Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Sender: kvalo=codeaurora.org@mg.codeaurora.org
DMARC-Filter: OpenDMARC Filter v1.3.2 smtp.codeaurora.org 7DB0FC433D3
From:   Kalle Valo <kvalo@codeaurora.org>
To:     Phillip Potter <phil@philpotter.co.uk>
Cc:     davem@davemloft.net, kuba@kernel.org,
        linux-wireless@vger.kernel.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, ath9k-devel@qca.qualcomm.com,
        fw@strlen.de
Subject: Re: [PATCH v2] ath9k: ath9k_htc_rx_msg: return when sk_buff is too small
References: <20210502212611.1818-1-phil@philpotter.co.uk>
Date:   Sun, 30 May 2021 12:00:53 +0300
In-Reply-To: <20210502212611.1818-1-phil@philpotter.co.uk> (Phillip Potter's
        message of "Sun, 2 May 2021 22:26:11 +0100")
Message-ID: <87fsy4ppdm.fsf@codeaurora.org>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/26.1 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

Phillip Potter <phil@philpotter.co.uk> writes:

> At the start of ath9k_htc_rx_msg, we check to see if the skb pointer is
> valid, but what we don't do is check if it is large enough to contain a
> valid struct htc_frame_hdr. We should check for this and return if not,
> as the buffer is invalid in this case. Fixes a KMSAN-found uninit-value bug
> reported by syzbot at:
> https://syzkaller.appspot.com/bug?id=7dccb7d9ad4251df1c49f370607a49e1f09644ee
>
> Reported-by: syzbot+e4534e8c1c382508312c@syzkaller.appspotmail.com
> Signed-off-by: Phillip Potter <phil@philpotter.co.uk>
> ---
>
> V2:
> * Free skb properly when this problem is detected, as pointed out by
>   Florian Westphal.
>
> ---
>  drivers/net/wireless/ath/ath9k/htc_hst.c | 5 +++++
>  1 file changed, 5 insertions(+)
>
> diff --git a/drivers/net/wireless/ath/ath9k/htc_hst.c b/drivers/net/wireless/ath/ath9k/htc_hst.c
> index 510e61e97dbc..1fe89b068ac4 100644
> --- a/drivers/net/wireless/ath/ath9k/htc_hst.c
> +++ b/drivers/net/wireless/ath/ath9k/htc_hst.c
> @@ -406,6 +406,11 @@ void ath9k_htc_rx_msg(struct htc_target *htc_handle,
>  	if (!htc_handle || !skb)
>  		return;
>  
> +	if (!pskb_may_pull(skb, sizeof(struct htc_frame_hdr))) {
> +		kfree_skb(skb);
> +		return;
> +	}

This does not look complete to me, I think the function is missing
proper length checks. For example, with ENDPOINT0 it reads two byte
msg_id after the htc header and it's not verified that skb really has
that. I did not check if ep_callbacks.rx handlers have proper length
handling, I recommend verifying that also while fixing this.

Also I want to point out that the skb is freed differently based on
endpoint, I did not check why and don't know if it causes:

	if (epid < 0 || epid >= ENDPOINT_MAX) {
		if (pipe_id != USB_REG_IN_PIPE)
			dev_kfree_skb_any(skb);
		else
			kfree_skb(skb);
		return;
	}

-- 
https://patchwork.kernel.org/project/linux-wireless/list/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches