Re: stackguard-like embedded protection

[Hiroaki Etoh <ETOH () JP IBM COM>]

On 2000/09/13 18:29:31 antirez wrote:

> On Mon, Sep 11, 2000 at 11:34:01AM +0900, Hiroaki Etoh wrote:
> > I don't think this solution is portable to any C compiler.  Because the
> ordering
> > of local variables is very dependent of the implementation of compilers, for
> > example, Visual C++ by Microsoft.  I don't know the current version of GCC
> > compiler preserves the ordering of local variables. Therefore, I implemented
> > ProPolice at the intermediate language level.
>
> I tested GCC i386 and sparc with a lot of optimizations and options:
> it seems that if the variable is declared with the specifier 'volatile'
> always goes next to the RET. I have no access to other compilers,
> but since implementing ProPolice at the intermediate language level
> you needs to implement it for all the compilers it seems that my
> solution can be more portable in some case. For example it can work
> in new compilers without changes, ProPolice can't.
> Also ProPolice needs to be implemented in a different way for different
> compilers: you can add #ifdefs to my header file to make it compatible
> with strange-behaviour compilers.

I tested GCC aix, xlC aix, and Microsoft Visual C++. They don't place the
'volatile' variable next to the RET.  Could someone test the following program
in the other compilers. I am very interested in such information.

#include <stdio.h>
int main()
{
   volatile int x=0;
   char buf[20];
   printf("%s!! x:%x buf:%x\n", ((char*)&x>buf)?"ok":"no", &x, buf);
}

Hiroaki Etoh,  Tokyo Research Laboratory, IBM Japan