[PATCH 0/4] Apple ADMAC peripheral reset clean-up

[PATCH 0/4] Apple ADMAC peripheral reset clean-up

[Martin Povišer]

On Apple SoCs the ADMAC peripheral shares a reset line with another
audio-related peripheral which is the MCA.

On the side of the MCA, the reset is described in binding (staged in
broonie-sound/for-6.1) and in the soon-to-be-merged series adding new
nodes to t8103/t600x SoC device trees [1]. The staged MCA driver does
make use of the reset, knowing it's a shared one.

This series catches up on the ADMAC side, adds the reset to the binding
and describes it in the t8103/t600x DTs (on top of [1]). It also pulls
the shared reset from the driver.

Since [1] already includes another ADMAC binding change and will
probably be fully merged into Hector's asahi-soc tree, I propose the
binding change here takes the same route. Leaving patches #3 and #4 to
go into dmaengine.

[1] https://lore.kernel.org/asahi/87pmful5r4.wl-maz@kernel.org/T/#t

Martin Povišer (4):
  dt-bindings: dma: apple,admac: Add reset
  arm64: dts: apple: Add ADMAC resets on t8103/t600x
  dmaengine: apple-admac: Do not use devres for IRQs
  dmaengine: apple-admac: Trigger shared reset

-- 
2.33.0

[PATCH 1/4] dt-bindings: dma: apple,admac: Add reset

[Martin Povišer]

On the SoCs there is usually a shared audio reset line, so add
a property for that.

Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
---
 Documentation/devicetree/bindings/dma/apple,admac.yaml | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/Documentation/devicetree/bindings/dma/apple,admac.yaml b/Documentation/devicetree/bindings/dma/apple,admac.yaml
index 3b1e667f7ea0..97282469e4af 100644
--- a/Documentation/devicetree/bindings/dma/apple,admac.yaml
+++ b/Documentation/devicetree/bindings/dma/apple,admac.yaml
@@ -56,6 +56,9 @@ properties:
   power-domains:
     maxItems: 1
 
+  resets:
+    maxItems: 1
+
 required:
   - compatible
   - reg
-- 
2.33.0

[PATCH 2/4] arm64: dts: apple: Add ADMAC resets on t8103/t600x

[Martin Povišer]

There's a shared reset for the ADMAC and MCA peripherals. The MCA node
already describes it, and being a shared reset it is important we
describe it on the ADMAC node too.

Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
---
 arch/arm64/boot/dts/apple/t600x-die0.dtsi | 1 +
 arch/arm64/boot/dts/apple/t8103.dtsi      | 1 +
 2 files changed, 2 insertions(+)

diff --git a/arch/arm64/boot/dts/apple/t600x-die0.dtsi b/arch/arm64/boot/dts/apple/t600x-die0.dtsi
index 639c90e108a7..0b8958a8db77 100644
--- a/arch/arm64/boot/dts/apple/t600x-die0.dtsi
+++ b/arch/arm64/boot/dts/apple/t600x-die0.dtsi
@@ -181,6 +181,7 @@ admac: dma-controller@39b400000 {
 				      <0>;
 		iommus = <&dart_sio_0 2>, <&dart_sio_1 2>;
 		power-domains = <&ps_sio_adma>;
+		resets = <&ps_audio_p>;
 	};
 
 	mca: mca@39b600000 {
diff --git a/arch/arm64/boot/dts/apple/t8103.dtsi b/arch/arm64/boot/dts/apple/t8103.dtsi
index 51bc901482db..358906a9274d 100644
--- a/arch/arm64/boot/dts/apple/t8103.dtsi
+++ b/arch/arm64/boot/dts/apple/t8103.dtsi
@@ -249,6 +249,7 @@ admac: dma-controller@238200000 {
 			#dma-cells = <1>;
 			iommus = <&dart_sio 2>;
 			power-domains = <&ps_sio_adma>;
+			resets = <&ps_audio_p>;
 		};
 
 		mca: i2s@238400000 {
-- 
2.33.0

[PATCH 3/4] dmaengine: apple-admac: Do not use devres for IRQs

[Martin Povišer]

This is in advance of adding support for triggering the reset signal to
the peripheral, since registering the IRQ handler will have to be
sequenced with it.

Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
---
 drivers/dma/apple-admac.c | 27 ++++++++++++++++++---------
 1 file changed, 18 insertions(+), 9 deletions(-)

diff --git a/drivers/dma/apple-admac.c b/drivers/dma/apple-admac.c
index d1f74a3aa999..7d1b76678032 100644
--- a/drivers/dma/apple-admac.c
+++ b/drivers/dma/apple-admac.c
@@ -96,6 +96,7 @@ struct admac_data {
 	struct device *dev;
 	__iomem void *base;
 
+	int irq;
 	int irq_index;
 	int nchannels;
 	struct admac_chan channels[];
@@ -724,12 +725,7 @@ static int admac_probe(struct platform_device *pdev)
 
 	if (irq < 0)
 		return dev_err_probe(&pdev->dev, irq, "no usable interrupt\n");
-
-	err = devm_request_irq(&pdev->dev, irq, admac_interrupt,
-			       0, dev_name(&pdev->dev), ad);
-	if (err)
-		return dev_err_probe(&pdev->dev, err,
-				     "unable to register interrupt\n");
+	ad->irq = irq;
 
 	ad->base = devm_platform_ioremap_resource(pdev, 0);
 	if (IS_ERR(ad->base))
@@ -774,17 +770,29 @@ static int admac_probe(struct platform_device *pdev)
 		tasklet_setup(&adchan->tasklet, admac_chan_tasklet);
 	}
 
-	err = dma_async_device_register(&ad->dma);
+	err = request_irq(irq, admac_interrupt, 0, dev_name(&pdev->dev), ad);
 	if (err)
-		return dev_err_probe(&pdev->dev, err, "failed to register DMA device\n");
+		return dev_err_probe(&pdev->dev, err,
+				     "unable to register interrupt\n");
+
+	err = dma_async_device_register(&ad->dma);
+	if (err) {
+		dev_err_probe(&pdev->dev, err, "failed to register DMA device\n");
+		goto free_irq;
+	}
 
 	err = of_dma_controller_register(pdev->dev.of_node, admac_dma_of_xlate, ad);
 	if (err) {
 		dma_async_device_unregister(&ad->dma);
-		return dev_err_probe(&pdev->dev, err, "failed to register with OF\n");
+		dev_err_probe(&pdev->dev, err, "failed to register with OF\n");
+		goto free_irq;
 	}
 
 	return 0;
+
+free_irq:
+	free_irq(ad->irq, ad);
+	return err;
 }
 
 static int admac_remove(struct platform_device *pdev)
@@ -793,6 +801,7 @@ static int admac_remove(struct platform_device *pdev)
 
 	of_dma_controller_free(pdev->dev.of_node);
 	dma_async_device_unregister(&ad->dma);
+	free_irq(ad->irq, ad);
 
 	return 0;
 }
-- 
2.33.0

[PATCH 4/4] dmaengine: apple-admac: Trigger shared reset

[Martin Povišer]

If a reset domain is attached to the device, obtain a shared reference
to it and trigger it. Typically on a chip the ADMAC controller will
share a reset domain with the MCA peripheral.

Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
---
 drivers/dma/apple-admac.c | 22 +++++++++++++++++++---
 1 file changed, 19 insertions(+), 3 deletions(-)

diff --git a/drivers/dma/apple-admac.c b/drivers/dma/apple-admac.c
index 7d1b76678032..317ca76ccafd 100644
--- a/drivers/dma/apple-admac.c
+++ b/drivers/dma/apple-admac.c
@@ -12,8 +12,9 @@
 #include <linux/module.h>
 #include <linux/of_device.h>
 #include <linux/of_dma.h>
-#include <linux/interrupt.h>
+#include <linux/reset.h>
 #include <linux/spinlock.h>
+#include <linux/interrupt.h>
 
 #include "dmaengine.h"
 
@@ -95,6 +96,7 @@ struct admac_data {
 	struct dma_device dma;
 	struct device *dev;
 	__iomem void *base;
+	struct reset_control *rstc;
 
 	int irq;
 	int irq_index;
@@ -732,6 +734,10 @@ static int admac_probe(struct platform_device *pdev)
 		return dev_err_probe(&pdev->dev, PTR_ERR(ad->base),
 				     "unable to obtain MMIO resource\n");
 
+	ad->rstc = devm_reset_control_get_optional_shared(&pdev->dev, NULL);
+	if (IS_ERR(ad->rstc))
+		return PTR_ERR(ad->rstc);
+
 	dma = &ad->dma;
 
 	dma_cap_set(DMA_PRIVATE, dma->cap_mask);
@@ -770,10 +776,17 @@ static int admac_probe(struct platform_device *pdev)
 		tasklet_setup(&adchan->tasklet, admac_chan_tasklet);
 	}
 
-	err = request_irq(irq, admac_interrupt, 0, dev_name(&pdev->dev), ad);
+	err = reset_control_reset(ad->rstc);
 	if (err)
 		return dev_err_probe(&pdev->dev, err,
-				     "unable to register interrupt\n");
+				     "unable to trigger reset\n");
+
+	err = request_irq(irq, admac_interrupt, 0, dev_name(&pdev->dev), ad);
+	if (err) {
+		dev_err_probe(&pdev->dev, err,
+				"unable to register interrupt\n");
+		goto free_reset;
+	}
 
 	err = dma_async_device_register(&ad->dma);
 	if (err) {
@@ -792,6 +805,8 @@ static int admac_probe(struct platform_device *pdev)
 
 free_irq:
 	free_irq(ad->irq, ad);
+free_reset:
+	reset_control_rearm(ad->rstc);
 	return err;
 }
 
@@ -802,6 +817,7 @@ static int admac_remove(struct platform_device *pdev)
 	of_dma_controller_free(pdev->dev.of_node);
 	dma_async_device_unregister(&ad->dma);
 	free_irq(ad->irq, ad);
+	reset_control_rearm(ad->rstc);
 
 	return 0;
 }
-- 
2.33.0

Re: [PATCH 1/4] dt-bindings: dma: apple,admac: Add reset

[Krzysztof Kozlowski]

On 18/09/2022 10:58, Martin Povišer wrote:
> On the SoCs there is usually a shared audio reset line, so add
> a property for that.
> 
> Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
> ---
>  Documentation/devicetree/bindings/dma/apple,admac.yaml | 3 +++


Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>


Best regards,
Krzysztof

Re: [PATCH 0/4] Apple ADMAC peripheral reset clean-up

[Hector Martin]

On 18/09/2022 18.58, Martin Povišer wrote:
> On Apple SoCs the ADMAC peripheral shares a reset line with another
> audio-related peripheral which is the MCA.
> 
> On the side of the MCA, the reset is described in binding (staged in
> broonie-sound/for-6.1) and in the soon-to-be-merged series adding new
> nodes to t8103/t600x SoC device trees [1]. The staged MCA driver does
> make use of the reset, knowing it's a shared one.
> 
> This series catches up on the ADMAC side, adds the reset to the binding
> and describes it in the t8103/t600x DTs (on top of [1]). It also pulls
> the shared reset from the driver.
> 
> Since [1] already includes another ADMAC binding change and will
> probably be fully merged into Hector's asahi-soc tree, I propose the
> binding change here takes the same route. Leaving patches #3 and #4 to
> go into dmaengine.
> 
> [1] https://lore.kernel.org/asahi/87pmful5r4.wl-maz@kernel.org/T/#t
> 
> Martin Povišer (4):
>   dt-bindings: dma: apple,admac: Add reset
>   arm64: dts: apple: Add ADMAC resets on t8103/t600x
>   dmaengine: apple-admac: Do not use devres for IRQs
>   dmaengine: apple-admac: Trigger shared reset
> 

Applied patches #1-#2 to asahi-soc/dt, thanks!

- Hector

Re: [PATCH 3/4] dmaengine: apple-admac: Do not use devres for IRQs

[Vinod Koul]

On 18-09-22, 11:58, Martin Povišer wrote:
> This is in advance of adding support for triggering the reset signal to
> the peripheral, since registering the IRQ handler will have to be
> sequenced with it.

Applied 3-4, thanks

-- 
~Vinod

Re: [PATCH 4/4] dmaengine: apple-admac: Trigger shared reset

[Philipp Zabel]

Hi Martin,

On So, 2022-09-18 at 11:58 +0200, Martin Povišer wrote:
> If a reset domain is attached to the device, obtain a shared reference
> to it and trigger it. Typically on a chip the ADMAC controller will
> share a reset domain with the MCA peripheral.
> 
> Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
> ---
>  drivers/dma/apple-admac.c | 22 +++++++++++++++++++---
>  1 file changed, 19 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/dma/apple-admac.c b/drivers/dma/apple-admac.c
> index 7d1b76678032..317ca76ccafd 100644
> --- a/drivers/dma/apple-admac.c
> +++ b/drivers/dma/apple-admac.c
> @@ -12,8 +12,9 @@
>  #include <linux/module.h>
>  #include <linux/of_device.h>
>  #include <linux/of_dma.h>
> -#include <linux/interrupt.h>
> +#include <linux/reset.h>
>  #include <linux/spinlock.h>
> +#include <linux/interrupt.h>

Moving around the linux/interrupt.h include doesn't belong in this
patch. Otherwise this looks good to me, assuming the MCA driver does
the same reset/rearm dance.

regards
Philipp