PPP module

I have been able to do basic internet/mail at home by using a cellphone tethered to a linux machine with a shared connection.
I have been studying !ComCentre in the hope of using either my phone or a Sim card stick from RISCOS, but the required PPP 1.42 module is so well hidden, Google or ROOL search finds nothing. Does anyone know why?

Found, PPP 1.42 on the Iyonix, a temporary download is here

Found, PPP 1.42 on the Iyonix, a temporary download is here

Thank you for your support.
Has anyone got a SurfStick working with ComCenter? I am interested in what the contents of the Internet configuration in Boot.Choices.Internet would look like now.
The linux connection for these modems shows information for a connection address that changes on every connection.
There is a fixed default route address of 10.64.64.64 each time. Thankfully the dns address’s given stay the same also.

Using usbmon/wireshark, I was able to gather the info needed for the PPP interface and the AT command interface to build the definition file. I am up against a CONNECTION: error now, which looks like Iyonix side network settings to me, as I seem to have got past hardware definition errors.

I’m using ComCentre with a ZTE Surfstick and with a old Nokia 6300.
My LTE Stick and my Smartphone are not working at this time.
On newer hardware ppp needs “alignment exceptions off” or ARMv5 mode on a Pi.
I mean no changes in Boot.choices.Internet. All in Boot.Choices.ComCentre.
Use you a Surfstick ore a Phone? The most Surfstick needs to be swappt to the ModemMode. Use USBDevSwp.
Any Sticks you can swap via dismount the CDROM (Many Sticks are a CDROM with WIN driver at first.).
I have my hardware not here. But if you start ComCentre you will have a new APP in !Boot.Configure. If the Stick or what ever is correct initialize, you find the settings here and in !Boot.Choices.ComCentre.

My phone (Samsung Galaxy S3) doesn’t work either. It ought to, in principle (no conflict with mass storage; USB tethering enabled in the phone). I can see that there are support files for other devices although the only phone appears to be the Nokia 6303. I tried copying and editing my own file for my phone, but there are three short fields in there that I don’t understand and don’t appear to be documented. A little random experimentation produced 100% failures. Oh well.

The solution that I use with that phone, that definitely does work, is to enable the phone’s mobile hotspot and use a small cheap USB-powered Ethernet wifi dongle.

USB tethering is not the same standard like the old phones and surfstick use. Not the same USB modem emulation.
Not supported by ComCentre, I mean und not supported by all Linux modemmanager.
I have also a Nokia 3109c working. The device file is nearly the same for all Nokia I try. Only the Product- and Vendor_ID must be changed.
Look at USBDevSwp. Inside you find devicefiles of theoretical supported devices sorted by Vendor_Product_ID before swapping to the modemmode. The same files are using by Linux. Inside you find the Vendor_Product_ID after swapping. It is possible that the same Vendor_Product before use different Vendor_Product after swapping. USBDevSwp use the first entry in the “after swapping list” by default. Is your device not the first in the list, move it on top.
After swapping USBDevSwp set any things for ComCentre…

My phone already has CDC endpoints, according to !USBInfo. I don’t think it needs to be swapped like a surf stick.

I have an old surf stick too (its SIM expired years ago) which clearly does need to be swapped – !USBInfo sees it initially as a mass storage device. The phone doesn’t behave like that.

I just tried tethering the phone to Ubuntu (just works, nothing to do). lsusb tells me that the VID:PID are the same after “Wired connection 2” has been established as before (04E8:6864), though curiously the Samsung model number is now different!

Look at USBDevSwp. Inside you find devicefiles of theoretical supported devices sorted by Vendor_Product_ID before swapping to the modemmode.

My phone has selection for modem mode when inserted to USB port. My Sierra stick modem is not supported by USBDevSwp, but I have got my own Mode switcher that changes the 1199 0003 bulk device to a 1199 68a3 modem.
So no, I dont need USBDevSwp.

but there are three short fields in there that I don’t understand and don’t appear to be documented.

In my case I have isolated the PPP and AT interfaces, but I am left wondering if I have non standard endpoint numbers, and their is no input for endpoints.
Wireshark shows another interface being used and I dont know what for yet, It may be a Status channel, but I get errors when I enter it as such. Once again I am concerned over the endpoint numbers.

I have realised after watching Thomas’s video, that the first step may be to try an AT command from !ModTerm before worrying about internet.

My apologies – I had always looked at the VID:PID after enabling tethering on the phone. It’s 04E8:6860 with tethering disabled, and 04E8:6864 with it enabled. USBInfo sees the device differently too.

Ronald: the phone has interfaces 0 (interrupt in, 224:1:3, RNDIS), 1 (bulk in and out, 10:0:0, CDC Data) and 2 (bulk in and out, 255:66:1, vendor specific). Trying to align that with your explanation, does that mean the fields are 1, 1 and i0? The endpoint numbers are (3) for interface 0, (2 and 1) for interface 1, and (5 and 4) for interface 2.

The endpoint numbers are (3) for interface 0, (2 and 1) for interface 1, and (5 and 4) for interface 2.

With that info you can look at the output from usbsniff or usbmon/wireshark(linux) and look for the AT or PPP commands in the packet descriptions, then look for the endpoints on those packets. That is the perhaps superficial way I’ve done it so far. I’m not sure how you woud recognise a Status interface, looks like it is not always implemented. Deeper diagnosing may require having the cdc documentation at hand.
The other thing worth mentioning is that you can see the resulting strings from your description file with *Show Com*
Double clicking the device description seems to enable it (Or running it from !Run) but that is not official, The Docs only mention using ComCentre_Register blah blah USB, I found that this didn’t work unless I dropped the word USB off. It is mentioned in conjunction with !DevSwp so it is a little confusing.

You not need USBDevSwp if you use a phone or a stick that already is in modemmode.
Sorry, I have set up my sticks and phone in 2011 ;-) and I forgot my ComCentre copy at home :-(
I have hacked my Surfstick. So it is in modemmode after plugin. To use I add two lines in the ComCentre !Run.
I start the 19d2_0063 obey from !ComCentre.Devices and MMODEM_19d20063 from Choices.ComCentre.Devices…
The first was created by myself with the help of USBInfo and the second was created via configure.
The key is the first obey. As I remember… sorry long time ago…
… BasicSetup 19d20063 3 1 b1… (is from my ZTE, this parameter must be changed)
At first the right ID… and than…
First Number is the “USB Interface PPP access”
Second Number is “USB Interface normally AT-Modem access”
Third “Number” “USB Interface answer channel”
known options:
b – bulk channel
i – interrupt channel
? – without → normally not.

Normally somthing like 0 3 b3 (e.g. Huawei) or 3 1 b1 ( e.g.ZTE) with splitted channel or 0 0 b3 or 3 3 b1 with the same channel.

… play around, try and error ;-)

The key is the first obey. As I remember… sorry long time ago

Yes this will set the system variables for the interfaces. Unfortunately it is limited to Bulk interfaces, and if AutoDevGen VID PID USB is anything to go by the requirement is that the interfaces have to be of the class 255.255.255 which rules out cdc class interfaces such as Dave or my phone.
My stick modem is 255.255.255 and bulk interfaces so there may be hope for it one day. Currently AutoDevGen or ModTerm is not successful with getting an AT response.
Reading this stick’s (Sierra) manufactures support for linux states the interface is a standard 115200 8N1 and this certainly worked (AT responses) from a linux terminal on both the stick and the cdc phone.
In linux the stick has a /dev/tty/USBn for each 4 interfaces whereas the cdc driver for the phone only provides /dev/ttyACM0
For the stick the terminal will only connect to the interface for AT commands.
The CDC phones I think use the AT interface as master and the cdc data interface as a slave. Hence only one device name.

The first success I have with all trying devices was with CCT_Info. It take some time but it shows all aviable mobile networks. Maybe it is enough to try something like … BasicSetup 1199_68a3.
I found this for Linux usbmodeswitch …

#######################################################
# Sierra Wireless
#
# Contributor: Vincent Teoh
#
DefaultVendor= 0×1199
DefaultProduct= 0×0003
#
TargetVendor= 0×1199
TargetProduct= 0×68a3
#
SierraMode=1
#
########################################################

… but not really heplful.

I found this for Linux usbmodeswitch

Sorry I gave the false default PID, My RISCOS switcher switches the 1199 0fff stick to 1199 68a3 modem.
Yes it is similar to the linux usb modeswitch source for Sierra. That job appears to be done, thanks.

The route of the problem I’m having is this part failing.

openin%=0
openin%=OPENIN"Devices#nopad;interface3;alternate0;bulk;size2049:$.USB7"
IF openin%<>0 THEN PRINT OK

I know the “cant open pipe” error message. But I can not remember. I will search in the old mails from Thomas. But it takes a (long) while.
At the first time I often forget to “Enable TCP/IP Protocol Suite” in the Internet configuration. “ARM V5 mode” or “ae off” is necessary for PPP… The error messages are not always clear.

openin%=OPENIN"Devices#nopad;interface3;alternate0;bulk;size2049:$.USB7"

I’ve just tried something similar with a usb device and it works fine

BTW Stronghelp does not mention &4F, only &40 and some others.

The F just determines what happens when there is an error.

Are you sure that you have the changed mode working if the usb stack is still seeing the mass storage interface the command you used will fail.

BTW Stronghelp does not mention &4F, only &40 and some others.

Doesn’t OS_Find use the bottom bits of R0 to indicate the path if R0 isn’t zero?

Are you sure that you have the changed mode working if the usb stack is still seeing the mass storage interface the command you used will fail.

Yes the lights on the stick do a different dance after the modeswitch, so it is easy to see what mode it is in even without checking *USBDevices At switching it jumps from USB6 to USB7 with a different name for example.
I have modified (and renamed) version of your PROCusb_control so that takes an extra parameter for the device handle which I get from FNusb_findDeviceByVendor by including a global devh%=item%+8
It’s working fine. I now have
DEF PROCdevhusb_control(devh%,bmRType%,bRequest%,wValue%,wIndex%,wLength%,buf%)
so there is no mixup down the track.
The bit of Test BASIC

openin% = 0
openin%=OPENIN("Devices#nopad;interface0;alternate0;bulk;size2049:$.USB7")
IF openin% <> 0 THEN PRINT "Opened OK"
CLOSE# openin%
openin%=0
PRINT "Closed OK"

I have a serial device here with 2 bulk and an interrupt endpoint on interface 0 alternate 0 so just copied your program and changed the usb number and it worked fine.

Edit:

Have you tried !Closefiles to check if the endpoint is alread open?

Colin, my impression is that ComCentre is suited to FTDI style interfaces without the interrupt endpoint and the Sierra is more like the Prolific style that has an interrupt endpoint. Though I think you mentioned you dont have to do anything with the interrupt endpoint.
I dont know. I’ll try !Closefiles but that would be suggesting that RISCOS USB had opened the endpoint, it is not an endpoint in use error anyway.
Another thought I had was that the stick is running something internally that blocks the attempt to open it. There is apparently a basic standard way of driving the stick and then the manufacturers high performance ‘direct ip’ driver, which was available for Linux, and may now be part of the standard distribution. (unverified)
It does seem nippier than using the phone. It could be the HSPA vs UMTS though.
I’ve bought another prepay sim for the stick now and the phone can stay a phone.

It might help to send a mail to Thomas. In my experience he is very helpful.

It might help to send a mail to Thomas

Yes but at this stage the problem is peculiar to the device.
I have found a few varieties of sierra.c and there is a few more USB controls to try as well as the &0001 I am using to switchmode, there is power &0000, mode &0B and NMEA &O7. There are variations depending on the age of sierra.c but it wont take much to try them.
I’m still unsure if the direct ip drivers are working in linux, one doc reckons that #ipconfig should show a usb entry, but mine is showing a ppp entry. (along with the eth ones)

Raik said

I’m using ComCentre with a ZTE Surfstick and with a old Nokia 6300.
My LTE Stick and my Smartphone are not working at this time.

Here is a serial related pdf with zte modem as example

but the required PPP 1.42 module is so well hidden, Google or ROOL search finds nothing

I am having the same trouble trying to use !ComCentre

Can anybody point me to a copy of the PPP module? (for a Ras Pi 2.2)