Skip to contentSkip to conversation

ASoC: pxa-ssp: make suspend call resilient against multiple invocations

PATCH In review Daniel Mack 2 participants 2 messages Nov 6, 2018

Subsystems (3)
Sound ASoC PXA2xx/PXA3xx support
lore
Share & tools
lore

Daniel Mack

author maintainer
#0 lore raw

PATCH 1/1 ASoC: pxa-ssp: make suspend call resilient against multiple invocations

The suspend callback is called twice when the machine is suspended, once through the ASoC layer and once through the platform device layer. Make sure not to confuse the clock layer with unbalanced calls, and bail out early if we are already suspended.

Signed-off-by Daniel Mack <[email protected]>
---
 sound/soc/pxa/pxa-ssp.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
sound/soc/pxa/pxa-ssp.c +11 -0
index adcf8ba9d287..fd93706702a6 100644
@@ -51,6 +51,7 @@ struct ssp_priv {
 	uint32_t	cr1;
 	uint32_t	to;
 	uint32_t	psp;
+	bool		suspended;
 #endif
 };
 

@@ -142,6 +143,9 @@ static int pxa_ssp_suspend(struct snd_soc_dai *cpu_dai)
 	struct ssp_priv *priv = snd_soc_dai_get_drvdata(cpu_dai);
 	struct ssp_device *ssp = priv->ssp;
 
+	if (priv->suspended)
+		return 0;
+
 	if (!cpu_dai->active)
 		clk_prepare_enable(ssp->clk);
 

@@ -152,6 +156,8 @@ static int pxa_ssp_suspend(struct snd_soc_dai *cpu_dai)
 
 	pxa_ssp_disable(ssp);
 	clk_disable_unprepare(ssp->clk);
+	priv->suspended = true;
+
 	return 0;
 }
 

@@ -161,6 +167,9 @@ static int pxa_ssp_resume(struct snd_soc_dai *cpu_dai)
 	struct ssp_device *ssp = priv->ssp;
 	uint32_t sssr = SSSR_ROR | SSSR_TUR | SSSR_BCE;
 
+	if (!priv->suspended)
+		return 0;
+
 	clk_prepare_enable(ssp->clk);
 
 	__raw_writel(sssr, ssp->mmio_base + SSSR);

@@ -174,6 +183,8 @@ static int pxa_ssp_resume(struct snd_soc_dai *cpu_dai)
 	else
 		clk_disable_unprepare(ssp->clk);
 
+	priv->suspended = false;
+
 	return 0;
 }
Signature · 1 line
2.17.2
maintainer
#1 lore raw
Quoted 4 lines · Daniel Mack

The suspend callback is called twice when the machine is suspended, once through the ASoC layer and once through the platform device layer. Make sure not to confuse the clock layer with unbalanced calls, and bail out early if we are already suspended.

This feels like something we should be fixing at the core level - as far as I can tell there's no PXA specific duplication going on here, it's the core that's triggering the suspend operation twice so it's likely that there's going to be problems on other platforms somehow. Or is there something PXA specific I'm missing here?