dbfunk@ICAEN.UIOWA.EDU (David B Funk) (02/02/90)
In posting <9001242216.AA17858@richter.mit.edu> David Krowitz writes: > Ack! > I'm try to set up a Macintosh running a VT100 emulator > which is wired to the SIO port on a DN3000 running SR10.2 > The /etc/ttys file has the line > > tty01 "/etc/getty 9600-baud" vt100 on > > in it, "init" has started a process running "getty", the > Apollo echo characters to the Mac, but the login prompt > never appears! One thing that I've noticed is that /com/tctl > reports that "Wait for DCD" is true, ie. line 1 is waiting > for a modem to single that it has a carrier present. The > help for /com/tctl does *not* say how to disable this, and > I suspect it is why we can't login. Anyone got any hints? To make Domain/OS JLRU, it honors the state of DCD when using SIO devices "/dev/tty0?". This (and the work-around) are discussed in the sr10 transition guide. (/install/doc/apollo/os.v.10.?__transition_guide) SIO Lines Although /dev/siox and /dev/ttyx (where x is the port number) can refer to the same physical port, the system treats them differently. The state of DCD (Data Carrier Detect, pin #8 on a standard 25-pin RS-232 connector) is ignored on open for /dev/siox, but is meaningful for /dev/ttyx. Ignoring the state of DCD on /dev/tty devices when calling ios_$open is possible by specifying ios_$no_open_delay_opt in the ios_$open call. For /dev/sio devices, the ios_$no_delay_opt is always implied. So when you use /dev/tty01, the login process waits for DCD to go true, if you use /dev/sio1, it doesn't wait. This is true whether you are using "getty" or "siologin". It is also true when using the port for output. If it is at all possible to get your "terminal" (Macintosh in this case) to produce a signal that is active when turned on, then connect it to pin #8. This is very desirable because it helps to avoid problems with "getty" when using /dev/sio?. When you use /dev/sio? the "wait for DCD" function is lost so "getty" will respond to any incoming character on the receive data line. The problem here is that when the "terminal" is turned off, this line will float and random noise can be interpreted as characters causing lots of "getty" activity and tying up the node. If DCD is honored, then this doesn't happen because DCD will be false and "getty" will just wait & not see the random garbage. Dave Funk
hildebrand@PAN.SSEC.HONEYWELL.COM (Jeff Hildebrand) (02/03/90)
<<forwarded message>> from: umix.cc.umich.edu!apollo-request subj: Re: Help with /etc/getty sender: David B Funk <David B Funk dbfunk@icaen.uiowa.edu> date: Fri, 2 Feb 90 02:50:06 CST to: apollo@umix.cc.umich.edu krowitz@richter.mit.edu sent: 02/02/1990 4:24 am (CST) sender: pan!daemon ------- **| In posting <9001242216.AA17858@richter.mit.edu> David Krowitz writes: > Ack! > I'm try to set up a Macintosh running a VT100 emulator > which is wired to the SIO port on a DN3000 running SR10.2 > The /etc/ttys file has the line > > tty01 "/etc/getty 9600-baud" vt100 on > > in it, "init" has started a process running "getty", the > Apollo echo characters to the Mac, but the login prompt > never appears! One thing that I've noticed is that /com/tctl > reports that "Wait for DCD" is true, ie. line 1 is waiting > for a modem to single that it has a carrier present. The > help for /com/tctl does *not* say how to disable this, and > I suspect it is why we can't login. Anyone got any hints? To make Domain/OS JLRU, it honors the state of DCD when using SIO devices "/dev/tty0?". This (and the work-around) are discussed in the sr10 transition guide. (/install/doc/apollo/os.v.10.?__transition_guide) SIO Lines Although /dev/siox and /dev/ttyx (where x is the port number) can refer to the same physical port, the system treats them differently. The state of DCD (Data Carrier Detect, pin #8 on a standard 25-pin RS-232 connector) is ignored on open for /dev/siox, but is meaningful for /dev/ttyx. Ignoring the state of DCD on /dev/tty devices when calling ios_$open is possible by specifying ios_$no_open_delay_opt in the ios_$open call. For /dev/sio devices, the ios_$no_delay_opt is always implied. So when you use /dev/tty01, the login process waits for DCD to go true, if you use /dev/sio1, it doesn't wait. This is true whether you are using "getty" or "siologin". It is also true when using the port for output. If it is at all possible to get your "terminal" (Macintosh in this case) to produce a signal that is active when turned on, then connect it to pin #8. This is very desirable because it helps to avoid problems with "getty" when using /dev/sio?. When you use /dev/sio? the "wait for DCD" function is lost so "getty" will respond to any incoming character on the receive data line. The problem here is that when the "terminal" is turned off, this line will float and random noise can be interpreted as characters causing lots of "getty" activity and tying up the node. If DCD is honored, then this doesn't happen because DCD will be false and "getty" will just wait & not see the random garbage. Dave Funk