Patch "ASoC: arizona: Poll for FLL clock OK rather than use interrupts" has been added to the 4.2-stable tree

PATCH [email protected] 1 participant 1 message Sep 16, 2015

Subsystems (4)

Stable branch Sound ASoC Wolfson microelectronics drivers

lore

Share & tools

lore

[email protected]

author maintainer Applied

2015-09-16 20:01 UTC #0 lore raw

This is a note to let you know that I've just added the patch titled

ASoC: arizona: Poll for FLL clock OK rather than use interrupts

to the 4.2-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
asoc-arizona-poll-for-fll-clock-ok-rather-than-use-interrupts.patch
and it can be found in the queue-4.2 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

Quoted 1 line

From 0e7659712836ca59b4735bc5cc94de38698a5e01 Mon Sep 17 00:00:00 2001

From: Charles Keepax <[email protected]>
Date: Tue, 25 Aug 2015 12:43:48 +0100
Subject: ASoC: arizona: Poll for FLL clock OK rather than use interrupts

From: Charles Keepax <[email protected]>

commit 0e7659712836ca59b4735bc5cc94de38698a5e01 upstream.

The extcon driver takes the DAPM mutex from within the interrupt thread in several places, which makes it possible to get into a situation where the interrupt thread is blocked waiting on the DAPM mutex whilst a DAPM sequence is running which is attempting to configure the FLL. In this case the FLL completion can't be completed as as the IRQ handler is ONE_SHOT, which cause the FLL lock to use the full time out (250mS) and report that the process timed out.

It is not really practical to make the extcon driver not take the DAPM mutex from within the interrupt thread, at least not without extensive modification. So this patch fixes the issue by switching the wait for the FLL lock to polling. A few fast polls are done first as the FLL should lock quickly for a good quality reference clock, (indeed it hits on the first poll on my system) and it will poll every 20mS after that until it times out.

Signed-off-by Charles Keepax <[email protected]>

Signed-off-by Mark Brown <[email protected]>

Signed-off-by Greg Kroah-Hartman <[email protected]>

---
 sound/soc/codecs/arizona.c |   47 ++++++++++++++++++---------------------------
 sound/soc/codecs/arizona.h |    1 
 2 files changed, 19 insertions(+), 29 deletions(-)

sound/soc/codecs/arizona.c +19 ~~-28~~

@@ -1756,17 +1756,6 @@ int arizona_init_dai(struct arizona_priv
 }
 EXPORT_SYMBOL_GPL(arizona_init_dai);
 
-static irqreturn_t arizona_fll_clock_ok(int irq, void *data)
-{
-	struct arizona_fll *fll = data;
-
-	arizona_fll_dbg(fll, "clock OK\n");
-
-	complete(&fll->ok);
-
-	return IRQ_HANDLED;
-}
-
 static struct {
 	unsigned int min;
 	unsigned int max;

@@ -2048,10 +2037,11 @@ static int arizona_is_enabled_fll(struct
 static int arizona_enable_fll(struct arizona_fll *fll)
 {
 	struct arizona *arizona = fll->arizona;
-	unsigned long time_left;
 	bool use_sync = false;
 	int already_enabled = arizona_is_enabled_fll(fll);
 	struct arizona_fll_cfg cfg;
+	int i;
+	unsigned int val;
 
 	if (already_enabled < 0)
 		return already_enabled;

@@ -2110,9 +2100,6 @@ static int arizona_enable_fll(struct ari
 	if (!already_enabled)
 		pm_runtime_get(arizona->dev);
 
-	/* Clear any pending completions */
-	try_wait_for_completion(&fll->ok);
-
 	regmap_update_bits_async(arizona->regmap, fll->base + 1,
 				 ARIZONA_FLL1_ENA, ARIZONA_FLL1_ENA);
 	if (use_sync)

@@ -2124,10 +2111,24 @@ static int arizona_enable_fll(struct ari
 		regmap_update_bits_async(arizona->regmap, fll->base + 1,
 					 ARIZONA_FLL1_FREERUN, 0);
 
-	time_left = wait_for_completion_timeout(&fll->ok,
-					  msecs_to_jiffies(250));
-	if (time_left == 0)
+	arizona_fll_dbg(fll, "Waiting for FLL lock...\n");
+	val = 0;
+	for (i = 0; i < 15; i++) {
+		if (i < 5)
+			usleep_range(200, 400);
+		else
+			msleep(20);
+
+		regmap_read(arizona->regmap,
+			    ARIZONA_INTERRUPT_RAW_STATUS_5,
+			    &val);
+		if (val & (ARIZONA_FLL1_CLOCK_OK_STS << (fll->id - 1)))
+			break;
+	}
+	if (i == 15)
 		arizona_fll_warn(fll, "Timed out waiting for lock\n");
+	else
+		arizona_fll_dbg(fll, "FLL locked (%d polls)\n", i);
 
 	return 0;
 }

@@ -2212,11 +2213,8 @@ EXPORT_SYMBOL_GPL(arizona_set_fll);
 int arizona_init_fll(struct arizona *arizona, int id, int base, int lock_irq,
 		     int ok_irq, struct arizona_fll *fll)
 {
-	int ret;
 	unsigned int val;
 
-	init_completion(&fll->ok);
-
 	fll->id = id;
 	fll->base = base;
 	fll->arizona = arizona;

@@ -2238,13 +2236,6 @@ int arizona_init_fll(struct arizona *ari
 	snprintf(fll->clock_ok_name, sizeof(fll->clock_ok_name),
 		 "FLL%d clock OK", id);
 
-	ret = arizona_request_irq(arizona, ok_irq, fll->clock_ok_name,
-				  arizona_fll_clock_ok, fll);
-	if (ret != 0) {
-		dev_err(arizona->dev, "Failed to get FLL%d clock OK IRQ: %d\n",
-			id, ret);
-	}
-
 	regmap_update_bits(arizona->regmap, fll->base + 1,
 			   ARIZONA_FLL1_FREERUN, 0);

sound/soc/codecs/arizona.h +0 -1

@@ -242,7 +242,6 @@ struct arizona_fll {
 	int id;
 	unsigned int base;
 	unsigned int vco_mult;
-	struct completion ok;
 
 	unsigned int fout;
 	int sync_src;

Patches currently in stable-queue which might be from [email protected] are

queue-4.2/asoc-arizona-poll-for-fll-clock-ok-rather-than-use-interrupts.patch queue-4.2/asoc-arizona-fix-gain-settings-of-fll-in-free-run-mode.patch