[alsa-devel] [PATCH] ASoC: DaVinci: Update suspend/resume support for McASP driver
Add clock enable and disable calls to resume and suspend respectively. Also add a member to the audio device data structure which tracks the clock status.
Tested on DA850/OMAP-L138 EVM. For the purpose of testing, the patches[1] which add suspend-to-RAM support to DA850/OMAP-L138 SoC were applied.
[1] http://linux.davincidsp.com/pipermail/davinci-linux-open-source/ 2009-November/016958.html
Signed-off-by: Chaithrika U S chaithrika@ti.com --- Applies to ALSA GIT tree on branch topic/asoc at http://git.kernel.org/?p=linux/kernel/git/tiwai/sound-2.6.git;a=shortlog; h=topic/asoc
sound/soc/davinci/davinci-mcasp.c | 18 ++++++++++++++++-- sound/soc/davinci/davinci-mcasp.h | 1 + sound/soc/davinci/davinci-pcm.c | 2 +- 3 files changed, 18 insertions(+), 3 deletions(-)
diff --git a/sound/soc/davinci/davinci-mcasp.c b/sound/soc/davinci/davinci-mcasp.c index 0a302e1..0d263f1 100644 --- a/sound/soc/davinci/davinci-mcasp.c +++ b/sound/soc/davinci/davinci-mcasp.c @@ -767,14 +767,27 @@ static int davinci_mcasp_trigger(struct snd_pcm_substream *substream, int ret = 0;
switch (cmd) { - case SNDRV_PCM_TRIGGER_START: case SNDRV_PCM_TRIGGER_RESUME: + if (!dev->clk_active) { + clk_enable(dev->clk); + dev->clk_active = 1; + } + + case SNDRV_PCM_TRIGGER_START: case SNDRV_PCM_TRIGGER_PAUSE_RELEASE: davinci_mcasp_start(dev, substream->stream); break;
- case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: + davinci_mcasp_stop(dev, substream->stream); + if (dev->clk_active) { + clk_disable(dev->clk); + dev->clk_active = 0; + } + + break; + + case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_PAUSE_PUSH: davinci_mcasp_stop(dev, substream->stream); break; @@ -866,6 +879,7 @@ static int davinci_mcasp_probe(struct platform_device *pdev) }
clk_enable(dev->clk); + dev->clk_active = 1;
dev->base = (void __iomem *)IO_ADDRESS(mem->start); dev->op_mode = pdata->op_mode; diff --git a/sound/soc/davinci/davinci-mcasp.h b/sound/soc/davinci/davinci-mcasp.h index 582c924..e755b51 100644 --- a/sound/soc/davinci/davinci-mcasp.h +++ b/sound/soc/davinci/davinci-mcasp.h @@ -44,6 +44,7 @@ struct davinci_audio_dev { int sample_rate; struct clk *clk; unsigned int codec_fmt; + u8 clk_active;
/* McASP specific data */ int tdm_slots; diff --git a/sound/soc/davinci/davinci-pcm.c b/sound/soc/davinci/davinci-pcm.c index ad4d7f4..80c7fdf 100644 --- a/sound/soc/davinci/davinci-pcm.c +++ b/sound/soc/davinci/davinci-pcm.c @@ -49,7 +49,7 @@ static void print_buf_info(int slot, char *name) static struct snd_pcm_hardware pcm_hardware_playback = { .info = (SNDRV_PCM_INFO_INTERLEAVED | SNDRV_PCM_INFO_BLOCK_TRANSFER | SNDRV_PCM_INFO_MMAP | SNDRV_PCM_INFO_MMAP_VALID | - SNDRV_PCM_INFO_PAUSE), + SNDRV_PCM_INFO_PAUSE | SNDRV_PCM_INFO_RESUME), .formats = (SNDRV_PCM_FMTBIT_S16_LE), .rates = (SNDRV_PCM_RATE_8000 | SNDRV_PCM_RATE_16000 | SNDRV_PCM_RATE_22050 | SNDRV_PCM_RATE_32000 |
At Wed, 2 Dec 2009 12:39:00 +0530, Chaithrika U S wrote:
diff --git a/sound/soc/davinci/davinci-pcm.c b/sound/soc/davinci/davinci-pcm.c index ad4d7f4..80c7fdf 100644 --- a/sound/soc/davinci/davinci-pcm.c +++ b/sound/soc/davinci/davinci-pcm.c @@ -49,7 +49,7 @@ static void print_buf_info(int slot, char *name) static struct snd_pcm_hardware pcm_hardware_playback = { .info = (SNDRV_PCM_INFO_INTERLEAVED | SNDRV_PCM_INFO_BLOCK_TRANSFER | SNDRV_PCM_INFO_MMAP | SNDRV_PCM_INFO_MMAP_VALID |
SNDRV_PCM_INFO_PAUSE),
SNDRV_PCM_INFO_PAUSE | SNDRV_PCM_INFO_RESUME),
Note that unless your driver supports the "full" resume, SNDRV_PCM_INFO_RESUME shouldn't be set. Here, the "full" resume means that the hardware gets back to a completely sane state and the PCM streams are resumed without extra PCM prepare call at PM resume. If the PCM (or whatever) needs another re-initialization (like in many cases), don't set this flag.
Just to be sure...
Takashi
On Wed, Dec 02, 2009 at 14:56:31, Takashi Iwai wrote:
At Wed, 2 Dec 2009 12:39:00 +0530, Chaithrika U S wrote:
diff --git a/sound/soc/davinci/davinci-pcm.c b/sound/soc/davinci/davinci-pcm.c index ad4d7f4..80c7fdf 100644 --- a/sound/soc/davinci/davinci-pcm.c +++ b/sound/soc/davinci/davinci-pcm.c @@ -49,7 +49,7 @@ static void print_buf_info(int slot, char *name) static struct snd_pcm_hardware pcm_hardware_playback = { .info = (SNDRV_PCM_INFO_INTERLEAVED | SNDRV_PCM_INFO_BLOCK_TRANSFER | SNDRV_PCM_INFO_MMAP | SNDRV_PCM_INFO_MMAP_VALID |
SNDRV_PCM_INFO_PAUSE),
SNDRV_PCM_INFO_PAUSE | SNDRV_PCM_INFO_RESUME),Note that unless your driver supports the "full" resume, SNDRV_PCM_INFO_RESUME shouldn't be set. Here, the "full" resume means that the hardware gets back to a completely sane state and the PCM streams are resumed without extra PCM prepare call at PM resume. If the PCM (or whatever) needs another re-initialization (like in many cases), don't set this flag.
Just to be sure...
Takashi
PCM prepare call is not needed in this case. Testing was done with audio loopback and after a suspend/resume cycle the audio output was proper. Hope my understanding is correct. Is this test sufficient to prove that driver supports full resume?
Regards, Chaithrika
At Thu, 3 Dec 2009 17:58:08 +0530, Chaithrika U S wrote:
On Wed, Dec 02, 2009 at 14:56:31, Takashi Iwai wrote:
At Wed, 2 Dec 2009 12:39:00 +0530, Chaithrika U S wrote:
diff --git a/sound/soc/davinci/davinci-pcm.c b/sound/soc/davinci/davinci-pcm.c index ad4d7f4..80c7fdf 100644 --- a/sound/soc/davinci/davinci-pcm.c +++ b/sound/soc/davinci/davinci-pcm.c @@ -49,7 +49,7 @@ static void print_buf_info(int slot, char *name) static struct snd_pcm_hardware pcm_hardware_playback = { .info = (SNDRV_PCM_INFO_INTERLEAVED | SNDRV_PCM_INFO_BLOCK_TRANSFER | SNDRV_PCM_INFO_MMAP | SNDRV_PCM_INFO_MMAP_VALID |
SNDRV_PCM_INFO_PAUSE),
SNDRV_PCM_INFO_PAUSE | SNDRV_PCM_INFO_RESUME),Note that unless your driver supports the "full" resume, SNDRV_PCM_INFO_RESUME shouldn't be set. Here, the "full" resume means that the hardware gets back to a completely sane state and the PCM streams are resumed without extra PCM prepare call at PM resume. If the PCM (or whatever) needs another re-initialization (like in many cases), don't set this flag.
Just to be sure...
Takashi
PCM prepare call is not needed in this case. Testing was done with audio loopback and after a suspend/resume cycle the audio output was proper. Hope my understanding is correct. Is this test sufficient to prove that driver supports full resume?
Yeah, if it worked with an actual app, it's fine :) That was just a slight concern.
thanks,
Takashi
On Thu, Dec 03, 2009 at 02:13:19PM +0100, Takashi Iwai wrote:
Yeah, if it worked with an actual app, it's fine :) That was just a slight concern.
Chaithrika, could you please fix the issue with the comment for fallthrough and resubmit?
On Thu, Dec 03, 2009 at 18:57:17, Mark Brown wrote:
On Thu, Dec 03, 2009 at 02:13:19PM +0100, Takashi Iwai wrote:
Yeah, if it worked with an actual app, it's fine :) That was just a slight concern.
Chaithrika, could you please fix the issue with the comment for fallthrough and resubmit?
Mark,
I will resubmit this patch soon.
Regards, Chaithrika
Hello.
Chaithrika U S wrote:
Add clock enable and disable calls to resume and suspend respectively. Also add a member to the audio device data structure which tracks the clock status.
Tested on DA850/OMAP-L138 EVM. For the purpose of testing, the patches[1] which add suspend-to-RAM support to DA850/OMAP-L138 SoC were applied.
[1] http://linux.davincidsp.com/pipermail/davinci-linux-open-source/ 2009-November/016958.html
Signed-off-by: Chaithrika U S chaithrika@ti.com
[...]
diff --git a/sound/soc/davinci/davinci-mcasp.c b/sound/soc/davinci/davinci-mcasp.c index 0a302e1..0d263f1 100644 --- a/sound/soc/davinci/davinci-mcasp.c +++ b/sound/soc/davinci/davinci-mcasp.c @@ -767,14 +767,27 @@ static int davinci_mcasp_trigger(struct snd_pcm_substream *substream, int ret = 0;
switch (cmd) {
- case SNDRV_PCM_TRIGGER_START: case SNDRV_PCM_TRIGGER_RESUME:
if (!dev->clk_active) {clk_enable(dev->clk);dev->clk_active = 1;}
You should add a comment in the cases where *break* is ommitted deliberately (if it indeed is), like:
/* FALL THRU */
- case SNDRV_PCM_TRIGGER_START: case SNDRV_PCM_TRIGGER_PAUSE_RELEASE: davinci_mcasp_start(dev, substream->stream); break;
WBR, Sergei
participants (4)
-
Chaithrika U S -
Mark Brown -
Sergei Shtylyov -
Takashi Iwai