summaryrefslogtreecommitdiff
path: root/fs/reiserfs
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2022-08-01 18:56:55 +0300
committerLinus Torvalds <torvalds@linux-foundation.org>2022-08-01 18:56:55 +0300
commitbdfae5ce3820e79e709bd530ab458302be2e0d9c (patch)
tree34dc6ec7a4984d9c2f9f4d61ef6a7a37f7089052 /fs/reiserfs
parente6a7cf70a3ca9dc83419dd3d8ef09a800da3d7c1 (diff)
parent77940f0d96cd2ec9fe2125f74f513a7254bcdd7f (diff)
downloadlinux-bdfae5ce3820e79e709bd530ab458302be2e0d9c.tar.xz
Merge tag 'fs.idmapped.vfsuid.v5.20' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux
Pull fs idmapping updates from Christian Brauner: "This introduces the new vfs{g,u}id_t types we agreed on. Similar to k{g,u}id_t the new types are just simple wrapper structs around regular {g,u}id_t types. They allow to establish a type safety boundary in the VFS for idmapped mounts preventing confusion betwen {g,u}ids mapped into an idmapped mount and {g,u}ids mapped into the caller's or the filesystem's idmapping. An initial set of helpers is introduced that allows to operate on vfs{g,u}id_t types. We will remove all references to non-type safe idmapped mounts helpers in the very near future. The patches do already exist. This converts the core attribute changing codepaths which become significantly easier to reason about because of this change. Just a few highlights here as the patches give detailed overviews of what is happening in the commit messages: - The kernel internal struct iattr contains type safe vfs{g,u}id_t values clearly communicating that these values have to take a given mount's idmapping into account. - The ownership values placed in struct iattr to change ownership are identical for idmapped and non-idmapped mounts going forward. This also allows to simplify stacking filesystems such as overlayfs that change attributes In other words, they always represent the values. - Instead of open coding checks for whether ownership changes have been requested and an actual update of the inode is required we now have small static inline wrappers that abstract this logic away removing a lot of code duplication from individual filesystems that all open-coded the same checks" * tag 'fs.idmapped.vfsuid.v5.20' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux: mnt_idmapping: align kernel doc and parameter order mnt_idmapping: use new helpers in mapped_fs{g,u}id() fs: port HAS_UNMAPPED_ID() to vfs{g,u}id_t mnt_idmapping: return false when comparing two invalid ids attr: fix kernel doc attr: port attribute changes to new types security: pass down mount idmapping to setattr hook quota: port quota helpers mount ids fs: port to iattr ownership update helpers fs: introduce tiny iattr ownership update helpers fs: use mount types in iattr fs: add two type safe mapping helpers mnt_idmapping: add vfs{g,u}id_t
Diffstat (limited to 'fs/reiserfs')
-rw-r--r--fs/reiserfs/inode.c4
1 files changed, 2 insertions, 2 deletions
diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c
index c68e9329a9b0..0df48d176732 100644
--- a/fs/reiserfs/inode.c
+++ b/fs/reiserfs/inode.c
@@ -3276,7 +3276,7 @@ int reiserfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
/* must be turned off for recursive notify_change calls */
ia_valid = attr->ia_valid &= ~(ATTR_KILL_SUID|ATTR_KILL_SGID);
- if (is_quota_modification(inode, attr)) {
+ if (is_quota_modification(mnt_userns, inode, attr)) {
error = dquot_initialize(inode);
if (error)
return error;
@@ -3359,7 +3359,7 @@ int reiserfs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
reiserfs_write_unlock(inode->i_sb);
if (error)
goto out;
- error = dquot_transfer(inode, attr);
+ error = dquot_transfer(mnt_userns, inode, attr);
reiserfs_write_lock(inode->i_sb);
if (error) {
journal_end(&th);