Re: [PATCH] ASoC: fsl_ssi: Override bit clock rate based on slot number

From: Arnaud Mouiche
Date: Tue Sep 12 2017 - 10:35:24 EST

Hello Nicolin

On 08/09/2017 07:23, Nicolin Chen wrote:
The set_sysclk() now is used to override the output bit clock rate.
But this is not a common way to implement a set_dai_sysclk(). And
this creates a problem when a general machine driver (simple-card
for example) tries to do set_dai_sysclk() by passing an input clock
rate for the baud clock instead of setting the bit clock rate as
fsl_ssi driver expected.

So this patch solves this problem by firstly removing set_sysclk()
since the hw_params() can calculate the bit clock rate. Secondly,
in order not to break those TDM use cases which previously might
have been using set_sysclk() to override the bit clock rate, this
patch changes the driver to override it based on the slot number.

The patch also removes an obsolete comment of the dir parameter.

Signed-off-by: Nicolin Chen <nicoleotsuka@xxxxxxxxx>
sound/soc/fsl/fsl_ssi.c | 26 ++++++++------------------
1 file changed, 8 insertions(+), 18 deletions(-)

diff --git a/sound/soc/fsl/fsl_ssi.c b/sound/soc/fsl/fsl_ssi.c
index 64598d1..3657c88 100644
--- a/sound/soc/fsl/fsl_ssi.c
+++ b/sound/soc/fsl/fsl_ssi.c
@@ -197,12 +197,12 @@ struct fsl_ssi_soc_data {
* @use_dma: DMA is used or FIQ with stream filter
* @use_dual_fifo: DMA with support for both FIFOs used
* @fifo_deph: Depth of the SSI FIFOs
+ * @slots: number of slots
* @rxtx_reg_val: Specific register settings for receive/transmit configuration
* @clk: SSI clock
* @baudclk: SSI baud clock for master mode
* @baudclk_streams: Active streams that are using baudclk
- * @bitclk_freq: bitclock frequency set by .set_dai_sysclk
* @dma_params_tx: DMA transmit parameters
* @dma_params_rx: DMA receive parameters
@@ -233,12 +233,12 @@ struct fsl_ssi_private {
bool use_dual_fifo;
bool has_ipg_clk_name;
unsigned int fifo_depth;
+ unsigned int slots;
struct fsl_ssi_rxtx_reg_val rxtx_reg_val;
struct clk *clk;
struct clk *baudclk;
unsigned int baudclk_streams;
- unsigned int bitclk_freq;
/* regcache for volatile regs */
u32 regcache_sfcsr;
@@ -700,8 +700,7 @@ static void fsl_ssi_shutdown(struct snd_pcm_substream *substream,
* Note: This function can be only called when using SSI as DAI master
* Quick instruction for parameters:
- * freq: Output BCLK frequency = samplerate * 32 (fixed) * channels
+ * freq: Output BCLK frequency = samplerate * 32 (fixed) * slots (or channels)

Slots are not necessarily 32 bits width.
Indeed, a simple grep of snd_soc_dai_set_tdm_slot shows 16, 24, 32 and 0 bits usage. (don't know the real meaning of 0 BTW...)
So, it should be good to also remember the slots width given in snd_soc_dai_set_tdm_slot() call.

Anyway, there is something wrong in the snd_soc_codec_set_sysclk usage by fsl_ssi
Let's get a look again at the description:

* snd_soc_dai_set_sysclk - configure DAI system or master clock.
* @dai: DAI
* @clk_id: DAI specific clock ID
* @freq: new clock frequency in Hz
* @dir: new clock direction - input/output.
* Configures the DAI master (MCLK) or system (SYSCLK) clocking.
int snd_soc_dai_set_sysclk(struct snd_soc_dai *dai, int clk_id,
unsigned int freq, int dir)

So, it can be used to configure 2 different clocks (and more) with different usages.

Lukasz complains that simple-card is configuring MCLK incorrectly. but simple-card, only want to configure the SYSCLK, whereas fsl_ssi understand "configure the MCLK..." (fsl_ssi doesn't check the clk_id at all)

I think the problem is here.
I would propose a new clk_id


And leave fsl_ssi_set_dai_sysclk(xxx, FSL_SSI_SYSCLK_MCLK, ...) override the computed mlck.
This will leave a way for obscure TDM users to specify a specific a random mclk frequency for obscure reasons...
Unfortunately, such generic clock_id (sysclk, mclk) were never defined widely.

Is it really needed ? It is true that, up to now, we are using fsl_ssi_set_dai_sysclk() in addition to snd_soc_dai_set_tdm_slot() only because this was the only way to deal with correct mclk setting. And if now, snd_soc_dai_set_tdm_slot() do its job correctly, fsl_ssi_set_dai_sysclk() will become useless (except for obscure TDM users of course)

Will it break TDM users ?
I will say yes, surely, but TDM users (at least myself) may consider something will break in TDM audio area every-time we jump to a new linux release...


static int fsl_ssi_set_bclk(struct snd_pcm_substream *substream,
struct snd_soc_dai *cpu_dai,
@@ -716,9 +715,9 @@ static int fsl_ssi_set_bclk(struct snd_pcm_substream *substream,
unsigned int freq;
bool baudclk_is_used;
- /* Prefer the explicitly set bitclock frequency */
- if (ssi_private->bitclk_freq)
- freq = ssi_private->bitclk_freq;
+ /* Generate bit clock based on the slot or channel number */
+ if (ssi_private->slots)
+ freq = ssi_private->slots * 32 * params_rate(hw_params);
freq = params_channels(hw_params) * 32 * params_rate(hw_params);
@@ -805,16 +804,6 @@ static int fsl_ssi_set_bclk(struct snd_pcm_substream *substream,
return 0;
-static int fsl_ssi_set_dai_sysclk(struct snd_soc_dai *cpu_dai,
- int clk_id, unsigned int freq, int dir)
- struct fsl_ssi_private *ssi_private = snd_soc_dai_get_drvdata(cpu_dai);
- ssi_private->bitclk_freq = freq;
- return 0;
* fsl_ssi_hw_params - program the sample size
@@ -1121,6 +1110,8 @@ static int fsl_ssi_set_dai_tdm_slot(struct snd_soc_dai *cpu_dai, u32 tx_mask,
regmap_update_bits(regs, CCSR_SSI_SCR, CCSR_SSI_SCR_SSIEN, val);
+ ssi_private->slots = slots;
return 0;
@@ -1191,7 +1182,6 @@ static const struct snd_soc_dai_ops fsl_ssi_dai_ops = {
.hw_params = fsl_ssi_hw_params,
.hw_free = fsl_ssi_hw_free,
.set_fmt = fsl_ssi_set_dai_fmt,
- .set_sysclk = fsl_ssi_set_dai_sysclk,
.set_tdm_slot = fsl_ssi_set_dai_tdm_slot,
.trigger = fsl_ssi_trigger,