Ath10k Issues on LEDE

For Developers
1

I'm running the following LEDE build on my Archer C7 v2: Lede Reboot CURRENT r2449+2
This is an airtime fairness build which has some patches applied to ath9k. Ath10k should be the same as a regular LEDE build. (downloaded from: https://kau.toke.dk/lede/airtime-fairness-builds/ar71xx/generic/)

My phone is experiencing issues on the 5ghz band. During heavy load, it would randomly disconnect and then immediately reconnect. During those occurrences I can find this in the system log:

Mon Dec 12 17:28:25 2016 daemon.info hostapd: wlan0: STA c0:bd:d1:d4:2c:26 IEEE 802.11: disconnected due to excessive missing ACKs
Mon Dec 12 17:28:28 2016 daemon.info hostapd: wlan0: STA c0:bd:d1:d4:2c:26 IEEE 802.11: authenticated
Mon Dec 12 17:28:28 2016 daemon.info hostapd: wlan0: STA c0:bd:d1:d4:2c:26 IEEE 802.11: associated (aid 6)

Sometimes however, the wifi would completely freeze on my phone. Pinging the router is no longer possible and the system log shows no messages related to this issue. Disconnecting and reconnecting to the wifi manually solves this problem.

My phone experiences no issues on other 5ghz networks, including the Archer C7 running the stock firmware. Does anyone happen to have a clue what my issue is, and more importantly, how I can solve it?

I've tried the "disassoc_low_ack 0" option in the wireless configuration file. While it successfully solves the first problem, the second problem of the wifi completely freezing is still there.

2

I found someone having a similar issue. He recommended disabling Wi-Fi scanning and Bluetooth scanning in the "improve accuracy" menu in Android. It seems like this did the trick. No more disconnects under heavy speed tests on my phone at 200Mbit/sec. I will continue to monitor the situation and hopefully it will stay like this. I hope that this message will be useful for people experiencing similar problems as I was :slight_smile:

3

Please try if using the latest version from my staging tree at https://git.lede-project.org/?p=lede/nbd/staging.git;a=summary helps.

1 Like
4

I didn't have time to test the latest build yet. I will have time to do that tomorrow. Shall I compile and flash the build that was latest at the the time of your post? Or should I flash the latest build at the time of compilation tomorrow?

5

Ah, I've just noticed that your latest Ath10k patches are now included in the main branch. I will compile the main branch then, and report back. Thanks again for the patches :slight_smile:

6

I flashed the latest snapshot a few hours ago and things are looking stable. Haven't had to use the "disassoc_low_ack 0" workaround and haven't seen any disconnects or freezes. I'll report back if I run into any more problems. Thank you very much for the patches, @nbd

7

@nbd stability is much better. No more disconnects due to excessive missing acks so far and no complete freezes either. I still have a few minor issues and some weird entries in the log, and I'm not completely sure whether these are normal, due to a client error, or due the a LEDE bug.

  1. This is the most important issue for me. After my phone (Galaxy s6) has been connected to the Ath10k AP for a while, suddenly the speeds drop significantly. I'm still seeing a 351 mbp/s link rate on my phone. But when running speedtests I'm getting only 20-30 mbp/s. A quick disconnect and reconnect on my phone restores these speeds back to 100+ mpb/s.

  2. The following two "issues" are not issues perse. Just some weird entries in the logs that I've noticed. I have no idea whether these are causing any troubles or whether these are related to my first issue. I'm getting these messages in the system log which I haven't seen before. Wlan1 is the Ath9k connection:

Sat Dec 17 08:48:41 2016 daemon.info hostapd: wlan1: STA 64:bc:0c:81:b2:9b WPA: received EAPOL-Key msg 2/4 in invalid state (10) - dropped
Sat Dec 17 08:48:41 2016 daemon.info hostapd: wlan1: STA 64:bc:0c:81:b2:9b WPA: received EAPOL-Key msg 2/4 in invalid state (10) - dropped
Sat Dec 17 08:48:41 2016 daemon.info hostapd: wlan1: STA 64:bc:0c:81:b2:9b WPA: received EAPOL-Key msg 2/4 in invalid state (10) - dropped
Sat Dec 17 08:48:41 2016 daemon.info hostapd: wlan1: STA 64:bc:0c:81:b2:9b WPA: received EAPOL-Key msg 2/4 in invalid state (10) - dropped
Sat Dec 17 08:48:41 2016 daemon.info hostapd: wlan1: STA 64:bc:0c:81:b2:9b WPA: received EAPOL-Key msg 2/4 in invalid state (10) - dropped
Sat Dec 17 08:48:41 2016 daemon.info hostapd: wlan1: STA 64:bc:0c:81:b2:9b WPA: received EAPOL-Key msg 2/4 in invalid state (10) - dropped

  1. I'm getting these messages about Ath10k in the kernel log every now and then:

[ 3210.073687] ath10k_pci 0000:01:00.0: failed to parse phyerr tlv payload at byte 0
[ 4191.138706] ath10k_pci 0000:01:00.0: failed to parse phyerr tlv payload at byte 0
[29546.739983] ath10k_pci 0000:01:00.0: failed to parse phyerr tlv payload at byte 0

Thanks again for the major help providing, and hopefully this feedback is helpful in debugging :slight_smile: