Re: [PATCH] sysctl: fix binary sysctl overflow

[ebiederm@xmission.com (Eric W. Biederman)]

Lucian Adrian Grijincu <lucian.grijincu@gmail.com> writes:

> We allocated a vector of CTL_MAXNAME elements and considered
> [0, CTL_MAXNAME] as valid indexes in the vector.
>
> This fixes valid index range to be [0, CTL_MAXNAME).
>
> I guess this should be backported to stable trees too.

No.  This part of the code is correct.  0 is a valid index,
and the array is only CTL_MAXNAME entries long.

I think the maximum depth of this table is only about 6, but
feel free to count and tell me CTL_MAXNAME is set too low.

Eric


>
> Signed-off-by: Lucian Adrian Grijincu <lucian.grijincu@gmail.com>
> ---
>  kernel/sysctl_binary.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/kernel/sysctl_binary.c b/kernel/sysctl_binary.c
> index b875bed..e2f377c 100644
> --- a/kernel/sysctl_binary.c
> +++ b/kernel/sysctl_binary.c
> @@ -1446,7 +1446,7 @@ static ssize_t do_sysctl(int __user *args_name, int nlen,
>  	int i;
>  
>  	/* Check args->nlen. */
> -	if (nlen < 0 || nlen > CTL_MAXNAME)
> +	if (nlen < 0 || nlen >= CTL_MAXNAME)
>  		return -ENOTDIR;
>  	/* Read in the sysctl name for simplicity */
>  	for (i = 0; i < nlen; i++)