[Bug libc/601] New: mktime(3) unreliable after calling tzset(3) when TZ is NULL or bogus

[Bug libc/601] New: mktime(3) unreliable after calling tzset(3) when TZ is NULL or bogus

[kreckel at ginac dot de]

#include <time.h>
#include <stdlib.h>
#include <unistd.h>
#include <stdio.h>

int main()
{
    unsigned sec;
    struct tm t;
    time_t t1, t2;
    const char* currTz;
    t.tm_year = 100; // 2000
    t.tm_mon = 0; // Jan
    t.tm_mday = 1;
    t.tm_hour = 0;
    t.tm_min = 0;
    t.tm_sec = 0;
    t.tm_isdst = 0;
    t1 = mktime( &t );
    for ( sec = 0; sec < 24*60*60; ++sec ) {
        t.tm_hour = sec / ( 60 * 60 );
        t.tm_min = ( sec / ( 60 ) ) % 60;
        t.tm_sec = sec % ( 60 );
        printf("%02d:%02d:%02d\t",t.tm_hour,t.tm_min,t.tm_sec);

        currTz = getenv( "TZ" );
        //setenv( "TZ", "#!@+*%&
NULLCannotBeUsedBecauseOfABugWhenLinkingWithLibPthreadStringCannotBeUsedEither",
1 );  // leads to errors
        setenv( "TZ", NULL, 1 );  // leads to errors
        //setenv( "TZ", "UTC", 1 );  // seems to work
        tzset();

        t2 = mktime( &t );

        if ( currTz )
            setenv( "TZ", currTz, 1 );
        else
            unsetenv( "TZ" );

        printf("%d\n",t2-t1);
        t1 = t2;
    }
}

The above program should print:
00:00:00        3600
00:00:01        1
00:00:02        1
00:00:03        1

The left column should always be the time difference to the previous line or, in
the case of the first line, the difference of the local time to UTC.  However,
it sometimes prints:
00:00:00        3601
00:00:01        0
00:00:02        1
00:00:03        1
[...]

or, even stranger things like
[...]
00:01:15        1
00:01:16        7201
00:01:17        -7199
00:01:18        1
[...]

This happens only if I set TZ to NULL or to some bogus (see code) value *inside*
the loop. However, according to tzset(3): "If the TZ variable does appear in the
environment but its value is NULL or its value cannot be interpreted using any 
of the formats specified below, Coordinated Universal Time (UTC) is used."

I have reproduced this bug with the glibc shipped with SuSE Professional Linux
8.1, SuSE Linux Linux 9.2 and Debian 3.0, so I assume it must be from upstream.

-- 
           Summary: mktime(3) unreliable after calling tzset(3) when TZ is
                    NULL or bogus
           Product: glibc
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: libc
        AssignedTo: gotom at debian dot or dot jp
        ReportedBy: kreckel at ginac dot de
                CC: glibc-bugs at sources dot redhat dot com


http://sources.redhat.com/bugzilla/show_bug.cgi?id=601

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/601] mktime(3) unreliable after calling tzset(3) when TZ is NULL or bogus

[jakub at redhat dot com]

------- Additional Comments From jakub at redhat dot com  2004-12-06 22:41 -------
Calling setenv with second argument NULL is a bug, just don't do that.
The man page you are citing is badly worded and inaccurate, e.g. it is not
possible to have a variable set in the environment, but its value NULL,
probably it means its value being an empty string.
For the "#!@+*%&" value you shouldn't rely on anything, certainly no standard
requires any specific behaviour and the man page is not what determines glibc
behaviour.
But I'll post a patch that will in that case make the offset 0 rather than -1L.

-- 


http://sources.redhat.com/bugzilla/show_bug.cgi?id=601

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/601] mktime(3) unreliable after calling tzset(3) when TZ is NULL or bogus

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-06 22:50 -------
Subject: Bug 601

CVSROOT:	/cvs/glibc
Module name:	libc
Changes by:	drepper@sources.redhat.com	2004-12-06 22:50:41

Modified files:
	time           : tzset.c 

Log message:
	(tzset_internal): If + or - is seen, but no offset after it, reset
	offset to 0.  [BZ #601]

Patches:
http://sources.redhat.com/cgi-bin/cvsweb.cgi/libc/time/tzset.c.diff?cvsroot=glibc&r1=1.78&r2=1.79



-- 


http://sources.redhat.com/bugzilla/show_bug.cgi?id=601

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/601] mktime(3) unreliable after calling tzset(3) when TZ is NULL or bogus

[aj at suse dot de]

------- Additional Comments From aj at suse dot de  2004-12-18 15:48 -------
The issue has been resolved.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://sources.redhat.com/bugzilla/show_bug.cgi?id=601

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.