Bug 43184

Summary: [Intel 6300] WLAN 802.11n connections stall regularly
Product: Networking Reporter: unggnu
Component: WirelessAssignee: networking_wireless (networking_wireless)
Status: CLOSED DUPLICATE    
Severity: normal CC: braket, emmanuel.grumbach, garkein, ilw, johannes, kernel, linville, stf_xl, timshel, wey-yi.w.guy
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 3.7.1 Subsystem:
Regression: Yes Bisected commit-id:
Attachments: dmesg output showing connection stall

Description unggnu 2012-05-01 07:49:55 UTC
Intel 6300 WLAN had huge issues in the past on Linux with 802.11n, like real connection drops and not supporting the higher frequencies of 802.11a but this seems to be fixed mostly.

Nevertheless there is still an annoying issue that even might be a general Linux WLAN stack issue.

Even though connection quality is shown as high I have regular connection stalls.
The possibility of this issue seems higher the further I am away from the AP and the more (small) packets are sent. 

During this stall I am still connected according to the Networkmanager but not packages could be transferred. Normally the connection stays this way until I disable WLAN and enable it again.
Then the wireless connection is established without problems in seconds and everything works fine for a short period of time. Than I have the same issue again.
So it can't be a problem of distance to the AP since the connection is always established directly and without problems and shortly afterwards everything works fine.

Dmesg doesn't show any errors but it seems that the connection gets out of sync or something at least that would explain why it happens more often if the wireless client is further away.

This issue has happened with several routers but only with 802.11n. This doesn't happen with 802.11g.

For example I had the same issue with a different 802.11n router from Buffalo in another place.
The interesting part is that this even happened there with a Samsung Galaxy S2. You have to disable the WLAN and enable it again on the Android Smartphone to be able to send packets. The Smartphone doesn't have this issue if I use 802.11g.

So it might be a regular issue of the Wireless stack with 802.11n networks at least with some routers?

I have an AVM 7390 router and using the 802.11a band on channel 100 but this also happens in the 2.4 GHz band.
I am using openSUSE Tumbleweed with the Kernel 3.3.3 but this happens with other distributions and older 3.x Kernel too but I don't think that his happened with 2.6.37 at least when I wasn't using 40 MHz (but 2.6.37 wasn't able to connect to higher 5GHz frequencies).

The easiest way to trigger this for me is to be a little further away from the ap and then starting a program which uses DHT like for example Ktorrent (DHT isn't enabled by default).
Shortly after this the connection normally stalls at least ones. But even without many small packets the connections stalls always during regular surfing if further away (The distance is still fine since establishing the connection and using it shortly afterwards is no problem).

I would really like to help debug this but I don't know how so please let me know if you need anything else.

02:00.0 Network controller: Intel Corporation Centrino Ultimate-N 6300 (rev 35)

dmesg extract (the MAC was changed by me):
[   18.944299] iwlwifi 0000:02:00.0: L1 Enabled; Disabling L0S
[   18.944492] iwlwifi 0000:02:00.0: Radio type=0x0-0x3-0x1
[   19.185541] iwlwifi 0000:02:00.0: L1 Enabled; Disabling L0S
[   19.185781] iwlwifi 0000:02:00.0: Radio type=0x0-0x3-0x1
[   19.305181] ADDRCONF(NETDEV_UP): wlan0: link is not ready
[   19.319743] NET: Registered protocol family 17
[   26.962662] wlan0: authenticate with c0:25:06:00:00:00 (try 1)
[   27.162231] wlan0: authenticate with c0:25:06:00:00:00 (try 2)
[   27.164713] wlan0: authenticated
[   27.243622] wlan0: associate with c0:25:06:00:00:00 (try 1)
[   27.443028] wlan0: associate with c0:25:06:00:00:00 (try 2)
[   27.444773] wlan0: RX AssocResp from c0:25:06:00:00:00 (capab=0x411 status=0 aid=1)
[   27.444776] wlan0: associated
[   27.444780] wlan0: moving STA c0:25:06:00:00:00 to state 1
[   27.444782] wlan0: moving STA c0:25:06:00:00:00 to state 2
[   27.444785] wlan0: No basic rates in AssocResp. Using min supported rate instead.
[   27.449815] ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
[   27.449895] cfg80211: Calling CRDA for country: DE
[   27.456541] cfg80211: Regulatory domain changed to country: DE
[   27.456545] cfg80211:   (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[   27.456548] cfg80211:   (2400000 KHz - 2483500 KHz @ 40000 KHz), (N/A, 2000 mBm)
[   27.456550] cfg80211:   (5150000 KHz - 5350000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[   27.456552] cfg80211:   (5470000 KHz - 5725000 KHz @ 40000 KHz), (N/A, 2698 mBm)
[   27.529464] wlan0: moving STA c0:25:06:00:00:00 to state 3
[   37.669340] wlan0: no IPv6 routers present
[  225.785379] wlan0: moving STA c0:25:06:00:00:00 to state 2
[  225.785385] wlan0: moving STA c0:25:06:00:00:00 to state 1
[  225.785389] wlan0: moving STA c0:25:06:00:00:00 to state 0
[  225.786161] cfg80211: Calling CRDA to update world regulatory domain
[  225.789911] cfg80211: World regulatory domain updated:
[  225.789915] cfg80211:   (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[  225.789919] cfg80211:   (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[  225.789923] cfg80211:   (2457000 KHz - 2482000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[  225.789926] cfg80211:   (2474000 KHz - 2494000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[  225.789930] cfg80211:   (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[  225.789933] cfg80211:   (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[  229.058862] wlan0: authenticate with c0:25:06:00:00:00 (try 1)
[  229.060169] wlan0: authenticated
[  229.060598] wlan0: associate with c0:25:06:00:00:00 (try 1)
[  229.061985] wlan0: RX ReassocResp from c0:25:06:00:00:00 (capab=0x411 status=0 aid=1)
[  229.061993] wlan0: associated
[  229.061999] wlan0: moving STA c0:25:06:00:00:00 to state 1
[  229.062004] wlan0: moving STA c0:25:06:00:00:00 to state 2
[  229.062009] wlan0: No basic rates in AssocResp. Using min supported rate instead.
[  229.067237] cfg80211: Calling CRDA for country: DE
[  229.070326] cfg80211: Regulatory domain changed to country: DE
[  229.070330] cfg80211:   (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[  229.070332] cfg80211:   (2400000 KHz - 2483500 KHz @ 40000 KHz), (N/A, 2000 mBm)
[  229.070334] cfg80211:   (5150000 KHz - 5350000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[  229.070336] cfg80211:   (5470000 KHz - 5725000 KHz @ 40000 KHz), (N/A, 2698 mBm)
[  230.073845] wlan0: moving STA c0:25:06:00:00:00 to state 3
[ 2119.782124] iwlwifi 0000:02:00.0: Queue 2 stuck for 2000 ms.
[ 2119.782131] iwlwifi 0000:02:00.0: Current SW read_ptr 116 write_ptr 121
[ 2119.782185] iwlwifi 0000:02:00.0: Current HW read_ptr 116 write_ptr 121
[ 2119.782189] iwlwifi 0000:02:00.0: On demand firmware reload
[ 2119.782685] ieee80211 phy0: Hardware restart was requested
[ 2119.782788] iwlwifi 0000:02:00.0: L1 Enabled; Disabling L0S
[ 2119.782982] iwlwifi 0000:02:00.0: Radio type=0x0-0x3-0x1
[ 2503.561649] wlan0: deauthenticating from c0:25:06:00:00:00 by local choice (reason=3)
[ 2503.573228] wlan0: moving STA c0:25:06:00:00:00 to state 2
[ 2503.573233] wlan0: moving STA c0:25:06:00:00:00 to state 1
[ 2503.573236] wlan0: moving STA c0:25:06:00:00:00 to state 0
[ 2503.575086] cfg80211: Calling CRDA to update world regulatory domain
[ 2503.579256] cfg80211: World regulatory domain updated:
[ 2503.579258] cfg80211:   (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[ 2503.579261] cfg80211:   (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[ 2503.579263] cfg80211:   (2457000 KHz - 2482000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[ 2503.579264] cfg80211:   (2474000 KHz - 2494000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[ 2503.579266] cfg80211:   (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[ 2503.579268] cfg80211:   (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[ 2503.582516] sched: RT throttling activated
[ 2504.294889] iwlwifi 0000:02:00.0: L1 Enabled; Disabling L0S
[ 2504.295078] iwlwifi 0000:02:00.0: Radio type=0x0-0x3-0x1
[ 2504.436218] ADDRCONF(NETDEV_UP): wlan0: link is not ready
[ 2512.026486] wlan0: authenticate with c0:25:06:00:00:00 (try 1)
[ 2512.027591] wlan0: authenticated
[ 2512.028574] wlan0: associate with c0:25:06:00:00:00 (try 1)
[ 2512.029911] wlan0: RX AssocResp from c0:25:06:00:00:00 (capab=0x411 status=0 aid=1)
[ 2512.029914] wlan0: associated
[ 2512.029918] wlan0: moving STA c0:25:06:00:00:00 to state 1
[ 2512.029920] wlan0: moving STA c0:25:06:00:00:00 to state 2
[ 2512.029922] wlan0: No basic rates in AssocResp. Using min supported rate instead.
[ 2512.034967] ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
[ 2512.035060] cfg80211: Calling CRDA for country: DE
[ 2512.038641] cfg80211: Regulatory domain changed to country: DE
[ 2512.038644] cfg80211:   (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[ 2512.038647] cfg80211:   (2400000 KHz - 2483500 KHz @ 40000 KHz), (N/A, 2000 mBm)
[ 2512.038649] cfg80211:   (5150000 KHz - 5350000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[ 2512.038651] cfg80211:   (5470000 KHz - 5725000 KHz @ 40000 KHz), (N/A, 2698 mBm)
[ 2512.112829] wlan0: moving STA c0:25:06:00:00:00 to state 3
Comment 1 John W. Linville 2012-05-01 13:26:44 UTC
Apparently DHT is some sort of extension to the bittorrent protocol?  It seems to enable a greater number of peers?  So that probably increases the number of small packets on the network?  But they should all be going through the same router in most cases (i.e. out to the Internet), so I don't know if that would make a difference to the aggregation algorithm.

Have you tried the 11n_disable option to iwlwifi?  While this would obviously, decrease your available bandwidth, I wonder if it would help with the "stalls"?  Does your network recover from these stalls?  Or are they permanent?

We seem to be getting a lot of reports about bad 802.11n performance with iwlwifi.  I wonder if there is some larger problem...?
Comment 2 unggnu 2012-05-01 17:19:51 UTC
DHT is a protocol to find peers without a central host so more small UDP packages are sent around.
Running Bittorrent with DHT and probably the usual high number of connections is just the easiest way for me to trigger this stall but there might be many others like being further away from the router or using 40 MHz 802.11n.
The interesting part is that if I am in the room directly behind the router the stalls doesn't happen in most cases even with Bittorrent running.

No, I haven't tried 11n_disable option but I am going to, thanks.

The stalls are mostly permanent I think. Normally I enable/disable WLAN to get the connection back but I know at least one case were the connection didn't recover in ~half an hour although my pc was still connected to the AP.

Yes, 802.11n was always an issue on Linux for me.
Like using 40 MHz (two channels at once) never really worked for me without disconnects or stalls.
Comment 3 unggnu 2012-05-02 19:26:39 UTC
I have only tested the 11n_disable option for two days but hadn't any stalls during this time even with Bittorrent running.
Since I am only using WLAN for my Internet connection even the speed is fine with me too.

But it is still sad that after so many years 802.11n still doesn't work fine on Linux. And I don't have some rare, not vendor supported hardware.
Comment 4 unggnu 2012-05-05 06:29:45 UTC
Just to confirm a longer test I hadn't any stalls since using the option. So how can I help to debug this?
Comment 5 unggnu 2012-06-01 20:01:10 UTC
Still an issue with 3.4 and it still does work without problems when n is disabled.
Comment 6 Lukas Elsner 2012-06-10 18:23:42 UTC
same problem here since several kernel versions.
stable workaround is disabling 11n, but not an acceptable solution, because most internet lines are faster than g-speed.

3.4 not working 
3.5-rc2 not working

but:

setting
CONFIG_IWLWIFI_DEBUG=y
CONFIG_IWLWIFI_DEBUG_EXPERIMENTAL_UCODE=y

will result in a stable connection which has (as result of my first tests) very low speed (~5mbit)
Comment 8 Daniel 2012-06-17 16:53:53 UTC
I have the same problems since upgrading to 3.4 (3.3 was fine). Also using AVM 7390 router on 5 GHz.

I tried both patches, but they didn't fix the issue. It still happens.
Comment 9 Jose Marino 2012-06-18 02:33:51 UTC
I see the same problem. Kernel 3.3 works fine but with 3.4 wireless stalls.

From lspci:
03:00.0 Network controller: Intel Corporation Centrino Ultimate-N 6300 (rev 35)

Firmware:
iwlwifi 0000:03:00.0: loaded firmware version 9.221.4.1 build 25532

I've seen this error on dmesg sometimes when the stall happens:
[   84.383705] iwlwifi 0000:03:00.0: Queue 2 stuck for 2000 ms.
[   84.383708] iwlwifi 0000:03:00.0: Current SW read_ptr 161 write_ptr 169
[   84.383760] iwlwifi 0000:03:00.0: Current HW read_ptr 161 write_ptr 169
[   84.383762] iwlwifi 0000:03:00.0: On demand firmware reload
[   84.388424] iwlwifi 0000:03:00.0: Failing on timeout while stopping DMA channel 1 [0x07fd0001]
[   84.388874] ieee80211 phy0: Hardware restart was requested
[   84.388939] iwlwifi 0000:03:00.0: L1 Enabled; Disabling L0S
[   84.395747] iwlwifi 0000:03:00.0: Radio type=0x0-0x3-0x1
Comment 10 Daniel 2012-06-18 20:53:09 UTC
Since upgrading to 3.4.3 the issue seems to be fixed for me. Can any of the other reporters confirm this?
Comment 11 Jose Marino 2012-06-19 03:51:08 UTC
I do agree with Daniel that 3.4.3 seems to make the situation better, but I still see connection issues that I do not see with 3.3.x

The connection is very unreliable. While pinging a local machine in the network I get:
135 packets transmitted, 108 received, 20% packet loss, time 174895ms
rtt min/avg/max/mdev = 1.459/90.582/1326.175/175.244 ms, pipe 2

I have been unable to get any useful debug info. I tried:
echo 0x0007BBC9 > /sys/module/iwlwifi/parameters/debug

and this is the output I see:
[  594.542058] iwlwifi 0000:03:00.0: I iwl_tt_handler Queueing thermal throttling work.
[  596.687066] iwlwifi 0000:03:00.0: I iwl_tt_handler Queueing thermal throttling work.
[  605.163945] iwlwifi 0000:03:00.0: I iwl_tt_handler Queueing thermal throttling work.
[  605.467038] usb usb1: usb auto-resume
[  605.467046] ehci_hcd 0000:00:1a.0: resume root hub
[  605.486652] hub 1-0:1.0: hub_resume
[  605.486669] hub 1-0:1.0: port 1: status 0507 change 0000
[  605.486694] usb 1-1: usb auto-resume
[  605.486753] hub 1-0:1.0: state 7 ports 3 chg 0000 evt 0000
[  605.512656] ehci_hcd 0000:00:1a.0: GetStatus port:1 status 001005 0  ACK POWER sig=se0 PE CONNECT
[  605.523600] usb 1-1: finish resume
[  605.524152] hub 1-1:1.0: hub_resume
[  605.525036] hub 1-1:1.0: port 4: status 0507 change 0000
[  605.525515] ehci_hcd 0000:00:1a.0: reused qh ffff8802248b7780 schedule
[  605.525525] usb 1-1: link qh256-0001/ffff8802248b7780 start 1 [1/0 us]
[  605.525572] hub 1-1:1.0: state 7 ports 6 chg 0000 evt 0000
[  605.526741] usb 1-1.4: usb auto-resume
[  605.563706] usb 1-1.4: finish resume
[  607.309948] iwlwifi 0000:03:00.0: I iwl_tt_handler Queueing thermal throttling work.
[  608.291784] usb 1-1.4: usb auto-suspend, wakeup 0
[  609.454331] iwlwifi 0000:03:00.0: I iwl_tt_handler Queueing thermal throttling work.
[  610.297558] hub 1-1:1.0: hub_suspend
[  610.297571] usb 1-1: unlink qh256-0001/ffff8802248b7780 start 1 [1/0 us]
[  610.297969] usb 1-1: usb auto-suspend, wakeup 1
[  612.303597] hub 1-0:1.0: hub_suspend
[  612.303608] usb usb1: bus auto-suspend, wakeup 1
[  612.303613] ehci_hcd 0000:00:1a.0: suspend root hub
[  613.744611] iwlwifi 0000:03:00.0: I iwl_tt_handler Queueing thermal throttling work.
[  615.891293] iwlwifi 0000:03:00.0: I iwl_tt_handler Queueing thermal throttling work.
[  618.034056] iwlwifi 0000:03:00.0: I iwl_tt_handler Queueing thermal throttling work.

Enabling iwlwifi debug seems to trigger the USB resume. I have nothing plugged to the USB ports.
Comment 12 Daniel 2012-06-19 18:26:43 UTC
Ok, I have to correct myself. Updating to 3.4.3 did not fully solve the issue.
Now, immediately after resuming from hibernate, it got stuck again:


[19634.521745] PM: restore of devices complete after 831.979 msecs
[19634.521856] PM: Image restored successfully.
[19634.521857] Restarting tasks ... done.
[19634.522666] PM: Basic memory bitmaps freed
[19634.522780] video LNXVIDEO:00: Restoring backlight state
[19639.112923] e1000e 0000:00:19.0: irq 41 for MSI/MSI-X
[19639.214509] e1000e 0000:00:19.0: irq 41 for MSI/MSI-X
[19639.214891] ADDRCONF(NETDEV_UP): eth0: link is not ready
[19639.217209] iwlwifi 0000:03:00.0: L1 Enabled; Disabling L0S
[19639.217448] iwlwifi 0000:03:00.0: Radio type=0x0-0x3-0x1
[19639.356186] ADDRCONF(NETDEV_UP): wlan0: link is not ready
[19648.086968] wlan0: authenticate with xx:xx:xx:xx:xx:xx
[19648.094632] wlan0: send auth to xx:xx:xx:xx:xx:xx (try 1/3)
[19648.095854] wlan0: authenticated
[19648.098786] wlan0: associate with xx:xx:xx:xx:xx:xx (try 1/3)
[19648.100183] wlan0: RX AssocResp from xx:xx:xx:xx:xx:xx (capab=0x411 status=0 aid=1)
[19648.100185] wlan0: associated
[19648.105239] ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
[19648.105287] cfg80211: Calling CRDA for country: DE
[19648.972293] cfg80211: Regulatory domain changed to country: DE
[19648.972296] cfg80211:   (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[19648.972298] cfg80211:   (2400000 KHz - 2483500 KHz @ 40000 KHz), (N/A, 2000 mBm)
[19648.972299] cfg80211:   (5150000 KHz - 5250000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[19648.972301] cfg80211:   (5250000 KHz - 5350000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[19648.972302] cfg80211:   (5470000 KHz - 5725000 KHz @ 40000 KHz), (N/A, 2698 mBm)
[19658.940775] wlan0: no IPv6 routers present
[19663.754567] iwlwifi 0000:03:00.0: Queue 2 stuck for 2000 ms.
[19663.754579] iwlwifi 0000:03:00.0: Current SW read_ptr 78 write_ptr 83
[19663.754650] iwlwifi 0000:03:00.0: Current HW read_ptr 78 write_ptr 83
[19663.754656] iwlwifi 0000:03:00.0: On demand firmware reload
[19663.755195] ieee80211 phy0: Hardware restart was requested
[19663.755289] iwlwifi 0000:03:00.0: L1 Enabled; Disabling L0S
[19663.755472] iwlwifi 0000:03:00.0: Radio type=0x0-0x3-0x1
[19704.112187] wlan0: deauthenticated from xx:xx:xx:xx:xx:xx (Reason: 6)
[19704.130624] cfg80211: Calling CRDA to update world regulatory domain
[19704.133712] cfg80211: World regulatory domain updated:
[19704.133714] cfg80211:   (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[19704.133716] cfg80211:   (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[19704.133718] cfg80211:   (2457000 KHz - 2482000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[19704.133720] cfg80211:   (2474000 KHz - 2494000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[19704.133721] cfg80211:   (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[19704.133723] cfg80211:   (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
Comment 13 Jose Marino 2012-06-21 15:04:51 UTC
I tried to do a bisect to find when exactly this bug was introduced. I could not really finish it, at some point it is not clear if I'm seeing the same bug or a different one that got fixed later on.

I'm copying the log of the partial bisect in hopes that it helps.

git bisect start
# bad: [76e10d158efb6d4516018846f60c2ab5501900bc] Linux 3.4
git bisect bad 76e10d158efb6d4516018846f60c2ab5501900bc
# good: [c16fa4f2ad19908a47c63d8fa436a1178438c7e7] Linux 3.3
git bisect good c16fa4f2ad19908a47c63d8fa436a1178438c7e7
# bad: [141124c02059eee9dbc5c86ea797b1ca888e77f7] Delete all instances of asm/system.h
git bisect bad 141124c02059eee9dbc5c86ea797b1ca888e77f7
# bad: [3b59bf081622b6446db77ad06c93fe23677bc533] Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
git bisect bad 3b59bf081622b6446db77ad06c93fe23677bc533
# bad: [74dd1521d0b4f940cdd3ce7b9d988836bef589b8] Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-next into for-davem
git bisect bad 74dd1521d0b4f940cdd3ce7b9d988836bef589b8
# good: [97767a87f3be8834192dc3fc9412aaccf708d87f] be2net: Remove unused OFFSET_IN_PAGE() macro
git bisect good 97767a87f3be8834192dc3fc9412aaccf708d87f
# bad: [cc4bf501a237f5232df6d4aeb7b24ac0362958c8] Merge branch 'wl12xx-next' into for-linville
git bisect bad cc4bf501a237f5232df6d4aeb7b24ac0362958c8
Comment 14 Daniel 2012-06-30 20:11:16 UTC
I haven't had the issue since I upgraded to 3.4.4. Seems to be fixed with that revision.
Comment 15 unggnu 2012-07-06 15:21:55 UTC
Still an issue with 3.4.4 after a suspend resume:
[  851.505743] wlan0: no IPv6 routers present
[  899.550886] iwlwifi 0000:02:00.0: Queue 2 stuck for 2000 ms.
[  899.550894] iwlwifi 0000:02:00.0: Current SW read_ptr 115 write_ptr 119
[  899.550950] iwlwifi 0000:02:00.0: Current HW read_ptr 115 write_ptr 119
[  899.550954] iwlwifi 0000:02:00.0: On demand firmware reload
[  899.551410] ieee80211 phy0: Hardware restart was requested
[  899.551514] iwlwifi 0000:02:00.0: L1 Enabled; Disabling L0S
[  899.551710] iwlwifi 0000:02:00.0: Radio type=0x0-0x3-0x1
Comment 16 Jose Marino 2012-08-10 15:41:36 UTC
Created attachment 77291 [details]
dmesg output showing connection stall

I still see stalls in kernel 3.5.1
In this particular instance I saw a couple of stutters, caused by firmware restarts, followed by the connection just stopping. While no packets were going through, the interface still showed as associated.
I attach the output from dmesg.
Comment 17 unggnu 2012-08-12 08:43:25 UTC
Don't take this the wrong way but it looks like your plan is to sit this out until nobody uses this piece of hardware anymore.

I mean what is missing? It is easy to reproduce after a few minutes if you connect to an AP in the 5 GHz band and it is a piece of hardware which isn't rare.
And I am willing to debug it if you tell me how.

And without disabling 802.11n this WLAN hardware is absolutely unusable under Linux because the connection stalls all the time.

And 54 Megabit gross WLAN bandwidth is not fast enough for modern Internet connections anymore.

At least have the decency to change the status from needinfo to confirmed.
Comment 18 unggnu 2012-10-14 16:17:49 UTC
The connection doesn't stall as fast as usual with Kernel 3.6 nevertheless the packet drop rate is pretty high at around 15-40% while using ping and the WLAN connection stopped after some time too.
So I am still stuck with 802.11a and 54 Mbit.
Comment 19 Jose Marino 2012-12-05 04:34:07 UTC
I haven't seen any problems in a while. I'm running 3.6.8.
Comment 20 unggnu 2012-12-08 16:30:00 UTC
@Jose
Have you checked your packet drop rate?
I haven't any stalls after over a day but the drop rate is still over 10% which is really noticeable during time critical applications and sometimes even while loading websites
Comment 21 unggnu 2012-12-15 10:53:53 UTC
The high packet loss is still an issue with 3.7. 10% - 30% packet loss is normal with 802.11n.
If I disable n and only use 54 Megabit there is only 0-1% packet loss from the same location.
Comment 22 Stanislaw Gruszka 2012-12-15 11:00:22 UTC
That could be pretty normal in noisy environment. Slower rates are more robust, so less packets are lost. Overall throughput is interesting, on 11n is should be better.
Comment 23 unggnu 2012-12-15 14:36:59 UTC
Downloads are normally slower with n in my case because of the packet loss. And the loading of websites often takes much longer or hangs for some seconds. That's why I always disable n again after testing.

Also n should use MIMO and thus be better equipped to deal with interferences.

And I am using the 5GHz band in which I have the only AP nearby according to my router.
Comment 24 unggnu 2013-01-30 19:37:26 UTC
So I have pinged my local router a 1000 times. Once with 802.11n disabled and once without. The kernel was 3.7.1.

802.11n disabled:
1000 packets transmitted, 881 received, 11% packet loss, time 999736ms
rtt min/avg/max/mdev = 1.747/291.081/2872.046/488.365 ms, pipe 3

vs


default:
1000 packets transmitted, 1000 received, 0% packet loss, time 1000385ms
rtt min/avg/max/mdev = 1.242/3.110/96.074/8.992 ms
Comment 25 unggnu 2013-01-30 19:39:30 UTC
I have switched the headlines. 802.11n disabled has zero packet loss and 802.11n 11%.
Comment 26 Emmanuel Grumbach 2014-03-18 08:34:07 UTC

*** This bug has been marked as a duplicate of bug 56581 ***