[Qemu-devel] Re: [PATCH 2/6] Use correct types to enable > 2G support (v2)

[Avi Kivity]

Izik Eidus wrote:
> > > Index: qemu/cpu-all.h
> > > ===================================================================
> > > --- qemu.orig/cpu-all.h 2008-02-01 15:24:45.000000000 -0600
> > > +++ qemu/cpu-all.h      2008-02-01 15:28:48.000000000 -0600
> > > @@ -695,7 +695,7 @@
> > >  
> > >  /* page related stuff */
> > >  
> > > -#define TARGET_PAGE_SIZE (1 << TARGET_PAGE_BITS)
> > > +#define TARGET_PAGE_SIZE (1ul << TARGET_PAGE_BITS)
> > >  #define TARGET_PAGE_MASK ~(TARGET_PAGE_SIZE - 1)
> > >  #define TARGET_PAGE_ALIGN(addr) (((addr) + TARGET_PAGE_SIZE - 1) & 
> > > TARGET_PAGE_MASK
> > >       
> > Do you recall what this change fixed?  As Paul pointed out in IRC, using 
> > the host type here doesn't really fix the problem (target_ulong would be 
> > more appropriate).  However, we're both curious what problem it's 
> > actually fixing since sign extending the int should just work.
> >     
>
> ok the commit say:
>
>    kvm: qemu: change the type of the various page masks to unsigned long
>     
>    prevents truncation with >=4GB of guest physical memory
>
>
> as far as i remember it was used to address something with
> cpu_physical_memory_rw() probably related to &TARGET_PAGE_SIZE
> or ~TARGET_PAGE_SIZE,
>
> the fact is that i dont know if it ever fixed anything
>   

It fixes TARGET_PAGE_MASK, defined one line downscreen.  I agree it 
should be changed to ((target_ulong)1 << TARGET_PAGE_BITS).

--
error compiling committee.c: too many arguments to function