Re: [PATCH 17/22] ASoC: rsnd: Add system suspend/resume support

[Kuninori Morimoto]

Hi John

> You're right. The ALSA framework already handles per-module stop/start
> through SNDRV_PCM_TRIGGER_SUSPEND/RESUME, which calls each module's
> .quit/.init (and thus rsnd_mod_power_off/rsnd_mod_power_on).
> 
> So the per-module clk_prepare/clk_unprepare cycle in rsnd_suspend_mod/
> rsnd_resume_mod is unnecessary.
> 
> What RZ/G3E actually needs beyond the existing ADG handling is:
> 
> 1- Reset handling for modules that have reset control
> 2- audmac-pp clock/reset toggle (infrastructure, like ADG)
> 
> However, there is no need to make these changes conditionally
> based on SoC family as the optional clock/reset APIs are used.
> Do you find any issues with my approach ?

OK

Since we have separated files for each modules, I think it's beter
to follow that style.

This is just an idea

core.c
	void rsnd_suspend_xxx(clk, reset)
	{
		clk_unprepare(clk);
		reset_control_assert(rstc);
	}

	int rsnd_suspend(struct device *dev)
	{

		/*
		 * Reverse order of probe:
		 * ADG -> DVC -> MIX -> CTU -> SRC -> SSIU -> SSI -> DMA
		 */
		rsnd_adg_suspend(...);
		rsnd_dvc_suspend(...);
		rsnd_mix_suspend(...);
		rsnd_ctu_suspend(...);
		rsnd_src_suspend(...);
		...
	}


ssi.c
	void rsnd_ssi_suspend(priv)
	{
		for_each_rsnd_ssi(ssi, priv, i) {
			mod = rsnd_mod_get(ssi);
			rsnd_suspend_xxx(mod->clk, mod->rstc);
		}
	}

src.c
	void rsnd_src_suspend(priv)
	{
		for_each_rsnd_src(src, priv, i) {
			mod = rsnd_mod_get(src);
			rsnd_suspend_xxx(mod->clk, mod->rstc);
		}

		rsnd_suspend_xxx(priv->clk_scu_x2, NULL);
		rsnd_suspend_xxx(priv->clk_scu, NULL);
	}

dma.c
	void rsnd_dma_suspend_xxx(priv)
	{
		rsnd_suspend_xxx(priv->clk_audmac_pp, NULL);
		rsnd_suspend_xxx(priv->rstc_audmac_pp, NULL);
	}


Thank you for your help !!

Best regards
---
Kuninori Morimoto