daleske@apr.UUCP (John D. Daleske) (12/06/84)
Does anyone know of good ways to use a phone line for both dial-in and dial-out? We have two lines and would like to use both for dial-in until a need arises to dial-out. I would prefer being able to use either or both for dial-out. If this is not feasible, then dedicating one line would certainly be no worse than our current situation. Our site: 4.2BSD UNIX on an Integrated Solutions Optimum. Two phone lines, one on a Rixon R212A and the other a Hayes 1200. John Daleske ... ihnp4!cbosgd!apr!daleske Advanced Programming Resources, Inc.
keith@reed.UUCP (Packard) (12/10/84)
Regarding ports with modem control. I just finished hacking the DL driver so that it supports lines that occasionally want modem control (so you can dial in) and occasionally don't want modem control (so you can dial out). This fix was *very* easy. I simply set bit 6 of the minor device number to tell the driver to not hang on open and also to not set the DATASET INTERRUPT bit in the reciever csr. Much easier than making the line hang on the first read. That involves nasty changes else where. keith packard ...!tektronix!reed!keith or ...!tektronix!tekmdp!keithp "even a 11/23 with unix sources is better than a vax without."
wls@astrovax.uucp (12/10/84)
I have just posted the changes to the 4.2 BSD uucp and tip to use the same
line for both dialing in and dialing out. Similar changes existed here when
we were running 4.1 BSD. The heart of it is a program called acucntrl that does
a lot of poking and hacking (which it must because no straightforward hooks
are provided into the operating system to do what must be done), and thus
is specific to 4.2 BSD with the Vax dz, dh, or dmf drivers. Thus I can't
guarantee it for other than the VAX although who knows. The modifications to
it if any for 4.X BSD on some other computer should be at least straight
forward. The modifications to it for System V are harder but should be doable
(though System V provides other hooks than those used here).
This may start some flaming but I should at least say what I think one of those
hooks should be. For a start, I think that the use of modem control should be
settable by an ioctl. Terminal lines with modem control enabled should hang
on first read rather than hanging on open, to allow one to gain the access to
do this ioctl. To enable/disable modem control I must poke in /dev/kmen to
toggle one of the driver bits, which accounts for the driver dependence.
Some better communication with init (process one) would also be nice. For
instance in 4.2 BSD it could be listening to a socket. There are some terrible
kluges resulting from the fact that one has little way of determining when
init has responded to a request.
--
Bill Sebok Princeton University, Astrophysics
{allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wlsignatz@aicchi.UUCP (Ihnat) (12/22/84)
Most fixes to the line turnaround problem are0 source fixes. This is, admittedly, the *proper* and *elegant* way to solve the problem; but, alas, not always available. We have a Plexus P/40 running System III, and absolutely couldn't dedicate a dial-out line; yet we needed/wanted it. SO...I've implemented a 'truly portable' and thoroughly non-elegant system of doing scheduled line turnaround. It doesn't care what version of USG Unix you're running--although it must have USG-like init states--and allows you to dial out on one of any number of uucp-supported lines. Specifically, it involves a shell-script(!), additional init states to shutdown getty, and a user-maintained tty information file. If anyone without a source license would like this stuff, and the information to make it work, mail to me; enough responses, as usual, and I'll post to net.sources. -- Dave Ihnat Analysts International Corporation (312) 882-4673 ihnp4!aicchi!ignatz