[PATCH 0/3] fs: Support compiling out the pivot_root syscall

[PATCH 0/3] fs: Support compiling out the pivot_root syscall

[bosrsf04]

From: Brian Ashworth <bosrsf04@gmail.com>

This patch series will allow for the pivot_root syscall to be made
optional. The first patch refactors the functions that are required
by pivot_root so they can be accessed outside of fs/namespace.c. The
second patch moves the pivot_root syscall to its own file. The third
adds a Kconfig option and conditional compilation to the Makefile.

The pivot_root syscall is not needed on systems that do not use
any intermediate filesystem. Allowing for pivot_root to be
ommitted from the kernel will aid in the tinification efforts.

Brian Ashworth (3):
  Makes functions used by pivot_root accessible
  Extracts pivot_root so it can be made optional
  Allows for the pivot_root syscall to be omitted

 fs/Makefile     |   2 +
 fs/mount.h      |  24 +++++++++
 fs/namespace.c  | 150 +++-----------------------------------------------------
 fs/pivot_root.c | 129 ++++++++++++++++++++++++++++++++++++++++++++++++
 init/Kconfig    |  10 ++++
 kernel/sys_ni.c |   1 +
 6 files changed, 172 insertions(+), 144 deletions(-)
 create mode 100644 fs/pivot_root.c

-- 
2.11.1

[PATCH 1/3] fs: Makes functions used by pivot_root accessible

[bosrsf04]

From: Brian Ashworth <bosrsf04@gmail.com>

add/remove: 1/0 grow/shrink: 0/2 up/down: 19/-44 (-25)
function                                     old     new   delta
detach_mnt                                     -      19     +19
attach_recursive_mnt                         349     331     -18
sys_pivot_root                               636     610     -26
Total: Before=1899598, After=1899573, chg -0.00%

Signed-off-by: Brian Ashworth <bosrsf04@gmail.com>
---
 fs/mount.h     | 24 ++++++++++++++++++++++++
 fs/namespace.c | 27 ++++++---------------------
 2 files changed, 30 insertions(+), 21 deletions(-)

diff --git a/fs/mount.h b/fs/mount.h
index 2c856fc47ae3..c228d0c233e8 100644
--- a/fs/mount.h
+++ b/fs/mount.h
@@ -3,6 +3,7 @@
 #include <linux/poll.h>
 #include <linux/ns_common.h>
 #include <linux/fs_pin.h>
+#include <linux/nsproxy.h>
 
 struct mnt_namespace {
 	atomic_t		count;
@@ -145,3 +146,26 @@ static inline bool is_local_mountpoint(struct dentry *dentry)
 
 	return __is_local_mountpoint(dentry);
 }
+
+/*
+ * Is the caller allowed to modify his namespace?
+ */
+static inline bool may_mount(void)
+{
+	return ns_capable(current->nsproxy->mnt_ns->user_ns, CAP_SYS_ADMIN);
+}
+
+static inline int check_mnt(struct mount *mnt)
+{
+	return mnt->mnt_ns == current->nsproxy->mnt_ns;
+}
+
+struct mountpoint *lock_mount(struct path *path);
+void unlock_mount(struct mountpoint *where);
+
+void detach_mnt(struct mount *mnt, struct path *old_path);
+void attach_mnt(struct mount *mnt, struct mount *parent, struct mountpoint *mp);
+
+void touch_mnt_namespace(struct mnt_namespace *ns);
+
+void put_mountpoint(struct mountpoint *mp);
diff --git a/fs/namespace.c b/fs/namespace.c
index 487ba30bb5c6..d49d615e30a1 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -789,7 +789,7 @@ static struct mountpoint *get_mountpoint(struct dentry *dentry)
 	return mp;
 }
 
-static void put_mountpoint(struct mountpoint *mp)
+void put_mountpoint(struct mountpoint *mp)
 {
 	if (!--mp->m_count) {
 		struct dentry *dentry = mp->m_dentry;
@@ -802,15 +802,10 @@ static void put_mountpoint(struct mountpoint *mp)
 	}
 }
 
-static inline int check_mnt(struct mount *mnt)
-{
-	return mnt->mnt_ns == current->nsproxy->mnt_ns;
-}
-
 /*
  * vfsmount lock must be held for write
  */
-static void touch_mnt_namespace(struct mnt_namespace *ns)
+void touch_mnt_namespace(struct mnt_namespace *ns)
 {
 	if (ns) {
 		ns->event = ++event;
@@ -846,7 +841,7 @@ static void unhash_mnt(struct mount *mnt)
 /*
  * vfsmount lock must be held for write
  */
-static void detach_mnt(struct mount *mnt, struct path *old_path)
+void detach_mnt(struct mount *mnt, struct path *old_path)
 {
 	old_path->dentry = mnt->mnt_mountpoint;
 	old_path->mnt = &mnt->mnt_parent->mnt;
@@ -881,9 +876,7 @@ void mnt_set_mountpoint(struct mount *mnt,
 /*
  * vfsmount lock must be held for write
  */
-static void attach_mnt(struct mount *mnt,
-			struct mount *parent,
-			struct mountpoint *mp)
+void attach_mnt(struct mount *mnt, struct mount *parent, struct mountpoint *mp)
 {
 	mnt_set_mountpoint(parent, mp, mnt);
 	hlist_add_head_rcu(&mnt->mnt_hash, m_hash(&parent->mnt, mp->m_dentry));
@@ -1639,14 +1632,6 @@ void __detach_mounts(struct dentry *dentry)
 	namespace_unlock();
 }
 
-/* 
- * Is the caller allowed to modify his namespace?
- */
-static inline bool may_mount(void)
-{
-	return ns_capable(current->nsproxy->mnt_ns->user_ns, CAP_SYS_ADMIN);
-}
-
 static inline bool may_mandlock(void)
 {
 #ifndef	CONFIG_MANDATORY_FILE_LOCKING
@@ -2049,7 +2034,7 @@ static int attach_recursive_mnt(struct mount *source_mnt,
 	return err;
 }
 
-static struct mountpoint *lock_mount(struct path *path)
+struct mountpoint *lock_mount(struct path *path)
 {
 	struct vfsmount *mnt;
 	struct dentry *dentry = path->dentry;
@@ -2078,7 +2063,7 @@ static struct mountpoint *lock_mount(struct path *path)
 	goto retry;
 }
 
-static void unlock_mount(struct mountpoint *where)
+void unlock_mount(struct mountpoint *where)
 {
 	struct dentry *dentry = where->m_dentry;
 
-- 
2.11.1

[PATCH 2/3] fs: Extracts pivot_root so it can be made optional

[bosrsf04]

From: Brian Ashworth <bosrsf04@gmail.com>

add/remove: 0/0 grow/shrink: 1/0 up/down: 4/0 (4)
function                                     old     new   delta
sys_pivot_root                               610     614      +4
Total: Before=1899573, After=1899577, chg +0.00%

Signed-off-by Brian Ashworth <bosrsf04@gmail.com>
---
 fs/Makefile     |   2 +-
 fs/namespace.c  | 123 -----------------------------------------------------
 fs/pivot_root.c | 129 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 130 insertions(+), 124 deletions(-)
 create mode 100644 fs/pivot_root.c

diff --git a/fs/Makefile b/fs/Makefile
index 7bbaca9c67b1..34cb58c4127d 100644
--- a/fs/Makefile
+++ b/fs/Makefile
@@ -11,7 +11,7 @@ obj-y :=	open.o read_write.o file_table.o super.o \
 		attr.o bad_inode.o file.o filesystems.o namespace.o \
 		seq_file.o xattr.o libfs.o fs-writeback.o \
 		pnode.o splice.o sync.o utimes.o \
-		stack.o fs_struct.o statfs.o fs_pin.o nsfs.o
+		stack.o fs_struct.o statfs.o fs_pin.o nsfs.o pivot_root.o
 
 ifeq ($(CONFIG_BLOCK),y)
 obj-y +=	buffer.o block_dev.o direct-io.o mpage.o
diff --git a/fs/namespace.c b/fs/namespace.c
index d49d615e30a1..36e4faf4c6e6 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -3043,129 +3043,6 @@ bool path_is_under(const struct path *path1, const struct path *path2)
 }
 EXPORT_SYMBOL(path_is_under);
 
-/*
- * pivot_root Semantics:
- * Moves the root file system of the current process to the directory put_old,
- * makes new_root as the new root file system of the current process, and sets
- * root/cwd of all processes which had them on the current root to new_root.
- *
- * Restrictions:
- * The new_root and put_old must be directories, and  must not be on the
- * same file  system as the current process root. The put_old  must  be
- * underneath new_root,  i.e. adding a non-zero number of /.. to the string
- * pointed to by put_old must yield the same directory as new_root. No other
- * file system may be mounted on put_old. After all, new_root is a mountpoint.
- *
- * Also, the current root cannot be on the 'rootfs' (initial ramfs) filesystem.
- * See Documentation/filesystems/ramfs-rootfs-initramfs.txt for alternatives
- * in this situation.
- *
- * Notes:
- *  - we don't move root/cwd if they are not at the root (reason: if something
- *    cared enough to change them, it's probably wrong to force them elsewhere)
- *  - it's okay to pick a root that isn't the root of a file system, e.g.
- *    /nfs/my_root where /nfs is the mount point. It must be a mountpoint,
- *    though, so you may need to say mount --bind /nfs/my_root /nfs/my_root
- *    first.
- */
-SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
-		const char __user *, put_old)
-{
-	struct path new, old, parent_path, root_parent, root;
-	struct mount *new_mnt, *root_mnt, *old_mnt;
-	struct mountpoint *old_mp, *root_mp;
-	int error;
-
-	if (!may_mount())
-		return -EPERM;
-
-	error = user_path_dir(new_root, &new);
-	if (error)
-		goto out0;
-
-	error = user_path_dir(put_old, &old);
-	if (error)
-		goto out1;
-
-	error = security_sb_pivotroot(&old, &new);
-	if (error)
-		goto out2;
-
-	get_fs_root(current->fs, &root);
-	old_mp = lock_mount(&old);
-	error = PTR_ERR(old_mp);
-	if (IS_ERR(old_mp))
-		goto out3;
-
-	error = -EINVAL;
-	new_mnt = real_mount(new.mnt);
-	root_mnt = real_mount(root.mnt);
-	old_mnt = real_mount(old.mnt);
-	if (IS_MNT_SHARED(old_mnt) ||
-		IS_MNT_SHARED(new_mnt->mnt_parent) ||
-		IS_MNT_SHARED(root_mnt->mnt_parent))
-		goto out4;
-	if (!check_mnt(root_mnt) || !check_mnt(new_mnt))
-		goto out4;
-	if (new_mnt->mnt.mnt_flags & MNT_LOCKED)
-		goto out4;
-	error = -ENOENT;
-	if (d_unlinked(new.dentry))
-		goto out4;
-	error = -EBUSY;
-	if (new_mnt == root_mnt || old_mnt == root_mnt)
-		goto out4; /* loop, on the same file system  */
-	error = -EINVAL;
-	if (root.mnt->mnt_root != root.dentry)
-		goto out4; /* not a mountpoint */
-	if (!mnt_has_parent(root_mnt))
-		goto out4; /* not attached */
-	root_mp = root_mnt->mnt_mp;
-	if (new.mnt->mnt_root != new.dentry)
-		goto out4; /* not a mountpoint */
-	if (!mnt_has_parent(new_mnt))
-		goto out4; /* not attached */
-	/* make sure we can reach put_old from new_root */
-	if (!is_path_reachable(old_mnt, old.dentry, &new))
-		goto out4;
-	/* make certain new is below the root */
-	if (!is_path_reachable(new_mnt, new.dentry, &root))
-		goto out4;
-	root_mp->m_count++; /* pin it so it won't go away */
-	lock_mount_hash();
-	detach_mnt(new_mnt, &parent_path);
-	detach_mnt(root_mnt, &root_parent);
-	if (root_mnt->mnt.mnt_flags & MNT_LOCKED) {
-		new_mnt->mnt.mnt_flags |= MNT_LOCKED;
-		root_mnt->mnt.mnt_flags &= ~MNT_LOCKED;
-	}
-	/* mount old root on put_old */
-	attach_mnt(root_mnt, old_mnt, old_mp);
-	/* mount new_root on / */
-	attach_mnt(new_mnt, real_mount(root_parent.mnt), root_mp);
-	touch_mnt_namespace(current->nsproxy->mnt_ns);
-	/* A moved mount should not expire automatically */
-	list_del_init(&new_mnt->mnt_expire);
-	put_mountpoint(root_mp);
-	unlock_mount_hash();
-	chroot_fs_refs(&root, &new);
-	error = 0;
-out4:
-	unlock_mount(old_mp);
-	if (!error) {
-		path_put(&root_parent);
-		path_put(&parent_path);
-	}
-out3:
-	path_put(&root);
-out2:
-	path_put(&old);
-out1:
-	path_put(&new);
-out0:
-	return error;
-}
-
 static void __init init_mount_tree(void)
 {
 	struct vfsmount *mnt;
diff --git a/fs/pivot_root.c b/fs/pivot_root.c
new file mode 100644
index 000000000000..a609b21a1438
--- /dev/null
+++ b/fs/pivot_root.c
@@ -0,0 +1,129 @@
+#include <linux/syscalls.h>
+#include <linux/security.h>
+#include <linux/namei.h>
+#include <linux/fs_struct.h>
+#include "pnode.h"
+#include "internal.h"
+
+/*
+ * pivot_root Semantics:
+ * Moves the root file system of the current process to the directory put_old,
+ * makes new_root as the new root file system of the current process, and sets
+ * root/cwd of all processes which had them on the current root to new_root.
+ *
+ * Restrictions:
+ * The new_root and put_old must be directories, and  must not be on the
+ * same file  system as the current process root. The put_old  must  be
+ * underneath new_root,  i.e. adding a non-zero number of /.. to the string
+ * pointed to by put_old must yield the same directory as new_root. No other
+ * file system may be mounted on put_old. After all, new_root is a mountpoint.
+ *
+ * Also, the current root cannot be on the 'rootfs' (initial ramfs) filesystem.
+ * See Documentation/filesystems/ramfs-rootfs-initramfs.txt for alternatives
+ * in this situation.
+ *
+ * Notes:
+ *  - we don't move root/cwd if they are not at the root (reason: if something
+ *    cared enough to change them, it's probably wrong to force them elsewhere)
+ *  - it's okay to pick a root that isn't the root of a file system, e.g.
+ *    /nfs/my_root where /nfs is the mount point. It must be a mountpoint,
+ *    though, so you may need to say mount --bind /nfs/my_root /nfs/my_root
+ *    first.
+ */
+SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
+		const char __user *, put_old)
+{
+	struct path new, old, parent_path, root_parent, root;
+	struct mount *new_mnt, *root_mnt, *old_mnt;
+	struct mountpoint *old_mp, *root_mp;
+	int error;
+
+	if (!may_mount())
+		return -EPERM;
+
+	error = user_path_dir(new_root, &new);
+	if (error)
+		goto out0;
+
+	error = user_path_dir(put_old, &old);
+	if (error)
+		goto out1;
+
+	error = security_sb_pivotroot(&old, &new);
+	if (error)
+		goto out2;
+
+	get_fs_root(current->fs, &root);
+	old_mp = lock_mount(&old);
+	error = PTR_ERR(old_mp);
+	if (IS_ERR(old_mp))
+		goto out3;
+
+	error = -EINVAL;
+	new_mnt = real_mount(new.mnt);
+	root_mnt = real_mount(root.mnt);
+	old_mnt = real_mount(old.mnt);
+	if (IS_MNT_SHARED(old_mnt) ||
+		IS_MNT_SHARED(new_mnt->mnt_parent) ||
+		IS_MNT_SHARED(root_mnt->mnt_parent))
+		goto out4;
+	if (!check_mnt(root_mnt) || !check_mnt(new_mnt))
+		goto out4;
+	if (new_mnt->mnt.mnt_flags & MNT_LOCKED)
+		goto out4;
+	error = -ENOENT;
+	if (d_unlinked(new.dentry))
+		goto out4;
+	error = -EBUSY;
+	if (new_mnt == root_mnt || old_mnt == root_mnt)
+		goto out4; /* loop, on the same file system  */
+	error = -EINVAL;
+	if (root.mnt->mnt_root != root.dentry)
+		goto out4; /* not a mountpoint */
+	if (!mnt_has_parent(root_mnt))
+		goto out4; /* not attached */
+	root_mp = root_mnt->mnt_mp;
+	if (new.mnt->mnt_root != new.dentry)
+		goto out4; /* not a mountpoint */
+	if (!mnt_has_parent(new_mnt))
+		goto out4; /* not attached */
+	/* make sure we can reach put_old from new_root */
+	if (!is_path_reachable(old_mnt, old.dentry, &new))
+		goto out4;
+	/* make certain new is below the root */
+	if (!is_path_reachable(new_mnt, new.dentry, &root))
+		goto out4;
+	root_mp->m_count++; /* pin it so it won't go away */
+	lock_mount_hash();
+	detach_mnt(new_mnt, &parent_path);
+	detach_mnt(root_mnt, &root_parent);
+	if (root_mnt->mnt.mnt_flags & MNT_LOCKED) {
+		new_mnt->mnt.mnt_flags |= MNT_LOCKED;
+		root_mnt->mnt.mnt_flags &= ~MNT_LOCKED;
+	}
+	/* mount old root on put_old */
+	attach_mnt(root_mnt, old_mnt, old_mp);
+	/* mount new_root on / */
+	attach_mnt(new_mnt, real_mount(root_parent.mnt), root_mp);
+	touch_mnt_namespace(current->nsproxy->mnt_ns);
+	/* A moved mount should not expire automatically */
+	list_del_init(&new_mnt->mnt_expire);
+	put_mountpoint(root_mp);
+	unlock_mount_hash();
+	chroot_fs_refs(&root, &new);
+	error = 0;
+out4:
+	unlock_mount(old_mp);
+	if (!error) {
+		path_put(&root_parent);
+		path_put(&parent_path);
+	}
+out3:
+	path_put(&root);
+out2:
+	path_put(&old);
+out1:
+	path_put(&new);
+out0:
+	return error;
+}
-- 
2.11.1

Re: [PATCH 2/3] fs: Extracts pivot_root so it can be made optional

[Theodore Ts'o]

If you're only going to be removing a single function, instead of
having to export a bunch of previously-static functions, my preference
would be to just insert a pair of #ifdef CONFIG_PIVOT_ROOT_SYSCALL / #endif
statements around the function in question.

Is it worth it to save 600-odd bytes?  Shrug; but if that's what you
are after, I'd suggest doing it the simplest and least-instrusive way
possible.

Cheers,

						- Ted

Re: [PATCH 2/3] fs: Extracts pivot_root so it can be made optional

[Al Viro]

On Sat, Feb 25, 2017 at 10:22:21AM -0500, Theodore Ts'o wrote:
> If you're only going to be removing a single function, instead of
> having to export a bunch of previously-static functions, my preference
> would be to just insert a pair of #ifdef CONFIG_PIVOT_ROOT_SYSCALL / #endif
> statements around the function in question.

Quite.  In this form: NAK for the reasons above.

[PATCH 0/1] fs: Support compiling out the pivot_root syscall

[Brian Ashworth]

This patch will allow for the pivot_root syscall to be made
optional. Based on feedback, this patch uses an ifdef in the
source rather than conditional compilation in the Makefile.

The pivot_root syscall is not needed on systems that do not use
any intermediate filesystem. Allowing for pivot_root to be
ommitted from the kernel will aid in the tinification efforts.

Brian Ashworth (1):
  Allows for the pivot_root syscall to be optional

 fs/namespace.c  |  2 ++
 init/Kconfig    | 10 ++++++++++
 kernel/sys_ni.c |  1 +
 3 files changed, 13 insertions(+)

-- 
2.11.1

[PATCH 1/1] fs: Allows for the pivot_root syscall to be optional

[Brian Ashworth]

The pivot_root syscall is not needed on systems that do not use
any intermediate filesystem. Allowing for pivot_root to be
ommitted from the kernel will aid in the tinification efforts.

Without CONFIG_PIVOT_ROOT_SYSCALL set
add/remove: 0/2 grow/shrink: 1/0 up/down: 45/-707 (-662)
function                                     old     new   delta
attach_recursive_mnt                         349     394     +45
attach_mnt                                    71       -     -71
sys_pivot_root                               636       -    -636
Total: Before=1899893, After=1899231, chg -0.03%

Signed-off-by: Brian Ashworth <bosrsf04@gmail.com>
---
 fs/namespace.c  |  2 ++
 init/Kconfig    | 10 ++++++++++
 kernel/sys_ni.c |  1 +
 3 files changed, 13 insertions(+)

diff --git a/fs/namespace.c b/fs/namespace.c
index 487ba30bb5c6..5e24a08bfb36 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -3058,6 +3058,7 @@ bool path_is_under(const struct path *path1, const struct path *path2)
 }
 EXPORT_SYMBOL(path_is_under);
 
+#ifdef CONFIG_PIVOT_ROOT_SYSCALL
 /*
  * pivot_root Semantics:
  * Moves the root file system of the current process to the directory put_old,
@@ -3180,6 +3181,7 @@ SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
 out0:
 	return error;
 }
+#endif  /* CONFIG_PIVOT_ROOT_SYSCALL */
 
 static void __init init_mount_tree(void)
 {
diff --git a/init/Kconfig b/init/Kconfig
index 8c39615165b7..4ea9ab25ec30 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1450,6 +1450,16 @@ config SYSCTL_SYSCALL
 
 	  If unsure say N here.
 
+config PIVOT_ROOT_SYSCALL
+	bool "Pivot_root syscall support" if EXPERT
+	default y
+	help
+	  pivot_root is a system call that allows the root to be moved and
+	  replaced by another root. This is needed for intermediate file
+	  systems such as initrd.
+
+	  If unsure say Y here.
+
 config POSIX_TIMERS
 	bool "Posix Clocks & timers" if EXPERT
 	default y
diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c
index 8acef8576ce9..7bd207571f87 100644
--- a/kernel/sys_ni.c
+++ b/kernel/sys_ni.c
@@ -178,6 +178,7 @@ cond_syscall(sys_setfsgid);
 cond_syscall(sys_capget);
 cond_syscall(sys_capset);
 cond_syscall(sys_copy_file_range);
+cond_syscall(sys_pivot_root);
 
 /* arch-specific weak syscall entries */
 cond_syscall(sys_pciconfig_read);
-- 
2.11.1

Re: [PATCH 1/1] fs: Allows for the pivot_root syscall to be optional

[Nicolas Pitre]

On Sat, 25 Feb 2017, Brian Ashworth wrote:

> The pivot_root syscall is not needed on systems that do not use
> any intermediate filesystem. Allowing for pivot_root to be
> ommitted from the kernel will aid in the tinification efforts.
> 
> Without CONFIG_PIVOT_ROOT_SYSCALL set
> add/remove: 0/2 grow/shrink: 1/0 up/down: 45/-707 (-662)
> function                                     old     new   delta
> attach_recursive_mnt                         349     394     +45
> attach_mnt                                    71       -     -71
> sys_pivot_root                               636       -    -636
> Total: Before=1899893, After=1899231, chg -0.03%

A -0.03% size difference doesn't seem much.  To bring up a more 
realistic scenario for tinification statistics, you could start from 
"make tinyconfig" instead.

> Signed-off-by: Brian Ashworth <bosrsf04@gmail.com>
> ---
>  fs/namespace.c  |  2 ++
>  init/Kconfig    | 10 ++++++++++
>  kernel/sys_ni.c |  1 +
>  3 files changed, 13 insertions(+)

I agree that this is a much more interesting diffstat than the previous 
one.

Acked-by: Nicolas Pitre <nico@linaro.org>


> diff --git a/fs/namespace.c b/fs/namespace.c
> index 487ba30bb5c6..5e24a08bfb36 100644
> --- a/fs/namespace.c
> +++ b/fs/namespace.c
> @@ -3058,6 +3058,7 @@ bool path_is_under(const struct path *path1, const struct path *path2)
>  }
>  EXPORT_SYMBOL(path_is_under);
>  
> +#ifdef CONFIG_PIVOT_ROOT_SYSCALL
>  /*
>   * pivot_root Semantics:
>   * Moves the root file system of the current process to the directory put_old,
> @@ -3180,6 +3181,7 @@ SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
>  out0:
>  	return error;
>  }
> +#endif  /* CONFIG_PIVOT_ROOT_SYSCALL */
>  
>  static void __init init_mount_tree(void)
>  {
> diff --git a/init/Kconfig b/init/Kconfig
> index 8c39615165b7..4ea9ab25ec30 100644
> --- a/init/Kconfig
> +++ b/init/Kconfig
> @@ -1450,6 +1450,16 @@ config SYSCTL_SYSCALL
>  
>  	  If unsure say N here.
>  
> +config PIVOT_ROOT_SYSCALL
> +	bool "Pivot_root syscall support" if EXPERT
> +	default y
> +	help
> +	  pivot_root is a system call that allows the root to be moved and
> +	  replaced by another root. This is needed for intermediate file
> +	  systems such as initrd.
> +
> +	  If unsure say Y here.
> +
>  config POSIX_TIMERS
>  	bool "Posix Clocks & timers" if EXPERT
>  	default y
> diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c
> index 8acef8576ce9..7bd207571f87 100644
> --- a/kernel/sys_ni.c
> +++ b/kernel/sys_ni.c
> @@ -178,6 +178,7 @@ cond_syscall(sys_setfsgid);
>  cond_syscall(sys_capget);
>  cond_syscall(sys_capset);
>  cond_syscall(sys_copy_file_range);
> +cond_syscall(sys_pivot_root);
>  
>  /* arch-specific weak syscall entries */
>  cond_syscall(sys_pciconfig_read);
> -- 
> 2.11.1
> 
> 

Re: [PATCH 1/1] fs: Allows for the pivot_root syscall to be optional

[bosrsf04]

Thanks for the ack on this. Whose tree should this patch go through?

[PATCH 3/3] fs: Allows for the pivot_root syscall to be omitted

[bosrsf04]

From: Brian Ashworth <bosrsf04@gmail.com>

The pivot_root syscall is not needed on systems that do not use
any intermediate filesystem. Allowing for pivot_root to be
ommitted from the kernel will aid in the tinification efforts.

Without CONFIG_PIVOT_ROOT_SYSCALL set
add/remove: 0/1 grow/shrink: 0/0 up/down: 0/-614 (-614)
function                                     old     new   delta
sys_pivot_root                               614       -    -614
Total: Before=1899577, After=1898963, chg -0.03%

Signed-off-by: Brian Ashworth <bosrsf04@gmail.com>
---
 fs/Makefile     |  4 +++-
 init/Kconfig    | 10 ++++++++++
 kernel/sys_ni.c |  1 +
 3 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/fs/Makefile b/fs/Makefile
index 34cb58c4127d..1a35bc90b0d5 100644
--- a/fs/Makefile
+++ b/fs/Makefile
@@ -11,7 +11,7 @@ obj-y :=	open.o read_write.o file_table.o super.o \
 		attr.o bad_inode.o file.o filesystems.o namespace.o \
 		seq_file.o xattr.o libfs.o fs-writeback.o \
 		pnode.o splice.o sync.o utimes.o \
-		stack.o fs_struct.o statfs.o fs_pin.o nsfs.o pivot_root.o
+		stack.o fs_struct.o statfs.o fs_pin.o nsfs.o
 
 ifeq ($(CONFIG_BLOCK),y)
 obj-y +=	buffer.o block_dev.o direct-io.o mpage.o
@@ -19,6 +19,8 @@ else
 obj-y +=	no-block.o
 endif
 
+obj-$(CONFIG_PIVOT_ROOT_SYSCALL) += pivot_root.o
+
 obj-$(CONFIG_PROC_FS) += proc_namespace.o
 
 obj-y				+= notify/
diff --git a/init/Kconfig b/init/Kconfig
index 4dd8bd232a1d..33ed0282765f 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1462,6 +1462,16 @@ config SYSCTL_SYSCALL
 
 	  If unsure say N here.
 
+config PIVOT_ROOT_SYSCALL
+	bool "Pivot_root syscall support" if EXPERT
+	default y
+	help
+	  pivot_root is a system call that allows the root to be moved and
+	  replaced by another root. This is needed for intermediate file
+	  systems such as initrd.
+
+	  If unsure say Y here.
+
 config POSIX_TIMERS
 	bool "Posix Clocks & timers" if EXPERT
 	default y
diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c
index 8acef8576ce9..7bd207571f87 100644
--- a/kernel/sys_ni.c
+++ b/kernel/sys_ni.c
@@ -178,6 +178,7 @@ cond_syscall(sys_setfsgid);
 cond_syscall(sys_capget);
 cond_syscall(sys_capset);
 cond_syscall(sys_copy_file_range);
+cond_syscall(sys_pivot_root);
 
 /* arch-specific weak syscall entries */
 cond_syscall(sys_pciconfig_read);
-- 
2.11.1

Re: [PATCH 0/3] fs: Support compiling out the pivot_root syscall

[Nicolas Pitre]

On Fri, 24 Feb 2017, bosrsf04@gmail.com wrote:

> From: Brian Ashworth <bosrsf04@gmail.com>
> 
> This patch series will allow for the pivot_root syscall to be made
> optional. The first patch refactors the functions that are required
> by pivot_root so they can be accessed outside of fs/namespace.c. The
> second patch moves the pivot_root syscall to its own file. The third
> adds a Kconfig option and conditional compilation to the Makefile.
> 
> The pivot_root syscall is not needed on systems that do not use
> any intermediate filesystem. Allowing for pivot_root to be
> ommitted from the kernel will aid in the tinification efforts.

This looks good.  Every such tiny bits count, and besides the size 
reduction possibility this also makes the code clearer.

Acked-by: Nicolas Pitre <nico@linaro.org>

> Brian Ashworth (3):
>   Makes functions used by pivot_root accessible
>   Extracts pivot_root so it can be made optional
>   Allows for the pivot_root syscall to be omitted
> 
>  fs/Makefile     |   2 +
>  fs/mount.h      |  24 +++++++++
>  fs/namespace.c  | 150 +++-----------------------------------------------------
>  fs/pivot_root.c | 129 ++++++++++++++++++++++++++++++++++++++++++++++++
>  init/Kconfig    |  10 ++++
>  kernel/sys_ni.c |   1 +
>  6 files changed, 172 insertions(+), 144 deletions(-)
>  create mode 100644 fs/pivot_root.c
> 
> -- 
> 2.11.1
> 
>