Re: [MiNT] XaAES regression in launch()

[Vincent Rivière <vincent.riviere@freesbee.fr>]

On 20/11/2012 14:04, Helmut Karlowski wrote:
> Are you sure the kfree in XaAES causes this?

Pretty sure, while not 100%.
If I comment that kfree() in make_argv(), the problem disappears.

> What is desk_cf, is it teradesk?

desk_cf.prg is TeraDesk rebuild with AHCC for 68020/ColdFire. Except the 
CPU, that's a standard build. It had worked flawlessly for years.

> For XaAES I always add
>
> DBG((0,fmt,...));
>
> use like printf. Output goes to xa_boot.log.

Thanks for the info.
I used the FORCE() macro, and a hacked sys/debug.c which sends debug 
output to the ColdFire serial port. That's very handy, and fortunately 
that didn't change the bug's behavior.

Other good thing, I can remotely examine the memory with the CodeWarrior 
debugger. So if I know that something is wrong at some address, I can 
break the execution and see what's wrong in the memory.

All that code around the strings[] array is quite messy (and poorly 
commented), that's not really surprising that something goes wrong.

I will add more traces.

--
Vincent Rivière