home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
kermit.columbia.edu
/
kermit.columbia.edu.tar
/
kermit.columbia.edu
/
newsgroups
/
misc.20010921-20020314
/
000076_arice@benchmark-systems.com_Fri Oct 26 12:15:42 EDT 2001.msg
< prev
next >
Wrap
Text File
|
2002-03-13
|
6KB
|
123 lines
Article: 12889 of comp.protocols.kermit.misc
Path: newsmaster.cc.columbia.edu!panix!newsfeed.mathworks.com!news.maxwell.syr.edu!newsfeed.stanford.edu!postnews1.google.com!not-for-mail
From: arice@benchmark-systems.com (Alan Rice)
Newsgroups: comp.protocols.kermit.misc,comp.unix.aix
Subject: Re: Kermit droping back to a login
Date: 26 Oct 2001 09:00:11 -0700
Organization: http://groups.google.com/
Lines: 104
Message-ID: <d86c5ce1.0110260800.7b127dbf@posting.google.com>
References: <d86c5ce1.0110221123.53a7102b@posting.google.com> <d86c5ce1.0110230446.686fdc5@posting.google.com> <9r3rb4$c7t$1@newsmaster.cc.columbia.edu> <d86c5ce1.0110250443.273b22ef@posting.google.com> <9r92os$373$1@newsmaster.cc.columbia.edu>
NNTP-Posting-Host: 63.170.151.15
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
X-Trace: posting.google.com 1004112011 10823 127.0.0.1 (26 Oct 2001 16:00:11 GMT)
X-Complaints-To: groups-abuse@google.com
NNTP-Posting-Date: 26 Oct 2001 16:00:11 GMT
Xref: newsmaster.cc.columbia.edu comp.protocols.kermit.misc:12889 comp.unix.aix:224934
jaltman@watsun.cc.columbia.edu (Jeffrey Altman) wrote in message news:<9r92os$373$1@newsmaster.cc.columbia.edu>...
> In article <d86c5ce1.0110250443.273b22ef@posting.google.com>,
> Alan Rice <arice@benchmark-systems.com> wrote:
> :Frank da Cruz wrote:
> : > If you are dialing directly into an AIX serial port, that's a different
> : > story, but it's also a scenario I can't duplicate for testing. The most
> : > likely suspect is Kermit's calls to the /dev/tty driver. When Kermit
> : > starts, it gets and saves your terminal modes, and then puts the terminal
> : > into "CBREAK" mode so it can read a character at a time from your keyboard.
> : > When it exits, it restores the original terminal modes. So the first
> : > question is: can you start Kermit and then immediately EXIT from it without
> : > getting logged out?
> :
> : I am dialing in to the box in question.
> :
> : I can start and stop kermit with no problems. But I can not issue a
> : SEND or a RECEIVE without it kicking me out to a login prompt.
> :
> : ALan R
>
> The problem is most likely the one that Frank described. The attempt
> to reset the terminal modes is causing the PTY to terminate. Therefore,
> the getty driver executes a new login() command on the device.
>
> This would be a bug in the PTY driver. There are many such bugs that
> have been fixed with patches from IBM. I'm not sure which version of AIX
> you are running, but the IBM AIX APAR database is the place to start.
> Search for "pty".
>
>
>
> Jeffrey Altman * Sr.Software Designer C-Kermit 8.0 Beta available
> The Kermit Project @ Columbia University includes Secure Telnet and FTP
> http://www.kermit-project.org/ using Kerberos, SRP, and
> kermit-support@kermit-project.org OpenSSL. SSH soon to follow.
Jeffery,
Yes it is a "bug" in the driver That Dale Talcott told me about back
in May. The problem is IBM calls it a standard and I don't see it
going away. And my product vendor that ships my hardware is using the
AIX level that includes this.
Here's a copy of the email I sent to Frank back it May.
:IY06484 modifies the pseudo-tty driver so that if you perform a
:zero-length write on one size, it shows up as a zero-length read on
:the other side. For every other file type in Unix, a return of 0
:from read() indicates end-of-file, but those lunatic people in the
:Single-Unix standards group decided that pseudo-tty's should be an
:exception to this principle. IY06484 brings AIX into compliance with
:this idiocy. (Can you tell that I am not pleased with this part of
:the standard? :-)
:This breaks several third-party programs, as well as IBM's own
:"script" program. The only fix is to track down programs that
:manipulate pseudo-tty's and change how they detect when the tty is
:closed. (Our list so far is script, sshd, expect, and PBS.)
:If this turns out to be your problem, you'll need to modify kermit.
:There's a chance the kermit people have already done this, so you
:might check that you are running the latest version.
:--
:Dale Talcott, Purdue University Computing Center
: aeh@quest.cc.purdue.edu
And here's Franks reply:
:> In the AIX user group Dale Talcott sent me the following email.
Would this
:> explain the behavior I'm seeing?
:>
:No, but it's interesting anyway, because we do have "script" and
"expect"
:like functions in C-Kermit, which work through ptys. But you're not
using
:them. I suppose we'll have to look and see if our pty code treats
0-length
:reads as EOF. I hope not. Hmmm, OK, I just looked and sure
enough...
: if (read(syncpipe[0], &c, 1) == 0) { /* Slave side died */
:But that's in just one place so maybe there's a workaround...
:Anyway, as I noted previously, I can't reproduce the problem you
reported
:(getting logged out when you send a FINISH command) when making a
Telnet
:connection with C-Kermit from AIX 4.3.3 (or any other platform), and
I have
:no way of making a dialout connection from AIX with C-Kermit.
However, I
:don't see what difference the connection type would make.
:So I take it the IBM C compiler is an extra-cost option on AIX?
Jerks.
:- Frank
If you need to I can provide you with access to the box here that is
broken and does not work. Tell me what we need to do to fix it.
Alan R