procfs: switch /proc/self away from proc_dir_entry

Just have it pinned in dcache all along and let procfs ->kill_sb()
drop it before kill_anon_super().

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

fs/proc/base.c

@@ -2794,7 +2794,7 @@ static struct tgid_iter next_tgid(struct pid_namespace *ns, struct tgid_iter ite
 	return iter;
 }
 
-#define TGID_OFFSET (FIRST_PROCESS_ENTRY)
+#define TGID_OFFSET (FIRST_PROCESS_ENTRY + 1)
 
 static int proc_pid_fill_cache(struct file *filp, void *dirent, filldir_t filldir,
 	struct tgid_iter iter)
@@ -2817,13 +2817,21 @@ int proc_pid_readdir(struct file * filp, void * dirent, filldir_t filldir)
 	struct tgid_iter iter;
 	struct pid_namespace *ns;
 	filldir_t __filldir;
+	loff_t pos = filp->f_pos;
 
-	if (filp->f_pos >= PID_MAX_LIMIT + TGID_OFFSET)
+	if (pos >= PID_MAX_LIMIT + TGID_OFFSET)
 		goto out;
 
-	ns = filp->f_dentry->d_sb->s_fs_info;
+	if (pos == TGID_OFFSET - 1) {
+		if (proc_fill_cache(filp, dirent, filldir, "self", 4,
+					NULL, NULL, NULL) < 0)
+			goto out;
+		iter.tgid = 0;
+	} else {
+		iter.tgid = pos - TGID_OFFSET;
+	}
 	iter.task = NULL;
-	iter.tgid = filp->f_pos - TGID_OFFSET;
+	ns = filp->f_dentry->d_sb->s_fs_info;
 	for (iter = next_tgid(ns, iter);
 	     iter.task;
 	     iter.tgid += 1, iter = next_tgid(ns, iter)) {

fs/proc/inode.c

@@ -506,5 +506,5 @@ int proc_fill_super(struct super_block *s)
 		return -ENOMEM;
 	}
 
-	return 0;
+	return proc_setup_self(s);
 }

fs/proc/internal.h

@@ -205,3 +205,4 @@ int proc_setattr(struct dentry *dentry, struct iattr *attr);
 extern const struct inode_operations proc_ns_dir_inode_operations;
 extern const struct file_operations proc_ns_dir_operations;
 
+extern int proc_setup_self(struct super_block *);

fs/proc/root.c

@@ -137,6 +137,8 @@ static void proc_kill_sb(struct super_block *sb)
 	struct pid_namespace *ns;
 
 	ns = (struct pid_namespace *)sb->s_fs_info;
+	if (ns->proc_self)
+		dput(ns->proc_self);
 	kill_anon_super(sb);
 	put_pid_ns(ns);
 }

fs/proc/self.c

@@ -1,6 +1,7 @@
-#include <linux/proc_fs.h>
 #include <linux/sched.h>
 #include <linux/namei.h>
+#include <linux/pid_namespace.h>
+#include "internal.h"
 
 /*
  * /proc/self:
@@ -48,12 +49,43 @@ static const struct inode_operations proc_self_inode_operations = {
 	.put_link	= proc_self_put_link,
 };
 
-void __init proc_self_init(void)
+static unsigned self_inum;
+
+int proc_setup_self(struct super_block *s)
 {
-	struct proc_dir_entry *proc_self_symlink;
-	umode_t mode;
+	struct inode *root_inode = s->s_root->d_inode;
+	struct pid_namespace *ns = s->s_fs_info;
+	struct dentry *self;
+
+	mutex_lock(&root_inode->i_mutex);
+	self = d_alloc_name(s->s_root, "self");
+	if (self) {
+		struct inode *inode = new_inode_pseudo(s);
+		if (inode) {
+			inode->i_ino = self_inum;
+			inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
+			inode->i_mode = S_IFLNK | S_IRWXUGO;
+			inode->i_uid = GLOBAL_ROOT_UID;
+			inode->i_gid = GLOBAL_ROOT_GID;
+			inode->i_op = &proc_self_inode_operations;
+			d_add(self, inode);
+		} else {
+			dput(self);
+			self = ERR_PTR(-ENOMEM);
+		}
+	} else {
+		self = ERR_PTR(-ENOMEM);
+	}
+	mutex_unlock(&root_inode->i_mutex);
+	if (IS_ERR(self)) {
+		pr_err("proc_fill_super: can't allocate /proc/self\n");
+		return PTR_ERR(self);
+	}
+	ns->proc_self = self;
+	return 0;
+}
 
-	mode = S_IFLNK | S_IRWXUGO;
-	proc_self_symlink = proc_create("self", mode, NULL, NULL );
-	proc_self_symlink->proc_iops = &proc_self_inode_operations;
+void __init proc_self_init(void)
+{
+	proc_alloc_inum(&self_inum);
 }

include/linux/pid_namespace.h

@@ -28,6 +28,7 @@ struct pid_namespace {
 	struct pid_namespace *parent;
 #ifdef CONFIG_PROC_FS
 	struct vfsmount *proc_mnt;
+	struct dentry *proc_self;
 #endif
 #ifdef CONFIG_BSD_PROCESS_ACCT
 	struct bsd_acct_struct *bacct;