Re: [PATCH v6] drm/dp: clamp PWM bit count to advertised MIN and MAX capabilities

[Johan Hovold <johan@kernel.org>]

On Thu, Jul 24, 2025 at 02:09:10PM +0300, Dmitry Baryshkov wrote:
> On 24/07/2025 12:42, Neil Armstrong wrote:
> > On 24/07/2025 11:32, Dmitry Baryshkov wrote:
> >> On Thu, 24 Jul 2025 at 12:08, <neil.armstrong@linaro.org> wrote:
> >>> On 20/05/2025 10:06, Johan Hovold wrote:
> >>>> On Fri, Apr 04, 2025 at 02:24:32PM +0100, Christopher Obbard wrote:
> >>>>> On Fri, 4 Apr 2025 at 09:54, Johan Hovold <johan@kernel.org> wrote:
> >>>>>> On Fri, Apr 04, 2025 at 08:54:29AM +0100, Christopher Obbard wrote:
> >>>>>>> On Mon, 31 Mar 2025 at 09:33, Johan Hovold <johan@kernel.org> wrote:
> >>>>>>>>> @@ -4035,6 +4036,32 @@ drm_edp_backlight_probe_max(struct 

> >>>>>>>>> drm_dp_aux *aux, struct drm_edp_backlight_inf
> >>>>>>>>>         }
> >>>>>>>>>
> >>>>>>>>>         pn &= DP_EDP_PWMGEN_BIT_COUNT_MASK;
> >>>>>>>>> +
> >>>>>>>>> +     ret = drm_dp_dpcd_read_byte(aux, 
> >>>>>>>>> DP_EDP_PWMGEN_BIT_COUNT_CAP_MIN, &pn_min);
> >>>>>>>>> +     if (ret < 0) {
> >>>>>>>>> +             drm_dbg_kms(aux->drm_dev, "%s: Failed to read 
> >>>>>>>>> pwmgen bit count cap min: %d\n",
> >>>>>>>>> +                         aux->name, ret);
> >>>>>>>>> +             return -ENODEV;
> >>>>>>>>> +     }
> >>>>>>>>> +     pn_min &= DP_EDP_PWMGEN_BIT_COUNT_MASK;
> >>>>>>>>> +
> >>>>>>>>> +     ret = drm_dp_dpcd_read_byte(aux, 
> >>>>>>>>> DP_EDP_PWMGEN_BIT_COUNT_CAP_MAX, &pn_max);
> >>>>>>>>> +     if (ret < 0) {
> >>>>>>>>> +             drm_dbg_kms(aux->drm_dev, "%s: Failed to read 
> >>>>>>>>> pwmgen bit count cap max: %d\n",
> >>>>>>>>> +                         aux->name, ret);
> >>>>>>>>> +             return -ENODEV;
> >>>>>>>>> +     }
> >>>>>>>>> +     pn_max &= DP_EDP_PWMGEN_BIT_COUNT_MASK;
> >>>>>>>>> +
> >>>>>>>>> +     /*
> >>>>>>>>> +      * Per VESA eDP Spec v1.4b, section 3.3.10.2:
> >>>>>>>>> +      * If DP_EDP_PWMGEN_BIT_COUNT is less than 
> >>>>>>>>> DP_EDP_PWMGEN_BIT_COUNT_CAP_MIN,
> >>>>>>>>> +      * the sink must use the MIN value as the effective PWM 
> >>>>>>>>> bit count.
> >>>>>>>>> +      * Clamp the reported value to the [MIN, MAX] capability 
> >>>>>>>>> range to ensure
> >>>>>>>>> +      * correct brightness scaling on compliant eDP panels.
> >>>>>>>>> +      */
> >>>>>>>>> +     pn = clamp(pn, pn_min, pn_max);
> >>>>>>>>
> >>>>>>>> You never make sure that pn_min <= pn_max so you could end up with
> >>>>>>>> pn < pn_min on broken hardware here. Not sure if it's something 
> >>>>>>>> you need
> >>>>>>>> to worry about at this point.
> >>>
> >>> I'm trying to figure out what would be the behavior in this case ?
> >>>
> >>> - Warn ?
> >>> - pn_max = pn_min ?
> >>> - use BIT_COUNT as-is and ignore MIN/MAX ?
> >>> - pm_max = max(pn_min, pn_max); pm_min = min(pn_min, pn_max); ?
> >>> - reverse clamp? clamp(pn, pn_max, pn_min); ?
> >>> - generic clamp? clamp(pn, min(pn_min, pn_max), max(pn_min, pn_max)); ?
> >>
> >> Per the standard, the min >= 1 and max >= min. We don't need to bother
> >> about anything here.
> > 
> > Yeah, I agree. But I think a:
> > if (likely(pn_min <= pn_max))
> > is simple and doesn't cost much..
> 
> Really, no need to.

It doesn't matter what the spec says, what matters is what may happen if
a device violates the spec (e.g. if a driver triggers a division by
zero).

Always sanitise your input.

(But there is no need for likely() as this is not a hot path.)

Johan