Integration of 3G Wireless Modem Detection - 4.1.2 & 4.2.1

[droope]

droope,
You did not say that you edited the "option-ZTE rules file to select your MF636. Several ZTE modems have the same initial vendor and product IDs and there is no way we can determine the correct rule to use. For now, the user must uncomment the rule appropriate for his/her modem. The file is /etc/udev/rules.d/51-usb_modeswitch-option-ZTE-...rules. The correct rule is the first one there, on line 15. Just remove the "#". If there is still a problem, I will need to work it. Yes, you probably need all of the dotpets, except for the two marked as "hso", which are for high-speed modems, and the nozomi. But including them does no harm.

Afterthoughts:
1. The safest version of puppy to use until you have a working modem setup is 4.1.2. which is what the project is built on. Since it is only kernel-specific, any of the puppys using kernel 2.6.25.16 should be usable with the appropriate dotpets. So, stick with 4.1.2 for now.
2. Since you are not specific about everything you did to try to get your modem working, it is difficult to guess what you left out. If you installed all of the dotpets posted in the first item in the thread, did you also do this?

1. Some USB wireless modems may require two updated driver modules that tempestuous has provided here: http://www.murga-linux.com/puppy/viewto ... 448#254448
They are particularly for ZTE modems, but may help if other modems that use the option driver are not recognized. From the above link, download and install these files:
- option-patched-01Apr09-k2.6.25.16.pet
- usb-storage-ZTE-patch-k2.6.25.16.pet

Hi rerwin,

First of all, I sincerely thank you very much for your efforts, and apologize for not knowing how to explain enough my gratitude. Sorry for al the questions, too.

I did this on a fresh puppy 4.1.2, to avoid problems:

1- Install the following pets:

option-patched-01April09-k2.6.25.16.pet
usb-storage-ZTE-patch-k2.6.25.16.pet
3G_pupdial-wireless-11.pet
usb_modeswitch-1.0.2.pet

2 - Edit the file /etc/udev/rules.d/51-usb_modeswitch-option-ZTE-...rules and uncomment line 15.

3-Go to connection wizard, and select GPRS modem. The modem y choose is ttyS0 (the only one puppy says is detected), choose the apropiate connection settings (i think). And press ok.

Then, I get a message telling me to go to network and choose gprs config... It's weird.

Adittional info: I have now noticed that puppy does not detect the modem as a optical drive anymore. I think the reason behind that is that it is not plugged in directly, but through a usb extension, that i can't explain. It's like a 2 - to - one usb, I plug 2 usb's to the PC, and the one on the other end to the ZTE modem. I will now try to plug it in directly.

Cheers!

[rerwin]

kultex,
Thanks for your log files. They are very revealing!

The order of the MODULES list has changed, such that my patch to module loading is not a factor. Considering that you are able to recover by simply modprobing option, I conclude that our problem must be about timing.

The way this works is that the loading of usb_storage triggers the mode switch, which waits 10 seconds before switching. The switching should change the interface class (ic) to FF and trigger loading of option. But if the device is very slow to actually switch to modem mode, option is loaded before the modem appears to exist, so the loading is probably aborted. During bootup, while all that is going on, the initialization script waits up to a minute for things to get settled.

I think the solution is to retry the loading of option after the device has had time to settle as a modem, so the driver can see it. To test that idea, I converted patch1 into patch 3, which changes the rule for writing the MODULES lines, adds a udev rule for your modem to invoke a new script that waits 15 seconds and then modprobes option again. If it still fails, try increasing the delay time in /etc/udev/rules.d/60-delayed_modprobe.rules. Change the 15 to 30 or 45, but preferably not more, because of the 60 seconds the init script waits.

Earlier, I commented that using modprobe alone would not start the init script. But your logs show that option gets loaded early on, so that starts the script, even though the loading may not work out. The patch might compensate for option's running too early. Please install it (after uninstalling patch1) and try again, attaching your logs to your report. Thanks.
Richard

[rerwin]

droope,

3-Go to connection wizard, and select GPRS modem.

You were doing fine up to that point. This 3G project does not support that GPRS Connect-Wizard option, only the "Internet by dialup analog modem" button, which should actually say "... analog/wireless modem". So use that option and tell us how it goes.

The Connect Wizard should show your modem as ttyUSB0, but it might take a second reboot for detection to complete (because of hardware delays). Thanks.
Richard

[BarryK]

droope,

3-Go to connection wizard, and select GPRS modem.

You were doing fine up to that point. This 3G project does not support that GPRS Connect-Wizard option, only the "Internet by dialup analog modem" button, which should actually say "... analog/wireless modem".

Yes indeed, I'll change it right now.

[kultex]

rerwin,

genius, congratulations - works perfect now, whatever you do - have it plugged in at boot, plug it in afterwards, replug it ..... I just add you 2 files - e220-11 is plugging in at boot and e220-12 is plugging in afterwards.

BarryK,

when you are working right now on PGPRS, may I have a wish? - I am using the Huawei 99% in the train and all the time, the connection brakes - so it would be very useful, to have at least the choice, to activate automatic reconnection.
The only linux, I know, where this works is in the netapplet of pclinuxos, but because of switching to a small SSD in my Portege, I had to swich to Puppy. And to have this also in Puppy would be a great pleasure.

But nevertheless, I am really happy, that the Huawei E220 and pupdial work now like a charm

[Aitch]

OOOH, thanks Rich/Kultex....

I just bought one (E220) on ebay for my IBM lappy....

still have to get the icon225 sorted, though....

Will be seeing the owner of that next week, probably, keep you updated, Rich

Aitch

[rerwin]

kultex, Aitch,
Thanks for running the patch3 test, which confirms that a delayed modprobe of option is necessary. I took Aitch's suggestion and reviewed the usb_modeswitch intro page, to determine the extent of the issue.

I conclude that the delayed loading applies to modems that do not change product ID as part of switching. That appears to pertain to the older Huawei modems, which use the "-H 1" parameter with usb_modeswitch. The site also implies that the issue has been resolved in a later kernel, so our new 2.6.29.6-based test puppys should have the correction.

The new logic I propose applies only to the standard 2.6.25.16-kernel puppys, at least until we see how the advanced kernel handles these modems. I have cleaned up the modeswitch script to remove debug code that may have been slowing boot-up by 30 seconds. When the storage part of the device is detected and triggers loading of the usb-storage driver, the script will wait 5 seconds for usb-storage to connect with the device; then wait 1 second before running usbmodeswitch, which itself waits 4 seconds before testing for success; and then modprobe the option driver. That;s a total of 10 seconds between storage-part detection and loading of the modem driver. (If we need to wait longer, increasing the "-s" option in the usb_modeswitch command after the last 'else' in /usb/sbin/usb_modeswitch.sh would extend the delay before loading.)

The attached "patch4" replaces all of the previous three patches, so they should be uninstalled (if possible). Patch4 will work even if the others are not uninstalled. Please try the new patch with the E220 and any other Huawei modem, in puppy 4.1.2 or 4.2.1 (standard kernel) AND in Barry's new 4.3 beta1 with the advanced (2.6.29.6) kernel. Thanks.
Richard

[kultex]

uninstalled all patches, installed patch 4 - still everything perfect
added 2 more files - e220-21 is plugging in at boot and e220-22 is plugging in afterwards.

[rerwin]

kultex,

uninstalled all patches, installed patch 4 - still everything perfect

That's music to my ears! Please ignore my request to try this with the new beta1. Barry left out a hotplug/usb piece of the kernel, so we cannot test 3G with it.

This result gives me the green light to put together a package for Barry to add to the beta2 release. Thank you very much for your timely help in testing all these experiments.
Richard

[kultex]

rerwin,

it was a great pleasure and I / we have to thank you, because you did the job. And as soon as there is something to test, I will do it (I give you my mail with pn)
kultex

[rerwin]

kultex,
Thank you for your offer. It has been very rewarding working with you. And I have something more to test.

I am attaching patch5, which is actually patch4 with minor adjustments in preparation for submission to Barry for the 4.3 beta2 and to go into 3G_pupdial-12. I have shortened the wait times slightly and removed some debug options to adjust the times. Now it has 5 seconds for the usb-storage driver to get it together before the mode switch, and four seconds for modeswitch to wait before checking for success and proceeding to modprobe the option driver (because it probably did not see the modem the first time it was modprobed normally). I deleted 3 seconds total.

I need you to try that patch as with patch4, to verify that the modem still works the same. I was generous with the delay times as we were getting started, but now I want to trim out the excess. The main reason is that the initialization script was waiting up to a minute for the driver to get loaded. While the driver apparently loads quickly, if the modem is removed, each boot-up without it will have the script wait the maximum time, which I have now reduced to 15 seconds, compared to the 9 seconds for the modeswitch-modprobe process. That's round-about logic, but there it is.

Thinking ahead to attempting an automated way to determine which ttyUSB number to use, because a few modems apparently use ttyUSB2 instead of the usual ttyUSB0. And there may be other such cases. I have added logging of udev information of several kinds, to see if there is any data that might help. I do this because the hso driver seems to provide a way to determine the ttyHS#, which is already inside the hso initialization script. The logging goes the /tmp/udevtrace-modem.log. so that is the only file I need to see. There may not be anything useful, but I need to rule out that possibility.

I hope a few people with the modems using ttyUSB2 will run thepatch and post the log file. Thanks for any help.
Richard

[kultex]

uninstalled patch 4, installed patch 5 - still everything perfect and much quicker ....
added 2 more files - e220-31 is plugging in at boot and e220-32 is plugging in afterwards.

[rerwin]

kultex,
Thanks for the good news! But I am puzzled by the fact that none of my debug info from the initialization script got logged. The installation process should have reset a value to cause the updated script to get placed where it executes.

Could you check the size of /etc/init.d/Option to verify it is "2776 B"? If it is not, try a run after editing a file entry to accomplish that? The file is /etc/modules/firmware.dep.2.6.25.16 (or whatever kernel you are using). There is an entry:
#option:option.ko
Just remove the #, save, reboot.
That should reload the "firmware", which is the script and maybe other files.

But in looking at the script, I suspect something else. It now waits 15 seconds for the driver to load, and exits if it hasn't by then. That is the simplest explanation for the absence of the debug info. You might edit that file to change line 12 from:
WAITMAX=15
to
WAITMAX=60
and check for debug data in /tmp/udevtrace-modem.log.

If the data is still absent, there may be a bigger issue. To be sure the script runs and selects the correct modem (ttyUSB0), please probe>ERASE in PupDial and reboot, to be sure puppy is not re-using the selection from before. If no modem is detected, that tells me that the script did not run or that it timed out waiting for the driver to load.

My expectation is that the line:

Code: Select all

usb_v12d1_p1003_d0000_dc00_icff_usb_interface_HUAWEI Technologies_HUAWEI Mobile_

should cause the driver to be loaded and set up the init script, but that the driver does not recognize the modem because the switching has not completed. The new additional modprobe then reattempts loading, and the driver does find the modem. I assume that the failure to find the modem the first time aborts the loading (but I am unsure of that). Even so, the second modprobe should complete within the 15 seconds.

Perhaps a cleaner test would be to simply make the edit to firmware.dep and delete the init script altogether. That would force re-installation of the script. If no modem is detected, with the modem plugged in at boot, that would mean either the script was not present where needed, or it timed out. Check for the existence of /etc/init.d/Option. I need to know whether the script makes it to /etc/init.d. If it is not, we need a redesign. Enough rambling. Please see what you can find. Thanks.
Richard

UPDATE: I am trying to imagine what might be happening with the modem startup. What if the device issues all of its device identification at once and that the presence of the entries containing icFF does not indicate that the device has switched modes? That should cause the modem driver to be modprobed, but maybe the driver does something special when it finds the modem not actually ready. On the chance that it refuses to load in that situation, maybe the modprobe instruction produces an error or status message. Although I have edited the modprobe in the script that issues it, to send errors to our log file, I may have coded that incorrectly. I have a better way to code that, so you might edit file /sbin/pup_event_backend_modprobe, line 151, to:

Code: Select all

exec /sbin/modprobe $MODULE >> /tmp/udevtrace-modem.log 2>&1   #rse

by changing the "2>>" to ">>" and inserting the "2>&1".
Richard

[droope]

Hi rerwin,

I am sorry, I must report failure And, also, I have lost access to the modem for a while (maybe a few months.) I've, however, managed to snatch the logs from the tmp folder, just in time:



Here I attach the rar, hope it is useful. I will not be able to test for some time, though!

Cheers

[rerwin]

droope,
You do not indicate that you used the "Internet by dialup modem" option, but that is immaterial, judging by the content of the logs you attached. I see no indication that the modem was plugged in at all during that session. I am looking for vendor 19d2 and product ID 2000. The only USB device indicated is a Kingston DataTraveler 2.0. From that I conclude that you had only a flash drive plugged, and not the modem. The only indication there was ever a modem plugged in is the line: "Script Option: timeout", which comes from the leftover initialization script.

If that DataTraveler is actually the modem, that is a "whole new ballgame". I guess we will have to wait until you are ready, to resume experimentation. Just let me know when.
Richard

[kultex]

rerwin,

I am also a bit confused:

there is no file: /etc/init.d/Option

There is an entry: option:option.ko, but without # - so there is nothing to do

in which script I shall change the time?

I erased the Modem - and it is not found by the system - even with probe, it does not found.

But it is present - here dmesg.

Code: Select all

usbcore: registered new interface driver usbserial
drivers/usb/serial/usb-serial.c: USB Serial support registered for generic
usbcore: registered new interface driver usbserial_generic
drivers/usb/serial/usb-serial.c: USB Serial Driver core
drivers/usb/serial/usb-serial.c: USB Serial support registered for GSM modem (1-port)
option 1-2:1.0: GSM modem (1-port) converter detected
usb 1-2: GSM modem (1-port) converter now attached to ttyUSB0
option 1-2:1.1: GSM modem (1-port) converter detected
usb 1-2: GSM modem (1-port) converter now attached to ttyUSB1
usbcore: registered new interface driver option
drivers/usb/serial/option.c: USB Driver for GSM modems: v0.7.1 

only if I edit by hand wvdial.conf the modem to /dev/ttyUSB0 it is detected and I can connect.

I hope this helps a little bit .....

[rerwin]

kultex,
Thanks for running that test. It confirms my fear that we have not actually mastered the startup timing. The logs show that the mode-switch took place and that the option driver was probably loaded by our second modprobe. It might be that a second bootup would take, without touching wvdial.conf.

I am concerned that we don't see confirmation that the original modprobe of the option driver was actually done, although the script probably got placed where it should be. (because that happens before the actual modprobe). That's been nagging me for a while.

I know it is late for you tonight, so I will later add suggestions and maybe an updated patch5 for you to try at your convenience. But I will go ahead and submit something for the pending 4.3 beta and hope to update that once we figure this out.

Thanks for continuing with this. I know you are going to be busy for awhile, so please work this only when you can spare the time. There is no deadline for getting it right.
Richard

[divisionmd]

Hello,

- Going bananas over my 3G device - spent alot of hours and now i am stuck.

- I have a 3G modem "Option Icon 225".

- And i have gotten so far that when i try to connect this happens:

--> WvDial: Internet dialer version 1.53
--> Initializing modem.
--> Sending: AT+CPIN=1234
AT+CPIN=1234
OK
--> Sending: ATZ
ATZ
OK
--> Modem initialized.
--> Sending: ATX1DT*99***1#
--> Waiting for carrier.


But nothing more happens... ?


- This is the log:

- /tmp/pup_event_module_devpath_log:

ODULE=hso DEVPATH=/devices/pci0000:00/0000:00:1d.0/usb3/3-1/3-1:1.0
# MODULE=fan DEVPATH=/devices/LNXSYSTM:00/LNXTHERM:00/PNP0C0B:00
# MODULE=usbhid DEVPATH=/devices/pci0000:00/0000:00:1d.0/usb3/3-2/3-2:1.0
# MODULE=usbhid DEVPATH=/devices/pci0000:00/0000:00:1d.0/usb3/3-2/3-2:1.1
# MODULE=evdev DEVPATH=/devices/platform/pcspkr/input/input2
# MODULE=evdev DEVPATH=/devices/LNXSYSTM:00/device:00/PNP0C0C:00/input/input4
# MODULE=evdev DEVPATH=/devices/LNXSYSTM:00/LNXPWRBN:00/input/input3
# MODULE=hso DEVPATH=/devices/pci0000:00/0000:00:1d.0/usb3/3-1/3-1:1.0


- /tmp/udevtrace-modem.log:

usb_v0af0_p6971_d0000_dcff_icff_usb_interface_Option N.V._Globetrotter HSDPA Modem_
tty__Add_device_node_link_for_HSOTYPE_Diagnostic
usb_v0af0_p6971_d0000_dcff_icff_usb_interface_Option N.V._Globetrotter HSDPA Modem_
tty_ttyHS2
tty__Add_device_node_link_for_HSOTYPE_Control
tty__Add_device_node_link_for_HSOTYPE_Application
tty_ttyHS1
tty_ttyHS0


- /tmp/udevtrace.log:

add_ttyHS2_tty_0x8086_3-1:1.1_usb:v0AF0p6971d0000dcFFdscFFdpFFicFFiscFFipFF
add_hso0_net_0x8086_3-1:1.0_usb:v0AF0p6971d0000dcFFdscFFdpFFicFFiscFFipFF
add_ttyHS1_tty_0x8086_3-1:1.0_usb:v0AF0p6971d0000dcFFdscFFdpFFicFFiscFFipFF
add_ttyHS0_tty_0x8086_3-1:1.0_usb:v0AF0p6971d0000dcFFdscFFdpFFicFFiscFFipFF
add_usbdev3.4_ep03_usb_endpoint_0x8086_3-1:1.0_usb:v0AF0p6971d0000dcFFdscFFdpFFicFFiscFFipFF
add_usbdev3.4_ep85_usb_endpoint_0x8086_3-1:1.0_usb:v0AF0p6971d0000dcFFdscFFdpFFicFFiscFFipFF
add_usbdev3.4_ep83_usb_endpoint_0x8086_3-1:1.0_usb:v0AF0p6971d0000dcFFdscFFdpFFicFFiscFFipFF


- /tmp/ozerocdoff.log:

/usr/sbin/ozerocdoff -wi 0x6971 -l /tmp/ozerocdoff.log
Successfully ZERO-CD disabled


- Before i connect i execute this:


modprobe hso
modprobe option
modprobe usbserial vendor=0x0af0 product=0x6971

echo "Echo PIN code ttyHS0"
echo AT+CPIN=1234 > /dev/ttyHS0
sleep 1
echo "AT+CPIN=1234" > /dev/ttyHS0
sleep 1

echo "Echo PIN code ttyHS2"
echo AT+CPIN=1234> /dev/ttyHS2
sleep 1
echo "AT+CPIN=1234" > /dev/ttyHS2
sleep 1

usb_modeswitch

then i get the message that the modem is switched.


- Modem works fine on Windows.

- Also tested to type "ATZ" to init string "2" but that did not help.


Thanks for all help,

Best regards,
Johan

[alix_board]

Hi @ all,

i have a Huawei E170 Wireless Modem and have already install the 3G pupdial wireless patch4. I do not have many experience with puppy, still !

How i can install now the modem? I have already try the Network Wizard (Autoload USB) - but it does not work. Maybe someone can give me an instruction?

Thanks a lot!

greets,
alix_board

[rerwin]

alix_board,
You have come to the right place to request help. I see that you earlier posted your information here: http://www.murga-linux.com/puppy/viewto ... 527#332131
That gives me something to work with.

The first problem is that the wireless modem support began on kernel 2.6.25.16. I have not tested on the "retro" 2.6.21.7 kernel and suspect that there is no "option" driver there. I have compiled updated drivers for the 2.6.25.16 kernel, but they won't help you until you upgrade to the "standard" 2.6.25.16 kernel.

Once you are on Puppy 4.1.2 or 4.2.1 "standard", please review the first post in this thread, to understand what to do. To summarize that, you need to download the "3G pupdial", "usb_modeswitch", patched option driver, patched usb-storage driver and 3G...patch4 dotpets mentioned there.

But be warned: we are having difficulties with the Huawei modems being recognized. Please reboot twice after installation, to give it a chance. If it doesn't work after you have done everything right, I will need your help/feedback to get the problem corrected. Be sure to have the modem plugged in when you boot up. You may need to unplug and replug it after you are up and running after two bootups.

Thanks for joining us.
Richard