[PATCH] Fix superio enable calls

[Rudolf Marek]

Hello,

While hunting yet another bug, I noticed that part of my serial log is missing. 
Especially I missed the Allocating & Setting resources.

It turns out that the code which enables specific LDN is somewhat buggy:

static void w83627dhg_pnp_enable(device_t dev)
  {
-	if (!dev->enabled)
-		return;
-
    pnp_enter_ext_func_mode(dev);
    pnp_set_logical_device(dev);
-	pnp_set_enable(dev, 0);
    pnp_exit_ext_func_mode(dev);
  }

So instead of enable the device the device gets disabled. However after some 
time the serial line gets back, most likely some "enable resources" might fix it.

I'm attaching patch which somewhat fixes the problem and changes the function to 
look same in all superio code. Some boards even did not convert the dev->enabled 
to 0,1 values.

Also makes me wonder some sio call pnp_enable directly, which has yet another 
semantics, it wont enable already enabled devices just disable those which 
declared not enabled. And no PnP enter conf magic is called (maybe this is the 
reason).

Signed-off-by: Rudolf Marek

Not abuild tested, works on dhg. I need to know if this fix is OK this way.

Thanks,
Rudolf

[Rudolf Marek]

Hi all,

Can please someone take a look into this?
Thanks,

Rudolf

[Peter Stuge]

Rudolf Marek wrote:
> Can please someone take a look into this?

Thanks for the ping. Good fix. Go for it!

//Peter

[Peter Stuge]

Rudolf Marek wrote:
> Hello,
>
> While hunting yet another bug, I noticed that part of my serial log is 
> missing. Especially I missed the Allocating & Setting resources.
>
> It turns out that the code which enables specific LDN is somewhat buggy:
>
> static void w83627dhg_pnp_enable(device_t dev)
>  {
> if (!dev->enabled)
> return;
> pnp_enter_ext_func_mode(dev);
>  pnp_set_logical_device(dev);
> pnp_set_enable(dev, 0);
> pnp_exit_ext_func_mode(dev);
>
> }
>
> So instead of enable the device the device gets disabled. However after 
> some time the serial line gets back, most likely some "enable resources" 
> might fix it.
>
> I'm attaching patch which somewhat fixes the problem and changes the 
> function to look same in all superio code. Some boards even did not convert 
> the dev->enabled to 0,1 values.
>
> Also makes me wonder some sio call pnp_enable directly, which has yet 
> another semantics, it wont enable already enabled devices just disable 
> those which declared not enabled. And no PnP enter conf magic is called 
> (maybe this is the reason).
>
> Signed-off-by: Rudolf Marek

Acked-by: Peter Stuge