r8169/RT2501USB Network interface hangs with debug output on syslog

[Steve Netting]

When under heavy load RT2501USB will hang. Applications report 'Transmit Buffer Full' to logs. Temporary solution is to start/stop wireless networking with NetworkManager. Easily reproducible on my system but hammering the interface (Second Life client reliably causes a crash within a few minutes).

lsusb:-

root@desktop:~# lsusb
Bus 005 Device 006: ID 148f:2573 Ralink Technology, Corp. RT2501USB Wireless Adapter
Bus 005 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 003 Device 004: ID 046d:c315 Logitech, Inc. Classic New Touch Keyboard
Bus 003 Device 003: ID 1241:1111 Belkin Mouse
Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 001 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub

Debug logged to syslog at time of hang:-

Oct 31 17:06:18 desktop NetworkManager: <debug> [1225472778.450624] periodic_update(): Roamed from BSSID 00:50:18:43:ED:84 (binkynet) to (none) ((none))
Oct 31 17:06:24 desktop NetworkManager: <debug> [1225472784.453670] periodic_update(): Roamed from BSSID (none) ((none)) to 00:50:18:43:ED:84 (binkynet)
Oct 31 17:08:18 desktop NetworkManager: <debug> [1225472898.534606] periodic_update(): Roamed from BSSID 00:50:18:43:ED:84 (binkynet) to (none) ((none))
Oct 31 17:08:24 desktop NetworkManager: <debug> [1225472904.539093] periodic_update(): Roamed from BSSID (none) ((none)) to 00:50:18:43:ED:84 (binkynet)
Oct 31 17:14:03 desktop NetworkManager: <info> (wlan0): supplicant connection state change: 7 -> 6
Oct 31 17:14:03 desktop NetworkManager: <info> (wlan0): supplicant connection state change: 6 -> 7
Oct 31 17:17:01 desktop kernel: [ 7772.294280] phy1 -> rt2x00queue_write_tx_frame: Error - Arrived at non-free entry in the non-full queue 2.
Oct 31 17:17:01 desktop kernel: [ 7772.294283] Please file bug report to http://rt2x00.serialmonkey.com.
Oct 31 17:17:01 desktop kernel: [ 7772.294335] phy1 -> rt2x00queue_write_tx_frame: Error - Arrived at non-free entry in the non-full queue 2.
Oct 31 17:17:01 desktop kernel: [ 7772.294339] Please file bug report to http://rt2x00.serialmonkey.com.
Oct 31 17:17:01 desktop /USR/SBIN/CRON[8512]: (root) CMD ( cd / && run-parts --report /etc/cron.hourly)

uname -a:-

Linux desktop 2.6.27-7-generic #1 SMP Thu Oct 30 04:18:38 UTC 2008 i686 GNU/Linux

Will also look into submitting a bug report at http://rt2x00.serialmonkey.com.

[Steve Netting (steve-netting)]

Previously reported, thought resolved on 2.6.26 but evidently not (or reintroduced?) on 2.6.27:-

http://rt2x00.serialmonkey.com/phpBB/viewtopic.php?f=5&t=4927

[Leann Ogasawara (leannogasawara)]

Hi Steve,

I think the following upstream LKML thread is interesting with regards to this bug:

http://lkml.org/lkml/2008/9/26/292

A patch has been accepted into the upstream kernel and is already in the upcoming ubuntu-jaunty kernel git tree:

ogasawara@yoji:~/ubuntu-jaunty$ git log -p 0262ab0df64a67d4c0ed7577a29b7d866819cc68
commit 0262ab0df64a67d4c0ed7577a29b7d866819cc68
Author: Ivo van Doorn <email address hidden>
Date: Fri Aug 29 21:04:26 2008 +0200

    rt2x00: Fix race conditions in flag handling

    Some of the flags should be accessed atomically to
    prevent race conditions. The flags that are most important
    are those that can change often and indicate the actual
    state of the device, queue or queue entry.

    The big flag rename was done to move all state flags to
    the same naming type as the other rt2x00dev flags and
    made sure all places where the flags were used were changed. ;)

    Thanks to Stephen for most of the queue flags updates,
    which fixes some of the most obvious consequences of the
    race conditions. Among those the notorious:

    rt2x00queue_write_tx_frame: Error - Arrived at non-free entry in the non-full queue 0.
    rt2x00queue_write_tx_frame: Error - Arrived at non-free entry in the non-full queue 0.
    rt2x00queue_write_tx_frame: Error - Arrived at non-free entry in the non-full queue 0.

    Signed-off-by: Stephen Blackheath <email address hidden>
    Signed-off-by: Ivo van Doorn <email address hidden>
    Signed-off-by: John W. Linville <email address hidden>

It would be great if you could test once the Jaunty images are in a decent enough shape and also contain the 2.6.28 based Jaunty kernel. We'll at least need confirmation that this is resolved in the actively developed kernel before the kernel team can consider backporting to Intrepid - http://wiki.ubuntu.com/StableReleaseUpdates . Thanks.

[Launchpad Janitor (janitor)]

Per a decision made by the Ubuntu Kernel Team, bugs will longer be assigned to the ubuntu-kernel-team in Launchpad as part of the bug triage process. The ubuntu-kernel-team is being unassigned from this bug report. Refer to https://wiki.ubuntu.com/KernelTeamBugPolicies for more information. Thanks.

[Andy Whitcroft (apw)]

@Steve Netting -- I have built some test kernels based on the current Intrepid -proposed kernel with the patch suggested by Leann applied. Could you test those and report back here, the kernels can be found at the URL below:

    http://people.ubuntu.com/~apw/lp291697-intrepid/

[Andy Whitcroft (apw)]

@Steve -- were you able to test these kernels? Without feedback we will not be able to justify including this fix in the official kernels. If you could report any testing here.

[Andy Whitcroft (apw)]

As there has been no feedback to these testing requests we will close this one off, Won't Fix. If you are able to do testing please indicate so here, and I can produce some up to date kernels. If you are able to test please move this bug back to New.

[Umuntu-Tim- (thembinkosi)]

I have lucid 2.6.32-25-generic, and have this bug.