[comp.unix.amiga] Kernel Panics from kermit

pepers@enme3.ucalgary.ca (Brad Pepers) (05/24/91)

It seems I've found a strange but consistent way to panic the unix kernel.
What I am doing is using kermit. The way I can consistently do it is to
ftp to prep.ai.mit.edu and do a ls listing on the gnu directory. Its large
so that may have something to do with it. Everytime I do this, the kernel
panics (tried three times). I also get other misc panics when doing other
things in kermit but nothing I can reliably reproduce. My setup is a 3000UXD
which pretty much everything as it was shipped (modified S70sioc for some
overscan). My modem is a SupraModem 2400MNP. I use kermit by entering it
with the "kermit" command and then doing the following:

	set line /dev/term/ser
	set speed 2400
	set modem hayes
	dial xxx-xxxx
	...wait for connection...
	connect

Is this a known problem? Will the 2.0 ser device fix this?

    Brad Pepers (pepers@enme.ucalgary.ca)

bernie@metapro.DIALix.oz.au (Bernd Felsche) (05/25/91)

In <1991May24.035949.4906@cpsc.ucalgary.ca>
   pepers@enme3.ucalgary.ca (Brad Pepers) writes:

>It seems I've found a strange but consistent way to panic the unix kernel.
>What I am doing is using kermit. The way I can consistently do it is to
>ftp to prep.ai.mit.edu and do a ls listing on the gnu directory. Its large
>so that may have something to do with it. Everytime I do this, the kernel
>panics (tried three times). I also get other misc panics when doing other
>things in kermit but nothing I can reliably reproduce. My setup is a 3000UXD
>which pretty much everything as it was shipped (modified S70sioc for some
>overscan). My modem is a SupraModem 2400MNP. I use kermit by entering it
>with the "kermit" command and then doing the following:

>Is this a known problem? Will the 2.0 ser device fix this?

I know that our Motorola VME system crashes when (presumably) clist
over-runs on receives. Other System V.3 users report the same problem.
From what I've read, CBM worked from an early V.4 port provided by
Motorola.

This can be allieviated somewhat by setting proper handshaking
(x-on/x-off). If the far end misses the x-off though, it's PANIC
again.

I trust that you've already mailed this report to bugs.
-- 
Bernd Felsche,                 _--_|\   #include <std/disclaimer.h>
Metapro Systems,              / sale \  Fax:   +61 9 472 3337
328 Albany Highway,           \_.--._/  Phone: +61 9 362 9355
Victoria Park,  Western Australia   v   Email: bernie@metapro.DIALix.oz.au

rhealey@digibd.com (Rob Healey) (05/27/91)

In article <1991May24.035949.4906@cpsc.ucalgary.ca> pepers@enme3.UUCP (Brad Pepers) writes:
>It seems I've found a strange but consistent way to panic the unix kernel.
[ Stuff delete ]

	Right problem, WRONGO conclution. Kermit isn't the villin here,
	Mr. /dev/term/ser is the evil perpitrator of you're kernel
	panics. cu will also tickle /dev/term/ser to panic the kernel.
	The ser driver in 1.1 needs to be taken out into a field and
	shot but I've bitched about that before so enough already.

	The workaround is to get ckermit for AmigaDOS 2.0, fetch
	the stuff under AmigaDOS, save it to floppy with BRU and then
	restore it from floppy with BRU under UNIX. Ugly yes but it works...

	The problem is that the input side of /dev/term/ser can't be 
	serviced often enough so characters are dropped over 1200
	baud.

	The problem, so I'm told, is fixed in 2.0... For now, use
	AmigaDOS for serial transfers.

	By the way, you can dial into 1.1 UNIX at high speed and things
	work great. Unless you type over 1200 baud... B^). It's just
	the other way that don't work.


	Once again keeping the traffic in c.u.a up, B^),

		-Rob
-- 

Rob Healey                                          rhealey@digibd.com
Digi International (DigiBoard)
Eden Prairie, MN                                    (612) 943-9020

kaba@wintermute.uucp (Kai Bartels) (05/27/91)

pepers@enme3.ucalgary.ca (Brad Pepers) writes:
>It seems I've found a strange but consistent way to panic the unix kernel.
>What I am doing is using kermit. The way I can consistently do it is to
>[large blocks of data]
>overscan). My modem is a SupraModem 2400MNP. I use kermit by entering it
>with the "kermit" command and then doing the following:
>	set line /dev/term/ser
>[...]
I ecountered a simmilar problem while using kermit (and Forsberg's XYZ-
modem):
the kermit (not the xzy-modem!) works just fine (well, it doesn't crash)
when I invoke it on the line, I'm doing the transfer on (term/ser). But
when I invoke either of the programs from a console instead using the -l
option to set the line (and redirection with xyz-modem), I get another
(kernel-) panic. :-(

>    Brad Pepers (pepers@enme.ucalgary.ca)
regards, Kai

-- 
"The sound of bombs has given way for children's cries"            <Fisher-Z>
BITNET: g14b@dhbrrz41               +               UUCP:kaba@wintermute.uucp
Snail:   Kai Bartels   +   Hudemuehler Str. 37   +   2800 Bremen 41   +   FRG