patches

linux终端仿真程序

C-KERMIT 6.0 PATCHES

Most recent update: Wed Dec 18 11:44:34 1996

This file contains source-level patches to C-Kermit 6.0.192.

You can apply these patches if you have source code and the necessary C
compiler and libraries to build a new executable.  There is no need to apply
any patch if it does not apply to your configuration, or if the symptom does
not affect you or your users.

Each patch is self-contained and independent from the others.  Line numbers
correspond to the source-code files of the 6.0.192 release.

The patches included in this file are believed to be safe, but come with no
guarantees.  They appear in chronological order as email messages separated by
lines of underscores.

CONTENTS:

Number Domain           Description
 0001   UNIX             C-Kermit mishandles timestamps on very old files
 0002	Solaris 2.5++    Compilation error on Solaris 2.5 with Pro C
 0003	VMS              CKERMIT.INI Fix for VMS
 0004	VMS/VAX/UCX 2.0  C-Kermit 6.0 can't TELNET on VAX/VMS with UCX 2.0
 0005	All              C-Kermit Might Send Packets Outside Window
 0006	All              MOVE from SEND-LIST does not delete original files
 0007	Solaris 2.5++    Higher serial speeds on Solaris 2.5
 0008   All              C-Kermit application file name can't contain spaces

Legend:
 ++ means "and above"
____________________
X-Patch-Number: 0001
Date: Sun, 1 Dec 1996 22:16:56 -0800
From: Paul Eggert <[e-mail]>
To: [e-mail]
Subject: C-Kermit mishandles timestamps on very old files

C-Kermit 6.0.192 rejects files whose timestamps are before 1970-01-01
local time.  But every Unix can represent files with timestamps before
that.  It's common, for example, in the US/Pacific time zone to have a
file with timestamp 1969-12-31 16:00:00 local time, since this equals
1970-01-01 00:00:00 UTC.  And many Unixes can represent times before
the epoch, e.g. on Solaris 2.5.1:

  $ uname -a
  SunOS shade 5.5.1 Generic_103640-03 sun4u sparc SUNW,Ultra-1
  $ echo $TZ
  UTC0
  $ ls -l test
  -rw-rw-r--   1 eggert   eggert         0 Dec 13  1901 test

It is certainly not a requirement of the ANSI C Library, or of Posix,
that timestamps before 1970 cannot be represented.  Here is a patch:

RCS file: RCS/ckufio.c,v
retrieving revision 6.0.192.0
retrieving revision 6.0.192.1
diff -c -r6.0.192.0 -r6.0.192.1
*** ckufio.c	1996/11/23 19:27:13	6.0.192.0
--- ckufio.c	1996/12/02 06:07:54	6.0.192.1
***************
*** 2659,2666 ****
      yy = time_stamp->tm_year;		/* Year - 1900 */
      yy += 1900;
      debug(F101,"zdatstr year","",yy);
-     if (yy < 1970)			/* By definition of C library */
-       return("");
  
      if (time_stamp->tm_mon  < 0 || time_stamp->tm_mon  > 11)
        return("");
--- 2659,2664 ----
[It is not known if this patch is safe on all UNIX platforms.]
____________________
X-Patch-Number: 0002
Date: Mon, 02 Dec 1996 13:47:13 -0500
To: [e-mail]
Subject: Compilation error on Solaris 2.5 with Pro C
From: "David J. Fiander" <[e-mail]>

I defined CK_ENVIRONMENT and found that the prototype for tn_snenv() in
ckcnet.h conflicted with the function definition in ckcnet.c.  the former
declared the first parameter to be a CHAR * (i.e. an unsigned char *),
while the latter declared to be a char * (i.e. signed char *).  The
Pro C compiler treats this as an error.  I changed the function definition
in ckcnet.c to declare the parameter as a CHAR * and the problem was
resolved.

- David

[Note: This problem does not occur with standard builds.
Resolution: CK_ENVIRONMENT, which controls the use of TELNET
NEW-ENVIRONMENT protocol, is not supported in the UNIX version of
C-Kermit 6.0.192.  The type mismatch is a mistake, however.]
____________________
X-Patch-Number: 0003
Date: Tue, 3 Dec 96 10:56:50 EST
From: Mike Freeman <[e-mail]>
Subject: CKERMIT.INI Fix for VMS

A minor fix to CKERMIT.INI from the 6-Sep release is shown below. 
Needed to get C-Kermit 6(192) to run on my Vax/VMS system with VmS
V5.5-2 and VAX C V3.1.

*** $ker:ckermit.ini;-1
--- $ker:ckermit.ini;
**************
*** 680,686
  
  ; In VMS and OpenVMS, allow for system-wide site customizations
  
! if equal "\v(system)" "VMS" {
      xif exist CKERMIT_INI:CKERMIT.SYS {
  	echo Executing CKERMIT_INI:CKERMIT.SYS
  	take CKERMIT_INI:CKERMIT.SYS
--- 680,686 -----
  
  ; In VMS and OpenVMS, allow for system-wide site customizations
  
! xif equal "\v(system)" "VMS" {
      xif exist CKERMIT_INI:CKERMIT.SYS {
  	echo Executing CKERMIT_INI:CKERMIT.SYS
  	take CKERMIT_INI:CKERMIT.SYS

MIke Freeman                    | Internet: [e-mail]
Amateur Radio Calsign: K7UIJ    | */ PGP Public Key Available */
____________________
X-Patch-Number: 0004
Date: Thu, 5 Dec 1996 15:20:00 EST
From: Frank da Cruz <[e-mail]>
Subject: C-Kermit 6.0 can't TELNET on VAX/VMS with UCX 2.0

PROBLEM:

On VAX/VMS systems with early versions of DEC TCP/IP (UCX), "set host" and
"telnet" commands might fail with "Connection refused", even though the system
(UCX) TELNET program to the same host works correctly.

DIAGNOSIS:

If you don't specify a port, Kermit substitutes the service name "telnet" and
then asks UCX to look it up; similary, if you specify a port by name, rather
than number, Kermit asks UCX to look it up.

On VAX/VMS with DEC TCP/IP UCX 2.0, 2.0A, 2.0B, or 2.0C, the getservbyname()
function fails to return the port number in network byte order as it should,
and instead returns the number with its bytes swapped (e.g. 5888 instead of
23).  Then Kermit tries to connect to port 5888 on the host; most hosts will
refuse the connection; if they don't, you probably didn't reach a Telnet port
anyway.

This bug was fixed in UCX 2.0D and thereafter.  However, there is no way for
Kermit to know which version of UCX is being used.

WORKAROUND:

Specify the port number (not name) in your SET HOST or TELNET command, e.g.:

  set host xyzcorp.com 23

CURE:

A new command:

  SET TCP UCX-PORT-BUG { ON, OFF }

It is OFF by default, so UCX versions of VMS C-Kermit will work with all
latter-day versions of UCX with no special effort.  But if you have an old
UCX system, and you tell C-Kermit to "set host" or "telnet" and it says
"Connection refused", tell it to SET TCP UCX-PORT-BUG ON and try again.  If
this works, then put this command into your (or the system-wide) CKERMIT.INI
file.  The following patch adds this command to VMS versions of C-Kermit that
were built for UCX support.  It has already been applied to the CKVV55-UCX20
binary in the Kermit archive.

PATCH:

*** /w/pub/ftp/kermit/f/ckcnet.c	Wed Nov 27 19:54:46 1996
--- ./ckcnet.c	Wed Dec  4 17:14:41 1996
***************
*** 1,4 ****
! char *cknetv = "Network support, 6.0.078, 6 Sep 1996";
  
  /*  C K C N E T  --  Network support  */
  
--- 1,4 ----
! char *cknetv = "Network support, 6.0.079, 4 Dec 1996";
  
  /*  C K C N E T  --  Network support  */
  
***************
*** 960,965 ****
--- 960,967 ----
  extern VOID C$$SOCK_TRANSLATE();
  #endif /* CK_ANSIC */
  
+ int ucx_port_bug = 0;			/* Explained below */
+ 
  struct servent *
  my_getservbyname (service, proto) char *service, *proto; {
      static struct servent sent;
***************
*** 1007,1035 ****
  	C$$SOCK_TRANSLATE(&s.sb.st[0]);
  	return NULL;
      }
-     /* sent.s_port is returned by UCX in network byte order. */
-     /* Calling htons here swaps the bytes, which ruins everything. */
- 
-     /* Oh yeah?  WHICH VERSION of UCX???  Let's try this... */
- 
- #ifndef __alpha /* Maybe it should be __DECC, or some version thereof... */
  /*
!   Hunter says: "In fact, the "#ifndef __alpha" isn't even needed, since
!   my_getservbyname() isn't included if "__DECC" is defined, and that's
!   always defined on Alpha."  But if it doesn't hurt either, better not risk
!   taking it out.
  */
! #ifndef TCPWARE
! #define DO_HTONS
! #endif /* TCPWARE */
! #endif /* __alpha */
! 
! #ifdef DO_HTONS
      sent.s_port = htons(sent.s_port);
!     debug(F111,"UCX getservbyname","port",ntohs(sent.s_port));
! #else
!     debug(F111,"UCX getservbyname","port",sent.s_port);
! #endif /* DO_HTONS */
      return &sent;
  }
  #endif /* __DECC */
--- 1009,1028 ----
  	C$$SOCK_TRANSLATE(&s.sb.st[0]);
  	return NULL;
      }
  /*