Message-ID: <20240620165746.GS3029315@google.com>
Date: Thu, 20 Jun 2024 17:57:46 +0100
From: Lee Jones <lee@...nel.org>
To: Christian Marangi <ansuelsmth@...il.com>
Cc: Pavel Machek <pavel@....cz>, Rob Herring <robh@...nel.org>,
	Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
	Conor Dooley <conor+dt@...nel.org>,
	Jacek Anaszewski <jacek.anaszewski@...il.com>,
	linux-leds@...r.kernel.org, devicetree@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v6 00/20] leds: leds-lp55xx: overhaul driver

On Thu, 20 Jun 2024, Christian Marangi wrote:

> On Thu, Jun 20, 2024 at 05:09:07PM +0100, Lee Jones wrote:
> > On Sun, 16 Jun 2024, Christian Marangi wrote:
> > 
> > > This long series is (as requested) a big overhaul of the lp55xx based
> > > LED driver.
> > > 
> > > As notice for these kind of LED chip there was the bad habit of copy
> > > the old driver and just modify it enough to make it work with the new
> > > model. Till v4 I was also doing the same by following the pattern and
> > > the code format of previous driver.
> > > 
> > > Since Lee didn't like this, here is the BIG series that generalize
> > > pretty much anything in the 4 model currently supported.
> > > 
> > > Indeed, although the LED chip have fundamental difference (page
> > > support), things can be generalized and produce slimmer drivers by
> > > putting everything in the lp55xx-common shared module.
> > > 
> > > This result in the new model lp5569 being very small with only the
> > > selftest portion to be custom.
> > > 
> > > Lee also wasn't clear by the meaning of ENGINE in these LED driver,
> > > so here some simple explaination. This is very common on these TI LED
> > > chip. The ENGINE (there are always 3) is just some kind of processor
> > > that execute a program (precompiled code ASM like) loaded in the SRAM.
> > > Sysfs is used to load the pattern, and to start and stop the engine.
> > > 
> > > These pattern can do all kind of complex thing with LEDs. Old LED chip
> > > had 32bytes of space for the pattern but newer one (like lp5569) have
> > > pages and if correctly configured can have massive pattern.
> > > These pattern can do all kind of magic like loops that make the LED
> > > pulse, change color and all kind of stuff.
> > > 
> > > (For Lee, sorry if you will have to repeat some review that I might
> > >  have missed in the lp5569 driver)
> > > 
> > > Changes v6:
> > > - Fix compilation warning for ret unused in read_poll_timeout
> > >   (no idea why this is flagged only on some particular arch...)
> > > - Fix missing bitfield.h in lp55x-common.c (again it seems this
> > >   header gets included in the flow if the arch use them or not...)
> > > Changes v5:
> > > - Big generalization patch
> > > - Rework lp5569 driver with new generalized functions
> > > - Drop all copyright header in lp5569 as the driver got reworked
> > >   entirely and it's not based on previous one anymore.
> > > Changes v4:
> > > - Fix reported buffer overflow due to a copypaste error
> > > - Add comments to describe fw size logic
> > > Changes v3:
> > > - Add ACK tag to DT patch
> > > - Enlarge and support program size up to 128bytes
> > > Changes v2:
> > > - Add ACK tag to DT patch
> > > - Fix compilation error with target that doesn't
> > >   include bitfield.h
> > > 
> > > Christian Marangi (20):
> > >   dt-bindings: leds-lp55xx: limit pwr-sel property to ti,lp8501
> > >   dt-bindings: leds-lp55xx: Add new ti,lp5569 compatible
> > >   leds: leds-lp55xx: generalize stop_all_engine OP
> > >   leds: leds-lp55xx: generalize probe/remove functions
> > >   leds: leds-lp55xx: generalize load_engine function
> > >   leds: leds-lp55xx: generalize load_engine_and_select_page function
> > >   leds: leds-lp55xx: generalize run_engine function
> > >   leds: leds-lp55xx: generalize update_program_memory function
> > >   leds: leds-lp55xx: generalize firmware_loaded function
> > >   leds: leds-lp55xx: generalize led_brightness function
> > >   leds: leds-lp55xx: generalize multicolor_brightness function
> > >   leds: leds-lp55xx: generalize set_led_current function
> > >   leds: leds-lp55xx: generalize turn_off_channels function
> > >   leds: leds-lp55xx: generalize stop_engine function
> > >   leds: leds-lp55xx: generalize sysfs engine_load and engine_mode
> > >   leds: leds-lp55xx: generalize sysfs engine_leds
> > >   leds: leds-lp55xx: generalize sysfs master_fader
> > >   leds: leds-lp55xx: support ENGINE program up to 128 bytes
> > >   leds: leds-lp55xx: drop deprecated defines
> > >   leds: leds-lp5569: Add support for Texas Instruments LP5569
> > > 
> > >  .../devicetree/bindings/leds/leds-lp55xx.yaml |  11 +
> > >  drivers/leds/Kconfig                          |  16 +-
> > >  drivers/leds/Makefile                         |   1 +
> > >  drivers/leds/leds-lp5521.c                    | 405 +---------
> > >  drivers/leds/leds-lp5523.c                    | 728 ++---------------
> > >  drivers/leds/leds-lp5562.c                    | 261 +------
> > >  drivers/leds/leds-lp5569.c                    | 542 +++++++++++++
> > >  drivers/leds/leds-lp55xx-common.c             | 730 +++++++++++++++++-
> > >  drivers/leds/leds-lp55xx-common.h             | 133 +++-
> > >  drivers/leds/leds-lp8501.c                    | 313 +-------
> > >  10 files changed, 1523 insertions(+), 1617 deletions(-)
> > >  create mode 100644 drivers/leds/leds-lp5569.c
> > 
> > Generally, I love it - what a difference!
> > 
> > Couple of small things within the patches and few general points.
> > 
> > * Ensure scripts/checkpatch.pl is happy before resubmitting.
> 
> Eh I always run patch with checkpatch.pl --strict but this time is quite
> problematic to fix the single error for the complex macro.
> 
> I will check what I can do but I think I'm really stretching the C
> precompiler there... That is really one of the very few case where error
> might be acceptable.
> 
> If I manage to mute that error, it might result in very bad code...

Just do whatever you think is the right thing and we'll roll with it.

-- 
Lee Jones [李琼斯]

Powered by blists - more mailing lists