[PATCH v4 1/2] ASoC: dt-bindings: sound-card-common: List sound widgets ignoring system suspend
Add a property to list audio sound widgets which are marked ignoring system suspend. Paths between these endpoints are still active over suspend of the main application processor that the current operating system is running.
Signed-off-by: Chancel Liu chancel.liu@nxp.com --- .../devicetree/bindings/sound/sound-card-common.yaml | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/Documentation/devicetree/bindings/sound/sound-card-common.yaml b/Documentation/devicetree/bindings/sound/sound-card-common.yaml index 3a941177f684..721950f65748 100644 --- a/Documentation/devicetree/bindings/sound/sound-card-common.yaml +++ b/Documentation/devicetree/bindings/sound/sound-card-common.yaml @@ -17,6 +17,13 @@ properties: pair of strings, the first being the connection's sink, the second being the connection's source.
+ ignore-suspend-widgets: + $ref: /schemas/types.yaml#/definitions/non-unique-string-array + description: | + A list of audio sound widgets which are marked ignoring system suspend. + Paths between these endpoints are still active over suspend of the main + application processor that the current operating system is running. + model: $ref: /schemas/types.yaml#/definitions/string description: User specified audio sound card name
Low power audio mode requires binding codec still power on while Acore enters into suspend so Mcore can continue playback music.
ASoC machine driver acquires DAPM endpoints through reading "ignore-suspend-widgets" property from DT and then forces the path between these endpoints ignoring suspend.
If the rpmsg sound card is in low power audio mode, the suspend/resume callback of binding codec is overridden to disable the suspend/resume.
Signed-off-by: Chancel Liu chancel.liu@nxp.com --- sound/soc/fsl/imx-rpmsg.c | 58 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 58 insertions(+)
diff --git a/sound/soc/fsl/imx-rpmsg.c b/sound/soc/fsl/imx-rpmsg.c index a9324712e3fa..e5bd63dab10c 100644 --- a/sound/soc/fsl/imx-rpmsg.c +++ b/sound/soc/fsl/imx-rpmsg.c @@ -20,8 +20,11 @@ struct imx_rpmsg { struct snd_soc_dai_link dai; struct snd_soc_card card; unsigned long sysclk; + bool lpa; };
+static struct dev_pm_ops lpa_pm; + static const struct snd_soc_dapm_widget imx_rpmsg_dapm_widgets[] = { SND_SOC_DAPM_HP("Headphone Jack", NULL), SND_SOC_DAPM_SPK("Ext Spk", NULL), @@ -38,6 +41,58 @@ static int imx_rpmsg_late_probe(struct snd_soc_card *card) struct device *dev = card->dev; int ret;
+ if (data->lpa) { + struct snd_soc_component *codec_comp; + struct device_node *codec_np; + struct device_driver *codec_drv; + struct device *codec_dev = NULL; + + codec_np = data->dai.codecs->of_node; + if (codec_np) { + struct platform_device *codec_pdev; + struct i2c_client *codec_i2c; + + codec_i2c = of_find_i2c_device_by_node(codec_np); + if (codec_i2c) + codec_dev = &codec_i2c->dev; + if (!codec_dev) { + codec_pdev = of_find_device_by_node(codec_np); + if (codec_pdev) + codec_dev = &codec_pdev->dev; + } + } + if (codec_dev) { + codec_comp = snd_soc_lookup_component_nolocked(codec_dev, NULL); + if (codec_comp) { + int i, num_widgets; + const char *widgets; + struct snd_soc_dapm_context *dapm; + + num_widgets = of_property_count_strings(data->card.dev->of_node, + "ignore-suspend-widgets"); + for (i = 0; i < num_widgets; i++) { + of_property_read_string_index(data->card.dev->of_node, + "ignore-suspend-widgets", + i, &widgets); + dapm = snd_soc_component_get_dapm(codec_comp); + snd_soc_dapm_ignore_suspend(dapm, widgets); + } + } + codec_drv = codec_dev->driver; + if (codec_drv->pm) { + memcpy(&lpa_pm, codec_drv->pm, sizeof(lpa_pm)); + lpa_pm.suspend = NULL; + lpa_pm.resume = NULL; + lpa_pm.freeze = NULL; + lpa_pm.thaw = NULL; + lpa_pm.poweroff = NULL; + lpa_pm.restore = NULL; + codec_drv->pm = &lpa_pm; + } + put_device(codec_dev); + } + } + if (!data->sysclk) return 0;
@@ -137,6 +192,9 @@ static int imx_rpmsg_probe(struct platform_device *pdev) goto fail; }
+ if (of_property_read_bool(np, "fsl,enable-lpa")) + data->lpa = true; + data->card.dev = &pdev->dev; data->card.owner = THIS_MODULE; data->card.dapm_widgets = imx_rpmsg_dapm_widgets;
On Mon, Oct 23, 2023 at 10:07:18AM +0800, Chancel Liu wrote:
Low power audio mode requires binding codec still power on while Acore enters into suspend so Mcore can continue playback music.
ASoC machine driver acquires DAPM endpoints through reading "ignore-suspend-widgets" property from DT and then forces the path between these endpoints ignoring suspend.
This breaks an x86 allmodconfig build:
/build/stage/linux/sound/soc/fsl/imx-rpmsg.c: In function ‘imx_rpmsg_late_probe’ : /build/stage/linux/sound/soc/fsl/imx-rpmsg.c:60:46: error: implicit declaration of function ‘of_find_device_by_node’; did you mean ‘of_find_i2c_device_by_node’? [-Werror=implicit-function-declaration] 60 | codec_pdev = of_find_device_by_node(code c_np); | ^~~~~~~~~~~~~~~~~~~~~~ | of_find_i2c_device_by_node /build/stage/linux/sound/soc/fsl/imx-rpmsg.c:60:44: error: assignment to ‘struct platform_device *’ from ‘int’ makes pointer from integer without a cast [-Werro r=int-conversion] 60 | codec_pdev = of_find_device_by_node(codec_np); | ^ cc1: all warnings being treated as errors
On Mon, Oct 23, 2023 at 10:07:17AM +0800, Chancel Liu wrote:
Add a property to list audio sound widgets which are marked ignoring system suspend. Paths between these endpoints are still active over suspend of the main application processor that the current operating system is running.
Perhaps it would be better to define components used for low power audio rather than the OS mode that gets used. Isn't LPA just audio handling that doesn't require the OS CPU to be involved? So the state of the CPU is kind of orthogonal.
Rob
On Thu, Oct 26, 2023 at 12:58:42PM -0500, Rob Herring wrote:
On Mon, Oct 23, 2023 at 10:07:17AM +0800, Chancel Liu wrote:
Add a property to list audio sound widgets which are marked ignoring system suspend. Paths between these endpoints are still active over suspend of the main application processor that the current operating system is running.
Perhaps it would be better to define components used for low power audio rather than the OS mode that gets used. Isn't LPA just audio handling that doesn't require the OS CPU to be involved? So the state of the CPU is kind of orthogonal.
Not really - you can have bypass paths that don't touch the AP but which will have the power cut if the AP goes into suspend and it can be a system integration decision which components that will be true for. Often it's an inflexible PMIC/MCU setup managing the lowest system power states. The pure low power audio bit we should already be able to figure out.
participants (3)
-
Chancel Liu -
Mark Brown -
Rob Herring