Smokegen core dump

Hi

Since some time i am getting this error reliably. ( Fresh rebuild from scratch
).

[ 28%] Built target smokekdecore
make -f kdeui/CMakeFiles/smokekdeui.dir/build.make
kdeui/CMakeFiles/smokekdeui.dir/depend
parsing "/home/mjansen/kde/trunk/src/smoke/smokekde/attica/attica_includes.h"
ASSERT failure in QList<T>::at: "index out of range", file
/qt/trunk/include/QtCore/qlist.h, line 456
/bin/sh: line 1:  7841 Aborted                 (core dumped)
/kde/trunk/kde/bin/smokegen -config
/home/mjansen/kde/trunk/bld/smoke/smokekde/attica/../kdecore/config.xml -
smokeconfig /home/mjansen/kde/trunk/src/smoke/smokekde/attica/smokeconfig.xml
-- /home/mjansen/kde/trunk/src/smoke/smokekde/attica/attica_includes.h
make[2]: *** [attica/smokedata.cpp] Error 134
make[2]: Leaving directory `/home/mjansen/kde/trunk/bld/smoke/smokekde'
make[1]: *** [attica/CMakeFiles/smokeattica.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....



The core files gives the following backtrace

#0  0x00002b7df05479e5 in raise () from /lib64/libc.so.6
#1  0x00002b7df0548ee6 in abort () from /lib64/libc.so.6
#2  0x00002b7def3b1d78 in qt_message_output (msgType=QtFatalMsg, buf=0x672a68
"ASSERT failure in QList<T>::at: \"index out of range\", file
/qt/trunk/include/QtCore/qlist.h, line 456") at
/home/mjansen/kde/trunk/src/qt/qt/src/corelib/global/qglobal.cpp:2282
#3  0x00002b7def3b1ef4 in qt_message(QtMsgType, const char *, typedef
__va_list_tag __va_list_tag *) (msgType=QtFatalMsg, msg=0x2b7def55bdb0 "ASSERT
failure in %s: \"%s\", file %s, line %d", ap=0x7fffc23cdf30) at
/home/mjansen/kde/trunk/src/qt/qt/src/corelib/global/qglobal.cpp:2328
#4  0x00002b7def3b2762 in qFatal (msg=0x2b7def55bdb0 "ASSERT failure in %s:
\"%s\", file %s, line %d") at
/home/mjansen/kde/trunk/src/qt/qt/src/corelib/global/qglobal.cpp:2511
#5  0x00002b7def3b196b in qt_assert_x (where=0x449dc9 "QList<T>::at",
what=0x449d23 "index out of range", file=0x449d88
"/qt/trunk/include/QtCore/qlist.h", line=456) at
/home/mjansen/kde/trunk/src/qt/qt/src/corelib/global/qglobal.cpp:2035
#6  0x000000000043669d in QList<QString>::at (this=0x65d2a0, i=1953391972) at
/qt/trunk/include/QtCore/qlist.h:456
#7  0x00002b7defb337f9 in IndexedString::byteArray (this=0x7fffc23ce100) at
/home/mjansen/kde/trunk/src/smoke/smokegen/parser/indexedstring.cpp:203
#8  0x00002b7defb4f037 in stringFromContents (contents=0x67537c, count=5) at
/home/mjansen/kde/trunk/src/smoke/smokegen/parser/rpp/chartools.cpp:65
#9  0x00002b7defb4ef93 in lineFromContents (size=28, contents=0x675320,
lineNumber=23) at
/home/mjansen/kde/trunk/src/smoke/smokegen/parser/rpp/chartools.cpp:56
#10 0x00002b7defb18897 in Parser::reportError (this=0x7fffc23ce6a0, msg=...)
at /home/mjansen/kde/trunk/src/smoke/smokegen/parser/parser.cpp:372
#11 0x00002b7defb185af in Parser::syntaxError (this=0x7fffc23ce6a0) at
/home/mjansen/kde/trunk/src/smoke/smokegen/parser/parser.cpp:339
#12 0x00002b7defb22e7b in Parser::parseDeclarationInternal
(this=0x7fffc23ce6a0, node=@0x7fffc23ce468) at
/home/mjansen/kde/trunk/src/smoke/smokegen/parser/parser.cpp:3584
#13 0x00002b7defb19827 in Parser::parseDeclaration (this=0x7fffc23ce6a0,
node=@0x7fffc23ce468) at
/home/mjansen/kde/trunk/src/smoke/smokegen/parser/parser.cpp:712
#14 0x00002b7defb1930c in Parser::parseTranslationUnit (this=0x7fffc23ce6a0,
node=@0x7fffc23ce4c8) at
/home/mjansen/kde/trunk/src/smoke/smokegen/parser/parser.cpp:613
#15 0x00002b7defb17c8f in Parser::parse (this=0x7fffc23ce6a0,
_session=0x7fffc23ce7e0) at
/home/mjansen/kde/trunk/src/smoke/smokegen/parser/parser.cpp:192
#16 0x0000000000423e0c in main (argc=7, argv=0x7fffc23cef68) at
/home/mjansen/kde/trunk/src/smoke/smokegen/main.cpp:224


In Frame 7
(gdb) p/x m_index
$1 = 0x746e6564

In frame 8
(gdb) p/x contents[a]
$2 = 0x746e6564
(gdb) p a
$3 = 4

In frame 10
(gdb) p/x *msg
Cannot resolve function operator* to any overloaded instance
gdb) p p->position.line
warning: can't find linker symbol for virtual table for `Problem' value
$17 = 23


Anyone an idea how to debug / fix that bug?

Btw i am on a 64bit system.

Mike

_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

[Arno Rehn-3]

On Sunday 30 January 2011 01:13:24 Michael Jansen wrote:

> Hi
>
> Since some time i am getting this error reliably. ( Fresh rebuild from
> scratch ).
>
> [ 28%] Built target smokekdecore
> make -f kdeui/CMakeFiles/smokekdeui.dir/build.make
> kdeui/CMakeFiles/smokekdeui.dir/depend
> parsing
> "/home/mjansen/kde/trunk/src/smoke/smokekde/attica/attica_includes.h"
> ASSERT failure in QList<T>::at: "index out of range", file
> /qt/trunk/include/QtCore/qlist.h, line 456
> /bin/sh: line 1:  7841 Aborted                 (core dumped)
> /kde/trunk/kde/bin/smokegen -config
> /home/mjansen/kde/trunk/bld/smoke/smokekde/attica/../kdecore/config.xml -
> smokeconfig
> /home/mjansen/kde/trunk/src/smoke/smokekde/attica/smokeconfig.xml --
> /home/mjansen/kde/trunk/src/smoke/smokekde/attica/attica_includes.h
> make[2]: *** [attica/smokedata.cpp] Error 134
> make[2]: Leaving directory `/home/mjansen/kde/trunk/bld/smoke/smokekde'
> make[1]: *** [attica/CMakeFiles/smokeattica.dir/all] Error 2
> make[1]: *** Waiting for unfinished jobs....

Thanks for the report. What version of attica is this, and which revision
smokegen? Last I checked it worked all fine for me.

--
Arno Rehn
arno@...
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

On Monday 31 January 2011 16:30:10 Arno Rehn wrote:

> On Sunday 30 January 2011 01:13:24 Michael Jansen wrote:
> > Hi
> >
> > Since some time i am getting this error reliably. ( Fresh rebuild from
> > scratch ).
> >
> > [ 28%] Built target smokekdecore
> > make -f kdeui/CMakeFiles/smokekdeui.dir/build.make
> > kdeui/CMakeFiles/smokekdeui.dir/depend
> > parsing
> > "/home/mjansen/kde/trunk/src/smoke/smokekde/attica/attica_includes.h"
> > ASSERT failure in QList<T>::at: "index out of range", file
> > /qt/trunk/include/QtCore/qlist.h, line 456
> > /bin/sh: line 1:  7841 Aborted                 (core dumped)
> > /kde/trunk/kde/bin/smokegen -config
> > /home/mjansen/kde/trunk/bld/smoke/smokekde/attica/../kdecore/config.xml
> > -
> > smokeconfig
> > /home/mjansen/kde/trunk/src/smoke/smokekde/attica/smokeconfig.xml --
> > /home/mjansen/kde/trunk/src/smoke/smokekde/attica/attica_includes.h
> > make[2]: *** [attica/smokedata.cpp] Error 134
> > make[2]: Leaving directory `/home/mjansen/kde/trunk/bld/smoke/smokekde'
> > make[1]: *** [attica/CMakeFiles/smokeattica.dir/all] Error 2
> > make[1]: *** Waiting for unfinished jobs....
>
> Thanks for the report. What version of attica is this, and which revision
> smokegen? Last I checked it worked all fine for me.

Sorry for the late answer. I was really busy with build-tool because of the
git conversion.

It is attica master from git://anongit.kde.org/attica .
smokegen is master too.

The core dump is still there btw. after updating both today.

Mike
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

> > Thanks for the report. What version of attica is this, and which
> > revision
> > smokegen? Last I checked it worked all fine for me.
>
> Sorry for the late answer. I was really busy with build-tool because of the
> git conversion.
>
> It is attica master from git://anongit.kde.org/attica .
> smokegen is master too.
>
> The core dump is still there btw. after updating both today.

And today it is gone. If you did not change anything we have a heisenbug.
Because it came and went before.

Mike

_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

[Arno Rehn-3]

On Wednesday 16 February 2011 20:28:51 Michael Jansen wrote:

> > > Thanks for the report. What version of attica is this, and which
> > > revision
> > > smokegen? Last I checked it worked all fine for me.
> >
> > Sorry for the late answer. I was really busy with build-tool because of
> > the git conversion.
> >
> > It is attica master from git://anongit.kde.org/attica .
> > smokegen is master too.
> >
> > The core dump is still there btw. after updating both today.
>
> And today it is gone. If you did not change anything we have a heisenbug.
> Because it came and went before.

Can you tell me which attica version (SHA1) you are using now and which you've
used before? I guess it's a bug somewhere in the parser. If I can see the API
diff, I might be able to track it down.

--
Arno Rehn
arno@...
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

On Wednesday 16 February 2011 20:43:36 Arno Rehn wrote:

> On Wednesday 16 February 2011 20:28:51 Michael Jansen wrote:
> > > > Thanks for the report. What version of attica is this, and which
> > > > revision
> > > > smokegen? Last I checked it worked all fine for me.
> > >
> > > Sorry for the late answer. I was really busy with build-tool because
> > > of
> > > the git conversion.
> > >
> > > It is attica master from git://anongit.kde.org/attica .
> > > smokegen is master too.
> > >
> > > The core dump is still there btw. after updating both today.
> >
> > And today it is gone. If you did not change anything we have a
> > heisenbug.
> > Because it came and went before.
>
> Can you tell me which attica version (SHA1) you are using now and which
> you've used before? I guess it's a bug somewhere in the parser. If I can
> see the API diff, I might be able to track it down.

That is the problem. Nothing changed there. Was the first thing i checked. The
last real code change is from mid of january.

Perhaps some indirectly needed header changed. But not chance to find that
one.

Mike
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

On Wednesday 16 February 2011 20:43:36 Arno Rehn wrote:

> On Wednesday 16 February 2011 20:28:51 Michael Jansen wrote:
> > > > Thanks for the report. What version of attica is this, and which
> > > > revision
> > > > smokegen? Last I checked it worked all fine for me.
> > >
> > > Sorry for the late answer. I was really busy with build-tool because
> > > of
> > > the git conversion.
> > >
> > > It is attica master from git://anongit.kde.org/attica .
> > > smokegen is master too.
> > >
> > > The core dump is still there btw. after updating both today.
> >
> > And today it is gone. If you did not change anything we have a
> > heisenbug.
> > Because it came and went before.
>
> Can you tell me which attica version (SHA1) you are using now and which
> you've used before? I guess it's a bug somewhere in the parser. If I can
> see the API diff, I might be able to track it down.

Now i have it again. No idea why it went thought at that time. What i can now
tell is that i am able to make it work by adding a newline at the end of
attica_includes.h. If a remote that again it coredumps again.

I got the idea because the line number shown was 23 (last in that file) and
the problem seems to be related to smokegen going over a upper limit.

Mike
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

On Wednesday 23 February 2011 22:49:25 Michael Jansen wrote:
> Now i have it again. No idea why it went thought at that time. What i
can now
> tell is that i am able to make it work by adding a newline at the end of
> attica_includes.h. If a remote that again it coredumps again.

we have this problem as well on freebsd (though i can reproduce it only
on i386, while another user gets it on amd64). we detected it for the
first time in rc1

> I got the idea because the line number shown was 23 (last in that file)
and
> the problem seems to be related to smokegen going over a upper
limit.

adding the newline didn't help, so we've tried applying the attached
patch. the build is proceeding successfully (though i doubt on the
correctness of the parsing), but we see lots of lines like this
(qDebug()ged)...

strings()->size(): 154, m_index: 2779096485

...with different numbers most of the time. most requested indexes are
below the limit (qstringlist size), but some of them are really insane. i
should also mention that the numbers change on different builds

but, as i said, the build is going on. obviously this patch is certainly not
the fix, it just silences the symptoms...
--
Alberto Villa, FreeBSD committer <avilla@...>
http://people.FreeBSD.org/~avilla

The story you are about to hear is true.
Only the names have been changed to protect the innocent.

[smoke1.diff]

---

--- generator/parser/indexedstring.cpp.orig 2011-02-23 22:12:38.695255708 +0100
+++ generator/parser/indexedstring.cpp 2011-02-24 02:36:09.035361151 +0100
@@ -195,12 +195,15 @@
 }

 QByteArray IndexedString::byteArray() const {
+  qDebug() << "strings()->size():" << strings()->size() << ", m_index:" << m_index;
   if(!m_index)
     return QByteArray();
   else if((m_index & 0xffff0000) == 0xffff0000)
     return QString(QChar((char)m_index & 0xff)).toUtf8();
-  else
+  else if (m_index < strings()->size())
     return strings()->at(m_index).toUtf8(); /*arrayFromItem(globalIndexedStringRepository->itemFromIndex(m_index));*/
+  else
+    return QByteArray();
 }

 unsigned int IndexedString::hashString(const char* str, unsigned short length) {

---

_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

[Arno Rehn-3]

On Wednesday 23 February 2011 23:49:25 you wrote:

> On Wednesday 16 February 2011 20:43:36 Arno Rehn wrote:
> > On Wednesday 16 February 2011 20:28:51 Michael Jansen wrote:
> > > > > Thanks for the report. What version of attica is this, and which
> > > > > revision
> > > > > smokegen? Last I checked it worked all fine for me.
> > > >
> > > > Sorry for the late answer. I was really busy with build-tool because
> > > > of
> > > > the git conversion.
> > > >
> > > > It is attica master from git://anongit.kde.org/attica .
> > > > smokegen is master too.
> > > >
> > > > The core dump is still there btw. after updating both today.
> > >
> > > And today it is gone. If you did not change anything we have a
> > > heisenbug.
> > > Because it came and went before.
> >
> > Can you tell me which attica version (SHA1) you are using now and which
> > you've used before? I guess it's a bug somewhere in the parser. If I can
> > see the API diff, I might be able to track it down.
>
> Now i have it again. No idea why it went thought at that time. What i can
> now tell is that i am able to make it work by adding a newline at the end
> of attica_includes.h. If a remote that again it coredumps again.
>
> I got the idea because the line number shown was 23 (last in that file) and
> the problem seems to be related to smokegen going over a upper limit.

Ok, thanks. I'll see what I can do.

--
Arno Rehn
arno@...
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

[Arno Rehn-3]

On Thursday 24 February 2011 03:34:10 Alberto Villa wrote:

> On Wednesday 23 February 2011 22:49:25 Michael Jansen wrote:
> > Now i have it again. No idea why it went thought at that time. What i
>
> can now
>
> > tell is that i am able to make it work by adding a newline at the end of
> > attica_includes.h. If a remote that again it coredumps again.
>
> we have this problem as well on freebsd (though i can reproduce it only
> on i386, while another user gets it on amd64). we detected it for the
> first time in rc1
>
> > I got the idea because the line number shown was 23 (last in that file)
>
> and
>
> > the problem seems to be related to smokegen going over a upper
>
> limit.
>
> adding the newline didn't help, so we've tried applying the attached
> patch. the build is proceeding successfully (though i doubt on the
> correctness of the parsing), but we see lots of lines like this
> (qDebug()ged)...
>
> strings()->size(): 154, m_index: 2779096485
>
> ...with different numbers most of the time. most requested indexes are
> below the limit (qstringlist size), but some of them are really insane. i
> should also mention that the numbers change on different builds
>
> but, as i said, the build is going on. obviously this patch is certainly
> not the fix, it just silences the symptoms...

Interesting. Thanks for the investigation. :) Seems like we found our culprit.

--
Arno Rehn
arno@...
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

On Thursday 24 February 2011 15:00:21 Arno Rehn wrote:
> Interesting. Thanks for the investigation. :) Seems like we found our
> culprit.

great!

by the way, shouldn't this line (generator/parser/rpp/chartools.h)
#define characterFromIndex(index) ((char)(index & 0xffff))
be
#define characterFromIndex(index) ((char)(index & 0xff))
just in case some wrong index is passed to it? as far as i know the (char)
cast should trim it correctly, but may it be that some weird
architectures/compilers/environments don't do the right thing?
--
Alberto Villa, FreeBSD committer <avilla@...>
http://people.FreeBSD.org/~avilla

The yankees, son, are up north.
The damnyankees are down here.


_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

[Arno Rehn-3]

On Thursday 24 February 2011 15:22:03 Alberto Villa wrote:

> On Thursday 24 February 2011 15:00:21 Arno Rehn wrote:
> > Interesting. Thanks for the investigation. :) Seems like we found our
> > culprit.
>
> great!
>
> by the way, shouldn't this line (generator/parser/rpp/chartools.h)
> #define characterFromIndex(index) ((char)(index & 0xffff))
> be
> #define characterFromIndex(index) ((char)(index & 0xff))
> just in case some wrong index is passed to it? as far as i know the (char)
> cast should trim it correctly, but may it be that some weird
> architectures/compilers/environments don't do the right thing?

Could be. I don't really know that much about the C++ parser itself. We
basically copied it over from KDevelop and removed all the KDE dependencies.
I'll check the current KDevelop source and see what they do. Maybe our bug has
already been fixed in their parser.

--
Arno Rehn
arno@...
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

On Thursday 24 February 2011 15:36:17 Arno Rehn wrote:
> Could be. I don't really know that much about the C++ parser itself. We
> basically copied it over from KDevelop and removed all the KDE
> dependencies. I'll check the current KDevelop source and see what they
do.
> Maybe our bug has already been fixed in their parser.

any news on this? our update to 4.6.x is blocked by this problem
--
Alberto Villa, FreeBSD committer <avilla@...>
http://people.FreeBSD.org/~avilla

After all, it is only the mediocre who are always at their best.
                -- Jean Giraudoux


_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

[Arno Rehn-3]

On Thursday 10 March 2011 02:14:38 Alberto Villa wrote:

> On Thursday 24 February 2011 15:36:17 Arno Rehn wrote:
> > Could be. I don't really know that much about the C++ parser itself. We
> > basically copied it over from KDevelop and removed all the KDE
> > dependencies. I'll check the current KDevelop source and see what they
>
> do.
>
> > Maybe our bug has already been fixed in their parser.
>
> any news on this? our update to 4.6.x is blocked by this problem

Sorry, I just can't reproduce the problem. I've tried with current attica
master, but it just works fine. Without access to a machine where it fails, I
just don't know how to fix it.
You can disable the attica bindings (cmake -DDISABLE_Attica=TRUE) for now - I
don't know of any project that uses them.

--
Arno Rehn
arno@...
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

On Thursday 10 March 2011 16:12:09 Arno Rehn wrote:
> Sorry, I just can't reproduce the problem. I've tried with current attica
> master, but it just works fine. Without access to a machine where it fails,
> I just don't know how to fix it.
> You can disable the attica bindings (cmake -DDISABLE_Attica=TRUE) for
now -
> I don't know of any project that uses them.

ok. meanwhile i'll keep searching a fix
--
Alberto Villa, FreeBSD committer <avilla@...>
http://people.FreeBSD.org/~avilla

How you look depends on where you go.


_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

[Arno Rehn-3]

On Friday 11 March 2011 02:53:55 Alberto Villa wrote:

> On Thursday 10 March 2011 16:12:09 Arno Rehn wrote:
> > Sorry, I just can't reproduce the problem. I've tried with current attica
> > master, but it just works fine. Without access to a machine where it
> > fails, I just don't know how to fix it.
> > You can disable the attica bindings (cmake -DDISABLE_Attica=TRUE) for
>
> now -
>
> > I don't know of any project that uses them.
>
> ok. meanwhile i'll keep searching a fix

Just so that I can try to reproduce: which distribution are you using?

--
Arno Rehn
arno@...
_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings

On Saturday 12 March 2011 20:34:07 Arno Rehn wrote:
> Just so that I can try to reproduce: which distribution are you using?

freebsd. but it doesn't happen everywhere
--
Alberto Villa, FreeBSD committer <avilla@...>
http://people.FreeBSD.org/~avilla

Let a fool hold his tongue and he will pass for a sage.
                -- Publilius Syrus


_______________________________________________
Kde-bindings mailing list
Kde-bindings@...
https://mail.kde.org/mailman/listinfo/kde-bindings