Sergio Freire a Ãcrit : > Duncan and all, > if i use the kernel driver, besides the "-k" switch in modem_run, what > else should i do? > I dont need to run pppoa3 since im using ATM on the kernel or should i > use it also? > Currently im using the following script to connect: > --------------------------------------------------- > > #!/bin/bash > > rmmod uhci-hcd > umount /proc/bus/usb > > modprobe usbcore > modprobe uhci-hcd > modprobe ehci-hcd > > killall adsl-connect > killall pppoa3 > killall pppd > sleep 5 > killall -9 adsl-connect > killall -9 pppoa3 > killall -9 pppd > modprobe tun > mount -t usbdevfs none /proc/bus/usb > echo Loading firmware... Please wait > /usr/local/sbin/modem_run -m -f /usr/lib/mgmt.o > /usr/local/sbin/pppoa3 -b -e 1 -m 1 -c -vpi 0 -vci 35 > echo Dialing ... > /sbin/adsl-start > route add default ppp0 > echo To stop the connection use adsl-stop... > echo To start the connection after stop, use adsl-start
The 1.3 release of the usermode driver introduce a new feature that can if not properly configured cause a lot of problem : hotplug. Currently, each time you plug the modem or you load the USB modules or whatever (each time the system discover that the modem has been connected), the speedtouch-start script is run THREE times (it's a bug in hotplug). To solve the problem of knowing if the firmware has been loaded or not, modem_run has been modified to have the following behaviour: case 1: firmware not loaded -> it loads firmware -> it waits for synch (30 s approx.) -> exit code is 0 case 2: firmware is loaded but no synch -> it exits immediately with exit code -1 case 3: firmware is loaded and synch is OK -> it exists immediately with exit code 0. This is achieve using a mutex library (mutex.c) which is implemented using IPC semaphore, to ensure that if you run several modem_run at the same time, one and only one will load the firmware into the modem. So, from that, there are two known issues: - your script (modprobe -r uhci-hcd, modprobe uhci-hcd) doest not take care of all the hotplug generates. That's why you got disconnected later (maybe the -c option of pppoa3 is not properly used). - under some circonstance, the mutex stay in memory and still indicate that the firmware has been loaded.... where it is not. I would say that it's not easy to maintain the "state" of the modem witch accuracy in user mode since process can be killed (with kill -9) and since USB events are difficult to track. Maybe we should find a more easier solution with kernel mode driver (even if we still need one for usermode driver). Benoit PAPILLAULT Liste de diffusion modem ALCATEL SpeedTouch USB Pour se désinscrire : mailto:[EMAIL PROTECTED]
