Re: [PATCH 04/10] ARM: tegra: pm: add platform suspend support

[Stephen Warren <swarren-3lzwWm7+Weoh9ZMKESR00Q@public.gmane.org>]

On 03/04/2013 04:40 AM, Joseph Lo wrote:
> Adding suspend to ram support for Tegra platform. There are three suspend

nit: s/ram/RAM/

> mode for Tegra. The difference were below.

Does this patch work for all 3 upstream Tegras, or just Tegra20 (or 20+30?).

> diff --git a/arch/arm/mach-tegra/common.c b/arch/arm/mach-tegra/common.c

> +	tegra_init_suspend();

That call isn't ifdef'd on anything ...

> diff --git a/arch/arm/mach-tegra/pm.c b/arch/arm/mach-tegra/pm.c

> +static void tegra_suspend_enter_lp2(void)
> +{
> +	tegra_set_cpu_in_lp2(0);
> +}
> +
> +static void tegra_suspend_exit_lp2(void)
> +{
> +	tegra_clear_cpu_in_lp2(0);
> +}

Are those two functions going to get larger? If not, why not just call
tegra_{set,clear}_cpu_in_lp2() from tegra_suspend_enter()?

Note: I realized that the LP0/LP1 functions might be larger, and feel
free to make them separate functions when they're implemented, but if
those two functions specifically are always going to be a single line,
then there's little point in separating them out, is there?

> +void __init tegra_init_suspend(void)
...
>  #endif

... but this function is only implemented inside an ifdef ...

> diff --git a/arch/arm/mach-tegra/pm.h b/arch/arm/mach-tegra/pm.h

> +void tegra_init_suspend(void);

... and no dummy inline is provided if the ifdef isn't defined.

So, turning off whatever that ifdef is would cause a build error.

> diff --git a/arch/arm/mach-tegra/pmc.c b/arch/arm/mach-tegra/pmc.c

> +unsigned long tegra_pmc_get_cpu_timer(void)
...
> +unsigned long tegra_pmc_get_cpu_off_timer(void)

I think you want to make those return u32, since the values they return
should be u32 to match DT cell type.

> +void tegra_pmc_pm_set(enum tegra_suspend_mode mode)
> +{
> +	u32 reg;
> +
> +	reg = tegra_pmc_readl(PMC_CTRL);
> +	reg |= TEGRA_POWER_CPU_PWRREQ_OE;
> +	reg &= ~TEGRA_POWER_EFFECT_LP0;
> +
> +	tegra_pmc_writel(reg, PMC_CTRL);
> +}

The "mode" parameter doesn't seem to be used here.

One overall comment: tegra_suspend_enter() only supports LP2 so far, but
I don't see a check anywhere (e.g. tegra_init_suspend() or
tegra_pmc_parse_dt()) which prevents suspend from being initialized if
the user requested some other type, or on SoCs where even LP2 isn't
supported yet (e.g. Tegra114).