linux-kernel - Re: [PATCH] Add renameat2 function [BZ #17662]

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <83bc54c8-ea97-22f7-360f-523e5979864a@redhat.com>
Date:   Mon, 2 Jul 2018 11:32:50 +0200
From:   Florian Weimer <fweimer@...hat.com>
To:     Yury Norov <ynorov@...iumnetworks.com>
Cc:     libc-alpha@...rceware.org,
        Alexander Viro <viro@...iv.linux.org.uk>,
        linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org,
        linux-api@...r.kernel.org
Subject: Re: [PATCH] Add renameat2 function [BZ #17662]

On 07/02/2018 10:46 AM, Yury Norov wrote:

> Is my understanding correct that glibc community finds <linux/fs.h>
> inappropriate for their use, and prefer to re-introduce (duplicate)
> its functionality locally? I think it's wrong. The right way to go
> is to make kernel headers comfortable for users instead of ignoring
> it.

In some cases, we already use UAPI headers (<linux/falloc.h> is an 
example), but it is not always possible.

> Are you OK to switch to kernel RENAME_* definitions if they will be
> located in separated small file? Like in the patch below.
> 
> Signed-off-by: Yury Norov <ynorov@...iumnetworks.com>
> ---
>   include/uapi/linux/fs.h     |  4 +---
>   include/uapi/linux/rename.h | 12 ++++++++++++
>   2 files changed, 13 insertions(+), 3 deletions(-)
>   create mode 100644 include/uapi/linux/rename.h
> 
> diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
> index c27576d471c2..46c03ea31a76 100644
> --- a/include/uapi/linux/fs.h
> +++ b/include/uapi/linux/fs.h
> @@ -44,9 +44,7 @@
>   #define SEEK_HOLE	4	/* seek to the next hole */
>   #define SEEK_MAX	SEEK_HOLE
>   
> -#define RENAME_NOREPLACE	(1 << 0)	/* Don't overwrite target */
> -#define RENAME_EXCHANGE		(1 << 1)	/* Exchange source and dest */
> -#define RENAME_WHITEOUT		(1 << 2)	/* Whiteout source */
> +#include <linux/rename.h>
>   
>   struct file_clone_range {
>   	__s64 src_fd;
> diff --git a/include/uapi/linux/rename.h b/include/uapi/linux/rename.h
> new file mode 100644
> index 000000000000..7178f0565657
> --- /dev/null
> +++ b/include/uapi/linux/rename.h
> @@ -0,0 +1,12 @@
> +/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
> +#ifndef _UAPI_LINUX_RENAME_H
> +#define _UAPI_LINUX_RENAME_H
> +
> +/*
> + * Definitions for rename syscall family.
> + */
> +#define RENAME_NOREPLACE	(1 << 0)	/* Don't overwrite target */
> +#define RENAME_EXCHANGE		(1 << 1)	/* Exchange source and dest */
> +#define RENAME_WHITEOUT		(1 << 2)	/* Whiteout source */
> +
> +#endif /* _UAPI_LINUX_RENAME_H */

This would help.

We would need to provide definitions for compatibility with older kernel 
headers locally, but on newer kernels, we could use the UAPI header file.

Thanks,
Florian