On Fri, 25 Oct 2002, Sam Griffiths-Jones wrote: > > Hi all > > I remember seeing something about this before, but the mail archive > seems to be having trouble -- I can't read messages older than a > couple of weeks. > > Tim's successful fix of the speedtouch 330 thing has exposed the next > problem. After only a couple of minutes connection time I get: > > Oct 24 23:54:44 lucy pppd[2365]: Child process /usr/bin/pppoa3 -v 2 -c > -m 1 -vpi 0 -vci 38 (pid 2368) terminated with signal 11 > > in my /var/log/messages. pppd stays active, and ifconfig reports > link still up, but I can't send/recieve anything. When I run the > sysv init script to stop the connection I get: > > Shutting down ADSL connection:n_hdlc: Device or resource busy > > And I can't rmmod n_hdlc for the same reason. Unplugging the modem > and starting again doesn't sort it either so the only option that I've > found to work is a reboot. Of course I only get a 2 minute connection > the next time round aswell. > > Expanded syslog follows. Any ideas? > > Sam > <snip> > > <IP stuff cut> > > Oct 24 23:54:16 lucy pppoa3[2374]: Extracted PPP packet sent to ppp(d) > Oct 24 23:54:16 lucy pppoa3[2373]: ATM cell queue sent to USB > <..... x lots> > > Oct 24 23:54:44 lucy pppd[2365]: Child process /usr/bin/pppoa3 -v 2 -c > -m 1 -vpi 0 -vci 38 (pid 2368) terminated with signal 11 > You haven't quite got enough logging here to be sure but it might be the OAM cell at the end of a frame problem or a buffer overflow in the frame.
The latest CVS version has a fix for both of these. (There is a strange oddity in the logging code - I'm not sure if its a bug or by design. - If REPORT_DUMP is NOT set then the log level is always LOG_INFO. If REPORT_DUMP is set then the log level depends on whether REPORT_ERROR, REPORT_INFO or REPORT_DEBUG is set.) To get the logging needed to check whether this is the bug outlined above you need to enable LOG_DEBUG logging in syslog. Hopefully you will see "AAL5 frame joined up (0 bytes long)" in the log just before the crash. (Or "Buffer overflow, too many cells for the same aal5 frame") Signal 11 is SIGSEGV which means that pppoa3 is writing to illegal memory. If the latest CVS (plus my patch) doesn't fix it can you enable maximum logging and then recreate the crash and then post the last 20 or so lines of the log before the "signal 11" line. Hopefully this will give a clue. (The log will get pretty big pretty quickly and the crash may not happen because things are running more slowly.) Regards, Tim. > > > > Liste de diffusion modem ALCATEL SpeedTouch USB > Pour se désinscrire : mailto:speedtouch-request@;ml.free.fr?subject=unsubscribe > > > -- God said, "div D = rho, div B = 0, curl E = - @B/@t, curl H = J + @D/@t," and there was light. http://tjw.hn.org/ http://www.locofungus.btinternet.co.uk/ Liste de diffusion modem ALCATEL SpeedTouch USB Pour se désinscrire : mailto:speedtouch-request@;ml.free.fr?subject=unsubscribe
