aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorTakashi Iwai <tiwai@suse.de>2013-04-12 04:27:39 -0400
committerTakashi Iwai <tiwai@suse.de>2013-04-12 04:27:39 -0400
commit232a73dda2f2dba9002b727eb3ec76d82882b90c (patch)
tree308cdeea9bf18de71af79dcd5a30a8f2be9af6da /fs
parent889d66848b12d891248b03abcb2a42047f8e172a (diff)
parentca62bed0bbf9baf88d8bb404d72dee3b44ef057e (diff)
Merge tag 'asoc-v3.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus
ASoC: Updates for v3.9 A few updates, more than I'd like, fixing some relatively small issues but mostly driver specific ones. Nothing wildly exciting so if it doesn't make v3.9 it won't be the end of the world but it'd be nice.
Diffstat (limited to 'fs')
-rw-r--r--fs/block_dev.c1
-rw-r--r--fs/btrfs/ctree.c30
-rw-r--r--fs/btrfs/disk-io.c14
-rw-r--r--fs/btrfs/extent-tree.c84
-rw-r--r--fs/btrfs/extent_io.c33
-rw-r--r--fs/btrfs/extent_io.h2
-rw-r--r--fs/btrfs/file-item.c6
-rw-r--r--fs/btrfs/file.c9
-rw-r--r--fs/btrfs/inode.c22
-rw-r--r--fs/btrfs/ordered-data.c2
-rw-r--r--fs/btrfs/qgroup.c3
-rw-r--r--fs/btrfs/scrub.c3
-rw-r--r--fs/btrfs/send.c10
-rw-r--r--fs/btrfs/volumes.c13
-rw-r--r--fs/dcache.c16
-rw-r--r--fs/ext4/extents.c11
-rw-r--r--fs/ext4/indirect.c4
-rw-r--r--fs/gfs2/file.c5
-rw-r--r--fs/gfs2/incore.h1
-rw-r--r--fs/gfs2/lock_dlm.c39
-rw-r--r--fs/gfs2/rgrp.c32
-rw-r--r--fs/internal.h5
-rw-r--r--fs/namespace.c54
-rw-r--r--fs/nfs/blocklayout/blocklayoutdm.c4
-rw-r--r--fs/nfs/idmap.c13
-rw-r--r--fs/nfs/nfs4filelayout.c1
-rw-r--r--fs/nfs/nfs4proc.c16
-rw-r--r--fs/nfs/pnfs.c81
-rw-r--r--fs/nfs/pnfs.h6
-rw-r--r--fs/nfsd/nfs4xdr.c2
-rw-r--r--fs/nfsd/nfscache.c11
-rw-r--r--fs/nfsd/vfs.c3
-rw-r--r--fs/pnode.c6
-rw-r--r--fs/pnode.h1
-rw-r--r--fs/proc/root.c4
-rw-r--r--fs/read_write.c28
-rw-r--r--fs/reiserfs/xattr.c4
-rw-r--r--fs/splice.c4
-rw-r--r--fs/sysfs/dir.c17
-rw-r--r--fs/sysfs/mount.c4
-rw-r--r--fs/ubifs/super.c12
41 files changed, 474 insertions, 142 deletions
diff --git a/fs/block_dev.c b/fs/block_dev.c
index aea605c98ba6..aae187a7f94a 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -551,6 +551,7 @@ struct block_device *bdgrab(struct block_device *bdev)
551 ihold(bdev->bd_inode); 551 ihold(bdev->bd_inode);
552 return bdev; 552 return bdev;
553} 553}
554EXPORT_SYMBOL(bdgrab);
554 555
555long nr_blockdev_pages(void) 556long nr_blockdev_pages(void)
556{ 557{
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index ecd25a1b4e51..ca9d8f1a3bb6 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -651,6 +651,8 @@ tree_mod_log_insert_root(struct btrfs_fs_info *fs_info,
651 if (tree_mod_dont_log(fs_info, NULL)) 651 if (tree_mod_dont_log(fs_info, NULL))
652 return 0; 652 return 0;
653 653
654 __tree_mod_log_free_eb(fs_info, old_root);
655
654 ret = tree_mod_alloc(fs_info, flags, &tm); 656 ret = tree_mod_alloc(fs_info, flags, &tm);
655 if (ret < 0) 657 if (ret < 0)
656 goto out; 658 goto out;
@@ -736,7 +738,7 @@ tree_mod_log_search(struct btrfs_fs_info *fs_info, u64 start, u64 min_seq)
736static noinline void 738static noinline void
737tree_mod_log_eb_copy(struct btrfs_fs_info *fs_info, struct extent_buffer *dst, 739tree_mod_log_eb_copy(struct btrfs_fs_info *fs_info, struct extent_buffer *dst,
738 struct extent_buffer *src, unsigned long dst_offset, 740 struct extent_buffer *src, unsigned long dst_offset,
739 unsigned long src_offset, int nr_items) 741 unsigned long src_offset, int nr_items, int log_removal)
740{ 742{
741 int ret; 743 int ret;
742 int i; 744 int i;
@@ -750,10 +752,12 @@ tree_mod_log_eb_copy(struct btrfs_fs_info *fs_info, struct extent_buffer *dst,
750 } 752 }
751 753
752 for (i = 0; i < nr_items; i++) { 754 for (i = 0; i < nr_items; i++) {
753 ret = tree_mod_log_insert_key_locked(fs_info, src, 755 if (log_removal) {
754 i + src_offset, 756 ret = tree_mod_log_insert_key_locked(fs_info, src,
755 MOD_LOG_KEY_REMOVE); 757 i + src_offset,
756 BUG_ON(ret < 0); 758 MOD_LOG_KEY_REMOVE);
759 BUG_ON(ret < 0);
760 }
757 ret = tree_mod_log_insert_key_locked(fs_info, dst, 761 ret = tree_mod_log_insert_key_locked(fs_info, dst,
758 i + dst_offset, 762 i + dst_offset,
759 MOD_LOG_KEY_ADD); 763 MOD_LOG_KEY_ADD);
@@ -927,7 +931,6 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
927 ret = btrfs_dec_ref(trans, root, buf, 1, 1); 931 ret = btrfs_dec_ref(trans, root, buf, 1, 1);
928 BUG_ON(ret); /* -ENOMEM */ 932 BUG_ON(ret); /* -ENOMEM */
929 } 933 }
930 tree_mod_log_free_eb(root->fs_info, buf);
931 clean_tree_block(trans, root, buf); 934 clean_tree_block(trans, root, buf);
932 *last_ref = 1; 935 *last_ref = 1;
933 } 936 }
@@ -1046,6 +1049,7 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
1046 btrfs_set_node_ptr_generation(parent, parent_slot, 1049 btrfs_set_node_ptr_generation(parent, parent_slot,
1047 trans->transid); 1050 trans->transid);
1048 btrfs_mark_buffer_dirty(parent); 1051 btrfs_mark_buffer_dirty(parent);
1052 tree_mod_log_free_eb(root->fs_info, buf);
1049 btrfs_free_tree_block(trans, root, buf, parent_start, 1053 btrfs_free_tree_block(trans, root, buf, parent_start,
1050 last_ref); 1054 last_ref);
1051 } 1055 }
@@ -1750,7 +1754,6 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
1750 goto enospc; 1754 goto enospc;
1751 } 1755 }
1752 1756
1753 tree_mod_log_free_eb(root->fs_info, root->node);
1754 tree_mod_log_set_root_pointer(root, child); 1757 tree_mod_log_set_root_pointer(root, child);
1755 rcu_assign_pointer(root->node, child); 1758 rcu_assign_pointer(root->node, child);
1756 1759
@@ -2995,7 +2998,7 @@ static int push_node_left(struct btrfs_trans_handle *trans,
2995 push_items = min(src_nritems - 8, push_items); 2998 push_items = min(src_nritems - 8, push_items);
2996 2999
2997 tree_mod_log_eb_copy(root->fs_info, dst, src, dst_nritems, 0, 3000 tree_mod_log_eb_copy(root->fs_info, dst, src, dst_nritems, 0,
2998 push_items); 3001 push_items, 1);
2999 copy_extent_buffer(dst, src, 3002 copy_extent_buffer(dst, src,
3000 btrfs_node_key_ptr_offset(dst_nritems), 3003 btrfs_node_key_ptr_offset(dst_nritems),
3001 btrfs_node_key_ptr_offset(0), 3004 btrfs_node_key_ptr_offset(0),
@@ -3066,7 +3069,7 @@ static int balance_node_right(struct btrfs_trans_handle *trans,
3066 sizeof(struct btrfs_key_ptr)); 3069 sizeof(struct btrfs_key_ptr));
3067 3070
3068 tree_mod_log_eb_copy(root->fs_info, dst, src, 0, 3071 tree_mod_log_eb_copy(root->fs_info, dst, src, 0,
3069 src_nritems - push_items, push_items); 3072 src_nritems - push_items, push_items, 1);
3070 copy_extent_buffer(dst, src, 3073 copy_extent_buffer(dst, src,
3071 btrfs_node_key_ptr_offset(0), 3074 btrfs_node_key_ptr_offset(0),
3072 btrfs_node_key_ptr_offset(src_nritems - push_items), 3075 btrfs_node_key_ptr_offset(src_nritems - push_items),
@@ -3218,12 +3221,18 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
3218 int mid; 3221 int mid;
3219 int ret; 3222 int ret;
3220 u32 c_nritems; 3223 u32 c_nritems;
3224 int tree_mod_log_removal = 1;
3221 3225
3222 c = path->nodes[level]; 3226 c = path->nodes[level];
3223 WARN_ON(btrfs_header_generation(c) != trans->transid); 3227 WARN_ON(btrfs_header_generation(c) != trans->transid);
3224 if (c == root->node) { 3228 if (c == root->node) {
3225 /* trying to split the root, lets make a new one */ 3229 /* trying to split the root, lets make a new one */
3226 ret = insert_new_root(trans, root, path, level + 1); 3230 ret = insert_new_root(trans, root, path, level + 1);
3231 /*
3232 * removal of root nodes has been logged by
3233 * tree_mod_log_set_root_pointer due to locking
3234 */
3235 tree_mod_log_removal = 0;
3227 if (ret) 3236 if (ret)
3228 return ret; 3237 return ret;
3229 } else { 3238 } else {
@@ -3261,7 +3270,8 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
3261 (unsigned long)btrfs_header_chunk_tree_uuid(split), 3270 (unsigned long)btrfs_header_chunk_tree_uuid(split),
3262 BTRFS_UUID_SIZE); 3271 BTRFS_UUID_SIZE);
3263 3272
3264 tree_mod_log_eb_copy(root->fs_info, split, c, 0, mid, c_nritems - mid); 3273 tree_mod_log_eb_copy(root->fs_info, split, c, 0, mid, c_nritems - mid,
3274 tree_mod_log_removal);
3265 copy_extent_buffer(split, c, 3275 copy_extent_buffer(split, c,
3266 btrfs_node_key_ptr_offset(0), 3276 btrfs_node_key_ptr_offset(0),
3267 btrfs_node_key_ptr_offset(mid), 3277 btrfs_node_key_ptr_offset(mid),
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 7d84651e850b..6d19a0a554aa 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1291,6 +1291,7 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
1291 0, objectid, NULL, 0, 0, 0); 1291 0, objectid, NULL, 0, 0, 0);
1292 if (IS_ERR(leaf)) { 1292 if (IS_ERR(leaf)) {
1293 ret = PTR_ERR(leaf); 1293 ret = PTR_ERR(leaf);
1294 leaf = NULL;
1294 goto fail; 1295 goto fail;
1295 } 1296 }
1296 1297
@@ -1334,11 +1335,16 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
1334 1335
1335 btrfs_tree_unlock(leaf); 1336 btrfs_tree_unlock(leaf);
1336 1337
1338 return root;
1339
1337fail: 1340fail:
1338 if (ret) 1341 if (leaf) {
1339 return ERR_PTR(ret); 1342 btrfs_tree_unlock(leaf);
1343 free_extent_buffer(leaf);
1344 }
1345 kfree(root);
1340 1346
1341 return root; 1347 return ERR_PTR(ret);
1342} 1348}
1343 1349
1344static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans, 1350static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
@@ -3253,7 +3259,7 @@ void btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
3253 if (btrfs_root_refs(&root->root_item) == 0) 3259 if (btrfs_root_refs(&root->root_item) == 0)
3254 synchronize_srcu(&fs_info->subvol_srcu); 3260 synchronize_srcu(&fs_info->subvol_srcu);
3255 3261
3256 if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { 3262 if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
3257 btrfs_free_log(NULL, root); 3263 btrfs_free_log(NULL, root);
3258 btrfs_free_log_root_tree(NULL, fs_info); 3264 btrfs_free_log_root_tree(NULL, fs_info);
3259 } 3265 }
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 9ac2eca681eb..3d551231caba 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -257,7 +257,8 @@ static int exclude_super_stripes(struct btrfs_root *root,
257 cache->bytes_super += stripe_len; 257 cache->bytes_super += stripe_len;
258 ret = add_excluded_extent(root, cache->key.objectid, 258 ret = add_excluded_extent(root, cache->key.objectid,
259 stripe_len); 259 stripe_len);
260 BUG_ON(ret); /* -ENOMEM */ 260 if (ret)
261 return ret;
261 } 262 }
262 263
263 for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) { 264 for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
@@ -265,13 +266,17 @@ static int exclude_super_stripes(struct btrfs_root *root,
265 ret = btrfs_rmap_block(&root->fs_info->mapping_tree, 266 ret = btrfs_rmap_block(&root->fs_info->mapping_tree,
266 cache->key.objectid, bytenr, 267 cache->key.objectid, bytenr,
267 0, &logical, &nr, &stripe_len); 268 0, &logical, &nr, &stripe_len);
268 BUG_ON(ret); /* -ENOMEM */ 269 if (ret)
270 return ret;
269 271
270 while (nr--) { 272 while (nr--) {
271 cache->bytes_super += stripe_len; 273 cache->bytes_super += stripe_len;
272 ret = add_excluded_extent(root, logical[nr], 274 ret = add_excluded_extent(root, logical[nr],
273 stripe_len); 275 stripe_len);
274 BUG_ON(ret); /* -ENOMEM */ 276 if (ret) {
277 kfree(logical);
278 return ret;
279 }
275 } 280 }
276 281
277 kfree(logical); 282 kfree(logical);
@@ -4438,7 +4443,7 @@ static void update_global_block_rsv(struct btrfs_fs_info *fs_info)
4438 spin_lock(&sinfo->lock); 4443 spin_lock(&sinfo->lock);
4439 spin_lock(&block_rsv->lock); 4444 spin_lock(&block_rsv->lock);
4440 4445
4441 block_rsv->size = num_bytes; 4446 block_rsv->size = min_t(u64, num_bytes, 512 * 1024 * 1024);
4442 4447
4443 num_bytes = sinfo->bytes_used + sinfo->bytes_pinned + 4448 num_bytes = sinfo->bytes_used + sinfo->bytes_pinned +
4444 sinfo->bytes_reserved + sinfo->bytes_readonly + 4449 sinfo->bytes_reserved + sinfo->bytes_readonly +
@@ -4793,14 +4798,49 @@ out_fail:
4793 * If the inodes csum_bytes is the same as the original 4798 * If the inodes csum_bytes is the same as the original
4794 * csum_bytes then we know we haven't raced with any free()ers 4799 * csum_bytes then we know we haven't raced with any free()ers
4795 * so we can just reduce our inodes csum bytes and carry on. 4800 * so we can just reduce our inodes csum bytes and carry on.
4796 * Otherwise we have to do the normal free thing to account for
4797 * the case that the free side didn't free up its reserve
4798 * because of this outstanding reservation.
4799 */ 4801 */
4800 if (BTRFS_I(inode)->csum_bytes == csum_bytes) 4802 if (BTRFS_I(inode)->csum_bytes == csum_bytes) {
4801 calc_csum_metadata_size(inode, num_bytes, 0); 4803 calc_csum_metadata_size(inode, num_bytes, 0);
4802 else 4804 } else {
4803 to_free = calc_csum_metadata_size(inode, num_bytes, 0); 4805 u64 orig_csum_bytes = BTRFS_I(inode)->csum_bytes;
4806 u64 bytes;
4807
4808 /*
4809 * This is tricky, but first we need to figure out how much we
4810 * free'd from any free-ers that occured during this
4811 * reservation, so we reset ->csum_bytes to the csum_bytes
4812 * before we dropped our lock, and then call the free for the
4813 * number of bytes that were freed while we were trying our
4814 * reservation.
4815 */
4816 bytes = csum_bytes - BTRFS_I(inode)->csum_bytes;
4817 BTRFS_I(inode)->csum_bytes = csum_bytes;
4818 to_free = calc_csum_metadata_size(inode, bytes, 0);
4819
4820
4821 /*
4822 * Now we need to see how much we would have freed had we not
4823 * been making this reservation and our ->csum_bytes were not
4824 * artificially inflated.
4825 */
4826 BTRFS_I(inode)->csum_bytes = csum_bytes - num_bytes;
4827 bytes = csum_bytes - orig_csum_bytes;
4828 bytes = calc_csum_metadata_size(inode, bytes, 0);
4829
4830 /*
4831 * Now reset ->csum_bytes to what it should be. If bytes is
4832 * more than to_free then we would have free'd more space had we
4833 * not had an artificially high ->csum_bytes, so we need to free
4834 * the remainder. If bytes is the same or less then we don't
4835 * need to do anything, the other free-ers did the correct
4836 * thing.
4837 */
4838 BTRFS_I(inode)->csum_bytes = orig_csum_bytes - num_bytes;
4839 if (bytes > to_free)
4840 to_free = bytes - to_free;
4841 else
4842 to_free = 0;
4843 }
4804 spin_unlock(&BTRFS_I(inode)->lock); 4844 spin_unlock(&BTRFS_I(inode)->lock);
4805 if (dropped) 4845 if (dropped)
4806 to_free += btrfs_calc_trans_metadata_size(root, dropped); 4846 to_free += btrfs_calc_trans_metadata_size(root, dropped);
@@ -7947,7 +7987,17 @@ int btrfs_read_block_groups(struct btrfs_root *root)
7947 * info has super bytes accounted for, otherwise we'll think 7987 * info has super bytes accounted for, otherwise we'll think
7948 * we have more space than we actually do. 7988 * we have more space than we actually do.
7949 */ 7989 */
7950 exclude_super_stripes(root, cache); 7990 ret = exclude_super_stripes(root, cache);
7991 if (ret) {
7992 /*
7993 * We may have excluded something, so call this just in
7994 * case.
7995 */
7996 free_excluded_extents(root, cache);
7997 kfree(cache->free_space_ctl);
7998 kfree(cache);
7999 goto error;
8000 }
7951 8001
7952 /* 8002 /*
7953 * check for two cases, either we are full, and therefore 8003 * check for two cases, either we are full, and therefore
@@ -8089,7 +8139,17 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
8089 8139
8090 cache->last_byte_to_unpin = (u64)-1; 8140 cache->last_byte_to_unpin = (u64)-1;
8091 cache->cached = BTRFS_CACHE_FINISHED; 8141 cache->cached = BTRFS_CACHE_FINISHED;
8092 exclude_super_stripes(root, cache); 8142 ret = exclude_super_stripes(root, cache);
8143 if (ret) {
8144 /*
8145 * We may have excluded something, so call this just in
8146 * case.
8147 */
8148 free_excluded_extents(root, cache);
8149 kfree(cache->free_space_ctl);
8150 kfree(cache);
8151 return ret;
8152 }
8093 8153
8094 add_new_free_space(cache, root->fs_info, chunk_offset, 8154 add_new_free_space(cache, root->fs_info, chunk_offset,
8095 chunk_offset + size); 8155 chunk_offset + size);
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index f173c5af6461..cdee391fc7bf 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -1257,6 +1257,39 @@ int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end)
1257 GFP_NOFS); 1257 GFP_NOFS);
1258} 1258}
1259 1259
1260int extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end)
1261{
1262 unsigned long index = start >> PAGE_CACHE_SHIFT;
1263 unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1264 struct page *page;
1265
1266 while (index <= end_index) {
1267 page = find_get_page(inode->i_mapping, index);
1268 BUG_ON(!page); /* Pages should be in the extent_io_tree */
1269 clear_page_dirty_for_io(page);
1270 page_cache_release(page);
1271 index++;
1272 }
1273 return 0;
1274}
1275
1276int extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end)
1277{
1278 unsigned long index = start >> PAGE_CACHE_SHIFT;
1279 unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1280 struct page *page;
1281
1282 while (index <= end_index) {
1283 page = find_get_page(inode->i_mapping, index);
1284 BUG_ON(!page); /* Pages should be in the extent_io_tree */
1285 account_page_redirty(page);
1286 __set_page_dirty_nobuffers(page);
1287 page_cache_release(page);
1288 index++;
1289 }
1290 return 0;
1291}
1292
1260/* 1293/*
1261 * helper function to set both pages and extents in the tree writeback 1294 * helper function to set both pages and extents in the tree writeback
1262 */ 1295 */
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 6068a1985560..258c92156857 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -325,6 +325,8 @@ int map_private_extent_buffer(struct extent_buffer *eb, unsigned long offset,
325 unsigned long *map_len); 325 unsigned long *map_len);
326int extent_range_uptodate(struct extent_io_tree *tree, 326int extent_range_uptodate(struct extent_io_tree *tree,
327 u64 start, u64 end); 327 u64 start, u64 end);
328int extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end);
329int extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end);
328int extent_clear_unlock_delalloc(struct inode *inode, 330int extent_clear_unlock_delalloc(struct inode *inode,
329 struct extent_io_tree *tree, 331 struct extent_io_tree *tree,
330 u64 start, u64 end, struct page *locked_page, 332 u64 start, u64 end, struct page *locked_page,
diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c
index ec160202be3e..c4628a201cb3 100644
--- a/fs/btrfs/file-item.c
+++ b/fs/btrfs/file-item.c
@@ -118,9 +118,11 @@ struct btrfs_csum_item *btrfs_lookup_csum(struct btrfs_trans_handle *trans,
118 csums_in_item = btrfs_item_size_nr(leaf, path->slots[0]); 118 csums_in_item = btrfs_item_size_nr(leaf, path->slots[0]);
119 csums_in_item /= csum_size; 119 csums_in_item /= csum_size;
120 120
121 if (csum_offset >= csums_in_item) { 121 if (csum_offset == csums_in_item) {
122 ret = -EFBIG; 122 ret = -EFBIG;
123 goto fail; 123 goto fail;
124 } else if (csum_offset > csums_in_item) {
125 goto fail;
124 } 126 }
125 } 127 }
126 item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item); 128 item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item);
@@ -728,7 +730,6 @@ int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
728 return -ENOMEM; 730 return -ENOMEM;
729 731
730 sector_sum = sums->sums; 732 sector_sum = sums->sums;
731 trans->adding_csums = 1;
732again: 733again:
733 next_offset = (u64)-1; 734 next_offset = (u64)-1;
734 found_next = 0; 735 found_next = 0;
@@ -899,7 +900,6 @@ next_sector:
899 goto again; 900 goto again;
900 } 901 }
901out: 902out:
902 trans->adding_csums = 0;
903 btrfs_free_path(path); 903 btrfs_free_path(path);
904 return ret; 904 return ret;
905 905
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 5b4ea5f55b8f..ade03e6f7bd2 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -2142,6 +2142,7 @@ static long btrfs_fallocate(struct file *file, int mode,
2142{ 2142{
2143 struct inode *inode = file_inode(file); 2143 struct inode *inode = file_inode(file);
2144 struct extent_state *cached_state = NULL; 2144 struct extent_state *cached_state = NULL;
2145 struct btrfs_root *root = BTRFS_I(inode)->root;
2145 u64 cur_offset; 2146 u64 cur_offset;
2146 u64 last_byte; 2147 u64 last_byte;
2147 u64 alloc_start; 2148 u64 alloc_start;
@@ -2169,6 +2170,11 @@ static long btrfs_fallocate(struct file *file, int mode,
2169 ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start); 2170 ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start);
2170 if (ret) 2171 if (ret)
2171 return ret; 2172 return ret;
2173 if (root->fs_info->quota_enabled) {
2174 ret = btrfs_qgroup_reserve(root, alloc_end - alloc_start);
2175 if (ret)
2176 goto out_reserve_fail;
2177 }
2172 2178
2173 /* 2179 /*
2174 * wait for ordered IO before we have any locks. We'll loop again 2180 * wait for ordered IO before we have any locks. We'll loop again
@@ -2272,6 +2278,9 @@ static long btrfs_fallocate(struct file *file, int mode,
2272 &cached_state, GFP_NOFS); 2278 &cached_state, GFP_NOFS);
2273out: 2279out:
2274 mutex_unlock(&inode->i_mutex); 2280 mutex_unlock(&inode->i_mutex);
2281 if (root->fs_info->quota_enabled)
2282 btrfs_qgroup_free(root, alloc_end - alloc_start);
2283out_reserve_fail:
2275 /* Let go of our reservation. */ 2284 /* Let go of our reservation. */
2276 btrfs_free_reserved_data_space(inode, alloc_end - alloc_start); 2285 btrfs_free_reserved_data_space(inode, alloc_end - alloc_start);
2277 return ret; 2286 return ret;
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index ca1b767d51f7..09c58a35b429 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -353,6 +353,7 @@ static noinline int compress_file_range(struct inode *inode,
353 int i; 353 int i;
354 int will_compress; 354 int will_compress;
355 int compress_type = root->fs_info->compress_type; 355 int compress_type = root->fs_info->compress_type;
356 int redirty = 0;
356 357
357 /* if this is a small write inside eof, kick off a defrag */ 358 /* if this is a small write inside eof, kick off a defrag */
358 if ((end - start + 1) < 16 * 1024 && 359 if ((end - start + 1) < 16 * 1024 &&
@@ -415,6 +416,17 @@ again:
415 if (BTRFS_I(inode)->force_compress) 416 if (BTRFS_I(inode)->force_compress)
416 compress_type = BTRFS_I(inode)->force_compress; 417 compress_type = BTRFS_I(inode)->force_compress;
417 418
419 /*
420 * we need to call clear_page_dirty_for_io on each
421 * page in the range. Otherwise applications with the file
422 * mmap'd can wander in and change the page contents while
423 * we are compressing them.
424 *
425 * If the compression fails for any reason, we set the pages
426 * dirty again later on.
427 */
428 extent_range_clear_dirty_for_io(inode, start, end);
429 redirty = 1;
418 ret = btrfs_compress_pages(compress_type, 430 ret = btrfs_compress_pages(compress_type,
419 inode->i_mapping, start, 431 inode->i_mapping, start,
420 total_compressed, pages, 432 total_compressed, pages,
@@ -554,6 +566,8 @@ cleanup_and_bail_uncompressed:
554 __set_page_dirty_nobuffers(locked_page); 566 __set_page_dirty_nobuffers(locked_page);
555 /* unlocked later on in the async handlers */ 567 /* unlocked later on in the async handlers */
556 } 568 }
569 if (redirty)
570 extent_range_redirty_for_io(inode, start, end);
557 add_async_extent(async_cow, start, end - start + 1, 571 add_async_extent(async_cow, start, end - start + 1,
558 0, NULL, 0, BTRFS_COMPRESS_NONE); 572 0, NULL, 0, BTRFS_COMPRESS_NONE);
559 *num_added += 1; 573 *num_added += 1;
@@ -1743,8 +1757,10 @@ static noinline int add_pending_csums(struct btrfs_trans_handle *trans,
1743 struct btrfs_ordered_sum *sum; 1757 struct btrfs_ordered_sum *sum;
1744 1758
1745 list_for_each_entry(sum, list, list) { 1759 list_for_each_entry(sum, list, list) {
1760 trans->adding_csums = 1;
1746 btrfs_csum_file_blocks(trans, 1761 btrfs_csum_file_blocks(trans,
1747 BTRFS_I(inode)->root->fs_info->csum_root, sum); 1762 BTRFS_I(inode)->root->fs_info->csum_root, sum);
1763 trans->adding_csums = 0;
1748 } 1764 }
1749 return 0; 1765 return 0;
1750} 1766}
@@ -3679,11 +3695,9 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir,
3679 * 1 for the dir item 3695 * 1 for the dir item
3680 * 1 for the dir index 3696 * 1 for the dir index
3681 * 1 for the inode ref 3697 * 1 for the inode ref
3682 * 1 for the inode ref in the tree log
3683 * 2 for the dir entries in the log
3684 * 1 for the inode 3698 * 1 for the inode
3685 */ 3699 */
3686 trans = btrfs_start_transaction(root, 8); 3700 trans = btrfs_start_transaction(root, 5);
3687 if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC) 3701 if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC)
3688 return trans; 3702 return trans;
3689 3703
@@ -8127,7 +8141,7 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
8127 * inodes. So 5 * 2 is 10, plus 1 for the new link, so 11 total items 8141 * inodes. So 5 * 2 is 10, plus 1 for the new link, so 11 total items
8128 * should cover the worst case number of items we'll modify. 8142 * should cover the worst case number of items we'll modify.
8129 */ 8143 */
8130 trans = btrfs_start_transaction(root, 20); 8144 trans = btrfs_start_transaction(root, 11);
8131 if (IS_ERR(trans)) { 8145 if (IS_ERR(trans)) {
8132 ret = PTR_ERR(trans); 8146 ret = PTR_ERR(trans);
8133 goto out_notrans; 8147 goto out_notrans;
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index dc08d77b717e..005c45db699e 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -557,6 +557,7 @@ void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput)
557 INIT_LIST_HEAD(&splice); 557 INIT_LIST_HEAD(&splice);
558 INIT_LIST_HEAD(&works); 558 INIT_LIST_HEAD(&works);
559 559
560 mutex_lock(&root->fs_info->ordered_operations_mutex);
560 spin_lock(&root->fs_info->ordered_extent_lock); 561 spin_lock(&root->fs_info->ordered_extent_lock);
561 list_splice_init(&root->fs_info->ordered_extents, &splice); 562 list_splice_init(&root->fs_info->ordered_extents, &splice);
562 while (!list_empty(&splice)) { 563 while (!list_empty(&splice)) {
@@ -600,6 +601,7 @@ void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput)
600 601
601 cond_resched(); 602 cond_resched();
602 } 603 }
604 mutex_unlock(&root->fs_info->ordered_operations_mutex);
603} 605}
604 606
605/* 607/*
diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index 5471e47d6559..b44124dd2370 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -1153,7 +1153,7 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
1153 ret = btrfs_find_all_roots(trans, fs_info, node->bytenr, 1153 ret = btrfs_find_all_roots(trans, fs_info, node->bytenr,
1154 sgn > 0 ? node->seq - 1 : node->seq, &roots); 1154 sgn > 0 ? node->seq - 1 : node->seq, &roots);
1155 if (ret < 0) 1155 if (ret < 0)
1156 goto out; 1156 return ret;
1157 1157
1158 spin_lock(&fs_info->qgroup_lock); 1158 spin_lock(&fs_info->qgroup_lock);
1159 quota_root = fs_info->quota_root; 1159 quota_root = fs_info->quota_root;
@@ -1275,7 +1275,6 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
1275 ret = 0; 1275 ret = 0;
1276unlock: 1276unlock:
1277 spin_unlock(&fs_info->qgroup_lock); 1277 spin_unlock(&fs_info->qgroup_lock);
1278out:
1279 ulist_free(roots); 1278 ulist_free(roots);
1280 ulist_free(tmp); 1279 ulist_free(tmp);
1281 1280
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 53c3501fa4ca..85e072b956d5 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -542,7 +542,6 @@ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
542 eb = path->nodes[0]; 542 eb = path->nodes[0];
543 ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item); 543 ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
544 item_size = btrfs_item_size_nr(eb, path->slots[0]); 544 item_size = btrfs_item_size_nr(eb, path->slots[0]);
545 btrfs_release_path(path);
546 545
547 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) { 546 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
548 do { 547 do {
@@ -558,7 +557,9 @@ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
558 ret < 0 ? -1 : ref_level, 557 ret < 0 ? -1 : ref_level,
559 ret < 0 ? -1 : ref_root); 558 ret < 0 ? -1 : ref_root);
560 } while (ret != 1); 559 } while (ret != 1);
560 btrfs_release_path(path);
561 } else { 561 } else {
562 btrfs_release_path(path);
562 swarn.path = path; 563 swarn.path = path;
563 swarn.dev = dev; 564 swarn.dev = dev;
564 iterate_extent_inodes(fs_info, found_key.objectid, 565 iterate_extent_inodes(fs_info, found_key.objectid,
diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c
index f7a8b861058b..c85e7c6b4598 100644
--- a/fs/btrfs/send.c
+++ b/fs/btrfs/send.c
@@ -3945,12 +3945,10 @@ static int is_extent_unchanged(struct send_ctx *sctx,
3945 found_key.type != key.type) { 3945 found_key.type != key.type) {
3946 key.offset += right_len; 3946 key.offset += right_len;
3947 break; 3947 break;
3948 } else { 3948 }
3949 if (found_key.offset != key.offset + right_len) { 3949 if (found_key.offset != key.offset + right_len) {
3950 /* Should really not happen */ 3950 ret = 0;
3951 ret = -EIO; 3951 goto out;
3952 goto out;
3953 }
3954 } 3952 }
3955 key = found_key; 3953 key = found_key;
3956 } 3954 }
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 5989a92236f7..2854c824ab64 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -4935,7 +4935,18 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
4935 em = lookup_extent_mapping(em_tree, chunk_start, 1); 4935 em = lookup_extent_mapping(em_tree, chunk_start, 1);
4936 read_unlock(&em_tree->lock); 4936 read_unlock(&em_tree->lock);
4937 4937
4938 BUG_ON(!em || em->start != chunk_start); 4938 if (!em) {
4939 printk(KERN_ERR "btrfs: couldn't find em for chunk %Lu\n",
4940 chunk_start);
4941 return -EIO;
4942 }
4943
4944 if (em->start != chunk_start) {
4945 printk(KERN_ERR "btrfs: bad chunk start, em=%Lu, wanted=%Lu\n",
4946 em->start, chunk_start);
4947 free_extent_map(em);
4948 return -EIO;
4949 }
4939 map = (struct map_lookup *)em->bdev; 4950 map = (struct map_lookup *)em->bdev;
4940 4951
4941 length = em->len; 4952 length = em->len;
diff --git a/fs/dcache.c b/fs/dcache.c
index fbfae008ba44..e8bc3420d63e 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -2542,7 +2542,6 @@ static int prepend_path(const struct path *path,
2542 bool slash = false; 2542 bool slash = false;
2543 int error = 0; 2543 int error = 0;
2544 2544
2545 br_read_lock(&vfsmount_lock);
2546 while (dentry != root->dentry || vfsmnt != root->mnt) { 2545 while (dentry != root->dentry || vfsmnt != root->mnt) {
2547 struct dentry * parent; 2546 struct dentry * parent;
2548 2547
@@ -2572,8 +2571,6 @@ static int prepend_path(const struct path *path,
2572 if (!error && !slash) 2571 if (!error && !slash)
2573 error = prepend(buffer, buflen, "/", 1); 2572 error = prepend(buffer, buflen, "/", 1);
2574 2573
2575out:
2576 br_read_unlock(&vfsmount_lock);
2577 return error; 2574 return error;
2578 2575
2579global_root: 2576global_root:
@@ -2590,7 +2587,7 @@ global_root:
2590 error = prepend(buffer, buflen, "/", 1); 2587 error = prepend(buffer, buflen, "/", 1);
2591 if (!error) 2588 if (!error)
2592 error = is_mounted(vfsmnt) ? 1 : 2; 2589 error = is_mounted(vfsmnt) ? 1 : 2;
2593 goto out; 2590 return error;
2594} 2591}
2595 2592
2596/** 2593/**
@@ -2617,9 +2614,11 @@ char *__d_path(const struct path *path,
2617 int error; 2614 int error;
2618 2615
2619 prepend(&res, &buflen, "\0", 1); 2616 prepend(&res, &buflen, "\0", 1);
2617 br_read_lock(&vfsmount_lock);
2620 write_seqlock(&rename_lock); 2618 write_seqlock(&rename_lock);
2621 error = prepend_path(path, root, &res, &buflen); 2619 error = prepend_path(path, root, &res, &buflen);
2622 write_sequnlock(&rename_lock); 2620 write_sequnlock(&rename_lock);
2621 br_read_unlock(&vfsmount_lock);
2623 2622
2624 if (error < 0) 2623 if (error < 0)
2625 return ERR_PTR(error); 2624 return ERR_PTR(error);
@@ -2636,9 +2635,11 @@ char *d_absolute_path(const struct path *path,
2636 int error; 2635 int error;
2637 2636
2638 prepend(&res, &buflen, "\0", 1); 2637 prepend(&res, &buflen, "\0", 1);
2638 br_read_lock(&vfsmount_lock);
2639 write_seqlock(&rename_lock); 2639 write_seqlock(&rename_lock);
2640 error = prepend_path(path, &root, &res, &buflen); 2640 error = prepend_path(path, &root, &res, &buflen);
2641 write_sequnlock(&rename_lock); 2641 write_sequnlock(&rename_lock);
2642 br_read_unlock(&vfsmount_lock);
2642 2643
2643 if (error > 1) 2644 if (error > 1)
2644 error = -EINVAL; 2645 error = -EINVAL;
@@ -2702,11 +2703,13 @@ char *d_path(const struct path *path, char *buf, int buflen)
2702 return path->dentry->d_op->d_dname(path->dentry, buf, buflen); 2703 return path->dentry->d_op->d_dname(path->dentry, buf, buflen);
2703 2704
2704 get_fs_root(current->fs, &root); 2705 get_fs_root(current->fs, &root);
2706 br_read_lock(&vfsmount_lock);
2705 write_seqlock(&rename_lock); 2707 write_seqlock(&rename_lock);
2706 error = path_with_deleted(path, &root, &res, &buflen); 2708 error = path_with_deleted(path, &root, &res, &buflen);
2709 write_sequnlock(&rename_lock);
2710 br_read_unlock(&vfsmount_lock);
2707 if (error < 0) 2711 if (error < 0)
2708 res = ERR_PTR(error); 2712 res = ERR_PTR(error);
2709 write_sequnlock(&rename_lock);
2710 path_put(&root); 2713 path_put(&root);
2711 return res; 2714 return res;
2712} 2715}
@@ -2830,6 +2833,7 @@ SYSCALL_DEFINE2(getcwd, char __user *, buf, unsigned long, size)
2830 get_fs_root_and_pwd(current->fs, &root, &pwd); 2833 get_fs_root_and_pwd(current->fs, &root, &pwd);
2831 2834
2832 error = -ENOENT; 2835 error = -ENOENT;
2836 br_read_lock(&vfsmount_lock);
2833 write_seqlock(&rename_lock); 2837 write_seqlock(&rename_lock);
2834 if (!d_unlinked(pwd.dentry)) { 2838 if (!d_unlinked(pwd.dentry)) {
2835 unsigned long len; 2839 unsigned long len;
@@ -2839,6 +2843,7 @@ SYSCALL_DEFINE2(getcwd, char __user *, buf, unsigned long, size)
2839 prepend(&cwd, &buflen, "\0", 1); 2843 prepend(&cwd, &buflen, "\0", 1);
2840 error = prepend_path(&pwd, &root, &cwd, &buflen); 2844 error = prepend_path(&pwd, &root, &cwd, &buflen);
2841 write_sequnlock(&rename_lock); 2845 write_sequnlock(&rename_lock);
2846 br_read_unlock(&vfsmount_lock);
2842 2847
2843 if (error < 0) 2848 if (error < 0)
2844 goto out; 2849 goto out;
@@ -2859,6 +2864,7 @@ SYSCALL_DEFINE2(getcwd, char __user *, buf, unsigned long, size)
2859 } 2864 }
2860 } else { 2865 } else {
2861 write_sequnlock(&rename_lock); 2866 write_sequnlock(&rename_lock);
2867 br_read_unlock(&vfsmount_lock);
2862 } 2868 }
2863 2869
2864out: 2870out:
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 56efcaadf848..9c6d06dcef8b 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -2999,20 +2999,23 @@ static int ext4_split_extent_at(handle_t *handle,
2999 if (split_flag & EXT4_EXT_DATA_VALID1) { 2999 if (split_flag & EXT4_EXT_DATA_VALID1) {
3000 err = ext4_ext_zeroout(inode, ex2); 3000 err = ext4_ext_zeroout(inode, ex2);
3001 zero_ex.ee_block = ex2->ee_block; 3001 zero_ex.ee_block = ex2->ee_block;
3002 zero_ex.ee_len = ext4_ext_get_actual_len(ex2); 3002 zero_ex.ee_len = cpu_to_le16(
3003 ext4_ext_get_actual_len(ex2));
3003 ext4_ext_store_pblock(&zero_ex, 3004 ext4_ext_store_pblock(&zero_ex,
3004 ext4_ext_pblock(ex2)); 3005 ext4_ext_pblock(ex2));
3005 } else { 3006 } else {
3006 err = ext4_ext_zeroout(inode, ex); 3007 err = ext4_ext_zeroout(inode, ex);
3007 zero_ex.ee_block = ex->ee_block; 3008 zero_ex.ee_block = ex->ee_block;
3008 zero_ex.ee_len = ext4_ext_get_actual_len(ex); 3009 zero_ex.ee_len = cpu_to_le16(
3010 ext4_ext_get_actual_len(ex));
3009 ext4_ext_store_pblock(&zero_ex, 3011 ext4_ext_store_pblock(&zero_ex,
3010 ext4_ext_pblock(ex)); 3012 ext4_ext_pblock(ex));
3011 } 3013 }
3012 } else { 3014 } else {
3013 err = ext4_ext_zeroout(inode, &orig_ex); 3015 err = ext4_ext_zeroout(inode, &orig_ex);
3014 zero_ex.ee_block = orig_ex.ee_block; 3016 zero_ex.ee_block = orig_ex.ee_block;
3015 zero_ex.ee_len = ext4_ext_get_actual_len(&orig_ex); 3017 zero_ex.ee_len = cpu_to_le16(
3018 ext4_ext_get_actual_len(&orig_ex));
3016 ext4_ext_store_pblock(&zero_ex, 3019 ext4_ext_store_pblock(&zero_ex,
3017 ext4_ext_pblock(&orig_ex)); 3020 ext4_ext_pblock(&orig_ex));
3018 } 3021 }
@@ -3272,7 +3275,7 @@ static int ext4_ext_convert_to_initialized(handle_t *handle,
3272 if (err) 3275 if (err)
3273 goto out; 3276 goto out;
3274 zero_ex.ee_block = ex->ee_block; 3277 zero_ex.ee_block = ex->ee_block;
3275 zero_ex.ee_len = ext4_ext_get_actual_len(ex); 3278 zero_ex.ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex));
3276 ext4_ext_store_pblock(&zero_ex, ext4_ext_pblock(ex)); 3279 ext4_ext_store_pblock(&zero_ex, ext4_ext_pblock(ex));
3277 3280
3278 err = ext4_ext_get_access(handle, inode, path + depth); 3281 err = ext4_ext_get_access(handle, inode, path + depth);
diff --git a/fs/ext4/indirect.c b/fs/ext4/indirect.c
index b505a145a593..a04183127ef0 100644
--- a/fs/ext4/indirect.c
+++ b/fs/ext4/indirect.c
@@ -1539,9 +1539,9 @@ static int free_hole_blocks(handle_t *handle, struct inode *inode,
1539 blk = *i_data; 1539 blk = *i_data;
1540 if (level > 0) { 1540 if (level > 0) {
1541 ext4_lblk_t first2; 1541 ext4_lblk_t first2;
1542 bh = sb_bread(inode->i_sb, blk); 1542 bh = sb_bread(inode->i_sb, le32_to_cpu(blk));
1543 if (!bh) { 1543 if (!bh) {
1544 EXT4_ERROR_INODE_BLOCK(inode, blk, 1544 EXT4_ERROR_INODE_BLOCK(inode, le32_to_cpu(blk),
1545 "Read failure"); 1545 "Read failure");
1546 return -EIO; 1546 return -EIO;
1547 } 1547 }
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index 019f45e45097..d79c2dadc536 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -923,8 +923,11 @@ static int gfs2_lock(struct file *file, int cmd, struct file_lock *fl)
923 cmd = F_SETLK; 923 cmd = F_SETLK;
924 fl->fl_type = F_UNLCK; 924 fl->fl_type = F_UNLCK;
925 } 925 }
926 if (unlikely(test_bit(SDF_SHUTDOWN, &sdp->sd_flags))) 926 if (unlikely(test_bit(SDF_SHUTDOWN, &sdp->sd_flags))) {
927 if (fl->fl_type == F_UNLCK)
928 posix_lock_file_wait(file, fl);
927 return -EIO; 929 return -EIO;
930 }
928 if (IS_GETLK(cmd)) 931 if (IS_GETLK(cmd))
929 return dlm_posix_get(ls->ls_dlm, ip->i_no_addr, file, fl); 932 return dlm_posix_get(ls->ls_dlm, ip->i_no_addr, file, fl);
930 else if (fl->fl_type == F_UNLCK) 933 else if (fl->fl_type == F_UNLCK)
diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h
index 156e42ec84ea..5c29216e9cc1 100644
--- a/fs/gfs2/incore.h
+++ b/fs/gfs2/incore.h
@@ -588,6 +588,7 @@ struct lm_lockstruct {
588 struct dlm_lksb ls_control_lksb; /* control_lock */ 588 struct dlm_lksb ls_control_lksb; /* control_lock */
589 char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */ 589 char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */
590 struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */ 590 struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */
591 char *ls_lvb_bits;
591 592
592 spinlock_t ls_recover_spin; /* protects following fields */ 593 spinlock_t ls_recover_spin; /* protects following fields */
593 unsigned long ls_recover_flags; /* DFL_ */ 594 unsigned long ls_recover_flags; /* DFL_ */
diff --git a/fs/gfs2/lock_dlm.c b/fs/gfs2/lock_dlm.c
index 9802de0f85e6..c8423d6de6c3 100644
--- a/fs/gfs2/lock_dlm.c
+++ b/fs/gfs2/lock_dlm.c
@@ -483,12 +483,8 @@ static void control_lvb_write(struct lm_lockstruct *ls, uint32_t lvb_gen,
483 483
484static int all_jid_bits_clear(char *lvb) 484static int all_jid_bits_clear(char *lvb)
485{ 485{
486 int i; 486 return !memchr_inv(lvb + JID_BITMAP_OFFSET, 0,
487 for (i = JID_BITMAP_OFFSET; i < GDLM_LVB_SIZE; i++) { 487 GDLM_LVB_SIZE - JID_BITMAP_OFFSET);
488 if (lvb[i])
489 return 0;
490 }
491 return 1;
492} 488}
493 489
494static void sync_wait_cb(void *arg) 490static void sync_wait_cb(void *arg)
@@ -580,7 +576,6 @@ static void gfs2_control_func(struct work_struct *work)
580{ 576{
581 struct gfs2_sbd *sdp = container_of(work, struct gfs2_sbd, sd_control_work.work); 577 struct gfs2_sbd *sdp = container_of(work, struct gfs2_sbd, sd_control_work.work);
582 struct lm_lockstruct *ls = &sdp->sd_lockstruct; 578 struct lm_lockstruct *ls = &sdp->sd_lockstruct;
583 char lvb_bits[GDLM_LVB_SIZE];
584 uint32_t block_gen, start_gen, lvb_gen, flags; 579 uint32_t block_gen, start_gen, lvb_gen, flags;
585 int recover_set = 0; 580 int recover_set = 0;
586 int write_lvb = 0; 581 int write_lvb = 0;
@@ -634,7 +629,7 @@ static void gfs2_control_func(struct work_struct *work)
634 return; 629 return;
635 } 630 }
636 631
637 control_lvb_read(ls, &lvb_gen, lvb_bits); 632 control_lvb_read(ls, &lvb_gen, ls->ls_lvb_bits);
638 633
639 spin_lock(&ls->ls_recover_spin); 634 spin_lock(&ls->ls_recover_spin);
640 if (block_gen != ls->ls_recover_block || 635 if (block_gen != ls->ls_recover_block ||
@@ -664,10 +659,10 @@ static void gfs2_control_func(struct work_struct *work)
664 659
665 ls->ls_recover_result[i] = 0; 660 ls->ls_recover_result[i] = 0;
666 661
667 if (!test_bit_le(i, lvb_bits + JID_BITMAP_OFFSET)) 662 if (!test_bit_le(i, ls->ls_lvb_bits + JID_BITMAP_OFFSET))
668 continue; 663 continue;
669 664
670 __clear_bit_le(i, lvb_bits + JID_BITMAP_OFFSET); 665 __clear_bit_le(i, ls->ls_lvb_bits + JID_BITMAP_OFFSET);
671 write_lvb = 1; 666 write_lvb = 1;
672 } 667 }
673 } 668 }
@@ -691,7 +686,7 @@ static void gfs2_control_func(struct work_struct *work)
691 continue; 686 continue;
692 if (ls->ls_recover_submit[i] < start_gen) { 687 if (ls->ls_recover_submit[i] < start_gen) {
693 ls->ls_recover_submit[i] = 0; 688 ls->ls_recover_submit[i] = 0;
694 __set_bit_le(i, lvb_bits + JID_BITMAP_OFFSET); 689 __set_bit_le(i, ls->ls_lvb_bits + JID_BITMAP_OFFSET);
695 } 690 }
696 } 691 }
697 /* even if there are no bits to set, we need to write the 692 /* even if there are no bits to set, we need to write the
@@ -705,7 +700,7 @@ static void gfs2_control_func(struct work_struct *work)
705 spin_unlock(&ls->ls_recover_spin); 700 spin_unlock(&ls->ls_recover_spin);
706 701
707 if (write_lvb) { 702 if (write_lvb) {
708 control_lvb_write(ls, start_gen, lvb_bits); 703 control_lvb_write(ls, start_gen, ls->ls_lvb_bits);
709 flags = DLM_LKF_CONVERT | DLM_LKF_VALBLK; 704 flags = DLM_LKF_CONVERT | DLM_LKF_VALBLK;
710 } else { 705 } else {
711 flags = DLM_LKF_CONVERT; 706 flags = DLM_LKF_CONVERT;
@@ -725,7 +720,7 @@ static void gfs2_control_func(struct work_struct *work)
725 */ 720 */
726 721
727 for (i = 0; i < recover_size; i++) { 722 for (i = 0; i < recover_size; i++) {
728 if (test_bit_le(i, lvb_bits + JID_BITMAP_OFFSET)) { 723 if (test_bit_le(i, ls->ls_lvb_bits + JID_BITMAP_OFFSET)) {
729 fs_info(sdp, "recover generation %u jid %d\n", 724 fs_info(sdp, "recover generation %u jid %d\n",
730 start_gen, i); 725 start_gen, i);
731 gfs2_recover_set(sdp, i); 726 gfs2_recover_set(sdp, i);
@@ -758,7 +753,6 @@ static void gfs2_control_func(struct work_struct *work)
758static int control_mount(struct gfs2_sbd *sdp) 753static int control_mount(struct gfs2_sbd *sdp)
759{ 754{
760 struct lm_lockstruct *ls = &sdp->sd_lockstruct; 755 struct lm_lockstruct *ls = &sdp->sd_lockstruct;
761 char lvb_bits[GDLM_LVB_SIZE];
762 uint32_t start_gen, block_gen, mount_gen, lvb_gen; 756 uint32_t start_gen, block_gen, mount_gen, lvb_gen;
763 int mounted_mode; 757 int mounted_mode;
764 int retries = 0; 758 int retries = 0;
@@ -857,7 +851,7 @@ locks_done:
857 * lvb_gen will be non-zero. 851 * lvb_gen will be non-zero.
858 */ 852 */
859 853
860 control_lvb_read(ls, &lvb_gen, lvb_bits); 854 control_lvb_read(ls, &lvb_gen, ls->ls_lvb_bits);
861 855
862 if (lvb_gen == 0xFFFFFFFF) { 856 if (lvb_gen == 0xFFFFFFFF) {
863 /* special value to force mount attempts to fail */ 857 /* special value to force mount attempts to fail */
@@ -887,7 +881,7 @@ locks_done:
887 * and all lvb bits to be clear (no pending journal recoveries.) 881 * and all lvb bits to be clear (no pending journal recoveries.)
888 */ 882 */
889 883
890 if (!all_jid_bits_clear(lvb_bits)) { 884 if (!all_jid_bits_clear(ls->ls_lvb_bits)) {
891 /* journals need recovery, wait until all are clear */ 885 /* journals need recovery, wait until all are clear */
892 fs_info(sdp, "control_mount wait for journal recovery\n"); 886 fs_info(sdp, "control_mount wait for journal recovery\n");
893 goto restart; 887 goto restart;
@@ -949,7 +943,6 @@ static int dlm_recovery_wait(void *word)
949static int control_first_done(struct gfs2_sbd *sdp) 943static int control_first_done(struct gfs2_sbd *sdp)
950{ 944{
951 struct lm_lockstruct *ls = &sdp->sd_lockstruct; 945 struct lm_lockstruct *ls = &sdp->sd_lockstruct;
952 char lvb_bits[GDLM_LVB_SIZE];
953 uint32_t start_gen, block_gen; 946 uint32_t start_gen, block_gen;
954 int error; 947 int error;
955 948
@@ -991,8 +984,8 @@ restart:
991 memset(ls->ls_recover_result, 0, ls->ls_recover_size*sizeof(uint32_t)); 984 memset(ls->ls_recover_result, 0, ls->ls_recover_size*sizeof(uint32_t));
992 spin_unlock(&ls->ls_recover_spin); 985 spin_unlock(&ls->ls_recover_spin);
993 986
994 memset(lvb_bits, 0, sizeof(lvb_bits)); 987 memset(ls->ls_lvb_bits, 0, GDLM_LVB_SIZE);
995 control_lvb_write(ls, start_gen, lvb_bits); 988 control_lvb_write(ls, start_gen, ls->ls_lvb_bits);
996 989
997 error = mounted_lock(sdp, DLM_LOCK_PR, DLM_LKF_CONVERT); 990 error = mounted_lock(sdp, DLM_LOCK_PR, DLM_LKF_CONVERT);
998 if (error) 991 if (error)
@@ -1022,6 +1015,12 @@ static int set_recover_size(struct gfs2_sbd *sdp, struct dlm_slot *slots,
1022 uint32_t old_size, new_size; 1015 uint32_t old_size, new_size;
1023 int i, max_jid; 1016 int i, max_jid;
1024 1017
1018 if (!ls->ls_lvb_bits) {
1019 ls->ls_lvb_bits = kzalloc(GDLM_LVB_SIZE, GFP_NOFS);
1020 if (!ls->ls_lvb_bits)
1021 return -ENOMEM;
1022 }
1023
1025 max_jid = 0; 1024 max_jid = 0;
1026 for (i = 0; i < num_slots; i++) { 1025 for (i = 0; i < num_slots; i++) {
1027 if (max_jid < slots[i].slot - 1) 1026 if (max_jid < slots[i].slot - 1)
@@ -1057,6 +1056,7 @@ static int set_recover_size(struct gfs2_sbd *sdp, struct dlm_slot *slots,
1057 1056
1058static void free_recover_size(struct lm_lockstruct *ls) 1057static void free_recover_size(struct lm_lockstruct *ls)
1059{ 1058{
1059 kfree(ls->ls_lvb_bits);
1060 kfree(ls->ls_recover_submit); 1060 kfree(ls->ls_recover_submit);
1061 kfree(ls->ls_recover_result); 1061 kfree(ls->ls_recover_result);
1062 ls->ls_recover_submit = NULL; 1062 ls->ls_recover_submit = NULL;
@@ -1205,6 +1205,7 @@ static int gdlm_mount(struct gfs2_sbd *sdp, const char *table)
1205 ls->ls_recover_size = 0; 1205 ls->ls_recover_size = 0;
1206 ls->ls_recover_submit = NULL; 1206 ls->ls_recover_submit = NULL;
1207 ls->ls_recover_result = NULL; 1207 ls->ls_recover_result = NULL;
1208 ls->ls_lvb_bits = NULL;
1208 1209
1209 error = set_recover_size(sdp, NULL, 0); 1210 error = set_recover_size(sdp, NULL, 0);
1210 if (error) 1211 if (error)
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index d1f51fd73f86..5a51265a4341 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -576,7 +576,7 @@ int gfs2_rs_alloc(struct gfs2_inode *ip)
576 RB_CLEAR_NODE(&ip->i_res->rs_node); 576 RB_CLEAR_NODE(&ip->i_res->rs_node);
577out: 577out:
578 up_write(&ip->i_rw_mutex); 578 up_write(&ip->i_rw_mutex);
579 return 0; 579 return error;
580} 580}
581 581
582static void dump_rs(struct seq_file *seq, const struct gfs2_blkreserv *rs) 582static void dump_rs(struct seq_file *seq, const struct gfs2_blkreserv *rs)
@@ -1181,12 +1181,9 @@ int gfs2_rgrp_send_discards(struct gfs2_sbd *sdp, u64 offset,
1181 const struct gfs2_bitmap *bi, unsigned minlen, u64 *ptrimmed) 1181 const struct gfs2_bitmap *bi, unsigned minlen, u64 *ptrimmed)
1182{ 1182{
1183 struct super_block *sb = sdp->sd_vfs; 1183 struct super_block *sb = sdp->sd_vfs;
1184 struct block_device *bdev = sb->s_bdev;
1185 const unsigned int sects_per_blk = sdp->sd_sb.sb_bsize /
1186 bdev_logical_block_size(sb->s_bdev);
1187 u64 blk; 1184 u64 blk;
1188 sector_t start = 0; 1185 sector_t start = 0;
1189 sector_t nr_sects = 0; 1186 sector_t nr_blks = 0;
1190 int rv; 1187 int rv;
1191 unsigned int x; 1188 unsigned int x;
1192 u32 trimmed = 0; 1189 u32 trimmed = 0;
@@ -1206,35 +1203,34 @@ int gfs2_rgrp_send_discards(struct gfs2_sbd *sdp, u64 offset,
1206 if (diff == 0) 1203 if (diff == 0)
1207 continue; 1204 continue;
1208 blk = offset + ((bi->bi_start + x) * GFS2_NBBY); 1205 blk = offset + ((bi->bi_start + x) * GFS2_NBBY);
1209 blk *= sects_per_blk; /* convert to sectors */
1210 while(diff) { 1206 while(diff) {
1211 if (diff & 1) { 1207 if (diff & 1) {
1212 if (nr_sects == 0) 1208 if (nr_blks == 0)
1213 goto start_new_extent; 1209 goto start_new_extent;
1214 if ((start + nr_sects) != blk) { 1210 if ((start + nr_blks) != blk) {
1215 if (nr_sects >= minlen) { 1211 if (nr_blks >= minlen) {
1216 rv = blkdev_issue_discard(bdev, 1212 rv = sb_issue_discard(sb,
1217 start, nr_sects, 1213 start, nr_blks,
1218 GFP_NOFS, 0); 1214 GFP_NOFS, 0);
1219 if (rv) 1215 if (rv)
1220 goto fail; 1216 goto fail;
1221 trimmed += nr_sects; 1217 trimmed += nr_blks;
1222 } 1218 }
1223 nr_sects = 0; 1219 nr_blks = 0;
1224start_new_extent: 1220start_new_extent:
1225 start = blk; 1221 start = blk;
1226 } 1222 }
1227 nr_sects += sects_per_blk; 1223 nr_blks++;
1228 } 1224 }
1229 diff >>= 2; 1225 diff >>= 2;
1230 blk += sects_per_blk; 1226 blk++;
1231 } 1227 }
1232 } 1228 }
1233 if (nr_sects >= minlen) { 1229 if (nr_blks >= minlen) {
1234 rv = blkdev_issue_discard(bdev, start, nr_sects, GFP_NOFS, 0); 1230 rv = sb_issue_discard(sb, start, nr_blks, GFP_NOFS, 0);
1235 if (rv) 1231 if (rv)
1236 goto fail; 1232 goto fail;
1237 trimmed += nr_sects; 1233 trimmed += nr_blks;
1238 } 1234 }
1239 if (ptrimmed) 1235 if (ptrimmed)
1240 *ptrimmed = trimmed; 1236 *ptrimmed = trimmed;
diff --git a/fs/internal.h b/fs/internal.h
index 507141fceb99..4be78237d896 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -125,3 +125,8 @@ extern int invalidate_inodes(struct super_block *, bool);
125 * dcache.c 125 * dcache.c
126 */ 126 */
127extern struct dentry *__d_alloc(struct super_block *, const struct qstr *); 127extern struct dentry *__d_alloc(struct super_block *, const struct qstr *);
128
129/*
130 * read_write.c
131 */
132extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
diff --git a/fs/namespace.c b/fs/namespace.c
index 50ca17d3cb45..d581e45c0a9f 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -798,6 +798,10 @@ static struct mount *clone_mnt(struct mount *old, struct dentry *root,
798 } 798 }
799 799
800 mnt->mnt.mnt_flags = old->mnt.mnt_flags & ~MNT_WRITE_HOLD; 800 mnt->mnt.mnt_flags = old->mnt.mnt_flags & ~MNT_WRITE_HOLD;
801 /* Don't allow unprivileged users to change mount flags */
802 if ((flag & CL_UNPRIVILEGED) && (mnt->mnt.mnt_flags & MNT_READONLY))
803 mnt->mnt.mnt_flags |= MNT_LOCK_READONLY;
804
801 atomic_inc(&sb->s_active); 805 atomic_inc(&sb->s_active);
802 mnt->mnt.mnt_sb = sb; 806 mnt->mnt.mnt_sb = sb;
803 mnt->mnt.mnt_root = dget(root); 807 mnt->mnt.mnt_root = dget(root);
@@ -1713,6 +1717,9 @@ static int change_mount_flags(struct vfsmount *mnt, int ms_flags)
1713 if (readonly_request == __mnt_is_readonly(mnt)) 1717 if (readonly_request == __mnt_is_readonly(mnt))
1714 return 0; 1718 return 0;
1715 1719
1720 if (mnt->mnt_flags & MNT_LOCK_READONLY)
1721 return -EPERM;
1722
1716 if (readonly_request) 1723 if (readonly_request)
1717 error = mnt_make_readonly(real_mount(mnt)); 1724 error = mnt_make_readonly(real_mount(mnt));
1718 else 1725 else
@@ -2339,7 +2346,7 @@ static struct mnt_namespace *dup_mnt_ns(struct mnt_namespace *mnt_ns,
2339 /* First pass: copy the tree topology */ 2346 /* First pass: copy the tree topology */
2340 copy_flags = CL_COPY_ALL | CL_EXPIRE; 2347 copy_flags = CL_COPY_ALL | CL_EXPIRE;
2341 if (user_ns != mnt_ns->user_ns) 2348 if (user_ns != mnt_ns->user_ns)
2342 copy_flags |= CL_SHARED_TO_SLAVE; 2349 copy_flags |= CL_SHARED_TO_SLAVE | CL_UNPRIVILEGED;
2343 new = copy_tree(old, old->mnt.mnt_root, copy_flags); 2350 new = copy_tree(old, old->mnt.mnt_root, copy_flags);
2344 if (IS_ERR(new)) { 2351 if (IS_ERR(new)) {
2345 up_write(&namespace_sem); 2352 up_write(&namespace_sem);
@@ -2732,6 +2739,51 @@ bool our_mnt(struct vfsmount *mnt)
2732 return check_mnt(real_mount(mnt)); 2739 return check_mnt(real_mount(mnt));
2733} 2740}
2734 2741
2742bool current_chrooted(void)
2743{
2744 /* Does the current process have a non-standard root */
2745 struct path ns_root;
2746 struct path fs_root;
2747 bool chrooted;
2748
2749 /* Find the namespace root */
2750 ns_root.mnt = &current->nsproxy->mnt_ns->root->mnt;
2751 ns_root.dentry = ns_root.mnt->mnt_root;
2752 path_get(&ns_root);
2753 while (d_mountpoint(ns_root.dentry) && follow_down_one(&ns_root))
2754 ;
2755
2756 get_fs_root(current->fs, &fs_root);
2757
2758 chrooted = !path_equal(&fs_root, &ns_root);
2759
2760 path_put(&fs_root);
2761 path_put(&ns_root);
2762
2763 return chrooted;
2764}
2765
2766void update_mnt_policy(struct user_namespace *userns)
2767{
2768 struct mnt_namespace *ns = current->nsproxy->mnt_ns;
2769 struct mount *mnt;
2770
2771 down_read(&namespace_sem);
2772 list_for_each_entry(mnt, &ns->list, mnt_list) {
2773 switch (mnt->mnt.mnt_sb->s_magic) {
2774 case SYSFS_MAGIC:
2775 userns->may_mount_sysfs = true;
2776 break;
2777 case PROC_SUPER_MAGIC:
2778 userns->may_mount_proc = true;
2779 break;
2780 }
2781 if (userns->may_mount_sysfs && userns->may_mount_proc)
2782 break;
2783 }
2784 up_read(&namespace_sem);
2785}
2786
2735static void *mntns_get(struct task_struct *task) 2787static void *mntns_get(struct task_struct *task)
2736{ 2788{
2737 struct mnt_namespace *ns = NULL; 2789 struct mnt_namespace *ns = NULL;
diff --git a/fs/nfs/blocklayout/blocklayoutdm.c b/fs/nfs/blocklayout/blocklayoutdm.c
index 737d839bc17b..6fc7b5cae92b 100644
--- a/fs/nfs/blocklayout/blocklayoutdm.c
+++ b/fs/nfs/blocklayout/blocklayoutdm.c
@@ -55,7 +55,8 @@ static void dev_remove(struct net *net, dev_t dev)
55 55
56 bl_pipe_msg.bl_wq = &nn->bl_wq; 56 bl_pipe_msg.bl_wq = &nn->bl_wq;
57 memset(msg, 0, sizeof(*msg)); 57 memset(msg, 0, sizeof(*msg));
58 msg->data = kzalloc(1 + sizeof(bl_umount_request), GFP_NOFS); 58 msg->len = sizeof(bl_msg) + bl_msg.totallen;
59 msg->data = kzalloc(msg->len, GFP_NOFS);
59 if (!msg->data) 60 if (!msg->data)
60 goto out; 61 goto out;
61 62
@@ -66,7 +67,6 @@ static void dev_remove(struct net *net, dev_t dev)
66 memcpy(msg->data, &bl_msg, sizeof(bl_msg)); 67 memcpy(msg->data, &bl_msg, sizeof(bl_msg));
67 dataptr = (uint8_t *) msg->data; 68 dataptr = (uint8_t *) msg->data;
68 memcpy(&dataptr[sizeof(bl_msg)], &bl_umount_request, sizeof(bl_umount_request)); 69 memcpy(&dataptr[sizeof(bl_msg)], &bl_umount_request, sizeof(bl_umount_request));
69 msg->len = sizeof(bl_msg) + bl_msg.totallen;
70 70
71 add_wait_queue(&nn->bl_wq, &wq); 71 add_wait_queue(&nn->bl_wq, &wq);
72 if (rpc_queue_upcall(nn->bl_device_pipe, msg) < 0) { 72 if (rpc_queue_upcall(nn->bl_device_pipe, msg) < 0) {
diff --git a/fs/nfs/idmap.c b/fs/nfs/idmap.c
index dc0f98dfa717..c516da5873fd 100644
--- a/fs/nfs/idmap.c
+++ b/fs/nfs/idmap.c
@@ -726,9 +726,9 @@ out1:
726 return ret; 726 return ret;
727} 727}
728 728
729static int nfs_idmap_instantiate(struct key *key, struct key *authkey, char *data) 729static int nfs_idmap_instantiate(struct key *key, struct key *authkey, char *data, size_t datalen)
730{ 730{
731 return key_instantiate_and_link(key, data, strlen(data) + 1, 731 return key_instantiate_and_link(key, data, datalen,
732 id_resolver_cache->thread_keyring, 732 id_resolver_cache->thread_keyring,
733 authkey); 733 authkey);
734} 734}
@@ -738,6 +738,7 @@ static int nfs_idmap_read_and_verify_message(struct idmap_msg *im,
738 struct key *key, struct key *authkey) 738 struct key *key, struct key *authkey)
739{ 739{
740 char id_str[NFS_UINT_MAXLEN]; 740 char id_str[NFS_UINT_MAXLEN];
741 size_t len;
741 int ret = -ENOKEY; 742 int ret = -ENOKEY;
742 743
743 /* ret = -ENOKEY */ 744 /* ret = -ENOKEY */
@@ -747,13 +748,15 @@ static int nfs_idmap_read_and_verify_message(struct idmap_msg *im,
747 case IDMAP_CONV_NAMETOID: 748 case IDMAP_CONV_NAMETOID:
748 if (strcmp(upcall->im_name, im->im_name) != 0) 749 if (strcmp(upcall->im_name, im->im_name) != 0)
749 break; 750 break;
750 sprintf(id_str, "%d", im->im_id); 751 /* Note: here we store the NUL terminator too */
751 ret = nfs_idmap_instantiate(key, authkey, id_str); 752 len = sprintf(id_str, "%d", im->im_id) + 1;
753 ret = nfs_idmap_instantiate(key, authkey, id_str, len);
752 break; 754 break;
753 case IDMAP_CONV_IDTONAME: 755 case IDMAP_CONV_IDTONAME:
754 if (upcall->im_id != im->im_id) 756 if (upcall->im_id != im->im_id)
755 break; 757 break;
756 ret = nfs_idmap_instantiate(key, authkey, im->im_name); 758 len = strlen(im->im_name);
759 ret = nfs_idmap_instantiate(key, authkey, im->im_name, len);
757 break; 760 break;
758 default: 761 default:
759 ret = -EINVAL; 762 ret = -EINVAL;
diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c
index 49eeb044c109..4fb234d3aefb 100644
--- a/fs/nfs/nfs4filelayout.c
+++ b/fs/nfs/nfs4filelayout.c
@@ -129,7 +129,6 @@ static void filelayout_fenceme(struct inode *inode, struct pnfs_layout_hdr *lo)
129{ 129{
130 if (!test_and_clear_bit(NFS_LAYOUT_RETURN, &lo->plh_flags)) 130 if (!test_and_clear_bit(NFS_LAYOUT_RETURN, &lo->plh_flags))
131 return; 131 return;
132 clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(inode)->flags);
133 pnfs_return_layout(inode); 132 pnfs_return_layout(inode);
134} 133}
135 134
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index b2671cb0f901..26431cf62ddb 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -2632,7 +2632,7 @@ nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
2632 int status; 2632 int status;
2633 2633
2634 if (pnfs_ld_layoutret_on_setattr(inode)) 2634 if (pnfs_ld_layoutret_on_setattr(inode))
2635 pnfs_return_layout(inode); 2635 pnfs_commit_and_return_layout(inode);
2636 2636
2637 nfs_fattr_init(fattr); 2637 nfs_fattr_init(fattr);
2638 2638
@@ -6416,22 +6416,8 @@ nfs4_layoutcommit_done(struct rpc_task *task, void *calldata)
6416static void nfs4_layoutcommit_release(void *calldata) 6416static void nfs4_layoutcommit_release(void *calldata)
6417{ 6417{
6418 struct nfs4_layoutcommit_data *data = calldata; 6418 struct nfs4_layoutcommit_data *data = calldata;
6419 struct pnfs_layout_segment *lseg, *tmp;
6420 unsigned long *bitlock = &NFS_I(data->args.inode)->flags;
6421 6419
6422 pnfs_cleanup_layoutcommit(data); 6420 pnfs_cleanup_layoutcommit(data);
6423 /* Matched by references in pnfs_set_layoutcommit */
6424 list_for_each_entry_safe(lseg, tmp, &data->lseg_list, pls_lc_list) {
6425 list_del_init(&lseg->pls_lc_list);
6426 if (test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT,
6427 &lseg->pls_flags))
6428 pnfs_put_lseg(lseg);
6429 }
6430
6431 clear_bit_unlock(NFS_INO_LAYOUTCOMMITTING, bitlock);
6432 smp_mb__after_clear_bit();
6433 wake_up_bit(bitlock, NFS_INO_LAYOUTCOMMITTING);
6434
6435 put_rpccred(data->cred); 6421 put_rpccred(data->cred);
6436 kfree(data); 6422 kfree(data);
6437} 6423}
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 48ac5aad6258..4bdffe0ba025 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -417,6 +417,16 @@ should_free_lseg(struct pnfs_layout_range *lseg_range,
417 lo_seg_intersecting(lseg_range, recall_range); 417 lo_seg_intersecting(lseg_range, recall_range);
418} 418}
419 419
420static bool pnfs_lseg_dec_and_remove_zero(struct pnfs_layout_segment *lseg,
421 struct list_head *tmp_list)
422{
423 if (!atomic_dec_and_test(&lseg->pls_refcount))
424 return false;
425 pnfs_layout_remove_lseg(lseg->pls_layout, lseg);
426 list_add(&lseg->pls_list, tmp_list);
427 return true;
428}
429
420/* Returns 1 if lseg is removed from list, 0 otherwise */ 430/* Returns 1 if lseg is removed from list, 0 otherwise */
421static int mark_lseg_invalid(struct pnfs_layout_segment *lseg, 431static int mark_lseg_invalid(struct pnfs_layout_segment *lseg,
422 struct list_head *tmp_list) 432 struct list_head *tmp_list)
@@ -430,11 +440,8 @@ static int mark_lseg_invalid(struct pnfs_layout_segment *lseg,
430 */ 440 */
431 dprintk("%s: lseg %p ref %d\n", __func__, lseg, 441 dprintk("%s: lseg %p ref %d\n", __func__, lseg,
432 atomic_read(&lseg->pls_refcount)); 442 atomic_read(&lseg->pls_refcount));
433 if (atomic_dec_and_test(&lseg->pls_refcount)) { 443 if (pnfs_lseg_dec_and_remove_zero(lseg, tmp_list))
434 pnfs_layout_remove_lseg(lseg->pls_layout, lseg);
435 list_add(&lseg->pls_list, tmp_list);
436 rv = 1; 444 rv = 1;
437 }
438 } 445 }
439 return rv; 446 return rv;
440} 447}
@@ -777,6 +784,21 @@ send_layoutget(struct pnfs_layout_hdr *lo,
777 return lseg; 784 return lseg;
778} 785}
779 786
787static void pnfs_clear_layoutcommit(struct inode *inode,
788 struct list_head *head)
789{
790 struct nfs_inode *nfsi = NFS_I(inode);
791 struct pnfs_layout_segment *lseg, *tmp;
792
793 if (!test_and_clear_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags))
794 return;
795 list_for_each_entry_safe(lseg, tmp, &nfsi->layout->plh_segs, pls_list) {
796 if (!test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags))
797 continue;
798 pnfs_lseg_dec_and_remove_zero(lseg, head);
799 }
800}
801
780/* 802/*
781 * Initiates a LAYOUTRETURN(FILE), and removes the pnfs_layout_hdr 803 * Initiates a LAYOUTRETURN(FILE), and removes the pnfs_layout_hdr
782 * when the layout segment list is empty. 804 * when the layout segment list is empty.
@@ -808,6 +830,7 @@ _pnfs_return_layout(struct inode *ino)
808 /* Reference matched in nfs4_layoutreturn_release */ 830 /* Reference matched in nfs4_layoutreturn_release */
809 pnfs_get_layout_hdr(lo); 831 pnfs_get_layout_hdr(lo);
810 empty = list_empty(&lo->plh_segs); 832 empty = list_empty(&lo->plh_segs);
833 pnfs_clear_layoutcommit(ino, &tmp_list);
811 pnfs_mark_matching_lsegs_invalid(lo, &tmp_list, NULL); 834 pnfs_mark_matching_lsegs_invalid(lo, &tmp_list, NULL);
812 /* Don't send a LAYOUTRETURN if list was initially empty */ 835 /* Don't send a LAYOUTRETURN if list was initially empty */
813 if (empty) { 836 if (empty) {
@@ -820,8 +843,6 @@ _pnfs_return_layout(struct inode *ino)
820 spin_unlock(&ino->i_lock); 843 spin_unlock(&ino->i_lock);
821 pnfs_free_lseg_list(&tmp_list); 844 pnfs_free_lseg_list(&tmp_list);
822 845
823 WARN_ON(test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags));
824
825 lrp = kzalloc(sizeof(*lrp), GFP_KERNEL); 846 lrp = kzalloc(sizeof(*lrp), GFP_KERNEL);
826 if (unlikely(lrp == NULL)) { 847 if (unlikely(lrp == NULL)) {
827 status = -ENOMEM; 848 status = -ENOMEM;
@@ -845,6 +866,33 @@ out:
845} 866}
846EXPORT_SYMBOL_GPL(_pnfs_return_layout); 867EXPORT_SYMBOL_GPL(_pnfs_return_layout);
847 868
869int
870pnfs_commit_and_return_layout(struct inode *inode)
871{
872 struct pnfs_layout_hdr *lo;
873 int ret;
874
875 spin_lock(&inode->i_lock);
876 lo = NFS_I(inode)->layout;
877 if (lo == NULL) {
878 spin_unlock(&inode->i_lock);
879 return 0;
880 }
881 pnfs_get_layout_hdr(lo);
882 /* Block new layoutgets and read/write to ds */
883 lo->plh_block_lgets++;
884 spin_unlock(&inode->i_lock);
885 filemap_fdatawait(inode->i_mapping);
886 ret = pnfs_layoutcommit_inode(inode, true);
887 if (ret == 0)
888 ret = _pnfs_return_layout(inode);
889 spin_lock(&inode->i_lock);
890 lo->plh_block_lgets--;
891 spin_unlock(&inode->i_lock);
892 pnfs_put_layout_hdr(lo);
893 return ret;
894}
895
848bool pnfs_roc(struct inode *ino) 896bool pnfs_roc(struct inode *ino)
849{ 897{
850 struct pnfs_layout_hdr *lo; 898 struct pnfs_layout_hdr *lo;
@@ -1458,7 +1506,6 @@ static void pnfs_ld_handle_write_error(struct nfs_write_data *data)
1458 dprintk("pnfs write error = %d\n", hdr->pnfs_error); 1506 dprintk("pnfs write error = %d\n", hdr->pnfs_error);
1459 if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags & 1507 if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags &
1460 PNFS_LAYOUTRET_ON_ERROR) { 1508 PNFS_LAYOUTRET_ON_ERROR) {
1461 clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(hdr->inode)->flags);
1462 pnfs_return_layout(hdr->inode); 1509 pnfs_return_layout(hdr->inode);
1463 } 1510 }
1464 if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) 1511 if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags))
@@ -1613,7 +1660,6 @@ static void pnfs_ld_handle_read_error(struct nfs_read_data *data)
1613 dprintk("pnfs read error = %d\n", hdr->pnfs_error); 1660 dprintk("pnfs read error = %d\n", hdr->pnfs_error);
1614 if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags & 1661 if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags &
1615 PNFS_LAYOUTRET_ON_ERROR) { 1662 PNFS_LAYOUTRET_ON_ERROR) {
1616 clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(hdr->inode)->flags);
1617 pnfs_return_layout(hdr->inode); 1663 pnfs_return_layout(hdr->inode);
1618 } 1664 }
1619 if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) 1665 if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags))
@@ -1746,11 +1792,27 @@ static void pnfs_list_write_lseg(struct inode *inode, struct list_head *listp)
1746 1792
1747 list_for_each_entry(lseg, &NFS_I(inode)->layout->plh_segs, pls_list) { 1793 list_for_each_entry(lseg, &NFS_I(inode)->layout->plh_segs, pls_list) {
1748 if (lseg->pls_range.iomode == IOMODE_RW && 1794 if (lseg->pls_range.iomode == IOMODE_RW &&
1749 test_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags)) 1795 test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags))
1750 list_add(&lseg->pls_lc_list, listp); 1796 list_add(&lseg->pls_lc_list, listp);
1751 } 1797 }
1752} 1798}
1753 1799
1800static void pnfs_list_write_lseg_done(struct inode *inode, struct list_head *listp)
1801{
1802 struct pnfs_layout_segment *lseg, *tmp;
1803 unsigned long *bitlock = &NFS_I(inode)->flags;
1804
1805 /* Matched by references in pnfs_set_layoutcommit */
1806 list_for_each_entry_safe(lseg, tmp, listp, pls_lc_list) {
1807 list_del_init(&lseg->pls_lc_list);
1808 pnfs_put_lseg(lseg);
1809 }
1810
1811 clear_bit_unlock(NFS_INO_LAYOUTCOMMITTING, bitlock);
1812 smp_mb__after_clear_bit();
1813 wake_up_bit(bitlock, NFS_INO_LAYOUTCOMMITTING);
1814}
1815
1754void pnfs_set_lo_fail(struct pnfs_layout_segment *lseg) 1816void pnfs_set_lo_fail(struct pnfs_layout_segment *lseg)
1755{ 1817{
1756 pnfs_layout_io_set_failed(lseg->pls_layout, lseg->pls_range.iomode); 1818 pnfs_layout_io_set_failed(lseg->pls_layout, lseg->pls_range.iomode);
@@ -1795,6 +1857,7 @@ void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data)
1795 1857
1796 if (nfss->pnfs_curr_ld->cleanup_layoutcommit) 1858 if (nfss->pnfs_curr_ld->cleanup_layoutcommit)
1797 nfss->pnfs_curr_ld->cleanup_layoutcommit(data); 1859 nfss->pnfs_curr_ld->cleanup_layoutcommit(data);
1860 pnfs_list_write_lseg_done(data->args.inode, &data->lseg_list);
1798} 1861}
1799 1862
1800/* 1863/*
diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index 94ba80417748..f5f8a470a647 100644
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -219,6 +219,7 @@ void pnfs_set_layoutcommit(struct nfs_write_data *wdata);
219void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data); 219void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data);
220int pnfs_layoutcommit_inode(struct inode *inode, bool sync); 220int pnfs_layoutcommit_inode(struct inode *inode, bool sync);
221int _pnfs_return_layout(struct inode *); 221int _pnfs_return_layout(struct inode *);
222int pnfs_commit_and_return_layout(struct inode *);
222void pnfs_ld_write_done(struct nfs_write_data *); 223void pnfs_ld_write_done(struct nfs_write_data *);
223void pnfs_ld_read_done(struct nfs_read_data *); 224void pnfs_ld_read_done(struct nfs_read_data *);
224struct pnfs_layout_segment *pnfs_update_layout(struct inode *ino, 225struct pnfs_layout_segment *pnfs_update_layout(struct inode *ino,
@@ -407,6 +408,11 @@ static inline int pnfs_return_layout(struct inode *ino)
407 return 0; 408 return 0;
408} 409}
409 410
411static inline int pnfs_commit_and_return_layout(struct inode *inode)
412{
413 return 0;
414}
415
410static inline bool 416static inline bool
411pnfs_ld_layoutret_on_setattr(struct inode *inode) 417pnfs_ld_layoutret_on_setattr(struct inode *inode)
412{ 418{
diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index 01168865dd37..a2720071f282 100644
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -264,7 +264,7 @@ nfsd4_decode_fattr(struct nfsd4_compoundargs *argp, u32 *bmval,
264 iattr->ia_valid |= ATTR_SIZE; 264 iattr->ia_valid |= ATTR_SIZE;
265 } 265 }
266 if (bmval[0] & FATTR4_WORD0_ACL) { 266 if (bmval[0] & FATTR4_WORD0_ACL) {
267 int nace; 267 u32 nace;
268 struct nfs4_ace *ace; 268 struct nfs4_ace *ace;
269 269
270 READ_BUF(4); len += 4; 270 READ_BUF(4); len += 4;
diff --git a/fs/nfsd/nfscache.c b/fs/nfsd/nfscache.c
index 62c1ee128aeb..ca05f6dc3544 100644
--- a/fs/nfsd/nfscache.c
+++ b/fs/nfsd/nfscache.c
@@ -102,7 +102,8 @@ nfsd_reply_cache_free_locked(struct svc_cacherep *rp)
102{ 102{
103 if (rp->c_type == RC_REPLBUFF) 103 if (rp->c_type == RC_REPLBUFF)
104 kfree(rp->c_replvec.iov_base); 104 kfree(rp->c_replvec.iov_base);
105 hlist_del(&rp->c_hash); 105 if (!hlist_unhashed(&rp->c_hash))
106 hlist_del(&rp->c_hash);
106 list_del(&rp->c_lru); 107 list_del(&rp->c_lru);
107 --num_drc_entries; 108 --num_drc_entries;
108 kmem_cache_free(drc_slab, rp); 109 kmem_cache_free(drc_slab, rp);
@@ -118,6 +119,10 @@ nfsd_reply_cache_free(struct svc_cacherep *rp)
118 119
119int nfsd_reply_cache_init(void) 120int nfsd_reply_cache_init(void)
120{ 121{
122 INIT_LIST_HEAD(&lru_head);
123 max_drc_entries = nfsd_cache_size_limit();
124 num_drc_entries = 0;
125
121 register_shrinker(&nfsd_reply_cache_shrinker); 126 register_shrinker(&nfsd_reply_cache_shrinker);
122 drc_slab = kmem_cache_create("nfsd_drc", sizeof(struct svc_cacherep), 127 drc_slab = kmem_cache_create("nfsd_drc", sizeof(struct svc_cacherep),
123 0, 0, NULL); 128 0, 0, NULL);
@@ -128,10 +133,6 @@ int nfsd_reply_cache_init(void)
128 if (!cache_hash) 133 if (!cache_hash)
129 goto out_nomem; 134 goto out_nomem;
130 135
131 INIT_LIST_HEAD(&lru_head);
132 max_drc_entries = nfsd_cache_size_limit();
133 num_drc_entries = 0;
134
135 return 0; 136 return 0;
136out_nomem: 137out_nomem:
137 printk(KERN_ERR "nfsd: failed to allocate reply cache\n"); 138 printk(KERN_ERR "nfsd: failed to allocate reply cache\n");
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 2a7eb536de0b..2b2e2396a869 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -1013,6 +1013,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1013 int host_err; 1013 int host_err;
1014 int stable = *stablep; 1014 int stable = *stablep;
1015 int use_wgather; 1015 int use_wgather;
1016 loff_t pos = offset;
1016 1017
1017 dentry = file->f_path.dentry; 1018 dentry = file->f_path.dentry;
1018 inode = dentry->d_inode; 1019 inode = dentry->d_inode;
@@ -1025,7 +1026,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
1025 1026
1026 /* Write the data. */ 1027 /* Write the data. */
1027 oldfs = get_fs(); set_fs(KERNEL_DS); 1028 oldfs = get_fs(); set_fs(KERNEL_DS);
1028 host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &offset); 1029 host_err = vfs_writev(file, (struct iovec __user *)vec, vlen, &pos);
1029 set_fs(oldfs); 1030 set_fs(oldfs);
1030 if (host_err < 0) 1031 if (host_err < 0)
1031 goto out_nfserr; 1032 goto out_nfserr;
diff --git a/fs/pnode.c b/fs/pnode.c
index 3e000a51ac0d..8b29d2164da6 100644
--- a/fs/pnode.c
+++ b/fs/pnode.c
@@ -9,6 +9,7 @@
9#include <linux/mnt_namespace.h> 9#include <linux/mnt_namespace.h>
10#include <linux/mount.h> 10#include <linux/mount.h>
11#include <linux/fs.h> 11#include <linux/fs.h>
12#include <linux/nsproxy.h>
12#include "internal.h" 13#include "internal.h"
13#include "pnode.h" 14#include "pnode.h"
14 15
@@ -220,6 +221,7 @@ static struct mount *get_source(struct mount *dest,
220int propagate_mnt(struct mount *dest_mnt, struct dentry *dest_dentry, 221int propagate_mnt(struct mount *dest_mnt, struct dentry *dest_dentry,
221 struct mount *source_mnt, struct list_head *tree_list) 222 struct mount *source_mnt, struct list_head *tree_list)
222{ 223{
224 struct user_namespace *user_ns = current->nsproxy->mnt_ns->user_ns;
223 struct mount *m, *child; 225 struct mount *m, *child;
224 int ret = 0; 226 int ret = 0;
225 struct mount *prev_dest_mnt = dest_mnt; 227 struct mount *prev_dest_mnt = dest_mnt;
@@ -237,6 +239,10 @@ int propagate_mnt(struct mount *dest_mnt, struct dentry *dest_dentry,
237 239
238 source = get_source(m, prev_dest_mnt, prev_src_mnt, &type); 240 source = get_source(m, prev_dest_mnt, prev_src_mnt, &type);
239 241
242 /* Notice when we are propagating across user namespaces */
243 if (m->mnt_ns->user_ns != user_ns)
244 type |= CL_UNPRIVILEGED;
245
240 child = copy_tree(source, source->mnt.mnt_root, type); 246 child = copy_tree(source, source->mnt.mnt_root, type);
241 if (IS_ERR(child)) { 247 if (IS_ERR(child)) {
242 ret = PTR_ERR(child); 248 ret = PTR_ERR(child);
diff --git a/fs/pnode.h b/fs/pnode.h
index 19b853a3445c..a0493d5ebfbf 100644
--- a/fs/pnode.h
+++ b/fs/pnode.h
@@ -23,6 +23,7 @@
23#define CL_MAKE_SHARED 0x08 23#define CL_MAKE_SHARED 0x08
24#define CL_PRIVATE 0x10 24#define CL_PRIVATE 0x10
25#define CL_SHARED_TO_SLAVE 0x20 25#define CL_SHARED_TO_SLAVE 0x20
26#define CL_UNPRIVILEGED 0x40
26 27
27static inline void set_mnt_shared(struct mount *mnt) 28static inline void set_mnt_shared(struct mount *mnt)
28{ 29{
diff --git a/fs/proc/root.c b/fs/proc/root.c
index c6e9fac26bac..9c7fab1d23f0 100644
--- a/fs/proc/root.c
+++ b/fs/proc/root.c
@@ -16,6 +16,7 @@
16#include <linux/sched.h> 16#include <linux/sched.h>
17#include <linux/module.h> 17#include <linux/module.h>
18#include <linux/bitops.h> 18#include <linux/bitops.h>
19#include <linux/user_namespace.h>
19#include <linux/mount.h> 20#include <linux/mount.h>
20#include <linux/pid_namespace.h> 21#include <linux/pid_namespace.h>
21#include <linux/parser.h> 22#include <linux/parser.h>
@@ -108,6 +109,9 @@ static struct dentry *proc_mount(struct file_system_type *fs_type,
108 } else { 109 } else {
109 ns = task_active_pid_ns(current); 110 ns = task_active_pid_ns(current);
110 options = data; 111 options = data;
112
113 if (!current_user_ns()->may_mount_proc)
114 return ERR_PTR(-EPERM);
111 } 115 }
112 116
113 sb = sget(fs_type, proc_test_super, proc_set_super, flags, ns); 117 sb = sget(fs_type, proc_test_super, proc_set_super, flags, ns);
diff --git a/fs/read_write.c b/fs/read_write.c
index a698eff457fb..e6ddc8dceb96 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -17,6 +17,7 @@
17#include <linux/splice.h> 17#include <linux/splice.h>
18#include <linux/compat.h> 18#include <linux/compat.h>
19#include "read_write.h" 19#include "read_write.h"
20#include "internal.h"
20 21
21#include <asm/uaccess.h> 22#include <asm/uaccess.h>
22#include <asm/unistd.h> 23#include <asm/unistd.h>
@@ -417,6 +418,33 @@ ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, lof
417 418
418EXPORT_SYMBOL(do_sync_write); 419EXPORT_SYMBOL(do_sync_write);
419 420
421ssize_t __kernel_write(struct file *file, const char *buf, size_t count, loff_t *pos)
422{
423 mm_segment_t old_fs;
424 const char __user *p;
425 ssize_t ret;
426
427 if (!file->f_op || (!file->f_op->write && !file->f_op->aio_write))
428 return -EINVAL;
429
430 old_fs = get_fs();
431 set_fs(get_ds());
432 p = (__force const char __user *)buf;
433 if (count > MAX_RW_COUNT)
434 count = MAX_RW_COUNT;
435 if (file->f_op->write)
436 ret = file->f_op->write(file, p, count, pos);
437 else
438 ret = do_sync_write(file, p, count, pos);
439 set_fs(old_fs);
440 if (ret > 0) {
441 fsnotify_modify(file);
442 add_wchar(current, ret);
443 }
444 inc_syscw(current);
445 return ret;
446}
447
420ssize_t vfs_write(struct file *file, const char __user *buf, size_t count, loff_t *pos) 448ssize_t vfs_write(struct file *file, const char __user *buf, size_t count, loff_t *pos)
421{ 449{
422 ssize_t ret; 450 ssize_t ret;
diff --git a/fs/reiserfs/xattr.c b/fs/reiserfs/xattr.c
index c196369fe408..4cce1d9552fb 100644
--- a/fs/reiserfs/xattr.c
+++ b/fs/reiserfs/xattr.c
@@ -187,8 +187,8 @@ fill_with_dentries(void *buf, const char *name, int namelen, loff_t offset,
187 if (dbuf->count == ARRAY_SIZE(dbuf->dentries)) 187 if (dbuf->count == ARRAY_SIZE(dbuf->dentries))
188 return -ENOSPC; 188 return -ENOSPC;
189 189
190 if (name[0] == '.' && (name[1] == '\0' || 190 if (name[0] == '.' && (namelen < 2 ||
191 (name[1] == '.' && name[2] == '\0'))) 191 (namelen == 2 && name[1] == '.')))
192 return 0; 192 return 0;
193 193
194 dentry = lookup_one_len(name, dbuf->xadir, namelen); 194 dentry = lookup_one_len(name, dbuf->xadir, namelen);
diff --git a/fs/splice.c b/fs/splice.c
index 718bd0056384..29e394e49ddd 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -31,6 +31,7 @@
31#include <linux/security.h> 31#include <linux/security.h>
32#include <linux/gfp.h> 32#include <linux/gfp.h>
33#include <linux/socket.h> 33#include <linux/socket.h>
34#include "internal.h"
34 35
35/* 36/*
36 * Attempt to steal a page from a pipe buffer. This should perhaps go into 37 * Attempt to steal a page from a pipe buffer. This should perhaps go into
@@ -1048,9 +1049,10 @@ static int write_pipe_buf(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
1048{ 1049{
1049 int ret; 1050 int ret;
1050 void *data; 1051 void *data;
1052 loff_t tmp = sd->pos;
1051 1053
1052 data = buf->ops->map(pipe, buf, 0); 1054 data = buf->ops->map(pipe, buf, 0);
1053 ret = kernel_write(sd->u.file, data + buf->offset, sd->len, sd->pos); 1055 ret = __kernel_write(sd->u.file, data + buf->offset, sd->len, &tmp);
1054 buf->ops->unmap(pipe, buf, data); 1056 buf->ops->unmap(pipe, buf, data);
1055 1057
1056 return ret; 1058 return ret;
diff --git a/fs/sysfs/dir.c b/fs/sysfs/dir.c
index 2fbdff6be25c..e14512678c9b 100644
--- a/fs/sysfs/dir.c
+++ b/fs/sysfs/dir.c
@@ -1020,6 +1020,8 @@ static int sysfs_readdir(struct file * filp, void * dirent, filldir_t filldir)
1020 ino = parent_sd->s_ino; 1020 ino = parent_sd->s_ino;
1021 if (filldir(dirent, ".", 1, filp->f_pos, ino, DT_DIR) == 0) 1021 if (filldir(dirent, ".", 1, filp->f_pos, ino, DT_DIR) == 0)
1022 filp->f_pos++; 1022 filp->f_pos++;
1023 else
1024 return 0;
1023 } 1025 }
1024 if (filp->f_pos == 1) { 1026 if (filp->f_pos == 1) {
1025 if (parent_sd->s_parent) 1027 if (parent_sd->s_parent)
@@ -1028,6 +1030,8 @@ static int sysfs_readdir(struct file * filp, void * dirent, filldir_t filldir)
1028 ino = parent_sd->s_ino; 1030 ino = parent_sd->s_ino;
1029 if (filldir(dirent, "..", 2, filp->f_pos, ino, DT_DIR) == 0) 1031 if (filldir(dirent, "..", 2, filp->f_pos, ino, DT_DIR) == 0)
1030 filp->f_pos++; 1032 filp->f_pos++;
1033 else
1034 return 0;
1031 } 1035 }
1032 mutex_lock(&sysfs_mutex); 1036 mutex_lock(&sysfs_mutex);
1033 for (pos = sysfs_dir_pos(ns, parent_sd, filp->f_pos, pos); 1037 for (pos = sysfs_dir_pos(ns, parent_sd, filp->f_pos, pos);
@@ -1058,10 +1062,21 @@ static int sysfs_readdir(struct file * filp, void * dirent, filldir_t filldir)
1058 return 0; 1062 return 0;
1059} 1063}
1060 1064
1065static loff_t sysfs_dir_llseek(struct file *file, loff_t offset, int whence)
1066{
1067 struct inode *inode = file_inode(file);
1068 loff_t ret;
1069
1070 mutex_lock(&inode->i_mutex);
1071 ret = generic_file_llseek(file, offset, whence);
1072 mutex_unlock(&inode->i_mutex);
1073
1074 return ret;
1075}
1061 1076
1062const struct file_operations sysfs_dir_operations = { 1077const struct file_operations sysfs_dir_operations = {
1063 .read = generic_read_dir, 1078 .read = generic_read_dir,
1064 .readdir = sysfs_readdir, 1079 .readdir = sysfs_readdir,
1065 .release = sysfs_dir_release, 1080 .release = sysfs_dir_release,
1066 .llseek = generic_file_llseek, 1081 .llseek = sysfs_dir_llseek,
1067}; 1082};
diff --git a/fs/sysfs/mount.c b/fs/sysfs/mount.c
index 8d924b5ec733..afd83273e6ce 100644
--- a/fs/sysfs/mount.c
+++ b/fs/sysfs/mount.c
@@ -19,6 +19,7 @@
19#include <linux/module.h> 19#include <linux/module.h>
20#include <linux/magic.h> 20#include <linux/magic.h>
21#include <linux/slab.h> 21#include <linux/slab.h>
22#include <linux/user_namespace.h>
22 23
23#include "sysfs.h" 24#include "sysfs.h"
24 25
@@ -111,6 +112,9 @@ static struct dentry *sysfs_mount(struct file_system_type *fs_type,
111 struct super_block *sb; 112 struct super_block *sb;
112 int error; 113 int error;
113 114
115 if (!(flags & MS_KERNMOUNT) && !current_user_ns()->may_mount_sysfs)
116 return ERR_PTR(-EPERM);
117
114 info = kzalloc(sizeof(*info), GFP_KERNEL); 118 info = kzalloc(sizeof(*info), GFP_KERNEL);
115 if (!info) 119 if (!info)
116 return ERR_PTR(-ENOMEM); 120 return ERR_PTR(-ENOMEM);
diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index ac838b844936..f21acf0ef01f 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -1568,6 +1568,12 @@ static int ubifs_remount_rw(struct ubifs_info *c)
1568 c->remounting_rw = 1; 1568 c->remounting_rw = 1;
1569 c->ro_mount = 0; 1569 c->ro_mount = 0;
1570 1570
1571 if (c->space_fixup) {
1572 err = ubifs_fixup_free_space(c);
1573 if (err)
1574 return err;
1575 }
1576
1571 err = check_free_space(c); 1577 err = check_free_space(c);
1572 if (err) 1578 if (err)
1573 goto out; 1579 goto out;
@@ -1684,12 +1690,6 @@ static int ubifs_remount_rw(struct ubifs_info *c)
1684 err = dbg_check_space_info(c); 1690 err = dbg_check_space_info(c);
1685 } 1691 }
1686 1692
1687 if (c->space_fixup) {
1688 err = ubifs_fixup_free_space(c);
1689 if (err)
1690 goto out;
1691 }
1692
1693 mutex_unlock(&c->umount_mutex); 1693 mutex_unlock(&c->umount_mutex);
1694 return err; 1694 return err;
1695 1695