Re: [ath6kl:pending-ath11k 198/205] drivers/net/wireless/ath/ath11k/mac.c:1274 ath11k_peer_assoc_h_he() error: memcpy() 'he_cap->he_cap_elem.mac_cap_info' too small (6 vs 8)

[Kalle Valo <kvalo@codeaurora.org>]

John Crispin <john@phrozen.org> writes:

> On 18/06/2019 13:06, Kalle Valo wrote:
>> (moving from ath10k to ath11k list)
>>
>> John Crispin <john@phrozen.org> writes:
>>
>>> On 18/06/2019 08:53, kbuild test robot wrote:
>>>> tree:
>>>> https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git
>>>> pending-ath11k
>>>> head:   0f82fec5679664bb91d6c167fd1a146f113e4197
>>>> commit: cbdb3159fdf450b7b3999a06600aa0e1fb78383f [198/205] ath11k:
>>>> set additional values inside wmi_peer_assoc_complete_cmd
>>>>
>>>> If you fix the issue, kindly add following tag
>>>> Reported-by: kbuild test robot <lkp@intel.com>
>>>> Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
>>>>
>>>> New smatch warnings:
>>>> drivers/net/wireless/ath/ath11k/mac.c:1274 ath11k_peer_assoc_h_he()
>>>> error: memcpy() 'he_cap->he_cap_elem.mac_cap_info' too small (6 vs
>>>> 8)
>>>>
>>>> Old smatch warnings:
>>>> drivers/net/wireless/ath/ath11k/mac.c:1276 ath11k_peer_assoc_h_he()
>>>> error: memcpy() 'he_cap->he_cap_elem.phy_cap_info' too small (11 vs
>>>> 12)
>>>>
>>>> #
>>>> https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git/commit/?id=cbdb3159fdf450b7b3999a06600aa0e1fb78383f
>>>> git remote add ath6kl
>>>> https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git
>>>> git remote update ath6kl
>>>> git checkout cbdb3159fdf450b7b3999a06600aa0e1fb78383f
>>>> vim +1274 drivers/net/wireless/ath/ath11k/mac.c
>>>>
>>>> 258bbf52 Kalle Valo   2019-02-05  1260
>>>> 258bbf52 Kalle Valo 2019-02-05 1261 static void
>>>> ath11k_peer_assoc_h_he(struct ath11k *ar,
>>>> 258bbf52 Kalle Valo 2019-02-05 1262 struct ieee80211_vif *vif,
>>>> 258bbf52 Kalle Valo 2019-02-05 1263 struct ieee80211_sta *sta,
>>>> 258bbf52 Kalle Valo 2019-02-05 1264 struct peer_assoc_params *arg)
>>>> 258bbf52 Kalle Valo   2019-02-05  1265  {
>>>> 17aca2d9 John Crispin 2019-06-03 1266 const struct
>>>> ieee80211_sta_he_cap *he_cap = &sta->he_cap;
>>>> 3db59a23 Kalle Valo   2019-06-12  1267  	u16 v;
>>>> 17aca2d9 John Crispin 2019-06-03  1268
>>>> 17aca2d9 John Crispin 2019-06-03  1269  	if (!he_cap->has_he)
>>>> 17aca2d9 John Crispin 2019-06-03  1270  		return;
>>>> 17aca2d9 John Crispin 2019-06-03  1271
>>>> 17aca2d9 John Crispin 2019-06-03  1272  	arg->he_flag = true;
>>>> 17aca2d9 John Crispin 2019-06-03  1273
>>>> 17aca2d9 John Crispin 2019-06-03 @1274
>>>> memcpy(&arg->peer_he_cap_macinfo, he_cap->he_cap_elem.mac_cap_info,
>>>> 17aca2d9 John Crispin 2019-06-03 1275
>>>> sizeof(arg->peer_he_cap_macinfo));
>>>>
>>>> Smatch thinks these are different sizes...  I don't have a copy of
>>>> struct peer_assoc_params so I can't check.
>>> Hi,
>>>
>>> its he_cap->he_cap_elem.mac_cap_info[6] and
>>> arg->peer_he_cap_macinfo[2] and we only copy the first 2 elements as
>>> the FW only cares for the first 2 bytes.
>> Are the remaining bytes to zero? (ie. does it follow "the reserved
>> fields should be zero" rule). In general this makes it safer to have
>> changes in firmware interface in the future.
>
>
> We copy the first 2 bytes of a 6 byte buffer into a 2 byte buffer.
> these 2 bytes are later copied into 2 non-consecutive fields inside a
> WMI cmd

Ah, good.

-- 
Kalle Valo

_______________________________________________
ath11k mailing list
ath11k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath11k