[PATCH] ASoC: bcm2835: fix hw_params error when device is in prepared state
Matthias Reichl
hias at horus.com
Wed Jan 17 04:50:50 PST 2018
If bcm2835 is configured as bitclock master calling hw_params()
after prepare() fails with EBUSY. This also makes it impossible to
use bcm2835 in full duplex mode.
The error is caused by the split clock setup: clk_set_rate
is called in hw_params, clk_prepare_enable in prepare. As hw_params
doesn't check if the clock was already enabled clk_set_rate
fails with EBUSY.
Fix this by moving clock startup from prepare to hw_params and
let hw_params properly deal with an already set up or enabled
clock.
Signed-off-by: Matthias Reichl <hias at horus.com>
---
sound/soc/bcm/bcm2835-i2s.c | 20 ++++++++++++++------
1 file changed, 14 insertions(+), 6 deletions(-)
diff --git a/sound/soc/bcm/bcm2835-i2s.c b/sound/soc/bcm/bcm2835-i2s.c
index 2e449d7173fc..d5f73a8ab893 100644
--- a/sound/soc/bcm/bcm2835-i2s.c
+++ b/sound/soc/bcm/bcm2835-i2s.c
@@ -130,6 +130,7 @@ struct bcm2835_i2s_dev {
struct regmap *i2s_regmap;
struct clk *clk;
bool clk_prepared;
+ int clk_rate;
};
static void bcm2835_i2s_start_clock(struct bcm2835_i2s_dev *dev)
@@ -419,10 +420,19 @@ static int bcm2835_i2s_hw_params(struct snd_pcm_substream *substream,
}
/* Clock should only be set up here if CPU is clock master */
- if (bit_clock_master) {
- ret = clk_set_rate(dev->clk, bclk_rate);
- if (ret)
- return ret;
+ if (bit_clock_master &&
+ (!dev->clk_prepared || dev->clk_rate != bclk_rate)) {
+ if (dev->clk_prepared)
+ bcm2835_i2s_stop_clock(dev);
+
+ if (dev->clk_rate != bclk_rate) {
+ ret = clk_set_rate(dev->clk, bclk_rate);
+ if (ret)
+ return ret;
+ dev->clk_rate = bclk_rate;
+ }
+
+ bcm2835_i2s_start_clock(dev);
}
/* Setup the frame format */
@@ -618,8 +628,6 @@ static int bcm2835_i2s_prepare(struct snd_pcm_substream *substream,
struct bcm2835_i2s_dev *dev = snd_soc_dai_get_drvdata(dai);
uint32_t cs_reg;
- bcm2835_i2s_start_clock(dev);
-
/*
* Clear both FIFOs if the one that should be started
* is not empty at the moment. This should only happen
--
2.11.0
More information about the linux-rpi-kernel
mailing list