btrfs: inode creation cleanups and fixes
v4 cover btrfs: inode creation cleanups and fixes
From: Omar Sandoval <[email protected]>
This series contains minor updates of the final four patches of my previous series [1].
Based on misc-next with the previous version of "btrfs: allocate inode outside of btrfs_new_inode()" removed.
Changes since v3 [2]:
- Fixed a struct btrfs_root leak [3] in patch 1 which was fixed later in the same series by patch 3 for the sake of git bisect.
Changes since v2:
- Mentioned reason for removal of btrfs_lookup_dentry() call in create_subvol() in commit message of patch 1.
- Made btrfs_init_inode_security() also take btrfs_new_inode_args in patch 2.
Thanks!
1: https://lore.kernel.org/linux-btrfs/[email protected]/
2: https://lore.kernel.org/linux-btrfs/[email protected]/
3: https://lore.kernel.org/linux-btrfs/CAL3q7H5ACv3ej1=7P2y7mA1vCJoAcHkCqro6_VBuAUxeaw25rw@mail.gmail.com/
Omar Sandoval (4):
btrfs: allocate inode outside of btrfs_new_inode()
btrfs: factor out common part of btrfs_{mknod,create,mkdir}()
btrfs: reserve correct number of items for inode creation
btrfs: move common inode creation code into btrfs_create_new_inode()
fs/btrfs/acl.c | 36 +-- fs/btrfs/ctree.h | 37 ++- fs/btrfs/inode.c | 800 +++++++++++++++++++++++------------------------ fs/btrfs/ioctl.c | 140 +++++---- fs/btrfs/props.c | 40 +-- fs/btrfs/props.h | 4 - 6 files changed, 489 insertions(+), 568 deletions(-)
Signature · 1 line
2.35.1
PATCH v4 1/4 btrfs: allocate inode outside of btrfs_new_inode()
From: Omar Sandoval <[email protected]>
Instead of calling new_inode() and inode_init_owner() inside of btrfs_new_inode(), do it in the callers. This allows us to pass in just the inode instead of the mnt_userns and mode and removes the need for memalloc_nofs_{save,restores}() since we do it before starting a transaction. In create_subvol(), it also means we no longer have to look up the inode again to instantiate it. This also paves the way for some more cleanups in later patches.
This also removes the comments about Smack checking i_op, which are no longer true since commit 5d6c31910bc0 ("xattr: Add __vfs_{get,set,remove}xattr helpers"). Now it checks inode->i_opflags & IOP_XATTR, which is set based on sb->s_xattr.
--- fs/btrfs/ctree.h | 5 +- fs/btrfs/inode.c | 287 +++++++++++++++++++++++++---------------------- fs/btrfs/ioctl.c | 22 ++-- 3 files changed, 169 insertions(+), 145 deletions(-)
fs/btrfs/ctree.h +3 -2
@@ -3254,10 +3254,11 @@ int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, long nr,int btrfs_set_extent_delalloc (struct btrfs_inode *inode,u64 start,u64 end,unsigned int extra_bits,struct extent_state **cached_state); +struct inode *btrfs_new_subvol_inode (struct user_namespace *mnt_userns, +struct inode *dir);int btrfs_create_subvol_root (struct btrfs_trans_handle *trans, -struct btrfs_root *new_root,struct btrfs_root *parent_root, -struct user_namespace *mnt_userns); +struct inode *inode);void btrfs_set_delalloc_extent (struct inode *inode,struct extent_state *state,unsigned *bits); void btrfs_clear_delalloc_extent(struct inode *inode,
fs/btrfs/inode.c +152 -135
527 lines not on this page · raw message
fs/btrfs/ioctl.c +14 -8
@@ -587,6 +587,12 @@ static noinline int create_subvol(struct user_namespace *mnt_userns,if (ret <0 ) goto out_root_item; + inode =btrfs_new_subvol_inode (mnt_userns, dir); +if (!inode) { + ret = -ENOMEM ; + goto out_anon_dev; + } +btrfs_init_block_rsv (&block_rsv,BTRFS_BLOCK_RSV_TEMP );/* * The same as the snapshot creation, please see the comment
@@ -594,13 +600,13 @@ static noinline int create_subvol(struct user_namespace *mnt_userns,*/ ret =btrfs_subvolume_reserve_metadata (root, &block_rsv,8 , false);if (ret) - goto out_anon_dev; + goto out_inode; trans =btrfs_start_transaction (root,0 );if (IS_ERR (trans)) { ret =PTR_ERR (trans);btrfs_subvolume_release_metadata (root, &block_rsv); - goto out_anon_dev; + goto out_inode; } trans->block_rsv = &block_rsv; trans->bytes_reserved = block_rsv.size;
@@ -683,16 +689,16 @@ static noinline int create_subvol(struct user_namespace *mnt_userns, }/* anon_dev is owned by new_root now. */ anon_dev =0 ; +BTRFS_I (inode)->root = new_root; +/* ... and new_root is owned by inode now. */ ret =btrfs_record_root_in_trans (trans, new_root);if (ret) { -btrfs_put_root (new_root);btrfs_abort_transaction (trans, ret); goto out; } - ret =btrfs_create_subvol_root (trans, new_root, root, mnt_userns); -btrfs_put_root (new_root); + ret =btrfs_create_subvol_root (trans, root, inode);if (ret) {/* We potentially lose an unused inode item here */ btrfs_abort_transaction (trans, ret);
@@ -745,11 +751,11 @@ static noinline int create_subvol(struct user_namespace *mnt_userns, ret =btrfs_commit_transaction (trans);if (!ret) { - inode =btrfs_lookup_dentry (dir, dentry); -if (IS_ERR (inode)) -return PTR_ERR (inode);d_instantiate (dentry, inode); + inode =NULL ; } +out_inode: +iput (inode); out_anon_dev:if (anon_dev)free_anon_bdev (anon_dev);
Signature · 1 line
2.35.1
PATCH v4 2/4 btrfs: factor out common part of btrfs_{mknod,create,mkdir}()
In reply to Omar SandovalFrom: Omar Sandoval <[email protected]>
btrfs_{mknod,create,mkdir}() are now identical other than the inode initialization and some inconsequential function call order differences. Factor out the common code to reduce code duplication.
--- fs/btrfs/inode.c | 152 ++++++++++------------------------------------- 1 file changed, 33 insertions(+), 119 deletions(-)
fs/btrfs/inode.c +33 -119
@@ -6347,82 +6347,15 @@ int btrfs_add_link(struct btrfs_trans_handle *trans,return ret; } -static int btrfs_mknod (struct user_namespace *mnt_userns,struct inode *dir, -struct dentry *dentry,umode_t mode,dev_t rdev) +static int btrfs_create_common (struct inode *dir,struct dentry *dentry, +struct inode *inode) {struct btrfs_fs_info *fs_info =btrfs_sb (dir->i_sb); -struct btrfs_trans_handle *trans;struct btrfs_root *root =BTRFS_I (dir)->root; -struct inode *inode; +struct btrfs_trans_handle *trans;int err;u64 index =0 ; - inode =new_inode (dir->i_sb); -if (!inode) -return -ENOMEM ; -inode_init_owner (mnt_userns, inode, dir, mode); - inode->i_op = &btrfs_special_inode_operations; -init_special_inode (inode, inode->i_mode, rdev); - -/* -* 2 for inode item and ref -* 2 for dir items -* 1 for xattr if selinux is on -*/ - trans =btrfs_start_transaction (root,5 ); -if (IS_ERR (trans)) { -iput (inode); -return PTR_ERR (trans); - } - - err =btrfs_new_inode (trans, root, inode, dir, dentry->d_name.name, - dentry->d_name.len, &index); -if (err) { -iput (inode); - inode =NULL ; - goto out_unlock; - } - - err =btrfs_init_inode_security (trans, inode, dir, &dentry->d_name); -if (err) - goto out_unlock; - - err =btrfs_add_link (trans,BTRFS_I (dir),BTRFS_I (inode), - dentry->d_name.name, dentry->d_name.len,0 , index); -if (err) - goto out_unlock; - -btrfs_update_inode (trans, root,BTRFS_I (inode)); -d_instantiate_new (dentry, inode); - -out_unlock: -btrfs_end_transaction (trans); -btrfs_btree_balance_dirty (fs_info); -if (err && inode) { -inode_dec_link_count (inode); -discard_new_inode (inode); - } -return err; -} - -static int btrfs_create (struct user_namespace *mnt_userns,struct inode *dir, -struct dentry *dentry,umode_t mode,bool excl) -{ -struct btrfs_fs_info *fs_info =btrfs_sb (dir->i_sb); -struct btrfs_trans_handle *trans; -struct btrfs_root *root =BTRFS_I (dir)->root; -struct inode *inode; -int err; -u64 index =0 ; - - inode =new_inode (dir->i_sb); -if (!inode) -return -ENOMEM ; -inode_init_owner (mnt_userns, inode, dir, mode); - inode->i_fop = &btrfs_file_operations; - inode->i_op = &btrfs_file_inode_operations; - inode->i_mapping->a_ops = &btrfs_aops; -/* * 2 for inode item and ref * 2 for dir items
@@ -6467,6 +6400,35 @@ static int btrfs_create(struct user_namespace *mnt_userns, struct inode *dir,return err; } +static int btrfs_mknod (struct user_namespace *mnt_userns,struct inode *dir, +struct dentry *dentry,umode_t mode,dev_t rdev) +{ +struct inode *inode; + + inode =new_inode (dir->i_sb); +if (!inode) +return -ENOMEM ; +inode_init_owner (mnt_userns, inode, dir, mode); + inode->i_op = &btrfs_special_inode_operations; +init_special_inode (inode, inode->i_mode, rdev); +return btrfs_create_common (dir, dentry, inode); +} + +static int btrfs_create (struct user_namespace *mnt_userns,struct inode *dir, +struct dentry *dentry,umode_t mode,bool excl) +{ +struct inode *inode; + + inode =new_inode (dir->i_sb); +if (!inode) +return -ENOMEM ; +inode_init_owner (mnt_userns, inode, dir, mode); + inode->i_fop = &btrfs_file_operations; + inode->i_op = &btrfs_file_inode_operations; + inode->i_mapping->a_ops = &btrfs_aops; +return btrfs_create_common (dir, dentry, inode); +} +static int btrfs_link (struct dentry *old_dentry,struct inode *dir,struct dentry *dentry) {
@@ -6548,12 +6510,7 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,static int btrfs_mkdir (struct user_namespace *mnt_userns,struct inode *dir,struct dentry *dentry,umode_t mode) { -struct btrfs_fs_info *fs_info =btrfs_sb (dir->i_sb);struct inode *inode; -struct btrfs_trans_handle *trans; -struct btrfs_root *root =BTRFS_I (dir)->root; -int err; -u64 index =0 ; inode =new_inode (dir->i_sb);if (!inode)
@@ -6561,50 +6518,7 @@ static int btrfs_mkdir(struct user_namespace *mnt_userns, struct inode *dir,inode_init_owner (mnt_userns, inode, dir,S_IFDIR | mode); inode->i_op = &btrfs_dir_inode_operations; inode->i_fop = &btrfs_dir_file_operations; - -/* -* 2 items for inode and ref -* 2 items for dir items -* 1 for xattr if selinux is on -*/ - trans =btrfs_start_transaction (root,5 ); -if (IS_ERR (trans)) { -iput (inode); -return PTR_ERR (trans); - } - - err =btrfs_new_inode (trans, root, inode, dir, dentry->d_name.name, - dentry->d_name.len, &index); -if (err) { -iput (inode); - inode =NULL ; - goto out_fail; - } - - err =btrfs_init_inode_security (trans, inode, dir, &dentry->d_name); -if (err) - goto out_fail; - - err =btrfs_update_inode (trans, root,BTRFS_I (inode)); -if (err) - goto out_fail; - - err =btrfs_add_link (trans,BTRFS_I (dir),BTRFS_I (inode), - dentry->d_name.name, - dentry->d_name.len,0 , index); -if (err) - goto out_fail; - -d_instantiate_new (dentry, inode); - -out_fail: -btrfs_end_transaction (trans); -if (err && inode) { -inode_dec_link_count (inode); -discard_new_inode (inode); - } -btrfs_btree_balance_dirty (fs_info); -return err; +return btrfs_create_common (dir, dentry, inode); } static noinline int uncompress_inline(struct btrfs_path *path,
Signature · 1 line
2.35.1
PATCH v4 3/4 btrfs: reserve correct number of items for inode creation
In reply to Omar SandovalFrom: Omar Sandoval <[email protected]>
The various inode creation code paths do not account for the compression property, POSIX ACLs, or the parent inode item when starting a transaction. Fix it by refactoring all of these code paths to use a new function, btrfs_new_inode_prepare(), which computes the correct number of items. To do so, it needs to know whether POSIX ACLs will be created, so move the ACL creation into that function. To reduce the number of arguments that need to be passed around for inode creation, define struct btrfs_new_inode_args containing all of the relevant information.
btrfs_new_inode_prepare() will also be a good place to set up the fscrypt context and encrypted filename in the future.
--- fs/btrfs/acl.c | 36 +------ fs/btrfs/ctree.h | 34 +++++-- fs/btrfs/inode.c | 251 ++++++++++++++++++++++++++++++++++------------- fs/btrfs/ioctl.c | 83 +++++++++++----- 4 files changed, 270 insertions(+), 134 deletions(-)
fs/btrfs/acl.c +2 -34
@@ -55,8 +55,8 @@ struct posix_acl *btrfs_get_acl(struct inode *inode, int type, bool rcu)return acl; } -static int __btrfs_set_acl (struct btrfs_trans_handle *trans, -struct inode *inode,struct posix_acl *acl,int type) +int __btrfs_set_acl (struct btrfs_trans_handle *trans,struct inode *inode, +struct posix_acl *acl,int type) {int ret, size =0 ;const char *name;
@@ -127,35 +127,3 @@ int btrfs_set_acl(struct user_namespace *mnt_userns, struct inode *inode, inode->i_mode = old_mode;return ret; } - -int btrfs_init_acl (struct btrfs_trans_handle *trans, -struct inode *inode,struct inode *dir) -{ -struct posix_acl *default_acl, *acl; -int ret =0 ; - -/* this happens with subvols */ -if (!dir) -return 0 ; - - ret =posix_acl_create (dir, &inode->i_mode, &default_acl, &acl); -if (ret) -return ret; - -if (default_acl) { - ret =__btrfs_set_acl (trans, inode, default_acl, -ACL_TYPE_DEFAULT ); -posix_acl_release (default_acl); - } - -if (acl) { -if (!ret) - ret =__btrfs_set_acl (trans, inode, acl, -ACL_TYPE_ACCESS ); -posix_acl_release (acl); - } - -if (!default_acl && !acl) -cache_no_acl (inode); -return ret; -}
fs/btrfs/ctree.h +28 -6
@@ -3254,11 +3254,32 @@ int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, long nr,int btrfs_set_extent_delalloc (struct btrfs_inode *inode,u64 start,u64 end,unsigned int extra_bits,struct extent_state **cached_state); +struct btrfs_new_inode_args { +/* Input */ +struct inode *dir; +struct dentry *dentry; +struct inode *inode; +bool orphan; +bool subvol; + +/* +* Output from btrfs_new_inode_prepare(), input to +* btrfs_create_new_inode(). +*/ +struct posix_acl *default_acl; +struct posix_acl *acl; +}; +int btrfs_new_inode_prepare (struct btrfs_new_inode_args *args, +unsigned int *trans_num_items); +int btrfs_create_new_inode (struct btrfs_trans_handle *trans, +struct btrfs_new_inode_args *args, +u64 *index); +void btrfs_new_inode_args_destroy (struct btrfs_new_inode_args *args);struct inode *btrfs_new_subvol_inode (struct user_namespace *mnt_userns,struct inode *dir);int btrfs_create_subvol_root (struct btrfs_trans_handle *trans,struct btrfs_root *parent_root, -struct inode *inode); +struct btrfs_new_inode_args *args);void btrfs_set_delalloc_extent (struct inode *inode,struct extent_state *state,unsigned *bits); void btrfs_clear_delalloc_extent(struct inode *inode,
@@ -3816,15 +3837,16 @@ static inline int __btrfs_fs_compat_ro(struct btrfs_fs_info *fs_info, u64 flag)struct posix_acl *btrfs_get_acl (struct inode *inode,int type,bool rcu);int btrfs_set_acl (struct user_namespace *mnt_userns,struct inode *inode,struct posix_acl *acl,int type); -int btrfs_init_acl (struct btrfs_trans_handle *trans, -struct inode *inode,struct inode *dir); +int __btrfs_set_acl (struct btrfs_trans_handle *trans,struct inode *inode, +struct posix_acl *acl,int type);#else #define btrfs_get_acl NULL#define btrfs_set_acl NULL -static inline int btrfs_init_acl (struct btrfs_trans_handle *trans, -struct inode *inode,struct inode *dir) +static inline int __btrfs_set_acl (struct btrfs_trans_handle *trans, +struct inode *inode,struct posix_acl *acl, +int type) { -return 0 ; +return -EOPNOTSUPP ; }#endif
fs/btrfs/inode.c +182 -69
454 lines not on this page · raw message
fs/btrfs/ioctl.c +58 -25
145 lines not on this page · raw message
Signature · 1 line
2.35.1
PATCH v4 4/4 btrfs: move common inode creation code into btrfs_create_new_inode()
In reply to Omar SandovalFrom: Omar Sandoval <[email protected]>
All of our inode creation code paths duplicate the calls to btrfs_init_inode_security() and btrfs_add_link(). Subvolume creation additionally duplicates property inheritance and the call to btrfs_set_inode_index(). Fix this by moving the common code into btrfs_create_new_inode(). This accomplishes a few things at once:
1. It reduces code duplication.
2. It allows us to set up the inode completely before inserting the inode item, removing calls to btrfs_update_inode().
3. It fixes a leak of an inode on disk in some error cases. For example, in btrfs_create(), if btrfs_new_inode() succeeds, then we have inserted an inode item and its inode ref. However, if something after that fails (e.g., btrfs_init_inode_security()), then we end the transaction and then decrement the link count on the inode. If the transaction is committed and the system crashes before the failed inode is deleted, then we leak that inode on disk. Instead, this refactoring aborts the transaction when we can't recover more gracefully.
4. It exposes various ways that subvolume creation diverges from mkdir in terms of inheriting flags, properties, permissions, and POSIX ACLs, a lot of which appears to be accidental. This patch explicitly does _not_ change the existing non-standard behavior, but it makes those differences more clear in the code and documents them so that we can discuss whether they should be changed.
--- fs/btrfs/ctree.h | 6 +- fs/btrfs/inode.c | 404 +++++++++++++++++++---------------------------- fs/btrfs/ioctl.c | 59 ++----- fs/btrfs/props.c | 40 +---- fs/btrfs/props.h | 4 - 5 files changed, 180 insertions(+), 333 deletions(-)
fs/btrfs/ctree.h +1 -5
@@ -3272,14 +3272,10 @@ struct btrfs_new_inode_args {int btrfs_new_inode_prepare (struct btrfs_new_inode_args *args,unsigned int *trans_num_items);int btrfs_create_new_inode (struct btrfs_trans_handle *trans, -struct btrfs_new_inode_args *args, -u64 *index); +struct btrfs_new_inode_args *args);void btrfs_new_inode_args_destroy (struct btrfs_new_inode_args *args);struct inode *btrfs_new_subvol_inode (struct user_namespace *mnt_userns,struct inode *dir); -int btrfs_create_subvol_root (struct btrfs_trans_handle *trans, -struct btrfs_root *parent_root, -struct btrfs_new_inode_args *args);void btrfs_set_delalloc_extent (struct inode *inode,struct extent_state *state,unsigned *bits); void btrfs_clear_delalloc_extent(struct inode *inode,
fs/btrfs/inode.c +165 -239
619 lines not on this page · raw message
fs/btrfs/ioctl.c +12 -47
92 lines not on this page · raw message
fs/btrfs/props.c +2 -38
@@ -334,9 +334,8 @@ static struct prop_handler prop_handlers[] = { }, }; -static int inherit_props (struct btrfs_trans_handle *trans, -struct inode *inode, -struct inode *parent) +int btrfs_inode_inherit_props (struct btrfs_trans_handle *trans, +struct inode *inode,struct inode *parent) {struct btrfs_root *root =BTRFS_I (inode)->root;struct btrfs_fs_info *fs_info = root->fs_info;
@@ -408,41 +407,6 @@ static int inherit_props(struct btrfs_trans_handle *trans,return 0 ; } -int btrfs_inode_inherit_props (struct btrfs_trans_handle *trans, -struct inode *inode, -struct inode *dir) -{ -if (!dir) -return 0 ; - -return inherit_props (trans, inode, dir); -} - -int btrfs_subvol_inherit_props (struct btrfs_trans_handle *trans, -struct btrfs_root *root, -struct btrfs_root *parent_root) -{ -struct super_block *sb = root->fs_info->sb; -struct inode *parent_inode, *child_inode; -int ret; - - parent_inode =btrfs_iget (sb,BTRFS_FIRST_FREE_OBJECTID , parent_root); -if (IS_ERR (parent_inode)) -return PTR_ERR (parent_inode); - - child_inode =btrfs_iget (sb,BTRFS_FIRST_FREE_OBJECTID , root); -if (IS_ERR (child_inode)) { -iput (parent_inode); -return PTR_ERR (child_inode); - } - - ret =inherit_props (trans, child_inode, parent_inode); -iput (child_inode); -iput (parent_inode); - -return ret; -} -void __init btrfs_props_init (void ) {int i;
fs/btrfs/props.h +0 -4
@@ -21,8 +21,4 @@ int btrfs_inode_inherit_props(struct btrfs_trans_handle *trans,struct inode *inode, struct inode *dir); -int btrfs_subvol_inherit_props (struct btrfs_trans_handle *trans, -struct btrfs_root *root, -struct btrfs_root *parent_root); -#endif
Signature · 1 line
2.35.1
Re: [PATCH v4 4/4] btrfs: move common inode creation code into btrfs_create_new_inode()
As Naohiro Aota noted in slack, this change turns out to make generic/650 and others have write-time corruption. It appears to be from moving btrfs_init_inode_security() (which inserts xattrs for the inode) to before inserting the new inode into the tree; if the xattrs for an acl or for selinux get written out before the inode they belong to, the write-time corruption checker complains that an xattr item appears before its requisite inode.
I'm not sure if it reintroduces the leak, but moving btrfs_init_inode_security() to after the inode is fully inserted into the tree seems to fix the issue, something like:
fs/btrfs/inode.c +0 -12
@@ -6238,18 +6238,6 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans, ret); } -/* -* Subvolumes don't inherit ACLs or get passed to the LSM. This is -* probably a bug. -*/ -if (!args->subvol) { - ret =btrfs_init_inode_security (trans, args); -if (ret) { -btrfs_abort_transaction (trans, ret); - goto discard; - } - } -/* * We could have gotten an inode number from somebody who was
fsynced
* and then removed in this same transaction, so let's just set
full
@@ -6327,6 +6315,18 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,
btrfs_mark_buffer_dirty(path->nodes[0]);
btrfs_release_path(path);+ /*
+ * Subvolumes don't inherit ACLs or get passed to the LSM. This is + * probably a bug. + */
+ if (!args->subvol) {
+ ret = btrfs_init_inode_security(trans, args);
+ if (ret) {
+ btrfs_abort_transaction(trans, ret);
+ goto discard;
+ }
+ }
+
inode_tree_add(inode);
trace_btrfs_inode_new(inode);Signature · 3 lines
I don't know what the right process to follow here is. Naohiro, might you be able to test with this incremental patch and confirm that it fixes? Omar, does this have the inode leak problem you were trying to fix? Sweet Tea
Quoted 871 lines · Omar Sandoval
From: Omar Sandoval <[email protected]>
All of our inode creation code paths duplicate the calls to
btrfs_init_inode_security()andbtrfs_add_link(). Subvolume creation additionally duplicates property inheritance and the call tobtrfs_set_inode_index(). Fix this by moving the common code intobtrfs_create_new_inode(). This accomplishes a few things at once:1. It reduces code duplication.
2. It allows us to set up the inode completely before inserting the inode item, removing calls tobtrfs_update_inode().
3. It fixes a leak of an inode on disk in some error cases. For example, inbtrfs_create(), ifbtrfs_new_inode()succeeds, then we have inserted an inode item and its inode ref. However, if something after that fails (e.g.,btrfs_init_inode_security()), then we end the transaction and then decrement the link count on the inode. If the transaction is committed and the system crashes before the failed inode is deleted, then we leak that inode on disk. Instead, this refactoring aborts the transaction when we can't recover more gracefully.
4. It exposes various ways that subvolume creation diverges from mkdir in terms of inheriting flags, properties, permissions, and POSIX ACLs, a lot of which appears to be accidental. This patch explicitly does _not_ change the existing non-standard behavior, but it makes those differences more clear in the code and documents them so that we can discuss whether they should be changed.Reviewed-by Sweet Tea Dorminy <[email protected]>Signed-off-by Omar Sandoval <[email protected]>--- fs/btrfs/ctree.h | 6 +- fs/btrfs/inode.c | 404 +++++++++++++++++++---------------------------- fs/btrfs/ioctl.c | 59 ++----- fs/btrfs/props.c | 40 +---- fs/btrfs/props.h | 4 - 5 files changed, 180 insertions(+), 333 deletions(-)fs/btrfs/ctree.h +1
-5@@ -3272,14 +3272,10 @@ struct btrfs_new_inode_args {int btrfs_new_inode_prepare (struct btrfs_new_inode_args *args,unsigned int *trans_num_items);int btrfs_create_new_inode (struct btrfs_trans_handle *trans, -struct btrfs_new_inode_args *args, -u64 *index); +struct btrfs_new_inode_args *args);void btrfs_new_inode_args_destroy (struct btrfs_new_inode_args *args);struct inode *btrfs_new_subvol_inode (struct user_namespace *mnt_userns,struct inode *dir); -int btrfs_create_subvol_root (struct btrfs_trans_handle *trans, -struct btrfs_root *parent_root, -struct btrfs_new_inode_args *args);void btrfs_set_delalloc_extent (struct inode *inode,struct extent_state *state,unsigned *bits); void btrfs_clear_delalloc_extent(struct inode *inode,fs/btrfs/inode.c +165
-239@@ -6122,9 +6122,6 @@ static void btrfs_inherit_iflags(struct inode *inode, struct inode *dir) {unsigned int flags; -if (!dir) -return ; - flags =BTRFS_I (dir)->flags;if (flags &BTRFS_INODE_NOCOMPRESS ) {@@ -6145,14 +6142,13 @@ static void btrfs_inherit_iflags(struct inode *inode, struct inode *dir) }int btrfs_create_new_inode (struct btrfs_trans_handle *trans, -struct btrfs_new_inode_args *args, -u64 *index) +struct btrfs_new_inode_args *args) { -struct inode *dir = args->subvol ?NULL : args->dir; +struct inode *dir = args->dir;struct inode *inode = args->inode; -const char *name; -int name_len; -struct btrfs_fs_info *fs_info =btrfs_sb (inode->i_sb); +const char *name = args->orphan ?NULL : args->dentry->d_name.name; +int name_len = args->orphan ?0 : args->dentry->d_name.len; +struct btrfs_fs_info *fs_info =btrfs_sb (dir->i_sb);struct btrfs_root *root;struct btrfs_inode_item *inode_item;struct btrfs_key *location;@@ -6165,49 +6161,32 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,unsigned long ptr;int ret; -if (args->subvol) { - name =".." ; - name_len =2 ; - }else if (args->orphan) { - name =NULL ; - name_len =0 ; - }else { - name = args->dentry->d_name.name; - name_len = args->dentry->d_name.len; - } - path =btrfs_alloc_path ();if (!path)return -ENOMEM ; -/* -* O_TMPFILE, set link count to 0, so that after this point, -* we fill in an inode item with the correct link count. -*/ -if (!name) -set_nlink (inode,0 ); -if (!args->subvol)BTRFS_I (inode)->root =btrfs_grab_root (BTRFS_I (dir)->root); root =BTRFS_I (inode)->root; ret =btrfs_get_free_objectid (root, &objectid); -if (ret) { -btrfs_free_path (path); -return ret; - } +if (ret) + goto out; inode->i_ino = objectid; -if (dir && name) { +if (args->orphan) { +/* +* O_TMPFILE, set link count to 0, so that after this point, we +* fill in an inode item with the correct link count. +*/ +set_nlink (inode,0 ); + }else {trace_btrfs_inode_request (dir); - ret =btrfs_set_inode_index (BTRFS_I (dir), index); -if (ret) { -btrfs_free_path (path); -return ret; - } - }else if (dir) { - *index =0 ; + ret =btrfs_set_inode_index (BTRFS_I (dir), + &BTRFS_I (inode)->dir_index); +if (ret) + goto out; }/* * index_cnt is ignored for everything but a dir, @@ -6215,11 +6194,16 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,* number */ BTRFS_I (inode)->index_cnt =2 ; -BTRFS_I (inode)->dir_index = *index;BTRFS_I (inode)->generation = trans->transid; inode->i_generation =BTRFS_I (inode)->generation; -btrfs_inherit_iflags (inode, dir); +/* +* Subvolumes don't inherit flags from their parent directory. +* Originally this was probably by accident, but we probably can't +* change it now. +*/ +if (!args->subvol) +btrfs_inherit_iflags (inode, dir);if (S_ISREG (inode->i_mode)) {if (btrfs_test_opt (fs_info,NODATASUM ))@@ -6229,6 +6213,55 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,BTRFS_INODE_NODATASUM ; } + location = &BTRFS_I (inode)->location; + location->objectid = objectid; + location->offset =0 ; + location->type =BTRFS_INODE_ITEM_KEY ; + + ret =btrfs_insert_inode_locked (inode); +if (ret <0 ) { +if (!args->orphan) +BTRFS_I (dir)->index_cnt--; + goto out; + } + +if (args->subvol) { +struct inode *parent; + +/* +* Subvolumes inherit properties from their parent subvolume, +* not the directory they were created in. +*/ + parent =btrfs_iget (fs_info->sb,BTRFS_FIRST_FREE_OBJECTID , +BTRFS_I (dir)->root); +if (IS_ERR (parent)) { + ret =PTR_ERR (parent); + }else { + ret =btrfs_inode_inherit_props (trans, inode, parent); +iput (parent); + } + }else { + ret =btrfs_inode_inherit_props (trans, inode, dir); + } +if (ret) { +btrfs_err (fs_info, +"error inheriting props for ino %llu (root %llu): %d" , +btrfs_ino (BTRFS_I (inode)), root->root_key.objectid, + ret); + } + +/* +* Subvolumes don't inherit ACLs or get passed to the LSM. This is +* probably a bug. +*/ +if (!args->subvol) { + ret =btrfs_init_inode_security (trans, args); +if (ret) { +btrfs_abort_transaction (trans, ret); + goto discard; + } + } +/* * We could have gotten an inode number from somebody who was fsynced * and then removed in this same transaction, so let's just set full @@ -6243,7 +6276,7 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans, sizes[0 ] =sizeof (struct btrfs_inode_item ); -if (name) { +if (!args->orphan) {/* * Start new inodes with an inode_ref. This is slightly more * efficient for small numbers of hard links since they will @@ -6252,53 +6285,61 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,*/ key[1 ].objectid = objectid; key[1 ].type =BTRFS_INODE_REF_KEY ; -if (dir) - key[1 ].offset =btrfs_ino (BTRFS_I (dir)); -else +if (args->subvol) { key[1 ].offset = objectid; - - sizes[1 ] = name_len +sizeof (*ref); + sizes[1 ] =2 +sizeof (*ref); + }else { + key[1 ].offset =btrfs_ino (BTRFS_I (dir)); + sizes[1 ] = name_len +sizeof (*ref); + } } - location = &BTRFS_I (inode)->location; - location->objectid = objectid; - location->offset =0 ; - location->type =BTRFS_INODE_ITEM_KEY ; - - ret =btrfs_insert_inode_locked (inode); -if (ret <0 ) - goto fail; - batch.keys = &key[0 ]; batch.data_sizes = &sizes[0 ]; - batch.total_data_size = sizes[0 ] + (name ? sizes[1 ] :0 ); - batch.nr = name ?2 :1 ; + batch.total_data_size = sizes[0 ] + (args->orphan ?0 : sizes[1 ]); + batch.nr = args->orphan ?1 :2 ; ret =btrfs_insert_empty_items (trans, root, path, &batch); -if (ret !=0 ) - goto fail_unlock; +if (ret !=0 ) { +btrfs_abort_transaction (trans, ret); + goto discard; + } inode->i_mtime =current_time (inode); inode->i_atime = inode->i_mtime; inode->i_ctime = inode->i_mtime;BTRFS_I (inode)->i_otime = inode->i_mtime; +/* +* We're going to fill the inode item now, so at this point the inode +* must be fully initialized. +*/ + inode_item =btrfs_item_ptr (path->nodes[0 ], path->slots[0 ], struct btrfs_inode_item);memzero_extent_buffer (path->nodes[0 ], (unsigned long )inode_item,sizeof (*inode_item));fill_inode_item (trans, path->nodes[0 ], inode_item, inode); -if (name) { +if (!args->orphan) { ref =btrfs_item_ptr (path->nodes[0 ], path->slots[0 ] +1 , struct btrfs_inode_ref); -btrfs_set_inode_ref_name_len (path->nodes[0 ], ref, name_len); -btrfs_set_inode_ref_index (path->nodes[0 ], ref, *index); ptr = (unsigned long )(ref +1 ); -write_extent_buffer (path->nodes[0 ], name, ptr, name_len); +if (args->subvol) { +btrfs_set_inode_ref_name_len (path->nodes[0 ], ref,2 ); +btrfs_set_inode_ref_index (path->nodes[0 ], ref,0 ); +write_extent_buffer (path->nodes[0 ],".." , ptr,2 ); + }else { +btrfs_set_inode_ref_name_len (path->nodes[0 ], ref, + name_len); +btrfs_set_inode_ref_index (path->nodes[0 ], ref, +BTRFS_I (inode)->dir_index); +write_extent_buffer (path->nodes[0 ], name, ptr, + name_len); + } }btrfs_mark_buffer_dirty (path->nodes[0 ]); -btrfs_free_path (path); +btrfs_release_path (path);inode_tree_add (inode);@@ -6307,24 +6348,28 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,btrfs_update_root_times (trans, root); - ret =btrfs_inode_inherit_props (trans, inode, dir); -if (ret) -btrfs_err (fs_info, -"error inheriting props for ino %llu (root %llu): %d" , -btrfs_ino (BTRFS_I (inode)), root->root_key.objectid, ret); +if (args->orphan) { + ret =btrfs_orphan_add (trans,BTRFS_I (inode)); + }else { + ret =btrfs_add_link (trans,BTRFS_I (dir),BTRFS_I (inode), name, + name_len,0 ,BTRFS_I (inode)->dir_index); + } +if (ret) { +btrfs_abort_transaction (trans, ret); + goto discard; + } -return 0 ; + ret =0 ; + goto out; -fail_unlock: +discard:/* * discard_new_inode() calls iput(), but the caller owns the reference * to the inode. */ ihold (inode);discard_new_inode (inode); -fail: -if (dir && name) -BTRFS_I (dir)->index_cnt--; +out:btrfs_free_path (path);return ret; }@@ -6431,52 +6476,28 @@ static int btrfs_create_common(struct inode *dir, struct dentry *dentry,unsigned int trans_num_items;struct btrfs_trans_handle *trans;int err; -u64 index =0 ; err =btrfs_new_inode_prepare (&new_inode_args, &trans_num_items); -if (err) { -iput (inode); -return err; - } +if (err) + goto out_inode; trans =btrfs_start_transaction (root, trans_num_items);if (IS_ERR (trans)) { -iput (inode); err =PTR_ERR (trans); goto out_new_inode_args; } - err =btrfs_create_new_inode (trans, &new_inode_args, &index); -if (err) { -iput (inode); - inode =NULL ; - goto out_unlock; - } + err =btrfs_create_new_inode (trans, &new_inode_args); +if (!err) +d_instantiate_new (dentry, inode); - err =btrfs_init_inode_security (trans, &new_inode_args); -if (err) - goto out_unlock; - - err =btrfs_update_inode (trans, root,BTRFS_I (inode)); -if (err) - goto out_unlock; - - err =btrfs_add_link (trans,BTRFS_I (dir),BTRFS_I (inode), - dentry->d_name.name, dentry->d_name.len,0 , index); -if (err) - goto out_unlock; - -d_instantiate_new (dentry, inode); - -out_unlock:btrfs_end_transaction (trans); -if (err && inode) { -inode_dec_link_count (inode); -discard_new_inode (inode); - }btrfs_btree_balance_dirty (fs_info); out_new_inode_args:btrfs_new_inode_args_destroy (&new_inode_args); +out_inode: +if (err) +iput (inode);return err; }@@ -8751,34 +8772,6 @@ struct inode *btrfs_new_subvol_inode(struct user_namespace *mnt_userns,return inode; } -/* -* create a new subvolume directory/inode (helper for the ioctl). -*/ -int btrfs_create_subvol_root (struct btrfs_trans_handle *trans, -struct btrfs_root *parent_root, -struct btrfs_new_inode_args *args) -{ -struct inode *inode = args->inode; -struct btrfs_root *new_root =BTRFS_I (inode)->root; -int err; -u64 index =0 ; - - err =btrfs_create_new_inode (trans, args, &index); -if (err) -return err; - -unlock_new_inode (inode); - - err =btrfs_subvol_inherit_props (trans, new_root, parent_root); -if (err) -btrfs_err (new_root->fs_info, -"error inheriting subvolume %llu properties: %d" , - new_root->root_key.objectid, err); - - err =btrfs_update_inode (trans, new_root,BTRFS_I (inode)); -return err; -} -struct inode *btrfs_alloc_inode (struct super_block *sb) {struct btrfs_fs_info *fs_info =btrfs_sb (sb);@@ -9267,41 +9260,6 @@ static struct inode *new_whiteout_inode(struct user_namespace *mnt_userns,return inode; } -static int btrfs_whiteout_for_rename (struct btrfs_trans_handle *trans, -struct btrfs_new_inode_args *args) -{ -struct inode *inode = args->inode; -struct inode *dir = args->dir; -struct btrfs_root *root =BTRFS_I (dir)->root; -struct dentry *dentry = args->dentry; -int ret; -u64 index; - - ret =btrfs_create_new_inode (trans, args, &index); -if (ret) { -iput (inode); -return ret; - } - - ret =btrfs_init_inode_security (trans, args); -if (ret) - goto out; - - ret =btrfs_add_link (trans,BTRFS_I (dir),BTRFS_I (inode), - dentry->d_name.name, dentry->d_name.len,0 , index); -if (ret) - goto out; - - ret =btrfs_update_inode (trans, root,BTRFS_I (inode)); -out: -unlock_new_inode (inode); -if (ret) -inode_dec_link_count (inode); -iput (inode); - -return ret; -} - static int btrfs_rename(struct user_namespace *mnt_userns, struct inode *old_dir, struct dentry *old_dentry, struct inode *new_dir, struct dentry *new_dentry,@@ -9513,11 +9471,14 @@ static int btrfs_rename(struct user_namespace *mnt_userns, rename_ctx.index, new_dentry->d_parent);if (flags &RENAME_WHITEOUT ) { - ret =btrfs_whiteout_for_rename (trans, &whiteout_args); - whiteout_args.inode =NULL ; + ret =btrfs_create_new_inode (trans, &whiteout_args);if (ret) {btrfs_abort_transaction (trans, ret); goto out_fail; + }else { +unlock_new_inode (whiteout_args.inode); +iput (whiteout_args.inode); + whiteout_args.inode =NULL ; } } out_fail:@@ -9756,7 +9717,6 @@ static int btrfs_symlink(struct user_namespace *mnt_userns, struct inode *dir, };unsigned int trans_num_items;int err; -u64 index =0 ;int name_len;int datasize;unsigned long ptr;@@ -9774,38 +9734,33 @@ static int btrfs_symlink(struct user_namespace *mnt_userns, struct inode *dir, inode->i_op = &btrfs_symlink_inode_operations;inode_nohighmem (inode); inode->i_mapping->a_ops = &btrfs_aops; +btrfs_i_size_write (BTRFS_I (inode), name_len); +inode_set_bytes (inode, name_len); new_inode_args.inode = inode; err =btrfs_new_inode_prepare (&new_inode_args, &trans_num_items); -if (err) { -iput (inode); -return err; - } +if (err) + goto out_inode;/* 1 additional item for the inline extent */ trans_num_items++; trans =btrfs_start_transaction (root, trans_num_items);if (IS_ERR (trans)) { -iput (inode); err =PTR_ERR (trans); goto out_new_inode_args; } - err =btrfs_create_new_inode (trans, &new_inode_args, &index); -if (err) { -iput (inode); - inode =NULL ; - goto out_unlock; - } - - err =btrfs_init_inode_security (trans, &new_inode_args); + err =btrfs_create_new_inode (trans, &new_inode_args);if (err) - goto out_unlock; + goto out; path =btrfs_alloc_path ();if (!path) { err = -ENOMEM ; - goto out_unlock; +btrfs_abort_transaction (trans, err); +discard_new_inode (inode); + inode =NULL ; + goto out; } key.objectid =btrfs_ino (BTRFS_I (inode)); key.offset =0 ;@@ -9814,8 +9769,11 @@ static int btrfs_symlink(struct user_namespace *mnt_userns, struct inode *dir, err =btrfs_insert_empty_item (trans, root, path, &key, datasize);if (err) { +btrfs_abort_transaction (trans, err);btrfs_free_path (path); - goto out_unlock; +discard_new_inode (inode); + inode =NULL ; + goto out; } leaf = path->nodes[0 ]; ei = btrfs_item_ptr(leaf, path->slots[0],@@ -9833,32 +9791,16 @@ static int btrfs_symlink(struct user_namespace *mnt_userns, struct inode *dir,btrfs_mark_buffer_dirty (leaf);btrfs_free_path (path); -inode_set_bytes (inode, name_len); -btrfs_i_size_write (BTRFS_I (inode), name_len); - err =btrfs_update_inode (trans, root,BTRFS_I (inode)); -/* -* Last step, add directory indexes for our symlink inode. This is the -* last step to avoid extra cleanup of these indexes if an error happens -* elsewhere above. -*/ -if (!err) - err =btrfs_add_link (trans,BTRFS_I (dir),BTRFS_I (inode), - dentry->d_name.name, dentry->d_name.len,0 , - index); -if (err) - goto out_unlock; -d_instantiate_new (dentry, inode); - -out_unlock: + err =0 ; +out:btrfs_end_transaction (trans); -if (err && inode) { -inode_dec_link_count (inode); -discard_new_inode (inode); - }btrfs_btree_balance_dirty (fs_info); out_new_inode_args:btrfs_new_inode_args_destroy (&new_inode_args); +out_inode: +if (err) +iput (inode);return err; }@@ -10121,7 +10063,6 @@ static int btrfs_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, .orphan = true, };unsigned int trans_num_items; -u64 index;int ret; inode =new_inode (dir->i_sb);@@ -10134,35 +10075,16 @@ static int btrfs_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, new_inode_args.inode = inode; ret =btrfs_new_inode_prepare (&new_inode_args, &trans_num_items); -if (ret) { -iput (inode); -return ret; - } +if (ret) + goto out_inode; trans =btrfs_start_transaction (root, trans_num_items);if (IS_ERR (trans)) { -iput (inode); ret =PTR_ERR (trans); goto out_new_inode_args; } - ret =btrfs_create_new_inode (trans, &new_inode_args, &index); -if (ret) { -iput (inode); - inode =NULL ; - goto out; - } - - ret =btrfs_init_inode_security (trans, &new_inode_args); -if (ret) - goto out; - - ret =btrfs_update_inode (trans, root,BTRFS_I (inode)); -if (ret) - goto out; - ret =btrfs_orphan_add (trans,BTRFS_I (inode)); -if (ret) - goto out; + ret =btrfs_create_new_inode (trans, &new_inode_args);/* * We set number of links to 0 in btrfs_create_new_inode(), and here we @@ -10172,16 +10094,20 @@ static int btrfs_tmpfile(struct user_namespace *mnt_userns, struct inode *dir,* d_tmpfile() -> inode_dec_link_count() -> drop_nlink() */ set_nlink (inode,1 ); -d_tmpfile (dentry, inode); -unlock_new_inode (inode); -mark_inode_dirty (inode); -out: + +if (!ret) { +d_tmpfile (dentry, inode); +unlock_new_inode (inode); +mark_inode_dirty (inode); + } +btrfs_end_transaction (trans); -if (ret && inode) -discard_new_inode (inode);btrfs_btree_balance_dirty (fs_info); out_new_inode_args:btrfs_new_inode_args_destroy (&new_inode_args); +out_inode: +if (ret) +iput (inode);return ret; }fs/btrfs/ioctl.c +12
-47@@ -574,8 +574,6 @@ static noinline int create_subvol(struct user_namespace *mnt_userns, struct inode *dir, struct dentry *dentry, struct btrfs_qgroup_inherit *inherit) { - const char *name = dentry->d_name.name; -int namelen = dentry->d_name.len; struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);struct btrfs_trans_handle *trans;struct btrfs_key key;@@ -595,7 +593,6 @@ static noinline int create_subvol(struct user_namespace *mnt_userns,int ret;dev_t anon_dev;u64 objectid; -u64 index =0 ; root_item =kzalloc (sizeof (*root_item),GFP_KERNEL );if (!root_item)@@ -712,7 +709,6 @@ static noinline int create_subvol(struct user_namespace *mnt_userns,free_extent_buffer (leaf); leaf =NULL ; - key.offset = (u64)-1 ; new_root =btrfs_get_new_fs_root (fs_info, objectid, anon_dev);if (IS_ERR (new_root)) { ret =PTR_ERR (new_root);@@ -730,47 +726,21 @@ static noinline int create_subvol(struct user_namespace *mnt_userns, goto out; } - ret =btrfs_create_subvol_root (trans, root, &new_inode_args); -if (ret) { -/* We potentially lose an unused inode item here */ -btrfs_abort_transaction (trans, ret); - goto out; - } - -/* -* insert the directory item -*/ - ret =btrfs_set_inode_index (BTRFS_I (dir), &index); -if (ret) { -btrfs_abort_transaction (trans, ret); - goto out; - } - - ret =btrfs_insert_dir_item (trans, name, namelen,BTRFS_I (dir), &key, -BTRFS_FT_DIR , index); -if (ret) { -btrfs_abort_transaction (trans, ret); - goto out; - } - -btrfs_i_size_write (BTRFS_I (dir), dir->i_size + namelen *2 ); - ret =btrfs_update_inode (trans, root,BTRFS_I (dir)); -if (ret) { -btrfs_abort_transaction (trans, ret); - goto out; - } - - ret =btrfs_add_root_ref (trans, objectid, root->root_key.objectid, -btrfs_ino (BTRFS_I (dir)), index, name, namelen); -if (ret) { -btrfs_abort_transaction (trans, ret); - goto out; - } - ret =btrfs_uuid_tree_add (trans, root_item->uuid,BTRFS_UUID_KEY_SUBVOL , objectid); -if (ret) +if (ret) {btrfs_abort_transaction (trans, ret); + goto out; + } + + ret =btrfs_create_new_inode (trans, &new_inode_args); +if (ret) { +btrfs_abort_transaction (trans, ret); + goto out; + } + +d_instantiate_new (dentry, new_inode_args.inode); + new_inode_args.inode =NULL ; out: trans->block_rsv =NULL ;@@ -781,11 +751,6 @@ static noinline int create_subvol(struct user_namespace *mnt_userns,btrfs_end_transaction (trans); else ret =btrfs_commit_transaction (trans); - -if (!ret) { -d_instantiate (dentry, new_inode_args.inode); - new_inode_args.inode =NULL ; - } out_new_inode_args:btrfs_new_inode_args_destroy (&new_inode_args); out_inode:fs/btrfs/props.c +2
-38@@ -334,9 +334,8 @@ static struct prop_handler prop_handlers[] = { }, }; -static int inherit_props (struct btrfs_trans_handle *trans, -struct inode *inode, -struct inode *parent) +int btrfs_inode_inherit_props (struct btrfs_trans_handle *trans, +struct inode *inode,struct inode *parent) {struct btrfs_root *root =BTRFS_I (inode)->root;struct btrfs_fs_info *fs_info = root->fs_info;@@ -408,41 +407,6 @@ static int inherit_props(struct btrfs_trans_handle *trans,return 0 ; } -int btrfs_inode_inherit_props (struct btrfs_trans_handle *trans, -struct inode *inode, -struct inode *dir) -{ -if (!dir) -return 0 ; - -return inherit_props (trans, inode, dir); -} - -int btrfs_subvol_inherit_props (struct btrfs_trans_handle *trans, -struct btrfs_root *root, -struct btrfs_root *parent_root) -{ -struct super_block *sb = root->fs_info->sb; -struct inode *parent_inode, *child_inode; -int ret; - - parent_inode =btrfs_iget (sb,BTRFS_FIRST_FREE_OBJECTID , parent_root); -if (IS_ERR (parent_inode)) -return PTR_ERR (parent_inode); - - child_inode =btrfs_iget (sb,BTRFS_FIRST_FREE_OBJECTID , root); -if (IS_ERR (child_inode)) { -iput (parent_inode); -return PTR_ERR (child_inode); - } - - ret =inherit_props (trans, child_inode, parent_inode); -iput (child_inode); -iput (parent_inode); - -return ret; -} -void __init btrfs_props_init (void ) {int i;fs/btrfs/props.h +0
-4@@ -21,8 +21,4 @@ int btrfs_inode_inherit_props(struct btrfs_trans_handle *trans,struct inode *inode, struct inode *dir); -int btrfs_subvol_inherit_props (struct btrfs_trans_handle *trans, -struct btrfs_root *root, -struct btrfs_root *parent_root); -#endif
Re: [PATCH v4 4/4] btrfs: move common inode creation code into btrfs_create_new_inode()
Quoted 7 lines · Sweet Tea Dorminy
As Naohiro Aota noted in slack, this change turns out to make generic/650 and others have write-time corruption. It appears to be from moving
btrfs_init_inode_security()(which inserts xattrs for the inode) to before inserting the new inode into the tree; if the xattrs for an acl or for selinux get written out before the inode they belong to, the write-time corruption checker complains that an xattr item appears before its requisite inode.
This may be the write-time tree checker being overzealous. Both the xattr insertions and the inode insertion happen within one transaction, so we won't end up with the xattr but no inode actually being committed. I don't see anything that would legitimately be upset if it came across this intermediate state in the meantime.
Quoted 46 lines
I'm not sure if it reintroduces the leak, but moving
btrfs_init_inode_security()to after the inode is fully inserted into the tree seems to fix the issue, something like:fs/btrfs/inode.c +0
-12@@ -6238,18 +6238,6 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans, ret); } -/* -* Subvolumes don't inherit ACLs or get passed to the LSM. This is -* probably a bug. -*/ -if (!args->subvol) { - ret =btrfs_init_inode_security (trans, args); -if (ret) { -btrfs_abort_transaction (trans, ret); - goto discard; - } - } -/* * We could have gotten an inode number from somebody who was fsynced
* and then removed in this same transaction, so let's just set fullfs/btrfs/inode.c +12
-0@@ -6327,6 +6315,18 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,btrfs_mark_buffer_dirty (path->nodes[0 ]);btrfs_release_path (path); +/* +* Subvolumes don't inherit ACLs or get passed to the LSM. This is +* probably a bug. +*/ +if (!args->subvol) { + ret =btrfs_init_inode_security (trans, args); +if (ret) { +btrfs_abort_transaction (trans, ret); + goto discard; + } + } +inode_tree_add (inode);
Quoted 7 lines
trace_btrfs_inode_new(inode);Signature · 3 lines
I don't know what the right process to follow here is. Naohiro, might you be able to test with this incremental patch and confirm that it fixes? Omar, does this have the inode leak problem you were trying to fix?
I _think_ this fix is fine, though. The only thing I'm concerned about is if btrfs_init_inode_security() changes the inode, then we won't write out those changes to disk. But I don't see any code paths in btrfs_init_inode_security() that change the inode.
I'd say that if this fixes Naohiro's issue, we should go with it.
Quoted 1 line
Sweet Tea
Re: [PATCH v4 4/4] btrfs: move common inode creation code into btrfs_create_new_inode()
Quoted 74 lines · Omar Sandoval
Quoted 7 lines · Sweet Tea Dorminy
As Naohiro Aota noted in slack, this change turns out to make generic/650 and others have write-time corruption. It appears to be from moving
btrfs_init_inode_security()(which inserts xattrs for the inode) to before inserting the new inode into the tree; if the xattrs for an acl or for selinux get written out before the inode they belong to, the write-time corruption checker complains that an xattr item appears before its requisite inode.This may be the write-time tree checker being overzealous. Both the xattr insertions and the inode insertion happen within one transaction, so we won't end up with the xattr but no inode actually being committed. I don't see anything that would legitimately be upset if it came across this intermediate state in the meantime.
Quoted 46 lines
I'm not sure if it reintroduces the leak, but moving
btrfs_init_inode_security()to after the inode is fully inserted into the tree seems to fix the issue, something like:fs/btrfs/inode.c +0
-12@@ -6238,18 +6238,6 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans, ret); } -/* -* Subvolumes don't inherit ACLs or get passed to the LSM. This is -* probably a bug. -*/ -if (!args->subvol) { - ret =btrfs_init_inode_security (trans, args); -if (ret) { -btrfs_abort_transaction (trans, ret); - goto discard; - } - } -/* * We could have gotten an inode number from somebody who was fsynced
* and then removed in this same transaction, so let's just set fullfs/btrfs/inode.c +12
-0@@ -6327,6 +6315,18 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,btrfs_mark_buffer_dirty (path->nodes[0 ]);btrfs_release_path (path); +/* +* Subvolumes don't inherit ACLs or get passed to the LSM. This is +* probably a bug. +*/ +if (!args->subvol) { + ret =btrfs_init_inode_security (trans, args); +if (ret) { +btrfs_abort_transaction (trans, ret); + goto discard; + } + } +inode_tree_add (inode);Quoted 7 lines
trace_btrfs_inode_new(inode);Signature · 3 lines
I don't know what the right process to follow here is. Naohiro, might you be able to test with this incremental patch and confirm that it fixes? Omar, does this have the inode leak problem you were trying to fix?I _think_ this fix is fine, though. The only thing I'm concerned about is if
btrfs_init_inode_security()changes the inode, then we won't write out those changes to disk. But I don't see any code paths inbtrfs_init_inode_security()that change the inode.
Quoted 1 line
I'd say that if this fixes Naohiro's issue, we should go with it.
As I just discussed with Sweet Tea offline, property inheritance also needs to move after inode creation for the same reason.
Re: [PATCH v4 4/4] btrfs: move common inode creation code into btrfs_create_new_inode()
Quoted 9 lines · Omar Sandoval
Quoted 6 lines
I _think_ this fix is fine, though. The only thing I'm concerned about is if
btrfs_init_inode_security()changes the inode, then we won't write out those changes to disk. But I don't see any code paths inbtrfs_init_inode_security()that change the inode.I'd say that if this fixes Naohiro's issue, we should go with it.
As I just discussed with Sweet Tea offline, property inheritance also needs to move after inode creation for the same reason.
Can you please resend the full updated patch, based on the one that's in misc-next once it's tested? Thanks.
All looks good to me still. Thanks!
Quoted 40 lines · Omar Sandoval
From: Omar Sandoval <[email protected]>
This series contains minor updates of the final four patches of my previous series [1].
Based on misc-next with the previous version of "btrfs: allocate inode outside of
btrfs_new_inode()" removed.Changes since v3 [2]:
- Fixed a struct
btrfs_rootleak [3] in patch 1 which was fixed later in the same series by patch 3 for the sake of git bisect.Changes since v2:
- Mentioned reason for removal of
btrfs_lookup_dentry()call increate_subvol()in commit message of patch 1.
- Madebtrfs_init_inode_security()also takebtrfs_new_inode_argsin patch 2.Thanks!
1: https://lore.kernel.org/linux-btrfs/[email protected]/
2: https://lore.kernel.org/linux-btrfs/[email protected]/
3: https://lore.kernel.org/linux-btrfs/CAL3q7H5ACv3ej1=7P2y7mA1vCJoAcHkCqro6_VBuAUxeaw25rw@mail.gmail.com/Omar Sandoval (4):
btrfs: allocate inode outside ofbtrfs_new_inode()
btrfs: factor out common part of btrfs_{mknod,create,mkdir}()
btrfs: reserve correct number of items for inode creation
btrfs: move common inode creation code intobtrfs_create_new_inode()fs/btrfs/acl.c | 36 +-- fs/btrfs/ctree.h | 37 ++- fs/btrfs/inode.c | 800 +++++++++++++++++++++++------------------------ fs/btrfs/ioctl.c | 140 +++++---- fs/btrfs/props.c | 40 +-- fs/btrfs/props.h | 4 - 6 files changed, 489 insertions(+), 568 deletions(-)
Quoted 26 lines · Omar Sandoval
From: Omar Sandoval <[email protected]>
This series contains minor updates of the final four patches of my previous series [1].
Based on misc-next with the previous version of "btrfs: allocate inode outside of
btrfs_new_inode()" removed.Changes since v3 [2]:
- Fixed a struct
btrfs_rootleak [3] in patch 1 which was fixed later in the same series by patch 3 for the sake of git bisect.Changes since v2:
- Mentioned reason for removal of
btrfs_lookup_dentry()call increate_subvol()in commit message of patch 1.
- Madebtrfs_init_inode_security()also takebtrfs_new_inode_argsin patch 2.Thanks!
1: https://lore.kernel.org/linux-btrfs/[email protected]/
2: https://lore.kernel.org/linux-btrfs/[email protected]/
3: https://lore.kernel.org/linux-btrfs/CAL3q7H5ACv3ej1=7P2y7mA1vCJoAcHkCqro6_VBuAUxeaw25rw@mail.gmail.com/
Quoted 5 lines
Omar Sandoval (4):
btrfs: allocate inode outside ofbtrfs_new_inode()
btrfs: factor out common part of btrfs_{mknod,create,mkdir}()
btrfs: reserve correct number of items for inode creation
btrfs: move common inode creation code intobtrfs_create_new_inode()
Added to misc-next, thanks.