Re: Slow connection with rtl8xxxu and 8192eu chipset

[Nikolay Borisov <n.borisov.lkml@gmail.com>]

On 28.11.2017 18:35, Jes Sorensen wrote:
> On 11/28/2017 11:26 AM, Nikolay Borisov wrote:
>>
>>
>> On 28.11.2017 18:16, Jes Sorensen wrote:
>>> On 11/14/2017 05:39 AM, Nikolay Borisov wrote:
>>>>
>>>>
>>>> On 14.11.2017 11:26, Nikolay Borisov wrote:
>>>>> (Please CC as I'm not subscribed)
>>>>>
>>>>> Hello,
>>>>>
>>>>> I have the tp-link tl-wn822N usb wifi dongle. lsbusb reports it as
>>>>>
>>>>> Bus 001 Device 003: ID 2357:0108
>>>>>
>>>>> Unfortunately with the in-kernel rtl8xxxu driver I don't get very good
>>>>> results:
>>>>>
>>>>> wifi1     IEEE 802.11  ESSID:"HOME"
>>>>>           Mode:Managed  Frequency:2.462 GHz  Access Point:
>>>>> 30:B5:C2:75:A4:CD
>>>>>           Bit Rate=1 Mb/s   Tx-Power=20 dBm
>>>>>           Retry short limit:7   RTS thr=2347 B   Fragment thr:off
>>>>>           Power Management:off
>>>>>           Link Quality=26/70  Signal level=-84 dBm
>>>>>           Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
>>>>>           Tx excessive retries:0  Invalid misc:165   Missed beacon:0
>>>>>
>>>>>
>>>>> At the same time if I use an out of tree driver acquired from github:
>>>>> https://github.com/Mange/rtl8192eu-linux-driver I get the following:
>>>>>
>>>>> wifi1     IEEE 802.11bgn  ESSID:"HOME"  Nickname:"<WIFI@REALTEK>"
>>>>>           Mode:Managed  Frequency:2.462 GHz  Access Point:
>>>>> 30:B5:C2:75:A4:CD
>>>>>           Bit Rate:144.4 Mb/s   Sensitivity:0/0
>>>>>           Retry:off   RTS thr:off   Fragment thr:off
>>>>>           Power Management:off
>>>>>           Link Quality=81/100  Signal level=100/100  Noise level=0/100
>>>>>           Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
>>>>>           Tx excessive retries:0  Invalid misc:0   Missed beacon:0
>>>>>
>>>>> Clearly this is a software problem of the in-kernel driver. I'm using
>>>>> v4.10.17 with commit c14239f23adb ("rtl8xxxu: Add another 8192eu device
>>>>> to the USB list") so that my device is recognised. Latest commit for
>>>>> that driver in my kernel is: c59f13bbead4 ("rtl8xxxu: Work around issue
>>>>> with 8192eu and 8723bu devices not reconnecting").
>>>>>
>>>>> Any ideas what I can do to further debug this, I'd really like to use
>>>>> the in-kernel driver ?
>>>>
>>>> I just tested with verbatim 4.14 and even though the wireless works, 
>>>> iwconfig reports something strange: 
>>>>
>>>> iwconfig wifi1
>>>> wifi1     no wireless extensions.
>>>>
>>>> However, my device works as expected (albeit still slow): 
>>>>
>>>> wifi1     Link encap:Ethernet  HWaddr 18:d6:c7:0d:47:3c  
>>>>           inet addr:10.20.1.175  Bcast:10.20.1.255  Mask:255.255.255.0
>>>>           inet6 addr: fe80::281d:5f27:eb1b:8ded/64 Scope:Link
>>>>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>>>>           RX packets:38903 errors:0 dropped:0 overruns:0 frame:0
>>>>           TX packets:24689 errors:0 dropped:0 overruns:0 carrier:0
>>>>           collisions:0 txqueuelen:1000 
>>>>           RX bytes:51524413 (51.5 MB)  TX bytes:5503039 (5.5 MB)
>>>
>>> iwconfig has been deprecated for a decade, but maybe you could elaborate
>>> on what you feel is strange in this output?
>>
>> Well there are 2 things:
>>
>> 1. The low bitrate - which I've confirmed doing actual data test, ie,
>> not just the number but the fact that the speeds are low.
>>
>> 2. The worse signal level/link quality.
> 
> I don't see either of those in the iwconfig output
> 
> The low TX bitrate is a known issue. Somehow the firmware rate handling
> doesn't seem to work (or it works differently than I assumed) similar to
> for the 8188eu. It's something I need to figure out when I have some time.
> 
> As for link quality is this something you measured or read out of 'iw' ?

Well there is:

Link Quality=26/70  Signal level=-84 dBm < - with rtl8xxxu driver

Link Quality=81/100  Signal level=100/100  Noise level=0/100 <-- with
vendor's driver. Another vector was the icon of network manager had only
2 bars with the in-tree driver and had full bars with out (yeah, this is
a bit subjective metric but it's the best i've got).

> 
> Thanks,
> Jes
>