proc: Usable inode numbers for the namespace file descriptors.

Assign a unique proc inode to each namespace, and use that
inode number to ensure we only allocate at most one proc
inode for every namespace in proc.

A single proc inode per namespace allows userspace to test
to see if two processes are in the same namespace.

This has been a long requested feature and only blocked because
a naive implementation would put the id in a global space and
would ultimately require having a namespace for the names of
namespaces, making migration and certain virtualization tricks
impossible.

We still don't have per superblock inode numbers for proc, which
appears necessary for application unaware checkpoint/restart and
migrations (if the application is using namespace file descriptors)
but that is now allowd by the design if it becomes important.

I have preallocated the ipc and uts initial proc inode numbers so
their structures can be statically initialized.

Signed-off-by: Eric W. Biederman <[email protected]>

fs/mount.h

@@ -4,6 +4,7 @@
 
 struct mnt_namespace {
 	atomic_t		count;
+	unsigned int		proc_inum;
 	struct mount *	root;
 	struct list_head	list;
 	struct user_namespace	*user_ns;

fs/namespace.c

@@ -2301,6 +2301,7 @@ long do_mount(const char *dev_name, const char *dir_name,
 
 static void free_mnt_ns(struct mnt_namespace *ns)
 {
+	proc_free_inum(ns->proc_inum);
 	put_user_ns(ns->user_ns);
 	kfree(ns);
 }
@@ -2317,10 +2318,16 @@ static atomic64_t mnt_ns_seq = ATOMIC64_INIT(1);
 static struct mnt_namespace *alloc_mnt_ns(struct user_namespace *user_ns)
 {
 	struct mnt_namespace *new_ns;
+	int ret;
 
 	new_ns = kmalloc(sizeof(struct mnt_namespace), GFP_KERNEL);
 	if (!new_ns)
 		return ERR_PTR(-ENOMEM);
+	ret = proc_alloc_inum(&new_ns->proc_inum);
+	if (ret) {
+		kfree(new_ns);
+		return ERR_PTR(ret);
+	}
 	new_ns->seq = atomic64_add_return(1, &mnt_ns_seq);
 	atomic_set(&new_ns->count, 1);
 	new_ns->root = NULL;
@@ -2799,10 +2806,17 @@ static int mntns_install(struct nsproxy *nsproxy, void *ns)
 	return 0;
 }
 
+static unsigned int mntns_inum(void *ns)
+{
+	struct mnt_namespace *mnt_ns = ns;
+	return mnt_ns->proc_inum;
+}
+
 const struct proc_ns_operations mntns_operations = {
 	.name		= "mnt",
 	.type		= CLONE_NEWNS,
 	.get		= mntns_get,
 	.put		= mntns_put,
 	.install	= mntns_install,
+	.inum		= mntns_inum,
 };

fs/proc/namespaces.c

@@ -82,21 +82,25 @@ static struct dentry *proc_ns_get_dentry(struct super_block *sb,
 		return ERR_PTR(-ENOMEM);
 	}
 
-	inode = new_inode(sb);
+	inode = iget_locked(sb, ns_ops->inum(ns));
 	if (!inode) {
 		dput(dentry);
 		ns_ops->put(ns);
 		return ERR_PTR(-ENOMEM);
 	}
 
 	ei = PROC_I(inode);
-	inode->i_ino = get_next_ino();
-	inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
-	inode->i_op = &ns_inode_operations;
-	inode->i_mode = S_IFREG | S_IRUGO;
-	inode->i_fop = &ns_file_operations;
-	ei->ns_ops = ns_ops;
-	ei->ns = ns;
+	if (inode->i_state & I_NEW) {
+		inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
+		inode->i_op = &ns_inode_operations;
+		inode->i_mode = S_IFREG | S_IRUGO;
+		inode->i_fop = &ns_file_operations;
+		ei->ns_ops = ns_ops;
+		ei->ns = ns;
+		unlock_new_inode(inode);
+	} else {
+		ns_ops->put(ns);
+	}
 
 	d_set_d_op(dentry, &ns_dentry_operations);
 	result = d_instantiate_unique(dentry, inode);
@@ -162,12 +166,12 @@ static int proc_ns_readlink(struct dentry *dentry, char __user *buffer, int bufl
 	if (!ns)
 		goto out_put_task;
 
-	snprintf(name, sizeof(name), "%s", ns_ops->name);
+	snprintf(name, sizeof(name), "%s:[%u]", ns_ops->name, ns_ops->inum(ns));
 	len = strlen(name);
 
 	if (len > buflen)
 		len = buflen;
-	if (copy_to_user(buffer, ns_ops->name, len))
+	if (copy_to_user(buffer, name, len))
 		len = -EFAULT;
 
 	ns_ops->put(ns);

include/linux/ipc_namespace.h

@@ -67,6 +67,8 @@ struct ipc_namespace {
 
 	/* user_ns which owns the ipc ns */
 	struct user_namespace *user_ns;
+
+	unsigned int	proc_inum;
 };
 
 extern struct ipc_namespace init_ipc_ns;

include/linux/pid_namespace.h

@@ -37,6 +37,7 @@ struct pid_namespace {
 	kgid_t pid_gid;
 	int hide_pid;
 	int reboot;	/* group exit code if this pidns was rebooted */
+	unsigned int proc_inum;
 };
 
 extern struct pid_namespace init_pid_ns;

include/linux/proc_fs.h

@@ -28,7 +28,11 @@ struct mm_struct;
  */
 
 enum {
-	PROC_ROOT_INO = 1,
+	PROC_ROOT_INO		= 1,
+	PROC_IPC_INIT_INO	= 0xEFFFFFFFU,
+	PROC_UTS_INIT_INO	= 0xEFFFFFFEU,
+	PROC_USER_INIT_INO	= 0xEFFFFFFDU,
+	PROC_PID_INIT_INO	= 0xEFFFFFFCU,
 };
 
 /*
@@ -263,6 +267,7 @@ struct proc_ns_operations {
 	void *(*get)(struct task_struct *task);
 	void (*put)(void *ns);
 	int (*install)(struct nsproxy *nsproxy, void *ns);
+	unsigned int (*inum)(void *ns);
 };
 extern const struct proc_ns_operations netns_operations;
 extern const struct proc_ns_operations utsns_operations;

include/linux/user_namespace.h

@@ -25,6 +25,7 @@ struct user_namespace {
 	struct user_namespace	*parent;
 	kuid_t			owner;
 	kgid_t			group;
+	unsigned int		proc_inum;
 };
 
 extern struct user_namespace init_user_ns;

include/linux/utsname.h

@@ -23,6 +23,7 @@ struct uts_namespace {
 	struct kref kref;
 	struct new_utsname name;
 	struct user_namespace *user_ns;
+	unsigned int proc_inum;
 };
 extern struct uts_namespace init_uts_ns;
 

include/net/net_namespace.h

@@ -56,6 +56,8 @@ struct net {
 
 	struct user_namespace   *user_ns;	/* Owning user namespace */
 
+	unsigned int		proc_inum;
+
 	struct proc_dir_entry 	*proc_net;
 	struct proc_dir_entry 	*proc_net_stat;
 

init/version.c

@@ -12,6 +12,7 @@
 #include <linux/utsname.h>
 #include <generated/utsrelease.h>
 #include <linux/version.h>
+#include <linux/proc_fs.h>
 
 #ifndef CONFIG_KALLSYMS
 #define version(a) Version_ ## a
@@ -34,6 +35,7 @@ struct uts_namespace init_uts_ns = {
 		.domainname	= UTS_DOMAINNAME,
 	},
 	.user_ns = &init_user_ns,
+	.proc_inum = PROC_UTS_INIT_INO,
 };
 EXPORT_SYMBOL_GPL(init_uts_ns);
 

ipc/msgutil.c

@@ -16,6 +16,7 @@
 #include <linux/msg.h>
 #include <linux/ipc_namespace.h>
 #include <linux/utsname.h>
+#include <linux/proc_fs.h>
 #include <asm/uaccess.h>
 
 #include "util.h"
@@ -30,6 +31,7 @@ DEFINE_SPINLOCK(mq_lock);
 struct ipc_namespace init_ipc_ns = {
 	.count		= ATOMIC_INIT(1),
 	.user_ns = &init_user_ns,
+	.proc_inum = PROC_IPC_INIT_INO,
 };
 
 atomic_t nr_ipc_ns = ATOMIC_INIT(1);

ipc/namespace.c

@@ -26,9 +26,16 @@ static struct ipc_namespace *create_ipc_ns(struct user_namespace *user_ns,
 	if (ns == NULL)
 		return ERR_PTR(-ENOMEM);
 
+	err = proc_alloc_inum(&ns->proc_inum);
+	if (err) {
+		kfree(ns);
+		return ERR_PTR(err);
+	}
+
 	atomic_set(&ns->count, 1);
 	err = mq_init_ns(ns);
 	if (err) {
+		proc_free_inum(ns->proc_inum);
 		kfree(ns);
 		return ERR_PTR(err);
 	}
@@ -111,6 +118,7 @@ static void free_ipc_ns(struct ipc_namespace *ns)
 	 */
 	ipcns_notify(IPCNS_REMOVED);
 	put_user_ns(ns->user_ns);
+	proc_free_inum(ns->proc_inum);
 	kfree(ns);
 }
 
@@ -172,10 +180,18 @@ static int ipcns_install(struct nsproxy *nsproxy, void *new)
 	return 0;
 }
 
+static unsigned int ipcns_inum(void *vp)
+{
+	struct ipc_namespace *ns = vp;
+
+	return ns->proc_inum;
+}
+
 const struct proc_ns_operations ipcns_operations = {
 	.name		= "ipc",
 	.type		= CLONE_NEWIPC,
 	.get		= ipcns_get,
 	.put		= ipcns_put,
 	.install	= ipcns_install,
+	.inum		= ipcns_inum,
 };

kernel/pid.c

@@ -80,6 +80,7 @@ struct pid_namespace init_pid_ns = {
 	.level = 0,
 	.child_reaper = &init_task,
 	.user_ns = &init_user_ns,
+	.proc_inum = PROC_PID_INIT_INO,
 };
 EXPORT_SYMBOL_GPL(init_pid_ns);
 

kernel/pid_namespace.c

@@ -107,6 +107,10 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns
 	if (ns->pid_cachep == NULL)
 		goto out_free_map;
 
+	err = proc_alloc_inum(&ns->proc_inum);
+	if (err)
+		goto out_free_map;
+
 	kref_init(&ns->kref);
 	ns->level = level;
 	ns->parent = get_pid_ns(parent_pid_ns);
@@ -133,6 +137,7 @@ static void destroy_pid_namespace(struct pid_namespace *ns)
 {
 	int i;
 
+	proc_free_inum(ns->proc_inum);
 	for (i = 0; i < PIDMAP_ENTRIES; i++)
 		kfree(ns->pidmap[i].page);
 	put_user_ns(ns->user_ns);
@@ -345,12 +350,19 @@ static int pidns_install(struct nsproxy *nsproxy, void *ns)
 	return 0;
 }
 
+static unsigned int pidns_inum(void *ns)
+{
+	struct pid_namespace *pid_ns = ns;
+	return pid_ns->proc_inum;
+}
+
 const struct proc_ns_operations pidns_operations = {
 	.name		= "pid",
 	.type		= CLONE_NEWPID,
 	.get		= pidns_get,
 	.put		= pidns_put,
 	.install	= pidns_install,
+	.inum		= pidns_inum,
 };
 
 static __init int pid_namespaces_init(void)

kernel/user.c

@@ -16,6 +16,7 @@
 #include <linux/interrupt.h>
 #include <linux/export.h>
 #include <linux/user_namespace.h>
+#include <linux/proc_fs.h>
 
 /*
  * userns count is 1 for root user, 1 for init_uts_ns,
@@ -51,6 +52,7 @@ struct user_namespace init_user_ns = {
 	},
 	.owner = GLOBAL_ROOT_UID,
 	.group = GLOBAL_ROOT_GID,
+	.proc_inum = PROC_USER_INIT_INO,
 };
 EXPORT_SYMBOL_GPL(init_user_ns);
 

kernel/user_namespace.c

@@ -58,6 +58,7 @@ int create_user_ns(struct cred *new)
 	struct user_namespace *ns, *parent_ns = new->user_ns;
 	kuid_t owner = new->euid;
 	kgid_t group = new->egid;
+	int ret;
 
 	/* The creator needs a mapping in the parent user namespace
 	 * or else we won't be able to reasonably tell userspace who
@@ -71,6 +72,12 @@ int create_user_ns(struct cred *new)
 	if (!ns)
 		return -ENOMEM;
 
+	ret = proc_alloc_inum(&ns->proc_inum);
+	if (ret) {
+		kmem_cache_free(user_ns_cachep, ns);
+		return ret;
+	}
+
 	kref_init(&ns->kref);
 	/* Leave the new->user_ns reference with the new user namespace. */
 	ns->parent = parent_ns;
@@ -103,6 +110,7 @@ void free_user_ns(struct kref *kref)
 		container_of(kref, struct user_namespace, kref);
 
 	parent = ns->parent;
+	proc_free_inum(ns->proc_inum);
 	kmem_cache_free(user_ns_cachep, ns);
 	put_user_ns(parent);
 }
@@ -808,12 +816,19 @@ static int userns_install(struct nsproxy *nsproxy, void *ns)
 	return commit_creds(cred);
 }
 
+static unsigned int userns_inum(void *ns)
+{
+	struct user_namespace *user_ns = ns;
+	return user_ns->proc_inum;
+}
+
 const struct proc_ns_operations userns_operations = {
 	.name		= "user",
 	.type		= CLONE_NEWUSER,
 	.get		= userns_get,
 	.put		= userns_put,
 	.install	= userns_install,
+	.inum		= userns_inum,
 };
 
 static __init int user_namespaces_init(void)

kernel/utsname.c

@@ -36,11 +36,18 @@ static struct uts_namespace *clone_uts_ns(struct user_namespace *user_ns,
 					  struct uts_namespace *old_ns)
 {
 	struct uts_namespace *ns;
+	int err;
 
 	ns = create_uts_ns();
 	if (!ns)
 		return ERR_PTR(-ENOMEM);
 
+	err = proc_alloc_inum(&ns->proc_inum);
+	if (err) {
+		kfree(ns);
+		return ERR_PTR(err);
+	}
+
 	down_read(&uts_sem);
 	memcpy(&ns->name, &old_ns->name, sizeof(ns->name));
 	ns->user_ns = get_user_ns(user_ns);
@@ -77,6 +84,7 @@ void free_uts_ns(struct kref *kref)
 
 	ns = container_of(kref, struct uts_namespace, kref);
 	put_user_ns(ns->user_ns);
+	proc_free_inum(ns->proc_inum);
 	kfree(ns);
 }
 
@@ -114,11 +122,18 @@ static int utsns_install(struct nsproxy *nsproxy, void *new)
 	return 0;
 }
 
+static unsigned int utsns_inum(void *vp)
+{
+	struct uts_namespace *ns = vp;
+
+	return ns->proc_inum;
+}
+
 const struct proc_ns_operations utsns_operations = {
 	.name		= "uts",
 	.type		= CLONE_NEWUTS,
 	.get		= utsns_get,
 	.put		= utsns_put,
 	.install	= utsns_install,
+	.inum		= utsns_inum,
 };
-