[alsa-devel] [PATCH v2 0/6] Runtime PM support (hda/tegra)
Background ========== The device power management operations can be controlled with the help of runtime power management (runtime PM) framework. In this case hda driver can register runtime PM callbacks with the power management core (PM core). Idea is to put the device in low power mode, when it is not getting used. The clocks and power domains can be turned off when no use case is running on the device. Current patch series adds necessary support to manage hda device at runtime.
Change log ========== v1: ------ Patch-1: "ALSA: hda/tegra: runtime power management support" * runtime PM enable/disable added in device probe/remove * runtime PM calls are forbidden if AZX_DCAPS_PM_RUNTIME is not enabled in driver flags * worker thread uses pm_runtime_get_sync to invoke ->runtime_resume and pm_runtime_put to invoke ->runtime_suspend. The callbacks are added in subsequent patches
Patch-2: "ALSA: hda/tegra: get clock handles early in probe" * Clock handles required for hda are acquired early in the probe. This is done to make use of runtime PM framework to enable/disable clocks in the callbacks.
Patch-3: "ALSA: hda/tegra: add runtime PM callbacks" * adds runtime PM callbacks for ->runtime_suspend(), ->runtime_resume() * only skeleton for the callbacks is added, implementation is done in the subsequent patches
Patch-4: "ALSA: hda/tegra: remove redundant clock enable API" * As worked thread is calling pm_runtime_get_sync() now, no need to enable the clock explicitly. This can be managed from callback. * Idea is to have clock enable and disable functionalities in runtime PM callbacks and the driver can make use of the above. Any device access should be preceded by ->runtime_resume() call.
Patch-5: "ALSA: hda/tegra: implement runtime suspend/resume" * clock enable and controller initialization is moved to ->runtime_reesume() and the reverse is done during ->runtime_suspend() * Also during system wide power transitions, pm_runtime_force_resume() is invoked in system resume() and similarly for suspend case.
Patch-6: "ALSA: hda/tegra: fix kernel panic" * kernel panic is happening befoe controller init is done * added a flag "probed" to indicate the completion of device init * this flag is checked before doing any device access.
Patch-7: "ALSA: hda/tegra: add driver flag for runtime PM" * driver flag AZX_DCAPS_PM_RUNTIME is added if CONFIG_PM is enabled. * otherwise runtime PM calls will be forbidden and act as dummy calls
v1-->v2 ------- Patch-1: "ALSA: hda/tegra: runtime power management support" * no change Patch-2: "ALSA: hda/tegra: get clock handles early in probe" * no change Patch-3: "ALSA: hda/tegra: add runtime PM callbacks" * no change Patch-4: "ALSA: hda/tegra: remove redundant clock enable API" * no change Patch-5: "ALSA: hda/tegra: implement runtime suspend/resume" * Patch-6(v1) commit is melded into this commit * instead of new flag "probed", chip->running status is used to know the completion of probe Patch-6: "ALSA: hda/tegra: add driver flag for runtime PM" * This commit is similar to Patch-7(v1) * AZX_DCAPS_PM_RUNTIME is unconditionally added and CONFIG_PM under which above flag was added is removed
=========
Sameer Pujar (6): ALSA: hda/tegra: runtime power management support ALSA: hda/tegra: get clock handles early in probe ALSA: hda/tegra: add runtime PM callbacks ALSA: hda/tegra: remove redundant clock enable API ALSA: hda/tegra: implement runtime suspend/resume ALSA: hda/tegra: add driver flag for runtime PM
sound/pci/hda/hda_tegra.c | 120 +++++++++++++++++++++++++++++++++------------- 1 file changed, 88 insertions(+), 32 deletions(-)
This patch enables runtime power management(runtime PM) support for hda. pm_runtime_enable() and pm_runtime_disable() are added during device probe and remove respectively. The runtime PM callbacks will be forbidden if hda controller does not have support for runtime PM. pm_runtime_get_sync() and pm_runtime_put() are added for hda register access. The callbacks for above will be added in subsequent patches.
Signed-off-by: Sameer Pujar spujar@nvidia.com Reviewed-by: Ravindra Lokhande rlokhande@nvidia.com Reviewed-by: Mohan Kumar D mkumard@nvidia.com --- sound/pci/hda/hda_tegra.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-)
diff --git a/sound/pci/hda/hda_tegra.c b/sound/pci/hda/hda_tegra.c index 97a176d..2f9dd23 100644 --- a/sound/pci/hda/hda_tegra.c +++ b/sound/pci/hda/hda_tegra.c @@ -32,6 +32,7 @@ #include <linux/slab.h> #include <linux/time.h> #include <linux/string.h> +#include <linux/pm_runtime.h>
#include <sound/core.h> #include <sound/initval.h> @@ -512,6 +513,11 @@ static int hda_tegra_probe(struct platform_device *pdev) card->private_data = chip;
dev_set_drvdata(&pdev->dev, card); + + pm_runtime_enable(hda->dev); + if (!azx_has_pm_runtime(chip)) + pm_runtime_forbid(hda->dev); + schedule_work(&hda->probe_work);
return 0; @@ -528,6 +534,7 @@ static void hda_tegra_probe_work(struct work_struct *work) struct platform_device *pdev = to_platform_device(hda->dev); int err;
+ pm_runtime_get_sync(hda->dev); err = hda_tegra_first_init(chip, pdev); if (err < 0) goto out_free; @@ -549,12 +556,18 @@ static void hda_tegra_probe_work(struct work_struct *work) snd_hda_set_power_save(&chip->bus, power_save * 1000);
out_free: + pm_runtime_put(hda->dev); return; /* no error return from async probe */ }
static int hda_tegra_remove(struct platform_device *pdev) { - return snd_card_free(dev_get_drvdata(&pdev->dev)); + int ret; + + ret = snd_card_free(dev_get_drvdata(&pdev->dev)); + pm_runtime_disable(&pdev->dev); + + return ret; }
static void hda_tegra_shutdown(struct platform_device *pdev)
Moved devm_clk_get() API calls to a separate function and the same can be called early in the probe. This is done before runtime PM for the device is enabled. The runtime resume/suspend callbacks can later enable/disable clocks respectively(the support would be added in subsequent patches). Clock handles should be available by the time runtime suspend/resume calls can happen.
Signed-off-by: Sameer Pujar spujar@nvidia.com Reviewed-by: Ravindra Lokhande rlokhande@nvidia.com Reviewed-by: Mohan Kumar D mkumard@nvidia.com --- sound/pci/hda/hda_tegra.c | 43 +++++++++++++++++++++++++++---------------- 1 file changed, 27 insertions(+), 16 deletions(-)
diff --git a/sound/pci/hda/hda_tegra.c b/sound/pci/hda/hda_tegra.c index 2f9dd23..28e1656 100644 --- a/sound/pci/hda/hda_tegra.c +++ b/sound/pci/hda/hda_tegra.c @@ -306,22 +306,6 @@ static int hda_tegra_init_chip(struct azx *chip, struct platform_device *pdev) struct resource *res; int err;
- hda->hda_clk = devm_clk_get(dev, "hda"); - if (IS_ERR(hda->hda_clk)) { - dev_err(dev, "failed to get hda clock\n"); - return PTR_ERR(hda->hda_clk); - } - hda->hda2codec_2x_clk = devm_clk_get(dev, "hda2codec_2x"); - if (IS_ERR(hda->hda2codec_2x_clk)) { - dev_err(dev, "failed to get hda2codec_2x clock\n"); - return PTR_ERR(hda->hda2codec_2x_clk); - } - hda->hda2hdmi_clk = devm_clk_get(dev, "hda2hdmi"); - if (IS_ERR(hda->hda2hdmi_clk)) { - dev_err(dev, "failed to get hda2hdmi clock\n"); - return PTR_ERR(hda->hda2hdmi_clk); - } - res = platform_get_resource(pdev, IORESOURCE_MEM, 0); hda->regs = devm_ioremap_resource(dev, res); if (IS_ERR(hda->regs)) @@ -341,6 +325,29 @@ static int hda_tegra_init_chip(struct azx *chip, struct platform_device *pdev) return 0; }
+static int hda_tegra_init_clk(struct hda_tegra *hda) +{ + struct device *dev = hda->dev; + + hda->hda_clk = devm_clk_get(dev, "hda"); + if (IS_ERR(hda->hda_clk)) { + dev_err(dev, "failed to get hda clock\n"); + return PTR_ERR(hda->hda_clk); + } + hda->hda2codec_2x_clk = devm_clk_get(dev, "hda2codec_2x"); + if (IS_ERR(hda->hda2codec_2x_clk)) { + dev_err(dev, "failed to get hda2codec_2x clock\n"); + return PTR_ERR(hda->hda2codec_2x_clk); + } + hda->hda2hdmi_clk = devm_clk_get(dev, "hda2hdmi"); + if (IS_ERR(hda->hda2hdmi_clk)) { + dev_err(dev, "failed to get hda2hdmi clock\n"); + return PTR_ERR(hda->hda2hdmi_clk); + } + + return 0; +} + static int hda_tegra_first_init(struct azx *chip, struct platform_device *pdev) { struct hdac_bus *bus = azx_bus(chip); @@ -507,6 +514,10 @@ static int hda_tegra_probe(struct platform_device *pdev) return err; }
+ err = hda_tegra_init_clk(hda); + if (err < 0) + goto out_free; + err = hda_tegra_create(card, driver_flags, hda); if (err < 0) goto out_free;
This patch adds skeleton of runtime suspend and resume callbacks.
Signed-off-by: Sameer Pujar spujar@nvidia.com Reviewed-by: Ravindra Lokhande rlokhande@nvidia.com Reviewed-by: Mohan Kumar D mkumard@nvidia.com --- sound/pci/hda/hda_tegra.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+)
diff --git a/sound/pci/hda/hda_tegra.c b/sound/pci/hda/hda_tegra.c index 28e1656..1189f97 100644 --- a/sound/pci/hda/hda_tegra.c +++ b/sound/pci/hda/hda_tegra.c @@ -265,8 +265,23 @@ static int hda_tegra_resume(struct device *dev) } #endif /* CONFIG_PM_SLEEP */
+#ifdef CONFIG_PM +static int hda_tegra_runtime_suspend(struct device *dev) +{ + return 0; +} + +static int hda_tegra_runtime_resume(struct device *dev) +{ + return 0; +} +#endif /* CONFIG_PM */ + static const struct dev_pm_ops hda_tegra_pm = { SET_SYSTEM_SLEEP_PM_OPS(hda_tegra_suspend, hda_tegra_resume) + SET_RUNTIME_PM_OPS(hda_tegra_runtime_suspend, + hda_tegra_runtime_resume, + NULL) };
static int hda_tegra_dev_disconnect(struct snd_device *device)
Explicit clock enable is not required during probe, as this would be managed by runtime PM calls. Clock can be enabled/disabled in runtime resume/suspend. This way it is easier to balance clock enable/disable counts.
Signed-off-by: Sameer Pujar spujar@nvidia.com Reviewed-by: Ravindra Lokhande rlokhande@nvidia.com Reviewed-by: Mohan Kumar D mkumard@nvidia.com --- sound/pci/hda/hda_tegra.c | 7 ------- 1 file changed, 7 deletions(-)
diff --git a/sound/pci/hda/hda_tegra.c b/sound/pci/hda/hda_tegra.c index 1189f97..f068b1e 100644 --- a/sound/pci/hda/hda_tegra.c +++ b/sound/pci/hda/hda_tegra.c @@ -319,7 +319,6 @@ static int hda_tegra_init_chip(struct azx *chip, struct platform_device *pdev) struct hdac_bus *bus = azx_bus(chip); struct device *dev = hda->dev; struct resource *res; - int err;
res = platform_get_resource(pdev, IORESOURCE_MEM, 0); hda->regs = devm_ioremap_resource(dev, res); @@ -329,12 +328,6 @@ static int hda_tegra_init_chip(struct azx *chip, struct platform_device *pdev) bus->remap_addr = hda->regs + HDA_BAR0; bus->addr = res->start + HDA_BAR0;
- err = hda_tegra_enable_clocks(hda); - if (err) { - dev_err(dev, "failed to get enable clocks\n"); - return err; - } - hda_tegra_init(hda);
return 0;
This patch moves clock enable/disable from system resume/suspend to runtime resume/suspend respectively. Along with this hda controller chip init or stop is also moved. System resume/suspend can invoke runtime callbacks and do necessary setup.
chip->running can be used to check for probe completion and device access during runtime_resume or runtime_suspend can be avoided if probe is not yet finished. This helps to avoid kernel panic during boot where runtime PM callbacks can happen from system PM.
Signed-off-by: Sameer Pujar spujar@nvidia.com Reviewed-by: Ravindra Lokhande rlokhande@nvidia.com Reviewed-by: Mohan Kumar D mkumard@nvidia.com --- sound/pci/hda/hda_tegra.c | 49 +++++++++++++++++++++++++++++++---------------- 1 file changed, 33 insertions(+), 16 deletions(-)
diff --git a/sound/pci/hda/hda_tegra.c b/sound/pci/hda/hda_tegra.c index f068b1e..a7fd4c6 100644 --- a/sound/pci/hda/hda_tegra.c +++ b/sound/pci/hda/hda_tegra.c @@ -233,32 +233,24 @@ static void hda_tegra_disable_clocks(struct hda_tegra *data) static int hda_tegra_suspend(struct device *dev) { struct snd_card *card = dev_get_drvdata(dev); - struct azx *chip = card->private_data; - struct hda_tegra *hda = container_of(chip, struct hda_tegra, chip); - struct hdac_bus *bus = azx_bus(chip); + int rc;
+ rc = pm_runtime_force_suspend(dev); + if (rc < 0) + return rc; snd_power_change_state(card, SNDRV_CTL_POWER_D3hot);
- azx_stop_chip(chip); - synchronize_irq(bus->irq); - azx_enter_link_reset(chip); - hda_tegra_disable_clocks(hda); - return 0; }
static int hda_tegra_resume(struct device *dev) { struct snd_card *card = dev_get_drvdata(dev); - struct azx *chip = card->private_data; - struct hda_tegra *hda = container_of(chip, struct hda_tegra, chip); - - hda_tegra_enable_clocks(hda); - - hda_tegra_init(hda); - - azx_init_chip(chip, 1); + int rc;
+ rc = pm_runtime_force_resume(dev); + if (rc < 0) + return rc; snd_power_change_state(card, SNDRV_CTL_POWER_D0);
return 0; @@ -268,11 +260,36 @@ static int hda_tegra_resume(struct device *dev) #ifdef CONFIG_PM static int hda_tegra_runtime_suspend(struct device *dev) { + struct snd_card *card = dev_get_drvdata(dev); + struct azx *chip = card->private_data; + struct hda_tegra *hda = container_of(chip, struct hda_tegra, chip); + struct hdac_bus *bus = azx_bus(chip); + + if (chip && chip->running) { + azx_stop_chip(chip); + synchronize_irq(bus->irq); + azx_enter_link_reset(chip); + } + hda_tegra_disable_clocks(hda); + return 0; }
static int hda_tegra_runtime_resume(struct device *dev) { + struct snd_card *card = dev_get_drvdata(dev); + struct azx *chip = card->private_data; + struct hda_tegra *hda = container_of(chip, struct hda_tegra, chip); + int rc; + + rc = hda_tegra_enable_clocks(hda); + if (rc != 0) + return rc; + if (chip && chip->running) { + hda_tegra_init(hda); + azx_init_chip(chip, 1); + } + return 0; } #endif /* CONFIG_PM */
AZX_DCAPS_PM_RUNTIME flag is added to indicate support for runtime PM. azx_has_pm_runtime() is used to check if above is enabled and thus forbid runtime PM calls if needed.
Signed-off-by: Sameer Pujar spujar@nvidia.com Reviewed-by: Ravindra Lokhande rlokhande@nvidia.com Reviewed-by: Mohan Kumar D mkumard@nvidia.com --- sound/pci/hda/hda_tegra.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/sound/pci/hda/hda_tegra.c b/sound/pci/hda/hda_tegra.c index a7fd4c6..c8d18dc 100644 --- a/sound/pci/hda/hda_tegra.c +++ b/sound/pci/hda/hda_tegra.c @@ -520,7 +520,8 @@ MODULE_DEVICE_TABLE(of, hda_tegra_match);
static int hda_tegra_probe(struct platform_device *pdev) { - const unsigned int driver_flags = AZX_DCAPS_CORBRP_SELF_CLEAR; + const unsigned int driver_flags = AZX_DCAPS_CORBRP_SELF_CLEAR | + AZX_DCAPS_PM_RUNTIME; struct snd_card *card; struct azx *chip; struct hda_tegra *hda;
On Tue, 22 Jan 2019 08:33:15 +0100, Sameer Pujar wrote:
Background
The device power management operations can be controlled with the help of runtime power management (runtime PM) framework. In this case hda driver can register runtime PM callbacks with the power management core (PM core). Idea is to put the device in low power mode, when it is not getting used. The clocks and power domains can be turned off when no use case is running on the device. Current patch series adds necessary support to manage hda device at runtime.
Change log
v1:
Patch-1: "ALSA: hda/tegra: runtime power management support" * runtime PM enable/disable added in device probe/remove * runtime PM calls are forbidden if AZX_DCAPS_PM_RUNTIME is not enabled in driver flags * worker thread uses pm_runtime_get_sync to invoke ->runtime_resume and pm_runtime_put to invoke ->runtime_suspend. The callbacks are added in subsequent patches
Patch-2: "ALSA: hda/tegra: get clock handles early in probe" * Clock handles required for hda are acquired early in the probe. This is done to make use of runtime PM framework to enable/disable clocks in the callbacks.
Patch-3: "ALSA: hda/tegra: add runtime PM callbacks" * adds runtime PM callbacks for ->runtime_suspend(), ->runtime_resume() * only skeleton for the callbacks is added, implementation is done in the subsequent patches
Patch-4: "ALSA: hda/tegra: remove redundant clock enable API" * As worked thread is calling pm_runtime_get_sync() now, no need to enable the clock explicitly. This can be managed from callback. * Idea is to have clock enable and disable functionalities in runtime PM callbacks and the driver can make use of the above. Any device access should be preceded by ->runtime_resume() call.
Patch-5: "ALSA: hda/tegra: implement runtime suspend/resume" * clock enable and controller initialization is moved to ->runtime_reesume() and the reverse is done during ->runtime_suspend() * Also during system wide power transitions, pm_runtime_force_resume() is invoked in system resume() and similarly for suspend case.
Patch-6: "ALSA: hda/tegra: fix kernel panic" * kernel panic is happening befoe controller init is done * added a flag "probed" to indicate the completion of device init * this flag is checked before doing any device access.
Patch-7: "ALSA: hda/tegra: add driver flag for runtime PM" * driver flag AZX_DCAPS_PM_RUNTIME is added if CONFIG_PM is enabled. * otherwise runtime PM calls will be forbidden and act as dummy calls
v1-->v2
Patch-1: "ALSA: hda/tegra: runtime power management support" * no change Patch-2: "ALSA: hda/tegra: get clock handles early in probe" * no change Patch-3: "ALSA: hda/tegra: add runtime PM callbacks" * no change Patch-4: "ALSA: hda/tegra: remove redundant clock enable API" * no change Patch-5: "ALSA: hda/tegra: implement runtime suspend/resume" * Patch-6(v1) commit is melded into this commit * instead of new flag "probed", chip->running status is used to know the completion of probe Patch-6: "ALSA: hda/tegra: add driver flag for runtime PM" * This commit is similar to Patch-7(v1) * AZX_DCAPS_PM_RUNTIME is unconditionally added and CONFIG_PM under which above flag was added is removed
=========
Sameer Pujar (6): ALSA: hda/tegra: runtime power management support ALSA: hda/tegra: get clock handles early in probe ALSA: hda/tegra: add runtime PM callbacks ALSA: hda/tegra: remove redundant clock enable API ALSA: hda/tegra: implement runtime suspend/resume ALSA: hda/tegra: add driver flag for runtime PM
Applied all patches now. Thanks.
Takashi
participants (2)
-
Sameer Pujar -
Takashi Iwai