linux-kernel - Re: [f2fs-dev] [PATCH v3] f2fs: move ioctl interface definitions to separated file

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

Message-ID: <20201103032234.GB2875@sol.localdomain>
Date:   Mon, 2 Nov 2020 19:22:34 -0800
From:   Eric Biggers <ebiggers@...nel.org>
To:     Chao Yu <yuchao0@...wei.com>
Cc:     jaegeuk@...nel.org, linux-kernel@...r.kernel.org,
        linux-f2fs-devel@...ts.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH v3] f2fs: move ioctl interface definitions to
 separated file

On Mon, Nov 02, 2020 at 02:21:31PM +0800, Chao Yu wrote:
> +#define F2FS_IOC_MOVE_RANGE		_IOWR(F2FS_IOCTL_MAGIC, 9,	\
> +						struct f2fs_move_range)
[...]
> +#define F2FS_IOC_GARBAGE_COLLECT_RANGE	_IOW(F2FS_IOCTL_MAGIC, 11,	\
> +						struct f2fs_gc_range)
[...]
> +
> +struct f2fs_gc_range {
> +	__u32 sync;
> +	__u64 start;
> +	__u64 len;
> +};
[...]
> +struct f2fs_move_range {
> +	__u32 dst_fd;		/* destination fd */
> +	__u64 pos_in;		/* start position in src_fd */
> +	__u64 pos_out;		/* start position in dst_fd */
> +	__u64 len;		/* size to move */
> +};

These two structs are weird because there is implicit padding between the __u32
field and the following __u64 field on some 32-bit architectures (e.g. x86_32)
but not others (e.g. arm32).

But f2fs_compat_ioctl() doesn't handle these two ioctls specially, but rather
just calls through to f2fs_ioctl().  That's wrong, and it means that
F2FS_IOC_MOVE_RANGE and F2FS_IOC_GARBAGE_COLLECT_RANGE won't work when called
from an x86_32 binary on an x86_64 kernel.

So something needs to be fixed.  I wonder if it's safe to just explicitly add
the padding field after the fact.  If no one is actually using these two ioctls
in a case where both userspace and the kernel lack the implicit padding (e.g.,
x86_32 userspace with x86_32 kernel), it should be fine...

- Eric