[PATCH v2 0/9] ASoC: stop dependent resume work after restore failures

[PATCH v2 0/9] ASoC: stop dependent resume work after restore failures

[Pengpeng Hou]

This revision addresses Mark's cache-mode concern and Charles's WM8978
bias rollback question. It also accounts for the next suspend after a
failed resume: the ASoC wrapper reports the callback error but marks the
component resumed, so leaving an IRQ disabled needs driver-local ownership
tracking to avoid disabling it a second time in the next suspend.

All nine changes retain best-effort card resume. They do not add framework
rollback or automatic retries. The RT5682 SoundWire branch and RT5677
retained VAD path keep their existing behavior. RT5640 handles both IRQ
producers of jack work and clears ownership when either IRQ is freed.
RT5677 also propagates normal-mode transport write errors and remembers
pending restoration across later changes to the requested VAD mode.

These issues were found by our static-analysis tool and manually reviewed.
The code and descriptions were prepared with gpt 5 assistance; the patches
carry the corresponding Assisted-by trailers.

Changes since v1:
- restore cache-only mode and preserve dirty state on replay errors
- avoid repeated suspend-owned IRQ disables after failed resume
- return WM8978 bias to OFF on failure, without claiming hardware rollback
  is guaranteed when the bias callback itself ignores register errors
- clear WM8978 PLL power as system suspend does
- quiesce both RT5640 jack IRQs and reject fresh setup while unrestored
- fix RT5677's normal-mode write error propagation and pending-restore state
- preserve the per-driver power, SoundWire and DSP VAD distinctions
- resend all members as one series rather than separate unthreaded posts

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

Base: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6 (sound/for-next)

1. ASoC: nau8821: restore safe state after resume errors
2. ASoC: nau8824: restore safe state after resume errors
3. ASoC: nau8825: restore safe state after resume errors
4. ASoC: rt5663: restore safe state after resume errors
5. ASoC: rt5682s: restore safe state after resume errors
6. ASoC: rt5682: restore safe state after resume errors
7. ASoC: wm8978: restore safe state after resume errors
8. ASoC: rt5640: restore safe state after resume errors
9. ASoC: rt5677: restore safe state after resume errors

[PATCH v2 1/9] ASoC: nau8821: restore safe state after resume errors

[Pengpeng Hou]

nau8821_resume() ignores regcache_sync() failure before continuing
dependent resume operations. Restore cache-only mode, leave the cache
dirty, and return the replay error before those operations.

Leave jack-detection IRQs disabled after a failed replay.

ASoC still marks the component resumed when the callback reports failure.
Track the suspend-owned IRQ disable so another suspend does not increment
the disable depth again after a failed resume. A later successful resume
releases that one disable reference. This does not add automatic retry or
change the framework's best-effort card resume policy.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: aab1ad11d69f ("ASoC: nau8821: new driver")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- keep suspend-owned IRQ disables balanced across failed resume cycles
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/nau8821.c |   16 +++++++++++++---
 sound/soc/codecs/nau8821.h |    1 +
 2 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/sound/soc/codecs/nau8821.c b/sound/soc/codecs/nau8821.c
index ecadbaee103798466755e2fbc2f3843f6ad2c05e..41c4cf947999020246e6431cf24c8527c7f6319a 100644
--- a/sound/soc/codecs/nau8821.c
+++ b/sound/soc/codecs/nau8821.c
@@ -1620,8 +1620,10 @@ static int __maybe_unused nau8821_suspend(struct snd_soc_component *component)
 {
 	struct nau8821 *nau8821 = snd_soc_component_get_drvdata(component);
 
-	if (nau8821->irq)
+	if (nau8821->irq && !nau8821->suspend_irq_disabled) {
 		disable_irq(nau8821->irq);
+		nau8821->suspend_irq_disabled = true;
+	}
 
 	if (nau8821->jdet_active)
 		cancel_delayed_work_sync(&nau8821->jdet_work);
@@ -1639,11 +1641,19 @@ static int __maybe_unused nau8821_suspend(struct snd_soc_component *component)
 static int __maybe_unused nau8821_resume(struct snd_soc_component *component)
 {
 	struct nau8821 *nau8821 = snd_soc_component_get_drvdata(component);
+	int ret;
 
 	regcache_cache_only(nau8821->regmap, false);
-	regcache_sync(nau8821->regmap);
-	if (nau8821->irq)
+	ret = regcache_sync(nau8821->regmap);
+	if (ret) {
+		regcache_cache_only(nau8821->regmap, true);
+		regcache_mark_dirty(nau8821->regmap);
+		return ret;
+	}
+	if (nau8821->suspend_irq_disabled) {
+		nau8821->suspend_irq_disabled = false;
 		enable_irq(nau8821->irq);
+	}
 
 	return 0;
 }
diff --git a/sound/soc/codecs/nau8821.h b/sound/soc/codecs/nau8821.h
index f9d7cd8cbd2116be93cf8692f1acb9022aeabaed..def3942f8785ece82a2eafd0e18dce51c7db72cc 100644
--- a/sound/soc/codecs/nau8821.h
+++ b/sound/soc/codecs/nau8821.h
@@ -563,6 +563,7 @@ struct nau8821 {
 	struct snd_soc_jack *jack;
 	struct delayed_work jdet_work;
 	bool jdet_active;
+	bool suspend_irq_disabled;
 	int irq;
 	int clk_id;
 	int micbias_voltage;

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

[PATCH v2 2/9] ASoC: nau8824: restore safe state after resume errors

[Pengpeng Hou]

nau8824_resume() ignores regcache_sync() failure before continuing
dependent resume operations. Restore cache-only mode, leave the cache
dirty, and return the replay error before those operations.

Do not acquire the jack-detection semaphore or enable its IRQ after a
failed replay.

ASoC still marks the component resumed when the callback reports failure.
Track the suspend-owned IRQ disable so another suspend does not increment
the disable depth again after a failed resume. A later successful resume
releases that one disable reference. This does not add automatic retry or
change the framework's best-effort card resume policy.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: dfeabded0496 ("ASoC: nau8824: new driver")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- keep suspend-owned IRQ disables balanced across failed resume cycles
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/nau8824.c |   17 ++++++++++++++---
 sound/soc/codecs/nau8824.h |    1 +
 2 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/sound/soc/codecs/nau8824.c b/sound/soc/codecs/nau8824.c
index 853eb9472af92df981b4668c6b0d0dd573c6c30c..522f8951b0598da3bf1def9119ec27a086427fde 100644
--- a/sound/soc/codecs/nau8824.c
+++ b/sound/soc/codecs/nau8824.c
@@ -1531,7 +1531,10 @@ static int __maybe_unused nau8824_suspend(struct snd_soc_component *component)
 	struct nau8824 *nau8824 = snd_soc_component_get_drvdata(component);
 
 	if (nau8824->irq) {
-		disable_irq(nau8824->irq);
+		if (!nau8824->suspend_irq_disabled) {
+			disable_irq(nau8824->irq);
+			nau8824->suspend_irq_disabled = true;
+		}
 		snd_soc_dapm_force_bias_level(nau8824->dapm, SND_SOC_BIAS_OFF);
 	}
 	regcache_cache_only(nau8824->regmap, true);
@@ -1546,7 +1549,12 @@ static int __maybe_unused nau8824_resume(struct snd_soc_component *component)
 	int ret;
 
 	regcache_cache_only(nau8824->regmap, false);
-	regcache_sync(nau8824->regmap);
+	ret = regcache_sync(nau8824->regmap);
+	if (ret) {
+		regcache_cache_only(nau8824->regmap, true);
+		regcache_mark_dirty(nau8824->regmap);
+		return ret;
+	}
 	if (nau8824->irq) {
 		/* Hold semaphore to postpone playback happening
 		 * until jack detection done.
@@ -1555,7 +1563,10 @@ static int __maybe_unused nau8824_resume(struct snd_soc_component *component)
 		ret = nau8824_sema_acquire(nau8824, 0);
 		if (ret)
 			nau8824->resume_lock = false;
-		enable_irq(nau8824->irq);
+		if (nau8824->suspend_irq_disabled) {
+			nau8824->suspend_irq_disabled = false;
+			enable_irq(nau8824->irq);
+		}
 	}
 
 	return 0;
diff --git a/sound/soc/codecs/nau8824.h b/sound/soc/codecs/nau8824.h
index d8e19515133c803c20dfcb167f073515daa6f39a..f44fc39836d634e8f0b98ac9292dabe8069612b2 100644
--- a/sound/soc/codecs/nau8824.h
+++ b/sound/soc/codecs/nau8824.h
@@ -437,6 +437,7 @@ struct nau8824 {
 	struct clk *mclk;
 	int fs;
 	int irq;
+	bool suspend_irq_disabled;
 	int resume_lock;
 	int micbias_voltage;
 	int vref_impedance;

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

[PATCH v2 3/9] ASoC: nau8825: restore safe state after resume errors

[Pengpeng Hou]

nau8825_resume() ignores regcache_sync() failure before continuing
dependent resume operations. Restore cache-only mode, leave the cache
dirty, and return the replay error before those operations.

Do not start crosstalk protection or enable its IRQ after a failed
replay.

ASoC still marks the component resumed when the callback reports failure.
Track the suspend-owned IRQ disable so another suspend does not increment
the disable depth again after a failed resume. A later successful resume
releases that one disable reference. This does not add automatic retry or
change the framework's best-effort card resume policy.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: 4454a8378be5 ("ASoC: nau8825: add pm function")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- keep suspend-owned IRQ disables balanced across failed resume cycles
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/nau8825.c |   17 ++++++++++++++---
 sound/soc/codecs/nau8825.h |    1 +
 2 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/sound/soc/codecs/nau8825.c b/sound/soc/codecs/nau8825.c
index 63d927f3140902e179bd491f757f4c47e8c314b1..11c548b99ea6d758ba3843090ee65b3d9f8f8245 100644
--- a/sound/soc/codecs/nau8825.c
+++ b/sound/soc/codecs/nau8825.c
@@ -2683,7 +2683,10 @@ static int __maybe_unused nau8825_suspend(struct snd_soc_component *component)
 {
 	struct nau8825 *nau8825 = snd_soc_component_get_drvdata(component);
 
-	disable_irq(nau8825->irq);
+	if (!nau8825->suspend_irq_disabled) {
+		disable_irq(nau8825->irq);
+		nau8825->suspend_irq_disabled = true;
+	}
 	snd_soc_dapm_force_bias_level(nau8825->dapm, SND_SOC_BIAS_OFF);
 	/* Power down codec power; don't suppoet button wakeup */
 	snd_soc_dapm_disable_pin(nau8825->dapm, "SAR");
@@ -2701,12 +2704,20 @@ static int __maybe_unused nau8825_resume(struct snd_soc_component *component)
 	int ret;
 
 	regcache_cache_only(nau8825->regmap, false);
-	regcache_sync(nau8825->regmap);
+	ret = regcache_sync(nau8825->regmap);
+	if (ret) {
+		regcache_cache_only(nau8825->regmap, true);
+		regcache_mark_dirty(nau8825->regmap);
+		return ret;
+	}
 	nau8825->xtalk_protect = true;
 	ret = nau8825_sema_acquire(nau8825, 0);
 	if (ret)
 		nau8825->xtalk_protect = false;
-	enable_irq(nau8825->irq);
+	if (nau8825->suspend_irq_disabled) {
+		nau8825->suspend_irq_disabled = false;
+		enable_irq(nau8825->irq);
+	}
 
 	return 0;
 }
diff --git a/sound/soc/codecs/nau8825.h b/sound/soc/codecs/nau8825.h
index 2abfbb5184da80f3637f06d57a7564fbf1c7574c..8238570ed48c2b147a4a1a0d7061603f8b9b4835 100644
--- a/sound/soc/codecs/nau8825.h
+++ b/sound/soc/codecs/nau8825.h
@@ -502,6 +502,7 @@ struct nau8825 {
 	struct semaphore xtalk_sem;
 	int sw_id;
 	int irq;
+	bool suspend_irq_disabled;
 	int mclk_freq; /* 0 - mclk is disabled */
 	int button_pressed;
 	int micbias_voltage;

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

Re: [PATCH v2 3/9] ASoC: nau8825: restore safe state after resume errors

[Ajay Kumar Nandam]

On 10/14/2026 8:54 AM, Pengpeng Hou wrote:
> nau8825_resume() ignores regcache_sync() failure before continuing
> dependent resume operations. Restore cache-only mode, leave the cache
> dirty, and return the replay error before those operations.
> 
> Do not start crosstalk protection or enable its IRQ after a failed
> replay.
> 
> ASoC still marks the component resumed when the callback reports failure.
> Track the suspend-owned IRQ disable so another suspend does not increment
> the disable depth again after a failed resume. A later successful resume
> releases that one disable reference. This does not add automatic retry or
> change the framework's best-effort card resume policy.
> 
> The issue was found by our static-analysis tool and manually reviewed.
> 
> Fixes: 4454a8378be5 ("ASoC: nau8825: add pm function")
> Assisted-by: gpt 5
> Signed-off-by: Pengpeng Hou <hppiscas@163.com>
> ---
> Changes since v1:
> - restore cache-only mode and dirty state on replay failure
> - keep suspend-owned IRQ disables balanced across failed resume cycles
> - send as one correctly threaded series, as Mark requested
> 
> Previous series:
> https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/
> 
>   sound/soc/codecs/nau8825.c |   17 ++++++++++++++---
>   sound/soc/codecs/nau8825.h |    1 +
>   2 files changed, 15 insertions(+), 3 deletions(-)
> 
> diff --git a/sound/soc/codecs/nau8825.c b/sound/soc/codecs/nau8825.c
> index 63d927f3140902e179bd491f757f4c47e8c314b1..11c548b99ea6d758ba3843090ee65b3d9f8f8245 100644
> --- a/sound/soc/codecs/nau8825.c
> +++ b/sound/soc/codecs/nau8825.c
> @@ -2683,7 +2683,10 @@ static int __maybe_unused nau8825_suspend(struct snd_soc_component *component)
>   {
>   	struct nau8825 *nau8825 = snd_soc_component_get_drvdata(component);
>   
> -	disable_irq(nau8825->irq);
> +	if (!nau8825->suspend_irq_disabled) {
> +		disable_irq(nau8825->irq);
> +		nau8825->suspend_irq_disabled = true;
> +	}

With this new 'suspend_irq_disabled' state, should the IRQ 
suspend/resume tracking also be tied to whether 'nau8825_setup_irq()' 
succeeded?

Thanks
Ajay Kumar Nandam

>   	snd_soc_dapm_force_bias_level(nau8825->dapm, SND_SOC_BIAS_OFF);
>   	/* Power down codec power; don't suppoet button wakeup */
>   	snd_soc_dapm_disable_pin(nau8825->dapm, "SAR");
> @@ -2701,12 +2704,20 @@ static int __maybe_unused nau8825_resume(struct snd_soc_component *component)
>   	int ret;
>   
>   	regcache_cache_only(nau8825->regmap, false);
> -	regcache_sync(nau8825->regmap);
> +	ret = regcache_sync(nau8825->regmap);
> +	if (ret) {
> +		regcache_cache_only(nau8825->regmap, true);
> +		regcache_mark_dirty(nau8825->regmap);
> +		return ret;
> +	}
>   	nau8825->xtalk_protect = true;
>   	ret = nau8825_sema_acquire(nau8825, 0);
>   	if (ret)
>   		nau8825->xtalk_protect = false;
> -	enable_irq(nau8825->irq);
> +	if (nau8825->suspend_irq_disabled) {
> +		nau8825->suspend_irq_disabled = false;
> +		enable_irq(nau8825->irq);
> +	}
>   
>   	return 0;
>   }
> diff --git a/sound/soc/codecs/nau8825.h b/sound/soc/codecs/nau8825.h
> index 2abfbb5184da80f3637f06d57a7564fbf1c7574c..8238570ed48c2b147a4a1a0d7061603f8b9b4835 100644
> --- a/sound/soc/codecs/nau8825.h
> +++ b/sound/soc/codecs/nau8825.h
> @@ -502,6 +502,7 @@ struct nau8825 {
>   	struct semaphore xtalk_sem;
>   	int sw_id;
>   	int irq;
> +	bool suspend_irq_disabled;
>   	int mclk_freq; /* 0 - mclk is disabled */
>   	int button_pressed;
>   	int micbias_voltage;
> 
> base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6
> 
> 

[PATCH v2 4/9] ASoC: rt5663: restore safe state after resume errors

[Pengpeng Hou]

rt5663_resume() ignores regcache_sync() failure before continuing
dependent resume operations. Restore cache-only mode, leave the cache
dirty, and return the replay error before those operations.

Do not schedule jack-detection work or enable its IRQ after a failed
replay.

ASoC still marks the component resumed when the callback reports failure.
Track the suspend-owned IRQ disable so another suspend does not increment
the disable depth again after a failed resume. A later successful resume
releases that one disable reference. This does not add automatic retry or
change the framework's best-effort card resume policy.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: df7c52168ee1 ("ASoC: add rt5663 codec driver")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- keep suspend-owned IRQ disables balanced across failed resume cycles
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/rt5663.c |   17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

diff --git a/sound/soc/codecs/rt5663.c b/sound/soc/codecs/rt5663.c
index ec46e6180cc0ee9fd66df0174bf0d0ac8022dc45..bd6a8e98bf44445e91edf81896f526a180df083f 100644
--- a/sound/soc/codecs/rt5663.c
+++ b/sound/soc/codecs/rt5663.c
@@ -75,6 +75,7 @@ struct rt5663_priv {
 
 	int jack_type;
 	unsigned int irq;
+	bool suspend_irq_disabled;
 };
 
 static const struct reg_sequence rt5663_patch_list[] = {
@@ -3187,8 +3188,10 @@ static int rt5663_suspend(struct snd_soc_component *component)
 {
 	struct rt5663_priv *rt5663 = snd_soc_component_get_drvdata(component);
 
-	if (rt5663->irq)
+	if (rt5663->irq && !rt5663->suspend_irq_disabled) {
 		disable_irq(rt5663->irq);
+		rt5663->suspend_irq_disabled = true;
+	}
 
 	cancel_delayed_work_sync(&rt5663->jack_detect_work);
 	cancel_delayed_work_sync(&rt5663->jd_unplug_work);
@@ -3202,14 +3205,22 @@ static int rt5663_suspend(struct snd_soc_component *component)
 static int rt5663_resume(struct snd_soc_component *component)
 {
 	struct rt5663_priv *rt5663 = snd_soc_component_get_drvdata(component);
+	int ret;
 
 	regcache_cache_only(rt5663->regmap, false);
-	regcache_sync(rt5663->regmap);
+	ret = regcache_sync(rt5663->regmap);
+	if (ret) {
+		regcache_cache_only(rt5663->regmap, true);
+		regcache_mark_dirty(rt5663->regmap);
+		return ret;
+	}
 
 	rt5663_irq(0, rt5663);
 
-	if (rt5663->irq)
+	if (rt5663->suspend_irq_disabled) {
+		rt5663->suspend_irq_disabled = false;
 		enable_irq(rt5663->irq);
+	}
 
 	return 0;
 }

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

[PATCH v2 5/9] ASoC: rt5682s: restore safe state after resume errors

[Pengpeng Hou]

rt5682s_resume() ignores regcache_sync() failure before continuing
dependent resume operations. Restore cache-only mode, leave the cache
dirty, and return the replay error before those operations.

Do not schedule jack-detection work or enable its IRQ after a failed
replay.

ASoC still marks the component resumed when the callback reports failure.
Track the suspend-owned IRQ disable so another suspend does not increment
the disable depth again after a failed resume. A later successful resume
releases that one disable reference. This does not add automatic retry or
change the framework's best-effort card resume policy.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: bdd229ab26be ("ASoC: rt5682s: Add driver for ALC5682I-VS codec")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- keep suspend-owned IRQ disables balanced across failed resume cycles
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/rt5682s.c |   16 +++++++++++++---
 sound/soc/codecs/rt5682s.h |    1 +
 2 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/sound/soc/codecs/rt5682s.c b/sound/soc/codecs/rt5682s.c
index 49ed7793d3a205cd0544ea8b50dd80cbba74b9da..0c669833df7c1565de6d4428fea47af597c7294d 100644
--- a/sound/soc/codecs/rt5682s.c
+++ b/sound/soc/codecs/rt5682s.c
@@ -2884,8 +2884,10 @@ static int rt5682s_suspend(struct snd_soc_component *component)
 {
 	struct rt5682s_priv *rt5682s = snd_soc_component_get_drvdata(component);
 
-	if (rt5682s->irq)
+	if (rt5682s->irq && !rt5682s->suspend_irq_disabled) {
 		disable_irq(rt5682s->irq);
+		rt5682s->suspend_irq_disabled = true;
+	}
 
 	cancel_delayed_work_sync(&rt5682s->jack_detect_work);
 	cancel_delayed_work_sync(&rt5682s->jd_check_work);
@@ -2902,17 +2904,25 @@ static int rt5682s_suspend(struct snd_soc_component *component)
 static int rt5682s_resume(struct snd_soc_component *component)
 {
 	struct rt5682s_priv *rt5682s = snd_soc_component_get_drvdata(component);
+	int ret;
 
 	regcache_cache_only(rt5682s->regmap, false);
-	regcache_sync(rt5682s->regmap);
+	ret = regcache_sync(rt5682s->regmap);
+	if (ret) {
+		regcache_cache_only(rt5682s->regmap, true);
+		regcache_mark_dirty(rt5682s->regmap);
+		return ret;
+	}
 
 	if (rt5682s->hs_jack) {
 		mod_delayed_work(system_power_efficient_wq,
 			&rt5682s->jack_detect_work, msecs_to_jiffies(0));
 	}
 
-	if (rt5682s->irq)
+	if (rt5682s->suspend_irq_disabled) {
+		rt5682s->suspend_irq_disabled = false;
 		enable_irq(rt5682s->irq);
+	}
 
 	return 0;
 }
diff --git a/sound/soc/codecs/rt5682s.h b/sound/soc/codecs/rt5682s.h
index 67f42898de9679679a97ff6a6b65e4b3100c9b59..a9da00e8df78293ec6c5f982682c3d55a23d88bb 100644
--- a/sound/soc/codecs/rt5682s.h
+++ b/sound/soc/codecs/rt5682s.h
@@ -1482,6 +1482,7 @@ struct rt5682s_priv {
 
 	int jack_type;
 	unsigned int irq;
+	bool suspend_irq_disabled;
 	int irq_work_delay_time;
 	int wclk_enabled;
 };

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

[PATCH v2 6/9] ASoC: rt5682: restore safe state after resume errors

[Pengpeng Hou]

rt5682_resume() ignores regcache_sync() failure before continuing
dependent resume operations. Restore cache-only mode, leave the cache
dirty, and return the replay error before those operations.

Leave the SoundWire no-op branch unchanged; stop the I2C jack restore
path on replay failure.

ASoC still marks the component resumed when the callback reports failure.
Track the suspend-owned IRQ disable so another suspend does not increment
the disable depth again after a failed resume. A later successful resume
releases that one disable reference. This does not add automatic retry or
change the framework's best-effort card resume policy.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: 0ddce71c21f0 ("ASoC: rt5682: add rt5682 codec driver")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- keep suspend-owned IRQ disables balanced across failed resume cycles
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/rt5682.c |   16 +++++++++++++---
 sound/soc/codecs/rt5682.h |    1 +
 2 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/sound/soc/codecs/rt5682.c b/sound/soc/codecs/rt5682.c
index 1aceb08b280d6cd4eca01badbc3107b0645c8949..db3f1383f84f5abaa2dc35681c59fe3f4a6188d4 100644
--- a/sound/soc/codecs/rt5682.c
+++ b/sound/soc/codecs/rt5682.c
@@ -2962,8 +2962,10 @@ static int rt5682_suspend(struct snd_soc_component *component)
 	if (rt5682->is_sdw)
 		return 0;
 
-	if (rt5682->irq)
+	if (rt5682->irq && !rt5682->suspend_irq_disabled) {
 		disable_irq(rt5682->irq);
+		rt5682->suspend_irq_disabled = true;
+	}
 
 	cancel_delayed_work_sync(&rt5682->jack_detect_work);
 	cancel_delayed_work_sync(&rt5682->jd_check_work);
@@ -3010,12 +3012,18 @@ static int rt5682_suspend(struct snd_soc_component *component)
 static int rt5682_resume(struct snd_soc_component *component)
 {
 	struct rt5682_priv *rt5682 = snd_soc_component_get_drvdata(component);
+	int ret;
 
 	if (rt5682->is_sdw)
 		return 0;
 
 	regcache_cache_only(rt5682->regmap, false);
-	regcache_sync(rt5682->regmap);
+	ret = regcache_sync(rt5682->regmap);
+	if (ret) {
+		regcache_cache_only(rt5682->regmap, true);
+		regcache_mark_dirty(rt5682->regmap);
+		return ret;
+	}
 
 	if (rt5682->hs_jack && (rt5682->jack_type & SND_JACK_HEADSET) == SND_JACK_HEADSET) {
 		snd_soc_component_update_bits(component, RT5682_SAR_IL_CMD_1,
@@ -3033,8 +3041,10 @@ static int rt5682_resume(struct snd_soc_component *component)
 	mod_delayed_work(system_power_efficient_wq,
 		&rt5682->jack_detect_work, msecs_to_jiffies(0));
 
-	if (rt5682->irq)
+	if (rt5682->suspend_irq_disabled) {
+		rt5682->suspend_irq_disabled = false;
 		enable_irq(rt5682->irq);
+	}
 
 	return 0;
 }
diff --git a/sound/soc/codecs/rt5682.h b/sound/soc/codecs/rt5682.h
index de43a5d99403b926c21b2415af9022b42bf9ae8b..74356cd7fb45d84a49b6d532538ff03c529763e8 100644
--- a/sound/soc/codecs/rt5682.h
+++ b/sound/soc/codecs/rt5682.h
@@ -1466,6 +1466,7 @@ struct rt5682_priv {
 
 	int jack_type;
 	int irq;
+	bool suspend_irq_disabled;
 	int irq_work_delay_time;
 };
 

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

[PATCH v2 7/9] ASoC: wm8978: restore safe state after resume errors

[Pengpeng Hou]

wm8978_resume() ignores cache replay, bias restoration and PLL update
results, allowing subsequent resume steps to run after a failed operation.

Stop at the first failure and request SND_SOC_BIAS_OFF on the error path.
Bias OFF intentionally preserves PLL power, so also clear the power
register as system suspend does. Keep the original failure if rollback
reports another error. Restore cache-only mode and mark the cache dirty
so later accesses do not write through a partially restored device.
Re-enable write-through before a later replay attempt. A positive
update_bits() result means that
bits changed, not an error.

The bias callback currently does not propagate every register write error;
the rollback is therefore best-effort, not a claim of verified hardware
power-down. The ASoC wrapper continues its established best-effort resume.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: 0d34e91596ef ("ASoC: add a WM8978 codec driver")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- restore bias OFF on the error path, as Charles requested
- also turn off PLL power, matching system suspend rather than bias OFF alone
- preserve positive update_bits() success
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/wm8978.c |   31 +++++++++++++++++++++++++++----
 1 file changed, 27 insertions(+), 4 deletions(-)

diff --git a/sound/soc/codecs/wm8978.c b/sound/soc/codecs/wm8978.c
index 258bc9f88f520e4de034393d39317e07e731d1ad..f5d3f26cc21790a8dfc4bf909a5fc5d56ba8f577 100644
--- a/sound/soc/codecs/wm8978.c
+++ b/sound/soc/codecs/wm8978.c
@@ -952,17 +952,40 @@ static int wm8978_resume(struct snd_soc_component *component)
 {
 	struct wm8978_priv *wm8978 = snd_soc_component_get_drvdata(component);
 	struct snd_soc_dapm_context *dapm = snd_soc_component_to_dapm(component);
+	int ret, err;
 
 	/* Sync reg_cache with the hardware */
-	regcache_sync(wm8978->regmap);
+	regcache_cache_only(wm8978->regmap, false);
+	ret = regcache_sync(wm8978->regmap);
+	if (ret)
+		goto restore_off;
 
-	snd_soc_dapm_force_bias_level(dapm, SND_SOC_BIAS_STANDBY);
+	ret = snd_soc_dapm_force_bias_level(dapm, SND_SOC_BIAS_STANDBY);
+	if (ret)
+		goto restore_off;
 
-	if (wm8978->f_pllout)
+	if (wm8978->f_pllout) {
 		/* Switch PLL on */
-		snd_soc_component_update_bits(component, WM8978_POWER_MANAGEMENT_1, 0x20, 0x20);
+		ret = snd_soc_component_update_bits(component,
+						    WM8978_POWER_MANAGEMENT_1,
+						    0x20, 0x20);
+		if (ret < 0)
+			goto restore_off;
+	}
 
 	return 0;
+
+restore_off:
+	err = snd_soc_dapm_force_bias_level(dapm, SND_SOC_BIAS_OFF);
+	if (err)
+		dev_err(component->dev, "Failed to restore bias OFF: %d\n", err);
+	/* Bias OFF preserves PLL; system suspend turns it off as well. */
+	err = snd_soc_component_write(component, WM8978_POWER_MANAGEMENT_1, 0);
+	if (err)
+		dev_err(component->dev, "Failed to disable PLL: %d\n", err);
+	regcache_cache_only(wm8978->regmap, true);
+	regcache_mark_dirty(wm8978->regmap);
+	return ret;
 }
 
 /*

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

Re: [PATCH v2 7/9] ASoC: wm8978: restore safe state after resume errors

[Charles Keepax]

On Wed, Oct 14, 2026 at 11:24:50AM +0800, Pengpeng Hou wrote:
> wm8978_resume() ignores cache replay, bias restoration and PLL update
> results, allowing subsequent resume steps to run after a failed operation.
> 
> Stop at the first failure and request SND_SOC_BIAS_OFF on the error path.
> Bias OFF intentionally preserves PLL power, so also clear the power
> register as system suspend does. Keep the original failure if rollback
> reports another error. Restore cache-only mode and mark the cache dirty
> so later accesses do not write through a partially restored device.
> Re-enable write-through before a later replay attempt. A positive
> update_bits() result means that
> bits changed, not an error.
> 
> The bias callback currently does not propagate every register write error;
> the rollback is therefore best-effort, not a claim of verified hardware
> power-down. The ASoC wrapper continues its established best-effort resume.
> 
> The issue was found by our static-analysis tool and manually reviewed.
> 
> Fixes: 0d34e91596ef ("ASoC: add a WM8978 codec driver")
> Assisted-by: gpt 5
> Signed-off-by: Pengpeng Hou <hppiscas@163.com>
> ---
> Changes since v1:
> - restore cache-only mode and dirty state on replay failure
> - restore bias OFF on the error path, as Charles requested

I was more interested in a rational for either approach. But
probably adding it seems more logical to me.

> - also turn off PLL power, matching system suspend rather than bias OFF alone
> - preserve positive update_bits() success
> - send as one correctly threaded series, as Mark requested
> 
> Previous series:
> https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/
> 
>  sound/soc/codecs/wm8978.c |   31 +++++++++++++++++++++++++++----
>  1 file changed, 27 insertions(+), 4 deletions(-)
> 
> diff --git a/sound/soc/codecs/wm8978.c b/sound/soc/codecs/wm8978.c
> index 258bc9f88f520e4de034393d39317e07e731d1ad..f5d3f26cc21790a8dfc4bf909a5fc5d56ba8f577 100644
> --- a/sound/soc/codecs/wm8978.c
> +++ b/sound/soc/codecs/wm8978.c
> @@ -952,17 +952,40 @@ static int wm8978_resume(struct snd_soc_component *component)
>  {
>  	struct wm8978_priv *wm8978 = snd_soc_component_get_drvdata(component);
>  	struct snd_soc_dapm_context *dapm = snd_soc_component_to_dapm(component);
> +	int ret, err;
>  
>  	/* Sync reg_cache with the hardware */
> -	regcache_sync(wm8978->regmap);
> +	regcache_cache_only(wm8978->regmap, false);

Nothing ever set cache_only true? Surely you either want a
separate patch to add cache_only support, or to not do this.
Although given the driver appears to only support system suspend
I would assume there isn't much point in cache only.

Thanks,
Charles

[PATCH v2 8/9] ASoC: rt5640: restore safe state after resume errors

[Pengpeng Hou]

rt5640_resume() ignores regcache_sync() failure before continuing
dependent resume operations. Restore cache-only mode, leave the cache
dirty, and return the replay error before those operations.

Both the codec IRQ and the optional GPIO jack IRQ can queue jack work.
Quiesce both before cancelling that work, retain their disable
references on replay failure, and clear each reference if jack teardown
frees that IRQ. Restore the optional LDO GPIO on failure and reject new
jack setup until a later replay succeeds; teardown remains allowed.

ASoC still marks the component resumed when the callback reports failure.
Track the suspend-owned IRQ disable so another suspend does not increment
the disable depth again after a failed resume. A later successful resume
releases that one disable reference. This does not add automatic retry or
change the framework's best-effort card resume policy.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: 997b05203b0a ("ASoC: add RT5640 CODEC driver")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- keep suspend-owned IRQ disables balanced across failed resume cycles
- cover both IRQ producers and jack teardown/re-registration after failure
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/rt5640.c |   37 ++++++++++++++++++++++++++++++++++---
 sound/soc/codecs/rt5640.h |    3 +++
 2 files changed, 37 insertions(+), 3 deletions(-)

diff --git a/sound/soc/codecs/rt5640.c b/sound/soc/codecs/rt5640.c
index d6068bb5cb4c841ebbc3892b5f6271fca246701d..8b592370d1294ae70d38cb40998f813b6eddc2ad 100644
--- a/sound/soc/codecs/rt5640.c
+++ b/sound/soc/codecs/rt5640.c
@@ -2433,11 +2433,13 @@ static void rt5640_disable_irq_and_cancel_work(void *data)
 	if (rt5640->jd_gpio_irq_requested) {
 		free_irq(rt5640->jd_gpio_irq, rt5640);
 		rt5640->jd_gpio_irq_requested = false;
+		rt5640->suspend_jd_gpio_irq_disabled = false;
 	}
 
 	if (rt5640->irq_requested) {
 		free_irq(rt5640->irq, rt5640);
 		rt5640->irq_requested = false;
+		rt5640->suspend_irq_disabled = false;
 	}
 
 	cancel_delayed_work_sync(&rt5640->jack_work);
@@ -2646,6 +2648,10 @@ static int rt5640_set_jack(struct snd_soc_component *component,
 {
 	struct rt5640_priv *rt5640 = snd_soc_component_get_drvdata(component);
 
+	/* Do not publish fresh IRQs while the register cache is unrestored. */
+	if (jack && rt5640->resume_failed)
+		return -EIO;
+
 	if (jack) {
 		if (rt5640->jd_src == RT5640_JD_SRC_HDA_HEADER)
 			rt5640_enable_hda_jack_detect(component, jack);
@@ -2803,7 +2809,15 @@ static int rt5640_suspend(struct snd_soc_component *component)
 
 	if (rt5640->jack) {
 		/* disable jack interrupts during system suspend */
-		disable_irq(rt5640->irq);
+		if (rt5640->irq_requested && !rt5640->suspend_irq_disabled) {
+			disable_irq(rt5640->irq);
+			rt5640->suspend_irq_disabled = true;
+		}
+		if (rt5640->jd_gpio_irq_requested &&
+		    !rt5640->suspend_jd_gpio_irq_disabled) {
+			disable_irq(rt5640->jd_gpio_irq);
+			rt5640->suspend_jd_gpio_irq_disabled = true;
+		}
 		cancel_delayed_work_sync(&rt5640->jack_work);
 		cancel_delayed_work_sync(&rt5640->bp_work);
 	}
@@ -2821,6 +2835,7 @@ static int rt5640_suspend(struct snd_soc_component *component)
 static int rt5640_resume(struct snd_soc_component *component)
 {
 	struct rt5640_priv *rt5640 = snd_soc_component_get_drvdata(component);
+	int ret;
 
 	if (rt5640->ldo1_en) {
 		gpiod_set_value_cansleep(rt5640->ldo1_en, 1);
@@ -2828,7 +2843,16 @@ static int rt5640_resume(struct snd_soc_component *component)
 	}
 
 	regcache_cache_only(rt5640->regmap, false);
-	regcache_sync(rt5640->regmap);
+	ret = regcache_sync(rt5640->regmap);
+	if (ret) {
+		rt5640->resume_failed = true;
+		regcache_cache_only(rt5640->regmap, true);
+		regcache_mark_dirty(rt5640->regmap);
+		if (rt5640->ldo1_en)
+			gpiod_set_value_cansleep(rt5640->ldo1_en, 0);
+		return ret;
+	}
+	rt5640->resume_failed = false;
 
 	if (rt5640->jack) {
 		if (rt5640->jd_src == RT5640_JD_SRC_HDA_HEADER) {
@@ -2857,7 +2881,14 @@ static int rt5640_resume(struct snd_soc_component *component)
 			}
 		}
 
-		enable_irq(rt5640->irq);
+		if (rt5640->suspend_irq_disabled) {
+			rt5640->suspend_irq_disabled = false;
+			enable_irq(rt5640->irq);
+		}
+		if (rt5640->suspend_jd_gpio_irq_disabled) {
+			rt5640->suspend_jd_gpio_irq_disabled = false;
+			enable_irq(rt5640->jd_gpio_irq);
+		}
 		queue_delayed_work(system_dfl_long_wq, &rt5640->jack_work, 0);
 	}
 
diff --git a/sound/soc/codecs/rt5640.h b/sound/soc/codecs/rt5640.h
index 8a12cee76bdc65b73c3a1c4b6ef53e0dc3bdb8b9..eceff517d44eb1b0d881c5a07c825cf4be7d714b 100644
--- a/sound/soc/codecs/rt5640.h
+++ b/sound/soc/codecs/rt5640.h
@@ -2154,7 +2154,10 @@ struct rt5640_priv {
 	bool hp_mute;
 	bool asrc_en;
 	bool irq_requested;
+	bool suspend_irq_disabled;
 	bool jd_gpio_irq_requested;
+	bool suspend_jd_gpio_irq_disabled;
+	bool resume_failed;
 
 	/* Jack and button detect data */
 	bool ovcd_irq_enabled;

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

[PATCH v2 9/9] ASoC: rt5677: restore safe state after resume errors

[Pengpeng Hou]

rt5677_resume() ignores regcache_sync() failure before continuing
dependent resume operations. Restore cache-only mode, leave the cache
dirty, and return the replay error before those operations.

First return physical regmap write errors from the normal-mode write
wrapper, which currently discards them, so cache replay can observe bus
failures. Restore reset and the optional LDO GPIO on replay failure.
Remember that cache restoration is still pending independently of the
requested DSP VAD mode: DAPM or controls can change that mode after
failed resume. Suppress DSP work while restoration is pending and
require replay on the next resume even if VAD has since been requested.
After successful replay, preserve the existing MCLK settling delay
before restarting requested DSP work. The already-running, retained VAD
path and DSP-mode transport are unchanged.

ASoC still marks the component resumed when the callback reports failure.
Track the suspend-owned IRQ disable so another suspend does not increment
the disable depth again after a failed resume. A later successful resume
releases that one disable reference. This does not add automatic retry or
change the framework's best-effort card resume policy.

The issue was found by our static-analysis tool and manually reviewed.

Fixes: 0e826e867264 ("ASoC: add RT5677 CODEC driver")
Assisted-by: gpt 5
Signed-off-by: Pengpeng Hou <hppiscas@163.com>
---
Changes since v1:
- restore cache-only mode and dirty state on replay failure
- keep suspend-owned IRQ disables balanced across failed resume cycles
- propagate the normal-mode transport write result to cache replay
- preserve pending cache restoration across changes to the VAD request
- keep DSP work paused until replay succeeds, retaining the MCLK delay
- send as one correctly threaded series, as Mark requested

Previous series:
https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/

 sound/soc/codecs/rt5677.c |   37 ++++++++++++++++++++++++++++++++-----
 sound/soc/codecs/rt5677.h |    2 ++
 2 files changed, 34 insertions(+), 5 deletions(-)

diff --git a/sound/soc/codecs/rt5677.c b/sound/soc/codecs/rt5677.c
index 7b80922236e5818e846d9ecc6e23793e6ae40714..fd1f840d6794ea95b47d803ba4977fdb81e184a9 100644
--- a/sound/soc/codecs/rt5677.c
+++ b/sound/soc/codecs/rt5677.c
@@ -885,6 +885,9 @@ static void rt5677_dsp_work(struct work_struct *work)
 	bool enable = rt5677->dsp_vad_en;
 	int i, val;
 
+	/* DAPM may request VAD even after component resume has failed. */
+	if (READ_ONCE(rt5677->cache_restore_pending))
+		return;
 
 	dev_info(rt5677->component->dev, "DSP VAD: enable=%d, activity=%d\n",
 			enable, activity);
@@ -4939,10 +4942,14 @@ static int rt5677_suspend(struct snd_soc_component *component)
 
 	if (rt5677->irq) {
 		cancel_delayed_work_sync(&rt5677->resume_irq_check);
-		disable_irq(rt5677->irq);
+		if (!rt5677->suspend_irq_disabled) {
+			disable_irq(rt5677->irq);
+			rt5677->suspend_irq_disabled = true;
+		}
 	}
 
 	if (!rt5677->dsp_vad_en) {
+		WRITE_ONCE(rt5677->cache_restore_pending, true);
 		regcache_cache_only(rt5677->regmap, true);
 		regcache_mark_dirty(rt5677->regmap);
 
@@ -4956,8 +4963,11 @@ static int rt5677_suspend(struct snd_soc_component *component)
 static int rt5677_resume(struct snd_soc_component *component)
 {
 	struct rt5677_priv *rt5677 = snd_soc_component_get_drvdata(component);
+	bool restore_cache = !rt5677->dsp_vad_en ||
+			     READ_ONCE(rt5677->cache_restore_pending);
+	int ret;
 
-	if (!rt5677->dsp_vad_en) {
+	if (restore_cache) {
 		rt5677->pll_src = 0;
 		rt5677->pll_in = 0;
 		rt5677->pll_out = 0;
@@ -4967,11 +4977,28 @@ static int rt5677_resume(struct snd_soc_component *component)
 			msleep(10);
 
 		regcache_cache_only(rt5677->regmap, false);
-		regcache_sync(rt5677->regmap);
+		ret = regcache_sync(rt5677->regmap);
+		if (ret) {
+			WRITE_ONCE(rt5677->cache_restore_pending, true);
+			regcache_cache_only(rt5677->regmap, true);
+			regcache_mark_dirty(rt5677->regmap);
+			gpiod_set_value_cansleep(rt5677->reset_pin, 1);
+			gpiod_set_value_cansleep(rt5677->pow_ldo2, 0);
+			return ret;
+		}
+		WRITE_ONCE(rt5677->cache_restore_pending, false);
+		if (rt5677->dsp_vad_en) {
+			/* Allow MCLK to settle, as when leaving bias OFF. */
+			schedule_delayed_work(&rt5677->dsp_work,
+					     msecs_to_jiffies(1000));
+		}
 	}
 
 	if (rt5677->irq) {
-		enable_irq(rt5677->irq);
+		if (rt5677->suspend_irq_disabled) {
+			rt5677->suspend_irq_disabled = false;
+			enable_irq(rt5677->irq);
+		}
 		schedule_delayed_work(&rt5677->resume_irq_check, 0);
 	}
 
@@ -5021,7 +5048,7 @@ static int rt5677_write(void *context, unsigned int reg, unsigned int val)
 			rt5677_dsp_mode_i2c_write(rt5677, reg, val);
 		}
 	} else {
-		regmap_write(rt5677->regmap_physical, reg, val);
+		return regmap_write(rt5677->regmap_physical, reg, val);
 	}
 
 	return 0;
diff --git a/sound/soc/codecs/rt5677.h b/sound/soc/codecs/rt5677.h
index 876f7674015c2c0a5421feb25a8c80313b18310a..e6e915453f2efbe78c9d2dd3e909397d9ad715ca 100644
--- a/sound/soc/codecs/rt5677.h
+++ b/sound/soc/codecs/rt5677.h
@@ -1788,6 +1788,7 @@ struct rt5677_priv {
 #endif
 	bool dsp_vad_en_request; /* DSP VAD enable/disable request */
 	bool dsp_vad_en; /* dsp_work parameter */
+	bool cache_restore_pending;
 	bool is_dsp_mode;
 	bool is_vref_slow;
 	struct delayed_work dsp_work;
@@ -1798,6 +1799,7 @@ struct rt5677_priv {
 	unsigned int irq_en;
 	struct delayed_work resume_irq_check;
 	int irq;
+	bool suspend_irq_disabled;
 
 	int (*set_dsp_vad)(struct snd_soc_component *component, bool on);
 };

base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6

Re: [PATCH v2 9/9] ASoC: rt5677: restore safe state after resume errors

[Ajay Kumar Nandam]

On 10/14/2026 8:54 AM, Pengpeng Hou wrote:
> rt5677_resume() ignores regcache_sync() failure before continuing
> dependent resume operations. Restore cache-only mode, leave the cache
> dirty, and return the replay error before those operations.
> 
> First return physical regmap write errors from the normal-mode write
> wrapper, which currently discards them, so cache replay can observe bus
> failures. Restore reset and the optional LDO GPIO on replay failure.
> Remember that cache restoration is still pending independently of the
> requested DSP VAD mode: DAPM or controls can change that mode after
> failed resume. Suppress DSP work while restoration is pending and
> require replay on the next resume even if VAD has since been requested.
> After successful replay, preserve the existing MCLK settling delay
> before restarting requested DSP work. The already-running, retained VAD
> path and DSP-mode transport are unchanged.
> 
> ASoC still marks the component resumed when the callback reports failure.
> Track the suspend-owned IRQ disable so another suspend does not increment
> the disable depth again after a failed resume. A later successful resume
> releases that one disable reference. This does not add automatic retry or
> change the framework's best-effort card resume policy.
> 
> The issue was found by our static-analysis tool and manually reviewed.
> 
> Fixes: 0e826e867264 ("ASoC: add RT5677 CODEC driver")
> Assisted-by: gpt 5
> Signed-off-by: Pengpeng Hou <hppiscas@163.com>
> ---
> Changes since v1:
> - restore cache-only mode and dirty state on replay failure
> - keep suspend-owned IRQ disables balanced across failed resume cycles
> - propagate the normal-mode transport write result to cache replay
> - preserve pending cache restoration across changes to the VAD request
> - keep DSP work paused until replay succeeds, retaining the MCLK delay
> - send as one correctly threaded series, as Mark requested
> 
> Previous series:
> https://lore.kernel.org/all/20260906034321.86080-1-hppiscas@163.com/
> 
>   sound/soc/codecs/rt5677.c |   37 ++++++++++++++++++++++++++++++++-----
>   sound/soc/codecs/rt5677.h |    2 ++
>   2 files changed, 34 insertions(+), 5 deletions(-)
> 
> diff --git a/sound/soc/codecs/rt5677.c b/sound/soc/codecs/rt5677.c
> index 7b80922236e5818e846d9ecc6e23793e6ae40714..fd1f840d6794ea95b47d803ba4977fdb81e184a9 100644
> --- a/sound/soc/codecs/rt5677.c
> +++ b/sound/soc/codecs/rt5677.c
> @@ -885,6 +885,9 @@ static void rt5677_dsp_work(struct work_struct *work)
>   	bool enable = rt5677->dsp_vad_en;
>   	int i, val;
>   
> +	/* DAPM may request VAD even after component resume has failed. */
> +	if (READ_ONCE(rt5677->cache_restore_pending))
> +		return;
>   
>   	dev_info(rt5677->component->dev, "DSP VAD: enable=%d, activity=%d\n",
>   			enable, activity);
> @@ -4939,10 +4942,14 @@ static int rt5677_suspend(struct snd_soc_component *component)
>   
>   	if (rt5677->irq) {
>   		cancel_delayed_work_sync(&rt5677->resume_irq_check);
> -		disable_irq(rt5677->irq);
> +		if (!rt5677->suspend_irq_disabled) {
> +			disable_irq(rt5677->irq);
> +			rt5677->suspend_irq_disabled = true;
> +		}

Similar to the nau8825 case, should this new suspend_irq_disabled 
tracking be based on whether the IRQ was successfully requested?

rt5677_init_irq() still sets rt5677->irq = i2c->irq even when
devm_request_threaded_irq() fails, and probe continues after only 
logging that error.

Thanks
Ajay Kumar Nandam

>   	}
>   
>   	if (!rt5677->dsp_vad_en) {
> +		WRITE_ONCE(rt5677->cache_restore_pending, true);
>   		regcache_cache_only(rt5677->regmap, true);
>   		regcache_mark_dirty(rt5677->regmap);
>   
> @@ -4956,8 +4963,11 @@ static int rt5677_suspend(struct snd_soc_component *component)
>   static int rt5677_resume(struct snd_soc_component *component)
>   {
>   	struct rt5677_priv *rt5677 = snd_soc_component_get_drvdata(component);
> +	bool restore_cache = !rt5677->dsp_vad_en ||
> +			     READ_ONCE(rt5677->cache_restore_pending);
> +	int ret;
>   
> -	if (!rt5677->dsp_vad_en) {
> +	if (restore_cache) {
>   		rt5677->pll_src = 0;
>   		rt5677->pll_in = 0;
>   		rt5677->pll_out = 0;
> @@ -4967,11 +4977,28 @@ static int rt5677_resume(struct snd_soc_component *component)
>   			msleep(10);
>   
>   		regcache_cache_only(rt5677->regmap, false);
> -		regcache_sync(rt5677->regmap);
> +		ret = regcache_sync(rt5677->regmap);
> +		if (ret) {
> +			WRITE_ONCE(rt5677->cache_restore_pending, true);
> +			regcache_cache_only(rt5677->regmap, true);
> +			regcache_mark_dirty(rt5677->regmap);
> +			gpiod_set_value_cansleep(rt5677->reset_pin, 1);
> +			gpiod_set_value_cansleep(rt5677->pow_ldo2, 0);
> +			return ret;
> +		}
> +		WRITE_ONCE(rt5677->cache_restore_pending, false);
> +		if (rt5677->dsp_vad_en) {
> +			/* Allow MCLK to settle, as when leaving bias OFF. */
> +			schedule_delayed_work(&rt5677->dsp_work,
> +					     msecs_to_jiffies(1000));
> +		}
>   	}
>   
>   	if (rt5677->irq) {
> -		enable_irq(rt5677->irq);
> +		if (rt5677->suspend_irq_disabled) {
> +			rt5677->suspend_irq_disabled = false;
> +			enable_irq(rt5677->irq);
> +		}
>   		schedule_delayed_work(&rt5677->resume_irq_check, 0);
>   	}
>   
> @@ -5021,7 +5048,7 @@ static int rt5677_write(void *context, unsigned int reg, unsigned int val)
>   			rt5677_dsp_mode_i2c_write(rt5677, reg, val);
>   		}
>   	} else {
> -		regmap_write(rt5677->regmap_physical, reg, val);
> +		return regmap_write(rt5677->regmap_physical, reg, val);
>   	}
>   
>   	return 0;
> diff --git a/sound/soc/codecs/rt5677.h b/sound/soc/codecs/rt5677.h
> index 876f7674015c2c0a5421feb25a8c80313b18310a..e6e915453f2efbe78c9d2dd3e909397d9ad715ca 100644
> --- a/sound/soc/codecs/rt5677.h
> +++ b/sound/soc/codecs/rt5677.h
> @@ -1788,6 +1788,7 @@ struct rt5677_priv {
>   #endif
>   	bool dsp_vad_en_request; /* DSP VAD enable/disable request */
>   	bool dsp_vad_en; /* dsp_work parameter */
> +	bool cache_restore_pending;
>   	bool is_dsp_mode;
>   	bool is_vref_slow;
>   	struct delayed_work dsp_work;
> @@ -1798,6 +1799,7 @@ struct rt5677_priv {
>   	unsigned int irq_en;
>   	struct delayed_work resume_irq_check;
>   	int irq;
> +	bool suspend_irq_disabled;
>   
>   	int (*set_dsp_vad)(struct snd_soc_component *component, bool on);
>   };
> 
> base-commit: 8ef28aef33f2d5da6959ad6b579ec3cfccc009f6
> 
>