Skip site navigation (1)Skip section navigation (2)
Date:      4 May 95 15:31:00 -0500
From:      sung!imdave@iwcs.att.com
To:        freebsd-bugs@FreeBSD.org
Subject:   sio bug in april snap
Message-ID:  <9505042032.AA27782@ig1.att.att.com>

next in thread | raw e-mail | index | archive | help

Hi,

I am attempting to get remote login over a modem working and
although the modem connects and I do get the login prompt,
tons of garbage start spewing out which is echoed back to the
freebsd machine which sees it and of course says "illegal
login" (although its real difficult to see this amongst the
spewing garbage.)  I don't know if this is a problem
with the modem (USR sportster 28.8) or with my setup --
I can call out OK, but I haven't tried yet tried running dos
and connecting *to* the machine.

Running April SNAP, compiled kernel with "-O -m486" flags,
using gettytab "std.38400".  /etc/ttys is unchanged except
for enabling the getty on /dev/ttyd0 and using std.38400.

A terminal connected via /dev/ttyd5 (a STB 4 port card) using
std.9600 works fine, with no spewing garbage.

Thinking it might be some tty setup problem, I enabled
rc.serial -- this is the bug report -- it seems that the
stty's in rc.serial cause a kernel panic (boy was I glad
I had made a boot floppy..., is there any way to tell
init at boot time not to run any of the scripts?):

ie.  stty </dev/ttyid1 clocal -hupcl 1200

causes:

fatal trap 12: page fault in kernel mode
fault virt addr: 0x80
fault code: supvr read, page not pres
ip: 0x8, 0xf0184ac3
code seg: base 0 lim 0xfffff type 1b
proc eflags: int enabled, re?? (I can't seem to read my own
	handwriting...), IOPL=0
curproc: stty

(It would be nice if the registers were printed too....)


The IP corresponds to the term = tp->t_termios in sioioctl()
(_sioioctl @ f0184a80 from nm -p /kernel) in sio.c

BYW, how the *@# can I get a disassembly to see where the
failing instruction is?  I tried gdb /kernel, but gdb couldn't
find "debugging symbols".  I couldn't find any disassembler
program -- is there none?  Also, no crash program.  Finally,
I had to compile sio.c with -S, and run "as -a" to get a
listing...  That's a real crock I think...  Do I have to
enable the kernel debugger?  That seems like maybe a lot
of baggage to insert in the kernel just for this.


A hint would be appreciated.


Thanks!

BTW, Freebsd is working great... now if I could only get the
serial connections working...



Dave Bodenstab
...att!iwcs!sung!imdave



PS,  as long as I'm bug reporting...  just for the fun of it
I tried two simultaneous dd's from /dev/fd0.1440 and /dev/fd1.1200
If I used full track reads, then the two dd's alternated reading
the floppies.  When I did not specify the blocksize, I got
messages from the floppy driver something like "input in output".
Sorry, I never wrote this one down, since it didn't seem too
important.





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9505042032.AA27782>