Re: [MiNT] stack overflow/stack underflow

["Helmut Karlowski" <helmut.karlowski@ish.de>]

Am 17.04.2011, 16:04 Uhr, schrieb Vincent Rivière  
<vincent.riviere@freesbee.fr>:

> Helmut Karlowski wrote:
> > did anybody ever see one of stack overflow/stack underflow - messages?
>
> No.
>
> Usually, when there is a stack overflow, the program crashes, sometimes  
> the OS, too.
>
> Which component is supposed to report those messages ?

I've asked because I've never seen this in real life.

Now I've tried to run the papyrus5-demo, and after fixing a first  
XaAES-bug, I ran into this:

17:01:16: pid 10 (papyrus): sendsig:stack overflow: new=122922E  
cur=2620678 call=2620004 newcur=122937A

I changed the message in the kernel (arch/sig_mach.c#145):

ALERT("sendsig:stack overflow: new=%lx cur=%lx call=%lx newcur=%lx",  
newstack, curproc->stack, call, &newcurrent);


The values do not point to an overflow but rather a mixup of different  
stacks. I don't know where this comes from.

I hacked my kernel to just ignore this error and everything works fine.  
Apart from the alerts it behaves like under MyAES now.

But it would be nice to have a clean solution. It could have something to  
do with USERDEFS, but I'm actually clueless.

--
Helmut Karlowski