Zekun Shen <bruceshenzk@xxxxxxxxx> writes:
> Function ep_rx_complete is being called without NULL checking
> in ath10k_htc_rx_completion_handler. Without such check, mal-
> formed packet is able to cause jump to NULL.
>
> ep->service_id seems a good candidate for sanity check as it is
> used in usb.c.
>
> Signed-off-by: Zekun Shen <bruceshenzk@xxxxxxxxx>
> ---
> drivers/net/wireless/ath/ath10k/htc.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/drivers/net/wireless/ath/ath10k/htc.c b/drivers/net/wireless/ath/ath10k/htc.c
> index 31df6dd04..e00794d97 100644
> --- a/drivers/net/wireless/ath/ath10k/htc.c
> +++ b/drivers/net/wireless/ath/ath10k/htc.c
> @@ -450,6 +450,11 @@ void ath10k_htc_rx_completion_handler(struct ath10k *ar, struct sk_buff *skb)
>
> ep = &htc->endpoint[eid];
>
> + if (ep->service_id == 0) {
> + ath10k_warn(ar, "HTC Rx: ep %d is not connect\n", eid);
> + goto out;
> + }
I think using ATH10K_HTC_SVC_ID_UNUSED is more descriptive than zero, as
ath10k_htc_reset_endpoint_states() uses it. I fixed in the pending
branch.
I think also ath10k_htc_process_credit_report() might have a similar
problem, can you take a look?
--
https://patchwork.kernel.org/project/linux-wireless/list/
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
