Radiotap Channel Bug

Radiotap Channel Bug

[Brandon Enochs]

With CentOS 6.5, the LP phy type currently suffers from a calculation
error in the way that it reports the channel on which a frame was
received up through the mac80211 stack.  It should use the
b43_channel_to_freq_2ghz function; however, it uses the
b43_freq_to_channel_2ghz instead.  As a result, the radiotap header
provides nonsensical values for the channel field.  The following
patch fixes this issue

Sincerely,
Brandon Enochs

diff -rupN ../linux-2.6.32-431.23.3.el6/drivers/net/wireless/b43/xmit.c
drivers/net/wireless/b43/xmit.c
--- ../linux-2.6.32-431.23.3.el6/drivers/net/wireless/b43/xmit.c
2014-07-16 05:33:22.000000000 -0400
+++ drivers/net/wireless/b43/xmit.c 2014-08-19 19:19:20.607959455 -0400
@@ -821,10 +821,10 @@ void b43_rx(struct b43_wldev *dev, struc
   * channel number in b43. */
  if (chanstat & B43_RX_CHAN_5GHZ) {
  status.band = IEEE80211_BAND_5GHZ;
- status.freq = b43_freq_to_channel_5ghz(chanid);
+ status.freq = b43_channel_to_freq_5ghz(chanid);
  } else {
  status.band = IEEE80211_BAND_2GHZ;
- status.freq = b43_freq_to_channel_2ghz(chanid);
+ status.freq = b43_channel_to_freq_2ghz(chanid);
  }
  break;
  default:

Radiotap Channel Bug

[Rafał Miłecki]

On 20 August 2014 01:36, Brandon Enochs <enochs.brandon@gmail.com> wrote:
> With CentOS 6.5, the LP phy type currently suffers from a calculation
> error in the way that it reports the channel on which a frame was
> received up through the mac80211 stack.  It should use the
> b43_channel_to_freq_2ghz function; however, it uses the
> b43_freq_to_channel_2ghz instead.  As a result, the radiotap header
> provides nonsensical values for the channel field.  The following
> patch fixes this issue

Consider updating kernel to some newer version. I know 2.6.32 is a
still supported longterm, but it may miss some things anyway.

It seems the frequency fix was applied back to 3.2 longterm only.

Radiotap Channel Bug

[Brandon Enochs]

Rafal,
Thanks, I'll request a back port in CentOS.

On Wed, Aug 20, 2014 at 1:04 AM, Rafa? Mi?ecki <zajec5@gmail.com> wrote:
> On 20 August 2014 01:36, Brandon Enochs <enochs.brandon@gmail.com> wrote:
>> With CentOS 6.5, the LP phy type currently suffers from a calculation
>> error in the way that it reports the channel on which a frame was
>> received up through the mac80211 stack.  It should use the
>> b43_channel_to_freq_2ghz function; however, it uses the
>> b43_freq_to_channel_2ghz instead.  As a result, the radiotap header
>> provides nonsensical values for the channel field.  The following
>> patch fixes this issue
>
> Consider updating kernel to some newer version. I know 2.6.32 is a
> still supported longterm, but it may miss some things anyway.
>
> It seems the frequency fix was applied back to 3.2 longterm only.

Radiotap Channel Bug

[Brandon Enochs]

Thanks,  I've submitted a bug report to the CentOS team to request a back port.

On Tue, Aug 19, 2014 at 7:36 PM, Brandon Enochs
<enochs.brandon@gmail.com> wrote:
> With CentOS 6.5, the LP phy type currently suffers from a calculation
> error in the way that it reports the channel on which a frame was
> received up through the mac80211 stack.  It should use the
> b43_channel_to_freq_2ghz function; however, it uses the
> b43_freq_to_channel_2ghz instead.  As a result, the radiotap header
> provides nonsensical values for the channel field.  The following
> patch fixes this issue
>
> Sincerely,
> Brandon Enochs
>
> diff -rupN ../linux-2.6.32-431.23.3.el6/drivers/net/wireless/b43/xmit.c
> drivers/net/wireless/b43/xmit.c
> --- ../linux-2.6.32-431.23.3.el6/drivers/net/wireless/b43/xmit.c
> 2014-07-16 05:33:22.000000000 -0400
> +++ drivers/net/wireless/b43/xmit.c 2014-08-19 19:19:20.607959455 -0400
> @@ -821,10 +821,10 @@ void b43_rx(struct b43_wldev *dev, struc
>    * channel number in b43. */
>   if (chanstat & B43_RX_CHAN_5GHZ) {
>   status.band = IEEE80211_BAND_5GHZ;
> - status.freq = b43_freq_to_channel_5ghz(chanid);
> + status.freq = b43_channel_to_freq_5ghz(chanid);
>   } else {
>   status.band = IEEE80211_BAND_2GHZ;
> - status.freq = b43_freq_to_channel_2ghz(chanid);
> + status.freq = b43_channel_to_freq_2ghz(chanid);
>   }
>   break;
>   default: