[Tioga-users] Segfault fixes ??

[Vincent Fourmond]

Edwin wrote:
> So, I can assure you it is the latest version I have installed, I double
> checked the make install and made sure I didn't have a copie left of any
> file. You can also see from the following that the error was in
> FigureMaker.so, so I made triple sure that that was the latest version.
> Program received signal SIGSEGV, Segmentation fault.
> 0xb7a93958 in End_Axis_Standard_State at plt ()
>     from /usr/lib/ruby/site_ruby/1.8/i686-linux/Tioga/FigureMaker.so
> 
> I am also still very puzzled by this. For now I am using tioga over ssh
> from a 64bit machine where it is working. I am a bit worried that it
> would be only on my two machine, but I have no problems with any other
> programs that I compiled myself.

  I'm glad it's working on a 64 bit machine. I had a closer look at the
function causing the segfault, I really can't understand how it could
cause segfaults, as it is not using shared symbols (which are kinda
slippery on different archs) and not even pointers that would have a
chance to get there uninitialized. It might be a compiler bug.

  I really won't have any opportunity to take a closer look over the
next two weeks as I'm away from my computer. However, you could try to
modify the Makefile produced by extconf.rb to replace -O2 (or any other
optimization options) by -O0, replace CC by another compiler if you have
one available and so on. It would help a lot to know if it is
compiler-related.

  If you feel like hacking the code, add

fprintf(stderr, "TF: %p\n", TF);

  in pdffile.c/Write_grestore. (before the extisting line). If the
values displayed change in one run, it really would look like a compiler
bug to me.

  Cheers, and good hunt !

	Vincent

-- 
Vincent Fourmond, PhD student (not for long anymore)
http://vincent.fourmond.neuf.fr/