| Message ID | 1534615923-8541-2-git-send-email-anshuman.gupta@intel.com (mailing list archive) |
|---|---|
| State | RFC, archived |
| Headers | show |
| Series | cover-letter hdac-codec runtime suspended at PM:Suspend. | expand |
Is this patch not in consideration, there are no review comment for it. this patch fixes an issue with hdac hdmi codec driver. On one of our platform HD audio controller takes arounf 300ms. Below are the snippet of dmesg log. [ 3778.461888] calling 0000:00:0e.0+ @ 3667, parent: pci0000:00, cb: pci_pm_resume [ 3778.775273] call 0000:00:0e.0+ returned 0 after 306016 usecs When HD audio controller is in runtime suspend state, with direct complete, we can improve overall system wide resume latency. On Sat, Aug 18, 2018 at 11:42:03PM +0530, Anshuman Gupta wrote: > Keep hdac hdmi codec to be in runtime suspended while entering to > system wide suspend. Currently hdac hdmi codec driver using its > suspend and resume operation in prepare and complete PM callbacks, > and it resumes the hd audio controller (parent of self) from runtime > suspend and blocks the direct complete for hd audio controller. > > If hdac-codec is already in runtime suspend state skip its power down > sequence in prepare, power up sequence in complete phase. It enables > direct complete path for hdac-codec and hd audio controller > PCI device. > > Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com> > --- > Changes in v2 > - Changed the commit message. > - Using pm_runtime_suspended instead of pm_runtime_status_suspended > in order to handle any race condition. > > sound/soc/codecs/hdac_hdmi.c | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/sound/soc/codecs/hdac_hdmi.c b/sound/soc/codecs/hdac_hdmi.c > index 84f7a7a..e965338 100644 > --- a/sound/soc/codecs/hdac_hdmi.c > +++ b/sound/soc/codecs/hdac_hdmi.c > @@ -1859,6 +1859,9 @@ static int hdmi_codec_prepare(struct device *dev) > struct hdac_ext_device *edev = to_hda_ext_device(dev); > struct hdac_device *hdev = &edev->hdev; > > + if (pm_runtime_suspended(dev)) > + return 1; > + > pm_runtime_get_sync(&edev->hdev.dev); > > /* > @@ -1880,6 +1883,9 @@ static void hdmi_codec_complete(struct device *dev) > struct hdac_hdmi_priv *hdmi = hdev_to_hdmi_priv(&edev->hdev); > struct hdac_device *hdev = &edev->hdev; > > + if (pm_runtime_suspended(dev)) > + return; > + > /* Power up afg */ > snd_hdac_codec_read(hdev, hdev->afg, 0, AC_VERB_SET_POWER_STATE, > AC_PWRST_D0); > -- > 2.7.4 > -- Thanks, Anshuman.
On Tue, 28 Aug 2018 18:34:15 +0200, Anshuman Gupta wrote: > > Is this patch not in consideration, there are no review comment for it. > this patch fixes an issue with hdac hdmi codec driver. > > On one of our platform HD audio controller takes arounf 300ms. > Below are the snippet of dmesg log. > > [ 3778.461888] calling 0000:00:0e.0+ @ 3667, parent: pci0000:00, cb: pci_pm_resume > [ 3778.775273] call 0000:00:0e.0+ returned 0 after 306016 usecs > > When HD audio controller is in runtime suspend state, > with direct complete, we can improve overall system wide resume latency. Actually, *this* should have been mentioned in the changelog, and the subject would be better phrased to reflect it. After all, you're trying to reduce / optimize the runtime PM latency. thanks, Takashi > On Sat, Aug 18, 2018 at 11:42:03PM +0530, Anshuman Gupta wrote: > > Keep hdac hdmi codec to be in runtime suspended while entering to > > system wide suspend. Currently hdac hdmi codec driver using its > > suspend and resume operation in prepare and complete PM callbacks, > > and it resumes the hd audio controller (parent of self) from runtime > > suspend and blocks the direct complete for hd audio controller. > > > > If hdac-codec is already in runtime suspend state skip its power down > > sequence in prepare, power up sequence in complete phase. It enables > > direct complete path for hdac-codec and hd audio controller > > PCI device. > > > > Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com> > > --- > > Changes in v2 > > - Changed the commit message. > > - Using pm_runtime_suspended instead of pm_runtime_status_suspended > > in order to handle any race condition. > > > > sound/soc/codecs/hdac_hdmi.c | 6 ++++++ > > 1 file changed, 6 insertions(+) > > > > diff --git a/sound/soc/codecs/hdac_hdmi.c b/sound/soc/codecs/hdac_hdmi.c > > index 84f7a7a..e965338 100644 > > --- a/sound/soc/codecs/hdac_hdmi.c > > +++ b/sound/soc/codecs/hdac_hdmi.c > > @@ -1859,6 +1859,9 @@ static int hdmi_codec_prepare(struct device *dev) > > struct hdac_ext_device *edev = to_hda_ext_device(dev); > > struct hdac_device *hdev = &edev->hdev; > > > > + if (pm_runtime_suspended(dev)) > > + return 1; > > + > > pm_runtime_get_sync(&edev->hdev.dev); > > > > /* > > @@ -1880,6 +1883,9 @@ static void hdmi_codec_complete(struct device *dev) > > struct hdac_hdmi_priv *hdmi = hdev_to_hdmi_priv(&edev->hdev); > > struct hdac_device *hdev = &edev->hdev; > > > > + if (pm_runtime_suspended(dev)) > > + return; > > + > > /* Power up afg */ > > snd_hdac_codec_read(hdev, hdev->afg, 0, AC_VERB_SET_POWER_STATE, > > AC_PWRST_D0); > > -- > > 2.7.4 > > > > -- > Thanks, > Anshuman. > >
diff --git a/sound/soc/codecs/hdac_hdmi.c b/sound/soc/codecs/hdac_hdmi.c index 84f7a7a..e965338 100644 --- a/sound/soc/codecs/hdac_hdmi.c +++ b/sound/soc/codecs/hdac_hdmi.c @@ -1859,6 +1859,9 @@ static int hdmi_codec_prepare(struct device *dev) struct hdac_ext_device *edev = to_hda_ext_device(dev); struct hdac_device *hdev = &edev->hdev; + if (pm_runtime_suspended(dev)) + return 1; + pm_runtime_get_sync(&edev->hdev.dev); /* @@ -1880,6 +1883,9 @@ static void hdmi_codec_complete(struct device *dev) struct hdac_hdmi_priv *hdmi = hdev_to_hdmi_priv(&edev->hdev); struct hdac_device *hdev = &edev->hdev; + if (pm_runtime_suspended(dev)) + return; + /* Power up afg */ snd_hdac_codec_read(hdev, hdev->afg, 0, AC_VERB_SET_POWER_STATE, AC_PWRST_D0);
Keep hdac hdmi codec to be in runtime suspended while entering to system wide suspend. Currently hdac hdmi codec driver using its suspend and resume operation in prepare and complete PM callbacks, and it resumes the hd audio controller (parent of self) from runtime suspend and blocks the direct complete for hd audio controller. If hdac-codec is already in runtime suspend state skip its power down sequence in prepare, power up sequence in complete phase. It enables direct complete path for hdac-codec and hd audio controller PCI device. Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com> --- Changes in v2 - Changed the commit message. - Using pm_runtime_suspended instead of pm_runtime_status_suspended in order to handle any race condition. sound/soc/codecs/hdac_hdmi.c | 6 ++++++ 1 file changed, 6 insertions(+)