[PATCH] v3: Add/use global PNP enter/exit functions for Super I/Os

[Uwe Hermann]

This is the start of a v3 superio refactoring, more will follow.
Build-tested on one or two targets.

Uwe.

[PATCH] v3: Add/use global PNP enter/exit functions forSuper I/Os

[Myles Watson]

> This is the start of a v3 superio refactoring, more will follow.
> Build-tested on one or two targets.

for fintek:
-static void pnp_enter_conf_state(struct device *dev)
-{
-	outb(0x87, dev->path.pnp.port);
-}

-	pnp_enter_conf_state(dev);
+	pnp_enter_8787(dev);

I don't know that much about SuperIOs, but does it matter that you replaced
one outb with two?

At least in pnp_enter_ite() it seems to matter.

Besides that it looks good.

Acked-by: Myles Watson

Thanks,
Myles

[PATCH] v3: Add/use global PNP enter/exit functions forSuper I/Os

[Corey Osgood]

On Tue, Nov 18, 2008 at 3:34 PM, Myles Watson  wrote:
> > This is the start of a v3 superio refactoring, more will follow.
> > Build-tested on one or two targets.
>
> for fintek:
> -static void pnp_enter_conf_state(struct device *dev)
> -{
>   outb(0x87, dev->path.pnp.port);
>
> -}
>   pnp_enter_conf_state(dev);
>   pnp_enter_8787(dev);
>
> I don't know that much about SuperIOs, but does it matter that you replaced
> one outb with two?

No, it doesn't, I've used two 0x87s on actual hardware.

-Corey
> At least in pnp_enter_ite() it seems to matter.
>
> Besides that it looks good.
>
> Acked-by: Myles Watson

[PATCH] v3: Add/use global PNP enter/exit functions forSuper I/Os

[Myles Watson]

On Tue, Nov 18, 2008 at 1:50 PM, Corey Osgood wrote:
> On Tue, Nov 18, 2008 at 3:34 PM, Myles Watson  wrote:
> > > This is the start of a v3 superio refactoring, more will follow.
> > > Build-tested on one or two targets.
> >
> > for fintek:
> > -static void pnp_enter_conf_state(struct device *dev)
> > -{
> >   outb(0x87, dev->path.pnp.port);
> >
> > -}
> >   pnp_enter_conf_state(dev);
> >   pnp_enter_8787(dev);
> >
> > I don't know that much about SuperIOs, but does it matter that you
> > replaced
> > one outb with two?
>
> No, it doesn't, I've used two 0x87s on actual hardware.
>
> -Corey

Good to know.  Interesting that the second one matters to the ite chip.

Thanks,
Myles
> > At least in pnp_enter_ite() it seems to matter.
> >
> > Besides that it looks good.
> >
> > Acked-by: Myles Watson

[PATCH] v3: Add/use global PNP enter/exit functions forSuper I/Os

[Uwe Hermann]

On Tue, Nov 18, 2008 at 02:16:32PM -0700, Myles Watson wrote:
> > > for fintek:
> > > -static void pnp_enter_conf_state(struct device *dev)
> > > -{
> > >   outb(0x87, dev->path.pnp.port);
> > >
> > > -}
> > >   pnp_enter_conf_state(dev);
> > >   pnp_enter_8787(dev);
> > >
> > > I don't know that much about SuperIOs, but does it matter that you
> > > replaced
> > > one outb with two?
> >
> > No, it doesn't, I've used two 0x87s on actual hardware.

Yep. Also, the same is true for some SMSC Super I/Os which use one
0x55, but 0x55,0x55 will also work. This will just make it "enter"
twice and doesn't seem to be a problem.
> Good to know.  Interesting that the second one matters to the ite chip.

Most ITE chips use a four byte sequence to enter config mode, and the
datasheet explicitly defines which four bytes you have to use (depends on
the chip _and_ on the config port the chip is located on, which is
board-specific). If you change one of the bytes, it won't work anymore,
whereas for the 0x87 vs. 0x87,0x87 is still works, it simply works
_twice_ :)
> > > Acked-by: Myles Watson

Thanks, r1044.

Uwe.