Alcatel Speedtouch USB with 2.6 kernel guide [superceded]

[474]

Ok guys, the plot thickens ...

I found out something very interesting about ppp-2.4.2, something which has changed since the beta release. The good news is that the ip-up and ip-down scripts are now coded so that the DNS servers allocated upon a successful connection are automatically populated in /etc/resolv.conf. Furthermore, when the pppd process is terminated your original resolv.conf is restored

The bad news is that it sets the permissions for resolv.conf to 600 for the duration of the connection, which means that non-root users are unable to read it ... thus, non-root users are not able to resolve names

The simple fix is to adjust /etc/ppp/ip-up as follows (this excerpt begins on line 30):

Code: Select all

# backup the old configuration and install the new one
cp -a $REALRESOLVCONF $REALRESOLVCONF.pppd-backup
mv $REALRESOLVCONF.tmp $REALRESOLVCONF
chmod 644 $REALRESOLVCONF
fi

You can bet your bottom dollar that I'll file a bug on this (assuming one hasn't already been filed). I am sure I'm not the only person to have noticed.

No luck unfortunatly. I did as you instructed kerframil and the error message no longer appears but I am still unable to access the internet.

Has the above got anything to do with it by any chance? If not, then the suggestions I had are probably null and void as you say the error message that you were originally experiencing (specifically, the "LCP: timeout" message) has gone away. If so, then perhaps we should get in touch so that I can take a closer look at exactly what is happening on your system (feel free to get hold of me on your chat network of choice).

i wanted to try it out, but didn't come far

Elementary, my dear Watson ... try the following:

Code: Select all

# emerge sync
# ACCEPT_KEYWORDS="~x86" USE="atm" emerge ppp
# sed -i -e "s/2\.4\.2b3/2\.4\.2/" /etc/ppp/peers/<file>

... where <file> should obviously be substituted as necessary.

By the way, the reason I am suggesting that you sync and re-emerge is because the developer who committed ppp-2.4.2-r1 made a mistake when he first committed the ebuild to Portage. This mistake meant that the pppatm.so file was not copied into the right place (outside of the Portage sandbox) during the final merge stage

[Bubsy]

Great guide! I've followed it and everything works perfectly!
I run modem_run & pppd and I happily surf the net.
Untill.... I reboot.

pppd gives me this error msg:

Code: Select all

Kernel doesn't support ppp_generic - needed for PPPoATM

I have all necessary PPP kernel options set as modules. I'm sure ppp_generic is there.
To fix this, I need to re-emerge ppp-2.4.2-r1 (and copy pppoatm.so to it's correct loc.). After this, pppd runs without a problem! (Untill I reboot again and I have to re-emerge ppp again...)
As long as I don't reboot, I can start/stop pppd without a problem.

[gen536]

Hi ! I followed your great guide, uploaded the firmware, then when I run pppd : (the r1 ebuild, with copied pppoatm.so)

Code: Select all

sferbox root # pppd call adsl
Plugin /usr/lib/pppd/2.4.2/pppoatm.so loaded.
PPPoATM plugin_init
PPPoATM setdevname - remove unwanted options
PPPoATM setdevname_pppoatm - SUCCESS:0.38
pppd: In file /etc/ppp/peers/adsl: unrecognized option 'pty'

It complains on this line from peers/adsl : tty "/usr/sbin/pppoa3 -m ..."

I loaded every ppp* 2.6 module I could find, can someone tell me what's wrong ?

Thanks a LOT

[474]

Great guide! I've followed it and everything works perfectly!
I run modem_run & pppd and I happily surf the net.
Untill.... I reboot.

I am getting this strong sensation that you are using udev on your machine ...
Even if you are not, the problem can be solved as follows:

Code: Select all

# echo "mknod /dev/ppp c 108 0" >> /etc/conf.d/local.start

[474]

It complains on this line from peers/adsl : tty "/usr/sbin/pppoa3 -m ..."

Please re-read the guide, in particular the part about options. You should not be using the pppoa3 binary at all. That is a user-space implementation of PPPoA which should only be used in the case that you are not using the Alcatel kernel module and kernel ATM implementation. This thread is about 2.6 kernels in which case we must use the kernel implementations. A ppp plugin is used to interface with the kernel modules in question.

To clarify, here is my current file (/etc/ppp/peers/zen):

Code: Select all

defaultroute
holdoff 4
maxfail 25
persist
asyncmap 0
lcp-echo-interval  2
lcp-echo-failure   7
name ********@zen
user ********@zen
                                                                                
plugin /usr/lib/pppd/2.4.2/pppoatm.so
0.38

And my /etc/ppp/options, which are inherited by any peer connection definitions, I believe:

Code: Select all

noauth
usepeerdns
lock
noipdefault

[474]

.... I need to re-emerge ppp-2.4.2-r1 (and copy pppoatm.so to it's correct loc.)

The pppoatm.so should be in the correct location already. As I mentioned before, a developer screwed up when he initially bumped the ebuild to the -r1 release. That's been resolved now (here's the segment of code in the ebuild which was once missing):

Code: Select all

if [ "`use atm`" -a "`use x86`" ]; then
         dolib.so pppd/plugins/pppoatm.so
fi

So, the file should be placed under /usr/lib/pppd/2.4.2.

[Bubsy]

I am getting this strong sensation that you are using udev on your machine ...

You're right. I am using udev.

Even if you are not, the problem can be solved as follows:

Code: Select all

# echo "mknod /dev/ppp c 108 0" >> /etc/conf.d/local.start

That solves it!!!

Thanks again for the great guide and the effort!!!

[Qubax]

if [ "`use atm`" -a "`use x86`" ]; then
dolib.so pppd/plugins/pppoatm.so
fi

is there, but as said above, pppoatm.so can be found in /usr/lib/, an not in /usr/lib/pppd/...
so either you move the file into the correct pppd directory (but won't be deleted if unmerging) an use the settings aobe, or set the correct path in the /etc/pppd/peers/<file> to

Code: Select all

plugin /usr/lib/pppoatm.so

--------------

edit:
persist still not working, but a "connection without automatic reconnection" is better than no "connection with automatic reconnection"

[474]

is there, but as said above, pppoatm.so can be found in /usr/lib/, an not in /usr/lib/pppd/...

Ah, I see. Well, I think most of the issues concerning this topic appear to have been cleared up now. Time to update the guide ...

[rensi]

The simple fix is to adjust /etc/ppp/ip-up as follows (this excerpt begins on line 30):

Code: Select all

# backup the old configuration and install the new one
cp -a $REALRESOLVCONF $REALRESOLVCONF.pppd-backup
mv $REALRESOLVCONF.tmp $REALRESOLVCONF
chmod 644 $REALRESOLVCONF
fi

This is how my /etcp/ppp/ip-up looks now

Code: Select all

#!/bin/sh

# this is a script which is executed after connecting the ppp interface.
# look at man pppd for details

# the followings parameters are available:
# $1 = interface-name
# $2 = tty-device
# $3 = speed
# $4 = local-IP-address
# $5 = remote-IP-address
# $6 = ipparam

if [ "$USEPEERDNS" ]; then

	# add the server supplied DNS entries to /etc/resolv.conf
	# (taken from debian's 0000usepeerdns)

	# follow any symlink to find the real file
	REALRESOLVCONF=$(readlink --canonicalize /etc/resolv.conf)

	if [ "$REALRESOLVCONF" != "/etc/ppp/resolv.conf" ]; then

	# merge the new nameservers with the other options from the old configuration
	{
		grep --invert-match '^nameserver[[:space:]]' $REALRESOLVCONF
		cat /etc/ppp/resolv.conf
	} > $REALRESOLVCONF.tmp

	# backup the old configuration and install the new one
	cp -a $REALRESOLVCONF $REALRESOLVCONF.pppd-backup
	mv $REALRESOLVCONF.tmp $REALRESOLVCONF
	chmod 644 $REALRESOLVCONF
	fi

fi

[ -f /etc/init.d/firewall ] && /etc/init.d/firewall start

[ -f /etc/ppp/ip-up.local ] && . /etc/ppp/ip-up.local $1 $2 $3 $4 $5 $6

but I can't use as normal user
any ideas?

[474]

rensi,

I am not sure whether you are:

1) Initiating ppp as root, dropping your priveleges back to a normal user then experiencing the problem or ...
2) Attempting to start ppp as a non-root user.

In the first case, ensure that the ppp-up and ppp-down scripts have the executable bit set:

Code: Select all

chmod 755 /etc/ppp/ip-*

In the second case, refer to this page which explains how to safely allow ppp access for non-priveleged users.

In all cases, check that resolv.conf looks sane (and has suitable permissions applied) and see if you can ping a known host by IP rather than hostname.

[rensi]

I am not sure whether you are:

1) Initiating ppp as root, dropping your priveleges back to a normal user then experiencing the problem or ...
2) Attempting to start ppp as a non-root user.

It was the first case! But changing ther permissions of /etc/resolv.conf from 600 to 644 seems to have fixed it!

Thanks
rensi

[50cc]

I can't find the pppoatm.so file. I downloaded the modified ebuild file and ran ebuild. Then i emerged ppp. But what i don't understand is this, i ran ebuild with a ppp version 2.4.2_beta3-r1 but when i emerge ppp verion 2.4.2-r1 is installed. Is the ok??

Note: I am not a Gentoo or Linux expert.

Thanks from the Netherlands

[50cc]

Ok, i just found out that the /etc/make.conf file has been modified somewhere since i installed Gentoo a few weeks ago. I added 'apm' to the USE variable. I did emerge -C ppp and emerge ppp. Now i do have a pppoatm.so in /usr/lib/ppp/2.4.2b3.

Off topic:
Any ideas what could have changed my /etc/make.conf file?

On topic:
This may be a very stupid question: I don't have speedtouch modem where i am now, and so obviously it is not connected. But because i am curious how far i would get this time i did modprobe speedtch and ran pppd, no i got this error in my /var/log/messages:

Code: Select all

Mar 18 10:16:43 tux drivers/usb/core/usb.c: registered new driver speedtch
Mar 18 10:17:00 tux pppd[12654]: Plugin /usr/lib/pppd/2.4.2b3/pppoatm.so loaded.
Mar 18 10:17:00 tux pppd[12654]: PPPoATM plugin_init
Mar 18 10:17:00 tux pppd[12654]: PPPoATM setdevname - remove unwanted options
Mar 18 10:17:00 tux pppd[12654]: PPPoATM setdevname_pppoatm - SUCCESS:8.48
Mar 18 10:17:00 tux pppd[12655]: pppd 2.4.2b3 started by root, uid 0
Mar 18 10:17:00 tux pppd[12655]: connect(8.48): No such device
Mar 18 10:17:00 tux pppd[12655]: Exit.

The line containing "connect( 8.48 ): No such device" is an error i also got on a SuSE 9.0 machine with the same modem.

Is this because the modem in not connected or is it a different problem?

Thanks (again)

[474]

Ok, i just found out that the /etc/make.conf file has been modified somewhere since i installed Gentoo a few weeks ago. I added 'apm' to the USE variable. I did emerge -C ppp and emerge ppp. Now i do have a pppoatm.so in /usr/lib/ppp/2.4.2b3.

Are you sure it's resident in that path? That makes me fear that you are using the ebuild I created which has been superceded. Allow me to make that perfectly clear: ppp-2.4.2 in the official portage tree now fully supports PPPoATM and this is what everyone should be using. However, the atm USE flag must be active prior to compilation of ppp for this to be the case (I'm assuming you meant atm, not apm because the latter has absolutely nothing to do with this topic).

Off topic:
Any ideas what could have changed my /etc/make.conf file?

This cannot happen without your consent. Perhaps this is a result of injudicious use of the etc-update utility?

On topic:
This may be a very stupid question: I don't have speedtouch modem where i am now, and so obviously it is not connected. But because i am curious how far i would get this time i did modprobe speedtch and ran pppd, no i got this error in my /var/log/messages:

[snip]

Code: Select all

Mar 18 10:17:00 tux pppd[12655]: connect(8.48): No such device
Mar 18 10:17:00 tux pppd[12655]: Exit.

[snip]

The line containing "connect( 8.48 ): No such device" is an error i also got on a SuSE 9.0 machine with the same modem.

Well, you can certainly load the speedtch module without having the device in question attached. However, the modem_run binary will not execute successfully because the device does not exist. Consequently, the firmware will not be uploaded and the DSL line will not be established. When you run pppd then it is quite right for the ppoatm.so plugin to claim that no device is available.

[clampinus]

It works weeeee !

IIn my case, the two big issues I had to solve were :
- getting pppoatm.so to be built and find where it was : use "atm" use flag and use path /usr/lib/pppoatm.so instead of the one provided, as the plugin seems to be installed in the wrong place currently
- adapting the VPI/VCI values to my own provider settings. I found those by using the diagnostic tool under Windows.

Thanks for this guide Kerf !

[50cc]

Am getting a bit closer (i think) i tested the speedtouch again, and after running pppd i got this in my /var/log/messages file:

Code: Select all

Mar 22 17:31:41 tux pppd[6606]: Plugin /usr/lib/pppd/2.4.2b3/pppoatm.so loaded.
Mar 22 17:31:41 tux pppd[6606]: PPPoATM plugin_init
Mar 22 17:31:41 tux pppd[6606]: PPPoATM setdevname - remove unwanted options
Mar 22 17:31:41 tux pppd[6606]: PPPoATM setdevname_pppoatm - SUCCESS:8.48
Mar 22 17:31:41 tux pppd[6608]: pppd 2.4.2b3 started by root, uid 0
Mar 22 17:31:41 tux pppd[6608]: connect(8.48): Resource temporarily unavailable
Mar 22 17:31:41 tux pppd[6608]: Exit.

This was with the speedtouch attached of course.

Any ideas?

[50cc]

I was reading a different thread on this forum which was discussing a user which was using the wrong VCI/VPI values. But i noticed that he had something different in his logs:

Code: Select all

Dec 21 19:23:35 [pppd] pppd 2.4.2b3 started by root, uid 0 
Dec 21 19:23:35 [pppd] Using interface ppp0 
Dec 21 19:23:35 [pppd] Connect: ppp0 <--> 0.35

In the log above pppd connects to ppp0 but if i do

Code: Select all

bash-2.05b# ls /dev/ppp* -l
crw-------    1 root     root     108,   0 Jan  1  1970 /dev/ppp

I don't have a ppp0 file, is this my problem maybe?

[50cc]

I finally got it working!! Thankyou for this guide!

What i was doing wrong (i think) was not running modem_run each time.

Does anyone now howto edit the hotplug files so that modem_run, modprobe speedtouch and pppd are started after the modem is plugged in? This would be pretty cool.

Thanks again.

[Hi-Rider]

IMHO you have mixed up two drivers in your guide:
kernel and user space:
http://www.linux-usb.org/SpeedTouch/docs/howto.html

http://cvs.sourceforge.net/viewcvs.py/* ... l?rev=HEAD

But finally you use userspace driver.
Does anybody trying to use kernel space driver. Does it really stable now ?

[50cc]

IMHO you have mixed up two drivers in your guide:
kernel and user space:
http://www.linux-usb.org/SpeedTouch/docs/howto.html

http://cvs.sourceforge.net/viewcvs.py/* ... l?rev=HEAD

But finally you use userspace driver.
Does anybody trying to use kernel space driver. Does it really stable now ?

I think your wrong, all the drivers come from the kernel all that's used from the userland files is the modem_run utility.

I'm busy converting a SuSE 9.0 maching to a Gentoo maching at this very moment. It wil be used as a fileserver/printserver/NAT/DHCP/firewall machine.

It will be connected to the Speedtouch, i'm hopping it will be stable. (the speedtouch part that is)

[Gentii]

Does anyone know what has changed with mm-sources 2.6.5 ?
I head about some hotplug fixes. But the speedtouch usb doesn't work anymore with these sources... I can start modem_run, then my modem is synchronized but modem_run doesn't stop. Even with ctrl+c it wont stop.
Is it possible to have this modem working on last mm-sources?

[OneOfOne]

I am getting this strong sensation that you are using udev on your machine ...

You're right. I am using udev.

Even if you are not, the problem can be solved as follows:

Code: Select all

# echo "mknod /dev/ppp c 108 0" >> /etc/conf.d/local.start

That solves it!!!

Thanks again for the great guide and the effort!!!

which kernel is that? afaik ppp sysfs patch went into main stream a while ago..

Does anyone know what has changed with mm-sources 2.6.5 ?
I head about some hotplug fixes. But the speedtouch usb doesn't work anymore with these sources... I can start modem_run, then my modem is synchronized but modem_run doesn't stop. Even with ctrl+c it wont stop.
Is it possible to have this modem working on last mm-sources?

it's the usb updates in it, if you dont depend on any of the major usb updates in it do this : cd /usr/src/linux-2.6.5-mm2 && wget http://www.kernel.org/pub/linux/kernel/ ... -usb.patch -q -O - | patch -p1 -R and recompile.
that *should* fix modem_run problems.

peace

[Azrael3000]

i followed your tutorial and compiled the kernel. all modules were working as supposed. then I started to emerge the whole stuff but the after another reboot the pppoatm module didn't want to start

dmesg says:

Code: Select all

pppoatm: Unknown symbol register_atm_ioctl 
pppoatm: Unknown symbol deregister_atm_ioctl

I disabled the pppoatm module in the kernel and enabled it again. While compiling I got the following message:

Code: Select all

Warning: /lib/modules/2.6.4-rc1/kernel/net/atm/pppoatm.ko needsunknown symbol register_atm_ioctl
Warning: /lib/modules/2.6.4-rc1/kernel/net/atm/pppoatm.ko needsunknown symbol deregister_atm_ioctl

thanks for your help

[danmilkman]

Followed the manual, but can't get it to work at all. Fails to load microcode, USBDEVFS_BULK errors?

What am I doin' wrong?

Help pleaze.