On Wed, Apr 01, 2020 at 01:03:41PM +0200, Jerome Pouiller wrote:
> From: Jérôme Pouiller <jerome.pouiller@xxxxxxxxxx>
>
> The last argument of hif_handle_tx_data() was now unused. In add,
> hif_handle_tx_data() has nothing to do with HIF layer and should be
> renamed. Finally, it not convenient to pass a wfx_vif as parameter. It
> is easier to let hif_handle_tx_data() find the interface itself.
>
> Signed-off-by: Jérôme Pouiller <jerome.pouiller@xxxxxxxxxx>
> ---
> drivers/staging/wfx/queue.c | 19 ++++++++++---------
> 1 file changed, 10 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/staging/wfx/queue.c b/drivers/staging/wfx/queue.c
> index 2553f77522d9..8647731e02c0 100644
> --- a/drivers/staging/wfx/queue.c
> +++ b/drivers/staging/wfx/queue.c
> @@ -319,13 +319,17 @@ bool wfx_tx_queues_is_empty(struct wfx_dev *wdev)
> return ret;
> }
>
> -static bool hif_handle_tx_data(struct wfx_vif *wvif, struct sk_buff *skb,
> - struct wfx_queue *queue)
> +static bool wfx_handle_tx_data(struct wfx_dev *wdev, struct sk_buff *skb)
> {
> struct hif_req_tx *req = wfx_skb_txreq(skb);
> struct ieee80211_key_conf *hw_key = wfx_skb_tx_priv(skb)->hw_key;
> struct ieee80211_hdr *frame =
> (struct ieee80211_hdr *)(req->frame + req->data_flags.fc_offset);
> + struct wfx_vif *wvif =
> + wdev_to_wvif(wdev, ((struct hif_msg *)skb->data)->interface);
^^^^^^^^^
This is on the TX side so it's probably okay, but one problem I have
noticed is that we do this on the RX side as well with checking that
if (skb->len < sizeof(struct hif_msg))
return -EINVAL;
So we could be reading beyond the end of the skb. If we got really
unlucky it could lead to an Oops.
regards,
dan carpenter
