Re: [PATCH v4 2/3] fs: don't let getdents return bogus names

[Dave Chinner <david@fromorbit.com>]

On Fri, Jan 18, 2019 at 05:14:39PM +0100, Jann Horn wrote:
> When you e.g. run `find` on a directory for which getdents returns
> "filenames" that contain slashes, `find` passes those "filenames" back to
> the kernel, which then interprets them as paths. That could conceivably
> cause userspace to do something bad when accessing something like an
> untrusted USB stick, but I'm not aware of any specific example.
> 
> Instead of returning bogus filenames to userspace, return -EUCLEAN.
> 
> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> Cc: stable@vger.kernel.org
> Signed-off-by: Jann Horn <jannh@google.com>
> ---
> I ordered this fix before the refactoring one so that it can easily be
> backported.
> 
> changed in v2:
>  - move bogus_dirent_name() out of the #ifdef (kbuild test robot)
> changed in v3:
>  - change calling convention (Al Viro)
>  - comment fix
> changed in v4:
>  - use EFSCORRUPTED instead of EUCLEAN (Dave Chinner)
> 
>  arch/alpha/kernel/osf_sys.c |  4 ++++
>  fs/readdir.c                | 35 +++++++++++++++++++++++++++++++++++
>  include/linux/fs.h          |  2 ++
>  3 files changed, 41 insertions(+)
> 
> diff --git a/arch/alpha/kernel/osf_sys.c b/arch/alpha/kernel/osf_sys.c
> index 792586038808..db1c2144d477 100644
> --- a/arch/alpha/kernel/osf_sys.c
> +++ b/arch/alpha/kernel/osf_sys.c
> @@ -40,6 +40,7 @@
>  #include <linux/vfs.h>
>  #include <linux/rcupdate.h>
>  #include <linux/slab.h>
> +#include <linux/fs.h>
>  
>  #include <asm/fpu.h>
>  #include <asm/io.h>
> @@ -117,6 +118,9 @@ osf_filldir(struct dir_context *ctx, const char *name, int namlen,
>  	unsigned int reclen = ALIGN(NAME_OFFSET + namlen + 1, sizeof(u32));
>  	unsigned int d_ino;
>  
> +	buf->error = check_dirent_name(name, namlen);
> +	if (unlikely(buf->error))
> +		return -EFSCORRUPTED;
>  	buf->error = -EINVAL;	/* only used if we fail */
>  	if (reclen > buf->count)
>  		return -EINVAL;
> diff --git a/fs/readdir.c b/fs/readdir.c
> index 2f6a4534e0df..58088510bb9c 100644
> --- a/fs/readdir.c
> +++ b/fs/readdir.c
> @@ -64,6 +64,26 @@ int iterate_dir(struct file *file, struct dir_context *ctx)
>  }
>  EXPORT_SYMBOL(iterate_dir);
>  
> +/*
> + * Most filesystems don't filter out bogus directory entry names, and userspace
> + * can get very confused by such names. Behave as if a filesystem error had
> + * happened while reading directory entries.
> + */
> +int check_dirent_name(const char *name, int namlen)
> +{
> +	if (namlen == 0) {
> +		pr_err_once("%s: filesystem returned bogus empty name\n",
> +			    __func__);
> +		return -EFSCORRUPTED;
> +	}
> +	if (memchr(name, '/', namlen)) {
> +		pr_err_once("%s: filesystem returned bogus name '%*pEhp' (contains slash)\n",
> +			    __func__, namlen, name);
> +		return -EFSCORRUPTED;
> +	}
> +	return 0;
> +}
> +
>  /*
>   * Traditional linux readdir() handling..
>   *
> @@ -98,6 +118,9 @@ static int fillonedir(struct dir_context *ctx, const char *name, int namlen,
>  
>  	if (buf->result)
>  		return -EINVAL;
> +	buf->result = check_dirent_name(name, namlen);
> +	if (unlikely(buf->result))
> +		return -EFSCORRUPTED;

Why bother returning an error from check_dirent_name() if you just
throw it away? i.e:

	if (!dirent_name_valid(name, namelen))
		return -EFSCORRUPTED;

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com