Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1737537 > unrolled thread

[PATCH 0/5] exec: binfmt_misc: fix use-after-free, kill iname[BINPRM_BUF_SIZE]

Back to article view | Back to linux.kernel

Contents

  [PATCH 0/5] exec: binfmt_misc: fix use-after-free, kill  iname[BINPRM_BUF_SIZE] Oleg Nesterov <oleg@redhat.com> - 2017-09-22 16:40 +0200
    [PATCH 1/5] exec: binfmt_misc: don't nullify Node->dentry in  kill_node() Oleg Nesterov <oleg@redhat.com> - 2017-09-22 16:40 +0200
    [PATCH 4/5] exec: binfmt_misc: fix race between load_misc_binary()  and kill_node() Oleg Nesterov <oleg@redhat.com> - 2017-09-22 16:40 +0200
    Re: [PATCH 0/5] exec: binfmt_misc: fix use-after-free, kill iname[BINPRM_BUF_SIZE] Kees Cook <keescook@chromium.org> - 2017-09-22 17:30 +0200

#1737537 — [PATCH 0/5] exec: binfmt_misc: fix use-after-free, kill iname[BINPRM_BUF_SIZE]

Note: 5/5 depends on

	-extern int bprm_change_interp(char *interp, struct linux_binprm *bprm);
	+extern int bprm_change_interp(const char *interp, struct linux_binprm *bprm);

change in

	[PATCH] exec: load_script: kill the onstack interp[BINPRM_BUF_SIZE] array
	https://marc.info/?l=linux-kernel&m=150575251328591

I sent before.

Looks like this code was always wrong, then 948b701a607f ("binfmt_misc: add persistent
opened binary handler for containers") added more problems.

Oleg.

 fs/binfmt_misc.c | 56 ++++++++++++++++++++++++++------------------------------
 1 file changed, 26 insertions(+), 30 deletions(-)

[toc] | [next] | [standalone]

#1737538 — [PATCH 1/5] exec: binfmt_misc: don't nullify Node->dentry in kill_node()

kill_node() nullifies/checks Node->dentry to avoid double free. This
complicates the next changes and this is very confusing:

- we do not need to check dentry != NULL under entries_lock, kill_node()
  is always called under inode_lock(d_inode(root)) and we rely on this
  inode_lock() anyway, without this lock the MISC_FMT_OPEN_FILE cleanup
  could race with itself.

- if kill_inode() was already called and ->dentry == NULL we should not
  even try to close e->interp_file.

We can change bm_entry_write() to simply check !list_empty(list) before
kill_node. Again, we rely on inode_lock(), in particular it saves us from
the race with bm_status_write(), another caller of kill_node().

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 fs/binfmt_misc.c | 22 +++++++++-------------
 1 file changed, 9 insertions(+), 13 deletions(-)

diff --git a/fs/binfmt_misc.c b/fs/binfmt_misc.c
index f471809..f4de5ae 100644
--- a/fs/binfmt_misc.c
+++ b/fs/binfmt_misc.c
@@ -600,11 +600,7 @@ static void kill_node(Node *e)
 	struct dentry *dentry;
 
 	write_lock(&entries_lock);
-	dentry = e->dentry;
-	if (dentry) {
-		list_del_init(&e->list);
-		e->dentry = NULL;
-	}
+	list_del_init(&e->list);
 	write_unlock(&entries_lock);
 
 	if ((e->flags & MISC_FMT_OPEN_FILE) && e->interp_file) {
@@ -612,12 +608,11 @@ static void kill_node(Node *e)
 		e->interp_file = NULL;
 	}
 
-	if (dentry) {
-		drop_nlink(d_inode(dentry));
-		d_drop(dentry);
-		dput(dentry);
-		simple_release_fs(&bm_mnt, &entry_count);
-	}
+	dentry = e->dentry;
+	drop_nlink(d_inode(dentry));
+	d_drop(dentry);
+	dput(dentry);
+	simple_release_fs(&bm_mnt, &entry_count);
 }
 
 /* /<entry> */
@@ -662,7 +657,8 @@ static ssize_t bm_entry_write(struct file *file, const char __user *buffer,
 		root = file_inode(file)->i_sb->s_root;
 		inode_lock(d_inode(root));
 
-		kill_node(e);
+		if (!list_empty(&e->list))
+			kill_node(e);
 
 		inode_unlock(d_inode(root));
 		break;
@@ -791,7 +787,7 @@ static ssize_t bm_status_write(struct file *file, const char __user *buffer,
 		inode_lock(d_inode(root));
 
 		while (!list_empty(&entries))
-			kill_node(list_entry(entries.next, Node, list));
+			kill_node(list_first_entry(&entries, Node, list));
 
 		inode_unlock(d_inode(root));
 		break;
-- 
2.5.0

[toc] | [prev] | [next] | [standalone]

#1737539 — [PATCH 4/5] exec: binfmt_misc: fix race between load_misc_binary() and kill_node()

load_misc_binary() makes a local copy of fmt->interpreter under entries_lock
to avoid the race with kill_node() but this is not enough; the whole Node
can be freed after we drop entries_lock, not only the ->interpreter string.

Add dget/dput(fmt->dentry) to ensure bm_evict_inode() can't destroy/free
this Node.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 fs/binfmt_misc.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/fs/binfmt_misc.c b/fs/binfmt_misc.c
index 45809c6..e59d006 100644
--- a/fs/binfmt_misc.c
+++ b/fs/binfmt_misc.c
@@ -138,20 +138,23 @@ static int load_misc_binary(struct linux_binprm *bprm)
 
 	retval = -ENOEXEC;
 	if (!enabled)
-		goto ret;
+		return retval;
 
 	/* to keep locking time low, we copy the interpreter string */
 	read_lock(&entries_lock);
 	fmt = check_file(bprm);
-	if (fmt)
+	if (fmt) {
+		dget(fmt->dentry);
 		strlcpy(iname, fmt->interpreter, BINPRM_BUF_SIZE);
+	}
 	read_unlock(&entries_lock);
 	if (!fmt)
-		goto ret;
+		return retval;
 
 	/* Need to be able to load the file after exec */
+	retval = -ENOENT;
 	if (bprm->interp_flags & BINPRM_FLAGS_PATH_INACCESSIBLE)
-		return -ENOENT;
+		goto ret;
 
 	if (!(fmt->flags & MISC_FMT_PRESERVE_ARGV0)) {
 		retval = remove_arg_zero(bprm);
@@ -235,6 +238,7 @@ static int load_misc_binary(struct linux_binprm *bprm)
 		goto error;
 
 ret:
+	dput(fmt->dentry);
 	return retval;
 error:
 	if (fd_binary > 0)
-- 
2.5.0

[toc] | [prev] | [next] | [standalone]

#1737583 — Re: [PATCH 0/5] exec: binfmt_misc: fix use-after-free, kill iname[BINPRM_BUF_SIZE]

On Fri, Sep 22, 2017 at 7:36 AM, Oleg Nesterov <oleg@redhat.com> wrote:
> Note: 5/5 depends on
>
>         -extern int bprm_change_interp(char *interp, struct linux_binprm *bprm);
>         +extern int bprm_change_interp(const char *interp, struct linux_binprm *bprm);
>
> change in
>
>         [PATCH] exec: load_script: kill the onstack interp[BINPRM_BUF_SIZE] array
>         https://marc.info/?l=linux-kernel&m=150575251328591
>
> I sent before.
>
> Looks like this code was always wrong, then 948b701a607f ("binfmt_misc: add persistent
> opened binary handler for containers") added more problems.
>
> Oleg.
>
>  fs/binfmt_misc.c | 56 ++++++++++++++++++++++++++------------------------------
>  1 file changed, 26 insertions(+), 30 deletions(-)

This all looks correct to me, thanks!

Acked-by: Kees Cook <keescook@chromium.org>

-Kees

-- 
Kees Cook
Pixel Security

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web