[Intel-gfx] Prepare for 2.3.2

[Jesse Barnes]

On Thursday, June 12, 2008 11:22 pm Lukas Hejtmanek wrote:
> On Fri, Jun 13, 2008 at 10:19:18AM +0800, Zhenyu Wang wrote:
> > I've pushed fixes from master I'd want to include in final 2.3.2
> > release, which targets Xorg 7.4.
> >
> > I'll run another test cycle mostly on new mobile chipset, and
> > testing help is needed for all previous chips.
> >
> > Please help to try current xf86-video-intel-2.3-branch and report
> > any regression you can find.
>
> with the current head I have problems with backlight again. The Xserver
> starts with maximum backlight.
>
> The following patch helps:

Did you try what I suggested in response your last patch?  I thought Zhenyu 
would push that, but I guess the code hasn't changed for awhile.

> +    memset(val, 0, sizeof(val));
>      if (read(fd, val, BACKLIGHT_VALUE_LEN) == -1)
>         goto out_err;

This looks like a good fix regardless of anything else.

> @@ -1081,7 +1090,10 @@ i830_lvds_set_property(xf86OutputPtr output, Atom
> property, "RRConfigureOutputProperty error, %d\n", ret); }
>         /* Set the current value of the backlight property */
> -       data = dev_priv->get_backlight(output);
> +       if ((INREG(PP_CONTROL)&POWER_TARGET_ON) && dev_priv->dpmsoff == 0)
> +          data = dev_priv->get_backlight(output);
> +       else
> +          data = dev_priv->backlight_duty_cycle;
>         ret = RRChangeOutputProperty(output->randr_output, backlight_atom,
>                                      XA_INTEGER, 32, PropModeReplace, 1,
> &data, FALSE, TRUE);
> @@ -1140,8 +1152,12 @@ i830_lvds_get_property(xf86OutputPtr output, Atom
> property) */
>      if (property == backlight_atom) {
>         int val;
> -       val = dev_priv->get_backlight(output);
> -       dev_priv->backlight_duty_cycle = val;
> +       if ((INREG(PP_CONTROL)&POWER_TARGET_ON) && dev_priv->dpmsoff == 0)
> +               val = dev_priv->get_backlight(output);
> +               dev_priv->backlight_duty_cycle = val;
> +       } else {
> +               val = dev_priv->backlight_duty_cycle;
> +       }
>         ret = RRChangeOutputProperty(output->randr_output, backlight_atom,
>                                      XA_INTEGER, 32, PropModeReplace, 1,
> &val, FALSE, TRUE);

These changes are to avoid returning a 0 backlight brightness when the LCD is 
off?  Or did you find cases where the value wasn't 0 even though the display 
was supposed to be off?

Thanks,
Jesse