Re: 9984 VHT

[Ben Greear <greearb@candelatech.com>]

On 02/02/2017 12:28 PM, Sebastian Gottschall wrote:
> Am 02.02.2017 um 20:32 schrieb Ben Greear:
>> On 02/02/2017 11:29 AM, Sebastian Gottschall wrote:
>>> Am 02.02.2017 um 20:18 schrieb Ben Greear:
>>>> On 02/02/2017 11:08 AM, Sebastian Gottschall wrote:
>>>>> Am 02.02.2017 um 20:05 schrieb Ben Greear:
>>>>>> On 02/02/2017 10:42 AM, Sebastian Gottschall wrote:
>>>>>>> Am 02.02.2017 um 19:24 schrieb Ben Greear:
>>>>>>>> On 02/02/2017 08:18 AM, Ben Greear wrote:
>>>>>>>>> On 02/01/2017 10:45 AM, Sebastian Gottschall wrote:
>>>>>>>>>> Am 01.02.2017 um 17:48 schrieb Ben Greear:
>>>>>>>>>>> On 01/30/2017 02:28 AM, Sebastian Gottschall wrote:
>>>>>>>>>>>> Hello
>>>>>>>>>>>>
>>>>>>>>>>>> with recent 9984 firmares vht160 seem to crash the firmware itself for no reason i see.
>>>>>>>>>>>> is there any internal structure change in these newer firmwares we need to consider?
>>>>>>>>>>>> i also tested the even more recent firmware firmware-5.bin_10.4-3.4-00068 from codeaurora. this one doesnt crash but the vdev_start returns with a
>>>>>>>>>>>> error
>>>>>>>>>>>>
>>>>>>>>>>>> i compared the more recent wmi headers from the qca drivers, but wasnt able to find anything which could explain the behaviour
>>>>>>>>>>>>
>>>>>>>>>>>> Sebastian
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Hello Sebastian,
>>>>>>>>>>>
>>>>>>>>>>> Could you share the hostapd/supplicant config file you are using?
>>>>>>>>>>>
>>>>>>>>>>> And, what kernel (or backports kernel) are you using? I'd like to
>>>>>>>>>>> see how 160Mhz works on my firmware....
>>>>>>>>>> always state of art for sure
>>>>>>>>>
>>>>>>>>> Using a somewhat hacked 4.9.2+ kernel and my CT firmware, I get failure to start CAC:
>>>>>>>>>
>>>>>>>>> 1486052184.818710: Mode: IEEE 802.11a  Channel: 100 Frequency: 5500 MHz
>>>>>>>>> 1486052184.818714: DFS 8 channels required radar detection
>>>>>>>>> 1486052184.818717: DFS all channels available, (SKIP CAC): no
>>>>>>>>> 1486052184.818719: DFS 0 chans unavailable - choose other channel: no
>>>>>>>>> 1486052184.818722: vap0: interface state COUNTRY_UPDATE->DFS
>>>>>>>>> 1486052184.818725: DFS start CAC on 5500 MHz
>>>>>>>>> 1486052184.818733: vap0: DFS-CAC-START freq=5500 chan=100 sec_chan=1, width=2, seg0=114, seg1=0, cac_time=60s
>>>>>>>>> 1486052184.818737: nl80211: Start radar detection (CAC) 5500 MHz (ht_enabled=1, vht_enabled=1, bandwidth=160 MHz, cf1=5570 MHz, cf2=0 MHz)
>>>>>>>>> 1486052184.818742:   * freq=5500
>>>>>>>>> 1486052184.818745:   * vht_enabled=1
>>>>>>>>> 1486052184.818747:   * ht_enabled=1
>>>>>>>>> 1486052184.818749:   * bandwidth=160
>>>>>>>>> 1486052184.818751:   * channel_width=5
>>>>>>>>> 1486052184.818754:   * center_freq1=5570
>>>>>>>>> 1486052184.818756:   * center_freq2=0
>>>>>>>>> 1486052184.823437: nl80211: Failed to start radar detection: -16 (Device or resource busy)
>>>>>>>>> 1486052184.823444: DFS start_dfs_cac() failed, -1
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> I'll go dig around to see if I can figure out why...
>>>>>>>>
>>>>>>>> I hacked ath10k to enable radar detection on 160Mhz bandwidths. Now hostapd
>>>>>>>> starts.
>>>>>>>>
>>>>>>>> I can reproduce the FW failure.  It is because FW 3.3-25 release started asserting
>>>>>>>> if the freq2 was zero in VHT160 mode.  It seems to use both freq1 and freq2, and not
>>>>>>>> how the driver or linux seems to normally use them.
>>>>>>> its even worse. starting from 3.3 it uses freq2 instead of freq1. but i made a patch for it. but it still wont work. vdev_start still fails
>>>>>>
>>>>>> Well it would have been helpful from the start to have known about this patch.
>>>>> i wasnt sure about it at that time. since it did not work
>>>>>>
>>>>>> Could you post it?
>>>>> --- wmi.c       (revision 3267)
>>>>> +++ wmi.c       (working copy)
>>>>> @@ -1637,11 +1637,12 @@
>>>>>                 flags |= WMI_CHAN_FLAG_DFS;
>>>>>
>>>>>         ch->mhz = __cpu_to_le32(arg->freq);
>>>>> +       ch->band_center_freq2 = 0;
>>>>>         ch->band_center_freq1 = __cpu_to_le32(arg->band_center_freq1);
>>>>>         if (arg->mode == MODE_11AC_VHT80_80)
>>>>>                 ch->band_center_freq2 = __cpu_to_le32(arg->band_center_freq2);
>>>>> -       else
>>>>> -               ch->band_center_freq2 = 0;
>>>>> +       if (arg->mode == MODE_11AC_VHT160)
>>>>> +               ch->band_center_freq2 = __cpu_to_le32(arg->band_center_freq1);
>>>>>         ch->min_power = arg->min_power;
>>>>>         ch->max_power = arg->max_power;
>>>>>         ch->reg_power = arg->max_reg_power;
>>>>
>>>> That would not appear to work with my FW, but, my FW's comments don't seem to match
>>>> it's code, so I am not sure where all the bugs lie.  Maybe re-check the source of this
>>>> patch above to make sure you got it right?
>>> its right if you already applied the vht160 patch which is in git already since a while (wireless-next)
>>> in openwrt the staging tree of nbd contains the same source which which fits to this patch
>>> but if it wont apply to you you have to update your whole wireless source to latest version since you did not have the required vht160 patches
>>
>> I mean the content of the patch is invalid, not a merge issue.
>>
>> Try setting freq1 to center-freq of the primary 80Mhz channel.
>> And set freq2 to center-freq of the 160Mhz channel
> so something like  that?
>
> +       if (arg->mode == MODE_11AC_VHT160)  {
> +               ch->band_center_freq1 = __cpu_to_le32(arg->freq);
> +               ch->band_center_freq2 = __cpu_to_le32(arg->band_center_freq1);
> +       }

No, I don't think so.  freq1 should be freq + 40 if freq < center-freq, otherwise
should be freq - 40.

freq2 settings above look correct.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k