Ath11k possible firmware bug - traffic interruptions when a client disconnects from WLAN

Just an update. I have been running OpenWrt SNAPSHOT, r23763-46ed38adeb on wrx36 with ath11k firmware downgraded to WLAN.HK.2.7.0.1-01744-QCAHKSWPL_SILICONZ-1 for two months (uptime 56 days) without problems.

Dual radio - channel 149 HE80 with sae-mixed encryption, channel 6 HE20 with psk2+ccmp encryption.

3 Likes

Where can I get the firmware version?

I'm now testing OpenWrt SNAPSHOT r24124-518923178c with WLAN.HK.2.7.0.1-01744-QCAHKSWPL_SILICONZ-1, starting today. We'll see :slight_smile:

1 Like
WLAN.HK.2.7.0.1-01744-QCAHKSWPL_SILICONZ-1
1 Like

Which directory do I need to place the fw files?

Edit; For the radio firmware simply replace content of /lib/firmware/IPQ8074 with e.g. content of the 1835 directory

See : https://github.com/egc112/OpenWRT-egc-add-on/tree/main/DL-WRX36 for instructions

1 Like

Ok, let's go :slight_smile:

Oh thanks... my wifi doesn't come up anymore.

At others: Do NOT insert 2.7.0.1-01744 into r24111 build. It won't work.

Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] qcom-q6v5-wcss-pil cd00000.q6v5_wcss: fatal error received: 
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] QC Image Version: QC_IMAGE_VERSION_STRING=WLAN.HK.2.7.0.1-01744-QCAHKSWPL_SILICONZ-1
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] Image Variant : IMAGE_VARIANT_STRING=8074.wlanfw.eval_v2Q
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] 
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292]     :Excep  :0 Exception detectedparam0 :zero, param1 :zero, param2 :zero.
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] Thread ID      : 0x00000069  Thread name    : WLAN RT0  Process ID     : 0
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] Register:
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] SP : 0x4bfacdc0
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] FP : 0x4bfacdd8
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] PC : 0x4b18d338
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] SSR : 0x00000001
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] BADVA : 0x009c9d7e
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] LR : 0x4b18d2b8
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] 
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] Stack Dump
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] from : 0x4bfacdc0
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] to   : 0x4bfad400
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   17.985292] 
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   18.031944] remoteproc remoteproc0: crash detected in cd00000.q6v5_wcss: type fatal error
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   18.054132] remoteproc remoteproc0: handling crash #1 in cd00000.q6v5_wcss
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   18.062217] remoteproc remoteproc0: recovering cd00000.q6v5_wcss
Oct 14 13:43:00 WifiAP-02-WZAX kernel: [   18.095005] remoteproc remoteproc0: stopped remote processor cd00000.q6v5_wcss

UPDATE: ath11k c000000.wifi: fw_version 0x290a84a5 fw_build_timestamp 2023-06-21 21:36 fw_build_id WLAN.HK.2.9.0.1-01837-QCAHKSWPL_SILICONZ-1

This works on r24111

That is the one I am using.

Not saying it is the best but works for me :slight_smile:

1 Like

Nope. WLAN.HK.2.7 and OpenWrt SNAPSHOT r24124-518923178c and toggle my iPhone wifi on/off a few times I get this:

Request timeout for icmp_seq 25
Request timeout for icmp_seq 26
Request timeout for icmp_seq 27
Request timeout for icmp_seq 28

Well I think I solved the broadcast issue I had with my airprinter. Not enabling ieee80211k - Enables Radio Resource Measurement (802.11k) support. .
I dunno why enabling ieee80211k blocks my airprinter to be discovered. Have to do some more testing about this issue. :face_with_monocle:

Edit: Nope... ieee80211k had nothing to do with it.

2 Likes

Snapshot r24111 with WLAN.HK.2.9.0.1-01837 still has the memory leak. Crashed after ~ 8 hours of uptime.

I'm back on r23845 which runs solid for days :).

I'm doing some searching for more or less the same topic, and see if there's any similar issues or fixes. I'm not sure if it's related, but I'll share what I found so far. All of this made me think that maybe we are on the wrong path, and the issue is somewhere else. Is it AQL/Airtime Fairness again?... (some of this goes way back too).

2 Likes

Hi @nbd , do you know when the code that sets he_bss_color in hostapd was changed?
Why I'm asking is that a clean install with OpenWrt SNAPSHOT, r24222-eacc885816 on a wax620 ipq807x sets he_bss_color=128 in /var/run/hostapd-phy1.conf.
128 is not valid.

# hostapd.conf 
...
# he_bss_color: BSS color (1-63)
#he_bss_color=1

I tried to find the commit that changed this, but I'm lost.

I might be wrong but I think a value greater than 63 makes hostapd set a random color.

I believe your are right. Found this... but does not say anything about it will become random. However, if hostapd works in range from 1-63 then 128 will not work?
Or is hostapd smart enough to know that if a number exceeds 63 it will generate a new one, but not save it in hostapd? I'm sorry if I don't understand this...

And this patch. But should this patch generate a random number if number 128 is defined in config? Or is the patch referring to /etc/config/wireless?:

A little update on my snapshot tour and hk wlan firmware.
This is still in progress, I'm still testing, but after I saw this one commit showing up a few days back I had to try it. I'm not using latest HK, and now I'm building with WLAN.HK.2.7.0.1-01744-QCAHKSWPL_SILICONZ-1. Running speedtest with latest HK is not stable in my experience. Latency is all over for some reason and not consistent. I'm gonna do some flent testing later between HK.2.7 and latest.

For now it's only 2 days but toggling my iPhones wifi on/off, I can't reproduce the latency issue. I have not experienced any drops either when stations are roaming or leaving.
No multicast to unicast option. Only added the arp setting as mentioned earlier.

-----------------------------------------------------
 OpenWrt SNAPSHOT, r24222-eacc885816
-----------------------------------------------------
1 Like

What is the arp setting, I'd like to try it.

Here you go. You put net.ipv4.conf.br-lan.arp_accept = 1 in /etc/sysctl.conf like so.

# nano /etc/sysctl.conf
net.ipv4.conf.br-lan.arp_accept = 1

Save sysctl.conf and run command sysctl -p or reboot.

Edit: There's also a ticket for this issue https://github.com/openwrt/openwrt/issues/9555.

1 Like

Any updates/changes with this, anyone?
I currently test @qosmio NSS wifi build and I no longer see interruptions when client disconnects from WLAN nor I can reproduce it deliberately disconnecting clients. I even disabled Multi to Unicast option but everything is OK.
At the time I was trying to reproduce the issue I was running both Ping an Iperf3 simultaneous tests on my Laptop connected the wifi. All tests were completely OK - no delays nor any interruption. At least this is after a reboot.
Will try again after several hours of work.

1 Like

There are some new observations on this topic here.

Edit - currently running latest ath11k firmware. Lets see after a few days, if there will be any change.