Re: [PATCH 2/4] ASoC: stm: stm32_i2s: Use guard() for spin locks

[Bui Duc Phuc]

Hi,


> > I'll send a v2 of the whole series with this fix.
> > Would that work for you?
>
> Yes.

Thank you for the confirmation.
By the way, I have an architectural question that came up during my cleanup...

I noticed that the STM drivers currently manage clk_prepare_enable()/
clk_disable_unprepare() directly from the dai_startup()/shutdown() paths.

After looking through various vendor audio drivers, I noticed that
PM/clock handling styles still vary quite a lot between implementations.

For example:

some Intel drivers enable clocks during dai_link init.

Samsung separates bus/interface clocks and operational clocks,
enabling them in different paths such as probe() and set_sysclk()

UX500 and STM enable clocks during DAI startup()/shutdown()

Tegra, Sunxi, and Rockchip often manage clocks through runtime
PM callbacks

>From a maintainer perspective, is there generally interest in gradually
converging these drivers toward more modern/common PM patterns,
or is preserving existing hardware-specific sequencing usually preferred
unless there is a concrete issue to solve?

Best regard,
Phuc