Thanks for your feedback.
Benoit PAPILLAULT wrote:
> <>Hi Frédéric!
>
> RUAUDEL Frédéric a écrit :
*snip*
> <>
> 1) speedtouch_clean_mutex.patch
>
> This patch fix a problem with the new mutex system of modem_run. When
> you stop modem_run, the mutex is not cleaned so when you try to start it
> up again, modem_run just print in /var/log/messages :
>
>
>You undo my change in a way. The mutex is designed to be kept in memory
>forever (you can remove it if the modem is unplugged, but this currently
>lead to some race conditions, so it's never removed). If you remove it,
>speedtouch-start will not work properly (if your stop the connection
>with speedtouch-stop and start it again using speedtouch-start, the
>mutex prevents modem_run from loading the firmware again).
>
>
So, it's a feature ;-)
Maybe it could be helpful to had a message indicating that modem_run
found that the firmware was already loaded and skip its work. After
reflexion, I maybe found this solution following a false problem,
because my first RPM have replaced my peer file so my login wasn't set
and the connection failed. As I didn't notice that first, maybe I was
disturbed by this behavior. If I have time, I will try to investigate on
this problem to see if it was a false issue or not.
But my solution works thanks to the redundant patch :)
>>modem_run version 1.3 started by root uid 0
>>
>>and then stop without any clue ! After looking at the code I found that
>>it was the mutex "deadbeef" ;-)
>>So I had an ipcrm in the speedtouch-start script but this is a dirty
>>hack and I think, if this is a bug and not a feature, that benoit will
>>certainly find a better solution :)
>>
>>
>
>What is important is the exit code from modem_run, 0 indicates that the
>line is OK and -1 indicates that there is some problem.
>
>
>
>>2) speedtouch_check_firmware_load.patch
>>
>>this patch check the state of the ADSL line present in the file
>>/proc/net/atm/speedtch*, and if the line is reported up and the firmware
>>loaded, it skip the launch of modem_run and directly launch pppd instead
>>of failing the load of the firmware and quit.
>>
>>
>
>Redundant with the modem_run's mutex.
>
>
>
>>3) speedtouch_fix_pppd_pid.patch
>>
>>the last one is probably FC2 specific or maybe a missing configuration
>>in my pppd package, but my pppd put its PID in /var/run/ppp0.pid and not
>>in /var/run/ppp-${PPPD_PEER}.pid. So I made this patch to make it work
>>with my current RPMS.
>>Feel free to tell me where I'm wrong, if I am.
>>
>>
>
>Two pid files are generated by pppd:
>/var/run/pppN.pid
>/var/run/ppp-${PPPD_PEER}.pid [only if you use the "link" option in your
>peer file).
>
>The later is easier to locate if you have multiple PPP connections at
>the same time.
>
>
>
I didn't know this link option that's why I made this patch, but I was
conscious of the problem with multiple PPP connections. So you just had
to put 'link' alone on a new line in the peer file to make it works ?
If its the case I will fix my RPM ASAP.
*snip*
>Benoit PAPILLAULT
>
>Liste de diffusion modem ALCATEL SpeedTouch USB
>Pour se désinscrire : mailto:[EMAIL PROTECTED]
>
>
>
>
Fred
Liste de diffusion modem ALCATEL SpeedTouch USB
Pour se désinscrire : mailto:[EMAIL PROTECTED]
