Ericsson F3507g (0bdb:1900) - does not display in NetworkManager, tho USB ACM devices are registered

which device in the lsusb output is your modem? I dont see anything that looks familiar fo rthis device.

The device doesn't correctly appear in my 'lsusb' output, though it did appear prior to my upgrade to Jaunty a couple of days ago. The upgrade to Jaunty *did* add the hopeful-looking cdc_wdm & cdc_acm messages in the 'dmesg' output.

I don't know if NetworkManager is necessarily at fault here - is there a lower subsystem this issue would be better addressed at?

Sorry, that should be 'doesn't currently appear' - ie it's just not there in my lsusb output.

please attach complete dmesg for plugging in. also attach complete lshal ... thanks.

that said: yes, its a problem if its not in lsusb ... means that something is really wrong.

> please attach complete dmesg for plugging in. also attach complete lshal ... thanks.

also your complete syslog please.

dmseg, lshal and syslog attached - I didn't know about the lshal command - couldn't see any sign of the WWAN card in it.

oh forgot:

udevadm info --query=all --attribute-walk --path=/sys/class/tty/ttyACM0

maybe also for ACM1 and 2

please.

Executing udevadm returns 'device path not found' for ACM0, 1, 2.

In fact, executing:

ls -R /sys/class/ | grep -i acm

returns zero results.

i really think your device is not detected at all. sounds more and more like a hardware issue. maybe remove the module and check whether you still see the cdc_acm messages in dmesg.

The Ericsson F3507g card is actually embedded into the Lenovo X200 - a bit fiddly to remove the module. However, rebooting to Vista shows that the device is working well - it is detected by Windows, and successfully connects to the T-mobile network - a screenshot of the connection details is attached.

To me, this indicates it's not a hardware issue - are there any further diagnostic steps I can take to determine why Ubuntu can't see the device?

I did a little history digging - I found that 3 days ago, before the Jaunty update, I was seeing this line in the output from lsusb:

Bus 008 Device 002: ID 0bdb:1900 Ericsson Business Mobile Networks BV

That line is no longer there in lsusb, I just double-checked after rebooting from Vista. If I get the chance I'll see if I can make an Intrepid Ibex LiveUSB and re-verify that the device shows under that pre-Jaunty release.

Running using an Intrepid Ibex LiveUSB, the Ericsson F3507g card is detected, and displayed in NetworkManager (0.7~~svn20081018t105859-0ubuntu1) as 3 devices - I'll attach the dmesg logs, etc.

Although the device show in Network Manager, it doesn't manage to successfully connect to the T-Mobile network when I select it - so running using Intrepid rather than Jaunty isn't a valid workaround for me :-(

For me this confirms that I don't have a hardware fault - and there must be something going wrong somewhere in the software stack. Thanks for your help so far - it's much appreciated!

chris also sees this from what i know. subscribed him as it worked for him in jaunty a few days ago. lets try to narrow down what upgrade broke this. I would think we should start to install different kernel versions.

One thing you could try the vanilla 2.6.29.1 kernel build from http://kernel.ubuntu.com/~kernel-ppa/mainline/

Just discovered : the card WORKS after resume-from-hibernate. This is with 'stock' Jaunty (linux kernel 2.6.28.11.15 and network-manager 0.7.1~rc4.1.cf199a964-0ubuntu2) - I've haven tried updating to the 2.6.29.1 kernel yet.

I've reproduced this twice now - it works beautifully after resume-from-hibernate, but has failed to work 3 times in a row after a simple restart.

Once the card is working post-hibernation, it also works post-suspend. I'm attaching the logs now...

This posted using the lovely lovely T-Mobile 3G network.

can you confirm that when it works you can also see your card in lsusb (which id does your card have?)?

yep, the device does show up in lsusb (full output attached) - this is the relevant line:

Bus 002 Device 004: ID 0bdb:1900 Ericsson Business Mobile Networks BV

Following up to this bug report. After rebooting my machine, I normally just suspend, it worked properly again. So I am not sure what caused my issue.

Chris

Roberto, please check the kernels from mainline. would be great if we can find a version that works even without hibernate. If all those versions don't make a difference, I would think that this could also be a bios bug or something - trying to upgrade that might be worth it.

I first upgraded to kernel 2.6.29.1 from mainline, then upgraded my BIOS from v1.05 to v2.06 (http://www-307.ibm.com/pc/support/site.wss/document.do?lndocid=MIGR-70347) - unfortunately neither of these updates caused any change in behaviour.

I might try downgrading to 2.6.27-10 kernel next, as bug #334413 suggests that versions after that revision had some problem which stopped the kernel from activating the card.

wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.29.1/linux-headers-2.6.29-02062901_2.6.29-02062901_all.deb
wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.29.1/linux-headers-2.6.29-02062901-generic_2.6.29-02062901_i386.deb
wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.29.1/linux-image-2.6.29-02062901-generic_2.6.29-02062901_i386.deb
sudo dpkg -i linux*.deb

maybe you changed your /etc/modprobe.d/ or /etc/modules things in the past? try to be sure that you dont load anything manually.

also please confirm that /dev/ttyACM0 actually exists.

I wanted to add that the Ericsson F3507G I have which shows up properly in NM has the same usb id. However I am using kernel 2.6.28-11-generic version 2.6.28-11.42. Note I don't actually use the device I just had it added to my X200 when I bought it in case I wanted to use it in the future. It appears from dmidecode that I am running the X200 BIOS 2.02.

Chris

Upgrading to BIOS v2.06 actually did change behaviour - it stopped the Ericsson F3507G card working on either Vista (where it previously worked) or my old Jaunty install (where it worked after resume from hibernate). That was quite... depressing, but in the end I wiped my Ubuntu partition, and did a fresh install from the final release of Jaunty. Lo and behold, the WWAN card started working on startup on Ubuntu! So I guess there must indeed have been something unpleasant in my /etc/modprobe.d/ or /etc/modules , which was wiped by the fresh install. However, the issue has now become reversed - in that the card no longer works in Ubuntu after resume from hibernate or suspend.

That's better than how it was before, even if it is rather mystifying. I don't know if it deserves a new bug report or not.

The device certainly does show up as /dev/ttyACM0 when it's working - and isn't there when it's failing after resume. It doesn't work at all in Vista, which no longer detects it as a device - presumably this must be down to the BIOS update. Chris, if you have Windows on your X200 - does it still detect the device with BIOS 2.02 ?

I am using
~/ uname -a
Linux kwinz-laptop 2.6.30-02063 #020630 SMP Wed Jun 10 09:04:38 UTC 2009 x86_64 GNU/Linux
and I have neither ttyACM* nor an entry in lsusb.

Worked with 2.6.28-11-generic.

Same problem here:
The card used to work with Jaunty 2.6.28-12, directly after booting/resume etc.
Currently i am running 2.6.28-13-generic #45-Ubuntu SMP Tue Jun 30 22:12:12 UTC 2009 x86_64 GNU/Linux and the card works, but not after a suspend/hibernate.

I also tried vanilla kernels from the ppa (same version as above and 2.6.31-rc1) and here the device isn't even recognized. no ttyACM[012].

Seems as if something got broken...?

Experiencing same problem on Lenovo X200 with embedded Ericsson F3507g card. Using 2.6.28-13-generic.

Used to work great, now does not show up in lsusb and /sys/devices/platform/thinkpad_acpi/wwan_enable does not exist.

I have not been able to make the card work by hibernating/resuming.

Makes the card inoperative, a real problem for those of us who depend on this functionality.

Correcting my previous post.

In fact I am running: 2.6.28-11-generic #43~undervolt1-Ubuntu SMP Fri Apr 17 18:32:02 UTC 2009 i686 GNU/Linux. Because I use the phc patches to undervolt my CPU.

Since my kernel has NOT been upgraded, it suggests that something else has gone wrong in the last few weeks, something that is not directly kernel related?

The thinkpad-acpi upstream maintainer here.

If the module does not show up in lsusb (run as *root*), it is blocked by the thinkpad firmware (powered down).

It will be powered down by the firmware if *any* of the following conditions are met:

1. Set to hidden, radio blocked or something like that in the BIOS setup screens (usually cannot be reenabled by software until reconfigured in BIOS);

2. Disabled in software, and not re-enabled (warning: this state *IS* kept across reboots and power down). In this case, thinkpad-acpi is supposed to be able to re-enable the device, if you use the rfkill sysfs class to do it;

3. Radio-kill physical switch is enabled (i.e. radios are killed). NOTHING can reenable the device until you disable the radio-kill switch. Mind you, I have been told there are kernel bugs in this area for some WLAN devices, but I haven't heard of any on the WWAN devices;

4. The kernel rfkill subsystem orders thinkpad-acpi to kill the radio (check /sys/class/rfkill/*), which it can do at any time, for any reason;

5. Broken backports of the rfkill subsystem, or other kernel bugs.

So, please check if any of these conditions are the issue, and do a full checking of the rfkill behaviour to make sure there are no problems with the rfkill subsystem. The thinkpad-acpi driver will log a lot of information if properly compiled with the THINKPAD_ACPI_DEBUG Kconfig option enabled, and loaded with the "debug=0xffff" parameter.

Ah, I forgot the *obvious*:

Make sure EHCI-HCD *and* UHCI-HCD are running properly, or the USB bus won't work, and obviously the device won't be there even if it is enabled.

Thank you for your guidance. Below my observations:

1. Set to hidden, radio blocked or something like that in the BIOS setup screens (usually cannot be reenabled by software until reconfigured in BIOS);

>Disabled/re-enabled wireless in BIOS. No effect.

2. Disabled in software, and not re-enabled (warning: this state *IS* kept across reboots and power down). In this case, thinkpad-acpi is supposed to be able to re-enable the device, if you use the rfkill sysfs class to do it;

>thinkpad-acpi does not show rfkill for the wwan, only for wifi card.

3. Radio-kill physical switch is enabled (i.e. radios are killed). NOTHING can reenable the device until you disable the radio-kill switch. Mind you, I have been told there are kernel bugs in this area for some WLAN devices, but I haven't heard of any on the WWAN devices;

>Physical switch is ON. Wifi works, wwan used to work up to a few weeks ago.

4. The kernel rfkill subsystem orders thinkpad-acpi to kill the radio (check /sys/class/rfkill/*), which it can do at any time, for any reason;

>No rfkill for wwan card (used to be there)

5. Broken backports of the rfkill subsystem, or other kernel bugs.

>Possible

6. Make sure EHCI-HCD *and* UHCI-HCD are running properly, or the USB bus won't work, and obviously the device won't be there even if it is enabled.

> built-in kernel for 2.6.28-11, lspci -v shows that they are loaded and working.
>>>Are we dependent on these two being build as modules?
>This was the case in 2.6.27-11 and changed with the newer kernel.

The following might help debug the problem:

dmesg shows that in fact the card IS detected during boot, the proper modules -cdc_acm and cdc_wdm- load and create the appropriate comm ports, but after a while the usb port is disconnected and ehci_hcd appears unhappy.

I have noted that in kernel 2.27.11 in my distribution (Ubuntu) the ehci_hcd was built as a module, whereas from 2.28.11 on it was built into the kernel.

.....
[ 3.920054] usb 2-4: new high speed USB device using ehci_hcd and address 2
....
[ 4.768172] usb 2-4: configuration #1 chosen from 2 choices
[ 4.816828] cdc_acm 2-4:1.1: ttyACM0: USB ACM device
[ 4.817162] cdc_wdm 2-4:1.5: cdc-wdm0: USB WDM device
[ 4.821853] usb0: register 'cdc_ether' at usb-0000:00:1d.7-4, CDC Ethernet
Device, 02:80:37:ec:02:00
[ 4.821869] usbcore: registered new interface driver cdc_ether
[ 4.822052] cdc_acm 2-4:1.3: ttyACM1: USB ACM device
[ 4.822520] cdc_wdm 2-4:1.6: cdc-wdm1: USB WDM device
[ 4.822532] usbcore: registered new interface driver cdc_wdm
[ 4.822668] cdc_acm 2-4:1.9: ttyACM2: USB ACM device
[ 4.823902] usbcore: registered new interface driver cdc_acm
[ 4.823904] cdc_acm: v0.26:USB Abstract Control Model driver for USB modems
and ISDN adapters
......
[ 9.840181] usb 2-4: USB disconnect, address 2
[ 9.841912] ehci_hcd 0000:00:1d.7: dma_pool_free buffer-2048, f5f48000/35f48000 (bad dma)
[ 9.842038] ehci_hcd 0000:00:1d.7: dma_pool_free buffer-2048, f5f48080/35f48080 (bad dma)

If it is of any interest,

I have a T500, which on Ubuntu 9.10 alpha 2 was not detecting the device (not shown in lsusb) but when I did the last update it started working. I can now see this line in the lsusb output.

Bus 002 Device 003: ID 0bdb:1900 Ericsson Business Mobile Networks BV

So maybe this could be marked as resolved in 9.10 ?

I have a Lenovo x200 running Jaunty 64, with similar issues, and I've looked at a few other bugs that also have strange "disappearing device" behaviour, such as:
https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/287893
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/334413

I've been observing this for about 3 weeks now trying to find a meaningful pattern that would help track this down, but I can't.

My observations:
Sometimes the card appears, sometimes it does not. When it does, it works fine using Network Manager, although I've yet to test the suspend / resume functionality that people have been having issues with.

When it does not:
 - There is no /dev/ttyACM[012].
 - It doesn't show up in lsusb.
 - There is nothing in dmesg hinting towards it (not even messages about cdc_acm).
 - Cycling Fn-F5 only affects Bluetooth and Wifi.
 - /sys/devices/platform/thinkpad_acpi/wwan_enable does not exist.

I'm currently running on 2.6.31-020631rc5-generic but have also tried v2.6.31-rc4, v2.6.31-rc3, v2.6.30, 2.6.28-14 and the stock Jaunty kernels before it. I have also tried the latest Karmic Alpha 3, but the results are the same. None of them seem to make the device show up or not show up any more or less frequently as far as I can tell.

I would say that on average, it shows up 60-70% of the time.

I'm running BIOS 3.05 (it seems 3.06 has been released, but the download link is broken - the changelog doesn't show anything related to this though). I've enabled everything possible in terms of internal components in the BIOS. The card's firmware is also the latest I think: R1D06

If there is anything else useful that I can provide, please let me know. I've attached a dmesg - this is when the card is not appearing.

I've upgraded my Lenovo X200 to Karmic Koala Alpha 6, and this behaviour is still present; on cold start-up the card appears in Network Manager and can create a connection, after suspend & resume the card disappears from Network Manager.

I've the same problems here. Lenovo X200 with Karmic Koala (updated in 2009 Oct 09). I can't get it working with the suspend/hibernate trick.
There is no /dev/ttyACM*

I'm debugging an x200s issue, and have tracked it down to the wwan card not being woken after suspend. After suspend, the card is still asleep so it doesn't show up in lsusb. It can be reactivated with:

  echo enabled > /proc/acpi/ibm/wan

Does this help anyone on this bug?

I think 334413 and 287893 are dups of this.

Hey, thanks Newman! As described by comment #46 , the card can be reactivated with:

echo enabled > /proc/acpi/ibm/wan

I'm using my Ericsson F3507g WWAN card to post this now, post-suspend...

Excellent, thanks Claire, this works for me too. My card now works reliably after startup (unlike issues I was having in comment 43), and running echo enabled > /proc/acpi/ibm/wan brings the card back reliably too as far as I can tell.

I was able to automate this by creating a script at /usr/lib/pm-utils/sleep.d/30wwan containing:

#!/bin/sh
#
# Bring WWAN card back up after suspend / hibernate.
#
. "${PM_FUNCTIONS}"

# Record the current operation to allow failure detection.
case "$1" in
        thaw|resume)
                echo enabled > /proc/acpi/ibm/wan
                ;;
esac

This may well be a horrible implementation, I just used one of the existing scripts as a guide. Corrections welcome, but it does work for me.

I should have mentioned that I'm using Karmic (clean install) and my card is a F3507g.

Also, in the script above, the line starting with "# Record the current..." is a relic from the script I used as a guide, has nothing to do with this script and should be removed.

I'm on a clean install of karmic, and the above method works for me with one change. the echo line should be:

  echo enable > /proc/acpi/ibm/wan

Judging from the existence of bug 756553; I'd venture to say that this dongle works at least in Natty at this point, so I'll mark this as Fix Released.

Note that I'd very much like to know if it works in Maverick too, and if you're still having issues in Natty, please don't hesitate to open a separate bug so we can look into it. Thanks!

Hi, I was having issues very similar to this one, I am keen but not especially competent with Linux. Most of the things discussed here are beyond my level.

My card was disappearing and would sometimes stay vanished for long periods then re-appear.
Things that would sometimes make it re-appear would then stop working.
It was very frustrating.

This is not a fix, my card still vanishes from time to time but one thing that so far has a 100% record for making the card re-appear is to re-start, enter the bios and do an F9 reset then F10 out and start as normal.

Every time so far, the card has then worked as normal, I am logged on using it now...

I hope that helps someone, like I said - I am hardly an expert.

Andrew: this is the wrong place to discuss unrelated problems. Please open a new bug report by typing "ubuntu-bug modemmanager" in a terminal.