[FS#1044] Bug after mac80211 update to backports-4.14-rc2

[LEDE Bugs]

The following task has a new comment added:

FS#1044 - Bug after mac80211 update to backports-4.14-rc2 
User who did this - Christian Lamparter (chunkeey)

----------
Luckily, not really a bug that would affect the system much.

Sat Oct  7 15:35:43 2017 kern.warn kernel: [  428.341238] WARNING: CPU: 0 PID: 1347 at backports-4.14-rc2-1/net/wireless/util.c:1257 cfg80211_calculate_bitrate+0x14c/0x1f4 [cfg80211]
Sat Oct  7 15:35:43 2017 kern.warn kernel: [  428.344875] invalid rate bw=2, mcs=9, nss=1
Sat Oct  7 15:35:43 2017 kern.warn kernel: [  428.532448] [] (nl80211_dump_station

"invalid rate bw=2, mcs=9, nss=1" is kind of a weird problem. "bw=2" means 10MHz channel bandwidth (RATE_INFO_BW_10). However, mcs=9 is a VHT rate and VHT is only available for 40, 80 and 160MHz wide channels.


Now, where does the RATE_INFO_BW_10 come from? It doesn't look like ath10k, mac80211 or cfg80211 would sets RATE_INFO_BW_10 directly. But the RATE_INFO_BW_* flags where recently changed. So it could be a API problem. The prime suspect there is ath10k (My guess is that ath10k's code wasn't updated correctly).

For example: ath10k_update_per_peer_tx_stats()'s

arsta->txrate.bw = txrate.bw + RATE_INFO_BW_20;

before the change

commit 842be75c77cb72ee546a2b19da9c285fb3ded660
Author: Johannes Berg 
Date:   Thu May 4 08:42:30 2017 +0200

    cfg80211: make RATE_INFO_BW_20 the default
    
this would have worked fine. But now RATE_INFO_BW_20 is 0
and the ath10k's own rate definitions can no longer be mapped
by simply adding +RATE_INFO_BW_20. 

There might be more issues. You should definitely post this to linux-wireless at vger.kernel.org / ath10k-devel.
----------

More information can be found at the following URL:
https://bugs.lede-project.org/index.php?do=details&task_id=1044#comment3574