linux-kernel - Re: [PATCH] rtlwifi: rtl8192cu: remove pointless memcpy

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <62048535-6455-e4ac-040a-ca2bf60fc0f7@lwfinger.net>
Date:   Fri, 9 Feb 2018 10:44:19 -0600
From:   Larry Finger <Larry.Finger@...inger.net>
To:     Arnd Bergmann <arnd@...db.de>, Ping-Ke Shih <pkshih@...ltek.com>
Cc:     Kalle Valo <kvalo@...eaurora.org>,
        Johannes Berg <johannes.berg@...el.com>,
        linux-wireless@...r.kernel.org, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] rtlwifi: rtl8192cu: remove pointless memcpy

On 02/09/2018 07:24 AM, Arnd Bergmann wrote:
> gcc-8 points out that source and destination of the memcpy() are
> always the same pointer, so the effect of memcpy() is undefined
> here (its arguments must not overlap):
> 
> drivers/net/wireless/realtek/rtlwifi/rtl8192cu/trx.c: In function '_rtl_rx_process':
> drivers/net/wireless/realtek/rtlwifi/rtl8192cu/trx.c:430:2: error: 'memcpy' source argument is the same as destination [-Werror=restrict]
> 
> Most likely this is harmless, but it's easy to just remove the
> line and get rid of the warning.
> 
> Signed-off-by: Arnd Bergmann <arnd@...db.de>
> ---
>   drivers/net/wireless/realtek/rtlwifi/rtl8192cu/trx.c | 1 -
>   1 file changed, 1 deletion(-)
> 
> diff --git a/drivers/net/wireless/realtek/rtlwifi/rtl8192cu/trx.c b/drivers/net/wireless/realtek/rtlwifi/rtl8192cu/trx.c
> index ac4a82de40c7..9ab56827124e 100644
> --- a/drivers/net/wireless/realtek/rtlwifi/rtl8192cu/trx.c
> +++ b/drivers/net/wireless/realtek/rtlwifi/rtl8192cu/trx.c
> @@ -427,7 +427,6 @@ static void _rtl_rx_process(struct ieee80211_hw *hw, struct sk_buff *skb)
>   		 (u32)hdr->addr1[0], (u32)hdr->addr1[1],
>   		 (u32)hdr->addr1[2], (u32)hdr->addr1[3],
>   		 (u32)hdr->addr1[4], (u32)hdr->addr1[5]);
> -	memcpy(IEEE80211_SKB_RXCB(skb), rx_status, sizeof(*rx_status));
>   	ieee80211_rx(hw, skb);
>   }

No, the warning is pointing to the wrong place. The routine in question does the 
following:

1. Loads the rx_status struct from skb->cb.
2. Overwrites the contents with 0.
3. Fills various members of the struct.
4. Writes the revised struct back into skb->cb.

Thus eliminating step 4 negates all the things done in step 3, and is wrong. The 
correct fix is to change step 1 to create a NULL-filled rx_status struct, and 
eliminate step 2.

NACK.

Larry