Skip to contentSkip to conversation

4.2: Can't mount sysfs in a mount ns & user ns

PATCH In review Lubomir Rintel 4 participants 7 messages Aug 13, 2015 – Aug 17, 2015 VFS

lore
Share & tools
lore
#0 lore raw

Hi,

4.0.6-300.fc22.x86_64:
[lkundrak@fedora22-1 ~]$ unshare -r --mount --net
[root@fedora22-1 ~]# mount --make-slave /sys
[root@fedora22-1 ~]# mount -t sysfs sysfs /sys
[root@fedora22-1 ~]#

4.2.0-0.rc6.git0.1.fc24.x86_64:
[lkundrak@fedora23-1 ~]$ unshare -r --mount --net
[root@fedora23-1 ~]# mount --make-slave /sys
[root@fedora23-1 ~]# mount -t sysfs sysfs /sys
mount: permission denied
[root@fedora23-1 ~]#

we use this in NetworkManager test suite, to ensure the devices we see via GUdev are the same as we see via rtnetlink.

I'm wondering if this is a bug or an intended change?

Thanks,
Lubo

#1 lore raw
Quoted 15 lines · Lubomir Rintel

Hi,

4.0.6-300.fc22.x86_64:
[lkundrak@fedora22-1 ~]$ unshare -r --mount --net
[root@fedora22-1 ~]# mount --make-slave /sys
[root@fedora22-1 ~]# mount -t sysfs sysfs /sys
[root@fedora22-1 ~]#

4.2.0-0.rc6.git0.1.fc24.x86_64:
[lkundrak@fedora23-1 ~]$ unshare -r --mount --net
[root@fedora23-1 ~]# mount --make-slave /sys
[root@fedora23-1 ~]# mount -t sysfs sysfs /sys
mount: permission denied
[root@fedora23-1 ~]#

Quoted 4 lines

we use this in NetworkManager test suite, to ensure the devices we see via GUdev are the same as we see via rtnetlink.

I'm wondering if this is a bug or an intended change?

There was an intentional tightening up of the permissions required to mount sysfs to prevent people in jails from gaining access to things they would not ordinarily have access to. The change was not expected to affect anyones legitimate use case.

What are the mount flags of the previous mount of sysfs? What is mounted on top of sysfs?

Or in short can I see /proc/self/mounts for the failing scenario?

Without a little more detail I can't see if there is a possible security violation in your code or if this is something I can fix.

Eric

#2 lore raw

Hello,

Quoted 29 lines · Eric W. Biederman
Quoted 15 lines · Lubomir Rintel

Hi,

4.0.6-300.fc22.x86_64:
[lkundrak@fedora22-1 ~]$ unshare -r --mount --net [root@fedora22-1 ~]# mount --make-slave /sys [root@fedora22-1 ~]# mount -t sysfs sysfs /sys [root@fedora22-1 ~]#

4.2.0-0.rc6.git0.1.fc24.x86_64:
[lkundrak@fedora23-1 ~]$ unshare -r --mount --net [root@fedora23-1 ~]# mount --make-slave /sys [root@fedora23-1 ~]# mount -t sysfs sysfs /sys
mount: permission denied
[root@fedora23-1 ~]#

Quoted 5 lines

we use this in NetworkManager test suite, to ensure the devices we see
via GUdev are the same as we see via rtnetlink.

I'm wondering if this is a bug or an intended change?

There was an intentional tightening up of the permissions required to mount sysfs to prevent people in jails from gaining access to things they would not ordinarily have access to. The change was not expected
to affect anyones legitimate use case.

Quoted 4 lines

What are the mount flags of the previous mount of sysfs? What is mounted on top of sysfs?

Or in short can I see /proc/self/mounts for the failing scenario?

Looks like this:

sysfs /sys sysfs rw,seclabel,nosuid,nodev,noexec,relatime 0 0
proc /proc proc rw,relatime 0 0
devtmpfs /dev devtmpfs rw,seclabel,nosuid,size=882904k,nr_inodes=220726,mode=755 0 0
securityfs /sys/kernel/security securityfs rw,nosuid,nodev,noexec,relatime 0 0
tmpfs /dev/shm tmpfs rw,seclabel,nosuid,nodev 0 0
devpts /dev/pts devpts rw,seclabel,nosuid,noexec,relatime,gid=5,mode=620,ptmxmode=000 0 0
tmpfs /run tmpfs rw,seclabel,nosuid,nodev,mode=755 0 0
tmpfs /sys/fs/cgroup tmpfs ro,seclabel,nosuid,nodev,noexec,mode=755 0 0
cgroup /sys/fs/cgroup/systemd cgroup rw,nosuid,nodev,noexec,relatime,xattr,release_agent=/usr/lib/systemd/systemd-cgroups-agent,name=systemd 0 0 pstore /sys/fs/pstore pstore rw,seclabel,nosuid,nodev,noexec,relatime 0 0
kdbusfs /sys/fs/kdbus kdbusfs rw,nosuid,nodev,noexec,relatime 0 0
cgroup /sys/fs/cgroup/hugetlb cgroup rw,nosuid,nodev,noexec,relatime,hugetlb 0 0
cgroup /sys/fs/cgroup/net_cls,net_prio cgroup rw,nosuid,nodev,noexec,relatime,net_cls,net_prio 0 0
cgroup /sys/fs/cgroup/devices cgroup rw,nosuid,nodev,noexec,relatime,devices 0 0
cgroup /sys/fs/cgroup/cpu,cpuacct cgroup rw,nosuid,nodev,noexec,relatime,cpu,cpuacct 0 0
cgroup /sys/fs/cgroup/perf_event cgroup rw,nosuid,nodev,noexec,relatime,perf_event 0 0
cgroup /sys/fs/cgroup/cpuset cgroup rw,nosuid,nodev,noexec,relatime,cpuset 0 0
cgroup /sys/fs/cgroup/memory cgroup rw,nosuid,nodev,noexec,relatime,memory 0 0
cgroup /sys/fs/cgroup/freezer cgroup rw,nosuid,nodev,noexec,relatime,freezer 0 0
cgroup /sys/fs/cgroup/blkio cgroup rw,nosuid,nodev,noexec,relatime,blkio 0 0
configfs /sys/kernel/config configfs rw,relatime 0 0
/dev/vda3 / btrfs rw,seclabel,relatime,space_cache,subvolid=5,subvol=/ 0 0
selinuxfs /sys/fs/selinux selinuxfs rw,relatime 0 0
systemd-1 /proc/sys/fs/binfmt_misc autofs rw,relatime,fd=36,pgrp=1,timeout=0,minproto=5,maxproto=5,direct 0 0
tmpfs /tmp tmpfs rw,seclabel 0 0
mqueue /dev/mqueue mqueue rw,seclabel,relatime 0 0
debugfs /sys/kernel/debug debugfs rw,seclabel,relatime 0 0
hugetlbfs /dev/hugepages hugetlbfs rw,seclabel,relatime 0 0
/dev/vda1 /boot ext4 rw,seclabel,relatime,data=ordered 0 0
tmpfs /run/user/42 tmpfs rw,seclabel,nosuid,nodev,relatime,size=178648k,mode=700,uid=42,gid=42 0 0
gvfsd-fuse /run/user/42/gvfs fuse.gvfsd-fuse rw,nosuid,nodev,relatime,user_id=42,group_id=42 0 0
fusectl /sys/fs/fuse/connections fusectl rw,relatime 0 0
tmpfs /run/user/8086 tmpfs rw,seclabel,nosuid,nodev,relatime,size=178648k,mode=700,uid=8086,gid=8086 0 0
gvfsd-fuse /run/user/8086/gvfs fuse.gvfsd-fuse rw,nosuid,nodev,relatime,user_id=8086,group_id=8086 0 0

Quoted 5 lines

Without a little more detail I can't see if there is a possible security
violation in your code or if this is something I can fix.

Eric

Thanks for the response

Lubo

#3 lore raw
Quoted 39 lines · Lubomir Rintel

Hello,

Quoted 29 lines · Eric W. Biederman
Quoted 15 lines · Lubomir Rintel

Hi,

4.0.6-300.fc22.x86_64:
[lkundrak@fedora22-1 ~]$ unshare -r --mount --net [root@fedora22-1 ~]# mount --make-slave /sys [root@fedora22-1 ~]# mount -t sysfs sysfs /sys [root@fedora22-1 ~]#

4.2.0-0.rc6.git0.1.fc24.x86_64:
[lkundrak@fedora23-1 ~]$ unshare -r --mount --net [root@fedora23-1 ~]# mount --make-slave /sys [root@fedora23-1 ~]# mount -t sysfs sysfs /sys
mount: permission denied
[root@fedora23-1 ~]#

Quoted 5 lines

we use this in NetworkManager test suite, to ensure the devices we see
via GUdev are the same as we see via rtnetlink.

I'm wondering if this is a bug or an intended change?

There was an intentional tightening up of the permissions required to mount sysfs to prevent people in jails from gaining access to things they would not ordinarily have access to. The change was not expected
to affect anyones legitimate use case.

Quoted 4 lines

What are the mount flags of the previous mount of sysfs? What is mounted on top of sysfs?

Or in short can I see /proc/self/mounts for the failing scenario?

Looks like this:

Quoted 11 lines

sysfs /sys sysfs rw,seclabel,nosuid,nodev,noexec,relatime 0 0
proc /proc proc rw,relatime 0 0
devtmpfs /dev devtmpfs rw,seclabel,nosuid,size=882904k,nr_inodes=220726,mode=755 0 0
securityfs /sys/kernel/security securityfs rw,nosuid,nodev,noexec,relatime 0 0
tmpfs /dev/shm tmpfs rw,seclabel,nosuid,nodev 0 0
devpts /dev/pts devpts rw,seclabel,nosuid,noexec,relatime,gid=5,mode=620,ptmxmode=000 0 0
tmpfs /run tmpfs rw,seclabel,nosuid,nodev,mode=755 0 0
tmpfs /sys/fs/cgroup tmpfs ro,seclabel,nosuid,nodev,noexec,mode=755 0 0
cgroup /sys/fs/cgroup/systemd cgroup rw,nosuid,nodev,noexec,relatime,xattr,release_agent=/usr/lib/systemd/systemd-cgroups-agent,name=systemd 0 0 pstore /sys/fs/pstore pstore rw,seclabel,nosuid,nodev,noexec,relatime 0 0
kdbusfs /sys/fs/kdbus kdbusfs rw,nosuid,nodev,noexec,relatime 0 0

^^^^^^^^^^^^^ This directory is probably not created with sysfs_create_mount_point
So I suspect this is your culprit.

Quoted 24 lines

cgroup /sys/fs/cgroup/hugetlb cgroup rw,nosuid,nodev,noexec,relatime,hugetlb 0 0
cgroup /sys/fs/cgroup/net_cls,net_prio cgroup rw,nosuid,nodev,noexec,relatime,net_cls,net_prio 0 0 cgroup /sys/fs/cgroup/devices cgroup rw,nosuid,nodev,noexec,relatime,devices 0 0
cgroup /sys/fs/cgroup/cpu,cpuacct cgroup rw,nosuid,nodev,noexec,relatime,cpu,cpuacct 0 0 cgroup /sys/fs/cgroup/perf_event cgroup rw,nosuid,nodev,noexec,relatime,perf_event 0 0 cgroup /sys/fs/cgroup/cpuset cgroup rw,nosuid,nodev,noexec,relatime,cpuset 0 0
cgroup /sys/fs/cgroup/memory cgroup rw,nosuid,nodev,noexec,relatime,memory 0 0
cgroup /sys/fs/cgroup/freezer cgroup rw,nosuid,nodev,noexec,relatime,freezer 0 0
cgroup /sys/fs/cgroup/blkio cgroup rw,nosuid,nodev,noexec,relatime,blkio 0 0
configfs /sys/kernel/config configfs rw,relatime 0 0
/dev/vda3 / btrfs rw,seclabel,relatime,space_cache,subvolid=5,subvol=/ 0 0
selinuxfs /sys/fs/selinux selinuxfs rw,relatime 0 0
systemd-1 /proc/sys/fs/binfmt_misc autofs rw,relatime,fd=36,pgrp=1,timeout=0,minproto=5,maxproto=5,direct 0 0 tmpfs /tmp tmpfs rw,seclabel 0 0
mqueue /dev/mqueue mqueue rw,seclabel,relatime 0 0
debugfs /sys/kernel/debug debugfs rw,seclabel,relatime 0 0
hugetlbfs /dev/hugepages hugetlbfs rw,seclabel,relatime 0 0
/dev/vda1 /boot ext4 rw,seclabel,relatime,data=ordered 0 0
tmpfs /run/user/42 tmpfs rw,seclabel,nosuid,nodev,relatime,size=178648k,mode=700,uid=42,gid=42 0 0 gvfsd-fuse /run/user/42/gvfs fuse.gvfsd-fuse rw,nosuid,nodev,relatime,user_id=42,group_id=42 0 0 fusectl /sys/fs/fuse/connections fusectl rw,relatime 0 0
tmpfs /run/user/8086 tmpfs rw,seclabel,nosuid,nodev,relatime,size=178648k,mode=700,uid=8086,gid=8086 0 0 gvfsd-fuse /run/user/8086/gvfs fuse.gvfsd-fuse rw,nosuid,nodev,relatime,user_id=8086,group_id=8086 0 0

Quoted 7 lines
Quoted 5 lines

Without a little more detail I can't see if there is a possible security
violation in your code or if this is something I can fix.

Eric

Thanks for the response

It looks like this is a kdbus thing. I don't see anything else that should be causing problems. Please try again with kdbus disabled and see what happens, and when it works please let the kdbus guys know that the need to use sysfs_create_mount_point.

Eric

#4 lore raw

PATCH 1/1 kdbus: create /sys/fs/kdbus with sysfs_create_mount_point()

Since 0cbee99269 user-namespace pull, if a kdbusfs is mounted on a location that's not created with sysfs_create_mount_point the user namespaces are not allowed to mount their sysfs instances.

---
Applies on top of char-misc/kdbus a36324913.

 ipc/kdbus/main.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)
ipc/kdbus/main.c +5 -8
index 1ad4dc8..c2117ea 100644
@@ -75,16 +75,13 @@
  *  '» struct kdbus_ep *ep (owned)
  */
 
-/* kdbus mount-point /sys/fs/kdbus */
-static struct kobject *kdbus_dir;
-
 static int __init kdbus_init(void)
 {
 	int ret;
 
-	kdbus_dir = kobject_create_and_add(KBUILD_MODNAME, fs_kobj);
-	if (!kdbus_dir)
-		return -ENOMEM;
+	ret = sysfs_create_mount_point(fs_kobj, KBUILD_MODNAME);
+	if (ret)
+		return ret;
 
 	ret = kdbus_fs_init();
 	if (ret < 0) {
@@ -96,14 +93,14 @@ static int __init kdbus_init(void)
 	return 0;
 
 exit_dir:
-	kobject_put(kdbus_dir);
+	sysfs_remove_mount_point(fs_kobj, KBUILD_MODNAME);
 	return ret;
 }
 
 static void __exit kdbus_exit(void)
 {
 	kdbus_fs_exit();
-	kobject_put(kdbus_dir);
+	sysfs_remove_mount_point(fs_kobj, KBUILD_MODNAME);
 	ida_destroy(&kdbus_node_ida);
 }
 
Signature · 1 line
2.4.3
#5 lore raw

Re: [PATCH] kdbus: create /sys/fs/kdbus with sysfs_create_mount_point()

Hi

Quoted 7 lines · Lubomir Rintel

Since 0cbee99269 user-namespace pull, if a kdbusfs is mounted on a location that's not created with sysfs_create_mount_point the user namespaces are not allowed to mount their sysfs instances.

---
Applies on top of char-misc/kdbus a36324913.

This does not apply on top of char-misc/kdbus. The tree lacks the necessary commits. But yeah, it's required for 4.2. I'll keep it starred until Greg back-merges 4.2.

Anyway, this is still: Reviewed-by: David Herrmann <[email protected]>

Thanks!
David

Quoted 47 lines
 ipc/kdbus/main.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)
ipc/kdbus/main.c +5 -9
index 1ad4dc8..c2117ea 100644
@@ -75,16 +75,13 @@
  *  '» struct kdbus_ep *ep (owned)
  */
 
-/* kdbus mount-point /sys/fs/kdbus */
-static struct kobject *kdbus_dir;
-
 static int __init kdbus_init(void)
 {
        int ret;
 
-       kdbus_dir = kobject_create_and_add(KBUILD_MODNAME, fs_kobj);
-       if (!kdbus_dir)
-               return -ENOMEM;
+       ret = sysfs_create_mount_point(fs_kobj, KBUILD_MODNAME);
+       if (ret)
+               return ret;
 
        ret = kdbus_fs_init();
        if (ret < 0) {
@@ -96,14 +93,14 @@ static int __init kdbus_init(void)
        return 0;
 
 exit_dir:
-       kobject_put(kdbus_dir);
+       sysfs_remove_mount_point(fs_kobj, KBUILD_MODNAME);
        return ret;
 }
 
 static void __exit kdbus_exit(void)
 {
        kdbus_fs_exit();
-       kobject_put(kdbus_dir);
+       sysfs_remove_mount_point(fs_kobj, KBUILD_MODNAME);
        ida_destroy(&kdbus_node_ida);
 }
 
--

2.4.3

#6 lore raw

Re: [PATCH] kdbus: create /sys/fs/kdbus with sysfs_create_mount_point()

Quoted 11 lines · David Herrmann

Hi

Quoted 7 lines · Lubomir Rintel

Since 0cbee99269 user-namespace pull, if a kdbusfs is mounted on a location that's not created with sysfs_create_mount_point the user namespaces are not allowed to mount their sysfs instances.

---
Applies on top of char-misc/kdbus a36324913.

Quoted 5 lines

This does not apply on top of char-misc/kdbus. The tree lacks the necessary commits. But yeah, it's required for 4.2. I'll keep it starred until Greg back-merges 4.2.

Anyway, this is still: Reviewed-by: David Herrmann <[email protected]>

Grumble. Keeping track of this is getting to be somewhat of a pain. Why isn't the kdbus-next branch already at 4.2?

So if one was carrying the contents of char-misc/kdbus as stand-alone patches on top of 4.2-rc7, should your Reviewed-by count as "yes, apply this patch in that instance?"

josh