The idea is solid, but the logic is broken - I think you should instead test for disable_crtscts earlier and only turn on that bit on the tios.c_cflag if it's turned off (a 'not' test).We have an old Ericsson GM12 with a custom-made serial cable that does not have the control signals wired. When running Kannel 1.2.0, no communications occurs. Turning off hardware flow control helps.Attached is a patch against Kannel 1.2.0 that add an extra directive to the modems group (disable-crtscts). Might help some people. Regards, Eric Yeo
Only in gateway-1.2.0-ttey/doc/alligata: alligata.tex Only in gateway-1.2.0-ttey/doc/alligata: alligata.tmp diff -r -c gateway-1.2.0/gw/smsc_at2.c gateway-1.2.0-ttey/gw/smsc_at2.c *** gateway-1.2.0/gw/smsc_at2.c Tue May 14 09:57:36 2002 --- gateway-1.2.0-ttey/gw/smsc_at2.c Fri Jul 12 15:01:54 2002 *************** *** 82,87 **** --- 82,91 ---- tios.c_cflag |= CRTSCTS; /* enable hardware flow control */ tios.c_cc[VSUSP] = 0; /* otherwhise we can not send CTRL Z */ + if (privdata->modem->disable_crtscts) { + tios.c_cflag &= ~CRTSCTS; /* disable hardware flow control */ + } +
or better : have a use_crtscts in the modem data structure, turn the bit on only if it is set, and on initialization, init the use_crtscts flag to true and set it to false only if "disable-crtscts" is set in the configuration.
-- Oded Arbel m-Wise mobile solutions
