Bug 130230 - USB transmission is very erratic with Auerswald telephony equipment
Summary: USB transmission is very erratic with Auerswald telephony equipment
Status: RESOLVED WONTFIX
Alias: None
Product: SUSE LINUX 10.0
Classification: openSUSE
Component: Kernel (show other bugs)
Version: Final
Hardware: x86 SuSE Linux 10.0
: P5 - None : Critical
Target Milestone: ---
Assignee: Wolfgang Mües
QA Contact: E-mail List
URL:
Whiteboard:
Keywords:
Depends on: 129938
Blocks:
  Show dependency treegraph
 
Reported: 2005-10-24 09:07 UTC by Manfred Hollstein
Modified: 2007-02-01 13:15 UTC (History)
0 users

See Also:
Found By: Other
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments
CARDBUS_IO_SIZE.patch (4.06 KB, patch)
2005-11-07 12:01 UTC, Olaf Hering
Details | Diff
git3-git4.log (16.97 KB, text/plain)
2005-11-08 19:25 UTC, Olaf Hering
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Manfred Hollstein 2005-10-24 09:07:53 UTC
Hi there,

I'm using the Auerswald COMpact 2206 USB telephony switch successfully
since quite some time (on SLES 9, SL-9.2 and other distros). Since
SL-10.0 the transmission is very erratic with lots of retries, most often
resulting in transmission completely breaking off. Here's an excerpt from
/var/log/messages:

Oct 24 09:37:54 merkur kernel: usb 1-2: new full speed USB device using uhci_hcd
and address 6
Oct 24 09:37:54 merkur kernel: usb 1-2: configuration #1 chosen from 2 choices
Oct 24 09:37:55 merkur kernel: drivers/usb/misc/auerswald.c: device is a COMpact
 2206,Ser# xxxxxxx
Oct 24 09:38:46 merkur kernel: drivers/usb/misc/auerswald.c: control read: trans
mission error -75, can not retry
Oct 24 09:38:47 merkur kernel: drivers/usb/misc/auerswald.c: control read: trans
mission error -75, can not retry
Oct 24 09:39:06 merkur kernel: drivers/usb/misc/auerswald.c: control read: trans
mission error -75, can not retry
Oct 24 09:39:08 merkur kernel: drivers/usb/misc/auerswald.c: control read: trans
mission error -75, can not retry

This means, I cannot program my telephony switch anymore... :-(

This incident may be related to #129938.
Comment 1 Olaf Hering 2005-10-26 23:12:24 UTC
can you try a newer kernel?
ftp.suse.com/pub/projects/kernel/kotd/i386/HEAD/kernel-default.i586.rpm
Comment 2 Manfred Hollstein 2005-10-27 09:20:29 UTC
Just tried kernel-default-2.6.14_rc5_git5-20051026104853.i586.rpm and it appears to fix THIS problem. Now my wireless device (Intel Corporation PRO/Wireless LAN 2100 3B Mini PCI Adapter (rev 04)) doesn't work anymore; instead, /var/log/messages gets flooded by the following message:

  kernel: eth1 (WE) : Driver using old /proc/net/wireless support, please fix driver !

Will try the kotd from SL100_BRANCH now and keep you informed.
Comment 3 Manfred Hollstein 2005-10-27 09:52:21 UTC
kernel-default-2.6.13-SL100_BRANCH_20051026063700 doesn't fix the problem unfortunately :-(
Comment 4 Olaf Hering 2005-10-27 11:11:53 UTC
I see no auerswald specific changes in 2.6.14. 
Comment 5 Manfred Hollstein 2005-10-27 11:40:37 UTC
So it must be a generic USB change then?!?
Comment 6 Olaf Hering 2005-10-27 13:32:07 UTC
very likely, can you try to narrow it down? try a minimal .config for your system with 2.6.14-rc1,2,3,4,5 to see where it works again.
Comment 7 Manfred Hollstein 2005-10-28 12:32:38 UTC
Problem is time which I don't have... :-( I'll be travelling most of the next two weeks. If someone can build some kernels for me and point me at them, that would help tremendously.
Comment 8 Olaf Hering 2005-11-02 13:48:49 UTC
There will be rpms for 2.6.14-rc{1,2,3,4,5} in

ftp://ftp.suse.com/pub/people/olh/kernel/bug130230/

plain rpms, no suse patches, no km_ packages. should be enough to test the auerswald driver.
Comment 9 Manfred Hollstein 2005-11-04 10:51:30 UTC
Thanks, I just downloaded the kernels (sitting in a meeting...) and will test them tomorrow.
Comment 10 Manfred Hollstein 2005-11-05 16:43:48 UTC
OK, tried the kernels now, here are the results:

rc1-_rc1: sort of OK, but still didn't work.
rc2-_rc2: catastrophe!
rc3-_rc3: better than rc1 and rc2, but still didn't work.
rc4-_rc4: much better than rc{1,2,3}, but still didn't work.
rc5-_rc5: much better than rc{1,2,3,4}, but still didn't work.
final: finally works OK!

HTH, cheers.

l8er
manfred
Comment 11 Olaf Hering 2005-11-06 18:29:14 UTC
I have uploaded more kernels. can you check if rc5-git4 works, and -git3 not?
Comment 12 Manfred Hollstein 2005-11-06 19:11:09 UTC
Just tried the kernels, and your suspicion was absolutely correct! -git3 almost failed immediately, while -git4 works properly.
Comment 13 Olaf Hering 2005-11-07 12:01:53 UTC
Created attachment 56580 [details]
CARDBUS_IO_SIZE.patch

there is now a new rpm with the attached patch.
Comment 14 Manfred Hollstein 2005-11-07 14:02:25 UTC
Hmm, just tried kernel-default-2.6.13-20051107_bug130230.i586.rpm, but it doesn't solve the problem; failed to succeed in any completely successfull communication. Data transfer terminates prematurely at between 72-96%
Comment 15 Olaf Hering 2005-11-07 14:28:39 UTC
ok, so it must be something else. there are not many changes between -git3 and -git4.

http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=4196c3af25d98204216a5d6c37ad2cb303a1f2bf

http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=e80eda94d3eaf1d12cfc97878eff77cd679dabc9

perhaps there is more than one patch required to fix this bug.
Comment 16 Olaf Hering 2005-11-08 10:54:13 UTC
there is now git3 + the CARDBUS_IO_SIZE patch.
Comment 17 Manfred Hollstein 2005-11-08 17:48:03 UTC
Unfortunately, this one doesn't fix the problem either; actually, it's worse than kernel-default-2.6.13-20051107_bug130230.i586.rpm from Comment #14
Comment 18 Olaf Hering 2005-11-08 19:25:21 UTC
Created attachment 56719 [details]
git3-git4.log

differences between git3 and 4.
I cant believe that the network changes are related. and the posix timer change looks also unrelated...
Comment 19 Olaf Hering 2005-11-09 12:19:57 UTC
jg, do you have a fix for this part of the bug:

...
Just tried kernel-default-2.6.14_rc5_git5-20051026104853.i586.rpm and it
appears to fix THIS problem. Now my wireless device (Intel Corporation
PRO/Wireless LAN 2100 3B Mini PCI Adapter (rev 04)) doesn't work anymore
...
Comment 20 Joachim Gleissner 2005-11-09 13:00:05 UTC
2.6.14 comes with ipw2x00 drivers, but older versions than we included in our 10.0 kernel before. If you don't use WPA it may sufficient to add PREFER_WPA_SUPPLICANT='no' to /etc/sysconfig/network/ifcfg-wlan-* .
Comment 21 Olaf Hering 2005-11-17 10:42:08 UTC
have you tried the PREFER_WPA_SUPPLICANT='no' thing with 2.6.14?
Comment 22 Manfred Hollstein 2005-11-21 11:54:29 UTC
No because I am using WPA :-( I can try, but it doesn't solve the problem with the old version, does it?
Comment 23 Manfred Hollstein 2005-12-09 14:04:59 UTC
(In reply to comment #18)
> Created an attachment (id=56719) [edit]
> git3-git4.log
> 
> differences between git3 and 4.
> I cant believe that the network changes are related. and the posix timer change
> looks also unrelated...
> 

Getting back to the original issue... I tried playing with my own kernels in the meantime (vanilla and patched ones, >= 2.6.14.2) and found out, that in some rare situations communication also collapses; so it didn't appear to have been fixed in newer kernels. I then tried the USB module provided by Auerswald at <http://www.commanderbasic.de/download/datei/linux/usbtreiber/v219/auerusb-2.1.9.tgz>
and guess what, it doesn't fail anymore. Built it for our latest update kernel 2.6.13-15.7, too, and it works here as well.

I haven't yet investigated who's maintaining the auerswald.ko module in the standard kernel sources, but those sources need to be updated!!!
Comment 24 Karsten Keil 2005-12-09 15:04:41 UTC
The author himself.
From kernel MAINTAINER file:
USB AUERSWALD DRIVER
P:      Wolfgang Muees
M:      wolfgang@iksw-muees.de
L:      linux-usb-users@lists.sourceforge.net
L:      linux-usb-devel@lists.sourceforge.net
S:      Maintained
Comment 25 Manfred Hollstein 2005-12-09 15:10:20 UTC
Yep, found that myself in the meantime; have contacted Wolfgang now. Will see if/when he'll update the driver sources...
Comment 26 Olaf Hering 2005-12-09 15:27:41 UTC
This tgz is from 2005-04-27. Perhaps one driver update broke it.
Comment 27 Manfred Hollstein 2005-12-09 17:18:03 UTC
No, it's the other way round actually. We have version 0.9.11 in our tree (the same as in SLES 9's kernel where the driver works flawlessly!), while the tgz contains driver version 2.1.9. This makes a difference!

I'd say some updates (generic or driver updates doesn't matter) broke the old version of the auerswald.ko driver in our tree (as well as in the vanilla sources); so Wolfgang (or someone else) should update the driver in the vanilla tree!
Comment 28 Wolfgang Mües 2005-12-28 09:47:34 UTC
Actually, the _real_ problem is not inside the auerswald driver, but in the error handling of the usb core. The auerswald devices are using the control endpoint to communicate, and the Linux usb kernels are very different in respect to recover from transmission errors. The version from the auerswald website has a better error recovery. 
Comment 29 Olaf Hering 2006-01-02 12:49:57 UTC
Can you merge your fixes with Linus this decade?
Comment 30 Andreas Jaeger 2007-02-01 13:15:18 UTC
Bug is stale, closing as WONTFIX.