Subject: krb5_timeofday() losses data from 64-bit time_t to krb5_int32

on windows, time_t is 64-bit even on 32-bit systems.
krb5_timeofday() assigns the 64-bit value to a 32-bit variable.

Here is a list of all 64-bit related warnings generated by the Windows
.NET 2005 compiler.


c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb5\asn.1\asn1_k_decode.c(288)
: warning C4244: '=' : conversion from 'time_t' to 'krb5_timestamp',
possible loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb5\ccache\cc_mslsa.c(431) :
warning C4244: '=' : conversion from 'time_t' to 'krb5_timestamp',
possible loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb5\ccache\cc_mslsa.c(432) :
warning C4244: '=' : conversion from 'time_t' to 'krb5_timestamp',
possible loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb5\ccache\cc_mslsa.c(433) :
warning C4244: '=' : conversion from 'time_t' to 'krb5_timestamp',
possible loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb5\os\c_ustime.c(52) :
warning C4244: '=' : conversion from '__time64_t' to 'krb5_int32',
possible loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb5\os\sendto_kdc.c(455) :
warning C4244: '=' : conversion from '__time64_t' to 'long', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb5\os\timeofday.c(50) :
warning C4244: '=' : conversion from 'time_t' to 'krb5_int32', possible
loss of data


c:\src\kerberos\mit-cvs\krb5-head\src\lib\des425\unix_time.c(43) :
warning C4244: '=' : conversion from 'time_t' to 'krb5_ui_4', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\des425\unix_time.c(44) :
warning C4244: 'return' : conversion from 'time_t' to 'krb5_ui_4',
possible loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb4\win_time.c(32) : warning
C4244: 'return' : conversion from '__time64_t' to 'unsigned int',
possible loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\lib\krb4\win_time.c(116) : warning
C4244:'=' : conversion from 'time_t' to 'long', possible loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\windows\cns\tktlist.c(64) :
warning C4133: 'function' : incompatible types - from 'long *' to 'const
time_t *'

c:\src\kerberos\mit-cvs\krb5-head\src\windows\cns\tktlist.c(176) :
warning C4244: 'function' : conversion from 'time_t' to 'long', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\windows\cns\tktlist.c(179) :
warning C4244: 'function' : conversion from 'time_t' to 'long', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\windows\gss\gss-misc.c(389) :
warning C4244: '=' : conversion from '__time64_t' to 'long', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\clients\klist\klist.c(238) :
warning C4244: '=' : conversion from 'time_t' to 'krb5_int32', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\windows\gss\gss-misc.c(389) :
warning C4244: '=' : conversion from '__time64_t' to 'long', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\clients\klist\klist.c(238) :
warning C4244: '=' : conversion from 'time_t' to 'krb5_int32', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\clients\kinit\kinit.c(335) :
warning C4244: '=' : conversion from 'time_t' to 'krb5_deltat', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\appl\gss-sample\gss-misc.c(418) :
warning C4244: '=' : conversion from '__time64_t' to 'long', possible
loss of data

c:\src\kerberos\mit-cvs\krb5-head\src\appl\gssftp\ftp\ftp.c(2317) :
warning C4244: '=' : conversion from '__time64_t' to 'long', possible
loss of data




These are not time_t issues but still:

c:\src\kerberos\mit-cvs\krb5-head\src\lib\gssapi\krb5\k5seal.c(246) :
warning C4244: 'function' : conversion from 'gssint_uint64' to
'krb5_ui_4', possible loss of data
c:\src\kerberos\mit-cvs\krb5-head\src\lib\gssapi\krb5\k5seal.c(260) :
warning C4244: '=' : conversion from 'gssint_uint64' to 'unsigned char',
possible loss of data
c:\src\kerberos\mit-cvs\krb5-head\src\lib\gssapi\krb5\k5seal.c(261) :
warning C4244: '=' : conversion from 'gssint_uint64' to 'unsigned char',
possible loss of data
c:\src\kerberos\mit-cvs\krb5-head\src\lib\gssapi\krb5\k5seal.c(262) :
warning C4244: '=' : conversion from 'gssint_uint64' to 'unsigned char',
possible loss of data
c:\src\kerberos\mit-cvs\krb5-head\src\lib\gssapi\krb5\k5seal.c(263) :
warning C4244: '=' : conversion from 'gssint_uint64' to 'unsigned char',
possible loss of data

From: Ken Raeburn <raeburn@MIT.EDU>
Subject: Re: [krbdev.mit.edu #2883] 64-bit time_t issues 
Date: Wed, 16 Mar 2005 15:21:28 -0500
To: rt-comment@krbdev.mit.edu
RT-Send-Cc: 

On Mar 15, 2005, at 01:06, Jeffrey Altman via RT wrote:
> Here is a list of all 64-bit related warnings generated by the Windows
> .NET 2005 compiler.
>
>
> c:\src\kerberos\mit-cvs\krb5-
> head\src\lib\krb5\asn.1\asn1_k_decode.c(288)
> : warning C4244: '=' : conversion from 'time_t' to 'krb5_timestamp',
> possible loss of data

Most of the krb5_timestamp issues should probably be dealt with by
making krb5_timestamp 64 bits on 64-bit Windows, as we were discussing.
  Though there might be interesting issues if 64-bit Windows supports
32-bit Windows software, and krb5_timestamp is used in any IPC (CCAPI?)
or file formats (ccache, keytab).  Which, actually, it appears to be.
So maybe this isn't quite so simple after all.

> These are not time_t issues but still:
>
> c:\src\kerberos\mit-cvs\krb5-head\src\lib\gssapi\krb5\k5seal.c(246) :
> warning C4244: 'function' : conversion from 'gssint_uint64' to
> 'krb5_ui_4', possible loss of data

All of these are in code implementing the original GSSAPI mech with
32-bit sequence numbers.  So maybe there should be an "assert(seqnum <=
0xffffffff)" but otherwise it's okay.

I have discovered that I can take care of the immediate 32-bit Windows
platform problem with 64-bit time_t by defining in win-mac.h:

/* To ensure backward compatibility of the ABI use 32-bit time_t on
 * 32-bit Windows.
 */
#if _INTEGRAL_MAX_BITS >= 64
#ifndef _WIN64
#define _USE_32BIT_TIME_T
#endif
#endif

This will force the time_t to be 32-bits on 32-bit Windows.

Looking longer term I am going to recommend we make the following
prototype changes:

--- krb5.hin    (revision 17479)
+++ krb5.hin    (working copy)
@@ -1802,11 +1802,11 @@
 /* From krb5/os but needed but by the outside world */
 krb5_error_code KRB5_CALLCONV krb5_us_timeofday
        (krb5_context,
-               krb5_int32 *,
+               krb5_timestamp *,
                krb5_int32 * );
 krb5_error_code KRB5_CALLCONV krb5_timeofday
        (krb5_context,
-               krb5_int32 * );
+               krb5_timestamp * );
                 /* get all the addresses of this host */
 krb5_error_code KRB5_CALLCONV krb5_os_localaddr
        (krb5_context,
@@ -2275,19 +2275,19 @@
 #endif

 krb5_error_code KRB5_CALLCONV krb5_set_real_time
-       (krb5_context, krb5_int32, krb5_int32);
+       (krb5_context, krb5_timestamp, krb5_int32);

 #if KRB5_PRIVATE
 krb5_error_code krb5_set_debugging_time
-       (krb5_context, krb5_int32, krb5_int32);
+       (krb5_context, krb5_timestamp, krb5_int32);
 krb5_error_code krb5_use_natural_time
        (krb5_context);
 #endif
 krb5_error_code KRB5_CALLCONV krb5_get_time_offsets
-       (krb5_context, krb5_int32 *, krb5_int32 *);
+       (krb5_context, krb5_timestamp *, krb5_int32 *);
 #if KRB5_PRIVATE
 krb5_error_code krb5_set_time_offsets
-       (krb5_context, krb5_int32, krb5_int32);
+       (krb5_context, krb5_timestamp, krb5_int32);
 #endif

 /* str_conv.c */

Currently krb5_timestamp is 32-bits.  Changing the prototypes to
krb5_timestamp will make it more obvious what the impact will be when
we redefine krb5_timestamp as a 64-bit value.   I believe krb5_timestamp
should be defined in terms of time_t.

Reading RFC 4120 I do not believe there are any bit length restrictions
on the number of seconds represented in a KerberosTime.  Therefore, we
should attempt to remove the restrictions in our own implementation.


Jeffrey Altman

Ken:

This is an improved test for the definition of 32-bit time_t

/* To ensure backward compatibility of the ABI use 32-bit time_t on
 * 32-bit Windows.
 */
#if _INTEGRAL_MAX_BITS >= 64 && _MSC_VER >= 1400 && !defined(_WIN64) &&
!defined(_USE_32BIT_TIME_T)
#if defined(_TIME_T_DEFINED) || defined(_INC_IO) || defined(_INC_TIME)
|| defined(_INC_WCHAR)
#error time_t has been defined as a 64-bit integer which is incompatible
with Kerberos on this platform.
#endif /* _TIME_T_DEFINED */
#define _USE_32BIT_TIME_T
#endif

Jeffrey Altman

In order to ensure that we do not affect third party application
developers that wish to use 64-bit time_t on 32-bit Windows, we will
only define _USE_32BIT_TIME_T if win-mac.h is loaded by k5-int.h.   This
means that we must ensure that our library source files include k5-int.h
instead of krb5.h. (see ticket 3236)

The current change for win-mac.h now reads:

/* To ensure backward compatibility of the ABI use 32-bit time_t on
 * 32-bit Windows.
 */
#ifdef _KRB5_INT_H
#ifdef KRB5_GENERAL__
#error krb5.h included before k5-int.h
#endif /* KRB5_GENERAL__ */
#if _INTEGRAL_MAX_BITS >= 64 && _MSC_VER >= 1400 && !defined(_WIN64) &&
!defined(_USE_32BIT_TIME_T)
#if defined(_TIME_T_DEFINED) || defined(_INC_IO) || defined(_INC_TIME)
|| defined(_INC_WCHAR)
#error time_t has been defined as a 64-bit integer which is incompatible
with Kerberos on this platform.
#endif /* _TIME_T_DEFINED */
#define _USE_32BIT_TIME_T
#endif
#endif

and we also include the following check in krb5.h to ensure that krb5.h
is not loading win-mac.h before k5-int.h would have done so.

#ifdef KRB5_GENERAL__
#error krb5.h included before k5-int.h
#endif /* KRB5_GENERAL__ */

From: jaltman@mit.edu
Subject: CVS Commit

  * Correct function prototypes that should have been using
    krb5_timestamp in order to prevent type conflicts if
    krb5_timestamp ever becomes a 64-bit value

  * Force the use of 32-bit time_t with Microsoft's VS 2005
    compiler on 32-bit platforms

  * Test for inclusion of krb5.h before k5-int.h


Commit By: jaltman



Revision: 17488
Changed Files:
U   trunk/src/include/ChangeLog
U   trunk/src/include/k5-int.h
U   trunk/src/include/krb5.hin
U   trunk/src/include/win-mac.h
U   trunk/src/lib/krb5/os/ChangeLog
U   trunk/src/lib/krb5/os/timeofday.c
U   trunk/src/lib/krb5/os/toffset.c
U   trunk/src/lib/krb5/os/ustime.c

From: tlyu@mit.edu
Subject: CVS Commit

pull up r17488 from trunk.  This breaks the unix build; corrected by
ticket #3236.

Commit By: tlyu



Revision: 17502
Changed Files:
U   branches/krb5-1-4/src/include/ChangeLog
U   branches/krb5-1-4/src/include/k5-int.h
U   branches/krb5-1-4/src/include/krb5.hin
U   branches/krb5-1-4/src/include/win-mac.h
U   branches/krb5-1-4/src/lib/krb5/os/ChangeLog
U   branches/krb5-1-4/src/lib/krb5/os/timeofday.c
U   branches/krb5-1-4/src/lib/krb5/os/toffset.c
U   branches/krb5-1-4/src/lib/krb5/os/ustime.c