On Fri, 26 Sep 2014, Geert Uytterhoeven wrote:

On Thu, Sep 25, 2014 at 6:28 PM, Geert Uytterhoeven
[off-list ref] wrote:

quoted

diff --git a/arch/arm/mach-shmobile/pm-rmobile.c b/arch/arm/mach-shmobile/pm-rmobile.c
index 717e6413d29cb998..ae846dfcd69293f7 100644
--- a/arch/arm/mach-shmobile/pm-rmobile.c
+++ b/arch/arm/mach-shmobile/pm-rmobile.c

@@ -101,6 +101,17 @@ static bool rmobile_pd_active_wakeup(struct device *dev)
        return true;
 }

+static void rmobile_pd_attach_dev(struct device *dev)
+{
+       if (pm_clk_no_clocks(dev))
+               pm_clk_add(dev, NULL);

With !CONFIG_PM_RUNTIME, this doesn't work well in the DT case,
as the clocks are no longer enabled by runtime PM
(some clocks may have been enabled by boot loader/reset state though).

In the legacy case, this works more or less, as genpd takes over after
the pm_clk domain has been initialized, which implies enabling the clocks
if !CONFIG_PM_RUNTIME.

Will fix...

The patch below fixes that.

I'll update the whole series, and will repost after v3.18-rc1, after the
prerequisites are in.

From cf40afcfab65fef8e6e6fa84bd3b49434afcb5eb Mon Sep 17 00:00:00 2001

From: Geert Uytterhoeven <geert+renesas@glider.be>
Date: Wed, 1 Oct 2014 20:05:53 +0200
Subject: [PATCH] ARM: shmobile: R-Mobile: Enable module clocks if !CONFIG_PM_RUNTIME

ARM: shmobile: R-Mobile: Enable module clocks if !CONFIG_PM_RUNTIME

If CONFIG_PM_RUNTIME is not set, the clocks must be enabled before the
devices can be used.  Also add missing pm_clk_create()/pm_clk_destroy()
calls.

Currently these clocks are enabled by the !CONFIG_PM_RUNTIME version of
pm_clk_notify, activated by the hack in drivers/sh/pm_runtime.c, but
this will go away soon (at least for DT, cfr. "[PATCH] drivers: sh:
Disable PM runtime for multi-platform r8a7740 with genpd").

On detach, disabling the clocks is already handled by pm_clk_destroy().
On system suspend/resume, disabling/enabling clocks is already handled
from the genpd->dev_ops.{start,stop}() callbacks.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
---
This depends on "PM / clock_ops: Fix crash in clocks management code if
!CONFIG_PM_RUNTIME".

 arch/arm/mach-shmobile/pm-rmobile.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/arch/arm/mach-shmobile/pm-rmobile.c b/arch/arm/mach-shmobile/pm-rmobile.c
index 80be5848adc8f15c..e5a2c568d96cdd6b 100644
--- a/arch/arm/mach-shmobile/pm-rmobile.c
+++ b/arch/arm/mach-shmobile/pm-rmobile.c

@@ -121,13 +121,23 @@ static bool rmobile_pd_active_wakeup(struct device *dev)
 
 static void rmobile_pd_attach_dev(struct device *dev)
 {
-	if (pm_clk_no_clocks(dev))
-		pm_clk_add(dev, NULL);
+	int error;
+
+	error = pm_clk_create(dev);
+	if (error) {
+		dev_err(dev, "pm_clk_create failed %d\n", error);
+		return;
+	}
+
+	pm_clk_add(dev, NULL);
+
+	if (!IS_ENABLED(CONFIG_PM_RUNTIME))
+		pm_clk_resume(dev);
 }
 
 static void rmobile_pd_detach_dev(struct device *dev)
 {
-	pm_clk_remove(dev, NULL);
+	pm_clk_destroy(dev);
 }
 
 static void rmobile_init_pm_domain(struct rmobile_pm_domain *rmobile_pd)

-- 
1.9.1

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds