Re: [gnugo-devel] Crash in regression test?

[Arend Bayer]

Dan wrote:

> SP Lee wrote:
>
> > ***assertion failure:
> > owl.c:5393 - &owl_stack[(*owl)->number_in_stack] == *owl near PASS***
>
> This seems to have to do with the patch paul_3_20.7. Do you
> still get the crash if you back out that patch?

I doubt this. The crash must be yet another case of problems with the
owl stack moving on a realloc(). I don't see how this could relate to
not copying all owl data from one stackp to the next.
At some point I have to clean this mess I created there. But probably
better post 3.4.

> It would be good to understand the mechanism of the crash.

Can you maybe get a backtrace? You would need gdb (gnu debugger) to get
this.

For example do:

gdb ../interface/gnugo
break abortgo
run --mode gtp --quiet <global.tst
        (should stop in abortgo at the point of the crash)
bt
        (will print the backtrace)


Arend