Merge branch 'proc-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace

Pull proc updates from Eric Biederman:
 "This has four sets of changes:

   - modernize proc to support multiple private instances

   - ensure we see the exit of each process tid exactly

   - remove has_group_leader_pid

   - use pids not tasks in posix-cpu-timers lookup

  Alexey updated proc so each mount of proc uses a new superblock. This
  allows people to actually use mount options with proc with no fear of
  messing up another mount of proc. Given the kernel's internal mounts
  of proc for things like uml this was a real problem, and resulted in
  Android's hidepid mount options being ignored and introducing security
  issues.

  The rest of the changes are small cleanups and fixes that came out of
  my work to allow this change to proc. In essence it is swapping the
  pids in de_thread during exec which removes a special case the code
  had to handle. Then updating the code to stop handling that special
  case"

* 'proc-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace:
  proc: proc_pid_ns takes super_block as an argument
  remove the no longer needed pid_alive() check in __task_pid_nr_ns()
  posix-cpu-timers: Replace __get_task_for_clock with pid_for_clock
  posix-cpu-timers: Replace cpu_timer_pid_type with clock_pid_type
  posix-cpu-timers: Extend rcu_read_lock removing task_struct references
  signal: Remove has_group_leader_pid
  exec: Remove BUG_ON(has_group_leader_pid)
  posix-cpu-timer:  Unify the now redundant code in lookup_task
  posix-cpu-timer: Tidy up group_leader logic in lookup_task
  proc: Ensure we see the exit of each process tid exactly once
  rculist: Add hlists_swap_heads_rcu
  proc: Use PIDTYPE_TGID in next_tgid
  Use proc_pid_ns() to get pid_namespace from the proc superblock
  proc: use named enums for better readability
  proc: use human-readable values for hidepid
  docs: proc: add documentation for "hidepid=4" and "subset=pid" options and new mount behavior
  proc: add option to mount only a pids subset
  proc: instantiate only pids that we can ptrace on 'hidepid=4' mount option
  proc: allow to mount many instances of proc in one pid namespace
  proc: rename struct proc_fs_info to proc_fs_opts

5 files changed, 52 insertions, 25 deletions

diff --git a/include/linux/pid.h b/include/linux/pid.h
index 93543cbc0e6b..176d6cf80e7c 100644
--- a/include/linux/pid.h
+++ b/include/linux/pid.h
@@ -102,6 +102,7 @@ extern void attach_pid(struct task_struct *task, enum pid_type);
 extern void detach_pid(struct task_struct *task, enum pid_type);
 extern void change_pid(struct task_struct *task, enum pid_type,
 			struct pid *pid);
+extern void exchange_tids(struct task_struct *task, struct task_struct *old);
 extern void transfer_pid(struct task_struct *old, struct task_struct *new,
 			 enum pid_type);
 
diff --git a/include/linux/pid_namespace.h b/include/linux/pid_namespace.h
index 4956e362e55e..5a5cb45ac57e 100644
--- a/include/linux/pid_namespace.h
+++ b/include/linux/pid_namespace.h
@@ -17,12 +17,6 @@
 
 struct fs_pin;
 
-enum { /* definitions for pid_namespace's hide_pid field */
-	HIDEPID_OFF	  = 0,
-	HIDEPID_NO_ACCESS = 1,
-	HIDEPID_INVISIBLE = 2,
-};
-
 struct pid_namespace {
 	struct kref kref;
 	struct idr idr;
@@ -32,17 +26,11 @@ struct pid_namespace {
 	struct kmem_cache *pid_cachep;
 	unsigned int level;
 	struct pid_namespace *parent;
-#ifdef CONFIG_PROC_FS
-	struct dentry *proc_self;
-	struct dentry *proc_thread_self;
-#endif
 #ifdef CONFIG_BSD_PROCESS_ACCT
 	struct fs_pin *bacct;
 #endif
 	struct user_namespace *user_ns;
 	struct ucounts *ucounts;
-	kgid_t pid_gid;
-	int hide_pid;
 	int reboot;	/* group exit code if this pidns was rebooted */
 	struct ns_common ns;
 } __randomize_layout;
diff --git a/include/linux/proc_fs.h b/include/linux/proc_fs.h
index ad4ff711fc02..d1eed1b43651 100644
--- a/include/linux/proc_fs.h
+++ b/include/linux/proc_fs.h
@@ -42,6 +42,34 @@ struct proc_ops {
 	unsigned long (*proc_get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
 } __randomize_layout;
 
+/* definitions for hide_pid field */
+enum proc_hidepid {
+	HIDEPID_OFF	  = 0,
+	HIDEPID_NO_ACCESS = 1,
+	HIDEPID_INVISIBLE = 2,
+	HIDEPID_NOT_PTRACEABLE = 4, /* Limit pids to only ptraceable pids */
+};
+
+/* definitions for proc mount option pidonly */
+enum proc_pidonly {
+	PROC_PIDONLY_OFF = 0,
+	PROC_PIDONLY_ON  = 1,
+};
+
+struct proc_fs_info {
+	struct pid_namespace *pid_ns;
+	struct dentry *proc_self;        /* For /proc/self */
+	struct dentry *proc_thread_self; /* For /proc/thread-self */
+	kgid_t pid_gid;
+	enum proc_hidepid hide_pid;
+	enum proc_pidonly pidonly;
+};
+
+static inline struct proc_fs_info *proc_sb_info(struct super_block *sb)
+{
+	return sb->s_fs_info;
+}
+
 #ifdef CONFIG_PROC_FS
 
 typedef int (*proc_write_t)(struct file *, char *, size_t);
@@ -177,9 +205,9 @@ int open_related_ns(struct ns_common *ns,
 		   struct ns_common *(*get_ns)(struct ns_common *ns));
 
 /* get the associated pid namespace for a file in procfs */
-static inline struct pid_namespace *proc_pid_ns(const struct inode *inode)
+static inline struct pid_namespace *proc_pid_ns(struct super_block *sb)
 {
-	return inode->i_sb->s_fs_info;
+	return proc_sb_info(sb)->pid_ns;
 }
 
 bool proc_ns_file(const struct file *file);
diff --git a/include/linux/rculist.h b/include/linux/rculist.h
index 7375bb3da140..df587d181844 100644
--- a/include/linux/rculist.h
+++ b/include/linux/rculist.h
@@ -506,6 +506,27 @@ static inline void hlist_replace_rcu(struct hlist_node *old,
 	WRITE_ONCE(old->pprev, LIST_POISON2);
 }
 
+/**
+ * hlists_swap_heads_rcu - swap the lists the hlist heads point to
+ * @left:  The hlist head on the left
+ * @right: The hlist head on the right
+ *
+ * The lists start out as [@left  ][node1 ... ] and
+                          [@right ][node2 ... ]
+ * The lists end up as    [@left  ][node2 ... ]
+ *                        [@right ][node1 ... ]
+ */
+static inline void hlists_swap_heads_rcu(struct hlist_head *left, struct hlist_head *right)
+{
+	struct hlist_node *node1 = left->first;
+	struct hlist_node *node2 = right->first;
+
+	rcu_assign_pointer(left->first, node2);
+	rcu_assign_pointer(right->first, node1);
+	WRITE_ONCE(node2->pprev, &left->first);
+	WRITE_ONCE(node1->pprev, &right->first);
+}
+
 /*
  * return the first or the next element in an RCU protected hlist
  */
diff --git a/include/linux/sched/signal.h b/include/linux/sched/signal.h
index 3e5b090c16d4..0ee5e696c5d8 100644
--- a/include/linux/sched/signal.h
+++ b/include/linux/sched/signal.h
@@ -654,17 +654,6 @@ static inline bool thread_group_leader(struct task_struct *p)
 	return p->exit_signal >= 0;
 }
 
-/* Do to the insanities of de_thread it is possible for a process
- * to have the pid of the thread group leader without actually being
- * the thread group leader.  For iteration through the pids in proc
- * all we care about is that we have a task with the appropriate
- * pid, we don't actually care if we have the right task.
- */
-static inline bool has_group_leader_pid(struct task_struct *p)
-{
-	return task_pid(p) == task_tgid(p);
-}
-
 static inline
 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
 {