[RFC 1/1] ath9k: Add radiotap vendor specific data example

[RFC 1/1] ath9k: Add radiotap vendor specific data example

[Wojciech Dubowik]

As an exmaple of vendor specific data we can extract rssi
and evm values for all the chains so one can see them on
per packet basis. More can be added depending on the needs i.e.
noise floor, Ness, STBC, key miss, etc.
AtheorsC OUI is used in this example but it should be replaced
by own when in use.

Signed-off-by: Wojciech Dubowik <Wojciech.Dubowik@neratec.com>
---
 drivers/net/wireless/ath/ath9k/recv.c |   33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/drivers/net/wireless/ath/ath9k/recv.c b/drivers/net/wireless/ath/ath9k/recv.c
index 6aafbb7..02c36c4 100644
--- a/drivers/net/wireless/ath/ath9k/recv.c
+++ b/drivers/net/wireless/ath/ath9k/recv.c
@@ -1246,6 +1246,39 @@ int ath_rx_tasklet(struct ath_softc *sc, int flush, bool hp)
 		if ((ah->caps.hw_caps & ATH9K_HW_CAP_ANT_DIV_COMB) && sc->ant_rx == 3)
 			ath_ant_comb_scan(sc, &rs);
 
+#if 0
+		/*TODO: Replace oui and vendor sub namespace with valid data
+		 * and add/remove fields below
+		 */
+		if( skb_headroom(skb) >  22 ) {
+			int8_t* vendor_data;
+
+			vendor_data = skb_push(skb, 22);
+
+			rxs->vendor_radiotap_len = 22;
+			rxs->vendor_radiotap_align = 4;
+			rxs->vendor_radiotap_oui[0] = 0x0;
+			rxs->vendor_radiotap_oui[1] = 0x3;
+			rxs->vendor_radiotap_oui[2] = 0x7f;
+			rxs->vendor_radiotap_subns = 0;
+			rxs->vendor_radiotap_bitmap = 0x00000001;
+
+			/* count bytes below and put it as the length*/
+			*vendor_data++ = rs.rs_rssi;
+			*vendor_data++ = rs.rs_rssi_ctl0;
+			*vendor_data++ = rs.rs_rssi_ctl1;
+			*vendor_data++ = rs.rs_rssi_ctl2;
+			put_unaligned_le32(rs.evm0, vendor_data);
+			vendor_data += 4;
+			put_unaligned_le32(rs.evm1, vendor_data);
+			vendor_data += 4;
+			put_unaligned_le32(rs.evm2, vendor_data);
+			vendor_data += 4;
+			put_unaligned_le32(rs.evm3, vendor_data);
+			vendor_data += 4;
+			put_unaligned_le16(rs.evm4, vendor_data);
+		}
+#endif
 		ieee80211_rx(hw, skb);
 
 requeue_drop_frag:
-- 
1.7.10.4

Re: [RFC 1/1] ath9k: Add radiotap vendor specific data example

[Mohammed Shafi]

On Fri, Nov 23, 2012 at 4:16 PM, Wojciech Dubowik
<Wojciech.Dubowik@neratec.com> wrote:
> As an exmaple of vendor specific data we can extract rssi
> and evm values for all the chains so one can see them on
> per packet basis. More can be added depending on the needs i.e.
> noise floor, Ness, STBC, key miss, etc.
> AtheorsC OUI is used in this example but it should be replaced
> by own when in use.
>
> Signed-off-by: Wojciech Dubowik <Wojciech.Dubowik@neratec.com>
> ---
>  drivers/net/wireless/ath/ath9k/recv.c |   33 +++++++++++++++++++++++++++++++++
>  1 file changed, 33 insertions(+)
>
> diff --git a/drivers/net/wireless/ath/ath9k/recv.c b/drivers/net/wireless/ath/ath9k/recv.c
> index 6aafbb7..02c36c4 100644
> --- a/drivers/net/wireless/ath/ath9k/recv.c
> +++ b/drivers/net/wireless/ath/ath9k/recv.c
> @@ -1246,6 +1246,39 @@ int ath_rx_tasklet(struct ath_softc *sc, int flush, bool hp)
>                 if ((ah->caps.hw_caps & ATH9K_HW_CAP_ANT_DIV_COMB) && sc->ant_rx == 3)
>                         ath_ant_comb_scan(sc, &rs);
>
> +#if 0

please remove ifdefs .


> +               /*TODO: Replace oui and vendor sub namespace with valid data
> +                * and add/remove fields below
> +                */
> +               if( skb_headroom(skb) >  22 ) {
> +                       int8_t* vendor_data;

we can use s8 ?


> +
> +                       vendor_data = skb_push(skb, 22);
> +
> +                       rxs->vendor_radiotap_len = 22;
> +                       rxs->vendor_radiotap_align = 4;
> +                       rxs->vendor_radiotap_oui[0] = 0x0;
> +                       rxs->vendor_radiotap_oui[1] = 0x3;
> +                       rxs->vendor_radiotap_oui[2] = 0x7f;
> +                       rxs->vendor_radiotap_subns = 0;
> +                       rxs->vendor_radiotap_bitmap = 0x00000001;
> +
> +                       /* count bytes below and put it as the length*/
> +                       *vendor_data++ = rs.rs_rssi;
> +                       *vendor_data++ = rs.rs_rssi_ctl0;
> +                       *vendor_data++ = rs.rs_rssi_ctl1;
> +                       *vendor_data++ = rs.rs_rssi_ctl2;
> +                       put_unaligned_le32(rs.evm0, vendor_data);
> +                       vendor_data += 4;
> +                       put_unaligned_le32(rs.evm1, vendor_data);
> +                       vendor_data += 4;
> +                       put_unaligned_le32(rs.evm2, vendor_data);
> +                       vendor_data += 4;
> +                       put_unaligned_le32(rs.evm3, vendor_data);
> +                       vendor_data += 4;
> +                       put_unaligned_le16(rs.evm4, vendor_data);
> +               }
> +#endif


would be nice check this stuff in sniffer.

>                 ieee80211_rx(hw, skb);
>
>  requeue_drop_frag:
> --
> 1.7.10.4
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html



-- 
thanks,
shafi

Re: [RFC 1/1] ath9k: Add radiotap vendor specific data example

[Wojciech Dubowik]

On 11/24/2012 12:22 PM, Mohammed Shafi wrote:
> On Fri, Nov 23, 2012 at 4:16 PM, Wojciech Dubowik
> <Wojciech.Dubowik@neratec.com> wrote:
>> As an exmaple of vendor specific data we can extract rssi
>> and evm values for all the chains so one can see them on
>> per packet basis. More can be added depending on the needs i.e.
>> noise floor, Ness, STBC, key miss, etc.
>> AtheorsC OUI is used in this example but it should be replaced
>> by own when in use.
>>
>> Signed-off-by: Wojciech Dubowik <Wojciech.Dubowik@neratec.com>
>> ---
>>   drivers/net/wireless/ath/ath9k/recv.c |   33 +++++++++++++++++++++++++++++++++
>>   1 file changed, 33 insertions(+)
>>
>> diff --git a/drivers/net/wireless/ath/ath9k/recv.c b/drivers/net/wireless/ath/ath9k/recv.c
>> index 6aafbb7..02c36c4 100644
>> --- a/drivers/net/wireless/ath/ath9k/recv.c
>> +++ b/drivers/net/wireless/ath/ath9k/recv.c
>> @@ -1246,6 +1246,39 @@ int ath_rx_tasklet(struct ath_softc *sc, int flush, bool hp)
>>                  if ((ah->caps.hw_caps & ATH9K_HW_CAP_ANT_DIV_COMB) && sc->ant_rx == 3)
>>                          ath_ant_comb_scan(sc, &rs);
>>
>> +#if 0
> please remove ifdefs .
I guess radiotap can bring quite some overhead so myabe instead of 
ifdefs I could
implement sysfs entry to enable/disable it? Could be also used to switch 
needed
fields suggested by Johannes.
>
>
>> +               /*TODO: Replace oui and vendor sub namespace with valid data
>> +                * and add/remove fields below
>> +                */
>> +               if( skb_headroom(skb) >  22 ) {
>> +                       int8_t* vendor_data;
> we can use s8 ?
Will do.
>
>
>> +
>> +                       vendor_data = skb_push(skb, 22);
>> +
>> +                       rxs->vendor_radiotap_len = 22;
>> +                       rxs->vendor_radiotap_align = 4;
>> +                       rxs->vendor_radiotap_oui[0] = 0x0;
>> +                       rxs->vendor_radiotap_oui[1] = 0x3;
>> +                       rxs->vendor_radiotap_oui[2] = 0x7f;
>> +                       rxs->vendor_radiotap_subns = 0;
>> +                       rxs->vendor_radiotap_bitmap = 0x00000001;
>> +
>> +                       /* count bytes below and put it as the length*/
>> +                       *vendor_data++ = rs.rs_rssi;
>> +                       *vendor_data++ = rs.rs_rssi_ctl0;
>> +                       *vendor_data++ = rs.rs_rssi_ctl1;
>> +                       *vendor_data++ = rs.rs_rssi_ctl2;
>> +                       put_unaligned_le32(rs.evm0, vendor_data);
>> +                       vendor_data += 4;
>> +                       put_unaligned_le32(rs.evm1, vendor_data);
>> +                       vendor_data += 4;
>> +                       put_unaligned_le32(rs.evm2, vendor_data);
>> +                       vendor_data += 4;
>> +                       put_unaligned_le32(rs.evm3, vendor_data);
>> +                       vendor_data += 4;
>> +                       put_unaligned_le16(rs.evm4, vendor_data);
>> +               }
>> +#endif
>
> would be nice check this stuff in sniffer.
You mean dissect it?
>
>>                  ieee80211_rx(hw, skb);
>>
>>   requeue_drop_frag:
>> --
>> 1.7.10.4
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>

Re: [RFC 1/1] ath9k: Add radiotap vendor specific data example

[Johannes Berg]

On Fri, 2012-11-23 at 11:46 +0100, Wojciech Dubowik wrote:

> +			rxs->vendor_radiotap_bitmap = 0x00000001;

I would recommend adding multiple fields so that it's extensible in the
future or some things can be left out:

field 0 (0x1):
	4 x u8, alignment 1 - RSSI data
field 1 (0x2):
	4 x u32 + 1 x u16, alignment ? - evm data


Also:
Ness, STBC are going to be standardised.
noise floor is already standardised I think

johannes

Re: [RFC 1/1] ath9k: Add radiotap vendor specific data example

[Wojciech Dubowik]

On 11/24/2012 02:44 PM, Johannes Berg wrote:
> On Fri, 2012-11-23 at 11:46 +0100, Wojciech Dubowik wrote:
>
>> +			rxs->vendor_radiotap_bitmap = 0x00000001;
> I would recommend adding multiple fields so that it's extensible in the
> future or some things can be left out:
>
> field 0 (0x1):
> 	4 x u8, alignment 1 - RSSI data
> field 1 (0x2):
> 	4 x u32 + 1 x u16, alignment ? - evm data
I will try to add fields which could be switched on/off through sysfs
with bitmask.

Wojtek
>
>
> Also:
> Ness, STBC are going to be standardised.
> noise floor is already standardised I think
>
> johannes
>