Re: [PATCH] kernel/sys.c: fix possible spectre-v1 in do_prlimit()

[Cyrill Gorcunov <gorcunov@gmail.com>]

On Mon, May 27, 2019 at 03:23:08PM +0800, Dianzhang Chen wrote:
> The `resource` in do_prlimit() is controlled by userspace via syscall: setrlimit(defined in kernel/sys.c), hence leading to a potential exploitation of the Spectre variant 1 vulnerability.
> The relevant code in do_prlimit() is as below：
> 
> if (resource >= RLIM_NLIMITS)
>         return -EINVAL;
> ...
> rlim = tsk->signal->rlim + resource;    // use resource as index
> ...
>             *old_rlim = *rlim;
> 
> Fix this by sanitizing resource before using it to index tsk->signal->rlim.
> 
> Signed-off-by: Dianzhang Chen <dianzhangchen0@gmail.com>
> ---
>  kernel/sys.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/kernel/sys.c b/kernel/sys.c
> index bdbfe8d..7eba1ca 100644
> --- a/kernel/sys.c
> +++ b/kernel/sys.c
> @@ -1532,6 +1532,8 @@ int do_prlimit(struct task_struct *tsk, unsigned int resource,
>  
>  	if (resource >= RLIM_NLIMITS)
>  		return -EINVAL;
> +
> +	resource = array_index_nospec(resource, RLIM_NLIMITS);
>  	if (new_rlim) {
>  		if (new_rlim->rlim_cur > new_rlim->rlim_max)
>  			return -EINVAL;

Could you please explain in details how array_index_nospec is different
from resource >= RLIM_NLIMITS? Since I don't get how it is related to
spectre issue.