[comp.os.minix] controlling ttys, SIGHUP

leo@krabat.marco.de (Matthias Pfaller) (01/22/91)

I tried a few things with cttys on our sysV box;
1. Minix does not send SIGHUP to childern if the processgroupleader dies.
   Our sysV box does; what is correct for POSIX?

2. For minix the children of a exited processgroupleader no longer can open
   /dev/tty; On our sysV box the children still can access /dev/tty, but
   do no longer receive signals (as it is in Minix). Again, what is correct?

	Matthias Pfaller (leo @ marco.de)

allbery@NCoast.ORG (Brandon S. Allbery KB8JRR) (01/24/91)

As quoted from <588@krabat.marco.de> by leo@krabat.marco.de (Matthias Pfaller):
+---------------
| I tried a few things with cttys on our sysV box;
| 1. Minix does not send SIGHUP to childern if the processgroupleader dies.
|    Our sysV box does; what is correct for POSIX?
+---------------

I posted something about this to the comp.unix.sysv386 yesterday; I'll give a
short summary here.  (The article I put in .sysv386 was *long*.)

Basically, what you see on the System V box happens via a "loopback" through
the modem.  The exact sequence of events is:

	process group leader dies
	getty spawns and forces DTR to drop
	modem hangs up and drops DCD to computer
	computer sends SIGHUP to children

Note that in the normal case (if getty does *not* force the dropping of DTR)
DTR will not be dropped, and therefore DCD will not be dropped, and therefore
SIGHUP will not be sent.  Unless someone System-V-ized getty in Minix, this
sequence of events won't happen regardless.

In any case, the only time SIGHUP is sent by the system is if it has modem
control enabled and DCD is dropped on the tty line.

+---------------
| 2. For minix the children of a exited processgroupleader no longer can open
|    /dev/tty; On our sysV box the children still can access /dev/tty, but
|    do no longer receive signals (as it is in Minix). Again, what is correct?
+---------------

Arguably, Minix is more correct.  I think Posix specifies this as well.  In
any case, having been on the receiving end of output from a forgotten
background process of a previous caller on one of ncoast's modems (under
System III, at the time), Minix's behavior is friendlier to the next person to
dial in.

++Brandon
-- 
Me: Brandon S. Allbery			    VHF/UHF: KB8JRR on 220, 2m, 440
Internet: allbery@NCoast.ORG		    Packet: KB8JRR @ WA8BXN
America OnLine: KB8JRR			    AMPR: KB8JRR.AmPR.ORG [44.70.4.88]
uunet!usenet.ins.cwru.edu!ncoast!allbery    Delphi: ALLBERY