Skip to contentSkip to conversation

security/selinux: fix SECURITY_LSM_NATIVE_LABELS on reused superblock

PATCH In review J. Bruce Fields 3 participants 8 messages Mar 5, 2019 – Mar 11, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 security/selinux: fix SECURITY_LSM_NATIVE_LABELS on reused superblock

From: "J. Bruce Fields" <[email protected]>

In the case when we're reusing a superblock, selinux_sb_clone_mnt_opts() fails to set set_kern_flags, with the result that nfs_clone_sb_security() incorrectly clears NFS_CAP_SECURITY_LABEL.

The result is that if you mount the same NFS filesystem twice, NFS security labels are turned off, even if they would work fine if you mounted the filesystem only once.

("fixes" may be not exactly the right tag, it may be more like "fixed-other-cases-but-missed-this-one".)

Cc Scott Mayhew <[email protected]>
Fixes 0b4d3452b8b4 "security/selinux: allow security_sb_clone_mnt_opts..."
---
 security/selinux/hooks.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
security/selinux/hooks.c +4 -1
index f0e36c3492ba..5e9304567233 100644
@@ -959,8 +959,11 @@ static int selinux_sb_clone_mnt_opts(const struct super_block *oldsb,
 	BUG_ON(!(oldsbsec->flags & SE_SBINITIALIZED));
 
 	/* if fs is reusing a sb, make sure that the contexts match */
-	if (newsbsec->flags & SE_SBINITIALIZED)
+	if (newsbsec->flags & SE_SBINITIALIZED) {
+		if ((kern_flags & SECURITY_LSM_NATIVE_LABELS) && !set_context)
+			*set_kern_flags |= SECURITY_LSM_NATIVE_LABELS;
 		return selinux_cmp_sb_context(oldsb, newsb);
+	}
 
 	mutex_lock(&newsbsec->lock);
 
Signature · 1 line
2.20.1
#1 lore raw

Whoops, probably should have cc'd linux-nfs as well.--b.

Quoted 41 lines · bfields

From: "J. Bruce Fields" <[email protected]>

In the case when we're reusing a superblock, selinux_sb_clone_mnt_opts() fails to set set_kern_flags, with the result that nfs_clone_sb_security() incorrectly clears NFS_CAP_SECURITY_LABEL.

The result is that if you mount the same NFS filesystem twice, NFS security labels are turned off, even if they would work fine if you mounted the filesystem only once.

("fixes" may be not exactly the right tag, it may be more like "fixed-other-cases-but-missed-this-one".)

Cc Scott Mayhew <[email protected]>
Fixes 0b4d3452b8b4 "security/selinux: allow security_sb_clone_mnt_opts..."
---
 security/selinux/hooks.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
security/selinux/hooks.c +4 -1
index f0e36c3492ba..5e9304567233 100644
@@ -959,8 +959,11 @@ static int selinux_sb_clone_mnt_opts(const struct super_block *oldsb,
 	BUG_ON(!(oldsbsec->flags & SE_SBINITIALIZED));
 
 	/* if fs is reusing a sb, make sure that the contexts match */
-	if (newsbsec->flags & SE_SBINITIALIZED)
+	if (newsbsec->flags & SE_SBINITIALIZED) {
+		if ((kern_flags & SECURITY_LSM_NATIVE_LABELS) && !set_context)
+			*set_kern_flags |= SECURITY_LSM_NATIVE_LABELS;
 		return selinux_cmp_sb_context(oldsb, newsb);
+	}
 
 	mutex_lock(&newsbsec->lock);
 
Signature · 1 line
2.20.1
Acked-by on v1 1/1
#2 lore raw
In reply to J. Bruce Fields
Quoted 10 lines · J. Bruce Fields

From: "J. Bruce Fields" <[email protected]>

In the case when we're reusing a superblock, selinux_sb_clone_mnt_opts() fails to set set_kern_flags, with the result that nfs_clone_sb_security() incorrectly clears NFS_CAP_SECURITY_LABEL.

The result is that if you mount the same NFS filesystem twice, NFS security labels are turned off, even if they would work fine if you mounted the filesystem only once.

Quoted 7 lines

("fixes" may be not exactly the right tag, it may be more like "fixed-other-cases-but-missed-this-one".)

Cc Scott Mayhew <[email protected]>
Fixes 0b4d3452b8b4 "security/selinux: allow security_sb_clone_mnt_opts..."

Do you have some tests you are using for the selinux nfs support? I have an open issue on the selinux-testsuite with an example script for running the regular selinux tests on a NFS mount but it can't fully succeed as noted there,
https://github.com/SELinuxProject/selinux-testsuite/issues/32

I've also have another script to test context= mount handling for nfs since that should take precedence over native labels; it looks like that might be broken again:

#!/bin/sh
cat > /etc/exports <<EOF
/home localhost(rw,no_root_squash,security_label)
EOF
exportfs -a
systemctl start nfs-server
mkdir -p /mnt/home
mount -t nfs -o vers=4.0,context=system_u:object_r:etc_t:s0 localhost:/home /mnt/home echo "Under NFSv4.0:"
ls -Za /mnt/home
touch /mnt/home/foo
ls -Z /mnt/home/foo
ls -Z /home/foo
rm /mnt/home/foo
umount /mnt/home
mount -t nfs -o vers=4.2,context=system_u:object_r:etc_t:s0 localhost:/home /mnt/home echo "Under NFSv4.2:"
ls -Za /mnt/home
touch /mnt/home/foo
ls -Z /mnt/home/foo
ls -Z /home/foo
rm /home/foo
umount /mnt/home
rmdir /mnt/home
rm /etc/exports
exportfs -ua
systemctl stop nfs-server

Quoted 22 lines
---
  security/selinux/hooks.c | 5 ++++-
  1 file changed, 4 insertions(+), 1 deletion(-)
security/selinux/hooks.c +4 -1
index f0e36c3492ba..5e9304567233 100644
@@ -959,8 +959,11 @@ static int selinux_sb_clone_mnt_opts(const struct super_block *oldsb,
  	BUG_ON(!(oldsbsec->flags & SE_SBINITIALIZED));
 
  	/* if fs is reusing a sb, make sure that the contexts match */
-	if (newsbsec->flags & SE_SBINITIALIZED)
+	if (newsbsec->flags & SE_SBINITIALIZED) {
+		if ((kern_flags & SECURITY_LSM_NATIVE_LABELS) && !set_context)
+			*set_kern_flags |= SECURITY_LSM_NATIVE_LABELS;
  		return selinux_cmp_sb_context(oldsb, newsb);
+	}
 
  	mutex_lock(&newsbsec->lock);
 
#3 lore raw
Quoted 14 lines · Stephen Smalley
Quoted 13 lines · J. Bruce Fields

From: "J. Bruce Fields" <[email protected]>

In the case when we're reusing a superblock, selinux_sb_clone_mnt_opts() fails to set set_kern_flags, with the result that nfs_clone_sb_security() incorrectly clears NFS_CAP_SECURITY_LABEL.

The result is that if you mount the same NFS filesystem twice, NFS security labels are turned off, even if they would work fine if you mounted the filesystem only once.

("fixes" may be not exactly the right tag, it may be more like "fixed-other-cases-but-missed-this-one".)

Quoted 8 lines
Quoted 4 lines
Cc Scott Mayhew <[email protected]>
Fixes 0b4d3452b8b4 "security/selinux: allow security_sb_clone_mnt_opts..."

Do you have some tests you are using for the selinux nfs support?

No. I'll ask around. Trond or Anna, do either of you do any selinux testing?

Quoted 4 lines

I have an open issue on the selinux-testsuite with an example script for running the regular selinux tests on a NFS mount but it can't fully succeed as noted there,
https://github.com/SELinuxProject/selinux-testsuite/issues/32

So if I just clone
https://github.com/SELinuxProject/selinux-testsuite.git and filter out those known failures, would that be a good starting point?

Quoted 3 lines

I've also have another script to test context= mount handling for nfs since that should take precedence over native labels; it looks
like that might be broken again:

Thanks for the report, I'll take a look. That's before or after applying this patch? Assuming the former, do you have an idea how recent a regression it is?

--b.

Quoted 49 lines

#!/bin/sh
cat > /etc/exports <<EOF
/home localhost(rw,no_root_squash,security_label) EOF
exportfs -a
systemctl start nfs-server
mkdir -p /mnt/home
mount -t nfs -o vers=4.0,context=system_u:object_r:etc_t:s0 localhost:/home /mnt/home
echo "Under NFSv4.0:"
ls -Za /mnt/home
touch /mnt/home/foo
ls -Z /mnt/home/foo
ls -Z /home/foo
rm /mnt/home/foo
umount /mnt/home
mount -t nfs -o vers=4.2,context=system_u:object_r:etc_t:s0 localhost:/home /mnt/home
echo "Under NFSv4.2:"
ls -Za /mnt/home
touch /mnt/home/foo
ls -Z /mnt/home/foo
ls -Z /home/foo
rm /home/foo
umount /mnt/home
rmdir /mnt/home
rm /etc/exports
exportfs -ua
systemctl stop nfs-server

Quoted 19 lines
---
 security/selinux/hooks.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
security/selinux/hooks.c +4 -1
index f0e36c3492ba..5e9304567233 100644
@@ -959,8 +959,11 @@ static int selinux_sb_clone_mnt_opts(const struct super_block *oldsb,
 	BUG_ON(!(oldsbsec->flags & SE_SBINITIALIZED));
 	/* if fs is reusing a sb, make sure that the contexts match */
-	if (newsbsec->flags & SE_SBINITIALIZED)
+	if (newsbsec->flags & SE_SBINITIALIZED) {
+		if ((kern_flags & SECURITY_LSM_NATIVE_LABELS) && !set_context)
+			*set_kern_flags |= SECURITY_LSM_NATIVE_LABELS;
 		return selinux_cmp_sb_context(oldsb, newsb);
+	}
 	mutex_lock(&newsbsec->lock);
 
#4 lore raw
Quoted 27 lines · J. Bruce Fields
Quoted 14 lines · Stephen Smalley
Quoted 13 lines · J. Bruce Fields

From: "J. Bruce Fields" <[email protected]>

In the case when we're reusing a superblock, selinux_sb_clone_mnt_opts() fails to set set_kern_flags, with the result that nfs_clone_sb_security() incorrectly clears NFS_CAP_SECURITY_LABEL.

The result is that if you mount the same NFS filesystem twice, NFS security labels are turned off, even if they would work fine if you mounted the filesystem only once.

("fixes" may be not exactly the right tag, it may be more like "fixed-other-cases-but-missed-this-one".)

Quoted 8 lines
Quoted 4 lines
Cc Scott Mayhew <[email protected]>
Fixes 0b4d3452b8b4 "security/selinux: allow security_sb_clone_mnt_opts..."

Do you have some tests you are using for the selinux nfs support?

No. I'll ask around. Trond or Anna, do either of you do any selinux testing?

Quoted 8 lines
Quoted 4 lines

I have an open issue on the selinux-testsuite with an example script for running the regular selinux tests on a NFS mount but it can't fully succeed as noted there,
https://github.com/SELinuxProject/selinux-testsuite/issues/32

So if I just clone
https://github.com/SELinuxProject/selinux-testsuite.git and filter out those known failures, would that be a good starting point?

Yes, although you might want to remove the inet_socket and sctp tests from tests/Makefile before running it over NFS; it looks like there are some interactions between NFS and labeled networking (netlabel and ipsec/xfrm) that require further test policy changes to permit.

Quoted 8 lines
Quoted 4 lines

I've also have another script to test context= mount handling for nfs since that should take precedence over native labels; it looks
like that might be broken again:

Thanks for the report, I'll take a look. That's before or after applying this patch? Assuming the former, do you have an idea how recent a regression it is?

Now I'm having difficulty reproducing it entirely. I thought on stock Fedora 29 (4.20.x) I was seeing the actual underlying security labels leaking through on files within the NFS mount despite using a context= mount, while correctly seeing the context mount values with your patch, but now I can't seem to repro. It was this bug that originally motivated Scott's commit that you are further fixing IIUC,
https://github.com/SELinuxProject/selinux-kernel/issues/35

Quoted 52 lines

--b.

Quoted 49 lines

#!/bin/sh
cat > /etc/exports <<EOF
/home localhost(rw,no_root_squash,security_label) EOF
exportfs -a
systemctl start nfs-server
mkdir -p /mnt/home
mount -t nfs -o vers=4.0,context=system_u:object_r:etc_t:s0 localhost:/home /mnt/home
echo "Under NFSv4.0:"
ls -Za /mnt/home
touch /mnt/home/foo
ls -Z /mnt/home/foo
ls -Z /home/foo
rm /mnt/home/foo
umount /mnt/home
mount -t nfs -o vers=4.2,context=system_u:object_r:etc_t:s0 localhost:/home /mnt/home
echo "Under NFSv4.2:"
ls -Za /mnt/home
touch /mnt/home/foo
ls -Z /mnt/home/foo
ls -Z /home/foo
rm /home/foo
umount /mnt/home
rmdir /mnt/home
rm /etc/exports
exportfs -ua
systemctl stop nfs-server

Quoted 19 lines
---
  security/selinux/hooks.c | 5 ++++-
  1 file changed, 4 insertions(+), 1 deletion(-)
security/selinux/hooks.c +4 -1
index f0e36c3492ba..5e9304567233 100644
@@ -959,8 +959,11 @@ static int selinux_sb_clone_mnt_opts(const struct super_block *oldsb,
  	BUG_ON(!(oldsbsec->flags & SE_SBINITIALIZED));
  	/* if fs is reusing a sb, make sure that the contexts match */
-	if (newsbsec->flags & SE_SBINITIALIZED)
+	if (newsbsec->flags & SE_SBINITIALIZED) {
+		if ((kern_flags & SECURITY_LSM_NATIVE_LABELS) && !set_context)
+			*set_kern_flags |= SECURITY_LSM_NATIVE_LABELS;
  		return selinux_cmp_sb_context(oldsb, newsb);
+	}
  	mutex_lock(&newsbsec->lock);
 
#5 lore raw
Quoted 10 lines · Stephen Smalley
Quoted 8 lines · J. Bruce Fields
Quoted 3 lines · Stephen Smalley

I've also have another script to test context= mount handling for nfs since that should take precedence over native labels; it looks
like that might be broken again:

Thanks for the report, I'll take a look. That's before or after applying this patch? Assuming the former, do you have an idea how recent a regression it is?

Quoted 8 lines

Now I'm having difficulty reproducing it entirely. I thought on stock Fedora 29 (4.20.x) I was seeing the actual underlying security labels leaking through on files within the NFS mount despite using a context= mount, while correctly seeing the context mount values with your patch, but now I can't seem to repro. It was this bug that originally motivated Scott's commit that you are further fixing IIUC,
https://github.com/SELinuxProject/selinux-kernel/issues/35

For what it's worth, I can't reproduce. (If I mount with -overs=4.2,context=system_u:object_r:etc_t:s0 then ls -Z, I only see system_u:object_r:etc_t:s0.)

--b.

#6 lore raw
Quoted 20 lines · J. Bruce Fields
Quoted 10 lines · Stephen Smalley
Quoted 8 lines · J. Bruce Fields
Quoted 3 lines · Stephen Smalley

I've also have another script to test context= mount handling for nfs since that should take precedence over native labels; it looks
like that might be broken again:

Thanks for the report, I'll take a look. That's before or after applying this patch? Assuming the former, do you have an idea how recent a regression it is?

Quoted 8 lines

Now I'm having difficulty reproducing it entirely. I thought on stock Fedora 29 (4.20.x) I was seeing the actual underlying security labels leaking through on files within the NFS mount despite using a context= mount, while correctly seeing the context mount values with your patch, but now I can't seem to repro. It was this bug that originally motivated Scott's commit that you are further fixing IIUC,
https://github.com/SELinuxProject/selinux-kernel/issues/35

Quoted 3 lines

For what it's worth, I can't reproduce. (If I mount with -overs=4.2,context=system_u:object_r:etc_t:s0 then ls -Z, I only see system_u:object_r:etc_t:s0.)

Yes, sorry for the noise.

maintainer
#7 lore raw
In reply to J. Bruce Fields
Quoted 13 lines · J. Bruce Fields

From: "J. Bruce Fields" <[email protected]>

In the case when we're reusing a superblock, selinux_sb_clone_mnt_opts() fails to set set_kern_flags, with the result that nfs_clone_sb_security() incorrectly clears NFS_CAP_SECURITY_LABEL.

The result is that if you mount the same NFS filesystem twice, NFS security labels are turned off, even if they would work fine if you mounted the filesystem only once.

("fixes" may be not exactly the right tag, it may be more like "fixed-other-cases-but-missed-this-one".)

Quoted 7 lines
Cc Scott Mayhew <[email protected]>
Fixes 0b4d3452b8b4 "security/selinux: allow security_sb_clone_mnt_opts..."
---
 security/selinux/hooks.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

Thanks for the fix. I just merged this into selinux/stable-5.1 and I'll send this up to Linus tomorrow.

Quoted 19 lines
security/selinux/hooks.c +4 -2
index f0e36c3492ba..5e9304567233 100644
@@ -959,8 +959,11 @@ static int selinux_sb_clone_mnt_opts(const struct super_block *oldsb,
        BUG_ON(!(oldsbsec->flags & SE_SBINITIALIZED));
 
        /* if fs is reusing a sb, make sure that the contexts match */
-       if (newsbsec->flags & SE_SBINITIALIZED)
+       if (newsbsec->flags & SE_SBINITIALIZED) {
+               if ((kern_flags & SECURITY_LSM_NATIVE_LABELS) && !set_context)
+                       *set_kern_flags |= SECURITY_LSM_NATIVE_LABELS;
                return selinux_cmp_sb_context(oldsb, newsb);
+       }
 
        mutex_lock(&newsbsec->lock);
 
--

2.20.1

Signature · 2 lines
paul moore
www.paul-moore.com