[PATCH v1 0/2] proc: Relax check of mount visibility

[PATCH v1 0/2] proc: Relax check of mount visibility

[Alexey Gladkov]

If only the dynamic part of procfs is mounted (subset=pid), then there is no
need to check if procfs is fully visible to the user in the new user namespace.

Alexey Gladkov (2):
  proc: Relax check of mount visibility
  Show /proc/self/net only for CAP_NET_ADMIN

 fs/namespace.c     | 27 ++++++++++++++++-----------
 fs/proc/proc_net.c |  6 ++++++
 fs/proc/root.c     | 16 +++++++++-------
 include/linux/fs.h |  1 +
 4 files changed, 32 insertions(+), 18 deletions(-)

-- 
2.25.4

[PATCH v1 1/2] proc: Relax check of mount visibility

[Alexey Gladkov]

Allow to mount of procfs with subset=pid option even if the entire
procfs is not fully accessible to the user.

Signed-off-by: Alexey Gladkov <gladkov.alexey@gmail.com>
---
 fs/namespace.c     | 27 ++++++++++++++++-----------
 fs/proc/root.c     | 16 +++++++++-------
 include/linux/fs.h |  1 +
 3 files changed, 26 insertions(+), 18 deletions(-)

diff --git a/fs/namespace.c b/fs/namespace.c
index 4a0f600a3328..ab9d607921da 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -3949,18 +3949,23 @@ static bool mnt_already_visible(struct mnt_namespace *ns,
 		    ((mnt_flags & MNT_ATIME_MASK) != (new_flags & MNT_ATIME_MASK)))
 			continue;
 
-		/* This mount is not fully visible if there are any
-		 * locked child mounts that cover anything except for
-		 * empty directories.
+		/* If this filesystem is completely dynamic, then it
+		 * makes no sense to check for any child mounts.
 		 */
-		list_for_each_entry(child, &mnt->mnt_mounts, mnt_child) {
-			struct inode *inode = child->mnt_mountpoint->d_inode;
-			/* Only worry about locked mounts */
-			if (!(child->mnt.mnt_flags & MNT_LOCKED))
-				continue;
-			/* Is the directory permanetly empty? */
-			if (!is_empty_dir_inode(inode))
-				goto next;
+		if (!(sb->s_iflags & SB_I_DYNAMIC)) {
+			/* This mount is not fully visible if there are any
+			 * locked child mounts that cover anything except for
+			 * empty directories.
+			 */
+			list_for_each_entry(child, &mnt->mnt_mounts, mnt_child) {
+				struct inode *inode = child->mnt_mountpoint->d_inode;
+				/* Only worry about locked mounts */
+				if (!(child->mnt.mnt_flags & MNT_LOCKED))
+					continue;
+				/* Is the directory permanetly empty? */
+				if (!is_empty_dir_inode(inode))
+					goto next;
+			}
 		}
 		/* Preserve the locked attributes */
 		*new_mnt_flags |= mnt_flags & (MNT_LOCK_READONLY | \
diff --git a/fs/proc/root.c b/fs/proc/root.c
index 5e444d4f9717..c6bf74de1906 100644
--- a/fs/proc/root.c
+++ b/fs/proc/root.c
@@ -145,18 +145,21 @@ static int proc_parse_param(struct fs_context *fc, struct fs_parameter *param)
 	return 0;
 }
 
-static void proc_apply_options(struct proc_fs_info *fs_info,
+static void proc_apply_options(struct super_block *s,
 			       struct fs_context *fc,
 			       struct user_namespace *user_ns)
 {
 	struct proc_fs_context *ctx = fc->fs_private;
+	struct proc_fs_info *fs_info = proc_sb_info(s);
 
 	if (ctx->mask & (1 << Opt_gid))
 		fs_info->pid_gid = make_kgid(user_ns, ctx->gid);
 	if (ctx->mask & (1 << Opt_hidepid))
 		fs_info->hide_pid = ctx->hidepid;
-	if (ctx->mask & (1 << Opt_subset))
+	if (ctx->mask & (1 << Opt_subset)) {
 		fs_info->pidonly = ctx->pidonly;
+		s->s_iflags |= SB_I_DYNAMIC;
+	}
 }
 
 static int proc_fill_super(struct super_block *s, struct fs_context *fc)
@@ -170,9 +173,6 @@ static int proc_fill_super(struct super_block *s, struct fs_context *fc)
 	if (!fs_info)
 		return -ENOMEM;
 
-	fs_info->pid_ns = get_pid_ns(ctx->pid_ns);
-	proc_apply_options(fs_info, fc, current_user_ns());
-
 	/* User space would break if executables or devices appear on proc */
 	s->s_iflags |= SB_I_USERNS_VISIBLE | SB_I_NOEXEC | SB_I_NODEV;
 	s->s_flags |= SB_NODIRATIME | SB_NOSUID | SB_NOEXEC;
@@ -183,6 +183,9 @@ static int proc_fill_super(struct super_block *s, struct fs_context *fc)
 	s->s_time_gran = 1;
 	s->s_fs_info = fs_info;
 
+	fs_info->pid_ns = get_pid_ns(ctx->pid_ns);
+	proc_apply_options(s, fc, current_user_ns());
+
 	/*
 	 * procfs isn't actually a stacking filesystem; however, there is
 	 * too much magic going on inside it to permit stacking things on
@@ -216,11 +219,10 @@ static int proc_fill_super(struct super_block *s, struct fs_context *fc)
 static int proc_reconfigure(struct fs_context *fc)
 {
 	struct super_block *sb = fc->root->d_sb;
-	struct proc_fs_info *fs_info = proc_sb_info(sb);
 
 	sync_filesystem(sb);
 
-	proc_apply_options(fs_info, fc, current_user_ns());
+	proc_apply_options(sb, fc, current_user_ns());
 	return 0;
 }
 
diff --git a/include/linux/fs.h b/include/linux/fs.h
index f5abba86107d..aff5ed9e8f82 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1413,6 +1413,7 @@ extern int send_sigurg(struct fown_struct *fown);
 #define SB_I_USERNS_VISIBLE		0x00000010 /* fstype already mounted */
 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE	0x00000020
 #define SB_I_UNTRUSTED_MOUNTER		0x00000040
+#define SB_I_DYNAMIC			0x00000080
 
 #define SB_I_SKIP_SYNC	0x00000100	/* Skip superblock at global sync */
 
-- 
2.25.4

[PATCH v1 2/2] Show /proc/self/net only for CAP_NET_ADMIN

[Alexey Gladkov]

Show /proc/self/net only for CAP_NET_ADMIN if procfs is mounted with
subset=pid option in user namespace. This is done to avoid possible
information leakage.

Signed-off-by: Alexey Gladkov <gladkov.alexey@gmail.com>
---
 fs/proc/proc_net.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/fs/proc/proc_net.c b/fs/proc/proc_net.c
index dba63b2429f0..11fa2c4b3529 100644
--- a/fs/proc/proc_net.c
+++ b/fs/proc/proc_net.c
@@ -275,6 +275,12 @@ static struct net *get_proc_task_net(struct inode *dir)
 	struct task_struct *task;
 	struct nsproxy *ns;
 	struct net *net = NULL;
+	struct proc_fs_info *fs_info = proc_sb_info(dir->i_sb);
+
+	if ((fs_info->pidonly == PROC_PIDONLY_ON) &&
+	    (current_user_ns() != &init_user_ns) &&
+	    !capable(CAP_NET_ADMIN))
+		return net;
 
 	rcu_read_lock();
 	task = pid_task(proc_pid(dir), PIDTYPE_PID);
-- 
2.25.4

Re: [PATCH v1 2/2] Show /proc/self/net only for CAP_NET_ADMIN

[Eric W. Biederman]

Alexey Gladkov <gladkov.alexey@gmail.com> writes:

> Show /proc/self/net only for CAP_NET_ADMIN if procfs is mounted with
> subset=pid option in user namespace. This is done to avoid possible
> information leakage.
>
> Signed-off-by: Alexey Gladkov <gladkov.alexey@gmail.com>
> ---
>  fs/proc/proc_net.c | 6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/fs/proc/proc_net.c b/fs/proc/proc_net.c
> index dba63b2429f0..11fa2c4b3529 100644
> --- a/fs/proc/proc_net.c
> +++ b/fs/proc/proc_net.c
> @@ -275,6 +275,12 @@ static struct net *get_proc_task_net(struct inode *dir)
>  	struct task_struct *task;
>  	struct nsproxy *ns;
>  	struct net *net = NULL;
> +	struct proc_fs_info *fs_info = proc_sb_info(dir->i_sb);
> +
> +	if ((fs_info->pidonly == PROC_PIDONLY_ON) &&
> +	    (current_user_ns() != &init_user_ns) &&
> +	    !capable(CAP_NET_ADMIN))
> +		return net;
>
>  	rcu_read_lock();
>  	task = pid_task(proc_pid(dir), PIDTYPE_PID);

Hmm.

I see 3 options going forward.

1) We just make PROC_PIDONLY_ON mean the net directory does not exist.
   No permission checks just always fail.

2) Move the permission checks into opendir/readdir and whichever
   is the appropriate method there and always allow the dentries
   to be cached.

3) Simply cache the mounters credentials and make access to the
   net directories contingent of the permisions of the mounter of
   proc.  Something like the code below.

static struct net *get_proc_task_net(struct inode *dir)
{
	struct task_struct *task;
	struct nsproxy *ns;
	struct net *net = NULL;

	rcu_read_lock();
	task = pid_task(proc_pid(dir), PIDTYPE_PID);
	if (task != NULL) {
		task_lock(task);
		ns = task->nsproxy;
		if (ns != NULL)
			net = get_net(ns->net_ns);
		task_unlock(task);
	}
	rcu_read_unlock();
	if ((fs_info->pidonly == PROC_PIDONLY_ON) &&
            !security_capable(fs_info->mounter_cred,
			      net->user_ns, CAP_SYS_ADMIN,
			      CAP_OPT_NONE)) {
		put_net(net);
		net = NULL;
	}
	return net;
}

Eric

Re: [PATCH v1 2/2] Show /proc/self/net only for CAP_NET_ADMIN

[Alexey Gladkov]

On Mon, Jul 27, 2020 at 11:29:36AM -0500, Eric W. Biederman wrote:
> Alexey Gladkov <gladkov.alexey@gmail.com> writes:
> 
> > Show /proc/self/net only for CAP_NET_ADMIN if procfs is mounted with
> > subset=pid option in user namespace. This is done to avoid possible
> > information leakage.
> >
> > Signed-off-by: Alexey Gladkov <gladkov.alexey@gmail.com>
> > ---
> >  fs/proc/proc_net.c | 6 ++++++
> >  1 file changed, 6 insertions(+)
> >
> > diff --git a/fs/proc/proc_net.c b/fs/proc/proc_net.c
> > index dba63b2429f0..11fa2c4b3529 100644
> > --- a/fs/proc/proc_net.c
> > +++ b/fs/proc/proc_net.c
> > @@ -275,6 +275,12 @@ static struct net *get_proc_task_net(struct inode *dir)
> >  	struct task_struct *task;
> >  	struct nsproxy *ns;
> >  	struct net *net = NULL;
> > +	struct proc_fs_info *fs_info = proc_sb_info(dir->i_sb);
> > +
> > +	if ((fs_info->pidonly == PROC_PIDONLY_ON) &&
> > +	    (current_user_ns() != &init_user_ns) &&
> > +	    !capable(CAP_NET_ADMIN))
> > +		return net;
> >
> >  	rcu_read_lock();
> >  	task = pid_task(proc_pid(dir), PIDTYPE_PID);
> 
> Hmm.
> 
> I see 3 options going forward.
> 
> 1) We just make PROC_PIDONLY_ON mean the net directory does not exist.
>    No permission checks just always fail.

I think it's wrong. Now if someone mounts a fully visible procfs then they
can see this directory. Hiding this directory completely will change the
current behavior.

> 2) Move the permission checks into opendir/readdir and whichever
>    is the appropriate method there and always allow the dentries
>    to be cached.

At first I did so, but then I transferred this check to get_proc_task_net
because if this function does not return anything, then 'net' directory
will exist but will simply be empty.

This allowed us to get rid of unnecessary wrappers for opendir/lookup.

> 3) Simply cache the mounters credentials and make access to the
>    net directories contingent of the permisions of the mounter of
>    proc.  Something like the code below.

Interesting idea. I like that :)

> static struct net *get_proc_task_net(struct inode *dir)
> {
> 	struct task_struct *task;
> 	struct nsproxy *ns;
> 	struct net *net = NULL;
> 
> 	rcu_read_lock();
> 	task = pid_task(proc_pid(dir), PIDTYPE_PID);
> 	if (task != NULL) {
> 		task_lock(task);
> 		ns = task->nsproxy;
> 		if (ns != NULL)
> 			net = get_net(ns->net_ns);
> 		task_unlock(task);
> 	}
> 	rcu_read_unlock();
> 	if ((fs_info->pidonly == PROC_PIDONLY_ON) &&

Is this check necessary? I mean, isn't it worth extending this check to
other cases?

>             !security_capable(fs_info->mounter_cred,
> 			      net->user_ns, CAP_SYS_ADMIN,
> 			      CAP_OPT_NONE)) {
> 		put_net(net);
> 		net = NULL;
> 	}
> 	return net;
> }
> 
> Eric
> 

-- 
Rgrds, legion

[PATCH v2 2/2] Show /proc/self/net only for CAP_NET_ADMIN

[Alexey Gladkov]

Cache the mounters credentials and make access to the net directories
contingent of the permissions of the mounter of proc.

Show /proc/self/net only if mounter has CAP_NET_ADMIN and if proc is
mounted with subset=pid option.

Signed-off-by: Alexey Gladkov <gladkov.alexey@gmail.com>
---
 fs/proc/proc_net.c      | 8 ++++++++
 fs/proc/root.c          | 7 +++++++
 include/linux/proc_fs.h | 1 +
 3 files changed, 16 insertions(+)

diff --git a/fs/proc/proc_net.c b/fs/proc/proc_net.c
index dba63b2429f0..c43fc5c907db 100644
--- a/fs/proc/proc_net.c
+++ b/fs/proc/proc_net.c
@@ -26,6 +26,7 @@
 #include <linux/uidgid.h>
 #include <net/net_namespace.h>
 #include <linux/seq_file.h>
+#include <linux/security.h>
 
 #include "internal.h"
 
@@ -275,6 +276,7 @@ static struct net *get_proc_task_net(struct inode *dir)
 	struct task_struct *task;
 	struct nsproxy *ns;
 	struct net *net = NULL;
+	struct proc_fs_info *fs_info = proc_sb_info(dir->i_sb);
 
 	rcu_read_lock();
 	task = pid_task(proc_pid(dir), PIDTYPE_PID);
@@ -287,6 +289,12 @@ static struct net *get_proc_task_net(struct inode *dir)
 	}
 	rcu_read_unlock();
 
+	if (net && (fs_info->pidonly == PROC_PIDONLY_ON) &&
+	    security_capable(fs_info->mounter_cred, net->user_ns, CAP_NET_ADMIN, CAP_OPT_NONE) < 0) {
+		put_net(net);
+		net = NULL;
+	}
+
 	return net;
 }
 
diff --git a/fs/proc/root.c b/fs/proc/root.c
index c6bf74de1906..eeeda375cf85 100644
--- a/fs/proc/root.c
+++ b/fs/proc/root.c
@@ -184,6 +184,8 @@ static int proc_fill_super(struct super_block *s, struct fs_context *fc)
 	s->s_fs_info = fs_info;
 
 	fs_info->pid_ns = get_pid_ns(ctx->pid_ns);
+	fs_info->mounter_cred = get_cred(fc->cred);
+
 	proc_apply_options(s, fc, current_user_ns());
 
 	/*
@@ -219,9 +221,13 @@ static int proc_fill_super(struct super_block *s, struct fs_context *fc)
 static int proc_reconfigure(struct fs_context *fc)
 {
 	struct super_block *sb = fc->root->d_sb;
+	struct proc_fs_info *fs_info = proc_sb_info(sb);
 
 	sync_filesystem(sb);
 
+	put_cred(fs_info->mounter_cred);
+	fs_info->mounter_cred = get_cred(fc->cred);
+
 	proc_apply_options(sb, fc, current_user_ns());
 	return 0;
 }
@@ -276,6 +282,7 @@ static void proc_kill_sb(struct super_block *sb)
 
 	kill_anon_super(sb);
 	put_pid_ns(fs_info->pid_ns);
+	put_cred(fs_info->mounter_cred);
 	kfree(fs_info);
 }
 
diff --git a/include/linux/proc_fs.h b/include/linux/proc_fs.h
index d1eed1b43651..671c6dafc4ee 100644
--- a/include/linux/proc_fs.h
+++ b/include/linux/proc_fs.h
@@ -63,6 +63,7 @@ struct proc_fs_info {
 	kgid_t pid_gid;
 	enum proc_hidepid hide_pid;
 	enum proc_pidonly pidonly;
+	struct cred *mounter_cred;
 };
 
 static inline struct proc_fs_info *proc_sb_info(struct super_block *sb)
-- 
2.25.4