(worked around, but not fixed) TP-Link USB dongle, kernel 3.15, no DHCP lease

[dibl]

On this old box:

Code Select Expand

System:    Host: decembersid Kernel: 3.14-5.towo.2-siduction-amd64 x86_64 (64 bit gcc: 4.8.3)
           Desktop: KDE 4.13.2 (Qt 4.8.6) Distro: siduction 13.2.1 December - kde - (201401272125)
Machine:   Mobo: N/A model: N/A Bios: Intel v: CF94510J.15A.0028.2006.1213.1629 date: 12/13/2006
CPU:       Dual core Intel Pentium D (-MCP-) cache: 1024 KB flags: (lm nx sse sse2 sse3) bmips: 11182
           Clock Speeds: 1: 2794 MHz 2: 2794 MHz
Graphics:  Card: Intel 82945G/GZ Integrated Graphics Controller bus-ID: 00:02.0
           Display Server: X.Org 1.15.1 drivers: intel (unloaded: fbdev,vesa) Resolution: 1440x900@60.1hz
           GLX Renderer: Mesa DRI Intel 945G GLX Version: 2.1 Mesa 10.1.4 Direct Rendering: Yes
Network:   Card-1: Intel PRO/100 VE Network Connection
           driver: e100 v: 3.5.24-k2-NAPI port: 1000 bus-ID: 06:08.0
           IF: eth1 state: down mac: 00:19:d1:32:d7:5c
           Card-2: Atheros TP-Link TL-WN821N v3 802.11n [Atheros AR7010+AR9287]
           driver: ath9k_htc usb-ID: 005-002
           IF: wlan0 state: up mac: 90:f6:52:0a:c0:16
Drives:    HDD Total Size: 250.1GB (4.0% used) ID-1: model: Hitachi_HDT72502
Info:      Processes: 181 Uptime: 3 min Memory: 431.3/1998.4MB Init: systemd runlevel: 5 Gcc sys: 4.9.0
           Client: Shell (bash 4.3.111) inxi: 2.1.28


I use the TP-Link USB dongle for network connectivity.  It has worked flawlessly on several siduction systems over the past two years.  After a dist-upgrade installed the first released 3.15 kernel, I shut it down and did not realize until today that it no longer connects after login.  At first I suspected the access point, but a netbook with a different wifi chip has no problem connecting.  Then I moved the antenna around to make sure a strong signal was available, and I moved the USB dongle to a different USB port.  I investigated first with wicd, and then with ceni, and everything works correctly through the WPA 1/2 authentication, and the process ends "awaiting IP address" -- there is no DHCP lease given/accepted.  Finally I booted a 3.14 kernel and it instantly connected -- no problem.  I pasted the dmesg output from 3.14 and also 3.15, although I don't see a difference until it fails to get the DHCP lease.  I also pasted in the output from

Code Select Expand

lsmod | grep ath from 3.14, and I compared it to the same output from 3.15 and the modules are identical in every way.  So, I am mystified by this.  If I can do any further troubleshooting to help diagnose the issue, I will be happy to do so.

[sunrat]

I use TP-Link TL-WN822N which I think is he same chip but with extra antennae and it has been very reliable. I might hold back on kernel 3.15 until you report this as resolved (no pun intended ).

[dibl]

You should go ahead and install a 3.15er and boot it.  That will help determine whether it is a general problem with 3.15, or whether there is some other hardware issue on my box.

[sunrat]

I just did dist-upgrade which brought in 3.15-0 and confirm I have exactly the same problem. No address issued from the router. I tried reconfigure wlan0 with Ceni to no avail. Reboot to 3.14-5 and it works perfectly.

Code Select Expand

$ inxi -b
System:    Host: brain Kernel: 3.14-5.towo-siduction-amd64 x86_64 (64 bit) Desktop: KDE 4.13.2
           Distro: siduction 13.2.0 December - kde - (201312310241)
Machine:   Mobo: Gigabyte model: EP45-UD3R v: x.x Bios: Award v: F3 date: 09/18/2008
CPU:       Dual core Intel Core2 Duo E8500 (-MCP-) clocked at 2000 MHz
Graphics:  Card: NVIDIA GF114 [GeForce GTX 560 Ti]
           Display Server: X.Org 1.15.1 drivers: nouveau (unloaded: fbdev,vesa) Resolution: 1920x1080@59.9hz
           GLX Renderer: Gallium 0.4 on NVCE GLX Version: 3.0 Mesa 10.1.4
Network:   Card: Atheros TP-Link TL-WN821N v3 802.11n [Atheros AR7010+AR9287] driver: ath9k_htc
Drives:    HDD Total Size: 1435.3GB (82.8% used)
Info:      Processes: 167 Uptime: 8 min Memory: 683.4/3958.4MB Client: Shell (bash) inxi: 2.1.28

Should not this topic be in Upgrade Warnings rather than release Bugs?

[DeepDayze]

Perhaps kernel 3.15 broke something in the ath9k driver? Any way to debug it?

My own current setup is a Lenovo SL510 that uses iwlwifi instead of athxx driver, and iwlwifi works fine with 3.15 so this has to be a driver problem with ath9k.

[dibl]

Any way to debug it?

Debugging a driver malfunction is basically above my skill level, but I can certainly follow instructions* from someone who knows what needs to be done.


*Detailed instructions for one with no engineering degree.  

[dibl]

If you read down this lkml message you can find these:

Rajkumar Manoharan (1):
      ath9k: fix race in setting ATH_OP_INVALID

Tim Harvey (2):
      ath9k: fix possible hang on flush
      ath9k: add a recv budget

I'm thinking one of these broke it.

[der_bud]

Perhaps kernel 3.15 broke something in the ath9k driver?

FWIW, no problem here with ath9k_hw for builtin pci atheros and kernel 3.15-0.towo.2-siduction-amd64, so not every ath affected.

Code Select Expand

Network:   Card: Qualcomm Atheros AR5418 Wireless Network Adapter [AR5008E 802.11(a)bgn] (PCI-Express)
           driver: ath9k bus-ID: 03:00.0
           IF: wlan0 state: up mac: xxx

[ayla]

Same here on my box (I'm just testing it).

Atheros TP-Link TL-WN821N v3 802.11n [Atheros AR7010+AR9287]

The stick worked in the past out of the box.

For what it may be worth here the relevant part of journalctl:

Code Select Expand

Jun 15 23:51:52 neskaya wpa_supplicant[1276]: ^[[1;39mwlan1: Trying to associate with c0:25:06:4f:68:d1 (SSID='darkover' freq=2462 MHz)
Jun 15 23:51:52 neskaya kernel: wlan1: authenticated
Jun 15 23:51:52 neskaya kernel: wlan1: associate with c0:25:06:4f:68:d1 (try 1/3)
Jun 15 23:51:52 neskaya NetworkManager[1078]: <info> (wlan1): supplicant interface state: authenticating -> associating
Jun 15 23:51:52 neskaya kernel: wlan1: RX AssocResp from c0:25:06:4f:68:d1 (capab=0x431 status=0 aid=1)
Jun 15 23:51:52 neskaya wpa_supplicant[1276]: ^[[1;39mwlan1: Associated with c0:25:06:4f:68:d1
Jun 15 23:51:52 neskaya kernel: wlan1: associated
Jun 15 23:51:52 neskaya kernel: cfg80211: Calling CRDA for country: DE
Jun 15 23:51:52 neskaya kernel: cfg80211: Regulatory domain changed to country: DE
Jun 15 23:51:52 neskaya kernel: cfg80211:  DFS Master region: unset
Jun 15 23:51:52 neskaya kernel: cfg80211:   (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp), (dfs_cac_time)
Jun 15 23:51:52 neskaya kernel: cfg80211:   (2400000 KHz - 2483500 KHz @ 40000 KHz), (N/A, 2000 mBm), (N/A)
Jun 15 23:51:52 neskaya kernel: cfg80211:   (5150000 KHz - 5350000 KHz @ 80000 KHz), (N/A, 2000 mBm), (N/A)
Jun 15 23:51:52 neskaya kernel: cfg80211:   (5470000 KHz - 5725000 KHz @ 80000 KHz), (N/A, 2698 mBm), (0 s)
Jun 15 23:51:52 neskaya kernel: cfg80211:   (57240000 KHz - 65880000 KHz @ 2160000 KHz), (N/A, 4000 mBm), (N/A)
Jun 15 23:51:52 neskaya NetworkManager[1078]: <info> (wlan1): supplicant interface state: associating -> 4-way handshake
Jun 15 23:51:52 neskaya wpa_supplicant[1276]: ^[[1;39mwlan1: WPA: Key negotiation completed with c0:25:06:4f:68:d1 [PTK=CCMP GTK=TKIP]
Jun 15 23:51:52 neskaya wpa_supplicant[1276]: ^[[1;39mwlan1: CTRL-EVENT-CONNECTED - Connection to c0:25:06:4f:68:d1 completed (reauth) [id=0 id_str=]
Jun 15 23:51:52 neskaya NetworkManager[1078]: <info> (wlan1): supplicant interface state: 4-way handshake -> completed
Jun 15 23:52:05 neskaya dhclient[3383]: DHCPDISCOVER on wlan1 to 255.255.255.255 port 67 interval 16
Jun 15 23:52:12 neskaya NetworkManager[1078]: ^[[1;39m<warn> (wlan1): DHCPv4 request timed out.
Jun 15 23:52:12 neskaya dhclient[3383]: Received signal 15, initiating shutdown.
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> (wlan1): canceled DHCP transaction, DHCP client pid 3383
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> Activation (wlan1) Stage 4 of 5 (IPv4 Configure Timeout) scheduled...
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> Activation (wlan1) Stage 4 of 5 (IPv4 Configure Timeout) started...
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> (wlan1): device state change: ip-config -> failed (reason 'ip-config-unavailable') [70 120 5]
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> NetworkManager state is now DISCONNECTED
Jun 15 23:52:12 neskaya NetworkManager[1078]: ^[[1;39m<warn> Activation (wlan1) failed for connection 'darkover'
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> Activation (wlan1) Stage 4 of 5 (IPv4 Configure Timeout) complete.
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> (wlan1): device state change: failed -> disconnected (reason 'none') [120 30 0]
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> (wlan1): deactivating device (reason 'none') [0]
Jun 15 23:52:12 neskaya kernel: wlan1: deauthenticating from c0:25:06:4f:68:d1 by local choice (Reason: 3=DEAUTH_LEAVING)
Jun 15 23:52:12 neskaya NetworkManager[1078]: <info> (wlan1): supplicant interface state: completed -> disconnected



Same box with my normally used card:

Qualcomm Atheros AR5418 Wireless Network Adapter [AR5008E 802.11(a)bgn] (PCI-Express)
           driver: ath9k

No problems, works as expected.

greets
ayla

[sunrat]

Debugging a driver malfunction is basically above my skill level, but I can certainly follow instructions* from someone who knows what needs to be done.
*Detailed instructions for one with no engineering degree.

Ditto. So it appears to be a specific problem with ath9k_htc.

Snips -

Code Select Expand

$ journalctl --since=2014-6-15 |grep -i network 
Jun 16 01:02:13 brain systemd[1]: Stopping Host and Network Name Lookups.
Jun 16 01:02:13 brain systemd[1]: Stopped target Host and Network Name Lookups.
Jun 16 01:02:13 brain systemd[1]: Stopped LSB: lightweight network traffic monitor.
Jun 16 01:02:13 brain dhclient[6227]: send_packet: Network is unreachable
Jun 16 01:02:14 brain dhclient[1811]: send_packet: Network is unreachable
Jun 16 01:02:14 brain ifdown[1697]: send_packet: Network is unreachable
Jun 16 01:02:14 brain systemd[1]: Stopping Network.
Jun 16 01:02:14 brain systemd[1]: Stopped target Network.
Jun 16 01:02:14 brain systemd[1]: Stopping LSB: Raise network interfaces....
Jun 16 01:02:14 brain networking[1962]: Deconfiguring network interfaces...done.
Jun 16 01:02:14 brain systemd[1]: Stopped LSB: Raise network interfaces..

Firmware appears to load OK and appears in lsusb -

Code Select Expand

$ journalctl --since=2014-6-15 |grep -i ath9k
Jun 16 01:02:38 brain kernel: usb 8-5: ath9k_htc: Firmware htc_7010.fw requested
Jun 16 01:02:38 brain kernel: usbcore: registered new interface driver ath9k_htc
Jun 16 01:02:38 brain kernel: usb 8-5: ath9k_htc: Transferred FW: htc_7010.fw, size: 72992
Jun 16 01:02:38 brain kernel: ath9k_htc 8-5:1.0: ath9k_htc: HTC initialized with 45 credits
Jun 16 01:02:39 brain kernel: ath9k_htc 8-5:1.0: ath9k_htc: FW Version: 1.3

Stand by while I reboot in to 3.15 again. Maybe dhclient is running before firmware loads.......
Result inconclusive -

Code Select Expand

roger@brain:~$ journalctl -b |grep -i dhclient
Jun 16 10:12:33 brain dhclient[1224]: Listening on LPF/wlan0/f8:d1:11:14:03:eb
Jun 16 10:12:33 brain dhclient[1224]: Sending on   LPF/wlan0/f8:d1:11:14:03:eb
Jun 16 10:12:33 brain dhclient[1224]: Sending on   Socket/fallback
Jun 16 10:12:33 brain dhclient[1224]: DHCPDISCOVER on wlan0 to 255.255.255.255 port 67 interval 3
---snip---
Jun 16 10:13:31 brain dhclient[1224]: DHCPDISCOVER on wlan0 to 255.255.255.255 port 67 interval 3
Jun 16 10:13:34 brain dhclient[1224]: No DHCPOFFERS received.
Jun 16 10:13:34 brain dhclient[1224]: No working leases in persistent database - sleeping.
roger@brain:~$ journalctl -b |grep -i ath9k_htc
Jun 16 10:12:31 brain kernel: usb 8-5: ath9k_htc: Firmware htc_7010.fw requested
Jun 16 10:12:31 brain kernel: usbcore: registered new interface driver ath9k_htc
Jun 16 10:12:31 brain kernel: usb 8-5: ath9k_htc: Transferred FW: htc_7010.fw, size: 72992
Jun 16 10:12:33 brain kernel: ath9k_htc 8-5:1.0: ath9k_htc: HTC initialized with 45 credits
Jun 16 10:12:33 brain kernel: ath9k_htc 8-5:1.0: ath9k_htc: FW Version: 1.3

[dibl]

Right -- this chip has no problem:

Code Select Expand

vCPU:       Single core Intel Atom N280 (-HT-) cache: 512 KB
           flags: (nx pae sse sse2 sse3 ssse3) bmips: 3326
           Clock Speeds: 1: 1333 MHz 2: 1333 MHz
Graphics:  Card: Intel Mobile 945GSE Express Integrated Graphics Controller
           bus-ID: 00:02.0
           Display Server: X.Org 1.15.1 drivers: intel (unloaded: fbdev,vesa)
           Resolution: 1024x600@60.0hz
           GLX Renderer: Mesa DRI Intel 945GME x86/MMX/SSE2
           GLX Version: 2.1 Mesa 10.1.4 Direct Rendering: Yes
Network:   Card-1: Realtek RTL8101E/RTL8102E PCI Express Fast Ethernet controller
           driver: r8169 v: 2.3LK-NAPI port: 2000 bus-ID: 04:00.0
           IF: eth0 state: down mac: 00:23:5a:fa:a0:32
           Card-2: Qualcomm Atheros AR9285 Wireless Network Adapter (PCI-Express)
           driver: ath9k bus-ID: 03:00.0
           IF: wlan0 state: up mac: 00:23:08:96:77:a1
Drives:    HDD Total Size: 40.0GB (16.7% used)
           ID-1: /dev/sda model: OCZ size: 40.0GB
Partition: ID-1: / size: 5.4G used: 4.5G (88%) fs: ext4 dev: /dev/sda1
           ID-2: swap-1 size: 1.63GB used: 0.00GB (0%) fs: swap dev: /dev/sda2
Info:      Processes: 158 Uptime: 3 min Memory: 380.0/2016.2MB
           Init: systemd runlevel: 5 Gcc sys: 4.9.0
           Client: Shell (bash 4.3.111) inxi: 2.1.28

[lamiska]

Hey guys, I have same problem on my Arch Linux after I upgraded to 3.15

NetGear WNA1100 Wireless-N 150 [Atheros AR9271] driver: ath9k_htc

Downgrading to 3.10 kernel solved the issue for me

[sunrat]

I'm a novice at reporting bugs as someone else has always previously reported any that I have found in the past. I couldn't find any filed for this one so attempted to use reportbug-ng but it defaults to bugs@debian so I didn't want to report a kernel bug to Debian when I am using siduction kernel. It may be generic to Linux 3.15 kernel but bugzilla.kernel.org only wants to know about mainline kernel bugs.
What could I do to help fix this issue?

[sunrat]

Just d-u today which brought in 3.15-4 kernel. Still no DHCP lease for ath9k_htc device.

[spro]

Just found this thread after battling my TP-Link USB card for the last few hours... I was setting up a fresh install of Arch and the network didn't work after installing. Turns out the installation ISO booted into 3.14, and the card worked, but I had 3.15 on the drive. Downgrading the kernel solved the problem.


So it's definitely more of a kernel-in-general bug, if that's interesting to know.