> -----Original Message----- > From: Linus Walleij <linus.walleij@xxxxxxxxxx> > Sent: Wednesday, January 24, 2024 5:51 AM > To: Paller, Kim Seer <KimSeer.Paller@xxxxxxxxxx> > Cc: Bartosz Golaszewski <brgl@xxxxxxxx>; linux-gpio@xxxxxxxxxxxxxxx; > devicetree@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Rob Herring > <robh+dt@xxxxxxxxxx>; Krzysztof Kozlowski > <krzysztof.kozlowski+dt@xxxxxxxxxx>; Conor Dooley <conor+dt@xxxxxxxxxx> > Subject: Re: [PATCH 2/2] gpio: gpio-adg1414: New driver > > [External] > > Hi Kim, > > On Tue, Jan 23, 2024 at 11:31 AM Paller, Kim Seer > <KimSeer.Paller@xxxxxxxxxx> wrote: > > > > Locking here is simple enough that you could use the SPI regmap and > > > get it to do the serialization for you. And then you could possibly > > > reuse the gpio-regmap abstraction and get an even smaller footprint. > > > > I could not seem to figure out how to use the SPI regmap in this case. > > Since the number of daisy-chained devices depends on the length of > > data transferred with continuous transaction, I could not determine > > how to implement that using the SPI regmap. Or maybe I misunderstood > > the statement. However, is it still acceptable to use the current approach? > > You just override or wrap with your own read/write callbacks if necessary > by defining a custom static struct regmap_bus. > > For example in drivers/gpu/drm/panel/panel-ilitek-ili9322.c > I do this. > > It may not save a lot of code in this case but it's still worth it because > we understand what regmap_read/write/update_bits do and reading > and understanding adg1414_set/get cognitively require more from us > as maintainers. Hi Linus, I'll check this out. Thanks, Kim
