Re: [PATCH v10 2/2] leds: Add driver for Qualcomm LPG

[Marijn Suijten <marijn.suijten@somainline.org>]

On 2022-01-28 18:50:42, Bjorn Andersson wrote:
> On Wed 27 Oct 16:19 CDT 2021, Marijn Suijten wrote:
> 
> > Hi Bjorn,
> > 
> > On 2021-10-22 10:25:35, Bjorn Andersson wrote:
> > > On Sat 09 Oct 21:39 PDT 2021, Bjorn Andersson wrote:
> > > 
> > > > The Light Pulse Generator (LPG) is a PWM-block found in a wide range of
> > > > PMICs from Qualcomm. These PMICs typically comes with 1-8 LPG instances,
> > > > with their output being routed to various other components, such as
> > > > current sinks or GPIOs.
> > > > 
> > > > Each LPG instance can operate on fixed parameters or based on a shared
> > > > lookup-table, altering the duty cycle over time. This provides the means
> > > > for hardware assisted transitions of LED brightness.
> > > > 
> > > > A typical use case for the fixed parameter mode is to drive a PWM
> > > > backlight control signal, the driver therefor allows each LPG instance
> > > > to be exposed to the kernel either through the LED framework or the PWM
> > > > framework.
> > > > 
> > > > A typical use case for the LED configuration is to drive RGB LEDs in
> > > > smartphones etc, for which the driver support multiple channels to be
> > > > ganged up to a MULTICOLOR LED. In this configuration the pattern
> > > > generators will be synchronized, to allow for multi-color patterns.
> > > > 
> > > > Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
> > > > ---
> > > 
> > > Any feedback on this?
> > 
> > I asked in #linux-msm whether anything is wrong with the patterns,
> > since my Sony Discovery (sdm630 with a pm660l) blinks way quicker on a
> > pattern that's supposed to stay on for 1s and off for 1s:
> > 
> >     echo "0 1000 255 1000" > /sys/class/leds/rgb\:status/hw_pattern
> > 
> > It however seems to be broken in the same way on an older version now
> > (this might be v9 or v8) which I don't remember to be the case.  Can you
> > double-check if this is all working fine on your side?  If so, I'll have
> > to find some time to debug it on my end.
> > 
> 
> I had missed the fact that LPG_RAMP_DURATION_REG is two registers for
> msg and lsb, for a total of 9 bits of duration. So what you saw was
> probably ticking at 232ms.
> 
> Note though that the pattern uses the last time as "high pause", so I
> expect that you should have seen 232 ms of off, followed by 464ms of
> light.

Visual inspection seems to confirm those numbers indeed!

> I've fixed this for v11, both rejecting invalid input and writing out
> all 9 bits.

Doesn't that 512ms limit, together with using only the last value for
hi_pause (and not the first value for lo_pause) force users to write
patterns in a certain way which is not easily conveyed to the caller
except by reading the comment in the driver?  I'd guess lo_pause can be
used even if not in ping-pong mode, it should just hold at the first
value for the given duration?

(That said hw_pattern is anyway already riddled with device-specific
information, such as only having one `delta_t` which functions as the
step size for every entry, and with the change above would need to be
sourced from another step that's not the first.)

Bit of a stretch, but perhaps worth noting anyway: should this be
written in documentation somewhere, together with pattern examples and
their desired outcome to function as testcases too?

- Marijn