Extend and tweak mapping support
Subsystems (10)
Christian Brauner
maintainercover Extend and tweak mapping support
From: Christian Brauner <[email protected]>
Hey,
This series extend the mapping infrastructure in order to support mapped mounts of mapped filesystems in the future.
Currently we only support mapped mounts of filesystems mounted without an idmapping. This was a consicous decision mentioned in multiple places. For example, see [1].
In our mapping documentation in [3] we explained in detail that it is perfectly fine to extend support for mapped mounts to filesystem's mounted with an idmapping should the need arise. The need has been there for some time now (cf. [2]).
Before we can port any such filesystem we need to first extend the mapping helpers to account for the filesystem's idmapping in the remapping helpers. This again, is explained at length in our documentation at [3].
Currently, the low-level mapping helpers implement the remapping algorithms described in [3] in a simplified manner. Because we could rely on the fact that all filesystems supporting mapped mounts are mounted without an idmapping the translation step from or into the filesystem idmapping could be skipped.
In order to support mapped mounts of filesystem's mountable with an idmapping the translation step we were able to skip before cannot be skipped anymore. A filesystem mounted with an idmapping is very likely to not use an identity mapping and will instead use a non-identity mapping. So the translation step from or into the filesystem's idmapping in the remapping algorithm cannot be skipped for such filesystems. More details with examples can be found in [3].
This series adds a few new as well as prepares and tweaks some already existing low-level mapping helpers to perform the full translation algorithm explained in [3]. The low-level helpers can be written in a way that they only perform the additional translation step when the filesystem is indeed mounted with an idmapping.
Since we don't yet support such a filesystem yet a kernel was compiled carrying a trivial patch making ext4 mountable with an idmapping:
# We're located on the host with the initial idmapping.
ubuntu@f2-vm:~$ cat /proc/self/uid_map
0 0 4294967295# Mount an ext4 filesystem with the initial idmapping.
ubuntu@f2-vm:~$ sudo mount -t ext4 /dev/loop0 /mnt
# The filesystem contains two files. One owned by id 0 and another one owned by
# id 1000 in the initial idmapping. ubuntu@f2-vm:~$ ls -al /mnt/ total 8 drwxrwxrwx 2 root root 4096 Nov 22 17:04 . drwxr-xr-x 24 root root 4096 Nov 20 11:24 .. -rw-r--r-- 1 root root 0 Nov 22 17:04 file_init_mapping_0 -rw-r--r-- 1 ubuntu ubuntu 0 Nov 22 17:04 file_init_mapping_1000
# Umount it again so we we can mount it in another namespace later.
ubuntu@f2-vm:~$ sudo umount /mnt
# Use the lxc-usernsexec binary to run a shell in a user and mount namespace
# with an idmapping of 0:10000:100000000.
#
# This idmapping will have the effect that files which are owned by i_{g,u}id
# 10000 and files that are owned by i_{g,u}id 11000 will be owned by {g,u}id
# 0 and {g,u}id 1000 with that namespace respectively.
ubuntu@f2-vm:~$ sudo lxc-usernsexec -m b:0:10000:100000000 -- bash# Verify that we're really running with the expected idmapping.
root@f2-vm:/home/ubuntu# cat /proc/self/uid_map
0 10000 100000000# Mount the ext4 filesystem in the user and mountns with the idmapping # 0:10000:100000000.
#
# Note, that this requires a test kernel that makes ext4 mountable in a
# non-initial userns. The patch is simply:
#
# diff --git a/fs/ext4/super.c b/fs/ext4/super.c
# index 4e33b5eca694..0221e8211e5b 100644
# --- a/fs/ext4/super.c
# +++ b/fs/ext4/super.c
# @@ -6584,7 +6584,7 @@ static struct file_system_type ext4_fs_type = {
# .name = "ext4",
# .mount = ext4_mount,
# .kill_sb = kill_block_super,
# - .fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
# + .fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP | FS_USERNS_MOUNT,
# };
# MODULE_ALIAS_FS("ext4");
root@f2-vm:/home/ubuntu# mount -t ext4 /dev/loop0 /mnt# We verify that we can still interact with the files and that they are
# correctly owned from witin the namespace.
#
# As mentioned before, a file owned by {g,u}id 0 within the userns is mapped to
# {g,u}id 10000 in the initial userns. So inode->i_{g,u}id will contain
# 10000. Similarly a file owned by {g,u}id 1000 within the userns means that
# inode->i_{g,u}id will contain 11000.
the inode->i_{g,u}id
root@f2-vm:/home/ubuntu# ls -al /mnt/
total 8
drwxrwxrwx 2 root root 4096 Nov 22 17:04 .
drwxr-xr-x 24 nobody nogroup 4096 Nov 20 11:24 ..
-rw-r--r-- 1 root root 0 Nov 22 17:04 file_init_mapping_0
-rw-r--r-- 1 ubuntu ubuntu 0 Nov 22 17:04 file_init_mapping_1000# Since we own the superblock we can also create idmapped mounts of the
# filesystem.
# First, we create an idmapped mount that maps {g,u}id 1000 to {g,u}id 0.
# Note that the idmapping 1000:0:1 as written from userspace is different from
# the idmapping the kernel actually stores. The idmapping is always translated
# into the initial userns. So the kernel actually stores the idmapping
# 1000:10000:1.
root@f2-vm:/home/ubuntu# /mount-idmapped --map-mount b:1000:0:1 /mnt /opt# Verify that files owned by {g,u}id 1000 are now owned by {g,u}id 0 and all
# others are owned by the overflow{g,u}id.
root@f2-vm:/home/ubuntu# ls -al /opt
total 8
drwxrwxrwx 2 nobody nogroup 4096 Nov 22 17:04 .
drwxr-xr-x 24 nobody nogroup 4096 Nov 20 11:24 ..
-rw-r--r-- 1 nobody nogroup 0 Nov 22 17:04 file_init_mapping_0
-rw-r--r-- 1 root root 0 Nov 22 17:04 file_init_mapping_1000# Umount the idmapped mount.
root@f2-vm:/mnt# umount /opt
# Now create another idmapped mount with a mapping of 0:20000:1000000. #
# Similar to what we said above the kernel doesn't store 0:20000:1000000 but # 0:30000:1000000.
root@f2-vm:/mnt# /mount-idmapped --map-mount b:0:20000:1000000 /mnt /opt
# Verify that files owned by {g,u}id 0 in the filesystem's idmapping are owned
# by {g,u}id 20000 in the mount's idmapping and that files owned by {g,u}id
# 10000 in the filesystem's idmapping are owned by {g,u}id 21000 in the
# mount's idmapping.
root@f2-vm:/mnt# ls -al /opt
total 8
drwxrwxrwx 2 20000 20000 4096 Nov 22 17:04 .
drwxr-xr-x 24 nobody nogroup 4096 Nov 20 11:24 ..
-rw-r--r-- 1 20000 20000 0 Nov 22 17:04 file_init_mapping_0
-rw-r--r-- 1 21000 21000 0 Nov 22 17:04 file_init_mapping_1000# Use the lxc-usernsexec binary to run a shell in a user and mount namespace
# with an idmapping of 0:20000:1000000. # # Similar to what we said above the kernel doesn't store 0:20000:1000000 but # 0:30000:1000000. root@f2-vm:/mnt# lxc-usernsexec -m b:0:20000:1000000 -- bash
# Verify that we managed to use the correct nested mapping.
root@f2-vm:/mnt# cat /proc/self/uid_map
0 20000 1000000# Verify that files owned by {g,u}id 20000 in the idmapped mount when accessed
# from the ancestor userns are now owned by {g,u}id 0 in the idmapped mount
# when accessed from the new namespace. Similarly, verify that files owned by
# {g,u}id 21000 in the idmapped mount when accessed from the ancestor userns
# are now owned by {g,u}id 1000 in the idmapped mount when accesed from the new
# namespace.
root@f2-vm:/mnt# ls -al /opt/
total 8
drwxrwxrwx 2 root root 4096 Nov 22 17:04 .
drwxr-xr-x 24 nobody nogroup 4096 Nov 20 11:24 ..
-rw-r--r-- 1 root root 0 Nov 22 17:04 file_init_mapping_0
-rw-r--r-- 1 ubuntu ubuntu 0 Nov 22 17:04 file_init_mapping_1000# Create a file as id 0 in the current namespace through the idmapped mount.
root@f2-vm:/mnt# touch /opt/file_from_nested_ns
# Verify that the newly created file is owned by {g,u}id 0.
root@f2-vm:/mnt# ls -al /opt total 8 drwxrwxrwx 2 root root 4096 Nov 22 17:32 . drwxr-xr-x 24 nobody nogroup 4096 Nov 20 11:24 .. -rw-r--r-- 1 root root 0 Nov 22 17:32 file_from_nested_ns -rw-r--r-- 1 root root 0 Nov 22 17:04 file_init_mapping_0 -rw-r--r-- 1 ubuntu ubuntu 0 Nov 22 17:04 file_init_mapping_1000
[1]: commit 2ca4dcc4909d ("fs/mount_setattr: tighten permission checks") [2]: https://github.com/containers/podman/issues/10374
[3]: Documentations/filesystems/idmappings.rst
Thanks!
Christian
Christian Brauner (10):
fs: add is_mapped_mnt() helper
fs: move mapping helpers
fs: tweak fsuidgid_has_mapping()
fs: account for filesystem mappings
docs: update mapping documentation
fs: use low-level mapping helpers
fs: remove unused low-level mapping helpers
fs: port higher-level mapping helpers
fs: add i_user_ns() helper
fs: support mapped mounts of mapped filesystems
Documentation/filesystems/idmappings.rst | 72 ------- fs/cachefiles/bind.c | 2 +- fs/ecryptfs/main.c | 2 +- fs/ksmbd/smbacl.c | 18 +- fs/ksmbd/smbacl.h | 4 +- fs/namespace.c | 38 ++-- fs/nfsd/export.c | 2 +- fs/open.c | 7 +- fs/overlayfs/super.c | 2 +- fs/posix_acl.c | 16 +- fs/proc_namespace.c | 2 +- fs/xfs/xfs_inode.c | 10 +- fs/xfs/xfs_symlink.c | 5 +- include/linux/fs.h | 141 ++++---------- include/linux/mnt_mapping.h | 234 +++++++++++++++++++++++ security/commoncap.c | 14 +- 16 files changed, 339 insertions(+), 230 deletions(-) create mode 100644 include/linux/mnt_mapping.h
base-commit: 136057256686de39cc3a07c2e39ef6bc43003ff6
Signature · 1 line
2.30.2
Christian Brauner
maintainerPATCH 1/10 fs: add is_mapped_mnt() helper
From: Christian Brauner <[email protected]>
Multiple places open-code the same check to determine whether a given mount is idmapped. Introduce a simple helper function that can be used instead. This allows us to get rid of the fragile open-coding. We will later change the check that is used to determine whether a given mount is idmapped. Introducing a helper allows us to do this in a single place instead of doing it for multiple places.
--- fs/cachefiles/bind.c | 2 +- fs/ecryptfs/main.c | 2 +- fs/namespace.c | 2 +- fs/nfsd/export.c | 2 +- fs/overlayfs/super.c | 2 +- fs/proc_namespace.c | 2 +- include/linux/fs.h | 14 ++++++++++++++ 7 files changed, 20 insertions(+), 6 deletions(-)
fs/cachefiles/bind.c +1 -1
@@ -117,7 +117,7 @@ static int cachefiles_daemon_add_cache(struct cachefiles_cache *cache) root = path.dentry; ret = -EINVAL ; -if (mnt_user_ns (path.mnt) != &init_user_ns) { +if (is_mapped_mnt (path.mnt)) {pr_warn ("File cache on idmapped mounts not supported" ); goto error_unsupported; }
fs/ecryptfs/main.c +1 -1
@@ -537,7 +537,7 @@ static struct dentry *ecryptfs_mount(struct file_system_type *fs_type, int flags goto out_free; } -if (mnt_user_ns (path.mnt) != &init_user_ns) { +if (is_mapped_mnt (path.mnt)) { rc = -EINVAL ;printk (KERN_ERR "Mounting on idmapped mounts currently disallowed\n" ); goto out_free;
fs/namespace.c +1 -1
@@ -3936,7 +3936,7 @@ static int can_idmap_mount(const struct mount_kattr *kattr, struct mount *mnt)* mapping. It makes things simpler and callers can just create * another bind-mount they can idmap if they want to. */ -if (mnt_user_ns (m) != &init_user_ns) +if (is_mapped_mnt (m))return -EPERM ;/* The underlying filesystem doesn't support idmapped mounts yet. */
fs/nfsd/export.c +1 -1
@@ -427,7 +427,7 @@ static int check_export(struct path *path, int *flags, unsigned char *uuid)return -EINVAL ; } -if (mnt_user_ns (path->mnt) != &init_user_ns) { +if (is_mapped_mnt (path->mnt)) {dprintk ("exp_export: export of idmapped mounts not yet supported.\n" );return -EINVAL ; }
fs/overlayfs/super.c +1 -1
@@ -873,7 +873,7 @@ static int ovl_mount_dir_noesc(const char *name, struct path *path)pr_err ("filesystem on '%s' not supported\n" , name); goto out_put; } -if (mnt_user_ns (path->mnt) != &init_user_ns) { +if (is_mapped_mnt (path->mnt)) {pr_err ("idmapped layers are currently not supported\n" ); goto out_put; }
fs/proc_namespace.c +1 -1
@@ -80,7 +80,7 @@ static void show_mnt_opts(struct seq_file *m, struct vfsmount *mnt)seq_puts (m, fs_infop->str); } -if (mnt_user_ns (mnt) != &init_user_ns) +if (is_mapped_mnt (mnt))seq_puts (m,",idmapped" ); }
include/linux/fs.h +14 -0
@@ -2725,6 +2725,20 @@ static inline struct user_namespace *file_mnt_user_ns(struct file *file) {return mnt_user_ns (file->f_path.mnt); } + +/** +* is_mapped_mnt - check whether a mount is mapped +* @mnt: the mount to check +* +* If @mnt has an idmapping attached to it @mnt is mapped. +* +* Return: true if mount is mapped, false if not. +*/ +static inline bool is_mapped_mnt (const struct vfsmount *mnt) +{ +return mnt_user_ns (mnt) != &init_user_ns; +} +extern long vfs_truncate (const struct path *,loff_t );int do_truncate (struct user_namespace *,struct dentry *,loff_t start,unsigned int time_attrs,struct file *filp);
Signature · 1 line
2.30.2
Amir Goldstein
maintainerRe: [PATCH 01/10] fs: add is_mapped_mnt() helper
Quoted 115 lines · Christian Brauner
From: Christian Brauner <[email protected]>
Multiple places open-code the same check to determine whether a given mount is idmapped. Introduce a simple helper function that can be used instead. This allows us to get rid of the fragile open-coding. We will later change the check that is used to determine whether a given mount is idmapped. Introducing a helper allows us to do this in a single place instead of doing it for multiple places.
Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>--- fs/cachefiles/bind.c | 2 +- fs/ecryptfs/main.c | 2 +- fs/namespace.c | 2 +- fs/nfsd/export.c | 2 +- fs/overlayfs/super.c | 2 +- fs/proc_namespace.c | 2 +- include/linux/fs.h | 14 ++++++++++++++ 7 files changed, 20 insertions(+), 6 deletions(-)fs/cachefiles/bind.c +1
-1@@ -117,7 +117,7 @@ static int cachefiles_daemon_add_cache(struct cachefiles_cache *cache) root = path.dentry; ret = -EINVAL ; -if (mnt_user_ns (path.mnt) != &init_user_ns) { +if (is_mapped_mnt (path.mnt)) {pr_warn ("File cache on idmapped mounts not supported" ); goto error_unsupported; }fs/ecryptfs/main.c +1
-1@@ -537,7 +537,7 @@ static struct dentry *ecryptfs_mount(struct file_system_type *fs_type, int flags goto out_free; } -if (mnt_user_ns (path.mnt) != &init_user_ns) { +if (is_mapped_mnt (path.mnt)) { rc = -EINVAL ;printk (KERN_ERR "Mounting on idmapped mounts currently disallowed\n" ); goto out_free;fs/namespace.c +1
-1@@ -3936,7 +3936,7 @@ static int can_idmap_mount(const struct mount_kattr *kattr, struct mount *mnt)* mapping. It makes things simpler and callers can just create * another bind-mount they can idmap if they want to. */ -if (mnt_user_ns (m) != &init_user_ns) +if (is_mapped_mnt (m))return -EPERM ;/* The underlying filesystem doesn't support idmapped mounts yet. */ fs/nfsd/export.c +1
-1@@ -427,7 +427,7 @@ static int check_export(struct path *path, int *flags, unsigned char *uuid)return -EINVAL ; } -if (mnt_user_ns (path->mnt) != &init_user_ns) { +if (is_mapped_mnt (path->mnt)) {dprintk ("exp_export: export of idmapped mounts not yet supported.\n" );return -EINVAL ; }fs/overlayfs/super.c +1
-1@@ -873,7 +873,7 @@ static int ovl_mount_dir_noesc(const char *name, struct path *path)pr_err ("filesystem on '%s' not supported\n" , name); goto out_put; } -if (mnt_user_ns (path->mnt) != &init_user_ns) { +if (is_mapped_mnt (path->mnt)) {pr_err ("idmapped layers are currently not supported\n" ); goto out_put; }fs/proc_namespace.c +1
-1@@ -80,7 +80,7 @@ static void show_mnt_opts(struct seq_file *m, struct vfsmount *mnt)seq_puts (m, fs_infop->str); } -if (mnt_user_ns (mnt) != &init_user_ns) +if (is_mapped_mnt (mnt))seq_puts (m,",idmapped" ); }include/linux/fs.h +4
-0@@ -2725,6 +2725,20 @@ static inline struct user_namespace *file_mnt_user_ns(struct file *file) {return mnt_user_ns (file->f_path.mnt); } + +/** +* is_mapped_mnt - check whether a mount is mapped +* @mnt: the mount to check
include/linux/fs.h static inline struct user_namespace *file_mnt_user_ns(struct file *file) · line 2741
Quoted 10 lines
include/linux/fs.h +10
-0+* +* If @mnt has an idmapping attached to it @mnt is mapped. +* +* Return: true if mount is mapped, false if not. +*/ +static inline bool is_mapped_mnt (const struct vfsmount *mnt) +{ +return mnt_user_ns (mnt) != &init_user_ns; +} +
Maybe is_idmapped_mnt?
Thanks,
Amir.
Re: [PATCH 01/10] fs: add is_mapped_mnt() helper
Quoted 120 lines · Amir Goldstein
Quoted 119 lines · Christian Brauner
From: Christian Brauner <[email protected]>
Multiple places open-code the same check to determine whether a given mount is idmapped. Introduce a simple helper function that can be used instead. This allows us to get rid of the fragile open-coding. We will later change the check that is used to determine whether a given mount is idmapped. Introducing a helper allows us to do this in a single place instead of doing it for multiple places.
Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>--- fs/cachefiles/bind.c | 2 +- fs/ecryptfs/main.c | 2 +- fs/namespace.c | 2 +- fs/nfsd/export.c | 2 +- fs/overlayfs/super.c | 2 +- fs/proc_namespace.c | 2 +- include/linux/fs.h | 14 ++++++++++++++ 7 files changed, 20 insertions(+), 6 deletions(-)fs/cachefiles/bind.c +1
-1@@ -117,7 +117,7 @@ static int cachefiles_daemon_add_cache(struct cachefiles_cache *cache) root = path.dentry; ret = -EINVAL ; -if (mnt_user_ns (path.mnt) != &init_user_ns) { +if (is_mapped_mnt (path.mnt)) {pr_warn ("File cache on idmapped mounts not supported" ); goto error_unsupported; }fs/ecryptfs/main.c +1
-1@@ -537,7 +537,7 @@ static struct dentry *ecryptfs_mount(struct file_system_type *fs_type, int flags goto out_free; } -if (mnt_user_ns (path.mnt) != &init_user_ns) { +if (is_mapped_mnt (path.mnt)) { rc = -EINVAL ;printk (KERN_ERR "Mounting on idmapped mounts currently disallowed\n" ); goto out_free;fs/namespace.c +1
-1@@ -3936,7 +3936,7 @@ static int can_idmap_mount(const struct mount_kattr *kattr, struct mount *mnt)* mapping. It makes things simpler and callers can just create * another bind-mount they can idmap if they want to. */ -if (mnt_user_ns (m) != &init_user_ns) +if (is_mapped_mnt (m))return -EPERM ;/* The underlying filesystem doesn't support idmapped mounts yet. */ fs/nfsd/export.c +1
-1@@ -427,7 +427,7 @@ static int check_export(struct path *path, int *flags, unsigned char *uuid)return -EINVAL ; } -if (mnt_user_ns (path->mnt) != &init_user_ns) { +if (is_mapped_mnt (path->mnt)) {dprintk ("exp_export: export of idmapped mounts not yet supported.\n" );return -EINVAL ; }fs/overlayfs/super.c +1
-1@@ -873,7 +873,7 @@ static int ovl_mount_dir_noesc(const char *name, struct path *path)pr_err ("filesystem on '%s' not supported\n" , name); goto out_put; } -if (mnt_user_ns (path->mnt) != &init_user_ns) { +if (is_mapped_mnt (path->mnt)) {pr_err ("idmapped layers are currently not supported\n" ); goto out_put; }fs/proc_namespace.c +1
-1@@ -80,7 +80,7 @@ static void show_mnt_opts(struct seq_file *m, struct vfsmount *mnt)seq_puts (m, fs_infop->str); } -if (mnt_user_ns (mnt) != &init_user_ns) +if (is_mapped_mnt (mnt))seq_puts (m,",idmapped" ); }include/linux/fs.h +8
-0@@ -2725,6 +2725,20 @@ static inline struct user_namespace *file_mnt_user_ns(struct file *file) {return mnt_user_ns (file->f_path.mnt); } + +/** +* is_mapped_mnt - check whether a mount is mapped +* @mnt: the mount to check +* +* If @mnt has an idmapping attached to it @mnt is mapped. +* +* Return: true if mount is mapped, false if not.
Quoted 8 lines
include/linux/fs.h static inline struct user_namespace *file_mnt_user_ns(struct file *file) · line 2741
Quoted 6 lines
include/linux/fs.h +6
-0+*/ +static inline bool is_mapped_mnt (const struct vfsmount *mnt) +{ +return mnt_user_ns (mnt) != &init_user_ns; +} +Maybe
is_idmapped_mnt?
Fine by me!
Christian Brauner
maintainerPATCH 2/10 fs: move mapping helpers
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
The low-level mapping helpers were so far crammed into fs.h. They are out of place there. The fs.h header should just contain the higher-level mapping helpers that interact directly with vfs objects such as struct super_block or struct inode and not the bare mapping helpers. Similarly, only vfs and specific fs code shall interact with low-level mapping helpers. And so they won't be made accessible automatically through regular {g,u}id helpers.
--- include/linux/fs.h | 91 +------------------------------- include/linux/mnt_mapping.h | 101 ++++++++++++++++++++++++++++++++++++ 2 files changed, 102 insertions(+), 90 deletions(-) create mode 100644 include/linux/mnt_mapping.h
include/linux/fs.h +1 -90
@@ -41,6 +41,7 @@#include <linux/stddef.h> #include <linux/mount.h> #include <linux/cred.h> +#include <linux/mnt_mapping.h> #include <asm/byteorder.h> #include <uapi/linux/fs.h>
@@ -1624,34 +1625,6 @@ static inline void i_gid_write(struct inode *inode, gid_t gid) inode->i_gid =make_kgid (inode->i_sb->s_user_ns, gid); } -/** -* kuid_into_mnt - map a kuid down into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kuid: kuid to be mapped -* -* Return: @kuid mapped according to @mnt_userns. -* If @kuid has no mapping INVALID_UID is returned. -*/ -static inline kuid_t kuid_into_mnt (struct user_namespace *mnt_userns, -kuid_t kuid) -{ -return make_kuid (mnt_userns,__kuid_val (kuid)); -} - -/** -* kgid_into_mnt - map a kgid down into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kgid: kgid to be mapped -* -* Return: @kgid mapped according to @mnt_userns. -* If @kgid has no mapping INVALID_GID is returned. -*/ -static inline kgid_t kgid_into_mnt (struct user_namespace *mnt_userns, -kgid_t kgid) -{ -return make_kgid (mnt_userns,__kgid_val (kgid)); -} -/** * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns * @mnt_userns: user namespace of the mount the inode was found from
@@ -1680,68 +1653,6 @@ static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,return kgid_into_mnt (mnt_userns, inode->i_gid); } -/** -* kuid_from_mnt - map a kuid up into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kuid: kuid to be mapped -* -* Return: @kuid mapped up according to @mnt_userns. -* If @kuid has no mapping INVALID_UID is returned. -*/ -static inline kuid_t kuid_from_mnt (struct user_namespace *mnt_userns, -kuid_t kuid) -{ -return KUIDT_INIT (from_kuid (mnt_userns, kuid)); -} - -/** -* kgid_from_mnt - map a kgid up into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kgid: kgid to be mapped -* -* Return: @kgid mapped up according to @mnt_userns. -* If @kgid has no mapping INVALID_GID is returned. -*/ -static inline kgid_t kgid_from_mnt (struct user_namespace *mnt_userns, -kgid_t kgid) -{ -return KGIDT_INIT (from_kgid (mnt_userns, kgid)); -} - -/** -* mapped_fsuid - return caller's fsuid mapped up into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* -* Use this helper to initialize a new vfs or filesystem object based on -* the caller's fsuid. A common example is initializing the i_uid field of -* a newly allocated inode triggered by a creation event such as mkdir or -* O_CREAT. Other examples include the allocation of quotas for a specific -* user. -* -* Return: the caller's current fsuid mapped up according to @mnt_userns. -*/ -static inline kuid_t mapped_fsuid (struct user_namespace *mnt_userns) -{ -return kuid_from_mnt (mnt_userns,current_fsuid ()); -} - -/** -* mapped_fsgid - return caller's fsgid mapped up into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* -* Use this helper to initialize a new vfs or filesystem object based on -* the caller's fsgid. A common example is initializing the i_gid field of -* a newly allocated inode triggered by a creation event such as mkdir or -* O_CREAT. Other examples include the allocation of quotas for a specific -* user. -* -* Return: the caller's current fsgid mapped up according to @mnt_userns. -*/ -static inline kgid_t mapped_fsgid (struct user_namespace *mnt_userns) -{ -return kgid_from_mnt (mnt_userns,current_fsgid ()); -} -/** * inode_fsuid_set - initialize inode's i_uid field with callers fsuid * @inode: inode to initialize
include/linux/mnt_mapping.h new file +101 -0
101 lines not on this page · raw message
Signature · 1 line
2.30.2
Amir Goldstein
maintainerRe: [PATCH 02/10] fs: move mapping helpers
Quoted 27 lines · Christian Brauner
From: Christian Brauner <[email protected]>
The low-level mapping helpers were so far crammed into
fs.h. They are out of place there. Thefs.hheader should just contain the higher-level mapping helpers that interact directly with vfs objects such as structsuper_blockor struct inode and not the bare mapping helpers. Similarly, only vfs and specific fs code shall interact with low-level mapping helpers. And so they won't be made accessible automatically through regular {g,u}id helpers.Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>--- include/linux/fs.h | 91 +------------------------------- include/linux/mnt_mapping.h | 101 ++++++++++++++++++++++++++++++++++++ 2 files changed, 102 insertions(+), 90 deletions(-) create mode 100644 include/linux/mnt_mapping.hinclude/linux/fs.h +0
-0
include/linux/fs.h line 44
Quoted 4 lines
include/linux/fs.h +1
-0@@ -41,6 +41,7 @@#include <linux/stddef.h> #include <linux/mount.h> #include <linux/cred.h> +#include <linux/mnt_mapping.h>
If I grepped correctly, there are ~20 files that use these helpers. Please put the include in those files, so changes to this header will not compile the world.
And how about mnt_idmapping.h or idmapped_mnt.h? Not sure if this naming issue was discussed already.
Thanks,
Amir.
Re: [PATCH 02/10] fs: move mapping helpers
Quoted 23 lines · Amir Goldstein
Quoted 22 lines · Christian Brauner
From: Christian Brauner <[email protected]>
The low-level mapping helpers were so far crammed into
fs.h. They are out of place there. Thefs.hheader should just contain the higher-level mapping helpers that interact directly with vfs objects such as structsuper_blockor struct inode and not the bare mapping helpers. Similarly, only vfs and specific fs code shall interact with low-level mapping helpers. And so they won't be made accessible automatically through regular {g,u}id helpers.Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>--- include/linux/fs.h | 91 +------------------------------- include/linux/mnt_mapping.h | 101 ++++++++++++++++++++++++++++++++++++ 2 files changed, 102 insertions(+), 90 deletions(-) create mode 100644 include/linux/mnt_mapping.h
Quoted 13 lines
include/linux/fs.h line 44
Quoted 9 lines
include/linux/fs.h +1
-0@@ -41,6 +41,7 @@#include <linux/stddef.h> #include <linux/mount.h> #include <linux/cred.h> +#include <linux/mnt_mapping.h> If I grepped correctly, there are ~20 files that use these helpers. Please put the include in those files, so changes to this header will not compile the world.
Ok, happy to.
Quoted 2 lines
And how about
mnt_idmapping.horidmapped_mnt.h?
I think then I'll opt for mnt_idmapping.h.
Quoted 1 line
Not sure if this naming issue was discussed already.
Nope!
Thanks!
Christian
Christian Brauner
maintainerPATCH 3/10 fs: tweak fsuidgid_has_mapping()
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
If the caller's fs{g,u}id aren't mapped in the mount's idmapping we can return early and skip the check whether the mapped fs{g,u}id also have a mapping in the filesystem's idmapping. If the fs{g,u}id aren't mapped in the mount's idmapping they consequently can't be mapped in the filesystem's idmapping. So there's no point in checking that.
--- include/linux/fs.h | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-)
include/linux/fs.h +11 -3
@@ -1695,10 +1695,18 @@ static inline void inode_fsgid_set(struct inode *inode,static inline bool fsuidgid_has_mapping (struct super_block *sb,struct user_namespace *mnt_userns) { -struct user_namespace *s_user_ns = sb->s_user_ns; +struct user_namespace *fs_userns = sb->s_user_ns; +kuid_t kuid; +kgid_t kgid; -return kuid_has_mapping (s_user_ns,mapped_fsuid (mnt_userns)) && -kgid_has_mapping (s_user_ns,mapped_fsgid (mnt_userns)); + kuid =mapped_fsuid (mnt_userns); +if (!uid_valid (kuid)) +return false; + kgid =mapped_fsgid (mnt_userns); +if (!gid_valid (kgid)) +return false; +return kuid_has_mapping (fs_userns, kuid) && +kgid_has_mapping (fs_userns, kgid); }extern struct timespec64 current_time (struct inode *inode);
Signature · 1 line
2.30.2
Re: [PATCH 03/10] fs: tweak fsuidgid_has_mapping()
Quoted 9 lines · Christian Brauner
From: Christian Brauner <[email protected]>
If the caller's fs{g,u}id aren't mapped in the mount's idmapping we can return early and skip the check whether the mapped fs{g,u}id also have a mapping in the filesystem's idmapping. If the fs{g,u}id aren't mapped in the mount's idmapping they consequently can't be mapped in the filesystem's idmapping. So there's no point in checking that.
Quoted 6 lines
Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>---
Quoted 32 lines
include/linux/fs.h | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-)include/linux/fs.h +11
-4@@ -1695,10 +1695,18 @@ static inline void inode_fsgid_set(struct inode *inode,static inline bool fsuidgid_has_mapping (struct super_block *sb,struct user_namespace *mnt_userns) { -struct user_namespace *s_user_ns = sb->s_user_ns; +struct user_namespace *fs_userns = sb->s_user_ns; +kuid_t kuid; +kgid_t kgid; -return kuid_has_mapping (s_user_ns,mapped_fsuid (mnt_userns)) && -kgid_has_mapping (s_user_ns,mapped_fsgid (mnt_userns)); + kuid =mapped_fsuid (mnt_userns); +if (!uid_valid (kuid)) +return false; + kgid =mapped_fsgid (mnt_userns); +if (!gid_valid (kgid)) +return false; +return kuid_has_mapping (fs_userns, kuid) && +kgid_has_mapping (fs_userns, kgid); }extern struct timespec64 current_time (struct inode *inode); --2.30.2
Christian Brauner
maintainerPATCH 4/10 fs: account for filesystem mappings
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
Currently we only support idmapped mounts for filesystems mounted without an idmapping. This was a conscious decision mentioned in multiple places (cf. e.g. [1]).
As explained at length in [3] it is perfectly fine to extend support for idmapped mounts to filesystem's mounted with an idmapping should the need arise. The need has been there for some time now. Various container projects in userspace need this to run unprivileged and nested unprivileged containers (cf. [2]).
Before we can port any filesystem that is mountable with an idmapping to support idmapped mounts we need to first extend the mapping helpers to account for the filesystem's idmapping. This again, is explained at length in our documentation at [3] but I'll give an overview here again.
Currently, the low-level mapping helpers implement the remapping algorithms described in [3] in a simplified manner. Because we could rely on the fact that all filesystems supporting idmapped mounts are mounted without an idmapping the translation step from or into the filesystem idmapping could be skipped.
In order to support idmapped mounts of filesystem's mountable with an idmapping the translation step we were able to skip before cannot be skipped anymore. A filesystem mounted with an idmapping is very likely to not use an identity mapping and will instead use a non-identity mapping. So the translation step from or into the filesystem's idmapping in the remapping algorithm cannot be skipped for such filesystems. More details with examples can be found in [3].
This patch adds a few new and prepares some already existing low-level mapping helpers to perform the full translation algorithm explained in [3]. The low-level helpers can be written in a way that they only perform the additional translation step when the filesystem is indeed mounted with an idmapping.
If the low-level helpers detect that they are not dealing with an idmapped mount they can simply return the relevant k{g,u}id unchanged; no remapping needs to be performed at all. The no_mapping() helper detects whether the shortcut can be used.
If the low-level helpers detected that they are dealing with an idmapped mount but the underlying filesystem is mounted without an idmapping we can rely on the previous shorcut and can continue to skip the translation step from or into the filesystem's idmapping.
These checks guarantee that only the minimal amount of work is performed. As before, if idmapped mounts aren't used the low-level helpers are idempotent and no work is performed at all.
This patch adds the helpers mapped_k{g,u}id_fs() and mapped_k{g,u}id_user(). Following patches will port all places to replace the old k{g,u}id_into_mnt() and k{g,u}id_from_mnt() with these two new helpers. After the conversion is done k{g,u}id_into_mnt() and k{g,u}id_from_mnt() will be removed. This also concludes the renaming of the mapping helpers we started in [4]. Now, all mapping helpers will started with the "mapped_" prefix making everything nice and consistent.
The mapped_k{g,u}id_fs() helpers replace the k{g,u}id_into_mnt() helpers. They are to be used when k{g,u}ids are to be mapped from the vfs, e.g. from from struct inode's i_{g,u}id. Conversely, the mapped_k{g,u}id_user() helpers replace the k{g,u}id_from_mnt() helpers. They are to be used when k{g,u}ids are to be written to disk, e.g. when entering from a system call to change ownership of a file.
This patch only introduces the helpers. It doesn't yet convert the relevant places to account for filesystem mounted with an idmapping.
[1]: commit 2ca4dcc4909d ("fs/mount_setattr: tighten permission checks") [2]: https://github.com/containers/podman/issues/10374
[3]: Documentations/filesystems/idmappings.rst
[4]: commit a65e58e791a1 ("fs: document and rename fsid helpers")
--- include/linux/fs.h | 4 +- include/linux/mnt_mapping.h | 193 +++++++++++++++++++++++++++++++++++- 2 files changed, 191 insertions(+), 6 deletions(-)
include/linux/fs.h +2 -2
@@ -1636,7 +1636,7 @@ static inline void i_gid_write(struct inode *inode, gid_t gid)static inline kuid_t i_uid_into_mnt (struct user_namespace *mnt_userns,const struct inode *inode) { -return kuid_into_mnt (mnt_userns, inode->i_uid); +return mapped_kuid_fs (mnt_userns, &init_user_ns, inode->i_uid); }/**
@@ -1650,7 +1650,7 @@ static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,static inline kgid_t i_gid_into_mnt (struct user_namespace *mnt_userns,const struct inode *inode) { -return kgid_into_mnt (mnt_userns, inode->i_gid); +return mapped_kgid_fs (mnt_userns, &init_user_ns, inode->i_gid); }/**
include/linux/mnt_mapping.h +189 -4
223 lines not on this page · raw message
Signature · 1 line
2.30.2
Christian Brauner
maintainerPATCH 5/10 docs: update mapping documentation
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
Now that we implement the full remapping algorithms described in our documentation remove the section about shortcircuting them.
--- Documentation/filesystems/idmappings.rst | 72 ------------------------ 1 file changed, 72 deletions(-)
Documentation/filesystems/idmappings.rst +0 -72
@@ -952,75 +952,3 @@ The raw userspace id that is put on disk is ``u1000`` so when the user takes their home directory back to their home computer where they are assigned ``u1000`` using the initial idmapping and mount the filesystem with the initial idmapping they will see all those files owned by ``u1000``. - -Shortcircuting --------------- - -Currently, the implementation of idmapped mounts enforces that the filesystem -is mounted with the initial idmapping. The reason is simply that none of the -filesystems that we targeted were mountable with a non-initial idmapping. But -that might change soon enough. As we've seen above, thanks to the properties of -idmappings the translation works for both filesystems mounted with the initial -idmapping and filesystem with non-initial idmappings. - -Based on this current restriction to filesystem mounted with the initial -idmapping two noticeable shortcuts have been taken: - -1. We always stash a reference to the initial user namespace in ``struct - vfsmount``. Idmapped mounts are thus mounts that have a non-initial user - namespace attached to them. - - In order to support idmapped mounts this needs to be changed. Instead of - stashing the initial user namespace the user namespace the filesystem was - mounted with must be stashed. An idmapped mount is then any mount that has - a different user namespace attached then the filesystem was mounted with. - This has no user-visible consequences. - -2. The translation algorithms in ``mapped_fs*id()`` and ``i_*id_into_mnt()`` - are simplified. - - Let's consider ``mapped_fs*id()`` first. This function translates the - caller's kernel id into a kernel id in the filesystem's idmapping via - a mount's idmapping. The full algorithm is:: - - mapped_fsuid(kid): - /* Map the kernel id up into a userspace id in the mount's idmapping. */ - from_kuid(mount-idmapping, kid) = uid - - /* Map the userspace id down into a kernel id in the filesystem's idmapping. */ - make_kuid(filesystem-idmapping, uid) = kuid - - We know that the filesystem is always mounted with the initial idmapping as - we enforce this in ``mount_setattr()``. So this can be shortened to:: - - mapped_fsuid(kid): - /* Map the kernel id up into a userspace id in the mount's idmapping. */ - from_kuid(mount-idmapping, kid) = uid - - /* Map the userspace id down into a kernel id in the filesystem's idmapping. */ - KUIDT_INIT(uid) = kuid - - Similarly, for ``i_*id_into_mnt()`` which translated the filesystem's kernel - id into a mount's kernel id:: - - i_uid_into_mnt(kid): - /* Map the kernel id up into a userspace id in the filesystem's idmapping. */ - from_kuid(filesystem-idmapping, kid) = uid - - /* Map the userspace id down into a kernel id in the mounts's idmapping. */ - make_kuid(mount-idmapping, uid) = kuid - - Again, we know that the filesystem is always mounted with the initial - idmapping as we enforce this in ``mount_setattr()``. So this can be - shortened to:: - - i_uid_into_mnt(kid): - /* Map the kernel id up into a userspace id in the filesystem's idmapping. */ - __kuid_val(kid) = uid - - /* Map the userspace id down into a kernel id in the mounts's idmapping. */ - make_kuid(mount-idmapping, uid) = kuid - -Handling filesystems mounted with non-initial idmappings requires that the -translation functions be converted to their full form. They can still be -shortcircuited on non-idmapped mounts. This has no user-visible consequences.
Signature · 1 line
2.30.2
Christian Brauner
maintainerPATCH 6/10 fs: use low-level mapping helpers
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
In a few places the vfs needs to interact with bare k{g,u}ids directly instead of struct inode. These are just a few. In previous patches we introduced low-level mapping helpers that are able to support filesystems mounted an idmapping. This patch simply converts the places to use these new helpers.
--- fs/ksmbd/smbacl.c | 18 ++---------------- fs/ksmbd/smbacl.h | 4 ++-- fs/open.c | 4 ++-- fs/posix_acl.c | 16 ++++++++++------ security/commoncap.c | 13 ++++++++----- 5 files changed, 24 insertions(+), 31 deletions(-)
fs/ksmbd/smbacl.c +2 -16
@@ -274,14 +274,7 @@ static int sid_to_id(struct user_namespace *user_ns,uid_t id; id =le32_to_cpu (psid->sub_auth[psid->num_subauth -1 ]); -/* -* Translate raw sid into kuid in the server's user -* namespace. -*/ - uid =make_kuid (&init_user_ns, id); - -/* If this is an idmapped mount, apply the idmapping. */ - uid =kuid_from_mnt (user_ns, uid); + uid =mapped_kuid_user (user_ns, &init_user_ns,KUIDT_INIT (id));if (uid_valid (uid)) { fattr->cf_uid = uid; rc =0 ;
@@ -291,14 +284,7 @@ static int sid_to_id(struct user_namespace *user_ns,gid_t id; id =le32_to_cpu (psid->sub_auth[psid->num_subauth -1 ]); -/* -* Translate raw sid into kgid in the server's user -* namespace. -*/ - gid =make_kgid (&init_user_ns, id); - -/* If this is an idmapped mount, apply the idmapping. */ - gid =kgid_from_mnt (user_ns, gid); + gid =mapped_kgid_user (user_ns, &init_user_ns,KGIDT_INIT (id));if (gid_valid (gid)) { fattr->cf_gid = gid; rc =0 ;
fs/ksmbd/smbacl.h +2 -2
@@ -216,7 +216,7 @@ static inline uid_t posix_acl_uid_translate(struct user_namespace *mnt_userns,kuid_t kuid;/* If this is an idmapped mount, apply the idmapping. */ - kuid =kuid_into_mnt (mnt_userns, pace->e_uid); + kuid =mapped_kuid_fs (mnt_userns, &init_user_ns, pace->e_uid);/* Translate the kuid into a userspace id ksmbd would see. */ return from_kuid (&init_user_ns, kuid);
@@ -228,7 +228,7 @@ static inline gid_t posix_acl_gid_translate(struct user_namespace *mnt_userns,kgid_t kgid;/* If this is an idmapped mount, apply the idmapping. */ - kgid =kgid_into_mnt (mnt_userns, pace->e_gid); + kgid =mapped_kgid_fs (mnt_userns, &init_user_ns, pace->e_gid);/* Translate the kgid into a userspace id ksmbd would see. */ return from_kgid (&init_user_ns, kgid);
fs/open.c +2 -2
@@ -652,8 +652,8 @@ int chown_common(const struct path *path, uid_t user, gid_t group) gid =make_kgid (current_user_ns (), group); mnt_userns =mnt_user_ns (path->mnt); - uid =kuid_from_mnt (mnt_userns, uid); - gid =kgid_from_mnt (mnt_userns, gid); + uid =mapped_kuid_user (mnt_userns, &init_user_ns, uid); + gid =mapped_kgid_user (mnt_userns, &init_user_ns, gid); retry_deleg: newattrs.ia_valid =ATTR_CTIME ;
fs/posix_acl.c +10 -6
@@ -374,7 +374,9 @@ posix_acl_permission(struct user_namespace *mnt_userns, struct inode *inode, goto check_perm;break ;case ACL_USER : - uid =kuid_into_mnt (mnt_userns, pa->e_uid); + uid =mapped_kuid_fs (mnt_userns, + &init_user_ns, + pa->e_uid);if (uid_eq (uid,current_fsuid ())) goto mask;break ;
@@ -387,7 +389,9 @@ posix_acl_permission(struct user_namespace *mnt_userns, struct inode *inode, }break ;case ACL_GROUP : - gid =kgid_into_mnt (mnt_userns, pa->e_gid); + gid =mapped_kgid_fs (mnt_userns, + &init_user_ns, + pa->e_gid);if (in_group_p (gid)) { found =1 ;if ((pa->e_perm & want) == want)
@@ -734,17 +738,17 @@ static void posix_acl_fix_xattr_userns(case ACL_USER : uid =make_kuid (from,le32_to_cpu (entry->e_id));if (from_user) - uid =kuid_from_mnt (mnt_userns, uid); + uid =mapped_kuid_user (mnt_userns, &init_user_ns, uid);else - uid =kuid_into_mnt (mnt_userns, uid); + uid =mapped_kuid_fs (mnt_userns, &init_user_ns, uid); entry->e_id =cpu_to_le32 (from_kuid (to, uid));break ;case ACL_GROUP : gid =make_kgid (from,le32_to_cpu (entry->e_id));if (from_user) - gid =kgid_from_mnt (mnt_userns, gid); + gid =mapped_kgid_user (mnt_userns, &init_user_ns, gid);else - gid =kgid_into_mnt (mnt_userns, gid); + gid =mapped_kgid_fs (mnt_userns, &init_user_ns, gid); entry->e_id =cpu_to_le32 (from_kgid (to, gid));break ;default :
security/commoncap.c +8 -5
@@ -418,7 +418,7 @@ int cap_inode_getsecurity(struct user_namespace *mnt_userns, kroot =make_kuid (fs_ns, root);/* If this is an idmapped mount shift the kuid. */ - kroot =kuid_into_mnt (mnt_userns, kroot); + kroot =mapped_kuid_fs (mnt_userns, &init_user_ns, kroot);/* If the root kuid maps to a valid uid in current ns, then return * this as a nscap. */
@@ -488,6 +488,7 @@ int cap_inode_getsecurity(struct user_namespace *mnt_userns,* @size: size of @ivalue * @task_ns: user namespace of the caller * @mnt_userns: user namespace of the mount the inode was found from +* @fs_userns: user namespace of the filesystem * * If the inode has been found through an idmapped mount the user namespace of * the vfsmount must be passed through @mnt_userns. This function will then
@@ -497,7 +498,8 @@ int cap_inode_getsecurity(struct user_namespace *mnt_userns,*/ static kuid_t rootid_from_xattr (const void *value,size_t size,struct user_namespace *task_ns, -struct user_namespace *mnt_userns) +struct user_namespace *mnt_userns, +struct user_namespace *fs_userns) {const struct vfs_ns_cap_data *nscap = value;kuid_t rootkid;
@@ -507,7 +509,7 @@ static kuid_t rootid_from_xattr(const void *value, size_t size, rootid =le32_to_cpu (nscap->rootid); rootkid =make_kuid (task_ns, rootid); -return kuid_from_mnt (mnt_userns, rootkid); +return mapped_kuid_user (mnt_userns, fs_userns, rootkid); }static bool validheader (size_t size,const struct vfs_cap_data *cap)
@@ -558,7 +560,8 @@ int cap_convert_nscap(struct user_namespace *mnt_userns, struct dentry *dentry,/* user is privileged, just write the v2 */ return size; - rootid =rootid_from_xattr (*ivalue, size, task_ns, mnt_userns); + rootid =rootid_from_xattr (*ivalue, size, task_ns, mnt_userns, + &init_user_ns);if (!uid_valid (rootid))return -EINVAL ;
@@ -699,7 +702,7 @@ int get_vfs_caps_from_disk(struct user_namespace *mnt_userns,/* Limit the caps to the mounter of the filesystem * or the more limited uid specified in the xattr. */ - rootkuid =kuid_into_mnt (mnt_userns, rootkuid); + rootkuid =mapped_kuid_fs (mnt_userns, &init_user_ns, rootkuid);if (!rootid_owns_currentns (rootkuid))return -ENODATA ;
Signature · 1 line
2.30.2
Christian Brauner
maintainerPATCH 7/10 fs: remove unused low-level mapping helpers
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
Now that we ported all places to use the new low-level mapping helpers that are able to support filesystems mounted with an idmapping we can remove the old low-level mapping helpers. With the removal of these old helpers we also conclude the renaming of the mapping helpers we started in [1].
[1]: commit a65e58e791a1 ("fs: document and rename fsid helpers")
--- include/linux/mnt_mapping.h | 56 ------------------------------------- 1 file changed, 56 deletions(-)
include/linux/mnt_mapping.h +0 -56
@@ -13,62 +13,6 @@ struct user_namespace;*/ extern struct user_namespace init_user_ns; -/** -* kuid_into_mnt - map a kuid down into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kuid: kuid to be mapped -* -* Return: @kuid mapped according to @mnt_userns. -* If @kuid has no mapping INVALID_UID is returned. -*/ -static inline kuid_t kuid_into_mnt (struct user_namespace *mnt_userns, -kuid_t kuid) -{ -return make_kuid (mnt_userns,__kuid_val (kuid)); -} - -/** -* kgid_into_mnt - map a kgid down into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kgid: kgid to be mapped -* -* Return: @kgid mapped according to @mnt_userns. -* If @kgid has no mapping INVALID_GID is returned. -*/ -static inline kgid_t kgid_into_mnt (struct user_namespace *mnt_userns, -kgid_t kgid) -{ -return make_kgid (mnt_userns,__kgid_val (kgid)); -} - -/** -* kuid_from_mnt - map a kuid up into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kuid: kuid to be mapped -* -* Return: @kuid mapped up according to @mnt_userns. -* If @kuid has no mapping INVALID_UID is returned. -*/ -static inline kuid_t kuid_from_mnt (struct user_namespace *mnt_userns, -kuid_t kuid) -{ -return KUIDT_INIT (from_kuid (mnt_userns, kuid)); -} - -/** -* kgid_from_mnt - map a kgid up into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kgid: kgid to be mapped -* -* Return: @kgid mapped up according to @mnt_userns. -* If @kgid has no mapping INVALID_GID is returned. -*/ -static inline kgid_t kgid_from_mnt (struct user_namespace *mnt_userns, -kgid_t kgid) -{ -return KGIDT_INIT (from_kgid (mnt_userns, kgid)); -} -/** * initial_idmapping - check whether this is the initial mapping * @ns: idmapping to check
Signature · 1 line
2.30.2
Re: [PATCH 07/10] fs: remove unused low-level mapping helpers
Quoted 9 lines · Christian Brauner
From: Christian Brauner <[email protected]>
Now that we ported all places to use the new low-level mapping helpers that are able to support filesystems mounted with an idmapping we can remove the old low-level mapping helpers. With the removal of these old helpers we also conclude the renaming of the mapping helpers we started in [1].
Quoted 7 lines
[1]: commit
a65e58e791a1("fs: document and rename fsid helpers")Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>---
Quoted 73 lines
include/linux/mnt_mapping.h | 56 ------------------------------------- 1 file changed, 56 deletions(-)include/linux/mnt_mapping.h +0
-57@@ -13,62 +13,6 @@ struct user_namespace;*/ extern struct user_namespace init_user_ns; -/** -* kuid_into_mnt - map a kuid down into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kuid: kuid to be mapped -* -* Return: @kuid mapped according to @mnt_userns. -* If @kuid has no mapping INVALID_UID is returned. -*/ -static inline kuid_t kuid_into_mnt (struct user_namespace *mnt_userns, -kuid_t kuid) -{ -return make_kuid (mnt_userns,__kuid_val (kuid)); -} - -/** -* kgid_into_mnt - map a kgid down into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kgid: kgid to be mapped -* -* Return: @kgid mapped according to @mnt_userns. -* If @kgid has no mapping INVALID_GID is returned. -*/ -static inline kgid_t kgid_into_mnt (struct user_namespace *mnt_userns, -kgid_t kgid) -{ -return make_kgid (mnt_userns,__kgid_val (kgid)); -} - -/** -* kuid_from_mnt - map a kuid up into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kuid: kuid to be mapped -* -* Return: @kuid mapped up according to @mnt_userns. -* If @kuid has no mapping INVALID_UID is returned. -*/ -static inline kuid_t kuid_from_mnt (struct user_namespace *mnt_userns, -kuid_t kuid) -{ -return KUIDT_INIT (from_kuid (mnt_userns, kuid)); -} - -/** -* kgid_from_mnt - map a kgid up into a mnt_userns -* @mnt_userns: user namespace of the relevant mount -* @kgid: kgid to be mapped -* -* Return: @kgid mapped up according to @mnt_userns. -* If @kgid has no mapping INVALID_GID is returned. -*/ -static inline kgid_t kgid_from_mnt (struct user_namespace *mnt_userns, -kgid_t kgid) -{ -return KGIDT_INIT (from_kgid (mnt_userns, kgid)); -} -/** * initial_idmapping - check whether this is the initial mapping * @ns: idmapping to check -- 2.30.2
Christian Brauner
maintainerPATCH 8/10 fs: port higher-level mapping helpers
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
Enable the mapped_fs{g,u}id() helpers to support filesystems mounted with an idmapping. Apart from core mapping helpers that use mapped_fs{g,u}id() to initialize struct inode's i_{g,u}id fields xfs is the only place that uses these low-level helpers directly.
The patch only extends the helpers to be able to take the filesystem idmapping into account. Since we don't actually yet pass the filesystem's idmapping in no functional changes happen. This will happen in a final patch.
--- fs/xfs/xfs_inode.c | 10 ++++++---- fs/xfs/xfs_symlink.c | 5 +++-- include/linux/fs.h | 8 ++++---- include/linux/mnt_mapping.h | 12 ++++++++---- 4 files changed, 21 insertions(+), 14 deletions(-)
fs/xfs/xfs_inode.c +6 -4
@@ -977,6 +977,7 @@ xfs_create(struct xfs_trans_res *tres;uint resblks;xfs_ino_t ino; +struct user_namespace *fs_userns = &init_user_ns;trace_xfs_create (dp, name);
@@ -988,8 +989,8 @@ xfs_create(/* * Make sure that we have allocated dquot(s) on disk. */ - error =xfs_qm_vop_dqalloc (dp,mapped_fsuid (mnt_userns), -mapped_fsgid (mnt_userns), prid, + error =xfs_qm_vop_dqalloc (dp,mapped_fsuid (mnt_userns, fs_userns), +mapped_fsgid (mnt_userns, fs_userns), prid,XFS_QMOPT_QUOTALL |XFS_QMOPT_INHERIT , &udqp, &gdqp, &pdqp);if (error)
@@ -1133,6 +1134,7 @@ xfs_create_tmpfile(struct xfs_trans_res *tres;uint resblks;xfs_ino_t ino; +struct user_namespace *fs_userns = &init_user_ns;if (xfs_is_shutdown (mp))return -EIO ;
@@ -1142,8 +1144,8 @@ xfs_create_tmpfile(/* * Make sure that we have allocated dquot(s) on disk. */ - error =xfs_qm_vop_dqalloc (dp,mapped_fsuid (mnt_userns), -mapped_fsgid (mnt_userns), prid, + error =xfs_qm_vop_dqalloc (dp,mapped_fsuid (mnt_userns, fs_userns), +mapped_fsgid (mnt_userns, fs_userns), prid,XFS_QMOPT_QUOTALL |XFS_QMOPT_INHERIT , &udqp, &gdqp, &pdqp);if (error)
fs/xfs/xfs_symlink.c +3 -2
@@ -163,6 +163,7 @@ xfs_symlink(struct xfs_dquot *pdqp =NULL ;uint resblks;xfs_ino_t ino; +struct user_namespace *fs_userns = &init_user_ns; *ipp =NULL ;
@@ -184,8 +185,8 @@ xfs_symlink(/* * Make sure that we have allocated dquot(s) on disk. */ - error =xfs_qm_vop_dqalloc (dp,mapped_fsuid (mnt_userns), -mapped_fsgid (mnt_userns), prid, + error =xfs_qm_vop_dqalloc (dp,mapped_fsuid (mnt_userns, fs_userns), +mapped_fsgid (mnt_userns, fs_userns), prid,XFS_QMOPT_QUOTALL |XFS_QMOPT_INHERIT , &udqp, &gdqp, &pdqp);if (error)
include/linux/fs.h +4 -4
@@ -1664,7 +1664,7 @@ static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,static inline void inode_fsuid_set (struct inode *inode,struct user_namespace *mnt_userns) { - inode->i_uid =mapped_fsuid (mnt_userns); + inode->i_uid =mapped_fsuid (mnt_userns, &init_user_ns); }/**
@@ -1678,7 +1678,7 @@ static inline void inode_fsuid_set(struct inode *inode,static inline void inode_fsgid_set (struct inode *inode,struct user_namespace *mnt_userns) { - inode->i_gid =mapped_fsgid (mnt_userns); + inode->i_gid =mapped_fsgid (mnt_userns, &init_user_ns); }/**
@@ -1699,10 +1699,10 @@ static inline bool fsuidgid_has_mapping(struct super_block *sb,kuid_t kuid;kgid_t kgid; - kuid =mapped_fsuid (mnt_userns); + kuid =mapped_fsuid (mnt_userns, &init_user_ns);if (!uid_valid (kuid))return false; - kgid =mapped_fsgid (mnt_userns); + kgid =mapped_fsgid (mnt_userns, &init_user_ns);if (!gid_valid (kgid))return false; return kuid_has_mapping(fs_userns, kuid) &&
include/linux/mnt_mapping.h +8 -4
@@ -196,6 +196,7 @@ static inline kgid_t mapped_kgid_user(struct user_namespace *mnt_userns,/** * mapped_fsuid - return caller's fsuid mapped up into a mnt_userns * @mnt_userns: the mount's idmapping +* @fs_userns: the filesystem's idmapping * * Use this helper to initialize a new vfs or filesystem object based on * the caller's fsuid. A common example is initializing the i_uid field of
@@ -205,14 +206,16 @@ static inline kgid_t mapped_kgid_user(struct user_namespace *mnt_userns,* * Return: the caller's current fsuid mapped up according to @mnt_userns. */ -static inline kuid_t mapped_fsuid (struct user_namespace *mnt_userns) +static inline kuid_t mapped_fsuid (struct user_namespace *mnt_userns, +struct user_namespace *fs_userns) { -return mapped_kuid_user (mnt_userns, &init_user_ns,current_fsuid ()); +return mapped_kuid_user (mnt_userns, fs_userns,current_fsuid ()); }/** * mapped_fsgid - return caller's fsgid mapped up into a mnt_userns * @mnt_userns: the mount's idmapping +* @fs_userns: the filesystem's idmapping * * Use this helper to initialize a new vfs or filesystem object based on * the caller's fsgid. A common example is initializing the i_gid field of
@@ -222,9 +225,10 @@ static inline kuid_t mapped_fsuid(struct user_namespace *mnt_userns)* * Return: the caller's current fsgid mapped up according to @mnt_userns. */ -static inline kgid_t mapped_fsgid (struct user_namespace *mnt_userns) +static inline kgid_t mapped_fsgid (struct user_namespace *mnt_userns, +struct user_namespace *fs_userns) { -return mapped_kgid_user (mnt_userns, &init_user_ns,current_fsgid ()); +return mapped_kgid_user (mnt_userns, fs_userns,current_fsgid ()); }#endif /* _LINUX_MNT_MAPPING_H */
Signature · 1 line
2.30.2
Re: [PATCH 08/10] fs: port higher-level mapping helpers
Quoted 38 lines · Christian Brauner
From: Christian Brauner <[email protected]>
Enable the mapped_fs{g,u}id() helpers to support filesystems mounted with an idmapping. Apart from core mapping helpers that use mapped_fs{g,u}id() to initialize struct inode's i_{g,u}id fields xfs is the only place that uses these low-level helpers directly.
The patch only extends the helpers to be able to take the filesystem idmapping into account. Since we don't actually yet pass the filesystem's idmapping in no functional changes happen. This will happen in a final patch.
Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>--- fs/xfs/xfs_inode.c | 10 ++++++---- fs/xfs/xfs_symlink.c | 5 +++-- include/linux/fs.h | 8 ++++---- include/linux/mnt_mapping.h | 12 ++++++++---- 4 files changed, 21 insertions(+), 14 deletions(-)fs/xfs/xfs_inode.c +1
-0@@ -977,6 +977,7 @@ xfs_create(struct xfs_trans_res *tres;uint resblks;xfs_ino_t ino; +struct user_namespace *fs_userns = &init_user_ns;trace_xfs_create (dp, name);
fs/xfs/xfs_inode.c xfs_create( · line 993
Quoted 7 lines
fs/xfs/xfs_inode.c +2
-2@@ -988,8 +989,8 @@ xfs_create(/* * Make sure that we have allocated dquot(s) on disk. */ - error = xfs_qm_vop_dqalloc(dp, mapped_fsuid(mnt_userns), - mapped_fsgid(mnt_userns), prid, + error = xfs_qm_vop_dqalloc(dp, mapped_fsuid(mnt_userns, fs_userns), + mapped_fsgid(mnt_userns, fs_userns), prid,
I am confused.
Do we intend to enable idmapped xfs sb?
If the answer is yes, then feel free to add:
I did a quick review pass of all the patches.
The ones I did not reply to I felt I needed to take a close look so will continue with the review later.
Thanks,
Amir.
Re: [PATCH 08/10] fs: port higher-level mapping helpers
Quoted 38 lines · Amir Goldstein
Quoted 37 lines · Christian Brauner
From: Christian Brauner <[email protected]>
Enable the mapped_fs{g,u}id() helpers to support filesystems mounted with an idmapping. Apart from core mapping helpers that use mapped_fs{g,u}id() to initialize struct inode's i_{g,u}id fields xfs is the only place that uses these low-level helpers directly.
The patch only extends the helpers to be able to take the filesystem idmapping into account. Since we don't actually yet pass the filesystem's idmapping in no functional changes happen. This will happen in a final patch.
Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>--- fs/xfs/xfs_inode.c | 10 ++++++---- fs/xfs/xfs_symlink.c | 5 +++-- include/linux/fs.h | 8 ++++---- include/linux/mnt_mapping.h | 12 ++++++++---- 4 files changed, 21 insertions(+), 14 deletions(-)fs/xfs/xfs_inode.c +1
-0@@ -977,6 +977,7 @@ xfs_create(struct xfs_trans_res *tres;uint resblks;xfs_ino_t ino; +struct user_namespace *fs_userns = &init_user_ns;trace_xfs_create (dp, name);
Quoted 11 lines
fs/xfs/xfs_inode.c xfs_create( · line 993
Quoted 8 lines
fs/xfs/xfs_inode.c +2
-2@@ -988,8 +989,8 @@ xfs_create(/* * Make sure that we have allocated dquot(s) on disk. */ - error = xfs_qm_vop_dqalloc(dp, mapped_fsuid(mnt_userns), - mapped_fsgid(mnt_userns), prid, + error = xfs_qm_vop_dqalloc(dp, mapped_fsuid(mnt_userns, fs_userns), + mapped_fsgid(mnt_userns, fs_userns), prid,I am confused.
Do we intend to enable idmapped xfs sb?
No, I don't think we need to given that we have idmapped mount support. I'm happy to just continue passing down the initial idmapping.
Quoted 7 lines
If the answer is yes, then feel free to add:
Reviewed-by Amir Goldstein <[email protected]>I did a quick review pass of all the patches.
The ones I did not reply to I felt I needed to take a close look so will continue with the review later.
Thank you, appreciate it!
Christian
Christian Brauner
maintainerPATCH 9/10 fs: add i_user_ns() helper
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
Since we'll be passing the filesystem's idmapping in even more places in the following patches and we do already dereference struct inode to get to the filesystem's idmapping multiple times add a tiny helper.
--- include/linux/fs.h | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-)
include/linux/fs.h +9 -4
@@ -1600,6 +1600,11 @@ struct super_block {struct list_head s_inodes_wb;/* writeback inodes */ } __randomize_layout; +static inline struct user_namespace *i_user_ns (const struct inode *inode) +{ +return inode->i_sb->s_user_ns; +} +/* Helper functions so that in most cases filesystems will * not need to deal directly with kuid_t and kgid_t and can * instead deal with the raw numeric values that are stored
@@ -1607,22 +1612,22 @@ struct super_block {*/ static inline uid_t i_uid_read (const struct inode *inode) { -return from_kuid (inode->i_sb->s_user_ns, inode->i_uid); +return from_kuid (i_user_ns (inode), inode->i_uid); }static inline gid_t i_gid_read (const struct inode *inode) { -return from_kgid (inode->i_sb->s_user_ns, inode->i_gid); +return from_kgid (i_user_ns (inode), inode->i_gid); }static inline void i_uid_write (struct inode *inode,uid_t uid) { - inode->i_uid =make_kuid (inode->i_sb->s_user_ns, uid); + inode->i_uid =make_kuid (i_user_ns (inode), uid); }static inline void i_gid_write (struct inode *inode,gid_t gid) { - inode->i_gid =make_kgid (inode->i_sb->s_user_ns, gid); + inode->i_gid =make_kgid (i_user_ns (inode), gid); }/**
Signature · 1 line
2.30.2
Re: [PATCH 09/10] fs: add i_user_ns() helper
Quoted 7 lines · Christian Brauner
From: Christian Brauner <[email protected]>
Since we'll be passing the filesystem's idmapping in even more places in the following patches and we do already dereference struct inode to get to the filesystem's idmapping multiple times add a tiny helper.
Quoted 6 lines
Cc Seth Forshee <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Al Viro <[email protected]>Signed-off-by Christian Brauner <[email protected]>---
Quoted 49 lines
include/linux/fs.h | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-)include/linux/fs.h +9
-5@@ -1600,6 +1600,11 @@ struct super_block {struct list_head s_inodes_wb;/* writeback inodes */ } __randomize_layout; +static inline struct user_namespace *i_user_ns (const struct inode *inode) +{ +return inode->i_sb->s_user_ns; +} +/* Helper functions so that in most cases filesystems will * not need to deal directly with kuid_t and kgid_t and can * instead deal with the raw numeric values that are stored @@ -1607,22 +1612,22 @@ struct super_block {*/ static inline uid_t i_uid_read (const struct inode *inode) { -return from_kuid (inode->i_sb->s_user_ns, inode->i_uid); +return from_kuid (i_user_ns (inode), inode->i_uid); }static inline gid_t i_gid_read (const struct inode *inode) { -return from_kgid (inode->i_sb->s_user_ns, inode->i_gid); +return from_kgid (i_user_ns (inode), inode->i_gid); }static inline void i_uid_write (struct inode *inode,uid_t uid) { - inode->i_uid =make_kuid (inode->i_sb->s_user_ns, uid); + inode->i_uid =make_kuid (i_user_ns (inode), uid); }static inline void i_gid_write (struct inode *inode,gid_t gid) { - inode->i_gid =make_kgid (inode->i_sb->s_user_ns, gid); + inode->i_gid =make_kgid (i_user_ns (inode), gid); }/** -- 2.30.2
Christian Brauner
maintainerPATCH 10/10 fs: support mapped mounts of mapped filesystems
In reply to Christian BraunerFrom: Christian Brauner <[email protected]>
In previous patches we added new and modified existing helpers to handle idmapped mounts of filesystems mounted with an idmapping. In this final patch we convert all relevant places in the vfs to actually pass the filesystem's idmapping into these helpers.
With this the vfs is in shape to handle idmapped mounts of filesystems mounted with an idmapping. Note that this is just the generic infrastructure. Actually adding support for idmapped mounts to a filesystem mountable with an idmapping is follow-up work.
In this patch we extend the definition of an idmapped mount from a mount that that has the initial idmapping attached to it to a mount that has an idmapping attached to it which is not the same as the idmapping the filesystem was mounted with.
As before we do not allow the initial idmapping to be attached to a mount. In addition this patch prevents that the idmapping the filesystem was mounted with can be attached to a mount created based on this filesystem.
This has multiple reasons and advantages. First, attaching the initial idmapping or the filesystem's idmapping doesn't make much sense as in both cases the values of the i_{g,u}id and other places where k{g,u}ids are used do not change. Second, a user that really wants to do this for whatever reason can just create a separate dedicated identical idmapping to attach to the mount. Third, we can continue to use the initial idmapping as an indicator that a mount is not idmapped allowing us to continue to keep passing the initial idmapping into the mapping helpers to tell them that something isn't an idmapped mount even if the filesystem is mounted with an idmapping.
Apart from the generic places in the vfs the only other place we need to change is xfs since it makes use of mapped_fs{g,u}id() directly. Technically, this isn't even required since xfs doesn't support being mounted with an idmapping but it is cleaner to convert it properly.
--- fs/namespace.c | 36 ++++++++++++++++++++++++------------ fs/open.c | 7 ++++--- fs/posix_acl.c | 8 ++++---- fs/xfs/xfs_inode.c | 4 ++-- fs/xfs/xfs_symlink.c | 2 +- include/linux/fs.h | 17 +++++++++-------- security/commoncap.c | 9 ++++----- 7 files changed, 48 insertions(+), 35 deletions(-)
fs/namespace.c +24 -12
@@ -561,7 +561,7 @@ static void free_vfsmnt(struct mount *mnt)struct user_namespace *mnt_userns; mnt_userns =mnt_user_ns (&mnt->mnt); -if (mnt_userns != &init_user_ns) +if (!initial_mapping (mnt_userns))put_user_ns (mnt_userns);kfree_const (mnt->mnt_devname);#ifdef CONFIG_SMP
@@ -965,6 +965,7 @@ static struct mount *skip_mnt_tree(struct mount *p)struct vfsmount *vfs_create_mount (struct fs_context *fc) {struct mount *mnt; +struct user_namespace *fs_userns;if (!fc->root)return ERR_PTR (-EINVAL );
@@ -982,6 +983,10 @@ struct vfsmount *vfs_create_mount(struct fs_context *fc) mnt->mnt_mountpoint = mnt->mnt.mnt_root; mnt->mnt_parent = mnt; + fs_userns = mnt->mnt.mnt_sb->s_user_ns; +if (!initial_mapping (fs_userns)) + mnt->mnt.mnt_userns =get_user_ns (fs_userns); +lock_mount_hash ();list_add_tail (&mnt->mnt_instance, &mnt->mnt.mnt_sb->s_mounts);unlock_mount_hash ();
@@ -1072,7 +1077,7 @@ static struct mount *clone_mnt(struct mount *old, struct dentry *root,atomic_inc (&sb->s_active); mnt->mnt.mnt_userns =mnt_user_ns (&old->mnt); -if (mnt->mnt.mnt_userns != &init_user_ns) +if (!initial_mapping (mnt->mnt.mnt_userns)) mnt->mnt.mnt_userns =get_user_ns (mnt->mnt.mnt_userns); mnt->mnt.mnt_sb = sb; mnt->mnt.mnt_root =dget (root);
@@ -3927,10 +3932,18 @@ static unsigned int recalc_flags(struct mount_kattr *kattr, struct mount *mnt)static int can_idmap_mount (const struct mount_kattr *kattr,struct mount *mnt) {struct vfsmount *m = &mnt->mnt; +struct user_namespace *fs_userns = m->mnt_sb->s_user_ns;if (!kattr->mnt_userns)return 0 ; +/* +* Creating an idmapped mount with the filesystem wide idmapping +* doesn't make sense so block that. We don't allow mushy semantics. +*/ +if (kattr->mnt_userns == fs_userns) +return -EINVAL ; +/* * Once a mount has been idmapped we don't allow it to change its * mapping. It makes things simpler and callers can just create
@@ -3943,12 +3956,8 @@ static int can_idmap_mount(const struct mount_kattr *kattr, struct mount *mnt)if (!(m->mnt_sb->s_type->fs_flags &FS_ALLOW_IDMAP ))return -EINVAL ; -/* Don't yet support filesystem mountable in user namespaces. */ -if (m->mnt_sb->s_user_ns != &init_user_ns) -return -EINVAL ; -/* We're not controlling the superblock. */ -if (!capable (CAP_SYS_ADMIN )) +if (!ns_capable (fs_userns,CAP_SYS_ADMIN ))return -EPERM ;/* Mount has already been visible in the filesystem hierarchy. */
@@ -4133,13 +4142,16 @@ static int build_mount_idmapped(const struct mount_attr *attr, size_t usize, }/* -* The init_user_ns is used to indicate that a vfsmount is not idmapped. -* This is simpler than just having to treat NULL as unmapped. Users -* wanting to idmap a mount to init_user_ns can just use a namespace -* with an identity mapping. +* The initial idmapping cannot be used to create an idmapped +* mount. Attaching the initial idmapping doesn't make much sense +* as it is an identity mapping. A user can just create a dedicated +* identity mapping to achieve the same result. We also use the +* initial idmapping as an indicator of a mount that is not +* idmapped. It can simply be passed into helpers that are aware of +* idmapped mounts as a convenient shortcut. */ mnt_userns = container_of(ns, struct user_namespace, ns); -if (mnt_userns == &init_user_ns) { +if (initial_mapping (mnt_userns)) { err = -EPERM ; goto out_fput; }
fs/open.c +4 -3
@@ -640,7 +640,7 @@ SYSCALL_DEFINE2(chmod, const char __user *, filename, umode_t, mode)int chown_common (const struct path *path,uid_t user,gid_t group) { -struct user_namespace *mnt_userns; +struct user_namespace *mnt_userns, *fs_userns;struct inode *inode = path->dentry->d_inode;struct inode *delegated_inode =NULL ;int error;
@@ -652,8 +652,9 @@ int chown_common(const struct path *path, uid_t user, gid_t group) gid =make_kgid (current_user_ns (), group); mnt_userns =mnt_user_ns (path->mnt); - uid =mapped_kuid_user (mnt_userns, &init_user_ns, uid); - gid =mapped_kgid_user (mnt_userns, &init_user_ns, gid); + fs_userns =i_user_ns (inode); + uid =mapped_kuid_user (mnt_userns, fs_userns, uid); + gid =mapped_kgid_user (mnt_userns, fs_userns, gid); retry_deleg: newattrs.ia_valid =ATTR_CTIME ;
fs/posix_acl.c +4 -4
@@ -375,8 +375,8 @@ posix_acl_permission(struct user_namespace *mnt_userns, struct inode *inode,break ;case ACL_USER : uid =mapped_kuid_fs (mnt_userns, - &init_user_ns, - pa->e_uid); +i_user_ns (inode), + pa->e_uid);if (uid_eq (uid,current_fsuid ())) goto mask;break ;
@@ -390,8 +390,8 @@ posix_acl_permission(struct user_namespace *mnt_userns, struct inode *inode,break ;case ACL_GROUP : gid =mapped_kgid_fs (mnt_userns, - &init_user_ns, - pa->e_gid); +i_user_ns (inode), + pa->e_gid);if (in_group_p (gid)) { found =1 ;if ((pa->e_perm & want) == want)
fs/xfs/xfs_inode.c +2 -2
@@ -977,7 +977,7 @@ xfs_create(struct xfs_trans_res *tres;uint resblks;xfs_ino_t ino; -struct user_namespace *fs_userns = &init_user_ns; +struct user_namespace *fs_userns =i_user_ns (VFS_I (dp));trace_xfs_create (dp, name);
@@ -1134,7 +1134,7 @@ xfs_create_tmpfile(struct xfs_trans_res *tres;uint resblks;xfs_ino_t ino; -struct user_namespace *fs_userns = &init_user_ns; +struct user_namespace *fs_userns =i_user_ns (VFS_I (dp));if (xfs_is_shutdown (mp))return -EIO ;
fs/xfs/xfs_symlink.c +1 -1
@@ -163,7 +163,7 @@ xfs_symlink(struct xfs_dquot *pdqp =NULL ;uint resblks;xfs_ino_t ino; -struct user_namespace *fs_userns = &init_user_ns; +struct user_namespace *fs_userns =i_user_ns (VFS_I (dp)); *ipp =NULL ;
include/linux/fs.h +9 -8
@@ -1641,7 +1641,7 @@ static inline void i_gid_write(struct inode *inode, gid_t gid)static inline kuid_t i_uid_into_mnt (struct user_namespace *mnt_userns,const struct inode *inode) { -return mapped_kuid_fs (mnt_userns, &init_user_ns, inode->i_uid); +return mapped_kuid_fs (mnt_userns,i_user_ns (inode), inode->i_uid); }/**
@@ -1655,7 +1655,7 @@ static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,static inline kgid_t i_gid_into_mnt (struct user_namespace *mnt_userns,const struct inode *inode) { -return mapped_kgid_fs (mnt_userns, &init_user_ns, inode->i_gid); +return mapped_kgid_fs (mnt_userns,i_user_ns (inode), inode->i_gid); }/**
@@ -1669,7 +1669,7 @@ static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,static inline void inode_fsuid_set (struct inode *inode,struct user_namespace *mnt_userns) { - inode->i_uid =mapped_fsuid (mnt_userns, &init_user_ns); + inode->i_uid =mapped_fsuid (mnt_userns,i_user_ns (inode)); }/**
@@ -1683,7 +1683,7 @@ static inline void inode_fsuid_set(struct inode *inode,static inline void inode_fsgid_set (struct inode *inode,struct user_namespace *mnt_userns) { - inode->i_gid =mapped_fsgid (mnt_userns, &init_user_ns); + inode->i_gid =mapped_fsgid (mnt_userns,i_user_ns (inode)); }/**
@@ -1704,10 +1704,10 @@ static inline bool fsuidgid_has_mapping(struct super_block *sb,kuid_t kuid;kgid_t kgid; - kuid =mapped_fsuid (mnt_userns, &init_user_ns); + kuid =mapped_fsuid (mnt_userns, fs_userns);if (!uid_valid (kuid))return false; - kgid =mapped_fsgid (mnt_userns, &init_user_ns); + kgid =mapped_fsgid (mnt_userns, fs_userns);if (!gid_valid (kgid))return false; return kuid_has_mapping(fs_userns, kuid) &&
@@ -2654,13 +2654,14 @@ static inline struct user_namespace *file_mnt_user_ns(struct file *file)* is_mapped_mnt - check whether a mount is mapped * @mnt: the mount to check * -* If @mnt has an idmapping attached to it @mnt is mapped. +* If @mnt has an idmapping attached different from the +* filesystem's idmapping then @mnt is mapped. * * Return: true if mount is mapped, false if not. */ static inline bool is_mapped_mnt (const struct vfsmount *mnt) { -return mnt_user_ns (mnt) != &init_user_ns; +return mnt_user_ns (mnt) != mnt->mnt_sb->s_user_ns; }extern long vfs_truncate (const struct path *,loff_t );
security/commoncap.c +4 -5
@@ -418,7 +418,7 @@ int cap_inode_getsecurity(struct user_namespace *mnt_userns, kroot =make_kuid (fs_ns, root);/* If this is an idmapped mount shift the kuid. */ - kroot =mapped_kuid_fs (mnt_userns, &init_user_ns, kroot); + kroot =mapped_kuid_fs (mnt_userns, fs_ns, kroot);/* If the root kuid maps to a valid uid in current ns, then return * this as a nscap. */
@@ -555,13 +555,12 @@ int cap_convert_nscap(struct user_namespace *mnt_userns, struct dentry *dentry,return -EINVAL ;if (!capable_wrt_inode_uidgid (mnt_userns, inode,CAP_SETFCAP ))return -EPERM ; -if (size ==XATTR_CAPS_SZ_2 && (mnt_userns == &init_user_ns)) +if (size ==XATTR_CAPS_SZ_2 && (mnt_userns == fs_ns))if (ns_capable (inode->i_sb->s_user_ns,CAP_SETFCAP ))/* user is privileged, just write the v2 */ return size; - rootid =rootid_from_xattr (*ivalue, size, task_ns, mnt_userns, - &init_user_ns); + rootid =rootid_from_xattr (*ivalue, size, task_ns, mnt_userns, fs_ns);if (!uid_valid (rootid))return -EINVAL ;
@@ -702,7 +701,7 @@ int get_vfs_caps_from_disk(struct user_namespace *mnt_userns,/* Limit the caps to the mounter of the filesystem * or the more limited uid specified in the xattr. */ - rootkuid =mapped_kuid_fs (mnt_userns, &init_user_ns, rootkuid); + rootkuid =mapped_kuid_fs (mnt_userns, fs_ns, rootkuid);if (!rootid_owns_currentns (rootkuid))return -ENODATA ;
Signature · 1 line
2.30.2
Quoted 6 lines · Christian Brauner
From: Christian Brauner <[email protected]>
Hey,
This series extend the mapping infrastructure in order to support mapped mounts of mapped filesystems in the future.
Friendly ping. Would be good to get a review for this.
Christian
Amir Goldstein
maintainerQuoted 79 lines · Christian Brauner
From: Christian Brauner <[email protected]>
Hey,
This series extend the mapping infrastructure in order to support mapped mounts of mapped filesystems in the future.
Currently we only support mapped mounts of filesystems mounted without an idmapping. This was a consicous decision mentioned in multiple places. For example, see [1].
In our mapping documentation in [3] we explained in detail that it is perfectly fine to extend support for mapped mounts to filesystem's mounted with an idmapping should the need arise. The need has been there for some time now (cf. [2]).
Before we can port any such filesystem we need to first extend the mapping helpers to account for the filesystem's idmapping in the remapping helpers. This again, is explained at length in our documentation at [3].
Currently, the low-level mapping helpers implement the remapping algorithms described in [3] in a simplified manner. Because we could rely on the fact that all filesystems supporting mapped mounts are mounted without an idmapping the translation step from or into the filesystem idmapping could be skipped.
In order to support mapped mounts of filesystem's mountable with an idmapping the translation step we were able to skip before cannot be skipped anymore. A filesystem mounted with an idmapping is very likely to not use an identity mapping and will instead use a non-identity mapping. So the translation step from or into the filesystem's idmapping in the remapping algorithm cannot be skipped for such filesystems. More details with examples can be found in [3].
This series adds a few new as well as prepares and tweaks some already existing low-level mapping helpers to perform the full translation algorithm explained in [3]. The low-level helpers can be written in a way that they only perform the additional translation step when the filesystem is indeed mounted with an idmapping.
Since we don't yet support such a filesystem yet a kernel was compiled carrying a trivial patch making ext4 mountable with an idmapping:
# We're located on the host with the initial idmapping.
ubuntu@f2-vm:~$ cat /proc/self/uid_map 0 0 4294967295# Mount an ext4 filesystem with the initial idmapping.
ubuntu@f2-vm:~$ sudo mount -t ext4 /dev/loop0 /mnt# The filesystem contains two files. One owned by id 0 and another one owned by
# id 1000 in the initial idmapping. ubuntu@f2-vm:~$ ls -al /mnt/ total 8 drwxrwxrwx 2 root root 4096 Nov 22 17:04 . drwxr-xr-x 24 root root 4096 Nov 20 11:24 ..(quoted diff) +0
-2-rw-r--r-- 1 root root 0 Nov 22 17:04 file_init_mapping_0 -rw-r--r-- 1 ubuntu ubuntu 0 Nov 22 17:04 file_init_mapping_1000# Umount it again so we we can mount it in another namespace later.
ubuntu@f2-vm:~$ sudo umount /mnt# Use the lxc-usernsexec binary to run a shell in a user and mount namespace
# with an idmapping of 0:10000:100000000. # # This idmapping will have the effect that files which are owned by i_{g,u}id # 10000 and files that are owned by i_{g,u}id 11000 will be owned by {g,u}id # 0 and {g,u}id 1000 with that namespace respectively. ubuntu@f2-vm:~$ sudo lxc-usernsexec -m b:0:10000:100000000 -- bash# Verify that we're really running with the expected idmapping.
root@f2-vm:/home/ubuntu# cat /proc/self/uid_map 0 10000 100000000# Mount the ext4 filesystem in the user and mountns with the idmapping # 0:10000:100000000.
#
# Note, that this requires a test kernel that makes ext4 mountable in a
Quoted 16 lines
# non-initial userns. The patch is simply:
#
# diff --git a/fs/ext4/super.c b/fs/ext4/super.c
# index 4e33b5eca694..0221e8211e5b 100644
# --- a/fs/ext4/super.c
# +++ b/fs/ext4/super.c
# @@ -6584,7 +6584,7 @@ static struct file_system_type ext4_fs_type = {
# .name = "ext4",
# .mount = ext4_mount,
# .kill_sb = kill_block_super,
# - .fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
# + .fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP | FS_USERNS_MOUNT,
# };
# MODULE_ALIAS_FS("ext4");
root@f2-vm:/home/ubuntu# mount -t ext4 /dev/loop0 /mntHi Christian,
I have a question not directly related to the patches, but to the test hack above.
I may be wrong, but it looks like an idmapped sb would be desired for some use cases(?).
My question is - could we use fsconfig() to allow CAP_SYS_ADMIN to attach a newly mounted sb (e.g. ext4) to user ns without allowing mount of ext4 from within userns? Wouldn't that add usability to some users without adding any new risks over the risks already subjected with idmapped mounts?
Thanks,
Amir.