Date:   Wed, 14 Jul 2021 19:12:08 +0300
From:   Pavel Tikhomirov <ptikhomirov@...tuozzo.com>
To:     Christian Brauner <christian.brauner@...ntu.com>
Cc:     linux-fsdevel@...r.kernel.org,
        "Eric W . Biederman" <ebiederm@...ssion.com>,
        Alexander Viro <viro@...iv.linux.org.uk>,
        Mattias Nissler <mnissler@...omium.org>,
        Aleksa Sarai <cyphar@...har.com>,
        Andrei Vagin <avagin@...il.com>, linux-api@...r.kernel.org,
        lkml <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v3 1/2] move_mount: allow to add a mount into an existing
 group



On 14.07.2021 17:43, Christian Brauner wrote:
> On Wed, Jul 14, 2021 at 04:27:53PM +0300, Pavel Tikhomirov wrote:
>> Previously a sharing group (shared and master ids pair) can be only
>> inherited when mount is created via bindmount. This patch adds an
>> ability to add an existing private mount into an existing sharing group.
>>
>> With this functionality one can first create the desired mount tree from
>> only private mounts (without the need to care about undesired mount
>> propagation or mount creation order implied by sharing group
>> dependencies), and next then setup any desired mount sharing between
>> those mounts in tree as needed.
>>
>> This allows CRIU to restore any set of mount namespaces, mount trees and
>> sharing group trees for a container.
>>
>> We have many issues with restoring mounts in CRIU related to sharing
>> groups and propagation:
>> - reverse sharing groups vs mount tree order requires complex mounts
>>    reordering which mostly implies also using some temporary mounts
>> (please see https://lkml.org/lkml/2021/3/23/569 for more info)
>>
>> - mount() syscall creates tons of mounts due to propagation
>> - mount re-parenting due to propagation
>> - "Mount Trap" due to propagation
>> - "Non Uniform" propagation, meaning that with different tricks with
>>    mount order and temporary children-"lock" mounts one can create mount
>>    trees which can't be restored without those tricks
>> (see https://www.linuxplumbersconf.org/event/7/contributions/640/)
>>
>> With this new functionality we can resolve all the problems with
>> propagation at once.
>>
>> Link: https://lore.kernel.org/r/20210714132754.94633-1-ptikhomirov@virtuozzo.com
>> Cc: Eric W. Biederman <ebiederm@...ssion.com>
>> Cc: Alexander Viro <viro@...iv.linux.org.uk>
>> Cc: Christian Brauner <christian.brauner@...ntu.com>
>> Cc: Mattias Nissler <mnissler@...omium.org>
>> Cc: Aleksa Sarai <cyphar@...har.com>
>> Cc: Andrei Vagin <avagin@...il.com>
>> Cc: linux-fsdevel@...r.kernel.org
>> Cc: linux-api@...r.kernel.org
>> Cc: lkml <linux-kernel@...r.kernel.org>
>> Signed-off-by: Pavel Tikhomirov <ptikhomirov@...tuozzo.com>
>> ---
>> This is a rework of "mnt: allow to add a mount into an existing group"
>> patch from Andrei. https://lkml.org/lkml/2017/4/28/20
>>
>> New do_set_group is similar to do_move_mount, but with some restrictions
>> of do_move_mount removed and that's why:
>>
>> 1) Allow "cross-namespace" sharing group set. If we allow operation only
>> with mounts from current+anon mount namespace one would still be able to
>> setns(from_mntns) + open_tree(from, OPEN_TREE_CLONE) + setns(to_mntns) +
>> move_mount(anon, to, MOVE_MOUNT_SET_GROUP) to set sharing group to mount
>> in different mount namespace with source mount. But with this approach
>> we would need to create anon mount namespace and mount copy each time,
>> which is just a waste of resources. So instead lets just check if we are
>> allowed to modify both mount namespaces (which looks equivalent to what
>> setns-es and open_tree check).
>>
>> 2) Skip checks wich only apply to actually moving mount which we have in
>> do_move_mount and open_tree. We don't need to check for MNT_LOCKED,
>> d_is_dir matching, unbindable, nsfs loops and ancestor relation as we
>> don't move mounts.
>>
>> Also let's add some new checks:
>>
>> 1) Don't allow to copy sharing from mount with narrow root to a wider
>> root, so that user does not have power to receive more propagations when
>> user already has. (proposed by Andrei)
>>
>> 2) Don't allow to copy sharing from mount with locked children for the
>> same reason, as user shouldn't see propagations to areas overmounted by
>> locked mounts (if the user could not already do it before sharing
>> adjustment).
>>
>> 3) If From is private for MOVE_MOUNT_SET_GROUP let's report an error
>> instead of just doing nothing, so that the user knows that there is
>> probably some logical usage error. (proposed by Christian)
>>
>> Security note: there would be no (new) loops in sharing groups tree,
>> because this new move_mount(MOVE_MOUNT_SET_GROUP) operation only adds
>> one _private_ mount to one group (without moving between groups), the
>> sharing groups tree itself stays unchanged after it.
>>
>> In Virtuozzo we have "mount-v2" implementation, based with the original
>> kernel patch from Andrei, tested for almost a year and it actually
>> decreased number of bugs with mounts a lot. One can take a look on the
>> implementation of sharing group restore in CRIU in "mount-v2" here:
>>
>> https://src.openvz.org/projects/OVZ/repos/criu/browse/criu/mount-v2.c#898
>>
>> This works almost the same with current version of patch if we replace
>> mount(MS_SET_GROUP) to move_mount(MOVE_MOUNT_SET_GROUP), please see
>> super-draft port for mainstream criu, this at least passes
>> non-user-namespaced mount tests (zdtm.py --mounts-v2 -f ns).
>>
>> https://github.com/Snorch/criu/commits/mount-v2-poc
>>
>> v2: Solve the problem mentioned by Andrei:
>> - check mnt_root of "to" is in the sub-tree of mnt_root of "from"
>> - also check "from" has no locked mounts in subroot of "to"
>> v3: Add checks:
>> - check paths to be mount root dentries
>> - return EINVAL if From is private (no sharing to copy)
>>
>> ---
>>   fs/namespace.c             | 75 +++++++++++++++++++++++++++++++++++++-
>>   include/uapi/linux/mount.h |  3 +-
>>   2 files changed, 76 insertions(+), 2 deletions(-)
>>
>> diff --git a/fs/namespace.c b/fs/namespace.c
>> index ab4174a3c802..a7828e695e03 100644
>> --- a/fs/namespace.c
>> +++ b/fs/namespace.c
>> @@ -2684,6 +2684,76 @@ static bool check_for_nsfs_mounts(struct mount *subtree)
>>   	return ret;
>>   }
>>   
>> +static int do_set_group(struct path *from_path, struct path *to_path)
>> +{
>> +	struct mount *from, *to;
>> +	int err;
>> +
>> +	from = real_mount(from_path->mnt);
>> +	to = real_mount(to_path->mnt);
>> +
>> +	namespace_lock();
>> +
>> +	err = -EINVAL;
>> +	/* To and From must be mounted */
>> +	if (!is_mounted(&from->mnt))
>> +		goto out;
>> +	if (!is_mounted(&to->mnt))
>> +		goto out;
>> +
>> +	err = -EPERM;
>> +	/* We should be allowed to modify mount namespaces of both mounts */
>> +	if (!ns_capable(from->mnt_ns->user_ns, CAP_SYS_ADMIN))
>> +		goto out;
>> +	if (!ns_capable(to->mnt_ns->user_ns, CAP_SYS_ADMIN))
>> +		goto out;
>> +
>> +	err = -EINVAL;
>> +	/* To and From paths should be mount roots */
>> +	if (from_path->dentry != from_path->mnt->mnt_root)
>> +		goto out;
>> +	if (to_path->dentry != to_path->mnt->mnt_root)
>> +		goto out;
>> +
>> +	/* Setting sharing groups is only allowed across same superblock */
>> +	if (from->mnt.mnt_sb != to->mnt.mnt_sb)
>> +		goto out;
>> +
>> +	/* From mount root should be wider than To mount root */
>> +	if (!is_subdir(to->mnt.mnt_root, from->mnt.mnt_root))
>> +		goto out;
>> +
>> +	/* From mount should not have locked children in place of To's root */
>> +	if (has_locked_children(from, to->mnt.mnt_root))
>> +		goto out;
>> +
>> +	/* Setting sharing groups is only allowed on private mounts */
>> +	if (IS_MNT_SHARED(to) || IS_MNT_SLAVE(to))
>> +		goto out;
>> +
>> +	/* From should not be private */
>> +	if (!IS_MNT_SHARED(from) && !IS_MNT_SLAVE(from))
>> +		goto out;
>> +
>> +	if (IS_MNT_SLAVE(from)) {
>> +		struct mount *m = from->mnt_master;
>> +
>> +		list_add(&to->mnt_slave, &m->mnt_slave_list);
>> +		to->mnt_master = m;
>> +	}
>> +
>> +	if (IS_MNT_SHARED(from)) {
>> +		to->mnt_group_id = from->mnt_group_id;
>> +		list_add(&to->mnt_share, &from->mnt_share);
>> +		set_mnt_shared(to);
> 
> Thanks for the updated patch.
> 
> Note, set_mnt_shared() writes to the vfsmount so this should be:
> 
> if (IS_MNT_SHARED(from)) {
> 	to->mnt_group_id = from->mnt_group_id;
> 	list_add(&to->mnt_share, &from->mnt_share);
> 	lock_mount_hash();
> 	set_mnt_shared(to);
> 	unlock_mount_hash();
> }

Thanks for pointing that out! Sent v4 to fix it.

> 
>> +	}
> 
> 
>> +	}
>> +
>> +	err = 0;
>> +out:
>> +	namespace_unlock();
>> +	return err;
>> +}
>> +
>>   static int do_move_mount(struct path *old_path, struct path *new_path)
>>   {
>>   	struct mnt_namespace *ns;
>> @@ -3669,7 +3739,10 @@ SYSCALL_DEFINE5(move_mount,
>>   	if (ret < 0)
>>   		goto out_to;
>>   
>> -	ret = do_move_mount(&from_path, &to_path);
>> +	if (flags & MOVE_MOUNT_SET_GROUP)
>> +		ret = do_set_group(&from_path, &to_path);
>> +	else
>> +		ret = do_move_mount(&from_path, &to_path);
>>   
>>   out_to:
>>   	path_put(&to_path);
>> diff --git a/include/uapi/linux/mount.h b/include/uapi/linux/mount.h
>> index dd7a166fdf9c..4d93967f8aea 100644
>> --- a/include/uapi/linux/mount.h
>> +++ b/include/uapi/linux/mount.h
>> @@ -73,7 +73,8 @@
>>   #define MOVE_MOUNT_T_SYMLINKS		0x00000010 /* Follow symlinks on to path */
>>   #define MOVE_MOUNT_T_AUTOMOUNTS		0x00000020 /* Follow automounts on to path */
>>   #define MOVE_MOUNT_T_EMPTY_PATH		0x00000040 /* Empty to path permitted */
>> -#define MOVE_MOUNT__MASK		0x00000077
>> +#define MOVE_MOUNT_SET_GROUP		0x00000100 /* Set sharing group instead */
>> +#define MOVE_MOUNT__MASK		0x00000177
>>   
>>   /*
>>    * fsopen() flags.
>> -- 
>> 2.31.1

-- 
Best regards, Tikhomirov Pavel
Software Developer, Virtuozzo.

Powered by blists - more mailing lists