Re: [PATCH 2/2] ath10k: support MAC address randomization in scan

[Arend van Spriel <arend.vanspriel@broadcom.com>]

On 4/17/2018 2:28 AM, Brian Norris wrote:
> Hi,
>
> On Mon, Apr 16, 2018 at 02:32:47PM +0300, Kalle Valo wrote:
>> cjhuang@codeaurora.org writes:
>>> On 2018-04-14 05:13, Arend van Spriel wrote:
>>>> On 4/13/2018 1:28 PM, Kalle Valo wrote:
>>>>> cjhuang@codeaurora.org writes:
>>>>>
>>>>>>>> +	if (test_bit(WMI_SERVICE_SPOOF_MAC_SUPPORT, ar->wmi.svc_map)) {
>>>>>>>> +		ret = ath10k_wmi_scan_prob_req_oui(ar, ar->mac_addr);
>>>>>>>> +		if (ret) {
>>>>>>>> +			ath10k_err(ar, "failed to set prob req oui: %i\n", ret);
>>>>>>>> +			goto err_dfs_detector_exit;
>>>>>>>> +		}
>>>>>>>> +
>>>>>>>> +		ar->hw->wiphy->features |=
>>>>>>>> +			NL80211_FEATURE_SCAN_RANDOM_MAC_ADDR;
>>>>>>>
>>>>>>> Do you support NL80211_FEATURE_SCHED_SCAN_RANDOM_MAC_ADDR too?
>>>>>>
>>>>>> I'll add this flag too.
>>>>>
>>>>> Are you going to send v2 or what's the plan?
>>>>
>>>> Maybe a stupid question, but does ath10k support scheduled scan?
>
> Not a stupid question. Sorry for not asking that first.
>
>> AFAICS ath10k does not support it (sched_scan_start() op).
>
> Right, that seems to be the case.
>
>>> The reason is AVL test case needs this flag to enable random mac
>>> address scan. Maybe Brian Can explain why this flag is necessary.
>
> I never actually claimed you *needed* this flag; I just wondered how you
> claimed to pass our test when you did not support this flag, since our
> network manager currently checks for both
> NL80211_FEATURE_SCAN_RANDOM_MAC_ADDR and
> NL80211_FEATURE_SCHED_SCAN_RANDOM_MAC_ADDR before enabling MAC
> randomization in scans. But that's no matter to worry about here.
>
>> If ath10k does not support scheduled scan what's the point? Shouldn't
>> the test case then be it fixed instead of making hacks in ath10k?
>
> Indeed. We're trying to work that out right now.
>
> It looks like the status quo for looking for SCHED_SCAN support is to
> check if NL80211_CMD_START_SCHED_SCAN shows up in the command support
> list. (IOW, that's what wpa_supplicant does.) We'll probably need to
> imitate that.

I believe checking command support is not really recommended. Instead, 
you better check NL80211_ATTR_SCHED_SCAN_MAX_REQS being non-zero (since 
kernel 4.12 that is).

Regards,
Arend