More QtScript crashes? Or QtCore/Grantlee?

[Milian Wolff]

On Wednesday 28 November 2012 08:31:37 Andreas Pakulat wrote:
> Hi,
> 
> On Wed, Nov 28, 2012 at 3:22 AM, Milian Wolff <mail at milianw.de> wrote:
> > I think I found more QtScript crashes... Or something else in
> > QtCore/Grantlee maybe - not sure :(
> > 
> > I tried to extend the file template preview to show previews for all files
> > in a tab widget. This reliably crashes for me, valgrind gave me this so
> > far:
> > 
> > ==14660== Jump to the invalid address stated on the next line
> > ==14660==    at 0x263C7D20: ???
> > ==14660==  Address 0x263c7d20 is not stack'd, malloc'd or (recently)
> > free'd
> > ==14660==
> > ==14660==
> > ==14660== Process terminating with default action of signal 11 (SIGSEGV)
> > ==14660==  Access not within mapped region at address 0x263C7D20
> > ==14660==    at 0x263C7D20: ???
> > ==14660==  If you believe this happened as a result of a stack
> > ==14660==  overflow in your program's main thread (unlikely but
> > ==14660==  possible), you can try to increase the size of the
> > ==14660==  main thread stack using the --main-stacksize= flag.
> > ==14660==  The main thread stack size used in this run was 8388608.
> > ==14660==
> > 
> > And quite a lot of this:
> > 
> > ==14660== Conditional jump or move depends on uninitialised value(s)
> > ==14660==    at 0xA95F86A: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0xA97E8E0: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0xA988A7C: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0xA93F9D3: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0x3952AA44: ???
> > ==14660==    by 0xA8F68AD: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0xA8F6B15: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0xAA26241: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0xAA2CEF8: QScriptEngine::evaluate(QString const&, QString
> > const&, int) (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0x21C1B92A: Kate::Script::require(QScriptContext*,
> > QScriptEngine*) (katescripthelpers.cpp:138)
> > ==14660==    by 0xAA496D7: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==    by 0xA95BE5F: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==  Uninitialised value was created by a stack allocation
> > ==14660==    at 0xA95F820: ??? (in /usr/lib/libQtScript.so.4.8.3)
> > ==14660==
> > 
> > 
> > In KDevelop I got this via gdb: http://paste.kde.org/615908/
> 
> Seeing self-built Qt from gdb and what looks like system packages in
> valgrind rings my BC-bell. But I guess you already checked that (by
> not using your self-built Qt)?

Ah yes, sorry I forgot to mention. I tried to "fix" it by using a selfcompiled 
Qt from git (4.8 branch). Same issue (see gdb log). Valgrind and initial crash 
was done with system Qt packages as provided by ArchLinux (which is 4.8.3 with 
the QtScript crash fixes by Olivier applied).

> > I really have no idea whats going on so far...
> > 
> > To reproduce: I've pushed the offending commit to
> > tabbed-filetemplate-preview branch in KDevplatform. Built it, then go to
> > 'new file from template' and browse a bit. After 3-5 clicks I get a
> > crash. I've added a test app but that
> > seems to work quite stable:
> The branch does not build, it misses a main.cpp in plugins/filetemplates/

Aaahhh ffs. This hits me far too often. Is there some way I can configure git 
to warn me that there are files of a given pattern that I have not added? I 
have tons of patches and other files in my source dirs which mess up the 
"Untracked files:" part of git status. Thus I most often just don't see that I 
missed to add an important file...

Cheers
-- 
Milian Wolff
mail at milianw.de
http://milianw.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20121128/c3566e30/attachment.sig>