11:29 p.m.
Hi,
I have reported this bug once already, but apparentlyy nobody took
notice. So here it is again. This happens on a fairly regular basis on
my system. Last time it happened twice in a row after exactly same
actions (copy a line from one buffer, switch to another, go to the top
and 2 lines down, paste - boom), but unfortunately I could not repro
it after that. My feeling is that the bug occurs when I do things
really fast (which would explain why I couldn't repro it). The
assertion is below. I have no idea what the code that causes it does,
but would be willing to help somebody who knows - e.g. I can add some
checking code to the problematic place etc. In any case, this bug is
as good a showstopper as any for a "release" so at least some
attention by one of XEmacs' core writers is I think justified.
LUCY:~$xemacs &
LUCY:~$Fatal error: assertion failed, file /usr/local/src/xemacs-21.2/src/insdel.c, line
1115, ( retval) >= BI_BUF_BEG (buf) && retval <= BI_BUF_Z (buf)
[1]+ Exit 1 xemacs
XEmacs 21.2 (beta44) "Thalia" [Lucid] (i686-pc-cygwin) of Tue Feb 13 2001 on
LUCY
LUCY:~$uname -a
CYGWIN_NT-5.0 LUCY 1.1.8(0.34/3/2) 2001-01-31 10:08 i686 unknown
--
Cheers,
-Dima.
1:06 a.m.
New subject: Crash (assertion failed) in 21.2.44 - 100% reproducible
I can now reproduce this 100%. Here are the steps:
LUCY:~$xemacs --vanilla -eval "(global-set-key \"\\C-m\"
'newline-and-indent)" bbb.c
The "bbb.c" file is pretty short, here it is in its entirety:
--------------------- cut here ---------------------
//---------------------------------------------------------------------
//
//---------------------------------------------------------------------
s_cmenu_all.h_active = s_menudoc.createElement("DIV");
function vedit_init (h_vedit)
{
split_init();
cmenu_init(vedit_cmenu_populate, vedit_cmenu_execute);
window.opener.vedit_load_callback(window);
}
--------------------- cut here ---------------------
Now, go to the beginning of the 2nd line (over the 1st "/") (or to the
beginning of 3rd of 4th line - all 3 produce the crash) and press C-m
(RET). BOOM:
Fatal error: assertion failed, file /usr/local/src/xemacs-21.2/src/insdel.c, line 1115, (
retval) >= BI_BUF_BEG (buf) && retval <= BI_BUF_Z (buf)
LUCY:~$
The culprit seems to be the comment and the fact that it starts from
the very beginning of the file. Inserting a blank like above the 1st
line removes the crash. I hope somebody who knows the C code involved
can look at it. I can provide further details if needed. Thanks.
Dmitry Yaitskov <dimas(a)home.com> wrote:
Hi,
I have reported this bug once already, but apparentlyy nobody took
notice. So here it is again. This happens on a fairly regular basis on
my system. Last time it happened twice in a row after exactly same
actions (copy a line from one buffer, switch to another, go to the top
and 2 lines down, paste - boom), but unfortunately I could not repro
it after that. My feeling is that the bug occurs when I do things
really fast (which would explain why I couldn't repro it). The
assertion is below. I have no idea what the code that causes it does,
but would be willing to help somebody who knows - e.g. I can add some
checking code to the problematic place etc. In any case, this bug is
as good a showstopper as any for a "release" so at least some
attention by one of XEmacs' core writers is I think justified.
LUCY:~$xemacs &
LUCY:~$Fatal error: assertion failed, file /usr/local/src/xemacs-21.2/src/insdel.c, line
1115, ( retval) >= BI_BUF_BEG (buf) && retval <= BI_BUF_Z (buf)
[1]+ Exit 1 xemacs
XEmacs 21.2 (beta44) "Thalia" [Lucid] (i686-pc-cygwin) of Tue Feb 13 2001 on
LUCY
LUCY:~$uname -a
CYGWIN_NT-5.0 LUCY 1.1.8(0.34/3/2) 2001-01-31 10:08 i686 unknown
--
Cheers,
-Dima.
3:44 a.m.
New subject: Crash (assertion failed) in 21.2.44 - 100% reproducible
>>>> "Dmitry" == Dmitry Yaitskov
<dimas(a)home.com> writes:
Dmitry> The culprit seems to be the comment and the fact that it
Dmitry> starts from the very beginning of the file. Inserting a
Dmitry> blank like above the 1st line removes the crash. I hope
Dmitry> somebody who knows the C code involved can look at it. I
Dmitry> can provide further details if needed. Thanks.
There seems to be a fencepost error in syntax.c. You could try
Vladimir Vukicevic's patch. This is (hopefully) being reviewed by
Matt Tucker now, and may not go in in its current form, but it might
help you get some work done.
Index: syntax.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/syntax.c,v
retrieving revision 1.7.2.18
diff -u -r1.7.2.18 syntax.c
--- syntax.c 2001/02/09 16:00:37 1.7.2.18
+++ syntax.c 2001/02/15 22:12:18
＠＠ -830,8 +830,6 ＠＠
as the comment sequence which began this comment
section */
{
- from++;
- UPDATE_SYNTAX_CACHE_FORWARD (from);
break;
}
}
--
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091
_________________ _________________ _________________ _________________
What are those straight lines for? "XEmacs rules."
5:39 a.m.
New subject: Crash (assertion failed) in 21.2.44 - 100% reproducible
Yes, that fixed it! Thanks a lot.
"Stephen J. Turnbull" <turnbull(a)sk.tsukuba.ac.jp> wrote:
>>>>> "Dmitry" == Dmitry Yaitskov
<dimas(a)home.com> writes:
Dmitry> The culprit seems to be the comment and the fact that it
Dmitry> starts from the very beginning of the file. Inserting a
Dmitry> blank like above the 1st line removes the crash. I hope
Dmitry> somebody who knows the C code involved can look at it. I
Dmitry> can provide further details if needed. Thanks.
There seems to be a fencepost error in syntax.c. You could try
Vladimir Vukicevic's patch. This is (hopefully) being reviewed by
Matt Tucker now, and may not go in in its current form, but it might
help you get some work done.
Index: syntax.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/syntax.c,v
retrieving revision 1.7.2.18
diff -u -r1.7.2.18 syntax.c
--- syntax.c 2001/02/09 16:00:37 1.7.2.18
+++ syntax.c 2001/02/15 22:12:18
＠＠ -830,8 +830,6 ＠＠
as the comment sequence which began this comment
section */
{
- from++;
- UPDATE_SYNTAX_CACHE_FORWARD (from);
break;
}
}
--
Cheers,
-Dima.
6:46 a.m.
New subject: Crash (assertion failed) in 21.2.44 - 100% reproducible
I wrote:
Yes, that fixed it! Thanks a lot.
Unfortunately, looks like it did not after all... I applied the patch
and ran make - which gave me an executable which did not crash when
run with --vanilla. That's when I replied to your message that it
fixed things. But w/out --vanilla, the new executable did not run very
well - it produced some weird messages when trying to load my init.el
- something to do with font-lock. So I rebuilt xemacs in a fresh build
directory (but still with the syntax.c patch applied of course) - and
it now loads my init.el file w/out problems, but again crashes exactly
as I described earlier - including when run --vanilla... :(
"Stephen J. Turnbull" <turnbull(a)sk.tsukuba.ac.jp>
wrote:
> >>>>> "Dmitry" == Dmitry Yaitskov <dimas(a)home.com>
writes:
>
> Dmitry> The culprit seems to be the comment and the fact that it
> Dmitry> starts from the very beginning of the file. Inserting a
> Dmitry> blank like above the 1st line removes the crash. I hope
> Dmitry> somebody who knows the C code involved can look at it. I
> Dmitry> can provide further details if needed. Thanks.
>
> There seems to be a fencepost error in syntax.c. You could try
> Vladimir Vukicevic's patch. This is (hopefully) being reviewed by
> Matt Tucker now, and may not go in in its current form, but it might
> help you get some work done.
>
>
> Index: syntax.c
> ===================================================================
> RCS file: /usr/CVSroot/XEmacs/xemacs/src/syntax.c,v
> retrieving revision 1.7.2.18
> diff -u -r1.7.2.18 syntax.c
> --- syntax.c 2001/02/09 16:00:37 1.7.2.18
> +++ syntax.c 2001/02/15 22:12:18
> ＠＠ -830,8 +830,6 ＠＠
> as the comment sequence which began this comment
> section */
> {
> - from++;
> - UPDATE_SYNTAX_CACHE_FORWARD (from);
> break;
> }
> }
--
Cheers,
-Dima.
4:45 p.m.
New subject: Crash (assertion failed) in 21.2.44 - 100% reproducible
There seems to be a fencepost error in syntax.c. You could try
Vladimir Vukicevic's patch. This is (hopefully) being reviewed by
Matt Tucker now, and may not go in in its current form, but it might
help you get some work done.
Actually, now that I've looked at my code again, I realized that it wasn't
right -- the code does indeed need to know if the last matching comment element
came at the end of the passed in region, not the space before it. With
my patch, a visual glitch appears if you're typing a string in a font-locked
mode; the font-locking will lag one character behind the end of the string.
So, a more correct patch follows... I've been working with this for a while
yesterday with no crashes.
- Vlad
Index: syntax.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/syntax.c,v
retrieving revision 1.7.2.18
diff -u -r1.7.2.18 syntax.c
--- syntax.c 2001/02/09 16:00:37 1.7.2.18
+++ syntax.c 2001/02/16 16:34:57
＠＠ -831,7 +831,11 ＠＠
section */
{
from++;
- UPDATE_SYNTAX_CACHE_FORWARD (from);
+ UPDATE_SYNTAX_CACHE_FORWARD (from);
+ if (from == stop)
+ {
+ return -1;
+ }
break;
}
}
6:26 p.m.
New subject: Crash (assertion failed) in 21.2.44 - 100% reproducible
No, this does not help - I get exactly the same crash using the sample
file that I posted. :(
Vladimir Vukicevic <vladimir(a)Intrepid.Com> wrote:
> There seems to be a fencepost error in syntax.c. You could try
> Vladimir Vukicevic's patch. This is (hopefully) being reviewed by
> Matt Tucker now, and may not go in in its current form, but it might
> help you get some work done.
Actually, now that I've looked at my code again, I realized that it wasn't
right -- the code does indeed need to know if the last matching comment element
came at the end of the passed in region, not the space before it. With
my patch, a visual glitch appears if you're typing a string in a font-locked
mode; the font-locking will lag one character behind the end of the string.
So, a more correct patch follows... I've been working with this for a while
yesterday with no crashes.
- Vlad
Index: syntax.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/syntax.c,v
retrieving revision 1.7.2.18
diff -u -r1.7.2.18 syntax.c
--- syntax.c 2001/02/09 16:00:37 1.7.2.18
+++ syntax.c 2001/02/16 16:34:57
＠＠ -831,7 +831,11 ＠＠
section */
{
from++;
- UPDATE_SYNTAX_CACHE_FORWARD (from);
+ UPDATE_SYNTAX_CACHE_FORWARD (from);
+ if (from == stop)
+ {
+ return -1;
+ }
break;
}
}
--
Cheers,
-Dima.
2:26 a.m.
New subject: Crash (assertion failed) in 21.2.44 - 100% reproducible
-- Dmitry Yaitskov <dimas(a)home.com> spake thusly:
Vladimir Vukicevic <vladimir(a)Intrepid.Com> wrote:
> > There seems to be a fencepost error in syntax.c. You could try
> > Vladimir Vukicevic's patch. This is (hopefully) being reviewed by
> > Matt Tucker now, and may not go in in its current form, but it
> > might help you get some work done.
>
> Actually, now that I've looked at my code again, I realized that it
> wasn't right -- the code does indeed need to know if the last
> matching comment element came at the end of the passed in region,
> not the space before it. With my patch, a visual glitch appears if
> you're typing a string in a font-locked mode; the font-locking will
> lag one character behind the end of the string. So, a more correct
> patch follows... I've been working with this for a while yesterday
> with no crashes.
No, this does not help - I get exactly the same crash using the sample
file that I posted. :(
This should be fixed in the latest CVS version. Please let me know if
it's not.
8488
days inactive
8492
days old
8 comments
4 participants
participants (4)
-
Dmitry Yaitskov -
Matt Tucker -
Stephen J. Turnbull -
Vladimir Vukicevic