Wireless network stopped working after 22 hours with ath5k, I have the
following message repeating, suspending to ram and back was enough to
get it talking again.
ath5k phy0: gain calibration timeout (2462MHz)
I previously was running 3.2-rc5 for a month and a half straight
without any wireless issues, I just upgraded to 3.3-rc1, the message
started 22 hours after booting the newer kernel. This ath5k is
running hostapd 1:0.7.3-4 with WPA-EAP and had no clients for the 12
hours before the calibration message started, and had no clients for
hours after until I tried to connect and found I couldn't.
While I was unable to connect, I flipped
the laptop rfkill switch and the gain calibration timeouts continued,
disabling rfkill and they still continued. That's when I did a
suspend to memory and resume and found it recovered. While it was
working I tried rfkill and it no longer was talking after disabling
rfkill, I thought it was resuming previously, but I don't use that
very often.
I revered the following two comments and will see if I run into any
more problems. I can test patches.
ath5k: We always do full calibration on AR5210
dafae6af0336958e9e2eb67cc5e4e31d6d13b308
ath5k: Calibration re-work
ce169aca0d823d38465127023e3d571816e6666c
Feb 24 21:54:23 Pulsar kernel: ath5k 0000:06:02.0: registered as 'phy0'
Feb 24 21:54:23 Pulsar kernel: ath: EEPROM regdomain: 0x63
Feb 24 21:54:23 Pulsar kernel: ath: EEPROM indicates we should expect a direct regpair map
Feb 24 21:54:23 Pulsar kernel: ath: Country alpha2 being used: 00
Feb 24 21:54:23 Pulsar kernel: ath: Regpair used: 0x63
Feb 24 21:54:23 Pulsar kernel: ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
Feb 24 21:54:23 Pulsar kernel: ath5k phy0: Atheros AR2413 chip found (MAC: 0x78, PHY: 0x45)
Feb 24 21:54:23 Pulsar kernel: device wlan0 entered promiscuous mode
Feb 24 21:54:23 Pulsar kernel: br0: port 2(wlan0) entering forwarding state
Feb 24 21:54:23 Pulsar kernel: br0: port 2(wlan0) entering forwarding state
Feb 24 21:58:59 Pulsar kernel: ath5k 0000:06:02.0: registered as 'phy0'
Feb 24 21:58:59 Pulsar kernel: ath: EEPROM regdomain: 0x63
Feb 24 21:58:59 Pulsar kernel: ath: EEPROM indicates we should expect a direct regpair map
Feb 24 21:58:59 Pulsar kernel: ath: Country alpha2 being used: 00
Feb 24 21:58:59 Pulsar kernel: ath: Regpair used: 0x63
Feb 24 21:58:59 Pulsar kernel: ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
Feb 24 21:58:59 Pulsar kernel: ath5k phy0: Atheros AR2413 chip found (MAC: 0x78, PHY: 0x45)
Feb 24 21:58:59 Pulsar kernel: device wlan0 entered promiscuous mode
Feb 24 21:58:59 Pulsar kernel: br0: port 2(wlan0) entering forwarding state
Feb 24 21:58:59 Pulsar kernel: br0: port 2(wlan0) entering forwarding state
Feb 24 23:02:31 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 1
Feb 24 23:02:31 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 2
Feb 24 23:02:40 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 3
Feb 24 23:15:37 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 2
Feb 24 23:15:37 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 1
Feb 24 23:15:37 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 0
Feb 25 08:48:40 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 1
Feb 25 08:48:40 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 2
Feb 25 08:48:46 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 3
Feb 25 08:56:35 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 2
Feb 25 08:56:35 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 1
Feb 25 08:56:35 Pulsar kernel: wlan0: moving STA xx:xx:xx:xx:xx:xx to state 0
Feb 25 19:52:50 Pulsar kernel: ath5k phy0: gain calibration timeout (2462MHz)
Feb 25 19:52:51 Pulsar kernel: ath5k phy0: calibration of channel 11 failed
Feb 25 19:52:52 Pulsar kernel: ath5k phy0: gain calibration timeout (2462MHz)
iw wlan0 survey dump
Survey data from wlan0
frequency: 2462 MHz
noise: -96 dBm
channel active time: 79250 ms
channel busy time: 118644 ms
channel receive time: 48278 ms
channel transmit time: 5285 ms
--
David Fries <[email protected]> PGP pub CB1EE8F0
http://fries.net/~david/
On 2012-02-27 4:12 AM, David Fries wrote:
> Wireless network stopped working after 22 hours with ath5k, I have the
> following message repeating, suspending to ram and back was enough to
> get it talking again.
> ath5k phy0: gain calibration timeout (2462MHz)
>
> I previously was running 3.2-rc5 for a month and a half straight
> without any wireless issues, I just upgraded to 3.3-rc1, the message
> started 22 hours after booting the newer kernel. This ath5k is
> running hostapd 1:0.7.3-4 with WPA-EAP and had no clients for the 12
> hours before the calibration message started, and had no clients for
> hours after until I tried to connect and found I couldn't.
> While I was unable to connect, I flipped
> the laptop rfkill switch and the gain calibration timeouts continued,
> disabling rfkill and they still continued. That's when I did a
> suspend to memory and resume and found it recovered. While it was
> working I tried rfkill and it no longer was talking after disabling
> rfkill, I thought it was resuming previously, but I don't use that
> very often.
>
> I revered the following two comments and will see if I run into any
> more problems. I can test patches.
> ath5k: We always do full calibration on AR5210
> dafae6af0336958e9e2eb67cc5e4e31d6d13b308
> ath5k: Calibration re-work
> ce169aca0d823d38465127023e3d571816e6666c
Please try applying the following two patches:
http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/440-ath5k_calibrate_no_queue_stop.patch;hb=HEAD
http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/441-ath5k_no_agc_recalibration.patch;hb=HEAD
- Felix
On Mon, Feb 27, 2012 at 01:07:14PM +0100, Felix Fietkau wrote:
> On 2012-02-27 4:12 AM, David Fries wrote:
> > Wireless network stopped working after 22 hours with ath5k, I have the
> > following message repeating, suspending to ram and back was enough to
> > get it talking again.
> > ath5k phy0: gain calibration timeout (2462MHz)
> >
> > I previously was running 3.2-rc5 for a month and a half straight
> > without any wireless issues, I just upgraded to 3.3-rc1, the message
> > started 22 hours after booting the newer kernel. This ath5k is
> > running hostapd 1:0.7.3-4 with WPA-EAP and had no clients for the 12
> > hours before the calibration message started, and had no clients for
> > hours after until I tried to connect and found I couldn't.
> > While I was unable to connect, I flipped
> > the laptop rfkill switch and the gain calibration timeouts continued,
> > disabling rfkill and they still continued. That's when I did a
> > suspend to memory and resume and found it recovered. While it was
> > working I tried rfkill and it no longer was talking after disabling
> > rfkill, I thought it was resuming previously, but I don't use that
> > very often.
> >
> > I revered the following two comments and will see if I run into any
> > more problems. I can test patches.
> > ath5k: We always do full calibration on AR5210
> > dafae6af0336958e9e2eb67cc5e4e31d6d13b308
> > ath5k: Calibration re-work
> > ce169aca0d823d38465127023e3d571816e6666c
> Please try applying the following two patches:
> http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/440-ath5k_calibrate_no_queue_stop.patch;hb=HEAD
> http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/441-ath5k_no_agc_recalibration.patch;hb=HEAD
Thanks, I'm running with those two patches. I'll give it a few days
before reporting to see if it goes down again.
--
David Fries <[email protected]> PGP pub CB1EE8F0
http://fries.net/~david/
On Mon, Feb 27, 2012 at 08:42:55AM -0600, David Fries wrote:
> On Mon, Feb 27, 2012 at 01:07:14PM +0100, Felix Fietkau wrote:
> > On 2012-02-27 4:12 AM, David Fries wrote:
> > > Wireless network stopped working after 22 hours with ath5k, I have the
> > > following message repeating, suspending to ram and back was enough to
> > > get it talking again.
> > > ath5k phy0: gain calibration timeout (2462MHz)
> > >
> > Please try applying the following two patches:
> > http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/440-ath5k_calibrate_no_queue_stop.patch;hb=HEAD
> > http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/441-ath5k_no_agc_recalibration.patch;hb=HEAD
No problems running with the above two patches after 12 days. No
calibration timeout messages either. Thanks.
--
David Fries <[email protected]> PGP pub CB1EE8F0
http://fries.net/~david/
On 10 March 2012 10:56, David Fries <[email protected]> wrote:
>> > http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/440-ath5k_calibrate_no_queue_stop.patch;hb=HEAD
>> > http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/441-ath5k_no_agc_recalibration.patch;hb=HEAD
>
> No problems running with the above two patches after 12 days. ?No
> calibration timeout messages either. ?Thanks.
Sweet. I did do a bit of digging (and forgot to email the list) - yes,
those initial calibrations should only be done once, after a (full)
chip cold reset. They shouldn't be done periodically.
Adrian
Felix Fietkau <nbd@...> writes:
>
> On 2012-02-27 4:12 AM, David Fries wrote:
> > Wireless network stopped working after 22 hours with ath5k, I have the
> > following message repeating, suspending to ram and back was enough to
> > get it talking again.
> > ath5k phy0: gain calibration timeout (2462MHz)
> >
> > I revered the following two comments and will see if I run into any
> > more problems. I can test patches.
> > ath5k: We always do full calibration on AR5210
> > dafae6af0336958e9e2eb67cc5e4e31d6d13b308
> > ath5k: Calibration re-work
> > ce169aca0d823d38465127023e3d571816e6666c
> Please try applying the following two patches:
>
http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/440-ath5k_calibrate_no_queue_stop.patch;hb=HEAD
>
http://nbd.name/gitweb.cgi?p=openwrt.git;a=blob_plain;f=package/mac80211/patches/441-ath5k_no_agc_recalibration.patch;hb=HEAD
Hello,
archlinux users are affected by this issue after upgrading the linux kernel in
3.3.1 :
https://bbs.archlinux.org/viewtopic.php?id=139270
Do you known if these two patches have been merged in the mainstream branch ?
thank you
hamelg <hamelg@...> writes:
> archlinux users are affected by this issue after upgrading the linux kernel in
> 3.3.1 :
> https://bbs.archlinux.org/viewtopic.php?id=139270
> Do you known if these two patches have been merged in the mainstream branch ?
I have checked : these patches are not in the v3.3.1.
They are merged in 3.4-rc1. Must I wait the 3.4 ?