10:59 a.m.
Happens in both xemacs-21.4 and latest XEmacs beta on Windows.
The problem is that list_to_time allows negative integers at the head of
the time list. Hence a negative time is returned. In the case localtime
on Windows returns NULL. The code does not check for it.
The patch below fixes the problem. It was tested on Windows 2000.
Nick.
12:13 p.m.
On Fri, 2002-03-29 at 12:59, Nick V. Pakoulin wrote:
Happens in both xemacs-21.4 and latest XEmacs beta on Windows.
The problem is that list_to_time allows negative integers at the head of
the time list. Hence a negative time is returned. In the case localtime
on Windows returns NULL. The code does not check for it.
The patch below fixes the problem. It was tested on Windows 2000.
[...]
I didn't test the patch, but 21.4.6 on Linux (from RedHat's rawhide)
doesn't crash:
M-x eval-expr
(format-time-string "%s" '(-1 0 0))
"4294901760"
--
Ville Skyttä
ville.skytta(a)xemacs.org
12:37 p.m.
Please don't post general commentary to xemacs-patches. redirected to
xemacs-beta (only).
>>>> "Ville" == Ville Skyttä
<ville.skytta(a)xemacs.org> writes:
Ville> On Fri, 2002-03-29 at 12:59, Nick V. Pakoulin wrote:
> Happens in both xemacs-21.4 and latest XEmacs beta on Windows.
> The problem is that list_to_time allows negative integers at
> the head of the time list. Hence a negative time is returned.
> In the case localtime on Windows returns NULL. The code does
> not check for it.
>
> The patch below fixes the problem. It was tested on Windows
> 2000.
Ville> I didn't test the patch, but 21.4.6 on Linux (from RedHat's
Ville> rawhide) doesn't crash:
That's right. This is a windows-only issue (bug?). I thought it was
addressed a long time ago, though. It's possible that the change that
Simon J and I made (released in 21.4.6) undid that, though.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Don't ask how you can "do" free software business;
ask what your business can "do for" free software.
12:42 p.m.
From MSDN Library:
localtime returns a pointer to the
structure result. If the value in
timer represents a date before midnight, January 1, 1970, localtime
returns NULL.
The same %^&* happens with gmtime, ctime. All calls of gmtime and ctime
that I found are protected by checks whether the result is sero (NULL).
But calls to localtime are not protected.
Nick.
intro: "VS" == Ville Skytt <Ville> writes:
VS> On Fri, 2002-03-29 at 12:59, Nick V. Pakoulin wrote:
> Happens in both xemacs-21.4 and latest XEmacs beta on Windows.
The
> problem is that list_to_time allows negative integers at the head of
> the time list. Hence a negative time is returned. In the case
> localtime on Windows returns NULL. The code does not check for it.
>
> The patch below fixes the problem. It was tested on Windows 2000.
VS> [...]
VS> I didn't test the patch, but 21.4.6 on Linux (from RedHat's rawhide)
VS> doesn't crash:
VS> M-x eval-expr (format-time-string "%s" '(-1 0 0))
"4294901760"
VS> -- Ville Skyttä ville.skytta(a)xemacs.org
12:48 p.m.
New subject: [PATCH] Crash on (format-time-string "%s" '(-1 0 0)) and (decode-time '(-1 0)) on Windows
`decode-time' also suffers from windows implementation of locatime. The
patch fixes both crashes.
Index: editfns.c
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/src/editfns.c,v
retrieving revision 1.36
diff -r1.36 editfns.c
1070a1071
struct tm * tm;
1075c1076,1079
< if (! lisp_to_time (time_, &value))
---
if (! lisp_to_time (time_, &value) ||
/* Might happen on MS Windows */
! (tm = localtime (&value))
)
1091c1095
< localtime (&value))
---
tm)
1120c1124,1127
< if (! lisp_to_time (specified_time, &time_spec))
---
if (! lisp_to_time (specified_time, &time_spec) ||
/* Might happen on MS Windows */
! (decoded_time = localtime (&time_spec))
)
1123d1129
< decoded_time = localtime (&time_spec);
9:04 p.m.
New subject: [PATCH] Crash on (format-time-string "%s" '(-1 0 0)) and (decode-time '(-1 0)) on Windows
Please provide a unified diff format patch (-u) and a ChangeLog entry. This
patch failed to apply for me.
Thanks
andy
-----Original Message-----
From: xemacs-beta-admin(a)xemacs.org
[mailto:xemacs-beta-admin＠xemacs.org]On Behalf Of Nick V. Pakoulin
Sent: Friday, March 29, 2002 4:49 AM
To: xemacs-beta(a)xemacs.org
Cc: xemacs-patches(a)xemacs.org
Subject: [PATCH] Crash on (format-time-string "%s" '(-1 0 0)) and
(decode-time '(-1 0)) on Windows
`decode-time' also suffers from windows implementation of locatime. The
patch fixes both crashes.
Index: editfns.c
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/src/editfns.c,v
retrieving revision 1.36
diff -r1.36 editfns.c
1070a1071
> struct tm * tm;
1075c1076,1079
< if (! lisp_to_time (time_, &value))
---
> if (! lisp_to_time (time_, &value) ||
> /* Might happen on MS Windows */
> ! (tm = localtime (&value))
> )
1091c1095
< localtime (&value))
---
> tm)
1120c1124,1127
< if (! lisp_to_time (specified_time, &time_spec))
---
> if (! lisp_to_time (specified_time, &time_spec) ||
> /* Might happen on MS Windows */
> ! (decoded_time = localtime (&time_spec))
> )
1123d1129
< decoded_time = localtime (&time_spec);
4:22 p.m.
New subject: [AC21.4 R21.5] Crash on (format-time-string "%s" '(-1 0 0)) and (decode-time '(-1 0)) on Windows
APPROVE COMMIT 21.4 RECOMMEND 21.5
Note that the patch presumably won't apply to 21.5, as the
implementation has been Mule-ized there.
>>>> "Nick" == Nick V Pakoulin
<npak(a)ispras.ru> writes:
Nick> `decode-time' also suffers from windows implementation of
Nick> locatime. The patch fixes both crashes.
The check for NULL is also correct for glibc, although I don't think
it can be triggered.
The current implementation of lisp_time doesn't look 64-bit clean, but
I'm no expert.
Index: src/ChangeLog
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/src/ChangeLog,v
retrieving revision 1.290.2.20.2.2
diff -u -U0 -r1.290.2.20.2.2 ChangeLog
--- src/ChangeLog 9 May 2002 19:37:37 -0000 1.290.2.20.2.2
+++ src/ChangeLog 13 May 2002 15:18:36 -0000
＠＠ -0,0 +1,20 ＠＠
+2002-05-14 Stephen J. Turnbull <stephen(a)xemacs.org>
+
+ * editfns.c (Fdecode_time):
+ (Fformat_time_string):
+ Check for invalid time. Thanks to Nick Pakoulin <npak(a)ispras.ru>.
+
Index: src/editfns.c
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/src/editfns.c,v
retrieving revision 1.27.2.3
diff -u -r1.27.2.3 editfns.c
--- src/editfns.c 17 Dec 2001 05:49:24 -0000 1.27.2.3
+++ src/editfns.c 13 May 2002 15:18:46 -0000
＠＠ -1107,10 +1107,11 ＠＠
{
time_t value;
size_t size;
+ struct tm * tm;
CHECK_STRING (format_string);
- if (! lisp_to_time (time_, &value))
+ if (! lisp_to_time (time_, &value) || ! (tm = localtime (&value)))
error ("Invalid time specification");
/* This is probably enough. */
＠＠ -1122,7 +1123,7 ＠＠
*buf = 1;
if (emacs_strftime (buf, size,
(const char *) XSTRING_DATA (format_string),
- localtime (&value))
+ tm)
|| !*buf)
return build_ext_string (buf, Qbinary);
/* If buffer was too small, make it bigger. */
＠＠ -1151,10 +1152,10 ＠＠
struct tm *decoded_time;
Lisp_Object list_args[9];
- if (! lisp_to_time (specified_time, &time_spec))
+ if (! lisp_to_time (specified_time, &time_spec)
+ || ! (decoded_time = localtime (&time_spec)))
error ("Invalid time specification");
- decoded_time = localtime (&time_spec);
list_args[0] = make_int (decoded_time->tm_sec);
list_args[1] = make_int (decoded_time->tm_min);
list_args[2] = make_int (decoded_time->tm_hour);
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
My nostalgia for Icon makes me forget about any of the bad things. I don't
have much nostalgia for Perl, so its faults I remember. Scott Gilbert c.l.py
8229
days inactive
8274
days old
6 comments
4 participants
participants (4)
-
Andy Piper -
Nickolay Pakoulin -
Stephen J. Turnbull -
Ville Skyttä