[v3] tzset robustness [BZ#17715]

Message ID 54E46D70.7000601@redhat.com
State Committed
Headers

Commit Message

Florian Weimer Feb. 18, 2015, 10:46 a.m. UTC
  On 02/16/2015 11:51 PM, Paul Eggert wrote:
> Florian Weimer wrote:
>> So I'm not sure what to do here.  Get rid of the alloca?  That's going
>> to be more difficult to review.
> 
> I haven't read the code carefully, but if the only reason for the alloca
> is to have a temporary string that one can munge by storing '\0' bytes
> at strategic locations, then I presume that one could rewrite the code
> to avoid the need to make a temporary copy,

Indeed.  I introduced __tzstring_len to avoid the need for the copy, and
broke down __tzset_parse_tz into several smaller functions.  Hopefully,
the control flow is more transparent.

>> I think TZNAME_MAX only applies to the “PST” part.
> 
> True, but I don't see how this contradicts what I wrote.
> 
> There are two limits here: the length limit for the entire TZ string
> (which POSIX does not allow a tight limit for), and the length limit for
> the time zone abbreviation (which is less than the TZ string limit, and
> which POSIX allows a tight limit for).  Currently glibc does not impose
> much of a limit on either quantity.  It sounds like you're thinking of
> limiting the latter to 255 bytes. If so, this new restriction should be
> done consistently and correctly, e.g., by altering how sysconf
> (_SC_TZNAME_MAX) behaves.

Wow, sysconf (_SZ_TZNAME_MAX) is really weird.  Instead of reporting
there is no limit, it returns the maximum of the name lengths seen so far.

> Currently neither glibc nor Solaris impose a tight limit on time zone
> abbreviation length, so this would be an incompatible change.  It
> probably wouldn't affect much real-world code, though.

I think the focus should be on TZ scrubbing anyway, but I see that
current consensus is against that.
  

Comments

Florian Weimer March 4, 2015, 12:22 p.m. UTC | #1
On 02/18/2015 11:46 AM, Florian Weimer wrote:
> On 02/16/2015 11:51 PM, Paul Eggert wrote:
>> Florian Weimer wrote:
>>> So I'm not sure what to do here.  Get rid of the alloca?  That's going
>>> to be more difficult to review.
>>
>> I haven't read the code carefully, but if the only reason for the alloca
>> is to have a temporary string that one can munge by storing '\0' bytes
>> at strategic locations, then I presume that one could rewrite the code
>> to avoid the need to make a temporary copy,
> 
> Indeed.  I introduced __tzstring_len to avoid the need for the copy, and
> broke down __tzset_parse_tz into several smaller functions.  Hopefully,
> the control flow is more transparent.

Ping?

Should I split this up into multiple parts?

One of the issues addressed was previously raised on oss-security:

  <http://www.openwall.com/lists/oss-security/2014/12/14/1>
  <https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=772705>
  
Florian Weimer March 16, 2015, 3 p.m. UTC | #2
On 03/04/2015 01:22 PM, Florian Weimer wrote:
> On 02/18/2015 11:46 AM, Florian Weimer wrote:
>> On 02/16/2015 11:51 PM, Paul Eggert wrote:
>>> Florian Weimer wrote:
>>>> So I'm not sure what to do here.  Get rid of the alloca?  That's going
>>>> to be more difficult to review.
>>>
>>> I haven't read the code carefully, but if the only reason for the alloca
>>> is to have a temporary string that one can munge by storing '\0' bytes
>>> at strategic locations, then I presume that one could rewrite the code
>>> to avoid the need to make a temporary copy,
>>
>> Indeed.  I introduced __tzstring_len to avoid the need for the copy, and
>> broke down __tzset_parse_tz into several smaller functions.  Hopefully,
>> the control flow is more transparent.
> 
> Ping?
> 
> Should I split this up into multiple parts?
> 
> One of the issues addressed was previously raised on oss-security:
> 
>   <http://www.openwall.com/lists/oss-security/2014/12/14/1>
>   <https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=772705>

Ping?
  
Florian Weimer April 24, 2015, 3:36 p.m. UTC | #3
On 02/18/2015 11:46 AM, Florian Weimer wrote:
> On 02/16/2015 11:51 PM, Paul Eggert wrote:
>> Florian Weimer wrote:
>>> So I'm not sure what to do here.  Get rid of the alloca?  That's going
>>> to be more difficult to review.
>>
>> I haven't read the code carefully, but if the only reason for the alloca
>> is to have a temporary string that one can munge by storing '\0' bytes
>> at strategic locations, then I presume that one could rewrite the code
>> to avoid the need to make a temporary copy,
> 
> Indeed.  I introduced __tzstring_len to avoid the need for the copy, and
> broke down __tzset_parse_tz into several smaller functions.  Hopefully,
> the control flow is more transparent.

I've committed this now, with a few whitespace fixes.
  
H.J. Lu April 26, 2015, 1:50 p.m. UTC | #4
On Fri, Apr 24, 2015 at 8:36 AM, Florian Weimer <fweimer@redhat.com> wrote:
> On 02/18/2015 11:46 AM, Florian Weimer wrote:
>> On 02/16/2015 11:51 PM, Paul Eggert wrote:
>>> Florian Weimer wrote:
>>>> So I'm not sure what to do here.  Get rid of the alloca?  That's going
>>>> to be more difficult to review.
>>>
>>> I haven't read the code carefully, but if the only reason for the alloca
>>> is to have a temporary string that one can munge by storing '\0' bytes
>>> at strategic locations, then I presume that one could rewrite the code
>>> to avoid the need to make a temporary copy,
>>
>> Indeed.  I introduced __tzstring_len to avoid the need for the copy, and
>> broke down __tzset_parse_tz into several smaller functions.  Hopefully,
>> the control flow is more transparent.
>
> I've committed this now, with a few whitespace fixes.
>

This caused:

https://sourceware.org/bugzilla/show_bug.cgi?id=18333
  

Patch

From d04aaa5f5873d78b4c40c9f226cb89f612c90156 Mon Sep 17 00:00:00 2001
From: Florian Weimer <fweimer@redhat.com>
Date: Wed, 18 Feb 2015 11:41:24 +0100
Subject: [PATCH] Make time zone file parser more robust [BZ #17715]

2015-02-18  Florian Weimer  <fweimer@redhat.com>

	[BZ #17715]
	* time/tzfile.c (__tzfile_read): Check for large values of
	tzh_ttisstdcnt and tzh_ttisgmtcnt.  Use malloc instead of alloca.
	* time/tzset.c (__tzstring_len): New function, based on the old
	__tzstring function.
	(__tzstring): Call __tzstring_len.
	(parse_tzname): New helper function extracted from
	__tzset_parse_tz.  Call __tzstring_len, without making a copy of
	the input string.
	(parse_offset): New helper function extracted from
	__tzset_parse_tz.  Replace switch with fallthrough with
	initialization before sscanf.
	(parse_rule): Likewise.
	(__tzset_parse_tz): Rewrite using the new helper functions.  Use
	new-style function definition.
	* timezone/Makefile (tests): Add tst-tzset.
	(tst-tzset.out): Dependencies on time zone files.
	(tst-tzset-ENV): Set TZDIR.
	(testdata/XT%): Copy crafted time zone files.
	* timezone/README: Mention crafted time zone files.
	* timezone/testdata/XT1, timezone/testdata/XT2,
	timezone/testdata/XT3, timezone/testdata/XT4: New time zone test
	files.
	* timezone/tst-tzset.c: New test.

diff --git a/NEWS b/NEWS
index 781f7a7..9883eb6 100644
--- a/NEWS
+++ b/NEWS
@@ -9,8 +9,14 @@  Version 2.22
 
 * The following bugs are resolved with this release:
 
-  4719, 15467, 15790, 16560, 17569, 17792, 17912, 17932, 17944, 17949,
-  17964, 17965, 17967, 17969.
+  4719, 15467, 15790, 16560, 17569, 17715, 17792, 17912, 17932, 17944,
+  17949, 17964, 17965, 17967, 17969.
+
+* The time zone file parser has been made more robust against crafted time
+  zone files, avoiding heap buffer overflows related to the processing of
+  the tzh_ttisstdcnt and tzh_ttisgmtcnt fields, and a stack overflow due to
+  large time zone data files.  Overly long time zone specifiers in the TZ
+  variable no longer result in stack overflows and crashes.
 
 Version 2.21
 
diff --git a/time/tzfile.c b/time/tzfile.c
index bcb408f..46d4fc7 100644
--- a/time/tzfile.c
+++ b/time/tzfile.c
@@ -200,6 +200,9 @@  __tzfile_read (const char *file, size_t extra, char **extrap)
   num_isstd = (size_t) decode (tzhead.tzh_ttisstdcnt);
   num_isgmt = (size_t) decode (tzhead.tzh_ttisgmtcnt);
 
+  if (__glibc_unlikely (num_isstd > num_types || num_isgmt > num_types))
+    goto lose;
+
   /* For platforms with 64-bit time_t we use the new format if available.  */
   if (sizeof (time_t) == 8 && trans_width == 4
       && tzhead.tzh_version[0] != '\0')
@@ -434,13 +437,21 @@  __tzfile_read (const char *file, size_t extra, char **extrap)
 	goto lose;
 
       tzspec_len = st.st_size - off - 1;
-      char *tzstr = alloca (tzspec_len);
+      if (tzspec_len == 0)
+	goto lose;
+      char *tzstr = malloc (tzspec_len);
+      if (tzstr == NULL)
+	goto lose;
       if (getc_unlocked (f) != '\n'
 	  || (__fread_unlocked (tzstr, 1, tzspec_len - 1, f)
 	      != tzspec_len - 1))
-	goto lose;
+	{
+	  free (tzstr);
+	  goto lose;
+	}
       tzstr[tzspec_len - 1] = '\0';
       tzspec = __tzstring (tzstr);
+      free (tzstr);
     }
 
   /* Don't use an empty TZ string.  */
diff --git a/time/tzset.c b/time/tzset.c
index 8bc7a2e..9948c70 100644
--- a/time/tzset.c
+++ b/time/tzset.c
@@ -18,6 +18,7 @@ 
 #include <ctype.h>
 #include <errno.h>
 #include <bits/libc-lock.h>
+#include <stdbool.h>
 #include <stddef.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -82,15 +83,14 @@  struct tzstring_l
 
 static struct tzstring_l *tzstring_list;
 
-/* Allocate a permanent home for S.  It will never be moved or deallocated,
-   but may share space with other strings.
-   Don't modify the returned string. */
-char *
-__tzstring (const char *s)
+/* Allocate a permanent home for the first LEN characters of S.  It
+   will never be moved or deallocated, but may share space with other
+   strings.  Don't modify the returned string. */
+static char *
+__tzstring_len (const char *s, size_t len)
 {
   char *p;
   struct tzstring_l *t, *u, *new;
-  size_t len = strlen (s);
 
   /* Walk the list and look for a match.  If this string is the same
      as the end of an already-allocated string, it can share space. */
@@ -98,7 +98,7 @@  __tzstring (const char *s)
     if (len <= t->len)
       {
 	p = &t->data[t->len - len];
-	if (strcmp (s, p) == 0)
+	if (memcmp (s, p, len) == 0)
 	  return p;
       }
 
@@ -109,7 +109,8 @@  __tzstring (const char *s)
 
   new->next = NULL;
   new->len = len;
-  strcpy (new->data, s);
+  memcpy (new->data, s, len);
+  new->data[len] = '\0';
 
   if (u)
     u->next = new;
@@ -118,6 +119,15 @@  __tzstring (const char *s)
 
   return new->data;
 }
+
+/* Allocate a permanent home for S.  It will never be moved or
+   deallocated, but may share space with other strings.  Don't modify
+   the returned string. */
+char *
+__tzstring (const char *s)
+{
+  return __tzstring_len (s, strlen (s));
+}
 
 /* Maximum length of a timezone name.  tzset_internal keeps this up to date
    (never decreasing it) when ! __use_tzfile.
@@ -164,234 +174,215 @@  compute_offset (unsigned int ss, unsigned int mm, unsigned int hh)
   return min (ss, 59) + min (mm, 59) * 60 + min (hh, 24) * 60 * 60;
 }
 
-
-/* Parse the POSIX TZ-style string.  */
-void
-__tzset_parse_tz (tz)
-     const char *tz;
+/* Parses the time zone name at *TZP, and writes a pointer to an
+   interned string to tz_rules[WHICHRULE].name.  On success, advances
+   *TZP, and returns true.  Returns false otherwise.  */
+static bool
+parse_tzname (const char **tzp, int whichrule)
 {
-  unsigned short int hh, mm, ss;
-
-  /* Clear out old state and reset to unnamed UTC.  */
-  memset (tz_rules, '\0', sizeof tz_rules);
-  tz_rules[0].name = tz_rules[1].name = "";
-
-  /* Get the standard timezone name.  */
-  char *tzbuf = strdupa (tz);
-
-  int consumed;
-  if (sscanf (tz, "%[A-Za-z]%n", tzbuf, &consumed) != 1)
+  const char *start = *tzp;
+  const char *p = start;
+  while (('a' <= *p && *p <= 'z')
+	 || ('A' <= *p && *p <= 'Z'))
+      ++p;
+  size_t len = p - start;
+  if (len < 3)
     {
-      /* Check for the quoted version.  */
-      char *wp = tzbuf;
-      if (__glibc_unlikely (*tz++ != '<'))
-	goto out;
-
-      while (isalnum (*tz) || *tz == '+' || *tz == '-')
-	*wp++ = *tz++;
-      if (__glibc_unlikely (*tz++ != '>' || wp - tzbuf < 3))
-	goto out;
-      *wp = '\0';
+      p = *tzp;
+      if (__glibc_unlikely (*p++ != '<'))
+	return false;
+      start = p;
+      while (('a' <= *p && *p <= 'z')
+	     || ('A' <= *p && *p <= 'Z')
+	     || ('0' <= *p && *p <= '9')
+	     || *p == '+' || *p == '-')
+	++p;
+      len = p - start;
+      if (*p++ != '>' || len < 3)
+	return false;
     }
-  else if (__glibc_unlikely (consumed < 3))
-    goto out;
-  else
-    tz += consumed;
-
-  tz_rules[0].name = __tzstring (tzbuf);
+  tz_rules[whichrule].name = __tzstring_len (start, len);
+  *tzp = p;
+  return true;
+}
 
-  /* Figure out the standard offset from UTC.  */
-  if (*tz == '\0' || (*tz != '+' && *tz != '-' && !isdigit (*tz)))
-    goto out;
+/* Parses the time zone offset at *TZP, and writes it to
+   tz_rules[WHICHRULE].offset.  Returns true if the parse was
+   successful.  */
+static bool
+parse_offset (const char **tzp, int whichrule)
+{
+  const char *tz = *tzp;
+  if (whichrule == 0
+      && (*tz == '\0' || (*tz != '+' && *tz != '-' && !isdigit (*tz))))
+    return false;
 
+  long sign;
   if (*tz == '-' || *tz == '+')
-    tz_rules[0].offset = *tz++ == '-' ? 1L : -1L;
+    sign = *tz++ == '-' ? 1L : -1L;
   else
-    tz_rules[0].offset = -1L;
-  switch (sscanf (tz, "%hu%n:%hu%n:%hu%n",
-		  &hh, &consumed, &mm, &consumed, &ss, &consumed))
-    {
-    default:
-      tz_rules[0].offset = 0;
-      goto out;
-    case 1:
-      mm = 0;
-    case 2:
-      ss = 0;
-    case 3:
-      break;
-    }
-  tz_rules[0].offset *= compute_offset (ss, mm, hh);
-  tz += consumed;
-
-  /* Get the DST timezone name (if any).  */
-  if (*tz != '\0')
-    {
-      if (sscanf (tz, "%[A-Za-z]%n", tzbuf, &consumed) != 1)
-	{
-	  /* Check for the quoted version.  */
-	  char *wp = tzbuf;
-	  const char *rp = tz;
-	  if (__glibc_unlikely (*rp++ != '<'))
-	    /* Punt on name, set up the offsets.  */
-	    goto done_names;
-
-	  while (isalnum (*rp) || *rp == '+' || *rp == '-')
-	    *wp++ = *rp++;
-	  if (__glibc_unlikely (*rp++ != '>' || wp - tzbuf < 3))
-	    /* Punt on name, set up the offsets.  */
-	    goto done_names;
-	  *wp = '\0';
-	  tz = rp;
-	}
-      else if (__glibc_unlikely (consumed < 3))
-	/* Punt on name, set up the offsets.  */
-	goto done_names;
+    sign = -1L;
+  *tzp = tz;
+
+  unsigned short int hh;
+  unsigned short mm = 0;
+  unsigned short ss = 0;
+  int consumed = 0;
+  if (sscanf (tz, "%hu%n:%hu%n:%hu%n",
+	      &hh, &consumed, &mm, &consumed, &ss, &consumed) > 0)
+    tz_rules[whichrule].offset = sign * compute_offset (ss, mm, hh);
+  else
+    /* Nothing could be parsed. */
+    if (whichrule == 0)
+      {
+	/* Standard time defaults to offset zero.  */
+	tz_rules[0].offset = 0;
+	return false;
+      }
       else
-	tz += consumed;
+	/* DST defaults to one hour later than standard time.  */
+	tz_rules[1].offset = tz_rules[0].offset + (60 * 60);
+  *tzp = tz + consumed;
+  return true;
+}
 
-      tz_rules[1].name = __tzstring (tzbuf);
+/* Parses the standard <-> DST rules at *TZP.  Updates
+   tz_rule[WHICHRULE].  On success, advances *TZP and returns true.
+   Otherwise, returns false.  */
+static bool
+parse_rule (const char **tzp, int whichrule)
+{
+  const char *tz = *tzp;
+  tz_rule *tzr = &tz_rules[whichrule];
 
-      /* Figure out the DST offset from GMT.  */
-      if (*tz == '-' || *tz == '+')
-	tz_rules[1].offset = *tz++ == '-' ? 1L : -1L;
-      else
-	tz_rules[1].offset = -1L;
+  /* Ignore comma to support string following the incorrect
+     specification in early POSIX.1 printings.  */
+  tz += *tz == ',';
 
-      switch (sscanf (tz, "%hu%n:%hu%n:%hu%n",
-		      &hh, &consumed, &mm, &consumed, &ss, &consumed))
+  /* Get the date of the change.  */
+  if (*tz == 'J' || isdigit (*tz))
+    {
+      char *end;
+      tzr->type = *tz == 'J' ? J1 : J0;
+      if (tzr->type == J1 && !isdigit (*++tz))
+	return false;
+      unsigned long int d = strtoul (tz, &end, 10);
+      if (end == tz || d > 365)
+	return false;
+      if (tzr->type == J1 && d == 0)
+	return false;
+      tzr->d = d;
+      tz = end;
+    }
+  else if (*tz == 'M')
+    {
+      tzr->type = M;
+      int consumed;
+      if (sscanf (tz, "M%hu.%hu.%hu%n",
+		  &tzr->m, &tzr->n, &tzr->d, &consumed) != 3
+	  || tzr->m < 1 || tzr->m > 12
+	  || tzr->n < 1 || tzr->n > 5 || tzr->d > 6)
+	return false;
+      tz += consumed;
+    }
+  else if (*tz == '\0')
+    {
+      /* Daylight time rules in the U.S. are defined in the U.S. Code,
+	 Title 15, Chapter 6, Subchapter IX - Standard Time.  These
+	 dates were established by Congress in the Energy Policy Act
+	 of 2005 [Pub. L. no. 109-58, 119 Stat 594 (2005)].
+	 Below is the equivalent of "M3.2.0,M11.1.0" [/2 not needed
+	 since 2:00AM is the default].  */
+      tzr->type = M;
+      if (tzr == &tz_rules[0])
 	{
-	default:
-	  /* Default to one hour later than standard time.  */
-	  tz_rules[1].offset = tz_rules[0].offset + (60 * 60);
-	  break;
-
-	case 1:
-	  mm = 0;
-	case 2:
-	  ss = 0;
-	case 3:
-	  tz_rules[1].offset *= compute_offset (ss, mm, hh);
-	  tz += consumed;
-	  break;
+	  tzr->m = 3;
+	  tzr->n = 2;
+	  tzr->d = 0;
 	}
-      if (*tz == '\0' || (tz[0] == ',' && tz[1] == '\0'))
+      else
 	{
-	  /* There is no rule.  See if there is a default rule file.  */
-	  __tzfile_default (tz_rules[0].name, tz_rules[1].name,
-			    tz_rules[0].offset, tz_rules[1].offset);
-	  if (__use_tzfile)
-	    {
-	      free (old_tz);
-	      old_tz = NULL;
-	      return;
-	    }
+	  tzr->m = 11;
+	  tzr->n = 1;
+	  tzr->d = 0;
 	}
     }
   else
+    return false;
+
+  if (*tz != '\0' && *tz != '/' && *tz != ',')
+    return false;
+  else if (*tz == '/')
     {
-      /* There is no DST.  */
-      tz_rules[1].name = tz_rules[0].name;
-      tz_rules[1].offset = tz_rules[0].offset;
-      goto out;
+      /* Get the time of day of the change.  */
+      int negative;
+      ++tz;
+      if (*tz == '\0')
+	return false;
+      negative = *tz == '-';
+      tz += negative;
+      /* Default to 2:00 AM.  */
+      unsigned short hh = 2;
+      unsigned short mm = 0;
+      unsigned short ss = 0;
+      int consumed = 0;
+      sscanf (tz, "%hu%n:%hu%n:%hu%n",
+	      &hh, &consumed, &mm, &consumed, &ss, &consumed);;
+      tz += consumed;
+      tzr->secs = (negative ? -1 : 1) * ((hh * 60 * 60) + (mm * 60) + ss);
     }
+  else
+    /* Default to 2:00 AM.  */
+    tzr->secs = 2 * 60 * 60;
 
- done_names:
-  /* Figure out the standard <-> DST rules.  */
-  for (unsigned int whichrule = 0; whichrule < 2; ++whichrule)
-    {
-      tz_rule *tzr = &tz_rules[whichrule];
+  tzr->computed_for = -1;
+  *tzp = tz;
+  return true;
+}
 
-      /* Ignore comma to support string following the incorrect
-	 specification in early POSIX.1 printings.  */
-      tz += *tz == ',';
+/* Parse the POSIX TZ-style string.  */
+void
+__tzset_parse_tz (const char *tz)
+{
+  /* Clear out old state and reset to unnamed UTC.  */
+  memset (tz_rules, '\0', sizeof tz_rules);
+  tz_rules[0].name = tz_rules[1].name = "";
 
-      /* Get the date of the change.  */
-      if (*tz == 'J' || isdigit (*tz))
-	{
-	  char *end;
-	  tzr->type = *tz == 'J' ? J1 : J0;
-	  if (tzr->type == J1 && !isdigit (*++tz))
-	    goto out;
-	  unsigned long int d = strtoul (tz, &end, 10);
-	  if (end == tz || d > 365)
-	    goto out;
-	  if (tzr->type == J1 && d == 0)
-	    goto out;
-	  tzr->d = d;
-	  tz = end;
-	}
-      else if (*tz == 'M')
-	{
-	  tzr->type = M;
-	  if (sscanf (tz, "M%hu.%hu.%hu%n",
-		      &tzr->m, &tzr->n, &tzr->d, &consumed) != 3
-	      || tzr->m < 1 || tzr->m > 12
-	      || tzr->n < 1 || tzr->n > 5 || tzr->d > 6)
-	    goto out;
-	  tz += consumed;
-	}
-      else if (*tz == '\0')
+  /* Get the standard timezone name.  */
+  if (parse_tzname (&tz, 0) && parse_offset (&tz, 0))
+    {
+      /* Get the DST timezone name (if any).  */
+      if (*tz != '\0')
 	{
-         /* Daylight time rules in the U.S. are defined in the
-            U.S. Code, Title 15, Chapter 6, Subchapter IX - Standard
-            Time.  These dates were established by Congress in the
-            Energy Policy Act of 2005 [Pub. L. no. 109-58, 119 Stat 594
-            (2005)].
-	    Below is the equivalent of "M3.2.0,M11.1.0" [/2 not needed
-	    since 2:00AM is the default].  */
-	  tzr->type = M;
-	  if (tzr == &tz_rules[0])
+	  if (parse_tzname (&tz, 1))
 	    {
-	      tzr->m = 3;
-	      tzr->n = 2;
-	      tzr->d = 0;
-	    }
-	  else
-	    {
-	      tzr->m = 11;
-	      tzr->n = 1;
-	      tzr->d = 0;
+	      parse_offset (&tz, 1);
+	      if (*tz == '\0' || (tz[0] == ',' && tz[1] == '\0'))
+		{
+		  /* There is no rule.  See if there is a default rule
+		     file.  */
+		  __tzfile_default (tz_rules[0].name, tz_rules[1].name,
+				    tz_rules[0].offset, tz_rules[1].offset);
+		  if (__use_tzfile)
+		    {
+		      free (old_tz);
+		      old_tz = NULL;
+		      return;
+		    }
+		}
 	    }
+	  /* Figure out the standard <-> DST rules.  */
+	  if (parse_rule (&tz, 0))
+	    parse_rule (&tz, 1);
 	}
       else
-	goto out;
-
-      if (*tz != '\0' && *tz != '/' && *tz != ',')
-	goto out;
-      else if (*tz == '/')
 	{
-	  /* Get the time of day of the change.  */
-	  int negative;
-	  ++tz;
-	  if (*tz == '\0')
-	    goto out;
-	  negative = *tz == '-';
-	  tz += negative;
-	  consumed = 0;
-	  switch (sscanf (tz, "%hu%n:%hu%n:%hu%n",
-			  &hh, &consumed, &mm, &consumed, &ss, &consumed))
-	    {
-	    default:
-	      hh = 2;		/* Default to 2:00 AM.  */
-	    case 1:
-	      mm = 0;
-	    case 2:
-	      ss = 0;
-	    case 3:
-	      break;
-	    }
-	  tz += consumed;
-	  tzr->secs = (negative ? -1 : 1) * ((hh * 60 * 60) + (mm * 60) + ss);
+	  /* There is no DST.  */
+	  tz_rules[1].name = tz_rules[0].name;
+	  tz_rules[1].offset = tz_rules[0].offset;
 	}
-      else
-	/* Default to 2:00 AM.  */
-	tzr->secs = 2 * 60 * 60;
-
-      tzr->computed_for = -1;
     }
 
- out:
   update_vars ();
 }
 
diff --git a/timezone/Makefile b/timezone/Makefile
index 17424b8..5f18545 100644
--- a/timezone/Makefile
+++ b/timezone/Makefile
@@ -25,7 +25,7 @@  include ../Makeconfig
 extra-objs := scheck.o ialloc.o
 
 others	:= zdump zic
-tests	:= test-tz tst-timezone
+tests	:= test-tz tst-timezone tst-tzset
 
 # pacificnew doesn't compile; if it is to be used, it should be included in
 # northamerica.
@@ -90,9 +90,11 @@  $(objpfx)tst-timezone.out: $(addprefix $(testdata)/, \
 				       Australia/Melbourne \
 				       America/Sao_Paulo Asia/Tokyo \
 				       Europe/London)
+$(objpfx)tst-tzset.out: $(addprefix $(testdata)/XT, 1 2 3 4)
 
 test-tz-ENV = TZDIR=$(testdata)
 tst-timezone-ENV = TZDIR=$(testdata)
+tst-tzset-ENV = TZDIR=$(testdata)
 
 # Note this must come second in the deps list for $(built-program-cmd) to work.
 zic-deps = $(objpfx)zic $(leapseconds) yearistype
@@ -114,6 +116,8 @@  $(testdata)/America/Sao_Paulo: southamerica $(zic-deps)
 $(testdata)/Asia/Tokyo: asia $(zic-deps)
 	$(build-testdata)
 
+$(testdata)/XT%: testdata/XT%
+	cp $< $@
 
 $(objpfx)tzselect: tzselect.ksh $(common-objpfx)config.make
 	sed -e 's|/bin/bash|$(BASH)|' \
diff --git a/timezone/README b/timezone/README
index 7a5e31c..2268f8e 100644
--- a/timezone/README
+++ b/timezone/README
@@ -15,3 +15,6 @@  version of the tzcode and tzdata packages.
 
 These packages may be found at ftp://ftp.iana.org/tz/releases/.  Commentary
 should be addressed to tz@iana.org.
+
+The subdirectory testdata contains manually edited data files for
+regression testing purposes.
diff --git a/timezone/testdata/XT1 b/timezone/testdata/XT1
new file mode 100644
index 0000000000000000000000000000000000000000..67d7ee0ba59389b4aaea0bdce15ff6bf7056c5ef
GIT binary patch
literal 127
zcmWHE%1kq2Kn4GS04TzYB+3Y6vq1O}A%;Lk2w{C7Jz#x5AR1vL!~iZJWxxdhA3F~_

literal 0
HcmV?d00001

diff --git a/timezone/testdata/XT2 b/timezone/testdata/XT2
new file mode 100644
index 0000000000000000000000000000000000000000..069189e34998662d526b3645891d515a31fd6495
GIT binary patch
literal 127
wcmWHE%1kq2zyQqufdEOA5y)nN@FPM%>O%<Y1L*<l`vK7iBOwNG0VxA60ROxXJ^%m!

literal 0
HcmV?d00001

diff --git a/timezone/testdata/XT3 b/timezone/testdata/XT3
new file mode 100644
index 0000000000000000000000000000000000000000..fbf5efff9ed5560de8c9fec46c84dc3dafc8f23f
GIT binary patch
literal 127
wcmWHE%1kq2zyJb35k@3Y5Ss<Uj|edaGC~OJ1L*<l`vK7iBOwNG0VxA60KSR`WdHyG

literal 0
HcmV?d00001

diff --git a/timezone/testdata/XT4 b/timezone/testdata/XT4
new file mode 100644
index 0000000000000000000000000000000000000000..990a9763da2c00d82ad235033005047903585b25
GIT binary patch
literal 127
wcmWHE%1kq2zyORu5dkDo5T6CYj|edVGC~OJ1L*<l`vK7iBOwNG0VxA60KRGmXaE2J

literal 0
HcmV?d00001

diff --git a/timezone/tst-tzset.c b/timezone/tst-tzset.c
new file mode 100644
index 0000000..dde03ca
--- /dev/null
+++ b/timezone/tst-tzset.c
@@ -0,0 +1,200 @@ 
+/* tzset tests with crafted time zone data.
+   Copyright (C) 2015 Free Software Foundation, Inc.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#define _GNU_SOURCE 1
+
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/resource.h>
+#include <time.h>
+#include <unistd.h>
+
+static int do_test (void);
+#define TEST_FUNCTION do_test ()
+#include "../test-skeleton.c"
+
+/* Returns the name of a large TZ file.  */
+static char *
+create_tz_file (off64_t size)
+{
+  char *path;
+  int fd = create_temp_file ("tst-tzset-", &path);
+  if (fd < 0)
+    exit (1);
+
+  // Reopen for large-file support.
+  close (fd);
+  fd = open64 (path, O_WRONLY);
+  if (fd < 0)
+    {
+      printf ("open64 (%s) failed: %m\n", path);
+      exit (1);
+    }
+  
+  static const char data[] = {
+    0x54, 0x5a, 0x69, 0x66, 0x32, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01,
+    0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01,
+    0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x58, 0x54, 0x47, 0x00, 0x00, 0x00,
+    0x54, 0x5a, 0x69, 0x66, 0x32, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01,
+    0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x01,
+    0x00, 0x00, 0x00, 0x04, 0xf8, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x58, 0x54, 0x47, 0x00, 0x00,
+    0x00, 0x0a, 0x58, 0x54, 0x47, 0x30, 0x0a
+  };
+  ssize_t ret = write (fd, data, sizeof (data));
+  if (ret < 0)
+    {
+      printf ("write failed: %m\n");
+      exit (1);
+    }
+  if ((size_t) ret != sizeof (data))
+    {
+      printf ("Short write\n");
+      exit (1);
+    }
+  if (lseek64 (fd, size, SEEK_CUR) < 0)
+    {
+      printf ("lseek failed: %m\n");
+      close (fd);
+      return NULL;
+    }
+  if (write (fd, "", 1) != 1)
+    {
+      printf ("Single-byte write failed\n");
+      close (fd);
+      return NULL;
+    }
+  if (close (fd) != 0)
+    {
+      printf ("close failed: %m\n");
+      exit (1);
+    }
+  return path;
+}
+
+static void
+test_tz_file (off64_t size)
+{
+  char *path = create_tz_file (size);
+  if (setenv ("TZ", path, 1) < 0)
+    {
+      printf ("setenv failed: %m\n");
+      exit (1);
+    }
+  tzset ();
+  free (path);
+}
+
+static int
+do_test (void)
+{
+  /* Limit the size of the process.  Otherwise, some of the tests will
+     consume a lot of resources.  */
+  {
+    struct rlimit limit;
+    if (getrlimit (RLIMIT_AS, &limit) != 0)
+      {
+	printf ("getrlimit (RLIMIT_AS) failed: %m\n");
+	return 1;
+      }
+    long target = 512 * 1024 * 1024;
+    if (limit.rlim_cur == RLIM_INFINITY || limit.rlim_cur > target)
+      {
+	limit.rlim_cur = 512 * 1024 * 1024;
+	if (setrlimit (RLIMIT_AS, &limit) != 0)
+	  {
+	    printf ("setrlimit (RLIMIT_AS) failed: %m\n");
+	    return 1;
+	  }
+      }
+  }
+  
+  int errors = 0;
+  for (int i = 1; i <= 4; ++i)
+    {
+      char tz[16];
+      snprintf (tz, sizeof (tz), "XT%d", i);
+      if (setenv ("TZ", tz, 1) < 0)
+	{
+	  printf ("setenv failed: %m\n");
+	  return 1;
+	}
+      tzset ();
+      if (strcmp (tzname[0], tz) == 0)
+	{
+	  printf ("Unexpected success for %s\n", tz);
+	  ++errors;
+	}
+    }
+
+  /* Large TZ files.  */
+
+  /* This will succeed on 64-bit architectures, and fail on 32-bit
+     architectures.  It used to crash on 32-bit.  */
+  test_tz_file (64 * 1024 * 1024);
+
+  /* This will fail on 64-bit and 32-bit architectures.  It used to
+     cause a test timeout on 64-bit and crash on 32-bit if the TZ file
+     open succeeded for some reason (it does not use O_LARGEFILE in
+     regular builds).  */
+  test_tz_file (4LL * 1024 * 1024 * 1024 - 6);
+
+  /* Large TZ variables.  */
+  {
+    size_t length = 64 * 1024 * 1024;
+    char *value = malloc (length + 1);
+    if (value == NULL)
+      {
+	puts ("malloc failed: %m");
+	return 1;
+      }
+    value[length] = '\0';
+
+    memset (value, ' ', length);
+    value[0] = 'U';
+    value[1] = 'T';
+    value[2] = 'C';
+    if (setenv ("TZ", value, 1) < 0)
+      {
+	printf ("setenv failed: %m\n");
+	return 1;
+      }
+    tzset ();
+
+    memset (value, '0', length);
+    value[0] = '<';
+    value[length - 1] = '>';
+    if (setenv ("TZ", value, 1) < 0)
+      {
+	printf ("setenv failed: %m\n");
+	return 1;
+      }
+    tzset ();
+  }
+  
+  return errors > 0;
+}
-- 
2.1.0