[PATCH v2] ASoC: codec: hdac_hdmi add device_link to card device

[PATCH v2] ASoC: codec: hdac_hdmi add device_link to card device

[libin.yang]

From: Libin Yang <libin.yang@intel.com>

In resume from S3, HDAC HDMI codec driver dapm event callback may be
operated before HDMI codec driver turns on the display audio power
domain because of the contest between display driver and hdmi codec driver.

This patch adds the device_link between soc card device (consumer) and
hdmi codec device (supplier) to make sure the sequence is always correct.

Signed-off-by: Libin Yang <libin.yang@intel.com>
---
 sound/soc/codecs/hdac_hdmi.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/sound/soc/codecs/hdac_hdmi.c b/sound/soc/codecs/hdac_hdmi.c
index 5eeb0fe..4de1fbf 100644
--- a/sound/soc/codecs/hdac_hdmi.c
+++ b/sound/soc/codecs/hdac_hdmi.c
@@ -1855,6 +1855,17 @@ static int hdmi_codec_probe(struct snd_soc_component *component)
 	hdmi->card = dapm->card->snd_card;
 
 	/*
+	 * Setup a device_link between card device and HDMI codec device.
+	 * The card device is the consumer and the HDMI codec device is
+	 * the supplier. With this setting, we can make sure that the audio
+	 * domain in display power will be always turned on before operating
+	 * on the HDMI audio codec registers.
+	 * Let's use the flag DL_FLAG_AUTOREMOVE_CONSUMER. This can make
+	 * sure the device link is freed when the machine driver is removed.
+	 */
+	device_link_add(component->card->dev, &hdev->dev, DL_FLAG_RPM_ACTIVE |
+			DL_FLAG_AUTOREMOVE_CONSUMER);
+	/*
 	 * hdac_device core already sets the state to active and calls
 	 * get_noresume. So enable runtime and set the device to suspend.
 	 */
-- 
2.7.4

Re: [PATCH v2] ASoC: codec: hdac_hdmi add device_link to card device

[Takashi Iwai]

On Sat, 13 Apr 2019 15:18:12 +0200,
libin.yang@intel.com wrote:
> 
> From: Libin Yang <libin.yang@intel.com>
> 
> In resume from S3, HDAC HDMI codec driver dapm event callback may be
> operated before HDMI codec driver turns on the display audio power
> domain because of the contest between display driver and hdmi codec driver.
> 
> This patch adds the device_link between soc card device (consumer) and
> hdmi codec device (supplier) to make sure the sequence is always correct.
> 
> Signed-off-by: Libin Yang <libin.yang@intel.com>

Reviewed-by: Takashi Iwai <tiwai@suse.de>

Is it an issue that has been present for older released kernels?
If so, it deserves for Cc-to-stable.


thanks,

Takashi

> ---
>  sound/soc/codecs/hdac_hdmi.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)
> 
> diff --git a/sound/soc/codecs/hdac_hdmi.c b/sound/soc/codecs/hdac_hdmi.c
> index 5eeb0fe..4de1fbf 100644
> --- a/sound/soc/codecs/hdac_hdmi.c
> +++ b/sound/soc/codecs/hdac_hdmi.c
> @@ -1855,6 +1855,17 @@ static int hdmi_codec_probe(struct snd_soc_component *component)
>  	hdmi->card = dapm->card->snd_card;
>  
>  	/*
> +	 * Setup a device_link between card device and HDMI codec device.
> +	 * The card device is the consumer and the HDMI codec device is
> +	 * the supplier. With this setting, we can make sure that the audio
> +	 * domain in display power will be always turned on before operating
> +	 * on the HDMI audio codec registers.
> +	 * Let's use the flag DL_FLAG_AUTOREMOVE_CONSUMER. This can make
> +	 * sure the device link is freed when the machine driver is removed.
> +	 */
> +	device_link_add(component->card->dev, &hdev->dev, DL_FLAG_RPM_ACTIVE |
> +			DL_FLAG_AUTOREMOVE_CONSUMER);
> +	/*
>  	 * hdac_device core already sets the state to active and calls
>  	 * get_noresume. So enable runtime and set the device to suspend.
>  	 */
> -- 
> 2.7.4
> 

Re: [PATCH v2] ASoC: codec: hdac_hdmi add device_link to card device

[Pierre-Louis Bossart]

On 4/16/19 5:23 AM, Takashi Iwai wrote:
> On Sat, 13 Apr 2019 15:18:12 +0200,
> libin.yang@intel.com wrote:
>>
>> From: Libin Yang <libin.yang@intel.com>
>>
>> In resume from S3, HDAC HDMI codec driver dapm event callback may be
>> operated before HDMI codec driver turns on the display audio power
>> domain because of the contest between display driver and hdmi codec driver.
>>
>> This patch adds the device_link between soc card device (consumer) and
>> hdmi codec device (supplier) to make sure the sequence is always correct.
>>
>> Signed-off-by: Libin Yang <libin.yang@intel.com>
> 
> Reviewed-by: Takashi Iwai <tiwai@suse.de>
> 
> Is it an issue that has been present for older released kernels?
> If so, it deserves for Cc-to-stable.

Yes and no. In theory the same problem impacts the Skylake driver. 
However its support for HDMI has been flaky at best, see e.g. the 
probe/timing issues on Linus' laptop and a variety of devices in 
December, and the only real distribution using it - ChromeOS - does not 
resume from S3. Not to mention that products typically use 3.18 to 4.9.

Re: [PATCH v2] ASoC: codec: hdac_hdmi add device_link to card device

[Yang, Libin]

>>>
>>> In resume from S3, HDAC HDMI codec driver dapm event callback may be
>>> operated before HDMI codec driver turns on the display audio power
>>> domain because of the contest between display driver and hdmi codec
>driver.
>>>
>>> This patch adds the device_link between soc card device (consumer)
>>> and hdmi codec device (supplier) to make sure the sequence is always
>correct.
>>>
>>> Signed-off-by: Libin Yang <libin.yang@intel.com>
>>
>> Reviewed-by: Takashi Iwai <tiwai@suse.de>
>>
>> Is it an issue that has been present for older released kernels?
>> If so, it deserves for Cc-to-stable.
>
>Yes and no. In theory the same problem impacts the Skylake driver.
>However its support for HDMI has been flaky at best, see e.g. the
>probe/timing issues on Linus' laptop and a variety of devices in December,
>and the only real distribution using it - ChromeOS - does not resume from S3.
>Not to mention that products typically use 3.18 to 4.9.

Agree with Pierre. I have tested on 4.1x before (I didn't remember the
exact version number), it works well. On some later stable versions,
for example 5.0.0, we can easily reproduce this issue.

And for other earlier version, I didn't have ever test on it.

Regards,
Libin

Re: [PATCH v2] ASoC: codec: hdac_hdmi add device_link to card device

[Yang, Libin]

Hi Takashi,

>-----Original Message-----
>From: Takashi Iwai [mailto:tiwai@suse.de]
>Sent: Tuesday, April 16, 2019 6:24 PM
>To: Yang, Libin <libin.yang@intel.com>
>Cc: alsa-devel@alsa-project.org; broonie@kernel.org; pierre-
>louis.bossart@linux.intel.com
>Subject: Re: [alsa-devel] [PATCH v2] ASoC: codec: hdac_hdmi add device_link
>to card device
>
>On Sat, 13 Apr 2019 15:18:12 +0200,
>libin.yang@intel.com wrote:
>>
>> From: Libin Yang <libin.yang@intel.com>
>>
>> In resume from S3, HDAC HDMI codec driver dapm event callback may be
>> operated before HDMI codec driver turns on the display audio power
>> domain because of the contest between display driver and hdmi codec
>driver.
>>
>> This patch adds the device_link between soc card device (consumer) and
>> hdmi codec device (supplier) to make sure the sequence is always correct.
>>
>> Signed-off-by: Libin Yang <libin.yang@intel.com>
>
>Reviewed-by: Takashi Iwai <tiwai@suse.de>

Thanks for the review.

>
>Is it an issue that has been present for older released kernels?
>If so, it deserves for Cc-to-stable.

It is for all released kernels, and the rates of reproduce are different. It's
graphics driver related.

Regards,
Libin

>
>
>thanks,
>
>Takashi
>
>> ---
>>  sound/soc/codecs/hdac_hdmi.c | 11 +++++++++++
>>  1 file changed, 11 insertions(+)
>>
>> diff --git a/sound/soc/codecs/hdac_hdmi.c
>> b/sound/soc/codecs/hdac_hdmi.c index 5eeb0fe..4de1fbf 100644
>> --- a/sound/soc/codecs/hdac_hdmi.c
>> +++ b/sound/soc/codecs/hdac_hdmi.c
>> @@ -1855,6 +1855,17 @@ static int hdmi_codec_probe(struct
>snd_soc_component *component)
>>  	hdmi->card = dapm->card->snd_card;
>>
>>  	/*
>> +	 * Setup a device_link between card device and HDMI codec device.
>> +	 * The card device is the consumer and the HDMI codec device is
>> +	 * the supplier. With this setting, we can make sure that the audio
>> +	 * domain in display power will be always turned on before operating
>> +	 * on the HDMI audio codec registers.
>> +	 * Let's use the flag DL_FLAG_AUTOREMOVE_CONSUMER. This can
>make
>> +	 * sure the device link is freed when the machine driver is removed.
>> +	 */
>> +	device_link_add(component->card->dev, &hdev->dev,
>DL_FLAG_RPM_ACTIVE |
>> +			DL_FLAG_AUTOREMOVE_CONSUMER);
>> +	/*
>>  	 * hdac_device core already sets the state to active and calls
>>  	 * get_noresume. So enable runtime and set the device to suspend.
>>  	 */
>> --
>> 2.7.4
>>

Re: [PATCH v2] ASoC: codec: hdac_hdmi add device_link to card device

[Pierre-Louis Bossart]

On 4/13/19 8:18 AM, libin.yang@intel.com wrote:
> From: Libin Yang <libin.yang@intel.com>
> 
> In resume from S3, HDAC HDMI codec driver dapm event callback may be
> operated before HDMI codec driver turns on the display audio power
> domain because of the contest between display driver and hdmi codec driver.
> 
> This patch adds the device_link between soc card device (consumer) and
> hdmi codec device (supplier) to make sure the sequence is always correct.
> 
> Signed-off-by: Libin Yang <libin.yang@intel.com>

Acked-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

Libin, please follow-up on the question from Takashi on CC:ing to 
stable. I would guess this probably would be needed for 4.14+, older 
stuff is unlikely to work anyways due to other issues.

> ---
>   sound/soc/codecs/hdac_hdmi.c | 11 +++++++++++
>   1 file changed, 11 insertions(+)
> 
> diff --git a/sound/soc/codecs/hdac_hdmi.c b/sound/soc/codecs/hdac_hdmi.c
> index 5eeb0fe..4de1fbf 100644
> --- a/sound/soc/codecs/hdac_hdmi.c
> +++ b/sound/soc/codecs/hdac_hdmi.c
> @@ -1855,6 +1855,17 @@ static int hdmi_codec_probe(struct snd_soc_component *component)
>   	hdmi->card = dapm->card->snd_card;
>   
>   	/*
> +	 * Setup a device_link between card device and HDMI codec device.
> +	 * The card device is the consumer and the HDMI codec device is
> +	 * the supplier. With this setting, we can make sure that the audio
> +	 * domain in display power will be always turned on before operating
> +	 * on the HDMI audio codec registers.
> +	 * Let's use the flag DL_FLAG_AUTOREMOVE_CONSUMER. This can make
> +	 * sure the device link is freed when the machine driver is removed.
> +	 */
> +	device_link_add(component->card->dev, &hdev->dev, DL_FLAG_RPM_ACTIVE |
> +			DL_FLAG_AUTOREMOVE_CONSUMER);
> +	/*
>   	 * hdac_device core already sets the state to active and calls
>   	 * get_noresume. So enable runtime and set the device to suspend.
>   	 */
>