selinuxns: extend namespace support to security.selinux xattrs

From the init namespace, I'll access the same file:

This is a proof-of-concept patch to demonstrate an approach to supporting SELinux namespaces for security.selinux xattr labels.

This follows on from the experimental SELinux namespace code posted by
Stephen: https://marc.info/?l=selinux&m=150696042210126&w=2

In the initial code, namespacing was only implemented for in-core inodes per this posting: https://marc.info/?l=selinux&m=150696324110943&w=2

In this patch, namespacing is extended to on-disk inode labels (xattrs), transparently to normal applications.

A summary of the approach is as follows:

1. Add a namespace "name" field to the SELinux namespace, which is specified by writing to the selinuxfs unshare node (instead of the current boolean value) during namespace creation.

e.g. if the namespace name is "vm8", run:

   # echo vm8 > /sys/fs/selinux/unshare

followed by the remaining steps from the original code.

This value can also be read back from the node, and the initial SELinux namespace is internally set to "init".

2. Transparently modify SELinux xattrs so that any non-initial namespace xattrs include the namespace name.

e.g. if the namespace name is "vm6", the "security.selinux" xattr is translated in the kernel to "security.selinux.ns.vm6" for disk read and write of xattrs.

This allows each SELinux namespace to independently manage its own xattr labels, transparently to applications. Namespaces only see their own xattrs, which are acted on by their own namespaced policies.

Note that the "init" namespace performs no translation for apps, it just uses regular old security.selinux xattrs.

Some examples:

[root at test]# cat /sys/fs/selinux/unshare
vm8

[root at test]# touch testfile

[root at test]# ls -Z testfile
-rw-r--r--. root root unconfined_u:object_r:unlabeled_t:s0 testfile

[root at test]# getfattr -n security.selinux testfile # file: testfile security.selinux="unconfined_u:object_r:unlabeled_t:s0"

  # restorecon -v testfile 
  restorecon reset /root/selinux/testfile context unconfined_u:object_r:unlabeled_t:s0->unconfined_u:object_r:admin_home_t:s0

[root at test]# getfattr -n security.selinux testfile # file: testfile security.selinux="unconfined_u:object_r:admin_home_t:s0"

[root at test]# chcon -t etc_t testfile

[root at test]# getfattr -n security.selinux testfile # file: testfile security.selinux="unconfined_u:object_r:etc_t:s0"

Ok, so this all looks pretty normal, but what's happening on disk is not.

Quoted 1 line

From the init namespace, I'll access the same file:

[root at test]# cat /sys/fs/selinux/unshare
init

[root at test]# ls -Z testfile
-rw-r--r--. root root system_u:object_r:unlabeled_t:s0 testfile

[root at test]# getfattr -n security.selinux testfile # file: testfile security.selinux="system_u:object_r:unlabeled_t:s0"

There is in fact no security.selinux xattr yet on this file as it was created in a different namespace and only initialized there. What you're seeing here is the default unlabeled label. Dumping out the xattrs shows what's on disk:

[root at test]# getfattr -d -m . testfile
# file: testfile security.selinux.ns.vm8="unconfined_u:object_r:etc_t:s0"

The xattr belonging to "vm8" is there but not being interpreted outside that namespace. Let's give it a proper label for the init ns:

  # restorecon -v testfile 
  restorecon reset /root/selinux/testfile context system_u:object_r:unlabeled_t:s0->system_u:object_r:admin_home_t:s0

[root at test]# getfattr -d -m . testfile
# file: testfile security.selinux="system_u:object_r:admin_home_t:s0" security.selinux.ns.vm8="unconfined_u:object_r:etc_t:s0"

[root at test]# ls -Z testfile
-rw-r--r--. root root system_u:object_r:admin_home_t:s0 testfile

And if you go into the vm8 namespace you'll see the label there is:

  [root at test]# echo  vm8 > /sys/fs/selinux/unshare 
  [root at test]# unshare -m -n
  [root at test]# umount /sys/fs/selinux && mount -t selinuxfs none /sys/fs/selinux && load_policy
  [root at test]# runcon unconfined_u:unconfined_r:unconfined_t:s0:c0.c1023 /bin/bash
  [root at test]# setenforce 1

[root at test]# ls -Z testfile
-rw-r--r--. root root unconfined_u:object_r:etc_t:s0 testfile

I hope that demonstrates the concept well enough: that there are zero changes for applications and the namespaced policy always uses xattrs belonging to that namespace.

The prototype code is far from complete, and also needs to implement support for listxattr and removexattr, as well as provide appropriate administrative access to raw (untranslated) xattrs, which you can currently see from any ns but not write at all. Other TODO items include accounting for dynamic xattr name size, nested policy enforcement, uniqueness of namespace names, and better audit support.

Comments?

Patch below...

---
From: James Morris <[email protected]>
Date: Mon, 30 Oct 2017 19:10:36 +1100
Subject: [PATCH] selinuxns: extend namespace support to security.selinux xattrs

Prototype code only.

Signed-off-by James Morris <[email protected]>

---
 fs/xattr.c                            | 12 ++++--
 include/linux/lsm_hooks.h             |  2 +
 include/linux/security.h              |  6 +++
 include/linux/xattr.h                 |  2 +-
 include/uapi/linux/xattr.h            |  3 ++
 security/integrity/evm/evm_crypto.c   |  2 +-
 security/integrity/ima/ima_appraise.c |  2 +-
 security/security.c                   |  6 +++
 security/selinux/hooks.c              | 75 ++++++++++++++++++++++++++++++-----
 security/selinux/include/security.h   |  7 +++-
 security/selinux/selinuxfs.c          | 63 ++++++++++++++++++-----------
 security/smack/smack_lsm.c            |  2 +-
 12 files changed, 142 insertions(+), 40 deletions(-)

fs/xattr.c +9 -3

index 4424f7f..d8107b7 100644
@@ -157,6 +157,7 @@
  *
  *  @dentry - object to perform setxattr on
  *  @name - xattr name to set
+ *  @nsname - namespaced xattr name, use instead of @name if set
  *  @value - value to set @name to
  *  @size - size of @value
  *  @flags - flags to pass into filesystem operations

@@ -168,7 +169,7 @@
  *  permission checks.
  */
 int __vfs_setxattr_noperm(struct dentry *dentry, const char *name,
-		const void *value, size_t size, int flags)
+		const char *nsname, const void *value, size_t size, int flags)
 {
 	struct inode *inode = dentry->d_inode;
 	int error = -EAGAIN;

@@ -178,7 +179,8 @@ int __vfs_setxattr_noperm(struct dentry *dentry, const char *name,
 	if (issec)
 		inode->i_flags &= ~S_NOSEC;
 	if (inode->i_opflags & IOP_XATTR) {
-		error = __vfs_setxattr(dentry, inode, name, value, size, flags);
+		error = __vfs_setxattr(dentry, inode, nsname?:name, value,
+					size, flags);
 		if (!error) {
 			fsnotify_xattr(dentry);
 			security_inode_post_setxattr(dentry, name, value,

@@ -211,6 +213,7 @@ int __vfs_setxattr_noperm(struct dentry *dentry, const char *name,
 {
 	struct inode *inode = dentry->d_inode;
 	int error;
+	char *nsname = NULL;
 
 	error = xattr_permission(inode, name, MAY_WRITE);
 	if (error)

@@ -221,8 +224,11 @@ int __vfs_setxattr_noperm(struct dentry *dentry, const char *name,
 	if (error)
 		goto out;
 
-	error = __vfs_setxattr_noperm(dentry, name, value, size, flags);
+	error = security_inode_translate_xattr_to_ns(name, &nsname);
+	if (error)
+		goto out;
 
+	error = __vfs_setxattr_noperm(dentry, name, nsname, value, size, flags);
 out:
 	inode_unlock(inode);
 	return error;

include/linux/lsm_hooks.h +2 -0

index c925812..e4eb43e 100644
@@ -1480,6 +1480,7 @@
 	void (*inode_getsecid)(struct inode *inode, u32 *secid);
 	int (*inode_copy_up)(struct dentry *src, struct cred **new);
 	int (*inode_copy_up_xattr)(const char *name);
+	int (*inode_translate_xattr_to_ns)(const char *name, char **tr);
 
 	int (*file_permission)(struct file *file, int mask);
 	int (*file_alloc_security)(struct file *file);

@@ -1760,6 +1761,7 @@ struct security_hook_heads {
 	struct list_head inode_getsecid;
 	struct list_head inode_copy_up;
 	struct list_head inode_copy_up_xattr;
+	struct list_head inode_translate_xattr_to_ns;
 	struct list_head file_permission;
 	struct list_head file_alloc_security;
 	struct list_head file_free_security;

include/linux/security.h +6 -0

index ce62659..64297e1 100644
@@ -302,6 +302,7 @@ void security_inode_post_setxattr(struct dentry *dentry, const char *name,
 void security_inode_getsecid(struct inode *inode, u32 *secid);
 int security_inode_copy_up(struct dentry *src, struct cred **new);
 int security_inode_copy_up_xattr(const char *name);
+int security_inode_translate_xattr_to_ns(const char *name, char **tr);
 int security_file_permission(struct file *file, int mask);
 int security_file_alloc(struct file *file);
 void security_file_free(struct file *file);

@@ -808,6 +809,11 @@ static inline int security_inode_copy_up_xattr(const char *name)
 	return -EOPNOTSUPP;
 }
 
+static inline int security_inode_translate_xattr_to_ns(const char *name, char **tr)
+{
+	return 0;
+}
+
 static inline int security_file_permission(struct file *file, int mask)
 {
 	return 0;

include/linux/xattr.h +1 -1

index e77605a..c25eb8a 100644
@@ -50,7 +50,7 @@ struct xattr {
 ssize_t vfs_getxattr(struct dentry *, const char *, void *, size_t);
 ssize_t vfs_listxattr(struct dentry *d, char *list, size_t size);
 int __vfs_setxattr(struct dentry *, struct inode *, const char *, const void *, size_t, int);
-int __vfs_setxattr_noperm(struct dentry *, const char *, const void *, size_t, int);
+int __vfs_setxattr_noperm(struct dentry *, const char *, const char *, const void *, size_t, int);
 int vfs_setxattr(struct dentry *, const char *, const void *, size_t, int);
 int __vfs_removexattr(struct dentry *, const char *);
 int vfs_removexattr(struct dentry *, const char *);

include/uapi/linux/xattr.h +3 -0

index 1590c49..528a5f7 100644
@@ -51,6 +51,9 @@
 
 #define XATTR_SELINUX_SUFFIX "selinux"
 #define XATTR_NAME_SELINUX XATTR_SECURITY_PREFIX XATTR_SELINUX_SUFFIX
+#define XATTR_SELINUX_NS_INFIX ".ns."
+#define XATTR_NAME_SELINUX_NS XATTR_NAME_SELINUX XATTR_SELINUX_NS_INFIX
+#define XATTR_NAME_SELINUX_NS_LEN (sizeof(XATTR_NAME_SELINUX_NS) - 1)
 
 #define XATTR_SMACK_SUFFIX "SMACK64"
 #define XATTR_SMACK_IPIN "SMACK64IPIN"

security/integrity/evm/evm_crypto.c +1 -1

index 1d32cd2..2249186 100644
@@ -260,7 +260,7 @@ int evm_update_evmxattr(struct dentry *dentry, const char *xattr_name,
 	if (rc == 0) {
 		xattr_data.type = EVM_XATTR_HMAC;
 		rc = __vfs_setxattr_noperm(dentry, XATTR_NAME_EVM,
-					   &xattr_data,
+					   NULL, &xattr_data,
 					   sizeof(xattr_data), 0);
 	} else if (rc == -ENODATA && (inode->i_opflags & IOP_XATTR)) {
 		rc = __vfs_removexattr(dentry, XATTR_NAME_EVM);

security/integrity/ima/ima_appraise.c +1 -1

index 809ba70..914cf5f 100644
@@ -71,7 +71,7 @@ static int ima_fix_xattr(struct dentry *dentry,
 		iint->ima_hash->xattr.ng.algo = algo;
 	}
 	rc = __vfs_setxattr_noperm(dentry, XATTR_NAME_IMA,
-				   &iint->ima_hash->xattr.data[offset],
+				   NULL, &iint->ima_hash->xattr.data[offset],
 				   (sizeof(iint->ima_hash->xattr) - offset) +
 				   iint->ima_hash->length, 0);
 	return rc;

security/security.c +6 -0

index 4bf0f57..7fce259 100644
@@ -856,6 +856,12 @@ int security_inode_copy_up_xattr(const char *name)
 }
 EXPORT_SYMBOL(security_inode_copy_up_xattr);
 
+int security_inode_translate_xattr_to_ns(const char *name, char **tr)
+{
+	return call_int_hook(inode_translate_xattr_to_ns, 0, name, tr);
+}
+EXPORT_SYMBOL(security_inode_translate_xattr_to_ns);
+
 int security_file_permission(struct file *file, int mask)
 {
 	int ret;

security/selinux/hooks.c +66 -9

index 3daad14..6b29d70 100644
@@ -646,6 +646,22 @@ static int selinux_is_sblabel_mnt(struct super_block *sb)
 		  !strcmp(sb->s_type->name, "cgroup2")));
 }
 
+static char *current_xattr_suffix(void)
+{
+	if (current_selinux_ns != init_selinux_ns)
+		return current_selinux_ns->xattr_name + XATTR_SECURITY_PREFIX_LEN;
+	else
+		return XATTR_SELINUX_SUFFIX;
+}
+
+static char *current_xattr_name(void)
+{
+	if (current_selinux_ns != init_selinux_ns)
+		return current_selinux_ns->xattr_name;
+	else
+		return XATTR_NAME_SELINUX;
+}
+
 static int sb_finish_set_opts(struct super_block *sb)
 {
 	struct superblock_security_struct *sbsec = superblock_security(sb);

@@ -654,6 +670,8 @@ static int sb_finish_set_opts(struct super_block *sb)
 	int rc = 0;
 
 	if (sbsec->behavior == SECURITY_FS_USE_XATTR) {
+		char *name;
+
 		/* Make sure that the xattr handler exists and that no
 		   error other than -ENODATA is returned by getxattr on
 		   the root directory.  -ENODATA is ok, as this may be

@@ -666,7 +684,9 @@ static int sb_finish_set_opts(struct super_block *sb)
 			goto out;
 		}
 
-		rc = __vfs_getxattr(root, root_inode, XATTR_NAME_SELINUX, NULL, 0);
+		name = current_xattr_name();
+
+		rc = __vfs_getxattr(root, root_inode, name, NULL, 0);
 		if (rc < 0 && rc != -ENODATA) {
 			if (rc == -EOPNOTSUPP)
 				printk(KERN_WARNING "SELinux: (dev %s, type "

@@ -1660,6 +1680,7 @@ static int inode_doinit_with_dentry(struct inode *inode,
 	char *context = NULL;
 	unsigned len = 0;
 	int rc = 0;
+	char *name;
 
 	if (isec->initialized == LABEL_INITIALIZED)
 		return 0;

@@ -1729,12 +1750,14 @@ static int inode_doinit_with_dentry(struct inode *inode,
 			goto out;
 		}
 		context[len] = '\0';
-		rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, context, len);
+
+		name = current_xattr_name();
+		rc = __vfs_getxattr(dentry, inode, name, context, len);
 		if (rc == -ERANGE) {
 			kfree(context);
 
 			/* Need a larger buffer.  Query for the right size. */
-			rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, NULL, 0);
+			rc = __vfs_getxattr(dentry, inode, name, NULL, 0);
 			if (rc < 0) {
 				dput(dentry);
 				goto out;

@@ -1747,7 +1770,7 @@ static int inode_doinit_with_dentry(struct inode *inode,
 				goto out;
 			}
 			context[len] = '\0';
-			rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, context, len);
+			rc = __vfs_getxattr(dentry, inode, name, context, len);
 		}
 		dput(dentry);
 		if (rc < 0) {

@@ -3167,7 +3190,7 @@ static int selinux_inode_init_security(struct inode *inode, struct inode *dir,
 		return -EOPNOTSUPP;
 
 	if (name)
-		*name = XATTR_SELINUX_SUFFIX;
+		*name = current_xattr_suffix();
 
 	if (value && len) {
 		rc = security_sid_to_context_force(current_selinux_ns, newsid,

@@ -3382,6 +3405,10 @@ static bool has_cap_mac_admin(bool audit)
 	return true;
 }
 
+/* TODO:
+ * - audit
+ * - handle raw namespaced xattrs
+ */
 static int selinux_inode_setxattr(struct dentry *dentry, const char *name,
 				  const void *value, size_t size, int flags)
 {

@@ -3392,8 +3419,12 @@ static int selinux_inode_setxattr(struct dentry *dentry, const char *name,
 	u32 newsid, sid = current_sid();
 	int rc = 0;
 
-	if (strcmp(name, XATTR_NAME_SELINUX))
+	if (strcmp(name, XATTR_NAME_SELINUX)) {
+		/* No raw namespaced xattrs, yet */
+		if (!strncmp(name, XATTR_NAME_SELINUX_NS, XATTR_NAME_SELINUX_NS_LEN))
+			return -EPERM;
 		return selinux_inode_setotherxattr(dentry, name);
+	}
 
 	sbsec = superblock_security(inode->i_sb);
 	if (!(sbsec->flags & SBLABEL_MNT))

@@ -3640,6 +3671,13 @@ static int selinux_inode_copy_up_xattr(const char *name)
 	return -EOPNOTSUPP;
 }
 
+static int selinux_inode_translate_xattr_to_ns(const char *name, char **tr)
+{
+	if(!strcmp(name, XATTR_NAME_SELINUX))
+		*tr = current_xattr_name();
+	return 0;
+}
+
 /* file security operations */
 
 static int selinux_revalidate_file_permission(struct file *file, int mask)

@@ -6430,10 +6468,11 @@ static int selinux_inode_notifysecctx(struct inode *inode, void *ctx, u32 ctxlen
 
 /*
  *	called with inode->i_mutex locked
+ *	TODO: namespace translation
  */
 static int selinux_inode_setsecctx(struct dentry *dentry, void *ctx, u32 ctxlen)
 {
-	return __vfs_setxattr_noperm(dentry, XATTR_NAME_SELINUX, ctx, ctxlen, 0);
+	return __vfs_setxattr_noperm(dentry, XATTR_NAME_SELINUX, NULL, ctx, ctxlen, 0);
 }
 
 static int selinux_inode_getsecctx(struct inode *inode, void **ctx, u32 *ctxlen)

@@ -6647,6 +6686,7 @@ static void selinux_ib_free_security(void *ib_sec)
 	LSM_HOOK_INIT(inode_getsecid, selinux_inode_getsecid),
 	LSM_HOOK_INIT(inode_copy_up, selinux_inode_copy_up),
 	LSM_HOOK_INIT(inode_copy_up_xattr, selinux_inode_copy_up_xattr),
+	LSM_HOOK_INIT(inode_translate_xattr_to_ns, selinux_inode_translate_xattr_to_ns),
 
 	LSM_HOOK_INIT(file_permission, selinux_file_permission),
 	LSM_HOOK_INIT(file_alloc_security, selinux_file_alloc_security),

@@ -6805,7 +6845,7 @@ static void selinux_ib_free_security(void *ib_sec)
 
 static void selinux_ns_free(struct work_struct *work);
 
-int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns)
+int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns, const char *name)
 {
 	struct selinux_ns *newns;
 	int rc;

@@ -6825,14 +6865,29 @@ int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns)
 	if (rc)
 		goto err;
 
+	newns->name = kstrdup(name, GFP_KERNEL);
+	if (!newns->name) {
+		rc = -ENOMEM;
+		goto err_avc;
+        }
+
+	newns->xattr_name = kasprintf(GFP_KERNEL, "%s.ns.%s", XATTR_NAME_SELINUX, name);
+	if (!newns->xattr_name) {
+		rc = -ENOMEM;
+		goto err_avc;
+	}
+
 	if (parent)
 		newns->parent = get_selinux_ns(parent);
 
 	*ns = newns;
 	return 0;
+err_avc:
+	selinux_avc_free(newns->avc);
 err:
 	selinux_ss_free(newns->ss);
 	kfree(newns);
+	kfree(newns->name);
 	return rc;
 }
 

@@ -6845,6 +6900,8 @@ static void selinux_ns_free(struct work_struct *work)
 		parent = ns->parent;
 		selinux_ss_free(ns->ss);
 		selinux_avc_free(ns->avc);
+		kfree(ns->name);
+		kfree(ns->xattr_name);
 		kfree(ns);
 		ns = parent;
 	} while (ns && refcount_dec_and_test(&ns->count));

@@ -6869,7 +6926,7 @@ static __init int selinux_init(void)
 
 	printk(KERN_INFO "SELinux:  Initializing.\n");
 
-	if (selinux_ns_create(NULL, &init_selinux_ns))
+	if (selinux_ns_create(NULL, &init_selinux_ns, SELINUX_NS_INIT_NAME))
 		panic("SELinux: Could not create initial namespace\n");
 
 	set_ns_enforcing(init_selinux_ns, selinux_enforcing_boot);

security/selinux/include/security.h +6 -1

index b80f9bd..f5f5a31 100644
@@ -92,6 +92,9 @@ enum {
 /* limitation of boundary depth  */
 #define POLICYDB_BOUNDS_MAXDEPTH	4
 
+/* Name of SELinux initial namespace */
+#define SELINUX_NS_INIT_NAME "init"
+
 struct selinux_avc;
 struct selinux_ss;
 

@@ -108,9 +111,11 @@ struct selinux_ns {
 	struct selinux_avc *avc;
 	struct selinux_ss *ss;
 	struct selinux_ns *parent;
+	char *name;
+	char *xattr_name;
 };
 
-int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns);
+int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns, const char *name);
 void __put_selinux_ns(struct selinux_ns *ns);
 
 int selinux_ss_create(struct selinux_ss **ss);

security/selinux/selinuxfs.c +40 -23

index 6c52d24..d190213 100644
@@ -334,9 +334,10 @@ static ssize_t sel_write_unshare(struct file *file, const char __user *buf,
 {
 	struct selinux_fs_info *fsi = file_inode(file)->i_sb->s_fs_info;
 	struct selinux_ns *ns = fsi->ns;
+	struct cred *cred;
+	struct task_security_struct *tsec;
 	char *page;
 	ssize_t length;
-	bool set;
 	int rc;
 
 	if (ns != current_selinux_ns)

@@ -359,30 +360,32 @@ static ssize_t sel_write_unshare(struct file *file, const char __user *buf,
 	if (IS_ERR(page))
 		return PTR_ERR(page);
 
-	length = -EINVAL;
-	if (kstrtobool(page, &set))
-		goto out;
+	/* strip any trailing newline */
+	if (page[strlen(page) - 1] == '\n')
+		page[strlen(page) - 1] = 0;
 
-	if (set) {
-		struct cred *cred = prepare_creds();
-		struct task_security_struct *tsec;
+	/* TODO: check for uniqueness! */
+	if (!strcmp(SELINUX_NS_INIT_NAME, page)) {
+		length = -EINVAL;
+		goto out;
+	}
 
-		if (!cred) {
-			length = -ENOMEM;
-			goto out;
-		}
-		tsec = cred->security;
-		if (selinux_ns_create(ns, &tsec->ns)) {
-			abort_creds(cred);
-			length = -ENOMEM;
-			goto out;
-		}
-		tsec->osid = tsec->sid = SECINITSID_KERNEL;
-		tsec->exec_sid = tsec->create_sid = tsec->keycreate_sid =
-			tsec->sockcreate_sid = SECSID_NULL;
-		tsec->parent_cred = get_current_cred();
-		commit_creds(cred);
+	cred = prepare_creds();
+	if (!cred) {
+		length = -ENOMEM;
+		goto out;
+	}
+	tsec = cred->security;
+	if (selinux_ns_create(ns, &tsec->ns, page)) {
+		abort_creds(cred);
+		length = -ENOMEM;
+		goto out;
 	}
+	tsec->osid = tsec->sid = SECINITSID_KERNEL;
+	tsec->exec_sid = tsec->create_sid = tsec->keycreate_sid =
+		tsec->sockcreate_sid = SECSID_NULL;
+	tsec->parent_cred = get_current_cred();
+	commit_creds(cred);
 
 	length = count;
 out:

@@ -390,8 +393,22 @@ static ssize_t sel_write_unshare(struct file *file, const char __user *buf,
 	return length;
 }
 
+static ssize_t sel_read_unshare(struct file *file, char __user *buf,
+				size_t count, loff_t *ppos)
+{
+	struct selinux_fs_info *fsi = file_inode(file)->i_sb->s_fs_info;
+	struct selinux_ns *ns = fsi->ns;
+	char *name = ns->name;
+
+	if (ns != current_selinux_ns)
+		return -EPERM;
+
+	return simple_read_from_buffer(buf, count, ppos, name, strlen(name));
+}
+
 static const struct file_operations sel_unshare_ops = {
 	.write		= sel_write_unshare,
+	.read		= sel_read_unshare,
 	.llseek		= generic_file_llseek,
 };
 

@@ -2021,7 +2038,7 @@ static int sel_fill_super(struct super_block *sb, void *data, int silent)
 		[SEL_POLICY] = {"policy", &sel_policy_ops, S_IRUGO},
 		[SEL_VALIDATE_TRANS] = {"validatetrans", &sel_transition_ops,
 					S_IWUGO},
-		[SEL_UNSHARE] = {"unshare", &sel_unshare_ops, 0222},
+		[SEL_UNSHARE] = {"unshare", &sel_unshare_ops, 0666},
 		/* last one */ {""}
 	};
 

security/smack/smack_lsm.c +0 -0

index 319add3..5ea841f 100644

This is a proof-of-concept patch to demonstrate an approach to supporting?
SELinux namespaces for security.selinux xattr labels.

This follows on from the experimental SELinux namespace code posted by?
Stephen: https://marc.info/?l=selinux&m=150696042210126&w=2

In the initial code, namespacing was only implemented for in-core inodes?
per this posting: https://marc.info/?l=selinux&m=150696324110943&w=2

In this patch, namespacing is extended to on-disk inode labels (xattrs),?
transparently to normal applications.

A summary of the approach is as follows:

1. Add a namespace "name" field to the SELinux namespace, which is? ???specified by writing to the selinuxfs unshare node (instead of the?
???current boolean value) during namespace creation.

???e.g. if the namespace name is "vm8", run:?

???# echo vm8 > /sys/fs/selinux/unshare

???followed by the remaining steps from the original code.

???This value can also be read back from the node, and the initial SELinux?
???namespace is internally set to "init".

???
2. Transparently modify SELinux xattrs so that any non-initial namespace?
???xattrs include the namespace name.?

???e.g. if the namespace name is "vm6", the "security.selinux" xattr is?
???translated in the kernel to "security.selinux.ns.vm6" for disk read and
???write of xattrs.

???This allows each SELinux namespace to independently manage its own?
???xattr labels, transparently to applications. Namespaces only see their?
???own xattrs, which are acted on by their own namespaced policies.

???Note that the "init" namespace performs no translation for apps, it?
???just uses regular old security.selinux xattrs.

Some examples:

? [root at test]# cat /sys/fs/selinux/unshare?
? vm8

? [root at test]# touch testfile

? [root at test]# ls -Z testfile?
? -rw-r--r--. root root unconfined_u:object_r:unlabeled_t:s0 testfile

? [root at test]# getfattr -n security.selinux testfile? ? # file: testfile
? security.selinux="unconfined_u:object_r:unlabeled_t:s0"

? # restorecon -v testfile?
? restorecon reset /root/selinux/testfile context
unconfined_u:object_r:unlabeled_t:s0-

Quoted 1 line

unconfined_u:object_r:admin_home_t:s0

? [root at test]# getfattr -n security.selinux testfile? ? # file: testfile
? security.selinux="unconfined_u:object_r:admin_home_t:s0"

? [root at test]# chcon -t etc_t testfile?

? [root at test]# getfattr -n security.selinux testfile? ? # file: testfile
? security.selinux="unconfined_u:object_r:etc_t:s0"

Ok, so this all looks pretty normal, but what's happening on disk is not.??

Quoted 1 line

From the init namespace, I'll access the same file:

? [root at test]# cat /sys/fs/selinux/unshare?
? init

? [root at test]# ls -Z testfile?
? -rw-r--r--. root root system_u:object_r:unlabeled_t:s0 testfile

? [root at test]# getfattr -n security.selinux testfile? ? # file: testfile
? security.selinux="system_u:object_r:unlabeled_t:s0"

There is in fact no security.selinux xattr yet on this file as it was?
created in a different namespace and only initialized there.??What you're?
seeing here is the default unlabeled label.??Dumping out the xattrs shows?
what's on disk:

? [root at test]# getfattr -d -m . testfile?
? # file: testfile
? security.selinux.ns.vm8="unconfined_u:object_r:etc_t:s0"

The xattr belonging to "vm8" is there but not being interpreted outside?
that namespace.??Let's give it a proper label for the init ns:

? # restorecon -v testfile?
? restorecon reset /root/selinux/testfile context system_u:object_r:unlabeled_t:s0->system_u:object_r:admin_home_t:s0

? [root at test]# getfattr -d -m . testfile?
? # file: testfile
? security.selinux="system_u:object_r:admin_home_t:s0" ? security.selinux.ns.vm8="unconfined_u:object_r:etc_t:s0"

? [root at test]# ls -Z testfile?
? -rw-r--r--. root root system_u:object_r:admin_home_t:s0 testfile

And if you go into the vm8 namespace you'll see the label there is:

? [root at test]# echo??vm8 > /sys/fs/selinux/unshare?
? [root at test]# unshare -m -n
? [root at test]# umount /sys/fs/selinux && mount -t selinuxfs none /sys/fs/selinux && load_policy
? [root at test]# runcon unconfined_u:unconfined_r:unconfined_t:s0:c0.c1023 /bin/bash ? [root at test]# setenforce 1

? [root at test]# ls -Z testfile?
? -rw-r--r--. root root unconfined_u:object_r:etc_t:s0???testfile

I hope that demonstrates the concept well enough: that there are zero?
changes for applications and the namespaced policy always uses xattrs?
belonging to that namespace.

The prototype code is far from complete, and also needs to implement? support for listxattr and removexattr, as well as provide appropriate?
administrative access to raw (untranslated) xattrs, which you can? currently see from any ns but not write at all.??Other TODO items include?
accounting for dynamic xattr name size, nested policy enforcement,? uniqueness of namespace names, and better audit support.

Comments?

Patch below...

---
From: James Morris <[email protected]>
Date: Mon, 30 Oct 2017 19:10:36 +1100
Subject: [PATCH] selinuxns: extend namespace support to security.selinux xattrs

Prototype code only.

Signed-off-by James Morris <[email protected]>

---

?fs/xattr.c????????????????????????????| 12 ++++--
?include/linux/lsm_hooks.h?????????????|??2 +
?include/linux/security.h??????????????|??6 +++
?include/linux/xattr.h?????????????????|??2 +-
?include/uapi/linux/xattr.h????????????|??3 ++
?security/integrity/evm/evm_crypto.c???|??2 +-
?security/integrity/ima/ima_appraise.c |??2 +-
?security/security.c???????????????????|??6 +++
?security/selinux/hooks.c??????????????| 75

++++++++++++++++++++++++++++++-----
?security/selinux/include/security.h???|??7 +++- ?security/selinux/selinuxfs.c??????????| 63 ++++++++++++++++++-------
----
?security/smack/smack_lsm.c????????????|??2 +-
?12 files changed, 142 insertions(+), 40 deletions(-)

fs/xattr.c +0 -0

index 4424f7f..d8107b7 100644
@@ -157,6 +157,7 @@

? *
? *??@dentry - object to perform setxattr on
? *??@name - xattr name to set
+ *??@nsname - namespaced xattr name, use instead of @name if set ? *??@value - value to set @name to
? *??@size - size of @value
? *??@flags - flags to pass into filesystem operations

fs/xattr.c +0 -0

@@ -168,7 +169,7 @@

? *??permission checks.
? */
?int __vfs_setxattr_noperm(struct dentry *dentry, const char *name,

fs/xattr.c +1 -1

-		const void *value, size_t size, int flags)
+		const char *nsname, const void *value, size_t size,

int flags)
?{
? struct inode *inode = dentry->d_inode;
? int error = -EAGAIN;

fs/xattr.c +2 -1

@@ -178,7 +179,8 @@ int __vfs_setxattr_noperm(struct dentry *dentry,
 const char *name,
 ?	if (issec)
 ?		inode->i_flags &= ~S_NOSEC;
 ?	if (inode->i_opflags & IOP_XATTR) {
-		error = __vfs_setxattr(dentry, inode, name, value,
 size, flags);
+		error = __vfs_setxattr(dentry, inode, nsname?:name,
 value,
+					size, flags);

? if (!error) {
? fsnotify_xattr(dentry);
? security_inode_post_setxattr(dentry, name,
value,

fs/xattr.c +1 -0

@@ -211,6 +213,7 @@ int __vfs_setxattr_noperm(struct dentry *dentry,
 const char *name,
 ?{
 ?	struct inode *inode = dentry->d_inode;
 ?	int error;
+	char *nsname = NULL;

?
? error = xattr_permission(inode, name, MAY_WRITE);
? if (error)

fs/xattr.c +0 -0

@@ -221,8 +224,11 @@ int __vfs_setxattr_noperm(struct dentry *dentry,
 const char *name,
 ?	if (error)
 ?		goto out;

?

fs/xattr.c +0 -1

-	error = __vfs_setxattr_noperm(dentry, name, value, size,

flags);

fs/xattr.c +3 -0

+	error = security_inode_translate_xattr_to_ns(name, &nsname);
+	if (error)
+		goto out;

?

fs/xattr.c +1 -0

+	error = __vfs_setxattr_noperm(dentry, name, nsname, value,

size, flags);
?out:
? inode_unlock(inode);
? return error;

include/linux/lsm_hooks.h +1 -0

index c925812..e4eb43e 100644
@@ -1480,6 +1480,7 @@
 ?	void (*inode_getsecid)(struct inode *inode, u32 *secid);
 ?	int (*inode_copy_up)(struct dentry *src, struct cred **new);
 ?	int (*inode_copy_up_xattr)(const char *name);
+	int (*inode_translate_xattr_to_ns)(const char *name, char
 **tr);

?
? int (*file_permission)(struct file *file, int mask);
? int (*file_alloc_security)(struct file *file);

include/linux/lsm_hooks.h +1 -0

@@ -1760,6 +1761,7 @@ struct security_hook_heads {
 ?	struct list_head inode_getsecid;
 ?	struct list_head inode_copy_up;
 ?	struct list_head inode_copy_up_xattr;
+	struct list_head inode_translate_xattr_to_ns;
 ?	struct list_head file_permission;
 ?	struct list_head file_alloc_security;
 ?	struct list_head file_free_security;

include/linux/security.h +1 -0

index ce62659..64297e1 100644
@@ -302,6 +302,7 @@ void security_inode_post_setxattr(struct dentry
 *dentry, const char *name,
 ?void security_inode_getsecid(struct inode *inode, u32 *secid);
 ?int security_inode_copy_up(struct dentry *src, struct cred **new);
 ?int security_inode_copy_up_xattr(const char *name);
+int security_inode_translate_xattr_to_ns(const char *name, char
 **tr);
 ?int security_file_permission(struct file *file, int mask);

?int security_file_alloc(struct file *file);
?void security_file_free(struct file *file);

include/linux/security.h +0 -0

@@ -808,6 +809,11 @@ static inline int
 security_inode_copy_up_xattr(const char *name)
 ?	return -EOPNOTSUPP;
 ?}

?
+static inline int security_inode_translate_xattr_to_ns(const char *name, char **tr)

include/linux/security.h +4 -0

+{
+	return 0;
+}
+

?static inline int security_file_permission(struct file *file, int mask)
?{
? return 0;

include/linux/xattr.h +1 -1

index e77605a..c25eb8a 100644
@@ -50,7 +50,7 @@ struct xattr {
 ?ssize_t vfs_getxattr(struct dentry *, const char *, void *, size_t);
 ?ssize_t vfs_listxattr(struct dentry *d, char *list, size_t size);
 ?int __vfs_setxattr(struct dentry *, struct inode *, const char *,
 const void *, size_t, int);
-int __vfs_setxattr_noperm(struct dentry *, const char *, const void
 *, size_t, int);
+int __vfs_setxattr_noperm(struct dentry *, const char *, const char
 *, const void *, size_t, int);

?int vfs_setxattr(struct dentry *, const char *, const void *, size_t, int);
?int __vfs_removexattr(struct dentry *, const char *); ?int vfs_removexattr(struct dentry *, const char *);

include/uapi/linux/xattr.h +0 -0

index 1590c49..528a5f7 100644
@@ -51,6 +51,9 @@

?
?#define XATTR_SELINUX_SUFFIX "selinux"
?#define XATTR_NAME_SELINUX XATTR_SECURITY_PREFIX
XATTR_SELINUX_SUFFIX

include/uapi/linux/xattr.h +2 -0

+#define XATTR_SELINUX_NS_INFIX ".ns."
+#define XATTR_NAME_SELINUX_NS XATTR_NAME_SELINUX

XATTR_SELINUX_NS_INFIX
+#define XATTR_NAME_SELINUX_NS_LEN (sizeof(XATTR_NAME_SELINUX_NS) -
1)
?
?#define XATTR_SMACK_SUFFIX "SMACK64"
?#define XATTR_SMACK_IPIN "SMACK64IPIN"

security/integrity/evm/evm_crypto.c +1 -1

index 1d32cd2..2249186 100644
@@ -260,7 +260,7 @@ int evm_update_evmxattr(struct dentry *dentry,
 const char *xattr_name,
 ?	if (rc == 0) {
 ?		xattr_data.type = EVM_XATTR_HMAC;
 ?		rc = __vfs_setxattr_noperm(dentry, XATTR_NAME_EVM,
-					???&xattr_data,
+					???NULL, &xattr_data,
 ?					???sizeof(xattr_data), 0);
 ?	} else if (rc == -ENODATA && (inode->i_opflags & IOP_XATTR))

{
? rc = __vfs_removexattr(dentry, XATTR_NAME_EVM);

security/integrity/ima/ima_appraise.c +1 -1

index 809ba70..914cf5f 100644
@@ -71,7 +71,7 @@ static int ima_fix_xattr(struct dentry *dentry,
 ?		iint->ima_hash->xattr.ng.algo = algo;
 ?	}
 ?	rc = __vfs_setxattr_noperm(dentry, XATTR_NAME_IMA,
-				???&iint->ima_hash-
 >xattr.data[offset],
+				???NULL, &iint->ima_hash-
 >xattr.data[offset],

? ???(sizeof(iint->ima_hash->xattr) -
offset) +
? ???iint->ima_hash->length, 0);
? return rc;

security/security.c +0 -0

index 4bf0f57..7fce259 100644
@@ -856,6 +856,12 @@ int security_inode_copy_up_xattr(const char
 *name)
 ?}
 ?EXPORT_SYMBOL(security_inode_copy_up_xattr);

?
+int security_inode_translate_xattr_to_ns(const char *name, char **tr)

security/security.c +2 -0

+{
+	return call_int_hook(inode_translate_xattr_to_ns, 0, name,

tr);

security/security.c +3 -0

+}
+EXPORT_SYMBOL(security_inode_translate_xattr_to_ns);
+

?int security_file_permission(struct file *file, int mask) ?{
? int ret;

security/selinux/hooks.c +0 -0

index 3daad14..6b29d70 100644
@@ -646,6 +646,22 @@ static int selinux_is_sblabel_mnt(struct
 super_block *sb)
 ?		??!strcmp(sb->s_type->name, "cgroup2")));
 ?}

?

security/selinux/hooks.c +4 -0

+static char *current_xattr_suffix(void)
+{
+	if (current_selinux_ns != init_selinux_ns)
+		return current_selinux_ns->xattr_name +

XATTR_SECURITY_PREFIX_LEN;

security/selinux/hooks.c +1 -0

+	else

+
?static int sb_finish_set_opts(struct super_block *sb) ?{
? struct superblock_security_struct *sbsec =
superblock_security(sb);

security/selinux/hooks.c +0 -0

@@ -654,6 +670,8 @@ static int sb_finish_set_opts(struct super_block
 *sb)
 ?	int rc = 0;

?
? if (sbsec->behavior == SECURITY_FS_USE_XATTR) {

security/selinux/hooks.c +2 -0

+		char *name;
+

? /* Make sure that the xattr handler exists and that
no
? ???error other than -ENODATA is returned by getxattr
on
? ???the root directory.??-ENODATA is ok, as this may
be

security/selinux/hooks.c +0 -0

@@ -666,7 +684,9 @@ static int sb_finish_set_opts(struct super_block
 *sb)
 ?			goto out;
 ?		}

?

security/selinux/hooks.c +0 -1

-		rc = __vfs_getxattr(root, root_inode,

XATTR_NAME_SELINUX, NULL, 0);

security/selinux/hooks.c +3 -0

+		name = current_xattr_name();
+
+		rc = __vfs_getxattr(root, root_inode, name, NULL,

0);
? if (rc < 0 && rc != -ENODATA) {
? if (rc == -EOPNOTSUPP)
? printk(KERN_WARNING "SELinux: (dev
%s, type "

security/selinux/hooks.c +1 -0

@@ -1660,6 +1680,7 @@ static int inode_doinit_with_dentry(struct
 inode *inode,
 ?	char *context = NULL;
 ?	unsigned len = 0;
 ?	int rc = 0;
+	char *name;

?
? if (isec->initialized == LABEL_INITIALIZED)
? return 0;

security/selinux/hooks.c +3 -1

@@ -1729,12 +1750,14 @@ static int inode_doinit_with_dentry(struct
 inode *inode,
 ?			goto out;
 ?		}
 ?		context[len] = '\0';
-		rc = __vfs_getxattr(dentry, inode,
 XATTR_NAME_SELINUX, context, len);
+
+		name = current_xattr_name();
+		rc = __vfs_getxattr(dentry, inode, name, context,
 len);
 ?		if (rc == -ERANGE) {
 ?			kfree(context);

?
? /* Need a larger buffer.??Query for the
right size. */

security/selinux/hooks.c +0 -1

-			rc = __vfs_getxattr(dentry, inode,

XATTR_NAME_SELINUX, NULL, 0);

security/selinux/hooks.c +1 -0

+			rc = __vfs_getxattr(dentry, inode, name,

NULL, 0);
? if (rc < 0) {
? dput(dentry);
? goto out;

security/selinux/hooks.c +1 -1

@@ -1747,7 +1770,7 @@ static int inode_doinit_with_dentry(struct
 inode *inode,
 ?				goto out;
 ?			}
 ?			context[len] = '\0';
-			rc = __vfs_getxattr(dentry, inode,
 XATTR_NAME_SELINUX, context, len);
+			rc = __vfs_getxattr(dentry, inode, name,
 context, len);

? }
? dput(dentry);
? if (rc < 0) {

security/selinux/hooks.c +0 -0

@@ -3167,7 +3190,7 @@ static int selinux_inode_init_security(struct
 inode *inode, struct inode *dir,
 ?		return -EOPNOTSUPP;

?
? if (name)

security/selinux/hooks.c +1 -1

-		*name = XATTR_SELINUX_SUFFIX;
+		*name = current_xattr_suffix();

?
? if (value && len) {
? rc =
security_sid_to_context_force(current_selinux_ns, newsid,

security/selinux/hooks.c +0 -0

@@ -3382,6 +3405,10 @@ static bool has_cap_mac_admin(bool audit)
 ?	return true;
 ?}

?
+/* TODO:
+ * - audit
+ * - handle raw namespaced xattrs
+ */
?static int selinux_inode_setxattr(struct dentry *dentry, const char *name,
? ??const void *value, size_t size,
int flags)
?{

security/selinux/hooks.c +0 -0

@@ -3392,8 +3419,12 @@ static int selinux_inode_setxattr(struct
 dentry *dentry, const char *name,
 ?	u32 newsid, sid = current_sid();
 ?	int rc = 0;

?

security/selinux/hooks.c +3 -1

-	if (strcmp(name, XATTR_NAME_SELINUX))
+	if (strcmp(name, XATTR_NAME_SELINUX)) {
+		/* No raw namespaced xattrs, yet */
+		if (!strncmp(name, XATTR_NAME_SELINUX_NS,

XATTR_NAME_SELINUX_NS_LEN))

security/selinux/hooks.c +1 -0

+			return -EPERM;

? return selinux_inode_setotherxattr(dentry, name);

security/selinux/hooks.c +1 -0

+	}

?
? sbsec = superblock_security(inode->i_sb);
? if (!(sbsec->flags & SBLABEL_MNT))

security/selinux/hooks.c +0 -0

@@ -3640,6 +3671,13 @@ static int selinux_inode_copy_up_xattr(const
 char *name)
 ?	return -EOPNOTSUPP;
 ?}

?
+static int selinux_inode_translate_xattr_to_ns(const char *name, char **tr)

security/selinux/hooks.c +6 -0

+{
+	if(!strcmp(name, XATTR_NAME_SELINUX))
+		*tr = current_xattr_name();
+	return 0;
+}
+

?/* file security operations */
?
?static int selinux_revalidate_file_permission(struct file *file, int mask)

security/selinux/hooks.c +0 -0

@@ -6430,10 +6468,11 @@ static int selinux_inode_notifysecctx(struct

inode *inode, void *ctx, u32 ctxlen
?
?/*
? * called with inode->i_mutex locked
+ * TODO: namespace translation
? */
?static int selinux_inode_setsecctx(struct dentry *dentry, void *ctx, u32 ctxlen)
?{

security/selinux/hooks.c +0 -1

-	return __vfs_setxattr_noperm(dentry, XATTR_NAME_SELINUX,

ctx, ctxlen, 0);

security/selinux/hooks.c +1 -0

+	return __vfs_setxattr_noperm(dentry, XATTR_NAME_SELINUX,

NULL, ctx, ctxlen, 0);
?}
?
?static int selinux_inode_getsecctx(struct inode *inode, void **ctx, u32 *ctxlen)

security/selinux/hooks.c +1 -0

@@ -6647,6 +6686,7 @@ static void selinux_ib_free_security(void
 *ib_sec)
 ?	LSM_HOOK_INIT(inode_getsecid, selinux_inode_getsecid),
 ?	LSM_HOOK_INIT(inode_copy_up, selinux_inode_copy_up),
 ?	LSM_HOOK_INIT(inode_copy_up_xattr,
 selinux_inode_copy_up_xattr),
+	LSM_HOOK_INIT(inode_translate_xattr_to_ns,
 selinux_inode_translate_xattr_to_ns),

?
? LSM_HOOK_INIT(file_permission, selinux_file_permission),
? LSM_HOOK_INIT(file_alloc_security,
selinux_file_alloc_security),

security/selinux/hooks.c +0 -0

@@ -6805,7 +6845,7 @@ static void selinux_ib_free_security(void
 *ib_sec)

?
?static void selinux_ns_free(struct work_struct *work); ?
-int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns)
+int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns, const char *name)
?{
? struct selinux_ns *newns;
? int rc;

security/selinux/hooks.c +0 -0

@@ -6825,14 +6865,29 @@ int selinux_ns_create(struct selinux_ns
 *parent, struct selinux_ns **ns)
 ?	if (rc)
 ?		goto err;

?

security/selinux/hooks.c +7 -0

+	newns->name = kstrdup(name, GFP_KERNEL);
+	if (!newns->name) {
+		rc = -ENOMEM;
+		goto err_avc;
+????????}
+
+	newns->xattr_name = kasprintf(GFP_KERNEL, "%s.ns.%s",

XATTR_NAME_SELINUX, name);

security/selinux/hooks.c +5 -0

+	if (!newns->xattr_name) {
+		rc = -ENOMEM;
+		goto err_avc;
+	}
+

? if (parent)
? newns->parent = get_selinux_ns(parent);
?
? *ns = newns;
? return 0;

security/selinux/hooks.c +2 -0

+err_avc:
+	selinux_avc_free(newns->avc);

?err:
? selinux_ss_free(newns->ss);
? kfree(newns);

security/selinux/hooks.c +1 -0

+	kfree(newns->name);

? return rc;
?}
?

security/selinux/hooks.c +2 -0

@@ -6845,6 +6900,8 @@ static void selinux_ns_free(struct work_struct
 *work)
 ?		parent = ns->parent;
 ?		selinux_ss_free(ns->ss);
 ?		selinux_avc_free(ns->avc);
+		kfree(ns->name);
+		kfree(ns->xattr_name);
 ?		kfree(ns);
 ?		ns = parent;

? } while (ns && refcount_dec_and_test(&ns->count));

security/selinux/hooks.c +0 -0

@@ -6869,7 +6926,7 @@ static __init int selinux_init(void)

?
? printk(KERN_INFO "SELinux:??Initializing.\n");
?

security/selinux/hooks.c +1 -1

-	if (selinux_ns_create(NULL, &init_selinux_ns))
+	if (selinux_ns_create(NULL, &init_selinux_ns,

SELINUX_NS_INIT_NAME))
? panic("SELinux: Could not create initial
namespace\n");
?
? set_ns_enforcing(init_selinux_ns, selinux_enforcing_boot);

security/selinux/include/security.h +0 -0

index b80f9bd..f5f5a31 100644
@@ -92,6 +92,9 @@ enum {

?/* limitation of boundary depth??*/
?#define POLICYDB_BOUNDS_MAXDEPTH 4
?

security/selinux/include/security.h +3 -0

+/* Name of SELinux initial namespace */
+#define SELINUX_NS_INIT_NAME "init"
+

?struct selinux_avc;
?struct selinux_ss;
?

security/selinux/include/security.h +2 -0

@@ -108,9 +111,11 @@ struct selinux_ns {
 ?	struct selinux_avc *avc;
 ?	struct selinux_ss *ss;
 ?	struct selinux_ns *parent;
+	char *name;
+	char *xattr_name;
 ?};

?
-int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns);
+int selinux_ns_create(struct selinux_ns *parent, struct selinux_ns **ns, const char *name);
?void __put_selinux_ns(struct selinux_ns *ns);
?
?int selinux_ss_create(struct selinux_ss **ss);

security/selinux/selinuxfs.c +0 -0

index 6c52d24..d190213 100644
@@ -334,9 +334,10 @@ static ssize_t sel_write_unshare(struct file
 *file, const char __user *buf,
 ?{

? struct selinux_fs_info *fsi = file_inode(file)->i_sb-

Quoted 1 line

s_fs_info;

? struct selinux_ns *ns = fsi->ns;

security/selinux/selinuxfs.c +2 -0

+	struct cred *cred;
+	struct task_security_struct *tsec;

? char *page;
? ssize_t length;

security/selinux/selinuxfs.c +0 -1

-	bool set;

? int rc;
?
? if (ns != current_selinux_ns)

security/selinux/selinuxfs.c +0 -0

@@ -359,30 +360,32 @@ static ssize_t sel_write_unshare(struct file
 *file, const char __user *buf,
 ?	if (IS_ERR(page))
 ?		return PTR_ERR(page);

?

security/selinux/selinuxfs.c +3 -3

-	length = -EINVAL;
-	if (kstrtobool(page, &set))
-		goto out;
+	/* strip any trailing newline */
+	if (page[strlen(page) - 1] == '\n')
+		page[strlen(page) - 1] = 0;

?

security/selinux/selinuxfs.c +5 -3

-	if (set) {
-		struct cred *cred = prepare_creds();
-		struct task_security_struct *tsec;
+	/* TODO: check for uniqueness! */
+	if (!strcmp(SELINUX_NS_INIT_NAME, page)) {
+		length = -EINVAL;
+		goto out;
+	}

?

security/selinux/selinuxfs.c +0 -12

-		if (!cred) {
-			length = -ENOMEM;
-			goto out;
-		}
-		tsec = cred->security;
-		if (selinux_ns_create(ns, &tsec->ns)) {
-			abort_creds(cred);
-			length = -ENOMEM;
-			goto out;
-		}
-		tsec->osid = tsec->sid = SECINITSID_KERNEL;
-		tsec->exec_sid = tsec->create_sid = tsec-

Quoted 1 line

keycreate_sid =

security/selinux/selinuxfs.c +10 -3

-			tsec->sockcreate_sid = SECSID_NULL;
-		tsec->parent_cred = get_current_cred();
-		commit_creds(cred);
+	cred = prepare_creds();
+	if (!cred) {
+		length = -ENOMEM;
+		goto out;
+	}
+	tsec = cred->security;
+	if (selinux_ns_create(ns, &tsec->ns, page)) {
+		abort_creds(cred);
+		length = -ENOMEM;
+		goto out;

? }

security/selinux/selinuxfs.c +5 -0

+	tsec->osid = tsec->sid = SECINITSID_KERNEL;
+	tsec->exec_sid = tsec->create_sid = tsec->keycreate_sid =
+		tsec->sockcreate_sid = SECSID_NULL;
+	tsec->parent_cred = get_current_cred();
+	commit_creds(cred);

?
? length = count;
?out:

security/selinux/selinuxfs.c +0 -0

@@ -390,8 +393,22 @@ static ssize_t sel_write_unshare(struct file
 *file, const char __user *buf,
 ?	return length;
 ?}

?

security/selinux/selinuxfs.c +4 -0

+static ssize_t sel_read_unshare(struct file *file, char __user *buf,
+				size_t count, loff_t *ppos)
+{
+	struct selinux_fs_info *fsi = file_inode(file)->i_sb-

Quoted 1 line

s_fs_info;

security/selinux/selinuxfs.c +7 -0

+	struct selinux_ns *ns = fsi->ns;
+	char *name = ns->name;
+
+	if (ns != current_selinux_ns)
+		return -EPERM;
+
+	return simple_read_from_buffer(buf, count, ppos, name,

strlen(name));

security/selinux/selinuxfs.c +2 -0

+}
+

?static const struct file_operations sel_unshare_ops = {
? .write = sel_write_unshare,

security/selinux/selinuxfs.c +1 -0

+	.read		= sel_read_unshare,

? .llseek = generic_file_llseek,
?};
?

security/selinux/selinuxfs.c +1 -1

@@ -2021,7 +2038,7 @@ static int sel_fill_super(struct super_block
 *sb, void *data, int silent)
 ?		[SEL_POLICY] = {"policy", &sel_policy_ops, S_IRUGO},
 ?		[SEL_VALIDATE_TRANS] = {"validatetrans",
 &sel_transition_ops,
 ?					S_IWUGO},
-		[SEL_UNSHARE] = {"unshare", &sel_unshare_ops, 0222},
+		[SEL_UNSHARE] = {"unshare", &sel_unshare_ops, 0666},
 ?		/* last one */ {""}

? };
?

security/smack/smack_lsm.c +0 -0

index 319add3..5ea841f 100644
@@ -4591,7 +4591,7 @@ static int smack_inode_notifysecctx(struct
 inode *inode, void *ctx, u32 ctxlen)

?
?static int smack_inode_setsecctx(struct dentry *dentry, void *ctx, u32 ctxlen)
?{

security/smack/smack_lsm.c +0 -1

-	return __vfs_setxattr_noperm(dentry, XATTR_NAME_SMACK, ctx,

ctxlen, 0);

security/smack/smack_lsm.c +1 -0

+	return __vfs_setxattr_noperm(dentry, XATTR_NAME_SMACK, NULL,

ctx, ctxlen, 0);
?}
?
?static int smack_inode_getsecctx(struct inode *inode, void **ctx, u32 *ctxlen)

Thanks, interesting approach. One drawback is that it doesn't presently support any form of inheritance of labels from the parent namespace, so files that are shared read-only from the init namespace will show up as unlabeled in the child namespace until they are assigned the namespaced attributes. This for example breaks running the selinux-testsuite with this patch applied (unless perhaps you run restorecon -R / after unsharing); otherwise just trying to invoke /usr/bin/runcon will fail since it is unlabeled in the child. It seems like we should provide some form of inheritance from the parent when there is no xattr for the namespace itself.

Another potential concern is that files created in a non-init namespace are left completely unlabeled in the init namespace (or in any parent).
As long as access to unlabeled is tightly controlled, this might
not be a problem, but I'm not sure that's guaranteed by the refpolicy or Fedora/RHEL policies. We might want to initialize an xattr at every level of the namespace hierarchy when a new file is created based on the process and parent directory labels and policy at that level. Otherwise, we lose all provenance information for the file outside of the namespace.

For example, suppose I want to leak information out of
my category set; I unshare and create files with the information, and they appear in the init namespace with no categories.

Quoted 10 lines · Stephen Smalley

Thanks, interesting approach. One drawback is that it doesn't presently
support any form of inheritance of labels from the parent namespace, so
files that are shared read-only from the init namespace will show up as
unlabeled in the child namespace until they are assigned the namespaced
attributes.??This for example breaks running the selinux-testsuite with

Quoted 8 lines

this patch applied (unless perhaps you run restorecon -R / after unsharing); otherwise just trying to invoke /usr/bin/runcon will fail
since it is unlabeled in the child.??It seems like we should provide
some form of inheritance from the parent when there is no xattr for the
namespace itself.

I was assuming that practical use of this would involve doing a filesystem?
relabel under the newly loaded policy, on first instantiation at least.

We could try adding an selinuxfs node to specify default handling of? unlabeled files in a child namespace, and write to that after mounting?
selinuxfs in that namespace.

e.g. echo inherit > /sys/fs/selinux/parent_ns_labels

or something.

Quoted 17 lines

Another potential concern is that files created in a non-init namespace
are left completely unlabeled in the init namespace (or in any parent).
????As long as access to unlabeled is tightly controlled, this might
not be a problem, but I'm not sure that's guaranteed by the refpolicy
or Fedora/RHEL policies.??We might want to initialize an xattr at every
level of the namespace hierarchy when a new file is created based on
the process and parent directory labels and policy at that level.? Otherwise, we lose all provenance information for the file outside of
the namespace.?

Ok.

Quoted 4 lines

For example, suppose I want to leak information out of my category set; I unshare and create files with the information, and
they appear in the init namespace with no categories.

Nice :)

Quoted 23 lines · James Morris

Quoted 12 lines · Stephen Smalley

Thanks, interesting approach. One drawback is that it doesn't presently
support any form of inheritance of labels from the parent namespace, so
files that are shared read-only from the init namespace will show up as
unlabeled in the child namespace until they are assigned the namespaced
attributes.??This for example breaks running the selinux- testsuite
with
this patch applied (unless perhaps you run restorecon -R / after

Quoted 8 lines

unsharing); otherwise just trying to invoke /usr/bin/runcon will fail
since it is unlabeled in the child.??It seems like we should provide
some form of inheritance from the parent when there is no xattr for
the
namespace itself.

Quoted 4 lines

I was assuming that practical use of this would involve doing a filesystem?
relabel under the newly loaded policy, on first instantiation at least.

I think that would be problematic, e.g. if you want to share /usr read-
only to all of the containers or similar, then they cannot relabel it themselves and even if you provided a way to set the xattrs from the init namespace, in the case where you are using the same or very similar policies, it seems wasteful to set the same xattr values for N
xattr names for N containers.??So I think inheritance support will be necessary.??This will be easier if we make the xattr names themselves hierarchical (e.g. if vm8 writes vm1 to /sys/fs/selinux/unshare, then the child xattr name would be security.selinux.ns.vm8.vm1, and we would
inherit from either security.selinux.ns.vm8 or security.selinux if security.selinux.ns.vm8.vm1 is not set.)

Quoted 36 lines

We could try adding an selinuxfs node to specify default handling of?
unlabeled files in a child namespace, and write to that after mounting?
selinuxfs in that namespace.

e.g. echo inherit > /sys/fs/selinux/parent_ns_labels

or something.

Quoted 11 lines

Another potential concern is that files created in a non-init namespace
are left completely unlabeled in the init namespace (or in any parent).
????As long as access to unlabeled is tightly controlled, this might
not be a problem, but I'm not sure that's guaranteed by the refpolicy
or Fedora/RHEL policies.??We might want to initialize an xattr at every

Quoted 8 lines

level of the namespace hierarchy when a new file is created based on
the process and parent directory labels and policy at that level.?
Otherwise, we lose all provenance information for the file outside
of
the namespace.?

Ok.

Quoted 7 lines

Quoted 4 lines

For example, suppose I want to leak information out of my category set; I unshare and create files with the information, and
they appear in the init namespace with no categories.

Nice :)

This btw would be a bit cleaner if we dropped the .ns. portion of the name, such that we would have:
security.selinux # xattr name in the init namespace security.selinux.vmN # xattr name in the vmN namespace security.selinux.vmN.vmM # xattr name in the vmN.vmM namespace

Quoted 6 lines · Stephen Smalley

This btw would be a bit cleaner if we dropped the .ns. portion of the
name, such that we would have:
security.selinux # xattr name in the init namespace security.selinux.vmN # xattr name in the vmN namespace security.selinux.vmN.vmM # xattr name in the vmN.vmM namespace

I used 'ns' to diffetentiate against other potential extensions of the?
xattr name.??If that's not a concern, then yes it will be cleaner.

Do we limit the number of nestings?

This btw would be a bit cleaner if we dropped the .ns. portion of the name, such that we would have:
security.selinux # xattr name in the init namespace security.selinux.vmN # xattr name in the vmN namespace security.selinux.vmN.vmM # xattr name in the vmN.vmM namespace

Quoted 6 lines · Stephen Smalley

This btw would be a bit cleaner if we dropped the .ns. portion of the
name, such that we would have:
security.selinux # xattr name in the init namespace security.selinux.vmN # xattr name in the vmN namespace security.selinux.vmN.vmM # xattr name in the vmN.vmM namespace

Ok, just to clarify, the namespace name in the last example is "vmN.vmM",?
not "vmM" ?

i.e. the namespaces are always hierarchical, and the security labels are?
identified by that hierarchy.??If you enter vmM from the init namespace,?
for example, the security labels for it are distinct from the labels under?
vmN.??On disk, you would have both:

security.selinux.vmM
security.selinux.vmN.vmM

which are independent.

Each of these instances would potentially inherit different labels, and?
have different provenance characteristics, so this seems necessary in any?
case.

Was also wondering if the name read for the init namespace ought to just be the empty string instead of the magic "init" value to make it consistent with the fact that there is no xattr suffix.