Date: Wed, 4 Sep 2002 12:11:10 +0100 From: "Duncan Barclay" <dmlb@dmlb.org> To: "John E Hein" <jhein@timing.com> Cc: "Mark Santcroos" <marks@ripe.net>, <emulation@FreeBSD.ORG> Subject: Re: TIOCSCTTY not implemented in linuxulator? Message-ID: <001b01c25403$c65353e0$6d6020c2@pc598cam> References: <XFMail.20020903224917.dmlb@computer.my.domain><20020904020952.GB7157@dhcp01.pn.xcllnt.net><20020904064620.GA713@laptop.6bone.nl><000801c253f7$2cae8ea0$6d6020c2@pc598cam> <15733.59121.143526.699051@brain.timing.com>
next in thread | previous in thread | raw e-mail | index | archive | help
From: "John E Hein" <jhein@timing.com>
Sent: Wednesday, September 04, 2002 11:56 AM
[sorry about quoting/format, using outlook at ${REALJOB}]
> Duncan Barclay wrote at 10:40 +0100 on Sep 4:
> > > Hi Duncan,
> > >
> > > Can you find out (ktrace) to which devices it is sending these
ioctls?
> >
> > It's the tty side of a pty/ttp pair. Matlab spawns a subprocess, and
it's
> > the subprocess that issues these ioctls.
>
> See this freebsd-emulation thread ("can't quit matlab"):
>
http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&frame=right&th=2
06ec3e38e06e5d2&seekm=15218.39637.311548.512594_localhost.econ.vt.edu%40ns.s
ol.net#link1
>
> Note that the 540e is TIOCSCTTY:
>
> /usr/src/linux-2.2.19/include/asm-i386/ioctls.h:#define TIOCSCTTY
0x540E
>
> Particularly look at the 3rd message in the thread that talks about the
> /dev/ptmx workaround and unix98 ptys.
> If someone could solve this, it'd be great.
>
Yup, this is the problem I'm trying to solve. Exiting without using the
/dev/ptmx problem.
ktrace output on spawned matlab process.
54357 matlab_helper NAMI "/compat/linux/dev/ttyp2"
54357 matlab_helper NAMI "/dev/ttyp2"
54357 matlab_helper RET linux_newstat 0
54357 matlab_helper CALL setsid
54357 matlab_helper RET setsid 54357/0xd455
54357 matlab_helper CALL linux_open(0xbfbfed58,0x2,0x60)
54357 matlab_helper NAMI "/compat/linux/dev/ttyp2"
54357 matlab_helper NAMI "/dev/ttyp2"
54357 matlab_helper RET linux_open 4
54357 matlab_helper CALL linux_ioctl(0x4,TIOCSER_TEMT,0x8049362)
54357 matlab_helper RET linux_ioctl -1 errno 22 Invalid argument
54357 matlab_helper CALL linux_ioctl(0x4,TIOCSER_TEMT,0x8049367)
54357 matlab_helper RET linux_ioctl -1 errno 22 Invalid argument
54357 matlab_helper CALL linux_ioctl(0x4,TIOCSCTTY,0x8049362)
54357 matlab_helper RET linux_ioctl -1 errno 22 Invalid argument
54357 matlab_helper CALL getpgrp
54357 matlab_helper RET getpgrp 54357/0xd455
54357 matlab_helper CALL linux_ioctl(0x4,TIOCSPGRP,0xbfbfecb4)
54357 matlab_helper RET linux_ioctl 0
54357 matlab_helper CALL read(0,0xbfbfecfc,0x4)
[end of trace]
Trying to quit matlab causes it to hang, kill -9 works. If I start
matlab with
/usr/local/bin/matlab -glnx86 -nosplash -nodesktop
so I get matlab running in a terminal, after the kill -9, the terminal
doesn't receive keyboard input.
Adding
case LINUX_TIOCSCTTY:
args->cmd = TIOCSCTTY;
return (ioctl(p, (struct ioctl_args *)args));
to /sys/compat/linux/linux_ioctl.h, correctly restores the terminal but does
not fix the exit problem itself. i.e. kill -9 is still needed.
Duncan
________________________________________________________________________
Duncan Barclay | God smiles upon the little children,
dmlb@dmlb.org | the alcoholics, and the permanently stoned.
dmlb@freebsd.org| Steven King
________________________________________________________________________
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-emulation" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?001b01c25403$c65353e0$6d6020c2>