| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <Z3hGZLKQUvT5-8Al@skv.local>
Date: Fri, 3 Jan 2025 23:19:48 +0300
From: Andrey Skvortsov <andrej.skvortzov@...il.com>
To: Vasily Khoruzhick <anarsoul@...il.com>
Cc: Ping-Ke Shih <pkshih@...ltek.com>, Kalle Valo <kvalo@...nel.org>,
Fiona Klute <fiona.klute@....de>,
Bitterblue Smith <rtl8821cerfe2@...il.com>,
linux-wireless@...r.kernel.org, linux-kernel@...r.kernel.org,
stable@...r.kernel.org
Subject: Re: [PATCH] wifi: rtw88: 8703b: Fix RX/TX issues
On 25-01-03 11:10, Vasily Khoruzhick wrote:
> On Fri, Jan 3, 2025 at 11:02 AM Andrey Skvortsov
> <andrej.skvortzov@...il.com> wrote:
> >
> > On 25-01-02 23:50, Vasily Khoruzhick wrote:
> > > Fix 3 typos in 8703b driver. 2 typos in calibration routines are not
> > > fatal and do not seem to have any impact, just fix them to match vendor
> > > driver.
> > >
> > > However the last one in rtw8703b_set_channel_bb() clears too many bits
> > > in REG_OFDM0_TX_PSD_NOISE, causing TX and RX issues (neither rate goes
> > > above MCS0-MCS1). Vendor driver clears only 2 most significant bits.
> > >
> > > With the last typo fixed, the driver is able to reach MCS7 on Pinebook
> > >
> > > Cc: stable@...r.kernel.org
> > > Fixes: 9bb762b3a957 ("wifi: rtw88: Add definitions for 8703b chip")
> > > Signed-off-by: Vasily Khoruzhick <anarsoul@...il.com>
> >
> > Tested-by: Andrey Skvortsov <andrej.skvortzov@...il.com>
> >
> > Thank you, Vasily, for fixing that. Performance is much better with
> > the fix. Here are iperf results made on PinePhone:
>
> Thanks for testing!
>
> > 1. without the patch using rtw88 driver
> > 1.98 Mbits/sec
> >
> > 2. with the patch using rtw88 driver
> > 14.0 Mbits/sec
> >
> > 3. using old vendor 8723cs driver
> > 23.6 Mbits/sec
>
> Interesting, I get 30-50 Mbit/s on both rtw88 and vendor driver on
> Pinebook for either TX or RX, however I am pretty close (a few meters)
> to the AP and it isn't a noisy env (Pinebook can pick up just 4 other
> 2.4G APs besides mine, none on the same channel). Could you try
> disabling bluetooth (run bluetoothctl, and do "power off") and re-run
> the test?
yes, previous test were done with Bluetooth enabled. With disabled
(power off) Bluetooth performance is slightly better.
Here is how I tested to make sure, that we are on the same page:
1. on device 'iperf -s'
2. on PC 'iperf -c <device> -e -i 10 -t 250'
Here are more detailed testing results:
1. without the patch using rtw88 driver
[ 1] (icwnd/mss/irtt=14/1448/66504) (ct=66.55 ms)
[ ID] Interval Transfer Bandwidth Write/Err Rtry InF(pkts)/Cwnd(pkts)/RTT(var) NetPwr
[ 1] 0.0000-10.0000 sec 2.38 MBytes 2.00 Mbits/sec 20/0 73 22K(16)/24K(17)/56167(15843) us 4.448847
[ 1] 10.0000-20.0000 sec 5.88 MBytes 4.93 Mbits/sec 47/0 38 178K(126)/178K(126)/462265(40709) us 1.332652
[ 1] 20.0000-30.0000 sec 2.76 MBytes 2.31 Mbits/sec 25/0 182 12K(9)/12K(9)/32519(11137) us 8.892008
[ 1] 30.0000-40.0000 sec 2.80 MBytes 2.35 Mbits/sec 23/0 48 45K(32)/45K(32)/186889(21282) us 1.569355
[ 1] 40.0000-50.0000 sec 6.31 MBytes 5.29 Mbits/sec 51/0 24 188K(133)/188K(133)/574842(11821) us 1.151013
[ 1] 50.0000-60.0000 sec 5.06 MBytes 4.25 Mbits/sec 43/0 8 346K(245)/346K(245)/412227(75277) us 1.287768
[ 1] 60.0000-70.0000 sec 4.29 MBytes 3.60 Mbits/sec 38/0 81 284K(201)/284K(201)/339931(17905) us 1.323076
2. with the patch using rtw88 driver
[ 1] (icwnd/mss/irtt=14/1448/40241) (ct=40.27 ms)
[ ID] Interval Transfer Bandwidth Write/Err Rtry InF(pkts)/Cwnd(pkts)/RTT(var) NetPwr
[ 1] 0.0000-10.0000 sec 22.0 MBytes 18.5 Mbits/sec 176/0 0 1069K(756)/1090K(771)/466471(10659) us 4.945374
[ 1] 10.0000-20.0000 sec 20.5 MBytes 17.2 Mbits/sec 164/0 0 2075K(1468)/3919K(2772)/1453362(10765) us 1.479040
[ 1] 20.0000-30.0000 sec 16.7 MBytes 14.0 Mbits/sec 137/0 140 2099K(1485)/2589K(1831)/1466588(44692) us 1.193315
[ 1] 30.0000-40.0000 sec 16.8 MBytes 14.1 Mbits/sec 136/1 169 446K(316)/1914K(1354)/1511938(187) us 1.166601
[ 1] 40.0000-50.0000 sec 14.8 MBytes 12.4 Mbits/sec 121/1 6 1278K(904)/1278K(904)/4145469(8168) us 0.374480
[ 1] 50.0000-60.0000 sec 22.0 MBytes 18.5 Mbits/sec 177/1 26 905K(640)/1084K(767)/428675(700) us 5.389748
[ 1] 60.0000-70.0000 sec 18.1 MBytes 15.2 Mbits/sec 146/0 247 229K(162)/254K(180)/129034(4979) us 14.74
3. using old vendor 8723cs driver
[ 1] (icwnd/mss/irtt=14/1448/91025) (ct=91.07 ms)
[ ID] Interval Transfer Bandwidth Write/Err Rtry InF(pkts)/Cwnd(pkts)/RTT(var) NetPwr
[ 1] 0.0000-10.0000 sec 38.3 MBytes 32.1 Mbits/sec 306/0 0 1938K(1371)/1949K(1379)/419011(4587) us 9.572087
[ 1] 10.0000-20.0000 sec 36.8 MBytes 30.8 Mbits/sec 294/0 0 2258K(1597)/3809K(2694)/856058(270) us 4.501467
[ 1] 20.0000-30.0000 sec 38.4 MBytes 32.2 Mbits/sec 307/0 0 2360K(1669)/3809K(2694)/727066(112) us 5.534450
[ 1] 30.0000-40.0000 sec 38.5 MBytes 32.3 Mbits/sec 308/0 0 2371K(1677)/3809K(2694)/786411(100) us 5.133470
[ 1] 40.0000-50.0000 sec 40.8 MBytes 34.2 Mbits/sec 326/0 0 2563K(1813)/3809K(2694)/652361(18130) us 6.549973
[ 1] 50.0000-60.0000 sec 35.4 MBytes 29.7 Mbits/sec 283/0 0 2524K(1785)/5953K(4210)/670626(151) us 5.531157
[ 1] 60.0000-70.0000 sec 39.5 MBytes 33.1 Mbits/sec 316/0 0 2502K(1770)/5953K(4210)/707670(140) us 5.852834
> I noticed that btcoex is implemented differently in rtw88
> and in the vendor driver, so it might not be working correctly.
yeah, I couldn't make good-enough VoIP-calls on WiFi
using PinePhone with rtw88, but I could make it just fine with old
vendor driver. But I've switched to rtw88 anyway.
--
Best regards,
Andrey Skvortsov
Powered by blists - more mailing lists