Merge tag 'selinux-pr-20190917' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux

Pull selinux updates from Paul Moore:

- Add LSM hooks, and SELinux access control hooks, for dnotify,
fanotify, and inotify watches. This has been discussed with both the
LSM and fs/notify folks and everybody is good with these new hooks.

- The LSM stacking changes missed a few calls to current_security() in
the SELinux code; we fix those and remove current_security() for
good.

- Improve our network object labeling cache so that we always return
the object's label, even when under memory pressure. Previously we
would return an error if we couldn't allocate a new cache entry, now
we always return the label even if we can't create a new cache entry
for it.

- Convert the sidtab atomic_t counter to a normal u32 with
READ/WRITE_ONCE() and memory barrier protection.

- A few patches to policydb.c to clean things up (remove forward
declarations, long lines, bad variable names, etc)

* tag 'selinux-pr-20190917' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux:
lsm: remove current_security()
selinux: fix residual uses of current_security() for the SELinux blob
selinux: avoid atomic_t usage in sidtab
fanotify, inotify, dnotify, security: add security hook for fs notifications
selinux: always return a secid from the network caches if we find one
selinux: policydb - rename type_val_to_struct_array
selinux: policydb - fix some checkpatch.pl warnings
selinux: shuffle around policydb.c to get rid of forward declarations

fs/notify/dnotify/dnotify.c

···
 #include <linux/sched/signal.h>
 #include <linux/dnotify.h>
 #include <linux/init.h>
+#include <linux/security.h>
 #include <linux/spinlock.h>
 #include <linux/slab.h>
 #include <linux/fdtable.h>
···
 		goto out_err;
 	}
 
+	/*
+	 * convert the userspace DN_* "arg" to the internal FS_*
+	 * defined in fsnotify
+	 */
+	mask = convert_arg(arg);
+
+	error = security_path_notify(&filp->f_path, mask,
+			FSNOTIFY_OBJ_TYPE_INODE);
+	if (error)
+		goto out_err;
+
 	/* expect most fcntl to add new rather than augment old */
 	dn = kmem_cache_alloc(dnotify_struct_cache, GFP_KERNEL);
 	if (!dn) {
···
 		error = -ENOMEM;
 		goto out_err;
 	}
-
-	/* convert the userspace DN_* "arg" to the internal FS_* defines in fsnotify */
-	mask = convert_arg(arg);
 
 	/* set up the new_fsn_mark and new_dn_mark */
 	new_fsn_mark = &new_dn_mark->fsn_mark;

fs/notify/fanotify/fanotify_user.c

···
 };
 
 static int fanotify_find_path(int dfd, const char __user *filename,
-			      struct path *path, unsigned int flags)
+			      struct path *path, unsigned int flags, __u64 mask,
+			      unsigned int obj_type)
 {
 	int ret;
 
···
 
 	/* you can only watch an inode if you have read permissions on it */
 	ret = inode_permission(path->dentry->d_inode, MAY_READ);
+	if (ret) {
+		path_put(path);
+		goto out;
+	}
+
+	ret = security_path_notify(path, mask, obj_type);
 	if (ret)
 		path_put(path);
+
 out:
 	return ret;
 }
···
 	__kernel_fsid_t __fsid, *fsid = NULL;
 	u32 valid_mask = FANOTIFY_EVENTS | FANOTIFY_EVENT_FLAGS;
 	unsigned int mark_type = flags & FANOTIFY_MARK_TYPE_BITS;
+	unsigned int obj_type;
 	int ret;
 
 	pr_debug("%s: fanotify_fd=%d flags=%x dfd=%d pathname=%p mask=%llx\n",
···
 
 	switch (mark_type) {
 	case FAN_MARK_INODE:
+		obj_type = FSNOTIFY_OBJ_TYPE_INODE;
+		break;
 	case FAN_MARK_MOUNT:
+		obj_type = FSNOTIFY_OBJ_TYPE_VFSMOUNT;
+		break;
 	case FAN_MARK_FILESYSTEM:
+		obj_type = FSNOTIFY_OBJ_TYPE_SB;
 		break;
 	default:
 		return -EINVAL;
···
 		goto fput_and_out;
 	}
 
-	ret = fanotify_find_path(dfd, pathname, &path, flags);
+	ret = fanotify_find_path(dfd, pathname, &path, flags,
+			(mask & ALL_FSNOTIFY_EVENTS), obj_type);
 	if (ret)
 		goto fput_and_out;
 

fs/notify/inotify/inotify_user.c

···
 #include <linux/poll.h>
 #include <linux/wait.h>
 #include <linux/memcontrol.h>
+#include <linux/security.h>
 
 #include "inotify.h"
 #include "../fdinfo.h"
···
 /*
  * find_inode - resolve a user-given path to a specific inode
  */
-static int inotify_find_inode(const char __user *dirname, struct path *path, unsigned flags)
+static int inotify_find_inode(const char __user *dirname, struct path *path,
+						unsigned int flags, __u64 mask)
 {
 	int error;
 
···
 		return error;
 	/* you can only watch an inode if you have read permissions on it */
 	error = inode_permission(path->dentry->d_inode, MAY_READ);
+	if (error) {
+		path_put(path);
+		return error;
+	}
+	error = security_path_notify(path, mask,
+				FSNOTIFY_OBJ_TYPE_INODE);
 	if (error)
 		path_put(path);
+
 	return error;
 }
 
···
 	if (mask & IN_ONLYDIR)
 		flags |= LOOKUP_DIRECTORY;
 
-	ret = inotify_find_inode(pathname, &path, flags);
+	ret = inotify_find_inode(pathname, &path, flags,
+			(mask & IN_ALL_EVENTS));
 	if (ret)
 		goto fput_and_out;
 

include/linux/cred.h

···
 #define current_fsgid() 	(current_cred_xxx(fsgid))
 #define current_cap()		(current_cred_xxx(cap_effective))
 #define current_user()		(current_cred_xxx(user))
-#define current_security()	(current_cred_xxx(security))
 
 extern struct user_namespace init_user_ns;
 #ifdef CONFIG_USER_NS

include/linux/lsm_hooks.h

···
  *	Check for permission to change root directory.
  *	@path contains the path structure.
  *	Return 0 if permission is granted.
+ * @path_notify:
+ *	Check permissions before setting a watch on events as defined by @mask,
+ *	on an object at @path, whose type is defined by @obj_type.
  * @inode_readlink:
  *	Check the permission to read the symbolic link.
  *	@dentry contains the dentry structure for the file link.
···
 	int (*path_chown)(const struct path *path, kuid_t uid, kgid_t gid);
 	int (*path_chroot)(const struct path *path);
 #endif
-
+	/* Needed for inode based security check */
+	int (*path_notify)(const struct path *path, u64 mask,
+				unsigned int obj_type);
 	int (*inode_alloc_security)(struct inode *inode);
 	void (*inode_free_security)(struct inode *inode);
 	int (*inode_init_security)(struct inode *inode, struct inode *dir,
···
 	struct hlist_head path_chown;
 	struct hlist_head path_chroot;
 #endif
+	/* Needed for inode based modules as well */
+	struct hlist_head path_notify;
 	struct hlist_head inode_alloc_security;
 	struct hlist_head inode_free_security;
 	struct hlist_head inode_init_security;

include/linux/security.h

···
 					struct qstr *name,
 					const struct cred *old,
 					struct cred *new);
-
+int security_path_notify(const struct path *path, u64 mask,
+					unsigned int obj_type);
 int security_inode_alloc(struct inode *inode);
 void security_inode_free(struct inode *inode);
 int security_inode_init_security(struct inode *inode, struct inode *dir,
···
 int security_secid_to_secctx(u32 secid, char **secdata, u32 *seclen);
 int security_secctx_to_secid(const char *secdata, u32 seclen, u32 *secid);
 void security_release_secctx(char *secdata, u32 seclen);
-
 void security_inode_invalidate_secctx(struct inode *inode);
 int security_inode_notifysecctx(struct inode *inode, void *ctx, u32 ctxlen);
 int security_inode_setsecctx(struct dentry *dentry, void *ctx, u32 ctxlen);
···
 
 static inline int security_move_mount(const struct path *from_path,
 				      const struct path *to_path)
+{
+	return 0;
+}
+
+static inline int security_path_notify(const struct path *path, u64 mask,
+				unsigned int obj_type)
 {
 	return 0;
 }

security/security.c

···
 	return call_int_hook(move_mount, 0, from_path, to_path);
 }
 
+int security_path_notify(const struct path *path, u64 mask,
+				unsigned int obj_type)
+{
+	return call_int_hook(path_notify, 0, path, mask, obj_type);
+}
+
 int security_inode_alloc(struct inode *inode)
 {
 	int rc = lsm_inode_alloc(inode);

security/selinux/hooks.c

···
 #include <linux/kernfs.h>
 #include <linux/stringhash.h>	/* for hashlen_string() */
 #include <uapi/linux/mount.h>
+#include <linux/fsnotify.h>
+#include <linux/fanotify.h>
 
 #include "avc.h"
 #include "objsec.h"
···
 	return -EACCES;
 }
 
+static int selinux_path_notify(const struct path *path, u64 mask,
+						unsigned int obj_type)
+{
+	int ret;
+	u32 perm;
+
+	struct common_audit_data ad;
+
+	ad.type = LSM_AUDIT_DATA_PATH;
+	ad.u.path = *path;
+
+	/*
+	 * Set permission needed based on the type of mark being set.
+	 * Performs an additional check for sb watches.
+	 */
+	switch (obj_type) {
+	case FSNOTIFY_OBJ_TYPE_VFSMOUNT:
+		perm = FILE__WATCH_MOUNT;
+		break;
+	case FSNOTIFY_OBJ_TYPE_SB:
+		perm = FILE__WATCH_SB;
+		ret = superblock_has_perm(current_cred(), path->dentry->d_sb,
+						FILESYSTEM__WATCH, &ad);
+		if (ret)
+			return ret;
+		break;
+	case FSNOTIFY_OBJ_TYPE_INODE:
+		perm = FILE__WATCH;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	/* blocking watches require the file:watch_with_perm permission */
+	if (mask & (ALL_FSNOTIFY_PERM_EVENTS))
+		perm |= FILE__WATCH_WITH_PERM;
+
+	/* watches on read-like events need the file:watch_reads permission */
+	if (mask & (FS_ACCESS | FS_ACCESS_PERM | FS_CLOSE_NOWRITE))
+		perm |= FILE__WATCH_READS;
+
+	return path_has_perm(current_cred(), path, perm);
+}
+
 /*
  * Copy the inode security context value to the user.
  *
···
 static int selinux_kernfs_init_security(struct kernfs_node *kn_dir,
 					struct kernfs_node *kn)
 {
-	const struct task_security_struct *tsec = current_security();
+	const struct task_security_struct *tsec = selinux_cred(current_cred());
 	u32 parent_sid, newsid, clen;
 	int rc;
 	char *context;
···
 	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(path_notify, selinux_path_notify),
 
 	LSM_HOOK_INIT(kernfs_init_security, selinux_kernfs_init_security),
 

security/selinux/include/classmap.h

···
 
 #define COMMON_FILE_PERMS COMMON_FILE_SOCK_PERMS, "unlink", "link", \
     "rename", "execute", "quotaon", "mounton", "audit_access", \
-    "open", "execmod"
+	"open", "execmod", "watch", "watch_mount", "watch_sb", \
+	"watch_with_perm", "watch_reads"
 
 #define COMMON_SOCK_PERMS COMMON_FILE_SOCK_PERMS, "bind", "connect", \
     "listen", "accept", "getopt", "setopt", "shutdown", "recvfrom",  \
···
 	{ "filesystem",
 	  { "mount", "remount", "unmount", "getattr",
 	    "relabelfrom", "relabelto", "associate", "quotamod",
-	    "quotaget", NULL } },
+	    "quotaget", "watch", NULL } },
 	{ "file",
 	  { COMMON_FILE_PERMS,
 	    "execute_no_trans", "entrypoint", NULL } },

security/selinux/include/objsec.h

···
 	u32 sockcreate_sid;	/* fscreate SID */
 };
 
-/*
- * get the subjective security ID of the current task
- */
-static inline u32 current_sid(void)
-{
-	const struct task_security_struct *tsec = current_security();
-
-	return tsec->sid;
-}
-
 enum label_initialized {
 	LABEL_INVALID,		/* invalid or not initialized */
 	LABEL_INITIALIZED,	/* initialized */
···
 						const struct kern_ipc_perm *ipc)
 {
 	return ipc->security + selinux_blob_sizes.lbs_ipc;
+}
+
+/*
+ * get the subjective security ID of the current task
+ */
+static inline u32 current_sid(void)
+{
+	const struct task_security_struct *tsec = selinux_cred(current_cred());
+
+	return tsec->sid;
 }
 
 #endif /* _SELINUX_OBJSEC_H_ */

security/selinux/netif.c

···
  */
 static int sel_netif_sid_slow(struct net *ns, int ifindex, u32 *sid)
 {
-	int ret;
+	int ret = 0;
 	struct sel_netif *netif;
-	struct sel_netif *new = NULL;
+	struct sel_netif *new;
 	struct net_device *dev;
 
 	/* NOTE: we always use init's network namespace since we don't
···
 	netif = sel_netif_find(ns, ifindex);
 	if (netif != NULL) {
 		*sid = netif->nsec.sid;
-		ret = 0;
 		goto out;
 	}
+
+	ret = security_netif_sid(&selinux_state, dev->name, sid);
+	if (ret != 0)
+		goto out;
 	new = kzalloc(sizeof(*new), GFP_ATOMIC);
-	if (new == NULL) {
-		ret = -ENOMEM;
-		goto out;
+	if (new) {
+		new->nsec.ns = ns;
+		new->nsec.ifindex = ifindex;
+		new->nsec.sid = *sid;
+		if (sel_netif_insert(new))
+			kfree(new);
 	}
-	ret = security_netif_sid(&selinux_state, dev->name, &new->nsec.sid);
-	if (ret != 0)
-		goto out;
-	new->nsec.ns = ns;
-	new->nsec.ifindex = ifindex;
-	ret = sel_netif_insert(new);
-	if (ret != 0)
-		goto out;
-	*sid = new->nsec.sid;
 
 out:
 	spin_unlock_bh(&sel_netif_lock);
 	dev_put(dev);
-	if (unlikely(ret)) {
+	if (unlikely(ret))
 		pr_warn("SELinux: failure in %s(), unable to determine network interface label (%d)\n",
 			__func__, ifindex);
-		kfree(new);
-	}
 	return ret;
 }
 

security/selinux/netnode.c

···
  */
 static int sel_netnode_sid_slow(void *addr, u16 family, u32 *sid)
 {
-	int ret = -ENOMEM;
+	int ret;
 	struct sel_netnode *node;
-	struct sel_netnode *new = NULL;
+	struct sel_netnode *new;
 
 	spin_lock_bh(&sel_netnode_lock);
 	node = sel_netnode_find(addr, family);
···
 		spin_unlock_bh(&sel_netnode_lock);
 		return 0;
 	}
+
 	new = kzalloc(sizeof(*new), GFP_ATOMIC);
-	if (new == NULL)
-		goto out;
 	switch (family) {
 	case PF_INET:
 		ret = security_node_sid(&selinux_state, PF_INET,
 					addr, sizeof(struct in_addr), sid);
-		new->nsec.addr.ipv4 = *(__be32 *)addr;
+		if (new)
+			new->nsec.addr.ipv4 = *(__be32 *)addr;
 		break;
 	case PF_INET6:
 		ret = security_node_sid(&selinux_state, PF_INET6,
 					addr, sizeof(struct in6_addr), sid);
-		new->nsec.addr.ipv6 = *(struct in6_addr *)addr;
+		if (new)
+			new->nsec.addr.ipv6 = *(struct in6_addr *)addr;
 		break;
 	default:
 		BUG();
 		ret = -EINVAL;
 	}
-	if (ret != 0)
-		goto out;
+	if (ret == 0 && new) {
+		new->nsec.family = family;
+		new->nsec.sid = *sid;
+		sel_netnode_insert(new);
+	} else
+		kfree(new);
 
-	new->nsec.family = family;
-	new->nsec.sid = *sid;
-	sel_netnode_insert(new);
-
-out:
 	spin_unlock_bh(&sel_netnode_lock);
-	if (unlikely(ret)) {
+	if (unlikely(ret))
 		pr_warn("SELinux: failure in %s(), unable to determine network node label\n",
 			__func__);
-		kfree(new);
-	}
 	return ret;
 }
 

security/selinux/netport.c

···
  */
 static int sel_netport_sid_slow(u8 protocol, u16 pnum, u32 *sid)
 {
-	int ret = -ENOMEM;
+	int ret;
 	struct sel_netport *port;
-	struct sel_netport *new = NULL;
+	struct sel_netport *new;
 
 	spin_lock_bh(&sel_netport_lock);
 	port = sel_netport_find(protocol, pnum);
···
 		spin_unlock_bh(&sel_netport_lock);
 		return 0;
 	}
-	new = kzalloc(sizeof(*new), GFP_ATOMIC);
-	if (new == NULL)
-		goto out;
+
 	ret = security_port_sid(&selinux_state, protocol, pnum, sid);
 	if (ret != 0)
 		goto out;
-
-	new->psec.port = pnum;
-	new->psec.protocol = protocol;
-	new->psec.sid = *sid;
-	sel_netport_insert(new);
+	new = kzalloc(sizeof(*new), GFP_ATOMIC);
+	if (new) {
+		new->psec.port = pnum;
+		new->psec.protocol = protocol;
+		new->psec.sid = *sid;
+		sel_netport_insert(new);
+	}
 
 out:
 	spin_unlock_bh(&sel_netport_lock);
-	if (unlikely(ret)) {
+	if (unlikely(ret))
 		pr_warn("SELinux: failure in %s(), unable to determine network port label\n",
 			__func__);
-		kfree(new);
-	}
 	return ret;
 }
 

security/selinux/ss/policydb.c

···
 }
 
 /*
- * Initialize the role table.
- */
-static int roles_init(struct policydb *p)
-{
-	char *key = NULL;
-	int rc;
-	struct role_datum *role;
-
-	role = kzalloc(sizeof(*role), GFP_KERNEL);
-	if (!role)
-		return -ENOMEM;
-
-	rc = -EINVAL;
-	role->value = ++p->p_roles.nprim;
-	if (role->value != OBJECT_R_VAL)
-		goto out;
-
-	rc = -ENOMEM;
-	key = kstrdup(OBJECT_R, GFP_KERNEL);
-	if (!key)
-		goto out;
-
-	rc = hashtab_insert(p->p_roles.table, key, role);
-	if (rc)
-		goto out;
-
-	return 0;
-out:
-	kfree(key);
-	kfree(role);
-	return rc;
-}
-
-static u32 filenametr_hash(struct hashtab *h, const void *k)
-{
-	const struct filename_trans *ft = k;
-	unsigned long hash;
-	unsigned int byte_num;
-	unsigned char focus;
-
-	hash = ft->stype ^ ft->ttype ^ ft->tclass;
-
-	byte_num = 0;
-	while ((focus = ft->name[byte_num++]))
-		hash = partial_name_hash(focus, hash);
-	return hash & (h->size - 1);
-}
-
-static int filenametr_cmp(struct hashtab *h, const void *k1, const void *k2)
-{
-	const struct filename_trans *ft1 = k1;
-	const struct filename_trans *ft2 = k2;
-	int v;
-
-	v = ft1->stype - ft2->stype;
-	if (v)
-		return v;
-
-	v = ft1->ttype - ft2->ttype;
-	if (v)
-		return v;
-
-	v = ft1->tclass - ft2->tclass;
-	if (v)
-		return v;
-
-	return strcmp(ft1->name, ft2->name);
-
-}
-
-static u32 rangetr_hash(struct hashtab *h, const void *k)
-{
-	const struct range_trans *key = k;
-	return (key->source_type + (key->target_type << 3) +
-		(key->target_class << 5)) & (h->size - 1);
-}
-
-static int rangetr_cmp(struct hashtab *h, const void *k1, const void *k2)
-{
-	const struct range_trans *key1 = k1, *key2 = k2;
-	int v;
-
-	v = key1->source_type - key2->source_type;
-	if (v)
-		return v;
-
-	v = key1->target_type - key2->target_type;
-	if (v)
-		return v;
-
-	v = key1->target_class - key2->target_class;
-
-	return v;
-}
-
-static int (*destroy_f[SYM_NUM]) (void *key, void *datum, void *datap);
-
-/*
- * Initialize a policy database structure.
- */
-static int policydb_init(struct policydb *p)
-{
-	int i, rc;
-
-	memset(p, 0, sizeof(*p));
-
-	for (i = 0; i < SYM_NUM; i++) {
-		rc = symtab_init(&p->symtab[i], symtab_sizes[i]);
-		if (rc)
-			goto out;
-	}
-
-	rc = avtab_init(&p->te_avtab);
-	if (rc)
-		goto out;
-
-	rc = roles_init(p);
-	if (rc)
-		goto out;
-
-	rc = cond_policydb_init(p);
-	if (rc)
-		goto out;
-
-	p->filename_trans = hashtab_create(filenametr_hash, filenametr_cmp, (1 << 10));
-	if (!p->filename_trans) {
-		rc = -ENOMEM;
-		goto out;
-	}
-
-	p->range_tr = hashtab_create(rangetr_hash, rangetr_cmp, 256);
-	if (!p->range_tr) {
-		rc = -ENOMEM;
-		goto out;
-	}
-
-	ebitmap_init(&p->filename_trans_ttypes);
-	ebitmap_init(&p->policycaps);
-	ebitmap_init(&p->permissive_map);
-
-	return 0;
-out:
-	hashtab_destroy(p->filename_trans);
-	hashtab_destroy(p->range_tr);
-	for (i = 0; i < SYM_NUM; i++) {
-		hashtab_map(p->symtab[i].table, destroy_f[i], NULL);
-		hashtab_destroy(p->symtab[i].table);
-	}
-	return rc;
-}
-
-/*
- * The following *_index functions are used to
- * define the val_to_name and val_to_struct arrays
- * in a policy database structure.  The val_to_name
- * arrays are used when converting security context
- * structures into string representations.  The
- * val_to_struct arrays are used when the attributes
- * of a class, role, or user are needed.
- */
-
-static int common_index(void *key, void *datum, void *datap)
-{
-	struct policydb *p;
-	struct common_datum *comdatum;
-
-	comdatum = datum;
-	p = datap;
-	if (!comdatum->value || comdatum->value > p->p_commons.nprim)
-		return -EINVAL;
-
-	p->sym_val_to_name[SYM_COMMONS][comdatum->value - 1] = key;
-
-	return 0;
-}
-
-static int class_index(void *key, void *datum, void *datap)
-{
-	struct policydb *p;
-	struct class_datum *cladatum;
-
-	cladatum = datum;
-	p = datap;
-	if (!cladatum->value || cladatum->value > p->p_classes.nprim)
-		return -EINVAL;
-
-	p->sym_val_to_name[SYM_CLASSES][cladatum->value - 1] = key;
-	p->class_val_to_struct[cladatum->value - 1] = cladatum;
-	return 0;
-}
-
-static int role_index(void *key, void *datum, void *datap)
-{
-	struct policydb *p;
-	struct role_datum *role;
-
-	role = datum;
-	p = datap;
-	if (!role->value
-	    || role->value > p->p_roles.nprim
-	    || role->bounds > p->p_roles.nprim)
-		return -EINVAL;
-
-	p->sym_val_to_name[SYM_ROLES][role->value - 1] = key;
-	p->role_val_to_struct[role->value - 1] = role;
-	return 0;
-}
-
-static int type_index(void *key, void *datum, void *datap)
-{
-	struct policydb *p;
-	struct type_datum *typdatum;
-
-	typdatum = datum;
-	p = datap;
-
-	if (typdatum->primary) {
-		if (!typdatum->value
-		    || typdatum->value > p->p_types.nprim
-		    || typdatum->bounds > p->p_types.nprim)
-			return -EINVAL;
-		p->sym_val_to_name[SYM_TYPES][typdatum->value - 1] = key;
-		p->type_val_to_struct_array[typdatum->value - 1] = typdatum;
-	}
-
-	return 0;
-}
-
-static int user_index(void *key, void *datum, void *datap)
-{
-	struct policydb *p;
-	struct user_datum *usrdatum;
-
-	usrdatum = datum;
-	p = datap;
-	if (!usrdatum->value
-	    || usrdatum->value > p->p_users.nprim
-	    || usrdatum->bounds > p->p_users.nprim)
-		return -EINVAL;
-
-	p->sym_val_to_name[SYM_USERS][usrdatum->value - 1] = key;
-	p->user_val_to_struct[usrdatum->value - 1] = usrdatum;
-	return 0;
-}
-
-static int sens_index(void *key, void *datum, void *datap)
-{
-	struct policydb *p;
-	struct level_datum *levdatum;
-
-	levdatum = datum;
-	p = datap;
-
-	if (!levdatum->isalias) {
-		if (!levdatum->level->sens ||
-		    levdatum->level->sens > p->p_levels.nprim)
-			return -EINVAL;
-
-		p->sym_val_to_name[SYM_LEVELS][levdatum->level->sens - 1] = key;
-	}
-
-	return 0;
-}
-
-static int cat_index(void *key, void *datum, void *datap)
-{
-	struct policydb *p;
-	struct cat_datum *catdatum;
-
-	catdatum = datum;
-	p = datap;
-
-	if (!catdatum->isalias) {
-		if (!catdatum->value || catdatum->value > p->p_cats.nprim)
-			return -EINVAL;
-
-		p->sym_val_to_name[SYM_CATS][catdatum->value - 1] = key;
-	}
-
-	return 0;
-}
-
-static int (*index_f[SYM_NUM]) (void *key, void *datum, void *datap) =
-{
-	common_index,
-	class_index,
-	role_index,
-	type_index,
-	user_index,
-	cond_index_bool,
-	sens_index,
-	cat_index,
-};
-
-#ifdef DEBUG_HASHES
-static void hash_eval(struct hashtab *h, const char *hash_name)
-{
-	struct hashtab_info info;
-
-	hashtab_stat(h, &info);
-	pr_debug("SELinux: %s:  %d entries and %d/%d buckets used, "
-	       "longest chain length %d\n", hash_name, h->nel,
-	       info.slots_used, h->size, info.max_chain_len);
-}
-
-static void symtab_hash_eval(struct symtab *s)
-{
-	int i;
-
-	for (i = 0; i < SYM_NUM; i++)
-		hash_eval(s[i].table, symtab_name[i]);
-}
-
-#else
-static inline void hash_eval(struct hashtab *h, char *hash_name)
-{
-}
-#endif
-
-/*
- * Define the other val_to_name and val_to_struct arrays
- * in a policy database structure.
- *
- * Caller must clean up on failure.
- */
-static int policydb_index(struct policydb *p)
-{
-	int i, rc;
-
-	if (p->mls_enabled)
-		pr_debug("SELinux:  %d users, %d roles, %d types, %d bools, %d sens, %d cats\n",
-			 p->p_users.nprim, p->p_roles.nprim, p->p_types.nprim,
-			 p->p_bools.nprim, p->p_levels.nprim, p->p_cats.nprim);
-	else
-		pr_debug("SELinux:  %d users, %d roles, %d types, %d bools\n",
-			 p->p_users.nprim, p->p_roles.nprim, p->p_types.nprim,
-			 p->p_bools.nprim);
-
-	pr_debug("SELinux:  %d classes, %d rules\n",
-		 p->p_classes.nprim, p->te_avtab.nel);
-
-#ifdef DEBUG_HASHES
-	avtab_hash_eval(&p->te_avtab, "rules");
-	symtab_hash_eval(p->symtab);
-#endif
-
-	p->class_val_to_struct = kcalloc(p->p_classes.nprim,
-					 sizeof(*p->class_val_to_struct),
-					 GFP_KERNEL);
-	if (!p->class_val_to_struct)
-		return -ENOMEM;
-
-	p->role_val_to_struct = kcalloc(p->p_roles.nprim,
-					sizeof(*p->role_val_to_struct),
-					GFP_KERNEL);
-	if (!p->role_val_to_struct)
-		return -ENOMEM;
-
-	p->user_val_to_struct = kcalloc(p->p_users.nprim,
-					sizeof(*p->user_val_to_struct),
-					GFP_KERNEL);
-	if (!p->user_val_to_struct)
-		return -ENOMEM;
-
-	p->type_val_to_struct_array = kvcalloc(p->p_types.nprim,
-					       sizeof(*p->type_val_to_struct_array),
-					       GFP_KERNEL);
-	if (!p->type_val_to_struct_array)
-		return -ENOMEM;
-
-	rc = cond_init_bool_indexes(p);
-	if (rc)
-		goto out;
-
-	for (i = 0; i < SYM_NUM; i++) {
-		p->sym_val_to_name[i] = kvcalloc(p->symtab[i].nprim,
-						 sizeof(char *),
-						 GFP_KERNEL);
-		if (!p->sym_val_to_name[i])
-			return -ENOMEM;
-
-		rc = hashtab_map(p->symtab[i].table, index_f[i], p);
-		if (rc)
-			goto out;
-	}
-	rc = 0;
-out:
-	return rc;
-}
-
-/*
  * The following *_destroy functions are used to
  * free any memory allocated for each kind of
  * symbol data in the policy database.
···
 static int filenametr_destroy(void *key, void *datum, void *p)
 {
 	struct filename_trans *ft = key;
+
 	kfree(ft->name);
 	kfree(key);
 	kfree(datum);
···
 static int range_tr_destroy(void *key, void *datum, void *p)
 {
 	struct mls_range *rt = datum;
+
 	kfree(key);
 	ebitmap_destroy(&rt->level[0].cat);
 	ebitmap_destroy(&rt->level[1].cat);
···
 	    i == OCON_NETIF || i == OCON_FSUSE)
 		kfree(c->u.name);
 	kfree(c);
+}
+
+/*
+ * Initialize the role table.
+ */
+static int roles_init(struct policydb *p)
+{
+	char *key = NULL;
+	int rc;
+	struct role_datum *role;
+
+	role = kzalloc(sizeof(*role), GFP_KERNEL);
+	if (!role)
+		return -ENOMEM;
+
+	rc = -EINVAL;
+	role->value = ++p->p_roles.nprim;
+	if (role->value != OBJECT_R_VAL)
+		goto out;
+
+	rc = -ENOMEM;
+	key = kstrdup(OBJECT_R, GFP_KERNEL);
+	if (!key)
+		goto out;
+
+	rc = hashtab_insert(p->p_roles.table, key, role);
+	if (rc)
+		goto out;
+
+	return 0;
+out:
+	kfree(key);
+	kfree(role);
+	return rc;
+}
+
+static u32 filenametr_hash(struct hashtab *h, const void *k)
+{
+	const struct filename_trans *ft = k;
+	unsigned long hash;
+	unsigned int byte_num;
+	unsigned char focus;
+
+	hash = ft->stype ^ ft->ttype ^ ft->tclass;
+
+	byte_num = 0;
+	while ((focus = ft->name[byte_num++]))
+		hash = partial_name_hash(focus, hash);
+	return hash & (h->size - 1);
+}
+
+static int filenametr_cmp(struct hashtab *h, const void *k1, const void *k2)
+{
+	const struct filename_trans *ft1 = k1;
+	const struct filename_trans *ft2 = k2;
+	int v;
+
+	v = ft1->stype - ft2->stype;
+	if (v)
+		return v;
+
+	v = ft1->ttype - ft2->ttype;
+	if (v)
+		return v;
+
+	v = ft1->tclass - ft2->tclass;
+	if (v)
+		return v;
+
+	return strcmp(ft1->name, ft2->name);
+
+}
+
+static u32 rangetr_hash(struct hashtab *h, const void *k)
+{
+	const struct range_trans *key = k;
+
+	return (key->source_type + (key->target_type << 3) +
+		(key->target_class << 5)) & (h->size - 1);
+}
+
+static int rangetr_cmp(struct hashtab *h, const void *k1, const void *k2)
+{
+	const struct range_trans *key1 = k1, *key2 = k2;
+	int v;
+
+	v = key1->source_type - key2->source_type;
+	if (v)
+		return v;
+
+	v = key1->target_type - key2->target_type;
+	if (v)
+		return v;
+
+	v = key1->target_class - key2->target_class;
+
+	return v;
+}
+
+/*
+ * Initialize a policy database structure.
+ */
+static int policydb_init(struct policydb *p)
+{
+	int i, rc;
+
+	memset(p, 0, sizeof(*p));
+
+	for (i = 0; i < SYM_NUM; i++) {
+		rc = symtab_init(&p->symtab[i], symtab_sizes[i]);
+		if (rc)
+			goto out;
+	}
+
+	rc = avtab_init(&p->te_avtab);
+	if (rc)
+		goto out;
+
+	rc = roles_init(p);
+	if (rc)
+		goto out;
+
+	rc = cond_policydb_init(p);
+	if (rc)
+		goto out;
+
+	p->filename_trans = hashtab_create(filenametr_hash, filenametr_cmp,
+					   (1 << 10));
+	if (!p->filename_trans) {
+		rc = -ENOMEM;
+		goto out;
+	}
+
+	p->range_tr = hashtab_create(rangetr_hash, rangetr_cmp, 256);
+	if (!p->range_tr) {
+		rc = -ENOMEM;
+		goto out;
+	}
+
+	ebitmap_init(&p->filename_trans_ttypes);
+	ebitmap_init(&p->policycaps);
+	ebitmap_init(&p->permissive_map);
+
+	return 0;
+out:
+	hashtab_destroy(p->filename_trans);
+	hashtab_destroy(p->range_tr);
+	for (i = 0; i < SYM_NUM; i++) {
+		hashtab_map(p->symtab[i].table, destroy_f[i], NULL);
+		hashtab_destroy(p->symtab[i].table);
+	}
+	return rc;
+}
+
+/*
+ * The following *_index functions are used to
+ * define the val_to_name and val_to_struct arrays
+ * in a policy database structure.  The val_to_name
+ * arrays are used when converting security context
+ * structures into string representations.  The
+ * val_to_struct arrays are used when the attributes
+ * of a class, role, or user are needed.
+ */
+
+static int common_index(void *key, void *datum, void *datap)
+{
+	struct policydb *p;
+	struct common_datum *comdatum;
+
+	comdatum = datum;
+	p = datap;
+	if (!comdatum->value || comdatum->value > p->p_commons.nprim)
+		return -EINVAL;
+
+	p->sym_val_to_name[SYM_COMMONS][comdatum->value - 1] = key;
+
+	return 0;
+}
+
+static int class_index(void *key, void *datum, void *datap)
+{
+	struct policydb *p;
+	struct class_datum *cladatum;
+
+	cladatum = datum;
+	p = datap;
+	if (!cladatum->value || cladatum->value > p->p_classes.nprim)
+		return -EINVAL;
+
+	p->sym_val_to_name[SYM_CLASSES][cladatum->value - 1] = key;
+	p->class_val_to_struct[cladatum->value - 1] = cladatum;
+	return 0;
+}
+
+static int role_index(void *key, void *datum, void *datap)
+{
+	struct policydb *p;
+	struct role_datum *role;
+
+	role = datum;
+	p = datap;
+	if (!role->value
+	    || role->value > p->p_roles.nprim
+	    || role->bounds > p->p_roles.nprim)
+		return -EINVAL;
+
+	p->sym_val_to_name[SYM_ROLES][role->value - 1] = key;
+	p->role_val_to_struct[role->value - 1] = role;
+	return 0;
+}
+
+static int type_index(void *key, void *datum, void *datap)
+{
+	struct policydb *p;
+	struct type_datum *typdatum;
+
+	typdatum = datum;
+	p = datap;
+
+	if (typdatum->primary) {
+		if (!typdatum->value
+		    || typdatum->value > p->p_types.nprim
+		    || typdatum->bounds > p->p_types.nprim)
+			return -EINVAL;
+		p->sym_val_to_name[SYM_TYPES][typdatum->value - 1] = key;
+		p->type_val_to_struct[typdatum->value - 1] = typdatum;
+	}
+
+	return 0;
+}
+
+static int user_index(void *key, void *datum, void *datap)
+{
+	struct policydb *p;
+	struct user_datum *usrdatum;
+
+	usrdatum = datum;
+	p = datap;
+	if (!usrdatum->value
+	    || usrdatum->value > p->p_users.nprim
+	    || usrdatum->bounds > p->p_users.nprim)
+		return -EINVAL;
+
+	p->sym_val_to_name[SYM_USERS][usrdatum->value - 1] = key;
+	p->user_val_to_struct[usrdatum->value - 1] = usrdatum;
+	return 0;
+}
+
+static int sens_index(void *key, void *datum, void *datap)
+{
+	struct policydb *p;
+	struct level_datum *levdatum;
+
+	levdatum = datum;
+	p = datap;
+
+	if (!levdatum->isalias) {
+		if (!levdatum->level->sens ||
+		    levdatum->level->sens > p->p_levels.nprim)
+			return -EINVAL;
+
+		p->sym_val_to_name[SYM_LEVELS][levdatum->level->sens - 1] = key;
+	}
+
+	return 0;
+}
+
+static int cat_index(void *key, void *datum, void *datap)
+{
+	struct policydb *p;
+	struct cat_datum *catdatum;
+
+	catdatum = datum;
+	p = datap;
+
+	if (!catdatum->isalias) {
+		if (!catdatum->value || catdatum->value > p->p_cats.nprim)
+			return -EINVAL;
+
+		p->sym_val_to_name[SYM_CATS][catdatum->value - 1] = key;
+	}
+
+	return 0;
+}
+
+static int (*index_f[SYM_NUM]) (void *key, void *datum, void *datap) =
+{
+	common_index,
+	class_index,
+	role_index,
+	type_index,
+	user_index,
+	cond_index_bool,
+	sens_index,
+	cat_index,
+};
+
+#ifdef DEBUG_HASHES
+static void hash_eval(struct hashtab *h, const char *hash_name)
+{
+	struct hashtab_info info;
+
+	hashtab_stat(h, &info);
+	pr_debug("SELinux: %s:  %d entries and %d/%d buckets used, longest chain length %d\n",
+		 hash_name, h->nel, info.slots_used, h->size,
+		 info.max_chain_len);
+}
+
+static void symtab_hash_eval(struct symtab *s)
+{
+	int i;
+
+	for (i = 0; i < SYM_NUM; i++)
+		hash_eval(s[i].table, symtab_name[i]);
+}
+
+#else
+static inline void hash_eval(struct hashtab *h, char *hash_name)
+{
+}
+#endif
+
+/*
+ * Define the other val_to_name and val_to_struct arrays
+ * in a policy database structure.
+ *
+ * Caller must clean up on failure.
+ */
+static int policydb_index(struct policydb *p)
+{
+	int i, rc;
+
+	if (p->mls_enabled)
+		pr_debug("SELinux:  %d users, %d roles, %d types, %d bools, %d sens, %d cats\n",
+			 p->p_users.nprim, p->p_roles.nprim, p->p_types.nprim,
+			 p->p_bools.nprim, p->p_levels.nprim, p->p_cats.nprim);
+	else
+		pr_debug("SELinux:  %d users, %d roles, %d types, %d bools\n",
+			 p->p_users.nprim, p->p_roles.nprim, p->p_types.nprim,
+			 p->p_bools.nprim);
+
+	pr_debug("SELinux:  %d classes, %d rules\n",
+		 p->p_classes.nprim, p->te_avtab.nel);
+
+#ifdef DEBUG_HASHES
+	avtab_hash_eval(&p->te_avtab, "rules");
+	symtab_hash_eval(p->symtab);
+#endif
+
+	p->class_val_to_struct = kcalloc(p->p_classes.nprim,
+					 sizeof(*p->class_val_to_struct),
+					 GFP_KERNEL);
+	if (!p->class_val_to_struct)
+		return -ENOMEM;
+
+	p->role_val_to_struct = kcalloc(p->p_roles.nprim,
+					sizeof(*p->role_val_to_struct),
+					GFP_KERNEL);
+	if (!p->role_val_to_struct)
+		return -ENOMEM;
+
+	p->user_val_to_struct = kcalloc(p->p_users.nprim,
+					sizeof(*p->user_val_to_struct),
+					GFP_KERNEL);
+	if (!p->user_val_to_struct)
+		return -ENOMEM;
+
+	p->type_val_to_struct = kvcalloc(p->p_types.nprim,
+					 sizeof(*p->type_val_to_struct),
+					 GFP_KERNEL);
+	if (!p->type_val_to_struct)
+		return -ENOMEM;
+
+	rc = cond_init_bool_indexes(p);
+	if (rc)
+		goto out;
+
+	for (i = 0; i < SYM_NUM; i++) {
+		p->sym_val_to_name[i] = kvcalloc(p->symtab[i].nprim,
+						 sizeof(char *),
+						 GFP_KERNEL);
+		if (!p->sym_val_to_name[i])
+			return -ENOMEM;
+
+		rc = hashtab_map(p->symtab[i].table, index_f[i], p);
+		if (rc)
+			goto out;
+	}
+	rc = 0;
+out:
+	return rc;
 }
 
 /*
···
 	kfree(p->class_val_to_struct);
 	kfree(p->role_val_to_struct);
 	kfree(p->user_val_to_struct);
-	kvfree(p->type_val_to_struct_array);
+	kvfree(p->type_val_to_struct);
 
 	avtab_destroy(&p->te_avtab);
 
···
 			return -EINVAL;
 		}
 
-		upper = p->type_val_to_struct_array[upper->bounds - 1];
+		upper = p->type_val_to_struct[upper->bounds - 1];
 		BUG_ON(!upper);
 
 		if (upper->attribute) {

security/selinux/ss/policydb.h

···
 	struct class_datum **class_val_to_struct;
 	struct role_datum **role_val_to_struct;
 	struct user_datum **user_val_to_struct;
-	struct type_datum **type_val_to_struct_array;
+	struct type_datum **type_val_to_struct;
 
 	/* type enforcement access vectors and transitions */
 	struct avtab te_avtab;

security/selinux/ss/services.c

···
 	struct type_datum *target;
 	u32 masked = 0;
 
-	source = policydb->type_val_to_struct_array[scontext->type - 1];
+	source = policydb->type_val_to_struct[scontext->type - 1];
 	BUG_ON(!source);
 
 	if (!source->bounds)
 		return;
 
-	target = policydb->type_val_to_struct_array[tcontext->type - 1];
+	target = policydb->type_val_to_struct[tcontext->type - 1];
 	BUG_ON(!target);
 
 	memset(&lo_avd, 0, sizeof(lo_avd));
···
 
 	index = new_context->type;
 	while (true) {
-		type = policydb->type_val_to_struct_array[index - 1];
+		type = policydb->type_val_to_struct[index - 1];
 		BUG_ON(!type);
 
 		/* not bounded anymore */

security/selinux/ss/sidtab.c

···
 #include <linux/slab.h>
 #include <linux/sched.h>
 #include <linux/spinlock.h>
-#include <linux/atomic.h>
+#include <asm/barrier.h>
 #include "flask.h"
 #include "security.h"
 #include "sidtab.h"
···
 
 	memset(s->roots, 0, sizeof(s->roots));
 
+	/* max count is SIDTAB_MAX so valid index is always < SIDTAB_MAX */
 	for (i = 0; i < SIDTAB_RCACHE_SIZE; i++)
-		atomic_set(&s->rcache[i], -1);
+		s->rcache[i] = SIDTAB_MAX;
 
 	for (i = 0; i < SECINITSID_NUM; i++)
 		s->isids[i].set = 0;
 
-	atomic_set(&s->count, 0);
-
+	s->count = 0;
 	s->convert = NULL;
 
 	spin_lock_init(&s->lock);
···
 
 static struct context *sidtab_lookup(struct sidtab *s, u32 index)
 {
-	u32 count = (u32)atomic_read(&s->count);
+	/* read entries only after reading count */
+	u32 count = smp_load_acquire(&s->count);
 
 	if (index >= count)
 		return NULL;
-
-	/* read entries after reading count */
-	smp_rmb();
 
 	return sidtab_do_lookup(s, index, 0);
 }
···
 static void sidtab_rcache_update(struct sidtab *s, u32 index, u32 pos)
 {
 	while (pos > 0) {
-		atomic_set(&s->rcache[pos], atomic_read(&s->rcache[pos - 1]));
+		WRITE_ONCE(s->rcache[pos], READ_ONCE(s->rcache[pos - 1]));
 		--pos;
 	}
-	atomic_set(&s->rcache[0], (int)index);
+	WRITE_ONCE(s->rcache[0], index);
 }
 
 static void sidtab_rcache_push(struct sidtab *s, u32 index)
···
 	u32 i;
 
 	for (i = 0; i < SIDTAB_RCACHE_SIZE; i++) {
-		int v = atomic_read(&s->rcache[i]);
+		u32 v = READ_ONCE(s->rcache[i]);
 
-		if (v < 0)
+		if (v >= SIDTAB_MAX)
 			continue;
 
-		if (context_cmp(sidtab_do_lookup(s, (u32)v, 0), context)) {
-			sidtab_rcache_update(s, (u32)v, i);
-			*index = (u32)v;
+		if (context_cmp(sidtab_do_lookup(s, v, 0), context)) {
+			sidtab_rcache_update(s, v, i);
+			*index = v;
 			return 0;
 		}
 	}
···
 				 u32 *index)
 {
 	unsigned long flags;
-	u32 count = (u32)atomic_read(&s->count);
-	u32 count_locked, level, pos;
+	u32 count, count_locked, level, pos;
 	struct sidtab_convert_params *convert;
 	struct context *dst, *dst_convert;
 	int rc;
···
 	if (rc == 0)
 		return 0;
 
+	/* read entries only after reading count */
+	count = smp_load_acquire(&s->count);
 	level = sidtab_level_from_count(count);
-
-	/* read entries after reading count */
-	smp_rmb();
 
 	pos = 0;
 	rc = sidtab_find_context(s->roots[level], &pos, count, level,
···
 	spin_lock_irqsave(&s->lock, flags);
 
 	convert = s->convert;
-	count_locked = (u32)atomic_read(&s->count);
+	count_locked = s->count;
 	level = sidtab_level_from_count(count_locked);
 
 	/* if count has changed before we acquired the lock, then catch up */
···
 		}
 
 		/* at this point we know the insert won't fail */
-		atomic_set(&convert->target->count, count + 1);
+		convert->target->count = count + 1;
 	}
 
 	if (context->len)
···
 	*index = count;
 
 	/* write entries before writing new count */
-	smp_wmb();
-
-	atomic_set(&s->count, count + 1);
+	smp_store_release(&s->count, count + 1);
 
 	rc = 0;
 out_unlock:
···
 		return -EBUSY;
 	}
 
-	count = (u32)atomic_read(&s->count);
+	count = s->count;
 	level = sidtab_level_from_count(count);
 
 	/* allocate last leaf in the new sidtab (to avoid race with
···
 	}
 
 	/* set count in case no new entries are added during conversion */
-	atomic_set(&params->target->count, count);
+	params->target->count = count;
 
 	/* enable live convert of new entries */
 	s->convert = params;

security/selinux/ss/sidtab.h

···
 #define SIDTAB_LEAF_ENTRIES \
 	(SIDTAB_NODE_ALLOC_SIZE / sizeof(struct sidtab_entry_leaf))
 
-#define SIDTAB_MAX_BITS 31 /* limited to INT_MAX due to atomic_t range */
-#define SIDTAB_MAX (((u32)1 << SIDTAB_MAX_BITS) - 1)
+#define SIDTAB_MAX_BITS 32
+#define SIDTAB_MAX U32_MAX
 /* ensure enough tree levels for SIDTAB_MAX entries */
 #define SIDTAB_MAX_LEVEL \
 	DIV_ROUND_UP(SIDTAB_MAX_BITS - size_to_shift(SIDTAB_LEAF_ENTRIES), \
···
 #define SIDTAB_RCACHE_SIZE 3
 
 struct sidtab {
+	/*
+	 * lock-free read access only for as many items as a prior read of
+	 * 'count'
+	 */
 	union sidtab_entry_inner roots[SIDTAB_MAX_LEVEL + 1];
-	atomic_t count;
+	/*
+	 * access atomically via {READ|WRITE}_ONCE(); only increment under
+	 * spinlock
+	 */
+	u32 count;
+	/* access only under spinlock */
 	struct sidtab_convert_params *convert;
 	spinlock_t lock;
 
-	/* reverse lookup cache */
-	atomic_t rcache[SIDTAB_RCACHE_SIZE];
+	/* reverse lookup cache - access atomically via {READ|WRITE}_ONCE() */
+	u32 rcache[SIDTAB_RCACHE_SIZE];
 
 	/* index == SID - 1 (no entry for SECSID_NULL) */
 	struct sidtab_isid_entry isids[SECINITSID_NUM];