[speedtouch] Re: Speedtouch 1.3.1 not working on Fedora Core 2

2004-10-17 Thread Clive Messer

Olivier,

Thanks for helping. The locked semaphore was my problem. Easy now I know
what I am looking for. I agree with you - it would be very nice if one
of the developers could add a modem_run option to remove the semaphore
and automatically remove it if loading the firmware fails.

Regards

Clive
-- 
Clive Messer
www.vacuumtube.org.uk

On Sat, 2004-10-16 at 16:46, Olivier Blin wrote:

 Hi,
 
 From speedtouch-1.3.1, modem_run locks a semaphore when it's run.
 So, if the first run fails, you won't be able to easily run modem_run
 again.
 Duncan posted a tip to remove the semaphore: get the id from
 /proc/sysvipc/sem, and try to remove it with ipcrm sem id.
 It would be nice to have a modem_run option to remove the semaphore,
 and to remove it if the firmware loading fails.
 
 Have you set the correct loader and firmware paths in speedtouch.conf
 ? If they're wrong, then modem_run will fail the first time and will
 be locked.
 You can add some debug options in the hotplug script (probably
 /etc/hotplug/usb/speedtouch) to see what happens, for example :
 exec  /dev/tty1 (or exec  /some-writable-file)
 set -x
 
 In Mandrakelinux, we got some other problems at boot because the usb
 controller triggers speedtch modules loading very early, and the
 hotplug script failed because it tried to set up a remover script, but
 /var was read-only at this point.
 
 If you fail to see what happens, then you can disable the hotplug
 script (move it away from /etc/hotplug/usb) and try to use modem_run
 once the system is booted.
 
 Regards


Liste de diffusion modem ALCATEL SpeedTouch USB
Pour se désinscrire : mailto:[EMAIL PROTECTED]




[speedtouch] Re: Speedtouch 1.3.1 not working on Fedora Core 2

2004-10-17 Thread Olivier Blin

Clive Messer [EMAIL PROTECTED] writes:

 Olivier,

 Thanks for helping. The locked semaphore was my problem. Easy now I know
 what I am looking for.

Actually, the real problem is that the first firmware loading fails.
Have you set the correct paths in speedtouch.conf ?

-- 
Olivier Blin


Liste de diffusion modem ALCATEL SpeedTouch USB
Pour se désinscrire : mailto:[EMAIL PROTECTED]




[speedtouch] Re: Speedtouch 1.3.1 not working on Fedora Core 2

2004-10-17 Thread Benoit PAPILLAULT
Clive Messer a crit :
 Olivier,
 
 Thanks for helping. The locked semaphore was my problem. Easy now I know
 what I am looking for. I agree with you - it would be very nice if one
 of the developers could add a modem_run option to remove the semaphore
 and automatically remove it if loading the firmware fails.
 
 Regards
 
 Clive

There are several problems in fact:

- the semaphore has been designed to be kept in memory, that might be 
the first problem.

- some errors in modem_run leaves the semaphore with an incorrect value, 
preventing modem_run to be run again. That's the second problem.

Benoit PAPILLAULT


Liste de diffusion modem ALCATEL SpeedTouch USB
Pour se dsinscrire : mailto:[EMAIL PROTECTED]




[speedtouch] Re: Speedtouch 1.3.1 not working on Fedora Core 2

2004-10-17 Thread Olivier Blin

Clive Messer [EMAIL PROTECTED] writes:

 Yes, the paths were/are correct. Have not changed them. Just removed the
 semaphore and 'speedtouch-start'. Need to check what happens when I
 reboot the box and the modem is detected by hotplug.

Maybe your usbfs isn't mounted at this point.
This happened in our Mandrakelinux package too.
The usb controller is loaded very early, it triggers the speedtch
module loading and the hotplug script, but /proc/bus/usb may not be
available yet, then modem_run will fail and keep the line_ko value
in the semaphore.
You can do some pretty evil hack before speedtouch-start is run in the
hotplug script, like this one :
while [ ! -f /proc/bus/usb/devices ]; do sleep 2; done

-- 
Olivier Blin


Liste de diffusion modem ALCATEL SpeedTouch USB
Pour se désinscrire : mailto:[EMAIL PROTECTED]




[speedtouch] Re: Speedtouch 1.3.1 not working on Fedora Core 2

2004-10-17 Thread Olivier Blin

Benoit PAPILLAULT [EMAIL PROTECTED] writes:

 There are several problems in fact:

 - the semaphore has been designed to be kept in memory, that might be 
 the first problem.

 - some errors in modem_run leaves the semaphore with an incorrect value, 
 preventing modem_run to be run again. That's the second problem.

Maybe modem_run could re-initialize the semaphore back to S_INIT if
the firmware loading fails (or if any other error occurs). Then next
modem_run instances will be able to try again.

-- 
Olivier Blin


Liste de diffusion modem ALCATEL SpeedTouch USB
Pour se désinscrire : mailto:[EMAIL PROTECTED]