| Message ID | 20230910-acl-fix-v2-1-38d6caa81419@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] fs: add a new SB_NOUMASK flag | expand |
On Sun, Sep 10, 2023 at 03:30:48PM -0400, Jeff Layton wrote: > SB_POSIXACL must be set when a filesystem supports POSIX ACLs, but NFSv4 > also sets this flag to prevent the VFS from applying the umask on > newly-created files. NFSv4 doesn't support POSIX ACLs however, which > causes confusion when other subsystems try to test for them. > > Split the umask-stripping opt-out into a separate SB_NOUMASK flag, and > have NFSv4 set that instead of SB_POSIXACL. Fix the appropriate places > in the VFS to check for that flag (in addition to SB_POSIXACL) when > stripping the umask. Oh, I see you only raised SB_POSIXACL to avoid umask stripping. That's a bit weird indeed. Hm, since this is an internal, non-user changeable flag I think it might be better in s_iflags as SB_I_NOUMASK? Also allows us to avoid wasting an s_flags bit. > > Signed-off-by: Jeff Layton <jlayton@kernel.org> > --- > Yet another approach to fixing this issue. I think this way is probably > the best, since makes the purpose of these flags clearer, and stops NFS > from relying on SB_POSIXACL to avoid umask stripping. > --- > Changes in v2: > - new approach: add a new SB_NOUMASK flag that NFSv4 can use instead of > SB_POSIXACL > - Link to v1: https://lore.kernel.org/r/20230908-acl-fix-v1-1-1e6b76c8dcc8@kernel.org > --- > fs/init.c | 4 ++-- > fs/namei.c | 2 +- > fs/nfs/super.c | 2 +- > include/linux/fs.h | 4 +++- > 4 files changed, 7 insertions(+), 5 deletions(-) > > diff --git a/fs/init.c b/fs/init.c > index 9684406a8416..157404bb7d19 100644 > --- a/fs/init.c > +++ b/fs/init.c > @@ -153,7 +153,7 @@ int __init init_mknod(const char *filename, umode_t mode, unsigned int dev) > if (IS_ERR(dentry)) > return PTR_ERR(dentry); > > - if (!IS_POSIXACL(path.dentry->d_inode)) > + if (!IS_NOUMASK(path.dentry->d_inode)) > mode &= ~current_umask(); > error = security_path_mknod(&path, dentry, mode, dev); > if (!error) > @@ -229,7 +229,7 @@ int __init init_mkdir(const char *pathname, umode_t mode) > dentry = kern_path_create(AT_FDCWD, pathname, &path, LOOKUP_DIRECTORY); > if (IS_ERR(dentry)) > return PTR_ERR(dentry); > - if (!IS_POSIXACL(path.dentry->d_inode)) > + if (!IS_NOUMASK(path.dentry->d_inode)) > mode &= ~current_umask(); Could you please just convert them over to mode_strip_umask()? Seems I forgot these two places back then.
diff --git a/fs/init.c b/fs/init.c index 9684406a8416..157404bb7d19 100644 --- a/fs/init.c +++ b/fs/init.c @@ -153,7 +153,7 @@ int __init init_mknod(const char *filename, umode_t mode, unsigned int dev) if (IS_ERR(dentry)) return PTR_ERR(dentry); - if (!IS_POSIXACL(path.dentry->d_inode)) + if (!IS_NOUMASK(path.dentry->d_inode)) mode &= ~current_umask(); error = security_path_mknod(&path, dentry, mode, dev); if (!error) @@ -229,7 +229,7 @@ int __init init_mkdir(const char *pathname, umode_t mode) dentry = kern_path_create(AT_FDCWD, pathname, &path, LOOKUP_DIRECTORY); if (IS_ERR(dentry)) return PTR_ERR(dentry); - if (!IS_POSIXACL(path.dentry->d_inode)) + if (!IS_NOUMASK(path.dentry->d_inode)) mode &= ~current_umask(); error = security_path_mkdir(&path, dentry, mode); if (!error) diff --git a/fs/namei.c b/fs/namei.c index 567ee547492b..905fdfec5d26 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -3117,7 +3117,7 @@ EXPORT_SYMBOL(unlock_rename); */ static inline umode_t mode_strip_umask(const struct inode *dir, umode_t mode) { - if (!IS_POSIXACL(dir)) + if (!IS_NOUMASK(dir)) mode &= ~current_umask(); return mode; } diff --git a/fs/nfs/super.c b/fs/nfs/super.c index 0d6473cb00cb..81cd11c4db3d 100644 --- a/fs/nfs/super.c +++ b/fs/nfs/super.c @@ -1071,7 +1071,7 @@ static void nfs_fill_super(struct super_block *sb, struct nfs_fs_context *ctx) sb->s_export_op = &nfs_export_ops; break; case 4: - sb->s_flags |= SB_POSIXACL; + sb->s_flags |= SB_NOUMASK; sb->s_time_gran = 1; sb->s_time_min = S64_MIN; sb->s_time_max = S64_MAX; diff --git a/include/linux/fs.h b/include/linux/fs.h index 4aeb3fa11927..51ad013ca716 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1119,13 +1119,14 @@ extern int send_sigurg(struct fown_struct *fown); #define SB_NOATIME BIT(10) /* Do not update access times. */ #define SB_NODIRATIME BIT(11) /* Do not update directory access times */ #define SB_SILENT BIT(15) -#define SB_POSIXACL BIT(16) /* VFS does not apply the umask */ +#define SB_POSIXACL BIT(16) /* POSIX ACLs are supported (implies NOUMASK) */ #define SB_INLINECRYPT BIT(17) /* Use blk-crypto for encrypted files */ #define SB_KERNMOUNT BIT(22) /* this is a kern_mount call */ #define SB_I_VERSION BIT(23) /* Update inode I_version field */ #define SB_LAZYTIME BIT(25) /* Update the on-disk [acm]times lazily */ /* These sb flags are internal to the kernel */ +#define SB_NOUMASK BIT(20) /* VFS does not apply the umask */ #define SB_DEAD BIT(21) #define SB_DYING BIT(24) #define SB_SUBMOUNT BIT(26) @@ -2111,6 +2112,7 @@ static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL) +#define IS_NOUMASK(inode) __IS_FLG(inode, SB_POSIXACL|SB_NOUMASK) #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
SB_POSIXACL must be set when a filesystem supports POSIX ACLs, but NFSv4 also sets this flag to prevent the VFS from applying the umask on newly-created files. NFSv4 doesn't support POSIX ACLs however, which causes confusion when other subsystems try to test for them. Split the umask-stripping opt-out into a separate SB_NOUMASK flag, and have NFSv4 set that instead of SB_POSIXACL. Fix the appropriate places in the VFS to check for that flag (in addition to SB_POSIXACL) when stripping the umask. Signed-off-by: Jeff Layton <jlayton@kernel.org> --- Yet another approach to fixing this issue. I think this way is probably the best, since makes the purpose of these flags clearer, and stops NFS from relying on SB_POSIXACL to avoid umask stripping. --- Changes in v2: - new approach: add a new SB_NOUMASK flag that NFSv4 can use instead of SB_POSIXACL - Link to v1: https://lore.kernel.org/r/20230908-acl-fix-v1-1-1e6b76c8dcc8@kernel.org --- fs/init.c | 4 ++-- fs/namei.c | 2 +- fs/nfs/super.c | 2 +- include/linux/fs.h | 4 +++- 4 files changed, 7 insertions(+), 5 deletions(-) --- base-commit: a48fa7efaf1161c1c898931fe4c7f0070964233a change-id: 20230908-acl-fix-6f8f86930f32 Best regards,