Re: [f2fs-dev] [PATCH] f2fs: file: add checks to f2fs_ioc_flush_device()

[Chao Yu <chao@kernel.org>]

On 2024/8/1 15:17, Roman Smirnov wrote:
> If invalid data is copied from user space and if GET_SEGNO()
> returns NULL_SEGNO an overflow is possible.

Can you explain in which condition that FDEV(range.dev_num).start_blk or
FDEV(range.dev_num).end_blk will be invalid?

 > > Add checks for invalid values.
> 
> Found by Linux Verification Center (linuxtesting.org) with Svace.
> 
> Signed-off-by: Roman Smirnov <r.smirnov@omp.ru>
> ---
>   fs/f2fs/file.c | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
> index 168f08507004..1b9c4fee9db1 100644
> --- a/fs/f2fs/file.c
> +++ b/fs/f2fs/file.c
> @@ -3093,6 +3093,9 @@ static int f2fs_ioc_flush_device(struct file *filp, unsigned long arg)
>   		start_segno = dev_start_segno;
>   	end_segno = min(start_segno + range.segments, dev_end_segno);
>   
> +	if (start_segno > F2FS_MAX_SEGMENT - range.segments || end_segno == NULL_SEGNO)
> +		return -EINVAL;

It missed to call mnt_drop_write_file() before return.

Thanks,

> +
>   	while (start_segno < end_segno) {
>   		if (!f2fs_down_write_trylock(&sbi->gc_lock)) {
>   			ret = -EBUSY;



_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel