid);
661 662
662 mutex_unlock(&ls->ls_mutex);
663 nfsd4_return_all_layouts(ls, &reaplist); 663 nfsd4_return_all_layouts(ls, &reaplist);
664 nfsd4_free_layouts(&reaplist); 664 nfsd4_free_layouts(&reaplist);
665 nfs4_put_stid(&ls->ls_stid); 665 nfs4_put_stid(&ls->ls_stid);
diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index a03f6f433075..3123408da935 100644
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -367,13 +367,11 @@ static int ocfs2_mknod(struct inode *dir,
367 goto leave; 367 goto leave;
368 } 368 }
369 369
370 status = posix_acl_create(dir, &mode, &default_acl, &acl); 370 status = posix_acl_create(dir, &inode->i_mode, &default_acl, &acl);
371 if (status) { 371 if (status) {
372 mlog_errno(status); 372 mlog_errno(status);
373 goto leave; 373 goto leave;
374 } 374 }
375 /* update inode->i_mode after mask with "umask". */
376 inode->i_mode = mode;
377 375
378 handle = ocfs2_start_trans(osb, ocfs2_mknod_credits(osb->sb, 376 handle = ocfs2_start_trans(osb, ocfs2_mknod_credits(osb->sb,
379 S_ISDIR(mode), 377 S_ISDIR(mode),
diff --git a/fs/overlayfs/copy_up.c b/fs/overlayfs/copy_up.c
index 871fcb67be97..0a8983492d91 100644
--- a/fs/overlayfs/copy_up.c
+++ b/fs/overlayfs/copy_up.c
@@ -195,8 +195,7 @@ int ovl_set_attr(struct dentry *upperdentry, struct kstat *stat)
195 195
196static int ovl_copy_up_locked(struct dentry *workdir, struct dentry *upperdir, 196static int ovl_copy_up_locked(struct dentry *workdir, struct dentry *upperdir,
197 struct dentry *dentry, struct path *lowerpath, 197 struct dentry *dentry, struct path *lowerpath,
198 struct kstat *stat, struct iattr *attr, 198 struct kstat *stat, const char *link)
199 const char *link)
200{ 199{
201 struct inode *wdir = workdir->d_inode; 200 struct inode *wdir = workdir->d_inode;
202 struct inode *udir = upperdir->d_inode; 201 struct inode *udir = upperdir->d_inode;
@@ -240,8 +239,6 @@ static int ovl_copy_up_locked(struct dentry *workdir, struct dentry *upperdir,
240 239
241 mutex_lock(&newdentry->d_inode->i_mutex); 240 mutex_lock(&newdentry->d_inode->i_mutex);
242 err = ovl_set_attr(newdentry, stat); 241 err = ovl_set_attr(newdentry, stat);
243 if (!err && attr)
244 err = notify_change(newdentry, attr, NULL);
245 mutex_unlock(&newdentry->d_inode->i_mutex); 242 mutex_unlock(&newdentry->d_inode->i_mutex);
246 if (err) 243 if (err)
247 goto out_cleanup; 244 goto out_cleanup;
@@ -286,8 +283,7 @@ out_cleanup:
286 * that point the file will have already been copied up anyway. 283 * that point the file will have already been copied up anyway.
287 */ 284 */
288int ovl_copy_up_one(struct dentry *parent, struct dentry *dentry, 285int ovl_copy_up_one(struct dentry *parent, struct dentry *dentry,
289 struct path *lowerpath, struct kstat *stat, 286 struct path *lowerpath, struct kstat *stat)
290 struct iattr *attr)
291{ 287{
292 struct dentry *workdir = ovl_workdir(dentry); 288 struct dentry *workdir = ovl_workdir(dentry);
293 int err; 289 int err;
@@ -345,26 +341,19 @@ int ovl_copy_up_one(struct dentry *parent, struct dentry *dentry,
345 } 341 }
346 upperdentry = ovl_dentry_upper(dentry); 342 upperdentry = ovl_dentry_upper(dentry);
347 if (upperdentry) { 343 if (upperdentry) {
348 unlock_rename(workdir, upperdir); 344 /* Raced with another copy-up? Nothing to do, then... */
349 err = 0; 345 err = 0;
350 /* Raced with another copy-up? Do the setattr here */ 346 goto out_unlock;
351 if (attr) {
352 mutex_lock(&upperdentry->d_inode->i_mutex);
353 err = notify_change(upperdentry, attr, NULL);
354 mutex_unlock(&upperdentry->d_inode->i_mutex);
355 }
356 goto out_put_cred;
357 } 347 }
358 348
359 err = ovl_copy_up_locked(workdir, upperdir, dentry, lowerpath, 349 err = ovl_copy_up_locked(workdir, upperdir, dentry, lowerpath,
360 stat, attr, link); 350 stat, link);
361 if (!err) { 351 if (!err) {
362 /* Restore timestamps on parent (best effort) */ 352 /* Restore timestamps on parent (best effort) */
363 ovl_set_timestamps(upperdir, &pstat); 353 ovl_set_timestamps(upperdir, &pstat);
364 } 354 }
365out_unlock: 355out_unlock:
366 unlock_rename(workdir, upperdir); 356 unlock_rename(workdir, upperdir);
367out_put_cred:
368 revert_creds(old_cred); 357 revert_creds(old_cred);
369 put_cred(override_cred); 358 put_cred(override_cred);
370 359
@@ -406,7 +395,7 @@ int ovl_copy_up(struct dentry *dentry)
406 ovl_path_lower(next, &lowerpath); 395 ovl_path_lower(next, &lowerpath);
407 err = vfs_getattr(&lowerpath, &stat); 396 err = vfs_getattr(&lowerpath, &stat);
408 if (!err) 397 if (!err)
409 err = ovl_copy_up_one(parent, next, &lowerpath, &stat, NULL); 398 err = ovl_copy_up_one(parent, next, &lowerpath, &stat);
410 399
411 dput(parent); 400 dput(parent);
412 dput(next); 401 dput(next);
diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c
index ec0c2a050043..4060ffde8722 100644
--- a/fs/overlayfs/inode.c
+++ b/fs/overlayfs/inode.c
@@ -12,8 +12,7 @@
12#include <linux/xattr.h> 12#include <linux/xattr.h>
13#include "overlayfs.h" 13#include "overlayfs.h"
14 14
15static int ovl_copy_up_last(struct dentry *dentry, struct iattr *attr, 15static int ovl_copy_up_truncate(struct dentry *dentry)
16 bool no_data)
17{ 16{
18 int err; 17 int err;
19 struct dentry *parent; 18 struct dentry *parent;
@@ -30,10 +29,8 @@ static int ovl_copy_up_last(struct dentry *dentry, struct iattr *attr,
30 if (err) 29 if (err)
31 goto out_dput_parent; 30 goto out_dput_parent;
32 31
33 if (no_data) 32 stat.size = 0;
34 stat.size = 0; 33 err = ovl_copy_up_one(parent, dentry, &lowerpath, &stat);
35
36 err = ovl_copy_up_one(parent, dentry, &lowerpath, &stat, attr);
37 34
38out_dput_parent: 35out_dput_parent:
39 dput(parent); 36 dput(parent);
@@ -49,13 +46,13 @@ int ovl_setattr(struct dentry *dentry, struct iattr *attr)
49 if (err) 46 if (err)
50 goto out; 47 goto out;
51 48
52 upperdentry = ovl_dentry_upper(dentry); 49 err = ovl_copy_up(dentry);
53 if (upperdentry) { 50 if (!err) {
51 upperdentry = ovl_dentry_upper(dentry);
52
54 mutex_lock(&upperdentry->d_inode->i_mutex); 53 mutex_lock(&upperdentry->d_inode->i_mutex);
55 err = notify_change(upperdentry, attr, NULL); 54 err = notify_change(upperdentry, attr, NULL);
56 mutex_unlock(&upperdentry->d_inode->i_mutex); 55 mutex_unlock(&upperdentry->d_inode->i_mutex);
57 } else {
58 err = ovl_copy_up_last(dentry, attr, false);
59 } 56 }
60 ovl_drop_write(dentry); 57 ovl_drop_write(dentry);
61out: 58out:
@@ -353,7 +350,7 @@ struct inode *ovl_d_select_inode(struct dentry *dentry, unsigned file_flags)
353 return ERR_PTR(err); 350 return ERR_PTR(err);
354 351
355 if (file_flags & O_TRUNC) 352 if (file_flags & O_TRUNC)
356 err = ovl_copy_up_last(dentry, NULL, true); 353 err = ovl_copy_up_truncate(dentry);
357 else 354 else
358 err = ovl_copy_up(dentry); 355 err = ovl_copy_up(dentry);
359 ovl_drop_write(dentry); 356 ovl_drop_write(dentry);
diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h
index ea5a40b06e3a..e17154aeaae4 100644
--- a/fs/overlayfs/overlayfs.h
+++ b/fs/overlayfs/overlayfs.h
@@ -194,7 +194,6 @@ void ovl_cleanup(struct inode *dir, struct dentry *dentry);
194/* copy_up.c */ 194/* copy_up.c */
195int ovl_copy_up(struct dentry *dentry); 195int ovl_copy_up(struct dentry *dentry);
196int ovl_copy_up_one(struct dentry *parent, struct dentry *dentry, 196int ovl_copy_up_one(struct dentry *parent, struct dentry *dentry,
197 struct path *lowerpath, struct kstat *stat, 197 struct path *lowerpath, struct kstat *stat);
198 struct iattr *attr);
199int ovl_copy_xattr(struct dentry *old, struct dentry *new); 198int ovl_copy_xattr(struct dentry *old, struct dentry *new);
200int ovl_set_attr(struct dentry *upper, struct kstat *stat); 199int ovl_set_attr(struct dentry *upper, struct kstat *stat);
diff --git a/fs/proc/base.c b/fs/proc/base.c
index bd3e9e68125b..4bd5d3118acd 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -2494,6 +2494,7 @@ static ssize_t proc_coredump_filter_write(struct file *file,
2494 mm = get_task_mm(task); 2494 mm = get_task_mm(task);
2495 if (!mm) 2495 if (!mm)
2496 goto out_no_mm; 2496 goto out_no_mm;
2497 ret = 0;
2497 2498
2498 for (i = 0, mask = 1; i < MMF_DUMP_FILTER_BITS; i++, mask <<= 1) { 2499 for (i = 0, mask = 1; i < MMF_DUMP_FILTER_BITS; i++, mask <<= 1) {
2499 if (val & mask) 2500 if (val & mask)
diff --git a/fs/splice.c b/fs/splice.c
index 801c21cd77fe..4cf700d50b40 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -809,6 +809,13 @@ static int splice_from_pipe_feed(struct pipe_inode_info *pipe, struct splice_des
809 */ 809 */
810static int splice_from_pipe_next(struct pipe_inode_info *pipe, struct splice_desc *sd) 810static int splice_from_pipe_next(struct pipe_inode_info *pipe, struct splice_desc *sd)
811{ 811{
812 /*
813 * Check for signal early to make process killable when there are
814 * always buffers available
815 */
816 if (signal_pending(current))
817 return -ERESTARTSYS;
818
812 while (!pipe->nrbufs) { 819 while (!pipe->nrbufs) {
813 if (!pipe->writers) 820 if (!pipe->writers)
814 return 0; 821 return 0;
@@ -884,6 +891,7 @@ ssize_t __splice_from_pipe(struct pipe_inode_info *pipe, struct splice_desc *sd,
884 891
885 splice_from_pipe_begin(sd); 892 splice_from_pipe_begin(sd);
886 do { 893 do {
894 cond_resched();
887 ret = splice_from_pipe_next(pipe, sd); 895 ret = splice_from_pipe_next(pipe, sd);
888 if (ret > 0) 896 if (ret > 0)
889 ret = splice_from_pipe_feed(pipe, sd, actor); 897 ret = splice_from_pipe_feed(pipe, sd, actor);
diff --git a/fs/sysv/inode.c b/fs/sysv/inode.c
index 590ad9206e3f..02fa1dcc5969 100644
--- a/fs/sysv/inode.c
+++ b/fs/sysv/inode.c
@@ -162,15 +162,8 @@ void sysv_set_inode(struct inode *inode, dev_t rdev)
162 inode->i_fop = &sysv_dir_operations; 162 inode->i_fop = &sysv_dir_operations;
163 inode->i_mapping->a_ops = &sysv_aops; 163 inode->i_mapping->a_ops = &sysv_aops;
164 } else if (S_ISLNK(inode->i_mode)) { 164 } else if (S_ISLNK(inode->i_mode)) {
165 if (inode->i_blocks) { 165 inode->i_op = &sysv_symlink_inode_operations;
166 inode->i_op = &sysv_symlink_inode_operations; 166 inode->i_mapping->a_ops = &sysv_aops;
167 inode->i_mapping->a_ops = &sysv_aops;
168 } else {
169 inode->i_op = &simple_symlink_inode_operations;
170 inode->i_link = (char *)SYSV_I(inode)->i_data;
171 nd_terminate_link(inode->i_link, inode->i_size,
172 sizeof(SYSV_I(inode)->i_data) - 1);
173 }
174 } else 167 } else
175 init_special_inode(inode, inode->i_mode, rdev); 168 init_special_inode(inode, inode->i_mode, rdev);
176} 169}