Re: [PATCH V7 6/6] backlight: qcom-wled: Add auto string detection logic

Re: [PATCH V7 6/6] backlight: qcom-wled: Add auto string detection logic

[kgunda]

On 2019-10-17 19:00, Lee Jones wrote:
> On Thu, 17 Oct 2019, kgunda@codeaurora.org wrote:
> 
>> On 2019-10-17 17:56, Lee Jones wrote:
>> > On Wed, 16 Oct 2019, Kiran Gunda wrote:
>> >
>> > > The auto string detection algorithm checks if the current WLED
>> > > sink configuration is valid. It tries enabling every sink and
>> > > checks if the OVP fault is observed. Based on this information
>> > > it detects and enables the valid sink configuration.
>> > > Auto calibration will be triggered when the OVP fault interrupts
>> > > are seen frequently thereby it tries to fix the sink configuration.
>> > >
>> > > The auto-detection also kicks in when the connected LED string
>> > > of the display-backlight malfunctions (because of damage) and
>> > > requires the damaged string to be turned off to prevent the
>> > > complete panel and/or board from being damaged.
>> > >
>> > > Signed-off-by: Kiran Gunda <kgunda@codeaurora.org>
>> > > ---
>> > >  drivers/video/backlight/qcom-wled.c | 410
>> > > +++++++++++++++++++++++++++++++++++-
>> > >  1 file changed, 404 insertions(+), 6 deletions(-)
>> > >
>> > > diff --git a/drivers/video/backlight/qcom-wled.c
>> > > b/drivers/video/backlight/qcom-wled.c
>> > > index b5b125c..ff7c409 100644
>> > > --- a/drivers/video/backlight/qcom-wled.c
>> > > +++ b/drivers/video/backlight/qcom-wled.c
> 
> [...]
> 
>> > > +		if (int_sts & WLED3_CTRL_REG_OVP_FAULT_STATUS)
>> > > +			dev_dbg(wled->dev, "WLED OVP fault detected with SINK %d\n",
>> > > +				i + 1);
>> >
>> > I haven't reviewed the whole patch, but this caught my eye.
>> >
>> > I think this should be upgraded to dev_warn().
>> >
>> Thought of keeping these messages silent, Because the string 
>> configuration
>> will be corrected in this
>> and informing it at end of the auto string detection.
> 
> [...]
> 
>> > > +	} else {
>> > > +		dev_warn(wled->dev, "New WLED string configuration found %x\n",
>> > > +			 sink_valid);
>> >
>> > Why would the user care about this?  Is it not normal?
>> >
>> Actually, it comes here if the user provided string configuration in 
>> the
>> device tree is in-correct.
>> That's why just informing the user about the right string 
>> configuration,
>> after the auto string detection.
> 
> Then I think we need to be more forthcoming.  Tell the user the
> configuration is incorrect and what you've done to rectify it.
> 
> "XXX is not a valid configuration - using YYY instead"
> 
Sure. Will update it in the next series.
> [...]
> 
>> > > +static int wled_configure_ovp_irq(struct wled *wled,
>> > > +				  struct platform_device *pdev)
>> > > +{
>> > > +	int rc;
>> > > +	u32 val;
>> > > +
>> > > +	wled->ovp_irq = platform_get_irq_byname(pdev, "ovp");
>> > > +	if (wled->ovp_irq < 0) {
>> > > +		dev_dbg(&pdev->dev, "ovp irq is not used\n");
>> >
>> > I assume this is optional.  What happens if no IRQ is provided?
>> >
>> Here OVP IRQ is used to detect the wrong string detection. If it is 
>> not
>> provided the auto string detection logic won't work.
> 
> "OVP IRQ not found - disabling automatic string detection"
> 
Sure. Will update it in the next series.
>> > If, for instance, polling mode is enabled, maybe something like this
>> > would be better?
>> >
>> >       dev_warn(&pdev->dev, "No IRQ found, falling back to polling
>> > mode\n");

Re: [PATCH V7 6/6] backlight: qcom-wled: Add auto string detection logic

[kgunda]

On 2019-10-17 19:09, Daniel Thompson wrote:
> On Thu, Oct 17, 2019 at 05:47:47PM +0530, kgunda@codeaurora.org wrote:
>> On 2019-10-17 16:59, Daniel Thompson wrote:
>> > On Wed, Oct 16, 2019 at 03:43:46PM +0530, Kiran Gunda wrote:
>> > > The auto string detection algorithm checks if the current WLED
>> > > sink configuration is valid. It tries enabling every sink and
>> > > checks if the OVP fault is observed. Based on this information
>> > > it detects and enables the valid sink configuration.
>> > > Auto calibration will be triggered when the OVP fault interrupts
>> > > are seen frequently thereby it tries to fix the sink configuration.
>> > >
>> > > The auto-detection also kicks in when the connected LED string
>> > > of the display-backlight malfunctions (because of damage) and
>> > > requires the damaged string to be turned off to prevent the
>> > > complete panel and/or board from being damaged.
>> > >
>> > > Signed-off-by: Kiran Gunda <kgunda@codeaurora.org>
>> >
>> > It's a complex bit of code but I'm OK with it in principle. Everything
>> > below is about small details and/or nitpicking.
>> >
>> >
>> > > +static void wled_ovp_work(struct work_struct *work)
>> > > +{
>> > > +	struct wled *wled = container_of(work,
>> > > +					 struct wled, ovp_work.work);
>> > > +	enable_irq(wled->ovp_irq);
>> > > +}
>> > > +
>> >
>> > A bit of commenting about why we have to wait 10ms before enabling the
>> > OVP interrupt would be appreciated.
>> >
>> >
>> Sure. Will add the comment in the next series.
>> > > +static irqreturn_t wled_ovp_irq_handler(int irq, void *_wled)
>> > > +{
>> > > +	struct wled *wled = _wled;
>> > > +	int rc;
>> > > +	u32 int_sts, fault_sts;
>> > > +
>> > > +	rc = regmap_read(wled->regmap,
>> > > +			 wled->ctrl_addr + WLED3_CTRL_REG_INT_RT_STS, &int_sts);
>> > > +	if (rc < 0) {
>> > > +		dev_err(wled->dev, "Error in reading WLED3_INT_RT_STS rc=%d\n",
>> > > +			rc);
>> > > +		return IRQ_HANDLED;
>> > > +	}
>> > > +
>> > > +	rc = regmap_read(wled->regmap, wled->ctrl_addr +
>> > > +			 WLED3_CTRL_REG_FAULT_STATUS, &fault_sts);
>> > > +	if (rc < 0) {
>> > > +		dev_err(wled->dev, "Error in reading WLED_FAULT_STATUS rc=%d\n",
>> > > +			rc);
>> > > +		return IRQ_HANDLED;
>> > > +	}
>> > > +
>> > > +	if (fault_sts &
>> > > +		(WLED3_CTRL_REG_OVP_FAULT_BIT | WLED3_CTRL_REG_ILIM_FAULT_BIT))
>> > > +		dev_dbg(wled->dev, "WLED OVP fault detected, int_sts=%x
>> > > fault_sts= %x\n",
>> > > +			int_sts, fault_sts);
>> > > +
>> > > +	if (fault_sts & WLED3_CTRL_REG_OVP_FAULT_BIT) {
>> > > +		mutex_lock(&wled->lock);
>> > > +		disable_irq_nosync(wled->ovp_irq);
>> >
>> > We're currently running the threaded ISR for this irq. Do we really need
>> > to disable it?
>> >
>> We need to disable this IRQ, during the auto string detection logic. 
>> Because
>> in the auto string detection we configure the current sinks one by one 
>> and
>> check the
>> status register for the OVPs and set the right string configuration. 
>> We
>> enable it later after
>> the auto string detection is completed.
> 
> This is a threaded oneshot interrupt handler. Why isn't the framework
> masking sufficient for you here?
> 
> 
> Daniel.
Right .. I overlooked that it is a oneshot interrupt earlier.
I will address it in the next series.