Re: [PATCH] Intel 830 support for agpgart

[David Weinehall <tao@acc.umu.se>]

On Tue, Oct 02, 2001 at 07:45:42PM -0400, Robert Love wrote:
> On Tue, 2001-10-02 at 09:10, Christof Efkemann wrote:
> > Yes, that seems to work as well.  Although there are two minor things I
> > noticed:
> > - First, intel_generic_setup sets num_aperture_sizes to 7, while the i830
> >   has only 4 valid values (32 to 256 MB).
> > - Second, when intel_generic_configure clears the error status register, it
> >   resets bits 8, 9 and 10.  With an i830 it should clear bits 2, 3 and 4.
> > 
> > So I'm not sure if this works in general, or could it cause errors on other
> > systems?
> 
> It will probably work fine on all systems, but its not the right way to
> go IMO.  Your original implementation was better.
> 
> However, I am still disliking the multiple function idea.  Same thing
> with the i840.  The only real difference is those defines.
> 
> A _very_ simple solution, IF we had separate CONFIG statements for each
> i8xx (or at least one for i830, one for i840, and one for the rest)
> would be:
> 
> /* all the normal defines here */
> #ifdef CONFIG_AGP_I830
> #undef whatever_define_i830_is_different_on
> #define whatever xxx
> /* etc */
> #endif
> #ifdef CONFIG_AGP_I840
> #undef whatever
> #define whatever xxx
> /* etc */
> #endif
> 
> and then, voila, we have but one setup function! we can remove all the
> unique i830 and i840 muck...
> 
> Is seperate config statements a problem?  We already have multiple ones
> for the i810/i815 on/off-board versions...Hmm.

If the only differences between the different cards are the nr of
aperture-sizes and the status-register settings, why not have a struct
which contains all the valid cards, and use a scan-routine?!


/David Weinehall
  _                                                                 _
 // David Weinehall <tao@acc.umu.se> /> Northern lights wander      \\
//  Project MCA Linux hacker        //  Dance across the winter sky //
\>  http://www.acc.umu.se/~tao/    </   Full colour fire           </