On May 24, 12:52pm, thorpej@zembu.com (Jason R Thorpe) wrote:
-- Subject: Re: CVS commit: basesrc

| On Thu, May 24, 2001 at 03:24:00PM -0400, Christos Zoulas wrote:
| 
|  > On May 24, 10:59am, thorpej@zembu.com (Jason R Thorpe) wrote:
|  > -- Subject: Re: CVS commit: basesrc
|  > 
|  > | But it causes <machine/vmparam.h> to be included.
|  > 
|  > Maybe it should be including that then?
| 
| Programs aren't supposed to include <machine/types.h> directly, are they?

I don't know. There seems to be a lot of flux in the uvm directory as
far as headers go. We need to decide on a policy on that. Traditionally
for each <sys/foo.h> that had a corresponding <machine/foo.h> you were
not supposed to include <machine/foo.h>. But there is no <sys/vmparam.h>,
and the place where you are supposed to be getting the typedefs for vm
types should not be called "param". "param" files were usually reserved
for #define constants. So I don't think that <machine/cpu.h> should need
<machine/param.h> in order to compile.

christos