[Users] crash on Windows 8.1 caused by libcairo-2.dll

Andrej Kacian andrej at kacian.sk
Sun Jul 2 12:16:57 CEST 2017

On Sun, 2 Jul 2017 12:29:56 +0300
Cristian Secară <liste at secarica.ro> wrote:

> > I've used the one at:
> > https://sourceforge.net/projects/mingw-w64/files/External%20binary%20packages%20%28Win64%20hosted%29/gdb/
> > 
> > You need to make sure you have the correct one (32 or 64bit, depending
> > on which Claws Mail build you're running).
> >   
> So far I tried only with the 64bit version.
> 1. Running the new build, it crashed as usual :)

As expected then. Oh well.

> 2. I turned back to my old (stable) Claws Mail, but replaced only
> libcairo-2.dll and libpng16-16.dll files from this test build; it
> crashed the same
> 3. I run gdb -- see the attached log files.

I see the same crash backtrace as I've already seen from someone else -
an impossible backtrace. In frame #1, cairo_status() is a simple
one-line function that only returns an integer variable, and can not in
any way call cairo_scaled_font_status() (frame #0).

I've had some weird issues in using 64-bit GDB on Windows myself in the
past, so while disappointing, it does not surprise me.

I would appreciate it if you could try to get a gdb backtrace with the
32-bit build and the two provided debug files, perhaps the backtrace
will be something real then.

> Maybe some useful info: in order to generate the crash, I just click at
> random on several mail folders, one after the other and somewhat quick,
> i.e. I don't wait for the display to finish rendering a particular
> folder. Using this method the crash occurs in less than one minute.

After one such crash, could you look in file claws-win32.log in your
TMPDIR, to see if there are any failed assertion warnings? There could
be something relevant there.

> Today I don't think I have much time to do more tests (for example with
> 32bit version), but probably I can to this the next days.
> I am not such in a hurry, since (for me) the two Inkscape files still
> makes Claws Mail rock solid :)

I'll have to talk to Inkscape guys what their secret formula is,
because this bug is really bugging me - especially when I can not
reproduce it myself. :)


More information about the Users mailing list