diff mbox series

[v2,3/4] ASoC: apple: mca: Start new platform driver

Message ID 20220819125430.4920-4-povik+lin@cutebit.org (mailing list archive)
State Superseded
Headers show
Series ASoC platform driver for Apple MCA | expand

Commit Message

Martin Povišer Aug. 19, 2022, 12:54 p.m. UTC
Add ASoC platform driver for the MCA peripheral found on Apple M1 and
other chips.

Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
---
 MAINTAINERS              |    8 +
 sound/soc/Kconfig        |    1 +
 sound/soc/Makefile       |    1 +
 sound/soc/apple/Kconfig  |    9 +
 sound/soc/apple/Makefile |    3 +
 sound/soc/apple/mca.c    | 1149 ++++++++++++++++++++++++++++++++++++++
 6 files changed, 1171 insertions(+)
 create mode 100644 sound/soc/apple/Kconfig
 create mode 100644 sound/soc/apple/Makefile
 create mode 100644 sound/soc/apple/mca.c

Comments

Mark Brown Aug. 22, 2022, 5:39 p.m. UTC | #1
On Fri, Aug 19, 2022 at 02:54:29PM +0200, Martin Povišer wrote:

This all looks good, one style nit and a couple of requests for
clarification below but basically this is fine.

> +++ b/sound/soc/apple/mca.c
> @@ -0,0 +1,1149 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * Apple SoCs MCA driver
> + *
> + * Copyright (C) The Asahi Linux Contributors
> + *
> + * The MCA peripheral is made up of a number of identical units called clusters.

Please make the entire comment block a C++ one so things look more
intentional.

> +#define USE_RXB_FOR_CAPTURE

What's this all about?

> +static int mca_fe_enable_clocks(struct mca_cluster *cl)
> +{
> +	struct mca_data *mca = cl->host;
> +	int ret;
> +
> +	ret = clk_prepare_enable(cl->clk_parent);
> +	if (ret) {
> +		dev_err(mca->dev,
> +			"cluster %d: unable to enable clock parent: %d\n",
> +			cl->no, ret);
> +		return ret;
> +	}
> +
> +	/*
> +	 * We can't power up the device earlier than this because
> +	 * the power state driver would error out on seeing the device
> +	 * as clock-gated.
> +	 */
> +	cl->pd_link = device_link_add(mca->dev, cl->pd_dev,
> +				      DL_FLAG_STATELESS | DL_FLAG_PM_RUNTIME |
> +					      DL_FLAG_RPM_ACTIVE);

I'm not clear on this dynamically adding and removing device links stuff
- it looks like the main (only?) purpose is to take a runtime PM
reference to the target device which is fine but it's not clear why
device links are involved given that the links are created and destroyed
every time the DAI is used, AFAICT always in the same fixed
relationship.  It's not a problem, it's just unclear.
Martin Povišer Aug. 23, 2022, 7:33 a.m. UTC | #2
> On 22. 8. 2022, at 19:39, Mark Brown <broonie@kernel.org> wrote:
> 
> On Fri, Aug 19, 2022 at 02:54:29PM +0200, Martin Povišer wrote:
> 
> This all looks good, one style nit and a couple of requests for
> clarification below but basically this is fine.
> 
>> +++ b/sound/soc/apple/mca.c
>> @@ -0,0 +1,1149 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>> +/*
>> + * Apple SoCs MCA driver
>> + *
>> + * Copyright (C) The Asahi Linux Contributors
>> + *
>> + * The MCA peripheral is made up of a number of identical units called clusters.
> 
> Please make the entire comment block a C++ one so things look more
> intentional.

Is this so that it does not look like the SPDX header was added
mechanically? I will do it, just curious what the reasoning is.

> 
>> +#define USE_RXB_FOR_CAPTURE
> 
> What's this all about?

There’s something we can configure one way or the other way in the
hardware (choosing RXA or RXB unit in a cluster for capture). Since this 
driver developed along reverse-engineering the hardware, this switch
got built in. I want to keep it for future experimentation. Also, as
the driver’s side gig is documenting the hardware, this way it
documents more.

>> +static int mca_fe_enable_clocks(struct mca_cluster *cl)
>> +{
>> +	struct mca_data *mca = cl->host;
>> +	int ret;
>> +
>> +	ret = clk_prepare_enable(cl->clk_parent);
>> +	if (ret) {
>> +		dev_err(mca->dev,
>> +			"cluster %d: unable to enable clock parent: %d\n",
>> +			cl->no, ret);
>> +		return ret;
>> +	}
>> +
>> +	/*
>> +	 * We can't power up the device earlier than this because
>> +	 * the power state driver would error out on seeing the device
>> +	 * as clock-gated.
>> +	 */
>> +	cl->pd_link = device_link_add(mca->dev, cl->pd_dev,
>> +				      DL_FLAG_STATELESS | DL_FLAG_PM_RUNTIME |
>> +					      DL_FLAG_RPM_ACTIVE);
> 
> I'm not clear on this dynamically adding and removing device links stuff
> - it looks like the main (only?) purpose is to take a runtime PM
> reference to the target device which is fine but it's not clear why
> device links are involved given that the links are created and destroyed
> every time the DAI is used, AFAICT always in the same fixed
> relationship.  It's not a problem, it's just unclear.

Indeed the only purpose is powering up the cluster’s power domain (there’s
one domain for each cluster). Adding the links is the only way I know to
do it. They need to be added dynamically (as opposed to statically linking,
say, the DAI’s ->dev to the cluster’s ->pd_dev, which I guess may do
something similar), because we need to sequence the power-up/power-down
with the enablement of the clocks.

Martin
Mark Brown Aug. 23, 2022, 11:31 a.m. UTC | #3
On Tue, Aug 23, 2022 at 09:33:36AM +0200, Martin Povišer wrote:
> > On 22. 8. 2022, at 19:39, Mark Brown <broonie@kernel.org> wrote:
> > On Fri, Aug 19, 2022 at 02:54:29PM +0200, Martin Povišer wrote:

> >> +// SPDX-License-Identifier: GPL-2.0-only
> >> +/*
> >> + * Apple SoCs MCA driver
> >> + *
> >> + * Copyright (C) The Asahi Linux Contributors
> >> + *
> >> + * The MCA peripheral is made up of a number of identical units called clusters.

> > Please make the entire comment block a C++ one so things look more
> > intentional.

> Is this so that it does not look like the SPDX header was added
> mechanically? I will do it, just curious what the reasoning is.

Yes, broadly.

> >> +	/*
> >> +	 * We can't power up the device earlier than this because
> >> +	 * the power state driver would error out on seeing the device
> >> +	 * as clock-gated.
> >> +	 */
> >> +	cl->pd_link = device_link_add(mca->dev, cl->pd_dev,
> >> +				      DL_FLAG_STATELESS | DL_FLAG_PM_RUNTIME |
> >> +					      DL_FLAG_RPM_ACTIVE);

> > I'm not clear on this dynamically adding and removing device links stuff
> > - it looks like the main (only?) purpose is to take a runtime PM
> > reference to the target device which is fine but it's not clear why
> > device links are involved given that the links are created and destroyed
> > every time the DAI is used, AFAICT always in the same fixed
> > relationship.  It's not a problem, it's just unclear.

> Indeed the only purpose is powering up the cluster’s power domain (there’s
> one domain for each cluster). Adding the links is the only way I know to
> do it. They need to be added dynamically (as opposed to statically linking,
> say, the DAI’s ->dev to the cluster’s ->pd_dev, which I guess may do
> something similar), because we need to sequence the power-up/power-down
> with the enablement of the clocks.

You could also just do the underlying runtime power management
operations directly couldn't you?  It's not clear what the device link
stuff is adding.
Martin Povišer Aug. 23, 2022, 11:51 a.m. UTC | #4
> On 23. 8. 2022, at 13:31, Mark Brown <broonie@kernel.org> wrote:
> 
> On Tue, Aug 23, 2022 at 09:33:36AM +0200, Martin Povišer wrote:
>>> On 22. 8. 2022, at 19:39, Mark Brown <broonie@kernel.org> wrote:
>>> On Fri, Aug 19, 2022 at 02:54:29PM +0200, Martin Povišer wrote:

>>>> +	/*
>>>> +	 * We can't power up the device earlier than this because
>>>> +	 * the power state driver would error out on seeing the device
>>>> +	 * as clock-gated.
>>>> +	 */
>>>> +	cl->pd_link = device_link_add(mca->dev, cl->pd_dev,
>>>> +				      DL_FLAG_STATELESS | DL_FLAG_PM_RUNTIME |
>>>> +					      DL_FLAG_RPM_ACTIVE);
> 
>>> I'm not clear on this dynamically adding and removing device links stuff
>>> - it looks like the main (only?) purpose is to take a runtime PM
>>> reference to the target device which is fine but it's not clear why
>>> device links are involved given that the links are created and destroyed
>>> every time the DAI is used, AFAICT always in the same fixed
>>> relationship.  It's not a problem, it's just unclear.
> 
>> Indeed the only purpose is powering up the cluster’s power domain (there’s
>> one domain for each cluster). Adding the links is the only way I know to
>> do it. They need to be added dynamically (as opposed to statically linking,
>> say, the DAI’s ->dev to the cluster’s ->pd_dev, which I guess may do
>> something similar), because we need to sequence the power-up/power-down
>> with the enablement of the clocks.
> 
> You could also just do the underlying runtime power management
> operations directly couldn't you?  It's not clear what the device link
> stuff is adding.

This seems to be the way to do it. Quoting from documentation of
dev_pm_domain_attach_by_id, by which we obtain the mca->dev and cl->pd_dev
the link is between:

 * This function should typically be invoked by a driver during the probe phase,
 * in case its device requires power management through multiple PM domains. The
 * driver may benefit from using the received device, to configure device-links
 * towards its original device. Depending on the use-case and if needed, the
 * links may be dynamically changed by the driver, which allows it to control
 * the power to the PM domains independently from each other.

--
Martin
Martin Povišer Aug. 23, 2022, 11:53 a.m. UTC | #5
> On 23. 8. 2022, at 13:51, Martin Povišer <povik+lin@cutebit.org> wrote:
> 
> 
> 
>> On 23. 8. 2022, at 13:31, Mark Brown <broonie@kernel.org> wrote:
>> 
>> On Tue, Aug 23, 2022 at 09:33:36AM +0200, Martin Povišer wrote:
>>>> On 22. 8. 2022, at 19:39, Mark Brown <broonie@kernel.org> wrote:
>>>> On Fri, Aug 19, 2022 at 02:54:29PM +0200, Martin Povišer wrote:
> 
>>>>> +	/*
>>>>> +	 * We can't power up the device earlier than this because
>>>>> +	 * the power state driver would error out on seeing the device
>>>>> +	 * as clock-gated.
>>>>> +	 */
>>>>> +	cl->pd_link = device_link_add(mca->dev, cl->pd_dev,
>>>>> +				 DL_FLAG_STATELESS | DL_FLAG_PM_RUNTIME |
>>>>> +					 DL_FLAG_RPM_ACTIVE);
>> 
>>>> I'm not clear on this dynamically adding and removing device links stuff
>>>> - it looks like the main (only?) purpose is to take a runtime PM
>>>> reference to the target device which is fine but it's not clear why
>>>> device links are involved given that the links are created and destroyed
>>>> every time the DAI is used, AFAICT always in the same fixed
>>>> relationship. It's not a problem, it's just unclear.
>> 
>>> Indeed the only purpose is powering up the cluster’s power domain (there’s
>>> one domain for each cluster). Adding the links is the only way I know to
>>> do it. They need to be added dynamically (as opposed to statically linking,
>>> say, the DAI’s ->dev to the cluster’s ->pd_dev, which I guess may do
>>> something similar), because we need to sequence the power-up/power-down
>>> with the enablement of the clocks.
>> 
>> You could also just do the underlying runtime power management
>> operations directly couldn't you? It's not clear what the device link
>> stuff is adding.
> 
> This seems to be the way to do it. Quoting from documentation of
> dev_pm_domain_attach_by_id, by which we obtain the mca->dev and cl->pd_dev
> the link is between:
> 
> * This function should typically be invoked by a driver during the probe phase,
> * in case its device requires power management through multiple PM domains. The
> * driver may benefit from using the received device, to configure device-links
> * towards its original device. Depending on the use-case and if needed, the
> * links may be dynamically changed by the driver, which allows it to control
> * the power to the PM domains independently from each other.
> 
> --
> Martin

Pardon, just the cl->pd_dev is from dev_pm_domain_attach_by_id, mca->dev is
the original device.

Martin
diff mbox series

Patch

diff --git a/MAINTAINERS b/MAINTAINERS
index 8a5012ba6ff9..5f91a6b62f2f 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1899,6 +1899,14 @@  F:	include/dt-bindings/pinctrl/apple.h
 F:	include/linux/apple-mailbox.h
 F:	include/linux/soc/apple/*
 
+ARM/APPLE MACHINE SOUND DRIVERS
+M:	Martin Povišer <povik+lin@cutebit.org>
+L:	asahi@lists.linux.dev
+L:	alsa-devel@alsa-project.org (moderated for non-subscribers)
+S:	Maintained
+F:	Documentation/devicetree/bindings/sound/apple,*
+F:	drivers/sound/apple/*
+
 ARM/ARTPEC MACHINE SUPPORT
 M:	Jesper Nilsson <jesper.nilsson@axis.com>
 M:	Lars Persson <lars.persson@axis.com>
diff --git a/sound/soc/Kconfig b/sound/soc/Kconfig
index 7d4747b6bab2..848fbae26c3b 100644
--- a/sound/soc/Kconfig
+++ b/sound/soc/Kconfig
@@ -68,6 +68,7 @@  config SND_SOC_ACPI
 # All the supported SoCs
 source "sound/soc/adi/Kconfig"
 source "sound/soc/amd/Kconfig"
+source "sound/soc/apple/Kconfig"
 source "sound/soc/atmel/Kconfig"
 source "sound/soc/au1x/Kconfig"
 source "sound/soc/bcm/Kconfig"
diff --git a/sound/soc/Makefile b/sound/soc/Makefile
index 453181ef6c94..507eaed1d6a1 100644
--- a/sound/soc/Makefile
+++ b/sound/soc/Makefile
@@ -34,6 +34,7 @@  obj-$(CONFIG_SND_SOC_ACPI) += snd-soc-acpi.o
 obj-$(CONFIG_SND_SOC)	+= snd-soc-core.o
 obj-$(CONFIG_SND_SOC)	+= codecs/
 obj-$(CONFIG_SND_SOC)	+= generic/
+obj-$(CONFIG_SND_SOC)	+= apple/
 obj-$(CONFIG_SND_SOC)	+= adi/
 obj-$(CONFIG_SND_SOC)	+= amd/
 obj-$(CONFIG_SND_SOC)	+= atmel/
diff --git a/sound/soc/apple/Kconfig b/sound/soc/apple/Kconfig
new file mode 100644
index 000000000000..0ba955657e98
--- /dev/null
+++ b/sound/soc/apple/Kconfig
@@ -0,0 +1,9 @@ 
+config SND_SOC_APPLE_MCA
+	tristate "Apple Silicon MCA driver"
+	depends on ARCH_APPLE || COMPILE_TEST
+	select SND_DMAENGINE_PCM
+	select COMMON_CLK
+	default ARCH_APPLE
+	help
+	  This option enables an ASoC platform driver for MCA peripherals found
+	  on Apple Silicon SoCs.
diff --git a/sound/soc/apple/Makefile b/sound/soc/apple/Makefile
new file mode 100644
index 000000000000..7a30bf452817
--- /dev/null
+++ b/sound/soc/apple/Makefile
@@ -0,0 +1,3 @@ 
+snd-soc-apple-mca-objs	:= mca.o
+
+obj-$(CONFIG_SND_SOC_APPLE_MCA)	+= snd-soc-apple-mca.o
diff --git a/sound/soc/apple/mca.c b/sound/soc/apple/mca.c
new file mode 100644
index 000000000000..6eeb35d49986
--- /dev/null
+++ b/sound/soc/apple/mca.c
@@ -0,0 +1,1149 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Apple SoCs MCA driver
+ *
+ * Copyright (C) The Asahi Linux Contributors
+ *
+ * The MCA peripheral is made up of a number of identical units called clusters.
+ * Each cluster has its separate clock parent, SYNC signal generator, carries
+ * four SERDES units and has a dedicated I2S port on the SoC's periphery.
+ *
+ * The clusters can operate independently, or can be combined together in a
+ * configurable manner. We mostly treat them as self-contained independent
+ * units and don't configure any cross-cluster connections except for the I2S
+ * ports. The I2S ports can be routed to any of the clusters (irrespective
+ * of their native cluster). We map this onto ASoC's (DPCM) notion of backend
+ * and frontend DAIs. The 'cluster guts' are frontends which are dynamically
+ * routed to backend I2S ports.
+ *
+ * DAI references in devicetree are resolved to backends. The routing between
+ * frontends and backends is determined by the machine driver in the DAPM paths
+ * it supplies.
+ */
+
+#include <linux/bitfield.h>
+#include <linux/clk.h>
+#include <linux/dma-mapping.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_clk.h>
+#include <linux/of_dma.h>
+#include <linux/platform_device.h>
+#include <linux/pm_domain.h>
+#include <linux/regmap.h>
+#include <linux/reset.h>
+#include <linux/slab.h>
+
+#include <sound/core.h>
+#include <sound/pcm.h>
+#include <sound/pcm_params.h>
+#include <sound/soc.h>
+#include <sound/dmaengine_pcm.h>
+
+#define USE_RXB_FOR_CAPTURE
+
+/* Relative to cluster base */
+#define REG_STATUS		0x0
+#define STATUS_MCLK_EN		BIT(0)
+#define REG_MCLK_CONF		0x4
+#define MCLK_CONF_DIV		GENMASK(11, 8)
+
+#define REG_SYNCGEN_STATUS	0x100
+#define SYNCGEN_STATUS_EN	BIT(0)
+#define REG_SYNCGEN_MCLK_SEL	0x104
+#define SYNCGEN_MCLK_SEL	GENMASK(3, 0)
+#define REG_SYNCGEN_HI_PERIOD	0x108
+#define REG_SYNCGEN_LO_PERIOD	0x10c
+
+#define REG_PORT_ENABLES	0x600
+#define PORT_ENABLES_CLOCKS	GENMASK(2, 1)
+#define PORT_ENABLES_TX_DATA	BIT(3)
+#define REG_PORT_CLOCK_SEL	0x604
+#define PORT_CLOCK_SEL		GENMASK(11, 8)
+#define REG_PORT_DATA_SEL	0x608
+#define PORT_DATA_SEL_TXA(cl)	(1 << ((cl)*2))
+#define PORT_DATA_SEL_TXB(cl)	(2 << ((cl)*2))
+
+#define REG_INTSTATE		0x700
+#define REG_INTMASK		0x704
+
+/* Bases of serdes units (relative to cluster) */
+#define CLUSTER_RXA_OFF	0x200
+#define CLUSTER_TXA_OFF	0x300
+#define CLUSTER_RXB_OFF	0x400
+#define CLUSTER_TXB_OFF	0x500
+
+#define CLUSTER_TX_OFF	CLUSTER_TXA_OFF
+
+#ifndef USE_RXB_FOR_CAPTURE
+#define CLUSTER_RX_OFF	CLUSTER_RXA_OFF
+#else
+#define CLUSTER_RX_OFF	CLUSTER_RXB_OFF
+#endif
+
+/* Relative to serdes unit base */
+#define REG_SERDES_STATUS	0x00
+#define SERDES_STATUS_EN	BIT(0)
+#define SERDES_STATUS_RST	BIT(1)
+#define REG_TX_SERDES_CONF	0x04
+#define REG_RX_SERDES_CONF	0x08
+#define SERDES_CONF_NCHANS	GENMASK(3, 0)
+#define SERDES_CONF_WIDTH_MASK	GENMASK(8, 4)
+#define SERDES_CONF_WIDTH_16BIT 0x40
+#define SERDES_CONF_WIDTH_20BIT 0x80
+#define SERDES_CONF_WIDTH_24BIT 0xc0
+#define SERDES_CONF_WIDTH_32BIT 0x100
+#define SERDES_CONF_BCLK_POL	0x400
+#define SERDES_CONF_LSB_FIRST	0x800
+#define SERDES_CONF_UNK1	BIT(12)
+#define SERDES_CONF_UNK2	BIT(13)
+#define SERDES_CONF_UNK3	BIT(14)
+#define SERDES_CONF_NO_DATA_FEEDBACK	BIT(15)
+#define SERDES_CONF_SYNC_SEL	GENMASK(18, 16)
+#define SERDES_CONF_SOME_RST	BIT(19)
+#define REG_TX_SERDES_BITSTART	0x08
+#define REG_RX_SERDES_BITSTART	0x0c
+#define REG_TX_SERDES_SLOTMASK	0x0c
+#define REG_RX_SERDES_SLOTMASK	0x10
+#define REG_RX_SERDES_PORT	0x04
+
+/* Relative to switch base */
+#define REG_DMA_ADAPTER_A(cl)	(0x8000 * (cl))
+#define REG_DMA_ADAPTER_B(cl)	(0x8000 * (cl) + 0x4000)
+#define DMA_ADAPTER_TX_LSB_PAD	GENMASK(4, 0)
+#define DMA_ADAPTER_TX_NCHANS	GENMASK(6, 5)
+#define DMA_ADAPTER_RX_MSB_PAD	GENMASK(12, 8)
+#define DMA_ADAPTER_RX_NCHANS	GENMASK(14, 13)
+#define DMA_ADAPTER_NCHANS	GENMASK(22, 20)
+
+#define SWITCH_STRIDE	0x8000
+#define CLUSTER_STRIDE	0x4000
+
+#define MAX_NCLUSTERS	6
+
+#define APPLE_MCA_FMTBITS (SNDRV_PCM_FMTBIT_S16_LE | \
+			   SNDRV_PCM_FMTBIT_S24_LE | \
+			   SNDRV_PCM_FMTBIT_S32_LE)
+
+struct mca_cluster {
+	int no;
+	__iomem void *base;
+	struct mca_data *host;
+	struct device *pd_dev;
+	struct clk *clk_parent;
+	struct dma_chan *dma_chans[SNDRV_PCM_STREAM_LAST + 1];
+
+	bool port_started[SNDRV_PCM_STREAM_LAST + 1];
+	int port_driver; /* The cluster driving this cluster's port */
+
+	bool clocks_in_use[SNDRV_PCM_STREAM_LAST + 1];
+	struct device_link *pd_link;
+
+	unsigned int bclk_ratio;
+
+	/* Masks etc. picked up via the set_tdm_slot method */
+	int tdm_slots;
+	int tdm_slot_width;
+	unsigned int tdm_tx_mask;
+	unsigned int tdm_rx_mask;
+};
+
+struct mca_data {
+	struct device *dev;
+
+	__iomem void *switch_base;
+
+	struct device *pd_dev;
+	struct reset_control *rstc;
+	struct device_link *pd_link;
+
+	int nclusters;
+	struct mca_cluster clusters[];
+};
+
+static void mca_modify(struct mca_cluster *cl, int regoffset, u32 mask, u32 val)
+{
+	__iomem void *ptr = cl->base + regoffset;
+	u32 newval;
+
+	newval = (val & mask) | (readl_relaxed(ptr) & ~mask);
+	writel_relaxed(newval, ptr);
+}
+
+/*
+ * Get the cluster of FE or BE DAI
+ */
+static struct mca_cluster *mca_dai_to_cluster(struct snd_soc_dai *dai)
+{
+	struct mca_data *mca = snd_soc_dai_get_drvdata(dai);
+	/*
+	 * FE DAIs are         0 ... nclusters - 1
+	 * BE DAIs are nclusters ... 2*nclusters - 1
+	 */
+	int cluster_no = dai->id % mca->nclusters;
+
+	return &mca->clusters[cluster_no];
+}
+
+/* called before PCM trigger */
+static void mca_fe_early_trigger(struct snd_pcm_substream *substream, int cmd,
+				 struct snd_soc_dai *dai)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+	bool is_tx = substream->stream == SNDRV_PCM_STREAM_PLAYBACK;
+	int serdes_unit = is_tx ? CLUSTER_TX_OFF : CLUSTER_RX_OFF;
+	int serdes_conf =
+		serdes_unit + (is_tx ? REG_TX_SERDES_CONF : REG_RX_SERDES_CONF);
+
+	switch (cmd) {
+	case SNDRV_PCM_TRIGGER_START:
+	case SNDRV_PCM_TRIGGER_RESUME:
+	case SNDRV_PCM_TRIGGER_PAUSE_RELEASE:
+		mca_modify(cl, serdes_unit + REG_SERDES_STATUS,
+			   SERDES_STATUS_EN | SERDES_STATUS_RST,
+			   SERDES_STATUS_RST);
+		mca_modify(cl, serdes_conf, SERDES_CONF_SOME_RST,
+			   SERDES_CONF_SOME_RST);
+		readl_relaxed(cl->base + serdes_conf);
+		mca_modify(cl, serdes_conf, SERDES_STATUS_RST, 0);
+		WARN_ON(readl_relaxed(cl->base + REG_SERDES_STATUS) &
+			SERDES_STATUS_RST);
+		break;
+	default:
+		break;
+	}
+}
+
+static int mca_fe_trigger(struct snd_pcm_substream *substream, int cmd,
+			  struct snd_soc_dai *dai)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+	bool is_tx = substream->stream == SNDRV_PCM_STREAM_PLAYBACK;
+	int serdes_unit = is_tx ? CLUSTER_TX_OFF : CLUSTER_RX_OFF;
+
+	switch (cmd) {
+	case SNDRV_PCM_TRIGGER_START:
+	case SNDRV_PCM_TRIGGER_RESUME:
+	case SNDRV_PCM_TRIGGER_PAUSE_RELEASE:
+		mca_modify(cl, serdes_unit + REG_SERDES_STATUS,
+			   SERDES_STATUS_EN | SERDES_STATUS_RST,
+			   SERDES_STATUS_EN);
+		break;
+
+	case SNDRV_PCM_TRIGGER_STOP:
+	case SNDRV_PCM_TRIGGER_SUSPEND:
+	case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
+		mca_modify(cl, serdes_unit + REG_SERDES_STATUS,
+			   SERDES_STATUS_EN, 0);
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int mca_fe_enable_clocks(struct mca_cluster *cl)
+{
+	struct mca_data *mca = cl->host;
+	int ret;
+
+	ret = clk_prepare_enable(cl->clk_parent);
+	if (ret) {
+		dev_err(mca->dev,
+			"cluster %d: unable to enable clock parent: %d\n",
+			cl->no, ret);
+		return ret;
+	}
+
+	/*
+	 * We can't power up the device earlier than this because
+	 * the power state driver would error out on seeing the device
+	 * as clock-gated.
+	 */
+	cl->pd_link = device_link_add(mca->dev, cl->pd_dev,
+				      DL_FLAG_STATELESS | DL_FLAG_PM_RUNTIME |
+					      DL_FLAG_RPM_ACTIVE);
+	if (!cl->pd_link) {
+		dev_err(mca->dev,
+			"cluster %d: unable to prop-up power domain\n", cl->no);
+		clk_disable_unprepare(cl->clk_parent);
+		return -EINVAL;
+	}
+
+	writel_relaxed(cl->no + 1, cl->base + REG_SYNCGEN_MCLK_SEL);
+	mca_modify(cl, REG_SYNCGEN_STATUS, SYNCGEN_STATUS_EN,
+		   SYNCGEN_STATUS_EN);
+	mca_modify(cl, REG_STATUS, STATUS_MCLK_EN, STATUS_MCLK_EN);
+
+	return 0;
+}
+
+static void mca_fe_disable_clocks(struct mca_cluster *cl)
+{
+	mca_modify(cl, REG_SYNCGEN_STATUS, SYNCGEN_STATUS_EN, 0);
+	mca_modify(cl, REG_STATUS, STATUS_MCLK_EN, 0);
+
+	device_link_del(cl->pd_link);
+	clk_disable_unprepare(cl->clk_parent);
+}
+
+static bool mca_fe_clocks_in_use(struct mca_cluster *cl)
+{
+	struct mca_data *mca = cl->host;
+	struct mca_cluster *be_cl;
+	int stream, i;
+
+	for (i = 0; i < mca->nclusters; i++) {
+		be_cl = &mca->clusters[i];
+
+		if (be_cl->port_driver != cl->no)
+			continue;
+
+		for_each_pcm_streams(stream)
+			if (be_cl->clocks_in_use[stream])
+				return true;
+	}
+	return false;
+}
+
+static int mca_be_prepare(struct snd_pcm_substream *substream,
+			  struct snd_soc_dai *dai)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+	struct mca_data *mca = cl->host;
+	struct mca_cluster *fe_cl;
+	int ret;
+
+	if (cl->port_driver < 0)
+		return -EINVAL;
+
+	fe_cl = &mca->clusters[cl->port_driver];
+
+	/*
+	 * Typically the CODECs we are paired with will require clocks
+	 * to be present at time of unmute with the 'mute_stream' op
+	 * or at time of DAPM widget power-up. We need to enable clocks
+	 * here at the latest (frontend prepare would be too late).
+	 */
+	if (!mca_fe_clocks_in_use(fe_cl)) {
+		ret = mca_fe_enable_clocks(fe_cl);
+		if (ret < 0)
+			return ret;
+	}
+
+	cl->clocks_in_use[substream->stream] = true;
+
+	return 0;
+}
+
+static int mca_be_hw_free(struct snd_pcm_substream *substream,
+			  struct snd_soc_dai *dai)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+	struct mca_data *mca = cl->host;
+	struct mca_cluster *fe_cl;
+
+	if (cl->port_driver < 0)
+		return -EINVAL;
+
+	fe_cl = &mca->clusters[cl->port_driver];
+	if (!mca_fe_clocks_in_use(fe_cl))
+		return 0; /* Nothing to do */
+
+	cl->clocks_in_use[substream->stream] = false;
+
+	if (!mca_fe_clocks_in_use(fe_cl))
+		mca_fe_disable_clocks(fe_cl);
+
+	return 0;
+}
+
+static unsigned int mca_crop_mask(unsigned int mask, int nchans)
+{
+	while (hweight32(mask) > nchans)
+		mask &= ~(1 << __fls(mask));
+
+	return mask;
+}
+
+static int mca_configure_serdes(struct mca_cluster *cl, int serdes_unit,
+				unsigned int mask, int slots, int nchans,
+				int slot_width, bool is_tx, int port)
+{
+	__iomem void *serdes_base = cl->base + serdes_unit;
+	u32 serdes_conf, serdes_conf_mask;
+
+	serdes_conf_mask = SERDES_CONF_WIDTH_MASK | SERDES_CONF_NCHANS;
+	serdes_conf = FIELD_PREP(SERDES_CONF_NCHANS, max(slots, 1) - 1);
+	switch (slot_width) {
+	case 16:
+		serdes_conf |= SERDES_CONF_WIDTH_16BIT;
+		break;
+	case 20:
+		serdes_conf |= SERDES_CONF_WIDTH_20BIT;
+		break;
+	case 24:
+		serdes_conf |= SERDES_CONF_WIDTH_24BIT;
+		break;
+	case 32:
+		serdes_conf |= SERDES_CONF_WIDTH_32BIT;
+		break;
+	default:
+		goto err;
+	}
+
+	serdes_conf_mask |= SERDES_CONF_SYNC_SEL;
+	serdes_conf |= FIELD_PREP(SERDES_CONF_SYNC_SEL, cl->no + 1);
+
+	if (is_tx) {
+		serdes_conf_mask |= SERDES_CONF_UNK1 | SERDES_CONF_UNK2 |
+				    SERDES_CONF_UNK3;
+		serdes_conf |= SERDES_CONF_UNK1 | SERDES_CONF_UNK2 |
+			       SERDES_CONF_UNK3;
+	} else {
+		serdes_conf_mask |= SERDES_CONF_UNK1 | SERDES_CONF_UNK2 |
+				    SERDES_CONF_UNK3 |
+				    SERDES_CONF_NO_DATA_FEEDBACK;
+		serdes_conf |= SERDES_CONF_UNK1 | SERDES_CONF_UNK2 |
+			       SERDES_CONF_NO_DATA_FEEDBACK;
+	}
+
+	mca_modify(cl,
+		   serdes_unit +
+			   (is_tx ? REG_TX_SERDES_CONF : REG_RX_SERDES_CONF),
+		   serdes_conf_mask, serdes_conf);
+
+	if (is_tx) {
+		writel_relaxed(0xffffffff,
+			       serdes_base + REG_TX_SERDES_SLOTMASK);
+		writel_relaxed(~((u32)mca_crop_mask(mask, nchans)),
+			       serdes_base + REG_TX_SERDES_SLOTMASK + 0x4);
+		writel_relaxed(0xffffffff,
+			       serdes_base + REG_TX_SERDES_SLOTMASK + 0x8);
+		writel_relaxed(~((u32)mask),
+			       serdes_base + REG_TX_SERDES_SLOTMASK + 0xc);
+	} else {
+		writel_relaxed(0xffffffff,
+			       serdes_base + REG_RX_SERDES_SLOTMASK);
+		writel_relaxed(~((u32)mca_crop_mask(mask, nchans)),
+			       serdes_base + REG_RX_SERDES_SLOTMASK + 0x4);
+		writel_relaxed(1 << port,
+			       serdes_base + REG_RX_SERDES_PORT);
+	}
+
+	return 0;
+
+err:
+	dev_err(cl->host->dev,
+		"unsupported SERDES configuration requested (mask=0x%x slots=%d slot_width=%d)\n",
+		mask, slots, slot_width);
+	return -EINVAL;
+}
+
+static int mca_fe_set_tdm_slot(struct snd_soc_dai *dai, unsigned int tx_mask,
+			       unsigned int rx_mask, int slots, int slot_width)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+
+	cl->tdm_slots = slots;
+	cl->tdm_slot_width = slot_width;
+	cl->tdm_tx_mask = tx_mask;
+	cl->tdm_rx_mask = rx_mask;
+
+	return 0;
+}
+
+static int mca_fe_set_fmt(struct snd_soc_dai *dai, unsigned int fmt)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+	struct mca_data *mca = cl->host;
+	bool fpol_inv = false;
+	u32 serdes_conf = 0;
+	u32 bitstart;
+
+	if ((fmt & SND_SOC_DAIFMT_CLOCK_PROVIDER_MASK) !=
+	    SND_SOC_DAIFMT_BP_FP)
+		goto err;
+
+	switch (fmt & SND_SOC_DAIFMT_FORMAT_MASK) {
+	case SND_SOC_DAIFMT_I2S:
+		fpol_inv = 0;
+		bitstart = 1;
+		break;
+	case SND_SOC_DAIFMT_LEFT_J:
+		fpol_inv = 1;
+		bitstart = 0;
+		break;
+	default:
+		goto err;
+	}
+
+	switch (fmt & SND_SOC_DAIFMT_INV_MASK) {
+	case SND_SOC_DAIFMT_NB_IF:
+	case SND_SOC_DAIFMT_IB_IF:
+		fpol_inv ^= 1;
+		break;
+	}
+
+	switch (fmt & SND_SOC_DAIFMT_INV_MASK) {
+	case SND_SOC_DAIFMT_NB_NF:
+	case SND_SOC_DAIFMT_NB_IF:
+		serdes_conf |= SERDES_CONF_BCLK_POL;
+		break;
+	}
+
+	if (!fpol_inv)
+		goto err;
+
+	mca_modify(cl, CLUSTER_TX_OFF + REG_TX_SERDES_CONF,
+		   SERDES_CONF_BCLK_POL, serdes_conf);
+	mca_modify(cl, CLUSTER_RX_OFF + REG_RX_SERDES_CONF,
+		   SERDES_CONF_BCLK_POL, serdes_conf);
+	writel_relaxed(bitstart,
+		       cl->base + CLUSTER_TX_OFF + REG_TX_SERDES_BITSTART);
+	writel_relaxed(bitstart,
+		       cl->base + CLUSTER_RX_OFF + REG_RX_SERDES_BITSTART);
+
+	return 0;
+
+err:
+	dev_err(mca->dev, "unsupported DAI format (0x%x) requested\n", fmt);
+	return -EINVAL;
+}
+
+static int mca_set_bclk_ratio(struct snd_soc_dai *dai, unsigned int ratio)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+
+	cl->bclk_ratio = ratio;
+
+	return 0;
+}
+
+static int mca_fe_get_port(struct snd_pcm_substream *substream)
+{
+	struct snd_soc_pcm_runtime *fe = asoc_substream_to_rtd(substream);
+	struct snd_soc_pcm_runtime *be;
+	struct snd_soc_dpcm *dpcm;
+
+	be = NULL;
+	for_each_dpcm_be(fe, substream->stream, dpcm) {
+		be = dpcm->be;
+		break;
+	}
+
+	if (!be)
+		return -EINVAL;
+
+	return mca_dai_to_cluster(asoc_rtd_to_cpu(be, 0))->no;
+}
+
+static int mca_fe_hw_params(struct snd_pcm_substream *substream,
+			    struct snd_pcm_hw_params *params,
+			    struct snd_soc_dai *dai)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+	struct mca_data *mca = cl->host;
+	struct device *dev = mca->dev;
+	unsigned int samp_rate = params_rate(params);
+	bool is_tx = substream->stream == SNDRV_PCM_STREAM_PLAYBACK;
+	bool refine_tdm = false;
+	unsigned long bclk_ratio;
+	unsigned int tdm_slots, tdm_slot_width, tdm_mask;
+	u32 regval, pad;
+	int ret, port, nchans_ceiled;
+
+	if (!cl->tdm_slot_width) {
+		/*
+		 * We were not given TDM settings from above, set initial
+		 * guesses which will later be refined.
+		 */
+		tdm_slot_width = params_width(params);
+		tdm_slots = params_channels(params);
+		refine_tdm = true;
+	} else {
+		tdm_slot_width = cl->tdm_slot_width;
+		tdm_slots = cl->tdm_slots;
+		tdm_mask = is_tx ? cl->tdm_tx_mask : cl->tdm_rx_mask;
+	}
+
+	if (cl->bclk_ratio)
+		bclk_ratio = cl->bclk_ratio;
+	else
+		bclk_ratio = tdm_slot_width * tdm_slots;
+
+	if (refine_tdm) {
+		int nchannels = params_channels(params);
+
+		if (nchannels > 2) {
+			dev_err(dev, "missing TDM for stream with two or more channels\n");
+			return -EINVAL;
+		}
+
+		if ((bclk_ratio % nchannels) != 0) {
+			dev_err(dev, "BCLK ratio (%ld) not divisible by no. of channels (%d)\n",
+				bclk_ratio, nchannels);
+			return -EINVAL;
+		}
+
+		tdm_slot_width = bclk_ratio / nchannels;
+
+		if (tdm_slot_width > 32 && nchannels == 1)
+			tdm_slot_width = 32;
+
+		if (tdm_slot_width < params_width(params)) {
+			dev_err(dev, "TDM slots too narrow (tdm=%d params=%d)\n",
+				tdm_slot_width, params_width(params));
+			return -EINVAL;
+		}
+
+		tdm_mask = (1 << tdm_slots) - 1;
+	}
+
+	port = mca_fe_get_port(substream);
+	if (port < 0)
+		return port;
+
+	ret = mca_configure_serdes(cl, is_tx ? CLUSTER_TX_OFF : CLUSTER_RX_OFF,
+				   tdm_mask, tdm_slots, params_channels(params),
+				   tdm_slot_width, is_tx, port);
+	if (ret)
+		return ret;
+
+	pad = 32 - params_width(params);
+
+	/*
+	 * TODO: Here the register semantics aren't clear.
+	 */
+	nchans_ceiled = min_t(int, params_channels(params), 4);
+	regval = FIELD_PREP(DMA_ADAPTER_NCHANS, nchans_ceiled) |
+		 FIELD_PREP(DMA_ADAPTER_TX_NCHANS, 0x2) |
+		 FIELD_PREP(DMA_ADAPTER_RX_NCHANS, 0x2) |
+		 FIELD_PREP(DMA_ADAPTER_TX_LSB_PAD, pad) |
+		 FIELD_PREP(DMA_ADAPTER_RX_MSB_PAD, pad);
+
+#ifndef USE_RXB_FOR_CAPTURE
+	writel_relaxed(regval, mca->switch_base + REG_DMA_ADAPTER_A(cl->no));
+#else
+	if (is_tx)
+		writel_relaxed(regval,
+			       mca->switch_base + REG_DMA_ADAPTER_A(cl->no));
+	else
+		writel_relaxed(regval,
+			       mca->switch_base + REG_DMA_ADAPTER_B(cl->no));
+#endif
+
+	if (!mca_fe_clocks_in_use(cl)) {
+		/*
+		 * Set up FSYNC duty cycle as even as possible.
+		 */
+		writel_relaxed((bclk_ratio / 2) - 1,
+			       cl->base + REG_SYNCGEN_HI_PERIOD);
+		writel_relaxed(((bclk_ratio + 1) / 2) - 1,
+			       cl->base + REG_SYNCGEN_LO_PERIOD);
+		writel_relaxed(FIELD_PREP(MCLK_CONF_DIV, 0x1),
+			       cl->base + REG_MCLK_CONF);
+
+		ret = clk_set_rate(cl->clk_parent, bclk_ratio * samp_rate);
+		if (ret) {
+			dev_err(mca->dev, "cluster %d: unable to set clock parent: %d\n",
+				cl->no, ret);
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
+static const struct snd_soc_dai_ops mca_fe_ops = {
+	.set_fmt = mca_fe_set_fmt,
+	.set_bclk_ratio = mca_set_bclk_ratio,
+	.set_tdm_slot = mca_fe_set_tdm_slot,
+	.hw_params = mca_fe_hw_params,
+	.trigger = mca_fe_trigger,
+};
+
+static bool mca_be_started(struct mca_cluster *cl)
+{
+	int stream;
+
+	for_each_pcm_streams(stream)
+		if (cl->port_started[stream])
+			return true;
+	return false;
+}
+
+static int mca_be_startup(struct snd_pcm_substream *substream,
+			  struct snd_soc_dai *dai)
+{
+	struct snd_soc_pcm_runtime *be = asoc_substream_to_rtd(substream);
+	struct snd_soc_pcm_runtime *fe;
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+	struct mca_cluster *fe_cl;
+	struct mca_data *mca = cl->host;
+	struct snd_soc_dpcm *dpcm;
+
+	fe = NULL;
+
+	for_each_dpcm_fe(be, substream->stream, dpcm) {
+		if (fe && dpcm->fe != fe) {
+			dev_err(mca->dev, "many FE per one BE unsupported\n");
+			return -EINVAL;
+		}
+
+		fe = dpcm->fe;
+	}
+
+	if (!fe)
+		return -EINVAL;
+
+	fe_cl = mca_dai_to_cluster(asoc_rtd_to_cpu(fe, 0));
+
+	if (mca_be_started(cl)) {
+		/*
+		 * Port is already started in the other direction.
+		 * Make sure there isn't a conflict with another cluster
+		 * driving the port.
+		 */
+		if (cl->port_driver != fe_cl->no)
+			return -EINVAL;
+
+		cl->port_started[substream->stream] = true;
+		return 0;
+	}
+
+	writel_relaxed(PORT_ENABLES_CLOCKS | PORT_ENABLES_TX_DATA,
+		       cl->base + REG_PORT_ENABLES);
+	writel_relaxed(FIELD_PREP(PORT_CLOCK_SEL, fe_cl->no + 1),
+		       cl->base + REG_PORT_CLOCK_SEL);
+	writel_relaxed(PORT_DATA_SEL_TXA(fe_cl->no),
+		       cl->base + REG_PORT_DATA_SEL);
+	cl->port_driver = fe_cl->no;
+	cl->port_started[substream->stream] = true;
+
+	return 0;
+}
+
+static void mca_be_shutdown(struct snd_pcm_substream *substream,
+			    struct snd_soc_dai *dai)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(dai);
+
+	cl->port_started[substream->stream] = false;
+
+	if (!mca_be_started(cl)) {
+		/*
+		 * Were we the last direction to shutdown?
+		 * Turn off the lights.
+		 */
+		writel_relaxed(0, cl->base + REG_PORT_ENABLES);
+		writel_relaxed(0, cl->base + REG_PORT_DATA_SEL);
+		cl->port_driver = -1;
+	}
+}
+
+static const struct snd_soc_dai_ops mca_be_ops = {
+	.prepare = mca_be_prepare,
+	.hw_free = mca_be_hw_free,
+	.startup = mca_be_startup,
+	.shutdown = mca_be_shutdown,
+};
+
+static int mca_set_runtime_hwparams(struct snd_soc_component *component,
+				    struct snd_pcm_substream *substream,
+				    struct dma_chan *chan)
+{
+	struct device *dma_dev = chan->device->dev;
+	struct snd_dmaengine_dai_dma_data dma_data = {};
+	int ret;
+
+	struct snd_pcm_hardware hw;
+
+	memset(&hw, 0, sizeof(hw));
+
+	hw.info = SNDRV_PCM_INFO_MMAP | SNDRV_PCM_INFO_MMAP_VALID |
+		  SNDRV_PCM_INFO_INTERLEAVED;
+	hw.periods_min = 2;
+	hw.periods_max = UINT_MAX;
+	hw.period_bytes_min = 256;
+	hw.period_bytes_max = dma_get_max_seg_size(dma_dev);
+	hw.buffer_bytes_max = SIZE_MAX;
+	hw.fifo_size = 16;
+
+	ret = snd_dmaengine_pcm_refine_runtime_hwparams(substream, &dma_data,
+							&hw, chan);
+
+	if (ret)
+		return ret;
+
+	return snd_soc_set_runtime_hwparams(substream, &hw);
+}
+
+static int mca_pcm_open(struct snd_soc_component *component,
+			struct snd_pcm_substream *substream)
+{
+	struct snd_soc_pcm_runtime *rtd = asoc_substream_to_rtd(substream);
+	struct mca_cluster *cl = mca_dai_to_cluster(asoc_rtd_to_cpu(rtd, 0));
+	struct dma_chan *chan = cl->dma_chans[substream->stream];
+	int ret;
+
+	if (rtd->dai_link->no_pcm)
+		return 0;
+
+	ret = mca_set_runtime_hwparams(component, substream, chan);
+	if (ret)
+		return ret;
+
+	return snd_dmaengine_pcm_open(substream, chan);
+}
+
+static int mca_hw_params(struct snd_soc_component *component,
+			 struct snd_pcm_substream *substream,
+			 struct snd_pcm_hw_params *params)
+{
+	struct snd_soc_pcm_runtime *rtd = asoc_substream_to_rtd(substream);
+	struct dma_chan *chan = snd_dmaengine_pcm_get_chan(substream);
+	struct dma_slave_config slave_config;
+	int ret;
+
+	if (rtd->dai_link->no_pcm)
+		return 0;
+
+	memset(&slave_config, 0, sizeof(slave_config));
+	ret = snd_hwparams_to_dma_slave_config(substream, params,
+					       &slave_config);
+	if (ret < 0)
+		return ret;
+
+	if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK)
+		slave_config.dst_port_window_size =
+			min_t(u32, params_channels(params), 4);
+	else
+		slave_config.src_port_window_size =
+			min_t(u32, params_channels(params), 4);
+
+	return dmaengine_slave_config(chan, &slave_config);
+}
+
+static int mca_close(struct snd_soc_component *component,
+		     struct snd_pcm_substream *substream)
+{
+	struct snd_soc_pcm_runtime *rtd = asoc_substream_to_rtd(substream);
+
+	if (rtd->dai_link->no_pcm)
+		return 0;
+
+	return snd_dmaengine_pcm_close(substream);
+}
+
+static int mca_trigger(struct snd_soc_component *component,
+		       struct snd_pcm_substream *substream, int cmd)
+{
+	struct snd_soc_pcm_runtime *rtd = asoc_substream_to_rtd(substream);
+
+	if (rtd->dai_link->no_pcm)
+		return 0;
+
+	/*
+	 * Before we do the PCM trigger proper, insert an opportunity
+	 * to reset the frontend's SERDES.
+	 */
+	mca_fe_early_trigger(substream, cmd, asoc_rtd_to_cpu(rtd, 0));
+
+	return snd_dmaengine_pcm_trigger(substream, cmd);
+}
+
+static snd_pcm_uframes_t mca_pointer(struct snd_soc_component *component,
+				     struct snd_pcm_substream *substream)
+{
+	struct snd_soc_pcm_runtime *rtd = asoc_substream_to_rtd(substream);
+
+	if (rtd->dai_link->no_pcm)
+		return -ENOTSUPP;
+
+	return snd_dmaengine_pcm_pointer(substream);
+}
+
+static int mca_pcm_new(struct snd_soc_component *component,
+		       struct snd_soc_pcm_runtime *rtd)
+{
+	struct mca_cluster *cl = mca_dai_to_cluster(asoc_rtd_to_cpu(rtd, 0));
+	unsigned int i;
+
+	if (rtd->dai_link->no_pcm)
+		return 0;
+
+	for_each_pcm_streams(i) {
+		struct snd_pcm_substream *substream =
+			rtd->pcm->streams[i].substream;
+		struct dma_chan *chan = cl->dma_chans[i];
+
+		if (!substream)
+			continue;
+
+		if (!chan) {
+			dev_err(component->dev, "missing DMA channel for stream %d on SERDES %d\n",
+				i, cl->no);
+			return -EINVAL;
+		}
+
+		snd_pcm_set_managed_buffer(substream, SNDRV_DMA_TYPE_DEV_IRAM,
+					   chan->device->dev, 512 * 1024 * 6,
+					   SIZE_MAX);
+	}
+
+	return 0;
+}
+
+static const struct snd_soc_component_driver mca_component = {
+	.name = "apple-mca",
+	.open = mca_pcm_open,
+	.close = mca_close,
+	.hw_params = mca_hw_params,
+	.trigger = mca_trigger,
+	.pointer = mca_pointer,
+	.pcm_construct = mca_pcm_new,
+};
+
+static void apple_mca_release(struct mca_data *mca)
+{
+	int i, stream;
+
+	for (i = 0; i < mca->nclusters; i++) {
+		struct mca_cluster *cl = &mca->clusters[i];
+
+		for_each_pcm_streams(stream) {
+			if (IS_ERR_OR_NULL(cl->dma_chans[stream]))
+				continue;
+
+			dma_release_channel(cl->dma_chans[stream]);
+		}
+
+		if (!IS_ERR_OR_NULL(cl->clk_parent))
+			clk_put(cl->clk_parent);
+
+		if (!IS_ERR_OR_NULL(cl->pd_dev))
+			dev_pm_domain_detach(cl->pd_dev, true);
+	}
+
+	if (mca->pd_link)
+		device_link_del(mca->pd_link);
+
+	if (!IS_ERR_OR_NULL(mca->pd_dev))
+		dev_pm_domain_detach(mca->pd_dev, true);
+
+	reset_control_assert(mca->rstc);
+}
+
+static int apple_mca_probe(struct platform_device *pdev)
+{
+	struct mca_data *mca;
+	struct mca_cluster *clusters;
+	struct snd_soc_dai_driver *dai_drivers;
+	struct resource *res;
+	void __iomem *base;
+	int nclusters;
+	int ret, i;
+
+	base = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
+	if (IS_ERR(base))
+		return PTR_ERR(base);
+
+	if (resource_size(res) < CLUSTER_STRIDE)
+		return -EINVAL;
+	nclusters = (resource_size(res) - CLUSTER_STRIDE) / CLUSTER_STRIDE + 1;
+
+	mca = devm_kzalloc(&pdev->dev, struct_size(mca, clusters, nclusters),
+			   GFP_KERNEL);
+	if (!mca)
+		return -ENOMEM;
+	mca->dev = &pdev->dev;
+	mca->nclusters = nclusters;
+	platform_set_drvdata(pdev, mca);
+	clusters = mca->clusters;
+
+	mca->switch_base =
+		devm_platform_ioremap_resource(pdev, 1);
+	if (IS_ERR(mca->switch_base))
+		return PTR_ERR(mca->switch_base);
+
+	mca->rstc = devm_reset_control_get_optional_shared(&pdev->dev, NULL);
+	if (IS_ERR(mca->rstc))
+		return PTR_ERR(mca->rstc);
+
+	dai_drivers = devm_kzalloc(
+		&pdev->dev, sizeof(*dai_drivers) * 2 * nclusters, GFP_KERNEL);
+	if (!dai_drivers)
+		return -ENOMEM;
+
+	mca->pd_dev = dev_pm_domain_attach_by_id(&pdev->dev, 0);
+	if (IS_ERR(mca->pd_dev))
+		return -EINVAL;
+
+	mca->pd_link = device_link_add(&pdev->dev, mca->pd_dev,
+				       DL_FLAG_STATELESS | DL_FLAG_PM_RUNTIME |
+					       DL_FLAG_RPM_ACTIVE);
+	if (!mca->pd_link) {
+		ret = -EINVAL;
+		/* Prevent an unbalanced reset assert */
+		mca->rstc = NULL;
+		goto err_release;
+	}
+
+	reset_control_deassert(mca->rstc);
+
+	for (i = 0; i < nclusters; i++) {
+		struct mca_cluster *cl = &clusters[i];
+		struct snd_soc_dai_driver *fe =
+			&dai_drivers[mca->nclusters + i];
+		struct snd_soc_dai_driver *be = &dai_drivers[i];
+		int stream;
+
+		cl->host = mca;
+		cl->no = i;
+		cl->base = base + CLUSTER_STRIDE * i;
+		cl->port_driver = -1;
+		cl->clk_parent = of_clk_get(pdev->dev.of_node, i);
+		if (IS_ERR(cl->clk_parent)) {
+			dev_err(&pdev->dev, "unable to obtain clock %d: %ld\n",
+				i, PTR_ERR(cl->clk_parent));
+			ret = PTR_ERR(cl->clk_parent);
+			goto err_release;
+		}
+		cl->pd_dev = dev_pm_domain_attach_by_id(&pdev->dev, i + 1);
+		if (IS_ERR(cl->pd_dev)) {
+			dev_err(&pdev->dev,
+				"unable to obtain cluster %d PD: %ld\n", i,
+				PTR_ERR(cl->pd_dev));
+			ret = PTR_ERR(cl->pd_dev);
+			goto err_release;
+		}
+
+		for_each_pcm_streams(stream) {
+			struct dma_chan *chan;
+			bool is_tx = (stream == SNDRV_PCM_STREAM_PLAYBACK);
+#ifndef USE_RXB_FOR_CAPTURE
+			char *name = devm_kasprintf(&pdev->dev, GFP_KERNEL,
+						    is_tx ? "tx%da" : "rx%da",
+						    i);
+#else
+			char *name = devm_kasprintf(&pdev->dev, GFP_KERNEL,
+						    is_tx ? "tx%da" : "rx%db",
+						    i);
+#endif
+
+			chan = of_dma_request_slave_channel(pdev->dev.of_node,
+							    name);
+			if (IS_ERR(chan)) {
+				if (PTR_ERR(chan) != -EPROBE_DEFER)
+					dev_err(&pdev->dev,
+						"no %s DMA channel: %ld\n",
+						name, PTR_ERR(chan));
+
+				ret = PTR_ERR(chan);
+				goto err_release;
+			}
+
+			cl->dma_chans[stream] = chan;
+		}
+
+		fe->id = i;
+		fe->name =
+			devm_kasprintf(&pdev->dev, GFP_KERNEL, "mca-pcm-%d", i);
+		if (!fe->name) {
+			ret = -ENOMEM;
+			goto err_release;
+		}
+		fe->ops = &mca_fe_ops;
+		fe->playback.channels_min = 1;
+		fe->playback.channels_max = 32;
+		fe->playback.rates = SNDRV_PCM_RATE_8000_192000;
+		fe->playback.formats = APPLE_MCA_FMTBITS;
+		fe->capture.channels_min = 1;
+		fe->capture.channels_max = 32;
+		fe->capture.rates = SNDRV_PCM_RATE_8000_192000;
+		fe->capture.formats = APPLE_MCA_FMTBITS;
+		fe->symmetric_rate = 1;
+
+		fe->playback.stream_name =
+			devm_kasprintf(&pdev->dev, GFP_KERNEL, "PCM%d TX", i);
+		fe->capture.stream_name =
+			devm_kasprintf(&pdev->dev, GFP_KERNEL, "PCM%d RX", i);
+
+		if (!fe->playback.stream_name || !fe->capture.stream_name) {
+			ret = -ENOMEM;
+			goto err_release;
+		}
+
+		be->id = i + nclusters;
+		be->name = devm_kasprintf(&pdev->dev, GFP_KERNEL, "mca-i2s-%d", i);
+		if (!be->name) {
+			ret = -ENOMEM;
+			goto err_release;
+		}
+		be->ops = &mca_be_ops;
+		be->playback.channels_min = 1;
+		be->playback.channels_max = 32;
+		be->playback.rates = SNDRV_PCM_RATE_8000_192000;
+		be->playback.formats = APPLE_MCA_FMTBITS;
+		be->capture.channels_min = 1;
+		be->capture.channels_max = 32;
+		be->capture.rates = SNDRV_PCM_RATE_8000_192000;
+		be->capture.formats = APPLE_MCA_FMTBITS;
+
+		be->playback.stream_name =
+			devm_kasprintf(&pdev->dev, GFP_KERNEL, "I2S%d TX", i);
+		be->capture.stream_name =
+			devm_kasprintf(&pdev->dev, GFP_KERNEL, "I2S%d RX", i);
+		if (!be->playback.stream_name || !be->capture.stream_name) {
+			ret = -ENOMEM;
+			goto err_release;
+		}
+	}
+
+	ret = devm_snd_soc_register_component(&pdev->dev, &mca_component,
+					      dai_drivers, nclusters * 2);
+	if (ret) {
+		dev_err(&pdev->dev, "unable to register ASoC component: %d\n",
+			ret);
+		goto err_release;
+	}
+
+	return 0;
+
+err_release:
+	apple_mca_release(mca);
+	return ret;
+}
+
+static int apple_mca_remove(struct platform_device *pdev)
+{
+	struct mca_data *mca = platform_get_drvdata(pdev);
+
+	apple_mca_release(mca);
+	return 0;
+}
+
+static const struct of_device_id apple_mca_of_match[] = {
+	{ .compatible = "apple,mca", },
+	{}
+};
+MODULE_DEVICE_TABLE(of, apple_mca_of_match);
+
+static struct platform_driver apple_mca_driver = {
+	.driver = {
+		.name = "apple-mca",
+		.of_match_table = apple_mca_of_match,
+	},
+	.probe = apple_mca_probe,
+	.remove = apple_mca_remove,
+};
+module_platform_driver(apple_mca_driver);
+
+MODULE_AUTHOR("Martin Povišer <povik+lin@cutebit.org>");
+MODULE_DESCRIPTION("ASoC Apple MCA driver");
+MODULE_LICENSE("GPL");