Re: amd64-current error in kern_exec.c

[Matthias Scheler <tron%zhadum.org.uk@localhost>]

On Fri, Aug 26, 2011 at 09:56:01AM +0100, Matthias Scheler wrote:
> On Fri, Aug 26, 2011 at 10:12:21AM +0200, Martin Husemann wrote:
> > On Thu, Aug 25, 2011 at 03:37:55PM -0700, Hisashi T Fujinaka wrote:
> > > 380 1.318        (reinoud  25-Aug-11):          /* Seems ok: check that 
> > > entry point is not too low */
> > > 381 1.318        (reinoud  25-Aug-11):          if (epp->ep_entry < 
> > > VM_MIN_ADDRESS) {
> > > 
> > > Apparently epp->ep_entry is unsigned and VM_MIN_ADDRESS is zero, and
> > > this results in an error.
> > 
> > Christos fixed it already.
> 
> No, he didn't, he broke at least the NetBSD/i386 build. He added
> the following check:
> 
> #if VM_MIN_ADDRESS > 0
> [...]
> #endif
> 
> "VM_MIN_ADDRESS" is define like this:
> 
>       ((vaddr_t)((0xc0000000UL/(1UL << 22)-1) << 22))
> 
> And the C++ pre-processor cannot handle that expression.

I got that bit slightly wrong while looking at the ".i" file.
Here is the correct definition:

#define VM_MIN_ADDRESS                ((vaddr_t)0)

The C++ pre-processor will not understand that either.

        Kind regards

-- 
Matthias Scheler                                  http://zhadum.org.uk/