Edimax Nano, disconnects and packet loss with Ubuntu image

[polygnwnd]

Am using the recommended Edimax Nano USB wireless adapter and it is performing terribly–

• iwconfig shows lots of “Invalid Misc” packets, but no errors.
• pinging the host from the wireless router results in high variability in response time, with about 25% packets lost and about 30% with >1000ms times
• host disappears from router entirely after about an hour

All tests run with OpenSprinkler host in the same room with router (clear line of sight), and this is not an area with lots of interference (laptop next to OSBo happily pulls 90mbps from the same router).

I see lots of discussion around the Linux drivers for this device (e.g. here and here) and am curious whether the driver in the OSBo custom image is known to work, or I need to compile an update

[polygnwnd]

Some more debug info (while connected via wired Ethernet):

uname

root@arm:~# uname -a

Linux arm 3.8.13-bone28 #1 SMP Fri Sep 13 01:11:14 UTC 2013 armv7l armv7l armv7l GNU/Linux

dmesg when USB wifi adapter inserted:

[  188.176953] usb usb1: usb wakeup-resume

[  188.177052] usb usb1: usb auto-resume

[  188.177105] hub 1-0:1.0: hub_resume

[  188.177225] hub 1-0:1.0: port 1: status 0101 change 0001

[  188.281147] hub 1-0:1.0: state 7 ports 1 chg 0002 evt 0000

[  188.281274] hub 1-0:1.0: port 1, status 0101, change 0000, 12 Mb/s

[  188.386435] usb 1-1: new high-speed USB device number 2 using musb-hdrc

[  188.506564] usb 1-1: default language 0x0409

[  188.507427] usb 1-1: udev 2, busnum 1, minor = 1

[  188.507473] usb 1-1: New USB device found, idVendor=7392, idProduct=7811

[  188.507511] usb 1-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3

[  188.507545] usb 1-1: Product: 802.11n WLAN Adapter

[  188.507576] usb 1-1: Manufacturer: Realtek

[  188.507608] usb 1-1: SerialNumber: 00xxxx00000x

[  188.511681] usb 1-1: usb_probe_device

[  188.511742] usb 1-1: configuration #1 chosen from 1 choice

[  188.513465] usb 1-1: adding 1-1:1.0 (config #1, interface 0)

[  188.519681] hub 1-0:1.0: state 7 ports 1 chg 0000 evt 0002

[  188.519802] hub 1-0:1.0: port 1 enable change, status 00000503

[  189.260354] cfg80211: Calling CRDA to update world regulatory domain

[  189.496848] rtl8192cu 1-1:1.0: usb_probe_interface

[  189.496919] rtl8192cu 1-1:1.0: usb_probe_interface - got id

[  189.497346] rtl8192cu: Chip version 0x10

[  189.611075] rtl8192cu: MAC address: 80:1f:02:bb:fe:10

[  189.611153] rtl8192cu: Board Type 0

[  189.611395] rtlwifi: rx_max_size 15360, rx_urb_num 8, in_ep 1

[  189.611783] rtl8192cu: Loading firmware rtlwifi/rtl8192cufw.bin

[  189.632273] usbcore: registered new interface driver rtl8192cu

[  189.773844] ieee80211 phy0: Selected rate control algorithm 'rtl_rc'

[  189.786898] rtlwifi: wireless switch is on

lsusb

root@arm:~# lsusb

Bus 001 Device 002: ID 7392:7811 Edimax Technology Co., Ltd EW-7811Un 802.11n Wireless Adapter [Realtek RTL8188CUS]

Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

lshw

root@arm:~# lshw -C network -sanitize

*-network:0

description: Ethernet interface

physical id: 1

logical name: eth0

serial: [REMOVED]

capabilities: ethernet physical

configuration: broadcast=yes driver=TI CPSW Driver v1.0 driverversion=1.0 ip=[REMOVED] link=yes multicast=yes

*-network:1

description: Ethernet interface

physical id: 2

logical name: usb0

serial: [REMOVED]

capabilities: ethernet physical

configuration: broadcast=yes driver=g_ether driverversion=29-May-2008 firmware=musb-hdrc ip=[REMOVED] link=no multicast=yes

*-network:2 DISABLED

description: Wireless interface

physical id: 3

bus info: usb@1:1

logical name: wlan1

serial: [REMOVED]

capabilities: ethernet physical wireless

configuration: broadcast=yes driver=rtl8192cu driverversion=3.8.13-bone28 firmware=N/A link=no multicast=yes wireless=IEEE 802.11bgn

lsmod

root@arm:~# lsmod

Module                  Size  Used by

arc4                    1660  2

rtl8192cu              87084  0

rtlwifi                76713  1 rtl8192cu

rtl8192c_common        59545  1 rtl8192cu

mac80211              494605  3 rtlwifi,rtl8192c_common,rtl8192cu

cfg80211              421373  2 mac80211,rtlwifi

rfkill                 18327  1 cfg80211

g_multi                56263  0

libcomposite           17089  1 g_multi

Now bringing up wifi interface

root@arm:~# ifup wlan1

Internet Systems Consortium DHCP Client 4.2.4

Copyright 2004-2012 Internet Systems Consortium.

All rights reserved.

For info, please visit https://www.isc.org/software/dhcp/



Listening on LPF/wlan1/80:1f:02:bb:fe:10

Sending on   LPF/wlan1/80:1f:02:bb:fe:10

Sending on   Socket/fallback

DHCPREQUEST of 192.168.1.180 on wlan1 to 255.255.255.255 port 67 (xid=0x230b0c4f)

DHCPREQUEST of 192.168.1.180 on wlan1 to 255.255.255.255 port 67 (xid=0x230b0c4f)

DHCPREQUEST of 192.168.1.180 on wlan1 to 255.255.255.255 port 67 (xid=0x230b0c4f)

DHCPACK of 192.168.1.180 from 192.168.1.1

bound to 192.168.1.180 -- renewal in 38388 seconds.

dmesg now says

[  744.386533] rtl8192cu: MAC auto ON okay!

[  744.454592] rtl8192cu: Tx queue select: 0x05

[  744.912888] IPv6: ADDRCONF(NETDEV_UP): wlan1: link is not ready

[  747.610632] wlan1: authenticate with [WIRELESS_AP_MAC]

[  747.633050] wlan1: send auth to [WIRELESS_AP_MAC] (try 1/3)

[  747.833633] wlan1: send auth to [WIRELESS_AP_MAC] (try 2/3)

[  748.034804] wlan1: send auth to [WIRELESS_AP_MAC] (try 3/3)

[  748.235955] wlan1: authentication with [WIRELESS_AP_MAC] timed out

[  749.145721] wlan1: authenticate with [WIRELESS_AP_MAC]

[  749.166245] wlan1: send auth to [WIRELESS_AP_MAC] (try 1/3)

[  749.366823] wlan1: send auth to [WIRELESS_AP_MAC] (try 2/3)

[  749.567988] wlan1: send auth to [WIRELESS_AP_MAC] (try 3/3)

[  749.769168] wlan1: authentication with [WIRELESS_AP_MAC] timed out

[  750.674845] wlan1: authenticate with [WIRELESS_AP_MAC]

[  750.696906] wlan1: send auth to [WIRELESS_AP_MAC] (try 1/3)

[  750.898088] wlan1: send auth to [WIRELESS_AP_MAC] (try 2/3)

[  751.099271] wlan1: send auth to [WIRELESS_AP_MAC] (try 3/3)

[  751.300425] wlan1: authentication with [WIRELESS_AP_MAC] timed out

[  752.208067] wlan1: authenticate with [WIRELESS_AP_MAC]

[  752.229125] wlan1: send auth to [WIRELESS_AP_MAC] (try 1/3)

[  752.429323] wlan1: send auth to [WIRELESS_AP_MAC] (try 2/3)

[  752.630493] wlan1: send auth to [WIRELESS_AP_MAC] (try 3/3)

[  752.831660] wlan1: authentication with [WIRELESS_AP_MAC] timed out

[  753.748137] wlan1: authenticate with [WIRELESS_AP_MAC]

[  753.768598] wlan1: send auth to [WIRELESS_AP_MAC] (try 1/3)

[  753.772072] wlan1: authenticated

[  753.775149] wlan1: associate with [WIRELESS_AP_MAC] (try 1/3)

[  753.789979] wlan1: RX AssocResp from [WIRELESS_AP_MAC] (capab=0x431 status=0 aid=1)

[  753.791445] wlan1: associated

[  753.791551] IPv6: ADDRCONF(NETDEV_CHANGE): wlan1: link becomes ready

iwconfig after a minute or so:

root@arm:~# iwconfig wlan1

wlan1     IEEE 802.11bgn  ESSID:"[MY_NETWORK_SSID]"

Mode:Managed  Frequency:2.462 GHz  Access Point: [WIRELESS_AP_MAC]

Bit Rate=72.2 Mb/s   Tx-Power=20 dBm

Retry  long limit:7   RTS thr=2347 B   Fragment thr:off

Encryption key:off

Power Management:off

Link Quality=70/70  Signal level=-35 dBm

Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0

Tx excessive retries:0  Invalid misc:12   Missed beacon:0

[polygnwnd]

So the kernel drivers seem awful, and there is a Realtek-supplied driver that works a lot better (at least for me). Latest version seems to be available from https://github.com/pvaret/rtl8192cu-fixes/blob/master/README.md

Setup is well documented elsewhere, but there were some minor variations to make it work on the OSBo image:

1. Install DKMS support for compiled kernel modules

   # build-essential is already on OSBo image
   
   apt-get install dkms 

2. Download precompiled BB headers for OSBo kernel version

   wget http://rcn-ee.net/deb/wheezy-armhf/v3.8.13-bone28/linux-headers-3.8.13-bone28_1.0wheezy_armhf.deb
   
   dpkg --install linux-headers-3.8.13-bone28_1.0wheezy_armhf.deb
   
   

3. Download and install patched driver

   
   
   git clone https://github.com/pvaret/rtl8192cu-fixes.git
   
   dkms add ./rtl8192cu-fixes
   
   dkms install install 8192cu/1.8
   
   # Watch your build fail due to missing timex.h dependency
   
   

4. Edit /usr/src/linux-headers-3.8.13-bone28/arch/arm/include/asm/timex.h on line 18 to change

   #include 

   to

   #include 

   This looks pointless, but allows the build to complete (run dkms install again)

5. Blacklist the native kernel module so the newly compiled module will be used instead

   cp ./rtl8192cu-fixes/blacklist-native-rtl8192.conf /etc/modprobe.d/

6. I’ve also disabled radio power saving as leaving it enabled fills the dmesg log with status messages and causes occasional disconnects from AP. Put the following in a new /etc/modprobe.d/8192cu.conf file

   options 8192cu rtw_power_mgnt=0 rtw_enusbss=0

Obviously these instructions and links/URLs will not age well– they work nicely as of 17 December 2013.

Now ping is not losing packets:

521 packets transmitted, 521 packets received, 0.0% packet loss

round-trip min/avg/max/stddev = 2.450/6.857/19.328/1.572 ms

[Ray]

Thanks for sharing your solution. Sounds great. I will give it a try and include it in the SD card image.

[amigoloko]

Dear Ray,

After the date: Tue Dec 17, 2013 1:12 pm and Wed Dec 18, 2013 8:34 am
of this topic,

where you able to repair/upgrade the kernel, and successfully included in all new sd images?

Regards

[Ray]

Yes, working on this and should get it done within a day or two.

[jgoldin9083]

I attempted the fix, however I noticed something strange. the wifi works consistantly while the ethernet is plugged in, but when the ethernet is unplugged the wifi ceases to work. Any ideas?

[Ray]

So I followed these steps and I am still not able to get the Edimax Nano connected reliably at all times. The only reliable solution is to use a powered USB hub. When using the powered USB hub, the WiFi works perfectly fine with no disconnection. I think it’s more of a hardware problem (either the 24V AC to 5V DC power conversion, or the PCB design of the BeagleBone Black), and not so much of a software problem.

[polygnwnd]

Could disabling HDMI (add cape disable parameter to u-boot), per https://groups.google.com/forum/#!msg/beagleboard/sIGi77LojUE/NBmoSC9_PlQJ

[Ray]

That’s interesting. I didn’t know about it. Will check it shortly. Thanks.

[mcg1103]

I also had all kinds of trouble maintaining a connection with the USB. I tried two USB wifi devices. The recommended edimax nano and one that I purchased on Ebay with a larger antenna. The one with the larger antenna had a problem with the power plug getting in the way but was able to get it plugged in. Neither worked reliably. I tried two beaglebone black boards (I use them at work so I had several at my disposal for testing). I could get it to connect once in a while but would not hold up a connection. I did not do the above mentioned fix…. found this after I found the solution below.

Finally I purchased a tp-link TL-WR702N. Running this mini router in client mode I was able to and plugging into the rj-45 ethernet port on the bbb I have had absolutely not trouble. It has worked after every reboot and has not stopped once. This device is small enough to fit in the recommended enclosure with a OSBo, one 16 station expansion board and the plug in style transformer. It is a tight fit and requires too many extra cables, but it is working.

I am running my OSBo outside, in the weather proof en-closure. I am interested to see how it holds up to the Phoenix summers. It is 99 degrees right now and it is still running 🙂 We will see what happens when the temp gets up to 115.

-Mark

[sbrown]

I also found the same problem with the driver. After sniffing from another pc running wireshark, it appeared that the aggregation sessions between the ap and the usb radio would deadlock. I tried running in non-ht mode and the problem disappeared. Next, I modified the driver to report no hardware aggregation support and the problem disappeared too. The latter allows you to run at ht speeds.

The problem seems to be that the AP starts an aggregation session and the STA starts one also before the AP session terminates. This seems to hang the driver.

I reported the problem to the linux-wireless list.

A patch is below. It’s not a fix, just a workaround until somebody more knowledgable has time to deal with it.

Steve
==========

— a/drivers/net/wireless/rtlwifi/rtl8192cu/sw.c
+++ b/drivers/net/wireless/rtlwifi/rtl8192cu/sw.c
@@ -59,6 +59,9 @@ static int rtl92cu_init_sw_vars(struct ieee80211_hw *hw)
struct rtl_priv *rtlpriv = rtl_priv(hw);
int err;

+ /* disable aggregation until the deadlock is fixed */
+ hw->flags &= ~IEEE80211_HW_AMPDU_AGGREGATION;
+
rtlpriv->dm.dm_initialgain_enable = true;
rtlpriv->dm.dm_flag = 0;
rtlpriv->dm.disable_framebursting = false;

[Ray]

Cool. Thanks for the update. Do you know if the latest Ubuntu for BeagleBone Black has fixed the driver issue? We will be making a new image soon using the latest Ubuntu version, and hopefully the driver issue is fixed in the latest version.

[sbrown]

No reply to my post on linux-wireless yet.
I built a kernel from here: https://github.com/RobertCNelson/bb-kernel.git
to test the work-around.

[Ray]

While preparing a new OSBo SD card image, I came across this website which describes a method to compile the RTL8192 driver:
http://gencarelle.com/blog/2013/07/19/problems-with-rtl8188cus/
I gave it a try and it seems to work much better than the default driver provided by the system. I followed the instructions and compiled it under kernel 3.8.13-bone56. The custom driver is now integrated into the new OSBo pre-configured SD card image, available for download at:
http://raysfiles.com/osbo/osbo2.zip
Will see from user feedback whether this indeed works more reliably or not.

[max_dieu]

Ray, I recently purchased and installed the OSBo, which came with kernel version 3.8.13-bone56 armv71. Last weekend, I ordered the Edimax Nano, to find out that I’m having the same problem that have been reported by this thread. Similarly to another member description, the Edimax nano doesn’t work unless I have the usb or Ethernet plugged in. Based on your most recent post on this subject, I don’t see anyone with feedback of the resolution you have compiled in the latest image. Any suggestion is appreciated.

[Ray]

Does the Edimax Nano work with the latest BBB Debian OS? If so, you can follow the instructions here to install the OpenSprinkler Unified Firmware:
https://opensprinkler.freshdesk.com/solution/categories/5000027039/folders/5000149652/articles/5000631599-installing-and-updating-the-unified-firmware
The pre-configured SD card is based on an old version of kernel so it may not support the latest Edimax dongle.

[Author]

Posts