Subject: Re: _btop macros are non-MI
To: Simon Burge <>
From: Jason R Thorpe <>
List: tech-kern
Date: 06/21/2000 22:20:00
On Thu, Jun 22, 2000 at 01:16:17PM +1000, Simon Burge wrote:

 > Is there any reason why each archtecture defines an <arch>_btop macro
 > and not just a single btop macro?  At the moment, we've got code
 > segments like

Technically, <machine_arch>_btop() is for machine-dependent page size,
whereas atop() ("address-to-page", provided by UVM) is for VM page size.

Now, it happens to be the case that for all of the platforms mentioned
below, that atop() would work (because MD page size == VM page size),
but due to the cookie nature of the mmap return value at the moment,
it's technically machine-dependent anyway...

Really, what this all calls for is bus_space_mmap() (tho, I'd like
a different interface than what the SPARC port is currently using,
I think...)

...but that's a different bridge to burn than the problem you're
currently trying to solve :-)

        -- Jason R. Thorpe <>