Re: [PATCH] net: sched: sch: null pointer dereference in htb_offload_move_qdisc()

[Larysa Zaremba <larysa.zaremba@intel.com>]

On Thu, Feb 16, 2023 at 02:49:40AM -0800, Alok Tiwari wrote:
> A possible case of null pointer dereference detected by static analyzer
> htb_destroy_class_offload() is calling htb_find() which can return NULL value
> for invalid class id, moved_cl=htb_find(classid, sch);
> in that case it should not pass 'moved_cl' to htb_offload_move_qdisc()
> if 'moved_cl' is NULL pointer return -EINVAL.
> 
> Signed-off-by: Alok Tiwari <alok.a.tiwari@oracle.com>

Conceptually, code looks fine, but it probably won't be accepted without:
- imperative verb in the title
- tree, to which patch must be applied [0]
- more nicely written and formatted commit message. If you are referring to a 
  static analyzer and not to common logic, maybe include, how the warning looks.
  Separate your text into sentences and paragraphs.

[0] https://www.kernel.org/doc/html/v5.3/networking/netdev-FAQ.html#q-how-do-i-indicate-which-tree-net-vs-net-next-my-patch-should-be-in

> ---
>  net/sched/sch_htb.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c
> index 92f2975b6a82..5a96d9ea3221 100644
> --- a/net/sched/sch_htb.c
> +++ b/net/sched/sch_htb.c
> @@ -1601,6 +1601,8 @@ static int htb_destroy_class_offload(struct Qdisc *sch, struct htb_class *cl,
>  		u32 classid = TC_H_MAJ(sch->handle) |
>  			      TC_H_MIN(offload_opt.classid);
>  		struct htb_class *moved_cl = htb_find(classid, sch);
> +		if (WARN_ON_ONCE(!moved_cl))
> +			return -EINVAL;
>  
>  		htb_offload_move_qdisc(sch, moved_cl, cl, destroying);
>  	}
> -- 
> 2.39.1
>