public inbox for libc-hacker@sourceware.org
 help / color / mirror / Atom feed
* [PATCH] tzfile.c fixes
@ 2007-10-16 11:34 Jakub Jelinek
  2007-10-16 15:31 ` Jakub Jelinek
  0 siblings, 1 reply; 2+ messages in thread
From: Jakub Jelinek @ 2007-10-16 11:34 UTC (permalink / raw)
  To: Ulrich Drepper; +Cc: Glibc hackers

Hi!

TZ=Asia/Tokyo date
since last tzfile.c changes shows 1hr more than actually is and JDT
for 32-bit date (and correct time and timezone for 64-bit date).
The difference is that for 32-bit libc we don't read the TZif2 part
of the zoneinfo file nor the POSIX string in it, and after the use_last
label the new code sets i to num_transitions - 1 and at found it
looks at i - 1 (i.e. num_transitions - 2).

The following patch fixes that, as well as assertion failure
on tst-mktime2 on 32-bit - e.g. for America/Sao_Paulo if time is in between
last transition and 0x7fffffff and last transition is dst, then
__tzname[0] == NULL, __tzname[1] = "something" and num_types > 1.
Generally, the forward search doesn't have to find the other __tzname
if i is sufficiently close to num_transitions, so I think it is best to
search backwards if the forward search doesn't help.

2007-10-16  Jakub Jelinek  <jakub@redhat.com>

	* time/tzfile.c (__tzfile_compute): For use_last case set
	i to num_transition rather than num_transitions - 1.  If forward
	search for other __tzname didn't find it, search backwards.

--- libc/time/tzfile.c.jj	2007-10-16 09:50:03.000000000 +0200
+++ libc/time/tzfile.c	2007-10-16 13:10:25.000000000 +0200
@@ -596,7 +596,7 @@ __tzfile_compute (time_t timer, int use_
 	  if (tzspec == NULL)
 	    {
 	    use_last:
-	      i = num_transitions - 1;
+	      i = num_transitions;
 	      goto found;
 	    }
 
@@ -666,7 +666,8 @@ __tzfile_compute (time_t timer, int use_
 	  i = hi;
 
 	found:
-	  /* assert (timer >= transitions[i - 1] && timer < transitions[i]); */
+	  /* assert (timer >= transitions[i - 1]
+		     && (i == num_transitions || timer < transitions[i]));  */
 	  __tzname[types[type_idxs[i - 1]].isdst]
 	    = __tzstring (&zone_names[types[type_idxs[i - 1]].idx]);
 	  size_t j = i;
@@ -681,12 +682,30 @@ __tzfile_compute (time_t timer, int use_
 		  __tzname[dst] = __tzstring (&zone_names[idx]);
 
 		  if (__tzname[1 - dst] != NULL)
-		    break;
+		    goto got_tznames;
 		}
 
 	      ++j;
 	    }
+	  j = i - 1;
+	  while (j > 0)
+	    {
+	      int type = type_idxs[--j];
+	      int dst = types[type].isdst;
+	      int idx = types[type].idx;
+
+	      if (__tzname[dst] == NULL)
+		{
+		  __tzname[dst] = __tzstring (&zone_names[idx]);
+
+		  if (__tzname[1 - dst] != NULL)
+		    goto got_tznames;
+		}
+	    }
+	  __tzname[1 - types[type_idxs[i - 1]].isdst]
+	    = __tzname[types[type_idxs[i - 1]].isdst];
 
+	got_tznames:
 	  i = type_idxs[i - 1];
 	}
 

	Jakub

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2007-10-16 15:31 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-10-16 11:34 [PATCH] tzfile.c fixes Jakub Jelinek
2007-10-16 15:31 ` Jakub Jelinek

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).