From 45ea6095c8f0d6caad5658306416a5d254f1205e Mon Sep 17 00:00:00 2001 From: "slyich@gmail.com" Date: Mon, 7 Nov 2011 16:08:01 -0500 Subject: btrfs: fix double-free 'tree_root' in 'btrfs_mount()' On error path 'tree_root' is treed in 'free_fs_info()'. No need to free it explicitely. Noticed by SLUB in debug mode: Complete reproducer under usermode linux (discovered on real machine): bdev=/dev/ubda btr_root=/btr /mkfs.btrfs $bdev mount $bdev $btr_root mkdir $btr_root/subvols/ cd $btr_root/subvols/ /btrfs su cr foo /btrfs su cr bar mount $bdev -osubvol=subvols/foo $btr_root/subvols/bar umount $btr_root/subvols/bar which gives device fsid 4d55aa28-45b1-474b-b4ec-da912322195e devid 1 transid 7 /dev/ubda ============================================================================= BUG kmalloc-2048: Object already free ----------------------------------------------------------------------------- INFO: Allocated in btrfs_mount+0x389/0x7f0 age=0 cpu=0 pid=277 INFO: Freed in btrfs_mount+0x51c/0x7f0 age=0 cpu=0 pid=277 INFO: Slab 0x0000000062886200 objects=15 used=9 fp=0x0000000070b4d2d0 flags=0x4081 INFO: Object 0x0000000070b4d2d0 @offset=21200 fp=0x0000000070b4a968 ... Call Trace: 70b31948: [<6008c522>] print_trailer+0xe2/0x130 70b31978: [<6008c5aa>] object_err+0x3a/0x50 70b319a8: [<6008e242>] free_debug_processing+0x142/0x2a0 70b319e0: [<600ebf6f>] btrfs_mount+0x55f/0x7f0 70b319f8: [<6008e5c1>] __slab_free+0x221/0x2d0 Signed-off-by: Sergei Trofimovich Cc: Arne Jansen Cc: Chris Mason Cc: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/super.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 57080dffdfc6..dcd5aef6b614 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -933,8 +933,12 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, * then open_ctree will properly initialize everything later. */ fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS); + if (!fs_info) { + error = -ENOMEM; + goto error_close_devices; + } tree_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS); - if (!fs_info || !tree_root) { + if (!tree_root) { error = -ENOMEM; goto error_close_devices; } @@ -964,7 +968,6 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, btrfs_close_devices(fs_devices); free_fs_info(fs_info); - kfree(tree_root); } else { char b[BDEVNAME_SIZE]; @@ -992,7 +995,6 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, error_close_devices: btrfs_close_devices(fs_devices); free_fs_info(fs_info); - kfree(tree_root); return ERR_PTR(error); } -- cgit v1.2.3 From 917c16b2b69fc2eeb432eabca73258f08c58361e Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Tue, 8 Nov 2011 14:49:59 -0500 Subject: Btrfs: fix oops on NULL trans handle in btrfs_truncate If we fail to reserve space in the transaction during truncate, we can error out with a NULL trans handle. The cleanup code needs an extra check to make sure we aren't trying to use the bad handle. Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 9d0eaa57d4ee..f60e2490bd0d 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6529,14 +6529,16 @@ end_trans: ret = btrfs_orphan_del(NULL, inode); } - trans->block_rsv = &root->fs_info->trans_block_rsv; - ret = btrfs_update_inode(trans, root, inode); - if (ret && !err) - err = ret; + if (trans) { + trans->block_rsv = &root->fs_info->trans_block_rsv; + ret = btrfs_update_inode(trans, root, inode); + if (ret && !err) + err = ret; - nr = trans->blocks_used; - ret = btrfs_end_transaction_throttle(trans, root); - btrfs_btree_balance_dirty(root, nr); + nr = trans->blocks_used; + ret = btrfs_end_transaction_throttle(trans, root); + btrfs_btree_balance_dirty(root, nr); + } out: btrfs_free_block_rsv(root, rsv); -- cgit v1.2.3 From 7fd2ae21a42d178982679b86086661292b4afe4a Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 8 Nov 2011 15:47:34 -0500 Subject: Btrfs: fix our reservations for updating an inode when completing io People have been reporting ENOSPC crashes in finish_ordered_io. This is because we try to steal from the delalloc block rsv to satisfy a reservation to update the inode. The problem with this is we don't explicitly save space for updating the inode when doing delalloc. This is kind of a problem and we've gotten away with this because way back when we just stole from the delalloc reserve without any questions, and this worked out fine because generally speaking the leaf had been modified either by the mtime update when we did the original write or because we just updated the leaf when we inserted the file extent item, only on rare occasions had the leaf not actually been modified, and that was still ok because we'd just use a block or two out of the over-reservation that is delalloc. Then came the delayed inode stuff. This is amazing, except it wants a full reservation for updating the inode since it may do it at some point down the road after we've written the blocks and we have to recow everything again. This worked out because the delayed inode stuff just stole from the global reserve, that is until recently when I changed that because it caused other problems. So here we are, we're doing everything right and being screwed for it. So take an extra reservation for the inode at delalloc reservation time and carry it through the life of the delalloc reservation. If we need it we can steal it in the delayed inode stuff. If we have already stolen it try and do a normal metadata reservation. If that fails try to steal from the delalloc reservation. If _that_ fails we'll get a WARN_ON() so I can start thinking of a better way to solve this and in the meantime we'll steal from the global reserve. With this patch I ran xfstests 13 in a loop for a couple of hours and didn't see any problems. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/btrfs_inode.h | 4 +-- fs/btrfs/delayed-inode.c | 65 +++++++++++++++++++++++++++++++++++++++++++++--- fs/btrfs/extent-tree.c | 22 +++++++++++++--- fs/btrfs/inode.c | 1 + 4 files changed, 83 insertions(+), 9 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h index 5a5d325a3935..634608d2a6d0 100644 --- a/fs/btrfs/btrfs_inode.h +++ b/fs/btrfs/btrfs_inode.h @@ -147,14 +147,12 @@ struct btrfs_inode { * the btrfs file release call will add this inode to the * ordered operations list so that we make sure to flush out any * new data the application may have written before commit. - * - * yes, its silly to have a single bitflag, but we might grow more - * of these. */ unsigned ordered_data_close:1; unsigned orphan_meta_reserved:1; unsigned dummy_inode:1; unsigned in_defrag:1; + unsigned delalloc_meta_reserved:1; /* * always compress this one file diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c index bbe8496d5339..313ee14cf3b7 100644 --- a/fs/btrfs/delayed-inode.c +++ b/fs/btrfs/delayed-inode.c @@ -617,12 +617,14 @@ static void btrfs_delayed_item_release_metadata(struct btrfs_root *root, static int btrfs_delayed_inode_reserve_metadata( struct btrfs_trans_handle *trans, struct btrfs_root *root, + struct inode *inode, struct btrfs_delayed_node *node) { struct btrfs_block_rsv *src_rsv; struct btrfs_block_rsv *dst_rsv; u64 num_bytes; int ret; + int release = false; src_rsv = trans->block_rsv; dst_rsv = &root->fs_info->delayed_block_rsv; @@ -652,11 +654,67 @@ static int btrfs_delayed_inode_reserve_metadata( if (!ret) node->bytes_reserved = num_bytes; return ret; + } else if (src_rsv == &root->fs_info->delalloc_block_rsv) { + spin_lock(&BTRFS_I(inode)->lock); + if (BTRFS_I(inode)->delalloc_meta_reserved) { + BTRFS_I(inode)->delalloc_meta_reserved = 0; + spin_unlock(&BTRFS_I(inode)->lock); + release = true; + goto migrate; + } + spin_unlock(&BTRFS_I(inode)->lock); + + /* Ok we didn't have space pre-reserved. This shouldn't happen + * too often but it can happen if we do delalloc to an existing + * inode which gets dirtied because of the time update, and then + * isn't touched again until after the transaction commits and + * then we try to write out the data. First try to be nice and + * reserve something strictly for us. If not be a pain and try + * to steal from the delalloc block rsv. + */ + ret = btrfs_block_rsv_add_noflush(root, dst_rsv, num_bytes); + if (!ret) + goto out; + + ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes); + if (!ret) + goto out; + + /* + * Ok this is a problem, let's just steal from the global rsv + * since this really shouldn't happen that often. + */ + WARN_ON(1); + ret = btrfs_block_rsv_migrate(&root->fs_info->global_block_rsv, + dst_rsv, num_bytes); + goto out; } +migrate: ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes); - if (!ret) - node->bytes_reserved = num_bytes; + if (unlikely(ret)) { + /* This shouldn't happen */ + BUG_ON(release); + return ret; + } + +out: + /* + * Migrate only takes a reservation, it doesn't touch the size of the + * block_rsv. This is to simplify people who don't normally have things + * migrated from their block rsv. If they go to release their + * reservation, that will decrease the size as well, so if migrate + * reduced size we'd end up with a negative size. But for the + * delalloc_meta_reserved stuff we will only know to drop 1 reservation, + * but we could in fact do this reserve/migrate dance several times + * between the time we did the original reservation and we'd clean it + * up. So to take care of this, release the space for the meta + * reservation here. I think it may be time for a documentation page on + * how block rsvs. work. + */ + if (release) + btrfs_block_rsv_release(root, src_rsv, num_bytes); + node->bytes_reserved = num_bytes; return ret; } @@ -1708,7 +1766,8 @@ int btrfs_delayed_update_inode(struct btrfs_trans_handle *trans, goto release_node; } - ret = btrfs_delayed_inode_reserve_metadata(trans, root, delayed_node); + ret = btrfs_delayed_inode_reserve_metadata(trans, root, inode, + delayed_node); if (ret) goto release_node; diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 18ea90c8943b..0b044e509e9f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4063,23 +4063,30 @@ int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans, */ static unsigned drop_outstanding_extent(struct inode *inode) { + unsigned drop_inode_space = 0; unsigned dropped_extents = 0; BUG_ON(!BTRFS_I(inode)->outstanding_extents); BTRFS_I(inode)->outstanding_extents--; + if (BTRFS_I(inode)->outstanding_extents == 0 && + BTRFS_I(inode)->delalloc_meta_reserved) { + drop_inode_space = 1; + BTRFS_I(inode)->delalloc_meta_reserved = 0; + } + /* * If we have more or the same amount of outsanding extents than we have * reserved then we need to leave the reserved extents count alone. */ if (BTRFS_I(inode)->outstanding_extents >= BTRFS_I(inode)->reserved_extents) - return 0; + return drop_inode_space; dropped_extents = BTRFS_I(inode)->reserved_extents - BTRFS_I(inode)->outstanding_extents; BTRFS_I(inode)->reserved_extents -= dropped_extents; - return dropped_extents; + return dropped_extents + drop_inode_space; } /** @@ -4165,9 +4172,18 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes) nr_extents = BTRFS_I(inode)->outstanding_extents - BTRFS_I(inode)->reserved_extents; BTRFS_I(inode)->reserved_extents += nr_extents; + } - to_reserve = btrfs_calc_trans_metadata_size(root, nr_extents); + /* + * Add an item to reserve for updating the inode when we complete the + * delalloc io. + */ + if (!BTRFS_I(inode)->delalloc_meta_reserved) { + nr_extents++; + BTRFS_I(inode)->delalloc_meta_reserved = 1; } + + to_reserve = btrfs_calc_trans_metadata_size(root, nr_extents); to_reserve += calc_csum_metadata_size(inode, num_bytes, 1); spin_unlock(&BTRFS_I(inode)->lock); diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index f60e2490bd0d..2b920596c126 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6607,6 +6607,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb) ei->orphan_meta_reserved = 0; ei->dummy_inode = 0; ei->in_defrag = 0; + ei->delalloc_meta_reserved = 0; ei->force_compress = BTRFS_COMPRESS_NONE; ei->delayed_node = NULL; -- cgit v1.2.3 From a90e8b6fb80db43b029e1e76205452afa8bdc77a Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Tue, 8 Nov 2011 16:47:55 +0200 Subject: Btrfs: fix memory leak in btrfs_parse_early_options() Don't leak subvol_name string in case multiple subvol= options are given. "The lastest option is effective" behavior (consistent with subvolid= and subvolrootid= options) is preserved. Signed-off-by: Ilya Dryomov --- fs/btrfs/super.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index dcd5aef6b614..6befcaf253bd 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -448,6 +448,7 @@ static int btrfs_parse_early_options(const char *options, fmode_t flags, token = match_token(p, tokens, args); switch (token) { case Opt_subvol: + kfree(*subvol_name); *subvol_name = match_strdup(&args[0]); break; case Opt_subvolid: -- cgit v1.2.3 From f23c8af8ca2789eeb0ab9ea90c214f9694d96cc5 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Tue, 8 Nov 2011 19:15:05 +0200 Subject: Btrfs: fix subvol_name leak on error in btrfs_mount() btrfs_parse_early_options() can fail due to error while scanning devices (-o device= option), but still strdup() subvol_name string: mount -o subvol=SUBV,device=BAD_DEVICE So free subvol_name string on error. Signed-off-by: Ilya Dryomov --- fs/btrfs/super.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 6befcaf253bd..58e9492230ce 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -905,8 +905,10 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, error = btrfs_parse_early_options(data, mode, fs_type, &subvol_name, &subvol_objectid, &subvol_rootid, &fs_devices); - if (error) + if (error) { + kfree(subvol_name); return ERR_PTR(error); + } if (subvol_name) { root = mount_subvol(subvol_name, flags, device_name, data); -- cgit v1.2.3 From 4d34b2789538befa45a68a191dc12e0886a69f7d Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Wed, 9 Nov 2011 00:08:15 +0200 Subject: Btrfs: avoid null dereference and leaks when bailing from open_ctree() Fix bugs introduced by 6c41761f. Firstly, after failing to allocate any of the tree roots (first 'goto fail' in open_ctree()) we would dereference a NULL fs_info pointer in free_fs_info(). Secondly, after failures from init_srcu_struct(), setup_bdi() and new_inode() we would leak all earlier allocated roots: fs_info fields haven't been initialized yet so free_fs_info() is rendered useless. Fix this by initializing fs_info pointer and fs_info fields before any allocations happen. Signed-off-by: Ilya Dryomov --- fs/btrfs/disk-io.c | 35 +++++++++++++++-------------------- 1 file changed, 15 insertions(+), 20 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index e53a5bb85670..91db90b526c2 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1890,31 +1890,32 @@ struct btrfs_root *open_ctree(struct super_block *sb, u64 features; struct btrfs_key location; struct buffer_head *bh; - struct btrfs_root *extent_root = kzalloc(sizeof(struct btrfs_root), - GFP_NOFS); - struct btrfs_root *csum_root = kzalloc(sizeof(struct btrfs_root), - GFP_NOFS); + struct btrfs_super_block *disk_super; struct btrfs_root *tree_root = btrfs_sb(sb); - struct btrfs_fs_info *fs_info = NULL; - struct btrfs_root *chunk_root = kzalloc(sizeof(struct btrfs_root), - GFP_NOFS); - struct btrfs_root *dev_root = kzalloc(sizeof(struct btrfs_root), - GFP_NOFS); + struct btrfs_fs_info *fs_info = tree_root->fs_info; + struct btrfs_root *extent_root; + struct btrfs_root *csum_root; + struct btrfs_root *chunk_root; + struct btrfs_root *dev_root; struct btrfs_root *log_tree_root; - int ret; int err = -EINVAL; int num_backups_tried = 0; int backup_index = 0; - struct btrfs_super_block *disk_super; + extent_root = fs_info->extent_root = + kzalloc(sizeof(struct btrfs_root), GFP_NOFS); + csum_root = fs_info->csum_root = + kzalloc(sizeof(struct btrfs_root), GFP_NOFS); + chunk_root = fs_info->chunk_root = + kzalloc(sizeof(struct btrfs_root), GFP_NOFS); + dev_root = fs_info->dev_root = + kzalloc(sizeof(struct btrfs_root), GFP_NOFS); - if (!extent_root || !tree_root || !tree_root->fs_info || - !chunk_root || !dev_root || !csum_root) { + if (!extent_root || !csum_root || !chunk_root || !dev_root) { err = -ENOMEM; goto fail; } - fs_info = tree_root->fs_info; ret = init_srcu_struct(&fs_info->subvol_srcu); if (ret) { @@ -1954,12 +1955,6 @@ struct btrfs_root *open_ctree(struct super_block *sb, mutex_init(&fs_info->reloc_mutex); init_completion(&fs_info->kobj_unregister); - fs_info->tree_root = tree_root; - fs_info->extent_root = extent_root; - fs_info->csum_root = csum_root; - fs_info->chunk_root = chunk_root; - fs_info->dev_root = dev_root; - fs_info->fs_devices = fs_devices; INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots); INIT_LIST_HEAD(&fs_info->space_info); btrfs_mapping_init(&fs_info->mapping_tree); -- cgit v1.2.3 From 586e46e2813c589d26258a599580421fb6fb576b Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Wed, 9 Nov 2011 13:26:37 +0200 Subject: Btrfs: close devices on all error paths in open_ctree() Fix a bug introduced by 7e662854 where we would leave devices busy on certain error paths in open_ctree(). fs_info is guaranteed to be non-NULL now so it's safe to dereference it on all error paths. Signed-off-by: Ilya Dryomov --- fs/btrfs/disk-io.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 91db90b526c2..b6a5c0dd0dd8 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2460,21 +2460,20 @@ fail_sb_buffer: btrfs_stop_workers(&fs_info->caching_workers); fail_alloc: fail_iput: + btrfs_mapping_tree_free(&fs_info->mapping_tree); + invalidate_inode_pages2(fs_info->btree_inode->i_mapping); iput(fs_info->btree_inode); - - btrfs_close_devices(fs_info->fs_devices); - btrfs_mapping_tree_free(&fs_info->mapping_tree); fail_bdi: bdi_destroy(&fs_info->bdi); fail_srcu: cleanup_srcu_struct(&fs_info->subvol_srcu); fail: + btrfs_close_devices(fs_info->fs_devices); free_fs_info(fs_info); return ERR_PTR(err); recovery_tree_root: - if (!btrfs_test_opt(tree_root, RECOVERY)) goto fail_tree_roots; -- cgit v1.2.3 From 04d21a244fdf79d0ac892eaaa9a46b682467277c Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Wed, 9 Nov 2011 14:41:22 +0200 Subject: Btrfs: rework error handling in btrfs_mount() Commits 6c41761f and 45ea6095 introduced the possibility of NULL pointer dereference on error paths, also we would leave all devices busy and leak fs_info with all sub-structures on error when trying to mount an already mounted fs to a different directory. Fix this by doing all allocations before trying to open any of the devices, adjust error path for mount-already-mounted-fs case. Signed-off-by: Ilya Dryomov --- fs/btrfs/super.c | 42 +++++++++++++++++++++--------------------- 1 file changed, 21 insertions(+), 21 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 58e9492230ce..629281c65ff5 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -891,7 +891,6 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, struct super_block *s; struct dentry *root; struct btrfs_fs_devices *fs_devices = NULL; - struct btrfs_root *tree_root = NULL; struct btrfs_fs_info *fs_info = NULL; fmode_t mode = FMODE_READ; char *subvol_name = NULL; @@ -920,15 +919,6 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, if (error) return ERR_PTR(error); - error = btrfs_open_devices(fs_devices, mode, fs_type); - if (error) - return ERR_PTR(error); - - if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) { - error = -EACCES; - goto error_close_devices; - } - /* * Setup a dummy root and fs_info for test/set super. This is because * we don't actually fill this stuff out until open_ctree, but we need @@ -936,28 +926,36 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, * then open_ctree will properly initialize everything later. */ fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS); - if (!fs_info) { - error = -ENOMEM; - goto error_close_devices; - } - tree_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS); - if (!tree_root) { + if (!fs_info) + return ERR_PTR(-ENOMEM); + + fs_info->tree_root = kzalloc(sizeof(struct btrfs_root), GFP_NOFS); + if (!fs_info->tree_root) { error = -ENOMEM; - goto error_close_devices; + goto error_fs_info; } - fs_info->tree_root = tree_root; + fs_info->tree_root->fs_info = fs_info; fs_info->fs_devices = fs_devices; - tree_root->fs_info = fs_info; fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS); fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS); if (!fs_info->super_copy || !fs_info->super_for_commit) { error = -ENOMEM; + goto error_fs_info; + } + + error = btrfs_open_devices(fs_devices, mode, fs_type); + if (error) + goto error_fs_info; + + if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) { + error = -EACCES; goto error_close_devices; } bdev = fs_devices->latest_bdev; - s = sget(fs_type, btrfs_test_super, btrfs_set_super, tree_root); + s = sget(fs_type, btrfs_test_super, btrfs_set_super, + fs_info->tree_root); if (IS_ERR(s)) { error = PTR_ERR(s); goto error_close_devices; @@ -966,7 +964,8 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, if (s->s_root) { if ((flags ^ s->s_flags) & MS_RDONLY) { deactivate_locked_super(s); - return ERR_PTR(-EBUSY); + error = -EBUSY; + goto error_close_devices; } btrfs_close_devices(fs_devices); @@ -997,6 +996,7 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, error_close_devices: btrfs_close_devices(fs_devices); +error_fs_info: free_fs_info(fs_info); return ERR_PTR(error); } -- cgit v1.2.3 From 2115133f8b9a8dbdb217d14080814df07ce90479 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 10 Nov 2011 20:39:08 -0500 Subject: Btrfs: tweak the delayed inode reservations again Josef sent along an incremental to the inode reservation code to make sure we try and fall back to directly updating the inode item if things go horribly wrong. This reworks that patch slightly, adding a fallback function that will always try to update the inode item directly without going through the delayed_inode code. Signed-off-by: Chris Mason --- fs/btrfs/delayed-inode.c | 9 +++---- fs/btrfs/inode.c | 64 +++++++++++++++++++++++++++++++++--------------- 2 files changed, 47 insertions(+), 26 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c index 313ee14cf3b7..6a1a6800776c 100644 --- a/fs/btrfs/delayed-inode.c +++ b/fs/btrfs/delayed-inode.c @@ -692,11 +692,6 @@ static int btrfs_delayed_inode_reserve_metadata( migrate: ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes); - if (unlikely(ret)) { - /* This shouldn't happen */ - BUG_ON(release); - return ret; - } out: /* @@ -712,9 +707,11 @@ out: * reservation here. I think it may be time for a documentation page on * how block rsvs. work. */ + if (!ret) + node->bytes_reserved = num_bytes; + if (release) btrfs_block_rsv_release(root, src_rsv, num_bytes); - node->bytes_reserved = num_bytes; return ret; } diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 2b920596c126..7d394335bcb5 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -93,6 +93,8 @@ static noinline int cow_file_range(struct inode *inode, struct page *locked_page, u64 start, u64 end, int *page_started, unsigned long *nr_written, int unlock); +static noinline int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct inode *inode); static int btrfs_init_inode_security(struct btrfs_trans_handle *trans, struct inode *inode, struct inode *dir, @@ -1741,7 +1743,7 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) trans = btrfs_join_transaction(root); BUG_ON(IS_ERR(trans)); trans->block_rsv = &root->fs_info->delalloc_block_rsv; - ret = btrfs_update_inode(trans, root, inode); + ret = btrfs_update_inode_fallback(trans, root, inode); BUG_ON(ret); } goto out; @@ -1791,7 +1793,7 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent); if (!ret || !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) { - ret = btrfs_update_inode(trans, root, inode); + ret = btrfs_update_inode_fallback(trans, root, inode); BUG_ON(ret); } ret = 0; @@ -2426,7 +2428,7 @@ static void fill_inode_item(struct btrfs_trans_handle *trans, /* * copy everything in the in-memory inode into the btree. */ -noinline int btrfs_update_inode(struct btrfs_trans_handle *trans, +static noinline int btrfs_update_inode_item(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *inode) { struct btrfs_inode_item *inode_item; @@ -2434,21 +2436,6 @@ noinline int btrfs_update_inode(struct btrfs_trans_handle *trans, struct extent_buffer *leaf; int ret; - /* - * If the inode is a free space inode, we can deadlock during commit - * if we put it into the delayed code. - * - * The data relocation inode should also be directly updated - * without delay - */ - if (!btrfs_is_free_space_inode(root, inode) - && root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID) { - ret = btrfs_delayed_update_inode(trans, root, inode); - if (!ret) - btrfs_set_inode_last_trans(trans, inode); - return ret; - } - path = btrfs_alloc_path(); if (!path) return -ENOMEM; @@ -2476,6 +2463,43 @@ failed: return ret; } +/* + * copy everything in the in-memory inode into the btree. + */ +noinline int btrfs_update_inode(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct inode *inode) +{ + int ret; + + /* + * If the inode is a free space inode, we can deadlock during commit + * if we put it into the delayed code. + * + * The data relocation inode should also be directly updated + * without delay + */ + if (!btrfs_is_free_space_inode(root, inode) + && root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID) { + ret = btrfs_delayed_update_inode(trans, root, inode); + if (!ret) + btrfs_set_inode_last_trans(trans, inode); + return ret; + } + + return btrfs_update_inode_item(trans, root, inode); +} + +static noinline int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct inode *inode) +{ + int ret; + + ret = btrfs_update_inode(trans, root, inode); + if (ret == -ENOSPC) + return btrfs_update_inode_item(trans, root, inode); + return ret; +} + /* * unlink helper that gets used here in inode.c and in the tree logging * recovery code. It remove a link in a directory with a given name, and @@ -5632,7 +5656,7 @@ again: if (test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags)) { ret = btrfs_ordered_update_i_size(inode, 0, ordered); if (!ret) - err = btrfs_update_inode(trans, root, inode); + err = btrfs_update_inode_fallback(trans, root, inode); goto out; } @@ -5670,7 +5694,7 @@ again: add_pending_csums(trans, inode, ordered->file_offset, &ordered->list); ret = btrfs_ordered_update_i_size(inode, 0, ordered); if (!ret || !test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags)) - btrfs_update_inode(trans, root, inode); + btrfs_update_inode_fallback(trans, root, inode); ret = 0; out_unlock: unlock_extent_cached(&BTRFS_I(inode)->io_tree, ordered->file_offset, -- cgit v1.2.3 From 924cd8fbe41851eda2b68bf2ed501b2777fd77b4 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 10 Nov 2011 20:45:04 -0500 Subject: Btrfs: fix nocow when deleting the item btrfs_previous_item() just search the b+ tree, do not COW the nodes or leaves, if we modify the result of it, the meta-data will be broken. fix it. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index f8e2943101a1..c37433d3cd82 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -999,7 +999,7 @@ static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans, key.objectid = device->devid; key.offset = start; key.type = BTRFS_DEV_EXTENT_KEY; - +again: ret = btrfs_search_slot(trans, root, &key, path, -1, 1); if (ret > 0) { ret = btrfs_previous_item(root, path, key.objectid, @@ -1012,6 +1012,9 @@ static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans, struct btrfs_dev_extent); BUG_ON(found_key.offset > start || found_key.offset + btrfs_dev_extent_length(leaf, extent) < start); + key = found_key; + btrfs_release_path(path); + goto again; } else if (ret == 0) { leaf = path->nodes[0]; extent = btrfs_item_ptr(leaf, path->slots[0], -- cgit v1.2.3 From ba38eb4de354d228f2792f93cde2c748a3a3f3b2 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 10 Nov 2011 20:45:04 -0500 Subject: Btrfs: fix no reserved space for writing out inode cache I-node cache forgets to reserve the space when writing out it. And when we do some stress test, such as synctest, it will trigger WARN_ON() in use_block_rsv(). WARNING: at fs/btrfs/extent-tree.c:5718 btrfs_alloc_free_block+0xbf/0x281 [btrfs]() ... Call Trace: [] warn_slowpath_common+0x80/0x98 [] warn_slowpath_null+0x15/0x17 [] btrfs_alloc_free_block+0xbf/0x281 [btrfs] [] ? __set_page_dirty_nobuffers+0xfe/0x108 [] __btrfs_cow_block+0x118/0x3b5 [btrfs] [] btrfs_cow_block+0x103/0x14e [btrfs] [] btrfs_search_slot+0x249/0x6a4 [btrfs] [] btrfs_lookup_inode+0x2a/0x8a [btrfs] [] btrfs_update_inode+0xaa/0x141 [btrfs] [] btrfs_save_ino_cache+0xea/0x202 [btrfs] [] ? btrfs_update_reloc_root+0x17e/0x197 [btrfs] [] commit_fs_roots+0xaa/0x158 [btrfs] [] btrfs_commit_transaction+0x405/0x731 [btrfs] [] ? wake_up_bit+0x25/0x25 [] ? btrfs_log_dentry_safe+0x43/0x51 [btrfs] [] btrfs_sync_file+0x16a/0x198 [btrfs] [] ? mntput+0x21/0x23 [] vfs_fsync_range+0x18/0x21 [] vfs_fsync+0x17/0x19 [] do_fsync+0x29/0x3e [] sys_fsync+0xb/0xf [] system_call_fastpath+0x16/0x1b Sometimes it causes BUG_ON() in the reservation code of the delayed inode is triggered. So we must reserve enough space for inode cache. Note: If we can not reserve the enough space for inode cache, we will give up writing out it. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/inode-map.c | 28 ++++++++++++++++++++++++---- 1 file changed, 24 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c index 53dcbdf446cd..f8962a957d65 100644 --- a/fs/btrfs/inode-map.c +++ b/fs/btrfs/inode-map.c @@ -398,6 +398,8 @@ int btrfs_save_ino_cache(struct btrfs_root *root, struct btrfs_free_space_ctl *ctl = root->free_ino_ctl; struct btrfs_path *path; struct inode *inode; + struct btrfs_block_rsv *rsv; + u64 num_bytes; u64 alloc_hint = 0; int ret; int prealloc; @@ -421,11 +423,26 @@ int btrfs_save_ino_cache(struct btrfs_root *root, if (!path) return -ENOMEM; + rsv = trans->block_rsv; + trans->block_rsv = &root->fs_info->trans_block_rsv; + + num_bytes = trans->bytes_reserved; + /* + * 1 item for inode item insertion if need + * 3 items for inode item update (in the worst case) + * 1 item for free space object + * 3 items for pre-allocation + */ + trans->bytes_reserved = btrfs_calc_trans_metadata_size(root, 8); + ret = btrfs_block_rsv_add_noflush(root, trans->block_rsv, + trans->bytes_reserved); + if (ret) + goto out; again: inode = lookup_free_ino_inode(root, path); if (IS_ERR(inode) && PTR_ERR(inode) != -ENOENT) { ret = PTR_ERR(inode); - goto out; + goto out_release; } if (IS_ERR(inode)) { @@ -434,7 +451,7 @@ again: ret = create_free_ino_inode(root, trans, path); if (ret) - goto out; + goto out_release; goto again; } @@ -477,11 +494,14 @@ again: } btrfs_free_reserved_data_space(inode, prealloc); + ret = btrfs_write_out_ino_cache(root, trans, path); out_put: iput(inode); +out_release: + btrfs_block_rsv_release(root, trans->block_rsv, trans->bytes_reserved); out: - if (ret == 0) - ret = btrfs_write_out_ino_cache(root, trans, path); + trans->block_rsv = rsv; + trans->bytes_reserved = num_bytes; btrfs_free_path(path); return ret; -- cgit v1.2.3 From 3254c87618354e58fa2a7b375c6664f567480c33 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 10 Nov 2011 20:45:05 -0500 Subject: Btrfs: fix unreleased path in btrfs_orphan_cleanup() When we did stress test for the space relocation, the deadlock happened. By debugging, We found it was caused by the carelessness that we forgot to unlock the read lock of the extent buffers in btrfs_orphan_cleanup() before we end the transaction handle, so the transaction commit task waited the task, which called btrfs_orphan_cleanup(), to unlock the extent buffer, but that task waited the commit task to end the transaction commit, and the deadlock happened. Fix it. Signed-ff-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 7d394335bcb5..e16215f480d0 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -2201,6 +2201,9 @@ int btrfs_orphan_cleanup(struct btrfs_root *root) if (ret) goto out; } + /* release the path since we're done with it */ + btrfs_release_path(path); + root->orphan_cleanup_state = ORPHAN_CLEANUP_DONE; if (root->orphan_block_rsv) -- cgit v1.2.3 From 61b520a9d0083b9b361638e456af45fd75150c87 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 10 Nov 2011 20:45:05 -0500 Subject: Btrfs: Abstract similar code for btrfs_block_rsv_add{, _noflush} btrfs_block_rsv_add{, _noflush}() have similar code, so abstract that code. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 28 ++++++++++++---------------- 1 file changed, 12 insertions(+), 16 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 0b044e509e9f..0f47b3e2010e 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3796,16 +3796,16 @@ void btrfs_free_block_rsv(struct btrfs_root *root, kfree(rsv); } -int btrfs_block_rsv_add(struct btrfs_root *root, - struct btrfs_block_rsv *block_rsv, - u64 num_bytes) +static inline int __block_rsv_add(struct btrfs_root *root, + struct btrfs_block_rsv *block_rsv, + u64 num_bytes, int flush) { int ret; if (num_bytes == 0) return 0; - ret = reserve_metadata_bytes(root, block_rsv, num_bytes, 1); + ret = reserve_metadata_bytes(root, block_rsv, num_bytes, flush); if (!ret) { block_rsv_add_bytes(block_rsv, num_bytes, 1); return 0; @@ -3814,22 +3814,18 @@ int btrfs_block_rsv_add(struct btrfs_root *root, return ret; } +int btrfs_block_rsv_add(struct btrfs_root *root, + struct btrfs_block_rsv *block_rsv, + u64 num_bytes) +{ + return __block_rsv_add(root, block_rsv, num_bytes, 1); +} + int btrfs_block_rsv_add_noflush(struct btrfs_root *root, struct btrfs_block_rsv *block_rsv, u64 num_bytes) { - int ret; - - if (num_bytes == 0) - return 0; - - ret = reserve_metadata_bytes(root, block_rsv, num_bytes, 0); - if (!ret) { - block_rsv_add_bytes(block_rsv, num_bytes, 1); - return 0; - } - - return ret; + return __block_rsv_add(root, block_rsv, num_bytes, 0); } int btrfs_block_rsv_check(struct btrfs_root *root, -- cgit v1.2.3 From 76b9e23d25d5c99f994bee3172de39492e452e93 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 10 Nov 2011 20:45:05 -0500 Subject: Btrfs: fix orphan backref nodes If the root node of a fs/file tree is in the block group that is being relocated, but the others are not in the other block groups. when we create a snapshot for this tree between the relocation tree creation ends and ->create_reloc_tree is set to 0, Btrfs will create some backref nodes that are the lowest nodes of the backrefs cache. But we forget to add them into ->leaves list of the backref cache and deal with them, and at last, they will triggered BUG_ON(). kernel BUG at fs/btrfs/relocation.c:239! This patch fixes it by adding them into ->leaves list of backref cache. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/relocation.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 24d654ce7a06..dff29d5e151a 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -1174,6 +1174,8 @@ static int clone_backref_node(struct btrfs_trans_handle *trans, list_add_tail(&new_edge->list[UPPER], &new_node->lower); } + } else { + list_add_tail(&new_node->lower, &cache->leaves); } rb_node = tree_insert(&cache->rb_root, new_node->bytenr, -- cgit v1.2.3 From 2f120c05e67ae34c93786b1050c6828904314429 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 10 Nov 2011 20:45:05 -0500 Subject: Btrfs: only map pages if we know we need them when reading the space cache People have been running into a warning when loading space cache because the page is already mapped when trying to read in a bitmap. The way we read in entries and pages is kind of convoluted, so fix it so that io_ctl_read_entry maps the entries if it needs to, and if it hits the end of the page it simply unmaps the page. That way we can unconditionally unmap the io_ctl before reading in the bitmap and we should stop hitting these warnings. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/free-space-cache.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 7a15fcfb3e1f..181760f9d2ab 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -537,6 +537,13 @@ static int io_ctl_read_entry(struct io_ctl *io_ctl, struct btrfs_free_space *entry, u8 *type) { struct btrfs_free_space_entry *e; + int ret; + + if (!io_ctl->cur) { + ret = io_ctl_check_crc(io_ctl, io_ctl->index); + if (ret) + return ret; + } e = io_ctl->cur; entry->offset = le64_to_cpu(e->offset); @@ -550,10 +557,7 @@ static int io_ctl_read_entry(struct io_ctl *io_ctl, io_ctl_unmap_page(io_ctl); - if (io_ctl->index >= io_ctl->num_pages) - return 0; - - return io_ctl_check_crc(io_ctl, io_ctl->index); + return 0; } static int io_ctl_read_bitmap(struct io_ctl *io_ctl, @@ -561,9 +565,6 @@ static int io_ctl_read_bitmap(struct io_ctl *io_ctl, { int ret; - if (io_ctl->cur && io_ctl->cur != io_ctl->orig) - io_ctl_unmap_page(io_ctl); - ret = io_ctl_check_crc(io_ctl, io_ctl->index); if (ret) return ret; @@ -699,6 +700,8 @@ int __load_free_space_cache(struct btrfs_root *root, struct inode *inode, num_entries--; } + io_ctl_unmap_page(&io_ctl); + /* * We add the bitmaps at the end of the entries in order that * the bitmap entries are added to the cache. -- cgit v1.2.3 From 62f30c5462374b991e7e3f42d49ce2265c1b82f1 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 10 Nov 2011 20:45:05 -0500 Subject: Btrfs: fix deadlock caused by the race between relocation We can not do flushable reservation for the relocation when we create snapshot, because it may make the transaction commit task and the flush task wait for each other and the deadlock happens. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/transaction.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 960835eaf4da..6a0574e923bc 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -882,8 +882,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans, btrfs_reloc_pre_snapshot(trans, pending, &to_reserve); if (to_reserve > 0) { - ret = btrfs_block_rsv_add(root, &pending->block_rsv, - to_reserve); + ret = btrfs_block_rsv_add_noflush(root, &pending->block_rsv, + to_reserve); if (ret) { pending->error = ret; goto fail; -- cgit v1.2.3 From 69f4cb526bd02ae5af35846f9a710c099eec3347 Mon Sep 17 00:00:00 2001 From: Arne Jansen Date: Fri, 11 Nov 2011 08:17:10 -0500 Subject: Btrfs: handle bio_add_page failure gracefully in scrub Currently scrub fails with ENOMEM when bio_add_page fails. Unfortunately dm based targets accept only one page per bio, thus making scrub always fails. This patch just submits the current bio when an error is encountered and starts a new one. Signed-off-by: Arne Jansen Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 64 +++++++++++++++++++++++++------------------------------- 1 file changed, 29 insertions(+), 35 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index ed11d3866afd..f4190f22edfb 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -944,50 +944,18 @@ static int scrub_checksum_super(struct scrub_bio *sbio, void *buffer) static int scrub_submit(struct scrub_dev *sdev) { struct scrub_bio *sbio; - struct bio *bio; - int i; if (sdev->curr == -1) return 0; sbio = sdev->bios[sdev->curr]; - - bio = bio_alloc(GFP_NOFS, sbio->count); - if (!bio) - goto nomem; - - bio->bi_private = sbio; - bio->bi_end_io = scrub_bio_end_io; - bio->bi_bdev = sdev->dev->bdev; - bio->bi_sector = sbio->physical >> 9; - - for (i = 0; i < sbio->count; ++i) { - struct page *page; - int ret; - - page = alloc_page(GFP_NOFS); - if (!page) - goto nomem; - - ret = bio_add_page(bio, page, PAGE_SIZE, 0); - if (!ret) { - __free_page(page); - goto nomem; - } - } - sbio->err = 0; sdev->curr = -1; atomic_inc(&sdev->in_flight); - submit_bio(READ, bio); + submit_bio(READ, sbio->bio); return 0; - -nomem: - scrub_free_bio(bio); - - return -ENOMEM; } static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len, @@ -995,6 +963,8 @@ static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len, u8 *csum, int force) { struct scrub_bio *sbio; + struct page *page; + int ret; again: /* @@ -1015,12 +985,22 @@ again: } sbio = sdev->bios[sdev->curr]; if (sbio->count == 0) { + struct bio *bio; + sbio->physical = physical; sbio->logical = logical; + bio = bio_alloc(GFP_NOFS, SCRUB_PAGES_PER_BIO); + if (!bio) + return -ENOMEM; + + bio->bi_private = sbio; + bio->bi_end_io = scrub_bio_end_io; + bio->bi_bdev = sdev->dev->bdev; + bio->bi_sector = sbio->physical >> 9; + sbio->err = 0; + sbio->bio = bio; } else if (sbio->physical + sbio->count * PAGE_SIZE != physical || sbio->logical + sbio->count * PAGE_SIZE != logical) { - int ret; - ret = scrub_submit(sdev); if (ret) return ret; @@ -1030,6 +1010,20 @@ again: sbio->spag[sbio->count].generation = gen; sbio->spag[sbio->count].have_csum = 0; sbio->spag[sbio->count].mirror_num = mirror_num; + + page = alloc_page(GFP_NOFS); + if (!page) + return -ENOMEM; + + ret = bio_add_page(sbio->bio, page, PAGE_SIZE, 0); + if (!ret) { + __free_page(page); + ret = scrub_submit(sdev); + if (ret) + return ret; + goto again; + } + if (csum) { sbio->spag[sbio->count].have_csum = 1; memcpy(sbio->spag[sbio->count].csum, csum, sdev->csum_size); -- cgit v1.2.3 From 8965593e41dd2d0e2a2f1e6f245336005ea94a2c Mon Sep 17 00:00:00 2001 From: David Sterba Date: Fri, 11 Nov 2011 10:14:57 -0500 Subject: btrfs: rename the option to nospace_cache Rename no_space_cache option to nospace_cache to be more consistent with the rest, where the simple prefix 'no' is used to negate an option. The option has been introduced during the -rc1 cycle and there are has not been widely used, so it's safe. Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/super.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 629281c65ff5..8bd9d6d0e07a 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -197,7 +197,7 @@ static match_table_t tokens = { {Opt_subvolrootid, "subvolrootid=%d"}, {Opt_defrag, "autodefrag"}, {Opt_inode_cache, "inode_cache"}, - {Opt_no_space_cache, "no_space_cache"}, + {Opt_no_space_cache, "nospace_cache"}, {Opt_recovery, "recovery"}, {Opt_err, NULL}, }; @@ -711,7 +711,7 @@ static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs) if (btrfs_test_opt(root, SPACE_CACHE)) seq_puts(seq, ",space_cache"); else - seq_puts(seq, ",no_space_cache"); + seq_puts(seq, ",nospace_cache"); if (btrfs_test_opt(root, CLEAR_CACHE)) seq_puts(seq, ",clear_cache"); if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED)) -- cgit v1.2.3