AQL and the ath10k is *lovely*

Quick question: Is there a simple way to extract some per AC stats from ath10K?

For ath9K the following works:

root@turris:~# cat /sys/kernel/debug/ieee80211/phy1/ath9k/xmit
                            BE         BK        VI        VO

MPDUs Queued:            17592         27       242    434014
MPDUs Completed:        307212         59       199    435288
MPDUs XRetried:           6084          0        78      1034
Aggregates:           11477580        178      1493         0
AMPDUs Queued HW:            0          0         0         0
AMPDUs Completed:    113013059       2511     30838         0
AMPDUs Retried:       11172087         48      1860         0
AMPDUs XRetried:         26587          0       227         0
TXERR Filtered:           2847          0         6         1
FIFO Underrun:               0          0         0         0
TXOP Exceeded:               0          0         0         0
TXTIMER Expiry:              0          0         0         0
DESC CFG Error:              0          0         0         0
DATA Underrun:               0          0         0         0
DELIM Underrun:              0          0         0         0
TX-Pkts-All:         113352942       2570     31342    436322
TX-Bytes-All:        749970768    1834235  28708511  90088314
HW-put-tx-buf:             252         16        41       259
HW-tx-start:          30583175       2097     20661    436322
HW-tx-proc-desc:      30592420       2097     20661    436313
TX-Failed:                   0          0         0         0

I am looking for something comparable for ath10K, so if anybody has any pointers, I would be grateful.

1 Like

I want to note here that thanks to @qosmio we now have this commit that adds mac80211 based AQL support for ath11k_nss.
So can you guys recommend any commands or tests to perform in order to see (and possibly improve/decrease) wifi bufferbloat.
Additionally this commit made threaded NAPI for ath10k and ath11k.

3 Likes

I was not able to make it work for non NSS build :confused: That was snapshot 6.6.35.

Edit: Bad description... what's happening is that as soon as one client connects to the wifi, it reboots. Not able to get any log.

Do before/after benchmarks exist for this chipset?

Sadly no, at least not any I'm aware of.
There were latency issues reported by some testers and for now the commit that added AQL support was reverted. Personally I thought of it as an alfa version to allow first AQL tests.
Currently the NSS-wifi branch is very often and actively updated by @qosmio and there are frequent commits so maybe when it settles down a bit we can make some tests to compare.
Especially if you can advise about those.

These days my volunteer time is very limited, tho I do try to check the forums at least once a week. On reviewing the threaded patch, I did not see anything wrong with it. I have however tried to get folk to reduce the NAPI_BUDGET to something saner than 64 for wifi drivers, like 8. Maybe the threading code works better than that, I don't know!

4 Likes

Our beloved Dave Täht has passed away. Your kindness, wisdom, and passion will forever be remembered. We will always miss you. Rest in peace.

9 Likes

this should be pushed up to the top of the list...

I had the pleasure of discussing things with him some time back - always the gentleman...

2 Likes
4 Likes

Thanks for sharing @vochong - we at LibreQoS are pretty devastated by his sudden departure to the other side....

This was the last Dave's podcast appearance: https://www.youtube.com/watch?v=CQX1PmRULU0

5 Likes