[arch-general] Last couple of updates Knocked my Wireless -- OUT (atheros/madwifi/ath_pci)
Listmates, I'm not sure where to start troubleshooting this problem. For months and months, ever since I installed Arch, my Atheros wifi (through ath0) has always just worked right out of the box. No biggie. You setup your /etc/network.d/mywpa.conf and add your config to the NETWORKS entry and then each time I would turn on my box, up came the network. That is, until the updates this past week or so. My config is still the same. The mywpa.conf file is: CONNECTION="wireless" INTERFACE=wlan0 SCAN="yes" SECURITY="wpa" ESSID="skyline" KEY="notmyrealkey" IP="dhcp" TIMEOUT=20 My rc.conf setting hasn't changed: eth0="dhcp" INTERFACES=(!eth0) gateway="default gw 192.168.6.13" ROUTES=(!gateway) NETWORKS=(!main mywpa.conf) DAEMONS=(syslog-ng network net-profiles hal @ntpd @sshd @crond @netfs @avahi- daemon @mysqld !postfix !dovecot !httpd !samba @cups kdm3) Now I boot my box and I get nothing but the "lo" interface brought up. What gives? To comply with the new warning about modprobe "blacklist", etc. all of the specified and empty files were moved from /etc/modprobe.d to /root/tmp. My wireless needs the ath_pci module and the ath5k module has always caused problems, so I added the "!ath5k" entry to the modules line before the ath_pci declaration and ath_pci is correctly loaded on boot, but still no wlan0 or ath0 devices are created. As a work-around while this is being sorted, I have been brining the device up with ifconfig ath0 up, then starting wpa_supplicant manually and then finally calling dhcpcd manually to get an IP. What in the world could have changed in the past week or so that is presenting my wireless from being brought up at boot?? Any help or pointers in the right direction would be much appreciated. Thanks. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com
On Mon, Sep 21, 2009 at 04:02, David C. Rankin <drankinatty@suddenlinkmail.com> wrote:
Listmates,
I'm not sure where to start troubleshooting this problem. For months and months, ever since I installed Arch, my Atheros wifi (through ath0) has always just worked right out of the box. No biggie. You setup your /etc/network.d/mywpa.conf and add your config to the NETWORKS entry and then each time I would turn on my box, up came the network.
That is, until the updates this past week or so. My config is still the same. The mywpa.conf file is:
[skipped]
What in the world could have changed in the past week or so that is presenting my wireless from being brought up at boot??
It is always useless when someone says something like "since recent updates" because it does not tell _anything_ about what packages were upgraded. Please post the relevant part of pacman.log otherwise all we can do is just guessing. -- Roman Kyrylych (Роман Кирилич)
On Monday 21 September 2009 04:59:45 am Roman Kyrylych wrote:
That is, until the updates this past week or so. My config is still the same. The mywpa.conf file is:
[skipped]
What in the world could have changed in the past week or so that is presenting my wireless from being brought up at boot??
It is always useless when someone says something like "since recent updates" because it does not tell anything about what packages were upgraded. Please post the relevant part of pacman.log otherwise all we can do is just guessing.
Thomas, Roman, From the pacman.log, it looks like this is where the problem started: [2009-09-17 22:39] upgraded madwifi-utils (0.9.4.3952-1 -> 0.9.4.4096-1) [2009-09-17 22:39] upgraded madwifi (0.9.4.3952-5 -> 0.9.4.4096-1) Prior to that, the network.d config I had posted: CONNECTION="wireless" INTERFACE=wlan0 SCAN="yes" SECURITY="wpa" ESSID="skyline" KEY="bravelightning" IP="dhcp" TIMEOUT=20 had been working great since I first installed 2009.02. Yes, I know is seems bizarre and strange, but the madwifi driver accomodated it somehow. With the update from madwifi 0.9.4.3952-5 -> 0.9.4.4096-1, the ability to make use of the wlan0 device went away. To get it working again i changed the network.d config to: CONNECTION="wireless" INTERFACE=ath0 SCAN="yes" SECURITY="wpa" ESSID="skyline" KEY="bravelightning" IP="dhcp" TIMEOUT=20 The network came right up on the next boot, but all isn't good. The network performance is about 1/20th of what it normally is. Generally I could get a solid 3-4 Meg/s over wireless when transferring files over the lan. (I'm always limited to 135K due to my ISP from the internet) I am now watching 3 meg files crawl across the wireless at 180-250K instead of 3400K. This also must be a side effect of the new madwifi driver. Looking back through the logs, the netwok had been using the "wlan0" device since day 1 and the wireless card had been working fine with the ath5k module. The only service that ever complained wlan0 interface was ntpd but the newtork was working without issue for everything else. Now it is slow... I have put the dmesg output for yesterday (a non-working log) at: http://www.3111skyline.com/download/Archlinux/config/wireless/dmesg-20090921... After changing the network.d config and having the wireless start on boot, I captured another dmesg for comparison: http://www.3111skyline.com/download/Archlinux/config/wireless/dmesg-20090922... WORKING.txt I also have the pacman.log (compressed) at: http://www.3111skyline.com/download/Archlinux/config/wireless/pacman.log.bz2 Whatever changes were made to the madwifi 0.9.4.4096-1 driver, it has killed my wireless performance. I have tested both with the ath_pci module (!ath5k) and with the ath5k module (by removing the !ath5k entry from the MODULES line in rc.conf) The ath5k driver then loads by default. However, performance is still really bad compared to where it was before the Sept. 17 madwifi upgrade: 01:48 alchemy:~/archlinux/config/wireless> lsmod | grep ath ath5k 138952 0 mac80211 179120 1 ath5k ath_rate_sample 15488 1 ath 9984 1 ath5k cfg80211 104376 3 ath5k,mac80211,ath led_class 5160 2 ath5k,sdhci ath_pci 250336 0 wlan 259744 6 wlan_tkip,wlan_ccmp,wlan_scan_sta,ath_rate_sample,ath_pci ath_hal 415520 3 ath_rate_sample,ath_pci I either have to make more changes to my configuration to accomodate the madwifi changes, or, there is a bug in the new driver. This is where my troubleshooting skills with wireless run out. I don't know where to begin to unwravel the driver slowness. Can you think of anything else I can make available that would help? Just let me know and I'll be happy to get it for you. Also, will try to install the prior version of the madwifi driver (Ver. 0.9.4.3952-5) and see if that helps the speed issue. Thank for your help and let me know if I can post anything else you think might help. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com
2009/9/22 David C. Rankin <drankinatty@suddenlinkmail.com>
The network came right up on the next boot, but all isn't good. The network performance is about 1/20th of what it normally is.
I had the same experience with the ath9k-driver since kernel 2.6.29 (ath9k-driver in kernel .28 worked perfectly and in .29 ndiswrapper helped me out). Several other people have asked questions about this (ath5k and ath9k) in the forum. Shortly after kernel .30 came out, my laptop died on me and my new laptop has another chipset, so I couldn't test kernel .31. I never tested the madwifi-drivers. My symptoms were: very weak signal strength, breaking connections and connection speeds that made me cry for ndiswrapper... I never got around to finding the culprit, so filing a bugreport was impossible. Vincent
On Tuesday 22 September 2009 02:56:18 am Vincent Van Houtte wrote:
The network came right up on the next boot, but all isn't good. The network performance is about 1/20th of what it normally is.
I had the same experience with the ath9k-driver since kernel 2.6.29 (ath9k-driver in kernel .28 worked perfectly and in .29 ndiswrapper helped me out). Several other people have asked questions about this (ath5k and ath9k) in the forum. Shortly after kernel .30 came out, my laptop died on me and my new laptop has another chipset, so I couldn't test kernel .31. I never tested the madwifi-drivers.
My symptoms were: very weak signal strength, breaking connections and connection speeds that made me cry for ndiswrapper... I never got around to finding the culprit, so filing a bugreport was impossible.
Vincent
Vincent, Thank you for the information. At least I know I'm not going crazy... It looks like I will have to downgrade the kernel to test the older madwifi package: 08:30 alchemy:~/archlinux/config/wireless/madwifi-good> sudo pacman -U madwifi-* loading package data... checking dependencies... error: failed to prepare transaction (could not satisfy dependencies) :: madwifi: requires kernel26<2.6.31 There is a huge problem with this kernel/madwifi combination. To test the speed, I pulled a 25M test file from the server which usually takes well less than 10 seconds. Now, with the new kernel/madwifi driver: 08:25 alchemy:~/archlinux/config/wireless> rsync -uav --progress archangel:~/dsj.wav . receiving incremental file list dsj.wav 23578844 100% 201.77kB/s 0:01:54 (xfer#1, to-check=0/1) sent 30 bytes received 23581816 bytes 204171.83 bytes/sec total size is 23578844 speedup is 1.00 Now it takes OVER 114 seconds. This is a 10-20 fold increase in time. I used to regularly transfer 500-800M files across the wireless connection. Now, that is out of the question. I admit, I'm not very good at troubleshooting wireless driver issues. How would I start trying to figure out what is causing the poor wireless performance. I do have a spare laptop drive with SuSE on it that I can pop in to compare things if that would help. I just need to know what I would need to start looking at. Which files/logs may hold the clue? Thanks. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com
Vincent,
Thank you for the information. At least I know I'm not going crazy... It looks like I will have to downgrade the kernel to test the older madwifi package: Could you try to recompile the older madwifi from svn using the latest kernel and see if that works? Also testing previous kernel with latest madwifi would be good. I'm fairly concerned about this, as I also use
On Tue, Sep 22, 2009 at 09:43, David C. Rankin <drankinatty@suddenlinkmail.com> wrote: the madwifi drivers, and my speed is abysmal as it is... A reduction like that would put me at dial-up speeds.
On Tuesday 22 September 2009 08:49:51 am Daenyth Blank wrote:
On Tue, Sep 22, 2009 at 09:43, David C. Rankin
<drankinatty@suddenlinkmail.com> wrote:
Vincent,
Thank you for the information. At least I know I'm not going crazy... It looks like I will have to downgrade the kernel to test the older madwifi package:
Could you try to recompile the older madwifi from svn using the latest kernel and see if that works? Also testing previous kernel with latest madwifi would be good. I'm fairly concerned about this, as I also use the madwifi drivers, and my speed is abysmal as it is... A reduction like that would put me at dial-up speeds.
Danny, I downgraded the kernel and madwifi, by downgrading the following: kernel-headers-2.6.30.5-1-x86_64.pkg.tar.gz kernel26-2.6.30.5-1-x86_64.pkg.tar.gz kernel26-firmware-2.6.30-1-x86_64.pkg.tar.gz madwifi-0.9.4.3952-5-x86_64.pkg.tar.gz madwifi-utils-0.9.4.3952-1-x86_64.pkg.tar.gz The wireless then started as usual under wlan0/ath0 and performance is back to FANTASTIC: alchemy:~> rsync -uav --progress archangel:~/dsj.wav . receiving incremental file list dsj.wav 23578844 100% 3.10MB/s 0:00:07 (xfer#1, to-check=0/1) sent 30 bytes received 23581816 bytes 3144246.13 bytes/sec total size is 23578844 speedup is 1.00 compared to the performance with the new kernel/driver: alchemy:~> rsync -uav --progress archangel:~/dsj.wav . receiving incremental file list dsj.wav 23578844 100% 201.77kB/s 0:01:54 (xfer#1, to-check=0/1) That's 3.10MB/s with the old kernel/madwifi combination and only 201.7kB/s with the new. The working setup under the old kernel was accomplished by: ifconfig wlan0 up iwconfig wlan0 essid "skyline" wpa_supplicant -iwlan0 -Dwext -c/etc/wpa_supplicant.conf -d -B dhcpcd wlan0 Done! [09:50 alchemy:/home/david/archlinux/config/wireless] # ifconfig lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:20 errors:0 dropped:0 overruns:0 frame:0 TX packets:20 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:1000 (1000.0 b) TX bytes:1000 (1000.0 b) wlan0 Link encap:Ethernet HWaddr 00:1B:9E:7C:F6:E7 inet addr:192.168.6.102 Bcast:192.168.6.255 Mask:255.255.255.0 inet6 addr: fe80::21b:9eff:fe7c:f6e7/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:17096 errors:0 dropped:0 overruns:0 frame:0 TX packets:8733 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:25090930 (23.9 Mb) TX bytes:784205 (765.8 Kb) wmaster0 Link encap:UNSPEC HWaddr 00-1B-9E-7C-F6- E7-00-00-00-00-00-00-00-00-00-00 UP RUNNING MTU:0 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:0 (0.0 b) TX bytes:0 (0.0 b) [09:50 alchemy:/home/david/archlinux/config/wireless] # iwconfig lo no wireless extensions. eth0 no wireless extensions. wmaster0 no wireless extensions. wlan0 IEEE 802.11bg ESSID:"skyline" Mode:Managed Frequency:2.437 GHz Access Point: 00:23:69:5C:FD:B6 Bit Rate=54 Mb/s Tx-Power=20 dBm Retry min limit:7 RTS thr:off Fragment thr:off Encryption key:***** [3] Security mode:open Power Management:off Link Quality=51/70 Signal level=-59 dBm Noise level=-103 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 Tx excessive retries:0 Invalid misc:0 Missed beacon:0 I know the wmaster0/wlan0 combination was weird, but it seemed to work fine. I'll change the config to wifi0/ath0 under the old kernel, confirm performance under the setup and then I'll reinstall the new kernel and try to build madwifi from svn as you have suggested. Let me know if you have any other thoughts in the mean time. Thanks. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com
On Tuesday 22 September 2009 09:56:49 am David C. Rankin wrote:
I know the wmaster0/wlan0 combination was weird, but it seemed to work fine. I'll change the config to wifi0/ath0 under the old kernel, confirm performance under the setup and then I'll reinstall the new kernel and try to build madwifi from svn as you have suggested. Let me know if you have any other thoughts in the mean time. Thanks.
Danny, all, I have upgraded back to the current kernel (2.6.31-1) and madwifi (0.9.4.4096-1) and I have managed to get wifi working at 100%, but according to Tom, my setup would never work -- it does. But I'm still fighting the problem with my clock getting adjusted backwards by the amount of the tzoffset on each shutdown/reboot (see: New kernel/updates - Is the timezone information messed up for dual-boot configs) [posted earlier today] I don't know if this is the right way to configure the madwifi driver or not, but from all the README's, etc., this is the way it has to work. If you see a better way, please let me know. I'll try to explain the current config in hopes that somebody can see a better way to configure the driver. With madwifi 0.9.4.4096-1, my box can use the ath5k driver just fine. So I have setup rc.conf as follows: MOD_AUTOLOAD="yes" MODULES=(ath5k vboxdrv vboxnetflt) HOSTNAME="alchemy" eth0="dhcp" INTERFACES=(!eth0) gateway="default gw 192.168.6.13" ROUTES=(!gateway) NETWORKS=(!main dcrwpa.conf) When ath5k is loaded, the following wireless extensions are created by default: [13:49 alchemy:/etc] # iwconfig lo no wireless extensions. eth0 no wireless extensions. wmaster0 no wireless extensions. wlan0 IEEE 802.11bg ESSID:"skyline" Mode:Managed Frequency:2.437 GHz Access Point: 00:23:69:5C:FD:B6 Bit Rate=54 Mb/s Tx-Power=20 dBm Retry long limit:7 RTS thr:off Fragment thr:off Encryption key:*** [3] Power Management:off Link Quality=49/70 Signal level=-61 dBm Noise level=-103 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 Tx excessive retries:0 Invalid misc:0 Missed beacon:0 (obviously wlan0 isn't configured when it is first created) Since the ath5k driver defaults to the wmaster0 placeholder for the wlan0 extension in the Arch setup, the network.d script has to reference wlan0 because the network scripts evidently call 'iwlist scan' to find available networks and the only way that will work is if wlan0 is used, anything else results in an error. So the network.d script that I derived from the examples is: [13:51 alchemy:/etc] # cat network.d/dcrwpa.conf CONNECTION="wireless" INTERFACE=wlan0 SCAN="yes" SECURITY="wpa" ESSID="skyline" KEY="myunsecretkey" IP="dhcp" TIMEOUT=20 This works at boot time to activate the wireless network automatically as it should. However it does it under wlan0 instead of ath0. (strange, yes). After the interface is up, the only remaining items to be done are to associate the interface with the access point: iwconfig wlan0 essid "skyline" Then the rest is up to wpa_supplicant and dhcpcd. Now if there is a better way to do this, please let me know. From what I can tell, since ath5k creates the wmaster0/wlan0 extensions by default, I don't know what else you would be able to do. However, the initial problem about the slowness with the connection seems to have been corrected by manually adjusting the time and setting the hwclock after it had been set backwards by 5 hours by something in the shutdown or init process: 13:42 alchemy:~/archlinux/config/wireless> rsync -uav --progress archangel:~/dsj.wav . receiving incremental file list dsj.wav 23578844 100% 2.99MB/s 0:00:07 (xfer#1, to-check=0/1) sent 30 bytes received 23581816 bytes 2774334.82 bytes/sec total size is 23578844 speedup is 1.00 I'm still trying to figure the time issue out. Again, if you can see a better way to do this, then please let me know. Thanks. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com
David C. Rankin wrote:
Listmates,
I'm not sure where to start troubleshooting this problem. For months and months, ever since I installed Arch, my Atheros wifi (through ath0) has always just worked right out of the box. <snip> The mywpa.conf file is:
CONNECTION="wireless" INTERFACE=wlan0 SCAN="yes" SECURITY="wpa" ESSID="skyline" KEY="notmyrealkey" IP="dhcp" TIMEOUT=20
That profile would never have worked with ath0, as it specifies wlan0 as the interface.
Now I boot my box and I get nothing but the "lo" interface brought up. What gives? Check dmesg and/or your logs for clues. What in the world could have changed in the past week or so that is presenting my wireless from being brought up at boot?
As already mentioned - until you tell us what you have done to your system in the past week or so, you might as well be asking how to build one of these: http://en.wikipedia.org/wiki/Perpetual_motion T.
On Monday 21 September 2009 09:25:24 am Tom K wrote:
David C. Rankin wrote:
Listmates,
I'm not sure where to start troubleshooting this problem. For months and months, ever since I installed Arch, my Atheros wifi (through ath0) has always just worked right out of the box.
<snip>
The mywpa.conf file is:
CONNECTION="wireless" INTERFACE=wlan0 SCAN="yes" SECURITY="wpa" ESSID="skyline" KEY="notmyrealkey" IP="dhcp" TIMEOUT=20
That profile would never have worked with ath0, as it specifies wlan0 as the interface.
Now I boot my box and I get nothing but the "lo" interface brought up. What gives?
Check dmesg and/or your logs for clues.
What in the world could have changed in the past week or so that is presenting my wireless from being brought up at boot?
As already mentioned - until you tell us what you have done to your system in the past week or so, you might as well be asking how to build one of these: http://en.wikipedia.org/wiki/Perpetual_motion
T.
Sorry guys, I'll get the detail, But yes this profile DOES work as wlan0 is simply an alias for ath0 under the madwifi driver: [13:48 alchemy:/home/david] # ifconfig ath0 Link encap:Ethernet HWaddr 00:1B:9E:7C:F6:E7 inet addr:192.168.6.102 Bcast:192.168.6.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:378638 errors:0 dropped:0 overruns:0 frame:0 TX packets:133854 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:429332196 (409.4 Mb) TX bytes:14984393 (14.2 Mb) eth0 Link encap:Ethernet HWaddr 00:1B:38:AF:36:B8 UP BROADCAST MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:0 (0.0 b) TX bytes:0 (0.0 b) Interrupt:252 Base address:0xa000 lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:499 errors:0 dropped:0 overruns:0 frame:0 TX packets:499 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:57057 (55.7 Kb) TX bytes:57057 (55.7 Kb) wifi0 Link encap:UNSPEC HWaddr 00-1B-9E-7C-F6- E7-00-00-00-00-00-00-00-00-00-00 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:747577 errors:0 dropped:0 overruns:0 frame:32135 TX packets:155592 errors:2 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:280 RX bytes:303248952 (289.2 Mb) TX bytes:21529160 (20.5 Mb) Interrupt:19 I'll get the pacman logs and post them. But I suspect the problem is related to the other post about wifi problems lately. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com
participants (5)
-
Daenyth Blank
-
David C. Rankin
-
Roman Kyrylych
-
Tom K
-
Vincent Van Houtte