[v4] ASoC: omap-mcbsp: Add PM QoS support for McBSP to prevent glitches
diff mbox

Message ID 1482551370-31493-1-git-send-email-matt@ranostay.consulting
State New
Headers show

Commit Message

Matt Ranostay Dec. 24, 2016, 3:49 a.m. UTC
We can get audio errors if hitting deeper idle states on omaps:

[alsa.c:230] error: Fatal problem with alsa output, error -5.
[audio.c:614] error: Error in writing audio (Input/output error?)!

This seems to happen with off mode idle enabled as power for the
whole SoC may get cut off between filling the McBSP fifo using DMA.
While active DMA blocks deeper idle states in hardware, McBSP
activity does not seem to do so.

Basing the QoS latency calculation on the FIFO size, threshold,
sample rate, and channels.

Based on the original patch by Tony Lindgren
Link: https://patchwork.kernel.org/patch/9305867/

Cc: Tony Lindgren <tony@atomide.com>
Cc: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Matt Ranostay <matt@ranostay.consulting>
---

Changes from v1:
* add calculations for latency per number of FIFO locations

Changes from v2:
* add missing mcbsp.h header change

Changes from v3:
* base the latency calculations on threshold, buffer size, sample
  rate, and channels

 sound/soc/omap/mcbsp.c      | 19 +++++++++++++++++++
 sound/soc/omap/mcbsp.h      |  3 +++
 sound/soc/omap/omap-mcbsp.c | 21 ++++++++++++++++++++-
 3 files changed, 42 insertions(+), 1 deletion(-)

Comments

Peter Ujfalusi Jan. 2, 2017, 2:18 p.m. UTC | #1
Matt,

On 12/24/2016 05:49 AM, Matt Ranostay wrote:
> We can get audio errors if hitting deeper idle states on omaps:
> 
> [alsa.c:230] error: Fatal problem with alsa output, error -5.
> [audio.c:614] error: Error in writing audio (Input/output error?)!
> 
> This seems to happen with off mode idle enabled as power for the
> whole SoC may get cut off between filling the McBSP fifo using DMA.
> While active DMA blocks deeper idle states in hardware, McBSP
> activity does not seem to do so.
> 
> Basing the QoS latency calculation on the FIFO size, threshold,
> sample rate, and channels.
> 
> Based on the original patch by Tony Lindgren
> Link: https://patchwork.kernel.org/patch/9305867/

I think it would be much better if...
 
> Cc: Tony Lindgren <tony@atomide.com>
> Cc: Peter Ujfalusi <peter.ujfalusi@ti.com>
> Signed-off-by: Matt Ranostay <matt@ranostay.consulting>
> ---
> 
> Changes from v1:
> * add calculations for latency per number of FIFO locations
> 
> Changes from v2:
> * add missing mcbsp.h header change
> 
> Changes from v3:
> * base the latency calculations on threshold, buffer size, sample
>   rate, and channels
> 
>  sound/soc/omap/mcbsp.c      | 19 +++++++++++++++++++
>  sound/soc/omap/mcbsp.h      |  3 +++
>  sound/soc/omap/omap-mcbsp.c | 21 ++++++++++++++++++++-
>  3 files changed, 42 insertions(+), 1 deletion(-)
> 
> diff --git a/sound/soc/omap/mcbsp.c b/sound/soc/omap/mcbsp.c
> index 06fec5699cc8..8f792e714894 100644
> --- a/sound/soc/omap/mcbsp.c
> +++ b/sound/soc/omap/mcbsp.c
> @@ -25,6 +25,7 @@
>  #include <linux/io.h>
>  #include <linux/slab.h>
>  #include <linux/pm_runtime.h>
> +#include <linux/pm_qos.h>
>  
>  #include <linux/platform_data/asoc-ti-mcbsp.h>
> @@ -640,9 +641,19 @@ void omap_mcbsp_free(struct omap_mcbsp *mcbsp)
>   */
>  void omap_mcbsp_start(struct omap_mcbsp *mcbsp, int tx, int rx)
>  {
> +	struct pm_qos_request *pm_qos_req = &mcbsp->pm_qos_req;
>  	int enable_srg = 0;
>  	u16 w;
>  
> +	/* Prevent omap hardware from hitting off between fifo fills */
> +	if (mcbsp->latency) {
> +		if (pm_qos_request_active(pm_qos_req))
> +			pm_qos_update_request(pm_qos_req, mcbsp->latency);
> +		else
> +			pm_qos_add_request(pm_qos_req, PM_QOS_CPU_DMA_LATENCY,
> +					mcbsp->latency);
> +	}

int stream1 = tx ? SNDRV_PCM_STREAM_PLAYBACK : SNDRV_PCM_STREAM_CAPTURE;
int stream2 = tx ? SNDRV_PCM_STREAM_CAPTURE : SNDRV_PCM_STREAM_PLAYBACK;
unsigned int latency;

if (!mcbsp->latency[stream2] ||
    mcbsp->latency[stream1] < mcbsp->latency[stream2])
	latency = mcbsp->latency[stream1];
else
	latency = mcbsp->latency[stream2];

if (latency) {
	if (pm_qos_request_active(pm_qos_req))
		pm_qos_update_request(pm_qos_req, latency);
	else
		pm_qos_add_request(pm_qos_req, PM_QOS_CPU_DMA_LATENCY,
				   latency);
}

> +
>  	if (mcbsp->st_data)
>  		omap_st_start(mcbsp);
>  
> @@ -731,6 +742,11 @@ void omap_mcbsp_stop(struct omap_mcbsp *mcbsp, int tx, int rx)
>  
>  	if (mcbsp->st_data)
>  		omap_st_stop(mcbsp);
> +
> +	if (mcbsp->active == 1 && pm_qos_request_active(&mcbsp->pm_qos_req)) {
> +		pm_qos_remove_request(&mcbsp->pm_qos_req);
> +		mcbsp->latency = 0;
> +	}

if (pm_qos_request_active(&mcbsp->pm_qos_req)) {
	int stream1 = tx ? SNDRV_PCM_STREAM_PLAYBACK :
			   SNDRV_PCM_STREAM_CAPTURE;
	int stream2 = tx ? SNDRV_PCM_STREAM_CAPTURE :
			   SNDRV_PCM_STREAM_PLAYBACK;

	mcbsp->latency[stream1] = 0;
	if (mcbsp->latency[stream2])
		pm_qos_update_request(pm_qos_req,
				      mcbsp->latency[stream2]);
	else
		pm_qos_remove_request(&mcbsp->pm_qos_req);
}

So we fall back to the qos needed for the remaining stream.

>  }
>  
>  int omap2_mcbsp_set_clks_src(struct omap_mcbsp *mcbsp, u8 fck_src_id)
> @@ -1098,6 +1114,9 @@ int omap_mcbsp_init(struct platform_device *pdev)
>  
>  void omap_mcbsp_cleanup(struct omap_mcbsp *mcbsp)
>  {
> +	if (pm_qos_request_active(&mcbsp->pm_qos_req))
> +		pm_qos_remove_request(&mcbsp->pm_qos_req);
> +
>  	if (mcbsp->pdata->buffer_size)
>  		sysfs_remove_group(&mcbsp->dev->kobj, &additional_attr_group);
>  
> diff --git a/sound/soc/omap/mcbsp.h b/sound/soc/omap/mcbsp.h
> index 61e93b1c185d..d312e72f7979 100644
> --- a/sound/soc/omap/mcbsp.h
> +++ b/sound/soc/omap/mcbsp.h
> @@ -323,8 +323,11 @@ struct omap_mcbsp {
>  
>  	unsigned int fmt;
>  	unsigned int in_freq;
> +	unsigned int latency;

unsigned int latency[2];

>  	int clk_div;
>  	int wlen;
> +
> +	struct pm_qos_request pm_qos_req;
>  };
>  
>  void omap_mcbsp_config(struct omap_mcbsp *mcbsp,
> diff --git a/sound/soc/omap/omap-mcbsp.c b/sound/soc/omap/omap-mcbsp.c
> index d018e966e533..ddaec37fb4c3 100644
> --- a/sound/soc/omap/omap-mcbsp.c
> +++ b/sound/soc/omap/omap-mcbsp.c
> @@ -226,6 +226,7 @@ static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
>  	int wlen, channels, wpf;
>  	int pkt_size = 0;
>  	unsigned int format, div, framesize, master;
> +	unsigned int buffer_size = mcbsp->pdata->buffer_size;
>  
>  	dma_data = snd_soc_dai_get_dma_data(cpu_dai, substream);
>  	channels = params_channels(params);
> @@ -240,7 +241,9 @@ static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
>  	default:
>  		return -EINVAL;
>  	}
> -	if (mcbsp->pdata->buffer_size) {
> +	if (buffer_size) {
> +		int latency;
> +
>  		if (mcbsp->dma_op_mode == MCBSP_DMA_MODE_THRESHOLD) {
>  			int period_words, max_thrsh;
>  			int divider = 0;
> @@ -271,6 +274,22 @@ static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
>  			/* Use packet mode for non mono streams */
>  			pkt_size = channels;
>  		}
> +
> +		latency = ((((buffer_size - pkt_size) / channels) * 1000)
> +				 / (params->rate_num / params->rate_den));
> +
> +		/*
> +		 * Use the lowest QoS latency value for active streams, but
> +		 * this has the disadvantage of using lower than needed latency
> +		 * if an stream with lower QoS ends before one with a higher
> +		 * one. Latency setting only gets reset when all streams
> +		 * complete.
> +		 */
> +		if (latency <= mcbsp->latency)
> +			mcbsp->latency = latency;
> +		else
> +			mcbsp->latency = 0;

mcbsp->latency[substream->stream] = latency;

> +
>  		omap_mcbsp_set_threshold(substream, pkt_size);
>  	}
>  
> 

What do you think?
Matt Ranostay Jan. 3, 2017, 5:10 a.m. UTC | #2
On Mon, Jan 2, 2017 at 6:18 AM, Peter Ujfalusi <peter.ujfalusi@ti.com> wrote:
> Matt,
>
> On 12/24/2016 05:49 AM, Matt Ranostay wrote:
>> We can get audio errors if hitting deeper idle states on omaps:
>>
>> [alsa.c:230] error: Fatal problem with alsa output, error -5.
>> [audio.c:614] error: Error in writing audio (Input/output error?)!
>>
>> This seems to happen with off mode idle enabled as power for the
>> whole SoC may get cut off between filling the McBSP fifo using DMA.
>> While active DMA blocks deeper idle states in hardware, McBSP
>> activity does not seem to do so.
>>
>> Basing the QoS latency calculation on the FIFO size, threshold,
>> sample rate, and channels.
>>
>> Based on the original patch by Tony Lindgren
>> Link: https://patchwork.kernel.org/patch/9305867/
>
> I think it would be much better if...
>
>> Cc: Tony Lindgren <tony@atomide.com>
>> Cc: Peter Ujfalusi <peter.ujfalusi@ti.com>
>> Signed-off-by: Matt Ranostay <matt@ranostay.consulting>
>> ---
>>
>> Changes from v1:
>> * add calculations for latency per number of FIFO locations
>>
>> Changes from v2:
>> * add missing mcbsp.h header change
>>
>> Changes from v3:
>> * base the latency calculations on threshold, buffer size, sample
>>   rate, and channels
>>
>>  sound/soc/omap/mcbsp.c      | 19 +++++++++++++++++++
>>  sound/soc/omap/mcbsp.h      |  3 +++
>>  sound/soc/omap/omap-mcbsp.c | 21 ++++++++++++++++++++-
>>  3 files changed, 42 insertions(+), 1 deletion(-)
>>
>> diff --git a/sound/soc/omap/mcbsp.c b/sound/soc/omap/mcbsp.c
>> index 06fec5699cc8..8f792e714894 100644
>> --- a/sound/soc/omap/mcbsp.c
>> +++ b/sound/soc/omap/mcbsp.c
>> @@ -25,6 +25,7 @@
>>  #include <linux/io.h>
>>  #include <linux/slab.h>
>>  #include <linux/pm_runtime.h>
>> +#include <linux/pm_qos.h>
>>
>>  #include <linux/platform_data/asoc-ti-mcbsp.h>
>> @@ -640,9 +641,19 @@ void omap_mcbsp_free(struct omap_mcbsp *mcbsp)
>>   */
>>  void omap_mcbsp_start(struct omap_mcbsp *mcbsp, int tx, int rx)
>>  {
>> +     struct pm_qos_request *pm_qos_req = &mcbsp->pm_qos_req;
>>       int enable_srg = 0;
>>       u16 w;
>>
>> +     /* Prevent omap hardware from hitting off between fifo fills */
>> +     if (mcbsp->latency) {
>> +             if (pm_qos_request_active(pm_qos_req))
>> +                     pm_qos_update_request(pm_qos_req, mcbsp->latency);
>> +             else
>> +                     pm_qos_add_request(pm_qos_req, PM_QOS_CPU_DMA_LATENCY,
>> +                                     mcbsp->latency);
>> +     }
>
> int stream1 = tx ? SNDRV_PCM_STREAM_PLAYBACK : SNDRV_PCM_STREAM_CAPTURE;
> int stream2 = tx ? SNDRV_PCM_STREAM_CAPTURE : SNDRV_PCM_STREAM_PLAYBACK;
> unsigned int latency;
>
> if (!mcbsp->latency[stream2] ||
>     mcbsp->latency[stream1] < mcbsp->latency[stream2])
>         latency = mcbsp->latency[stream1];
> else
>         latency = mcbsp->latency[stream2];
>
> if (latency) {
>         if (pm_qos_request_active(pm_qos_req))
>                 pm_qos_update_request(pm_qos_req, latency);
>         else
>                 pm_qos_add_request(pm_qos_req, PM_QOS_CPU_DMA_LATENCY,
>                                    latency);
> }
>
>> +
>>       if (mcbsp->st_data)
>>               omap_st_start(mcbsp);
>>
>> @@ -731,6 +742,11 @@ void omap_mcbsp_stop(struct omap_mcbsp *mcbsp, int tx, int rx)
>>
>>       if (mcbsp->st_data)
>>               omap_st_stop(mcbsp);
>> +
>> +     if (mcbsp->active == 1 && pm_qos_request_active(&mcbsp->pm_qos_req)) {
>> +             pm_qos_remove_request(&mcbsp->pm_qos_req);
>> +             mcbsp->latency = 0;
>> +     }

Seems reasonable solution for this.  Although I suspect if there are
multiple McBSP channel pairs it could be an issue...
But baby steps I guess to get this hashed out for a correct QoS

I'll test this to confirm it works as well as the current solution.

- Matt

>
> if (pm_qos_request_active(&mcbsp->pm_qos_req)) {
>         int stream1 = tx ? SNDRV_PCM_STREAM_PLAYBACK :
>                            SNDRV_PCM_STREAM_CAPTURE;
>         int stream2 = tx ? SNDRV_PCM_STREAM_CAPTURE :
>                            SNDRV_PCM_STREAM_PLAYBACK;
>
>         mcbsp->latency[stream1] = 0;
>         if (mcbsp->latency[stream2])
>                 pm_qos_update_request(pm_qos_req,
>                                       mcbsp->latency[stream2]);
>         else
>                 pm_qos_remove_request(&mcbsp->pm_qos_req);
> }
>
> So we fall back to the qos needed for the remaining stream.
>
>>  }
>>
>>  int omap2_mcbsp_set_clks_src(struct omap_mcbsp *mcbsp, u8 fck_src_id)
>> @@ -1098,6 +1114,9 @@ int omap_mcbsp_init(struct platform_device *pdev)
>>
>>  void omap_mcbsp_cleanup(struct omap_mcbsp *mcbsp)
>>  {
>> +     if (pm_qos_request_active(&mcbsp->pm_qos_req))
>> +             pm_qos_remove_request(&mcbsp->pm_qos_req);
>> +
>>       if (mcbsp->pdata->buffer_size)
>>               sysfs_remove_group(&mcbsp->dev->kobj, &additional_attr_group);
>>
>> diff --git a/sound/soc/omap/mcbsp.h b/sound/soc/omap/mcbsp.h
>> index 61e93b1c185d..d312e72f7979 100644
>> --- a/sound/soc/omap/mcbsp.h
>> +++ b/sound/soc/omap/mcbsp.h
>> @@ -323,8 +323,11 @@ struct omap_mcbsp {
>>
>>       unsigned int fmt;
>>       unsigned int in_freq;
>> +     unsigned int latency;
>
> unsigned int latency[2];
>
>>       int clk_div;
>>       int wlen;
>> +
>> +     struct pm_qos_request pm_qos_req;
>>  };
>>
>>  void omap_mcbsp_config(struct omap_mcbsp *mcbsp,
>> diff --git a/sound/soc/omap/omap-mcbsp.c b/sound/soc/omap/omap-mcbsp.c
>> index d018e966e533..ddaec37fb4c3 100644
>> --- a/sound/soc/omap/omap-mcbsp.c
>> +++ b/sound/soc/omap/omap-mcbsp.c
>> @@ -226,6 +226,7 @@ static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
>>       int wlen, channels, wpf;
>>       int pkt_size = 0;
>>       unsigned int format, div, framesize, master;
>> +     unsigned int buffer_size = mcbsp->pdata->buffer_size;
>>
>>       dma_data = snd_soc_dai_get_dma_data(cpu_dai, substream);
>>       channels = params_channels(params);
>> @@ -240,7 +241,9 @@ static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
>>       default:
>>               return -EINVAL;
>>       }
>> -     if (mcbsp->pdata->buffer_size) {
>> +     if (buffer_size) {
>> +             int latency;
>> +
>>               if (mcbsp->dma_op_mode == MCBSP_DMA_MODE_THRESHOLD) {
>>                       int period_words, max_thrsh;
>>                       int divider = 0;
>> @@ -271,6 +274,22 @@ static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
>>                       /* Use packet mode for non mono streams */
>>                       pkt_size = channels;
>>               }
>> +
>> +             latency = ((((buffer_size - pkt_size) / channels) * 1000)
>> +                              / (params->rate_num / params->rate_den));
>> +
>> +             /*
>> +              * Use the lowest QoS latency value for active streams, but
>> +              * this has the disadvantage of using lower than needed latency
>> +              * if an stream with lower QoS ends before one with a higher
>> +              * one. Latency setting only gets reset when all streams
>> +              * complete.
>> +              */
>> +             if (latency <= mcbsp->latency)
>> +                     mcbsp->latency = latency;
>> +             else
>> +                     mcbsp->latency = 0;
>
> mcbsp->latency[substream->stream] = latency;
>
>> +
>>               omap_mcbsp_set_threshold(substream, pkt_size);
>>       }
>>
>>
>
> What do you think?
>
> --
> P├ęter

Patch
diff mbox

diff --git a/sound/soc/omap/mcbsp.c b/sound/soc/omap/mcbsp.c
index 06fec5699cc8..8f792e714894 100644
--- a/sound/soc/omap/mcbsp.c
+++ b/sound/soc/omap/mcbsp.c
@@ -25,6 +25,7 @@ 
 #include <linux/io.h>
 #include <linux/slab.h>
 #include <linux/pm_runtime.h>
+#include <linux/pm_qos.h>
 
 #include <linux/platform_data/asoc-ti-mcbsp.h>
 
@@ -640,9 +641,19 @@  void omap_mcbsp_free(struct omap_mcbsp *mcbsp)
  */
 void omap_mcbsp_start(struct omap_mcbsp *mcbsp, int tx, int rx)
 {
+	struct pm_qos_request *pm_qos_req = &mcbsp->pm_qos_req;
 	int enable_srg = 0;
 	u16 w;
 
+	/* Prevent omap hardware from hitting off between fifo fills */
+	if (mcbsp->latency) {
+		if (pm_qos_request_active(pm_qos_req))
+			pm_qos_update_request(pm_qos_req, mcbsp->latency);
+		else
+			pm_qos_add_request(pm_qos_req, PM_QOS_CPU_DMA_LATENCY,
+					mcbsp->latency);
+	}
+
 	if (mcbsp->st_data)
 		omap_st_start(mcbsp);
 
@@ -731,6 +742,11 @@  void omap_mcbsp_stop(struct omap_mcbsp *mcbsp, int tx, int rx)
 
 	if (mcbsp->st_data)
 		omap_st_stop(mcbsp);
+
+	if (mcbsp->active == 1 && pm_qos_request_active(&mcbsp->pm_qos_req)) {
+		pm_qos_remove_request(&mcbsp->pm_qos_req);
+		mcbsp->latency = 0;
+	}
 }
 
 int omap2_mcbsp_set_clks_src(struct omap_mcbsp *mcbsp, u8 fck_src_id)
@@ -1098,6 +1114,9 @@  int omap_mcbsp_init(struct platform_device *pdev)
 
 void omap_mcbsp_cleanup(struct omap_mcbsp *mcbsp)
 {
+	if (pm_qos_request_active(&mcbsp->pm_qos_req))
+		pm_qos_remove_request(&mcbsp->pm_qos_req);
+
 	if (mcbsp->pdata->buffer_size)
 		sysfs_remove_group(&mcbsp->dev->kobj, &additional_attr_group);
 
diff --git a/sound/soc/omap/mcbsp.h b/sound/soc/omap/mcbsp.h
index 61e93b1c185d..d312e72f7979 100644
--- a/sound/soc/omap/mcbsp.h
+++ b/sound/soc/omap/mcbsp.h
@@ -323,8 +323,11 @@  struct omap_mcbsp {
 
 	unsigned int fmt;
 	unsigned int in_freq;
+	unsigned int latency;
 	int clk_div;
 	int wlen;
+
+	struct pm_qos_request pm_qos_req;
 };
 
 void omap_mcbsp_config(struct omap_mcbsp *mcbsp,
diff --git a/sound/soc/omap/omap-mcbsp.c b/sound/soc/omap/omap-mcbsp.c
index d018e966e533..ddaec37fb4c3 100644
--- a/sound/soc/omap/omap-mcbsp.c
+++ b/sound/soc/omap/omap-mcbsp.c
@@ -226,6 +226,7 @@  static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
 	int wlen, channels, wpf;
 	int pkt_size = 0;
 	unsigned int format, div, framesize, master;
+	unsigned int buffer_size = mcbsp->pdata->buffer_size;
 
 	dma_data = snd_soc_dai_get_dma_data(cpu_dai, substream);
 	channels = params_channels(params);
@@ -240,7 +241,9 @@  static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
 	default:
 		return -EINVAL;
 	}
-	if (mcbsp->pdata->buffer_size) {
+	if (buffer_size) {
+		int latency;
+
 		if (mcbsp->dma_op_mode == MCBSP_DMA_MODE_THRESHOLD) {
 			int period_words, max_thrsh;
 			int divider = 0;
@@ -271,6 +274,22 @@  static int omap_mcbsp_dai_hw_params(struct snd_pcm_substream *substream,
 			/* Use packet mode for non mono streams */
 			pkt_size = channels;
 		}
+
+		latency = ((((buffer_size - pkt_size) / channels) * 1000)
+				 / (params->rate_num / params->rate_den));
+
+		/*
+		 * Use the lowest QoS latency value for active streams, but
+		 * this has the disadvantage of using lower than needed latency
+		 * if an stream with lower QoS ends before one with a higher
+		 * one. Latency setting only gets reset when all streams
+		 * complete.
+		 */
+		if (latency <= mcbsp->latency)
+			mcbsp->latency = latency;
+		else
+			mcbsp->latency = 0;
+
 		omap_mcbsp_set_threshold(substream, pkt_size);
 	}