On Sat, 30 Jan 1999, Eduardo E. Horvath wrote:

> On Fri, 29 Jan 1999, Matthew Jacob wrote:
> 
> > >    
> > >    One of the things I'm considering is to change the drivers to access all
> > >    device registers through the passthrough ASIs so the registers never need
> > >    to be mapped in.  The problem is that then I would need to rewrite all
> > >    drivers to properly use the bus_space_{read,write}*() macros for all
> > >    register accesses.
> > > 
> > > 
> > > 
> > > problem!?  this would be excellent, not a problem!  :-)
> 
> That's an awful lot of drivers that need to be gone through.
> 
> > And on that note, I just finished converting isp_sbus- I dunno how it
> > managed to not have bus_space_XXXX stuff...
> > 
> > I'd wonder about the performance of using ASI's....
> 
> Should be faster since they explicitly bypass the MMU and don't take TLB
> miss traps.  However, the current scheme using gcc macros is not optimal
> since gcc can't do as good a job at scheduling the instructions.  The best
> solution would be to modify gcc so pointers could be explicitly assocated
> with ASIs.

i'm kinda new with inline asm and gcc, but can't you 'anonymoutize' the
asms to take any register as input and hopefully only clobber one
register?

i'd show some example code, but it'd most likly be incorrect...

-Alfred