Oliver Neukum <oliver@xxxxxxxxxx> writes: > Am Freitag 27 Februar 2009 10:55:47 schrieb Bjørn Mork: >> This patch adds a dummy implementation of the TCGETS and TCSETS ioctls >> to cdc-wdm. This is necessary for userspace applications like "chat". > > How common is that? I have no statistics but have briefly tested some applications found in Debian stable by stracing them (haven't yet looked at the source to verify these results): bjorn@nemi:~$ dpkg -l ppp wvdial gsm-utils Desired=Unknown/Install/Remove/Purge/Hold | Status=Not/Inst/Cfg-files/Unpacked/Failed-cfg/Half-inst/trig-aWait/Trig-pend |/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad) ||/ Name Version Description +++-===========================-===========================-======================================= ii gsm-utils 1.10-13 GSM mobile phone access applications ii ppp 2.4.4rel-10.1 Point-to-Point Protocol (PPP) - daemon ii wvdial 1.60.1+nmu2 PPP dialer with built-in intelligence The results were: chat (from the Debian package ppp) requires TCGETS and TCSETS wvdial additionally requires TCFLSH. It also calls TIOCGSERIAL and TIOCMGET but does not seem to require them. gsmctl and gsmsmsd (from the Debian package gsm-utils) require the additional ioctls TIOCMBIC, TIOCMBIS and TCSBRK. But these utilities also assume that ATZ will always succeed, which prevents me from testing them any further. I conclude that these apps must be fixed in any case and that there therefore is no need to implement workarounds in the driver. I've not tried to get GUI applications like kppp etc to work, as they probably are tuned too much towards ppp-connections to be useful with cdc-wdm devices. >> Note: I have not understood why cdc-wdm isn't a fullblown TTY driver, >> like cdc-acm is. Could anyone care to explain? > > The device is no tty. The operations specific to ttys make no sense. > Your patch essentially proves that implementing no-ops. OK. Thanks for explaining. >> The correct fix for the described problem may be changing cdc-wdm to use >> the TTY layer, but I guess that means renaming the devices and possibly >> other very userspace visible changes? Let me know if this is desired, >> and I'll try to provide a patch. > > No, the correct fix is to fix "chat". The question is how practicable that > is. Agreed. I certainly had (and have) my doubts about adding workarounds like this to a driver. However, in this case there is a number of old applications which have never seen a non-tty modem-lookalike, but which still are potentionally usable with the cdc-wdm driver. But it's so easy to just post a patch like this and let someone else decide whether it's wise or not. You just gotta love the open source development model :-) Bjørn -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
