Merge tag 'dmaengine-6.10-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine

Pull dmaengine updates from Vinod Koul:
 "New HW support:
   - Freescale i.MX8ULP edma support in edma driver
   - StarFive JH8100 DMA support in Synopsis axi-dmac driver

  Updates:
   - Tracing support for freescale edma driver, updates to dpaa2 driver
   - Remove unused QCom hidma DT support
   - Support for i2c dma in imx-sdma
   - Maintainers update for idxd and edma drivers"

* tag 'dmaengine-6.10-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine: (42 commits)
  MAINTAINERS: Update role for IDXD driver
  dmaengine: fsl-edma: use _Generic to handle difference type
  dmaengine: fsl-edma: add trace event support
  dmaengine: idxd: Avoid unnecessary destruction of file_ida
  dmaengine: xilinx: xdma: fix module autoloading
  dt-bindings: dma: fsl-edma: allow 'power-domains' property
  dt-bindings: dma: fsl-edma: remove 'clocks' from required
  dmaengine: fsl-dpaa2-qdma: Fix kernel-doc check warning
  dmaengine: imx-sdma: Add i2c dma support
  dmaengine: imx-sdma: utilize compiler to calculate ADDRS_ARRAY_SIZE_V<n>
  dt-bindings: fsl-imx-sdma: Add I2C peripheral types ID
  dt-bindings: fsl-dma: fsl-edma: clean up unused "fsl,imx8qm-adma" compatible string
  dmaengine: fsl-edma: clean up unused "fsl,imx8qm-adma" compatible string
  dt-bindings: dma: Drop unused QCom hidma binding
  dmaengine: qcom: Drop hidma DT support
  dmaengine: pl08x: Use kcalloc() instead of kzalloc()
  dmaengine: fsl-dpaa2-qdma: Update DPDMAI interfaces to version 3
  dmaengine: fsl-edma: fix miss mutex unlock at an error return path
  dmaengine: pch_dma: remove unused function chan2parent
  dmaengine: fsl-dpaa2-qdma: Add dpdmai_cmd_open
  ...

24 files changed, 499 insertions, 413 deletions

diff --git a/drivers/dma/Makefile b/drivers/dma/Makefile
index dfd40d14e408..802ca916f05f 100644
--- a/drivers/dma/Makefile
+++ b/drivers/dma/Makefile
@@ -31,10 +31,12 @@ obj-$(CONFIG_DW_AXI_DMAC) += dw-axi-dmac/
 obj-$(CONFIG_DW_DMAC_CORE) += dw/
 obj-$(CONFIG_DW_EDMA) += dw-edma/
 obj-$(CONFIG_EP93XX_DMA) += ep93xx_dma.o
+fsl-edma-trace-$(CONFIG_TRACING) := fsl-edma-trace.o
+CFLAGS_fsl-edma-trace.o := -I$(src)
 obj-$(CONFIG_FSL_DMA) += fsldma.o
-fsl-edma-objs := fsl-edma-main.o fsl-edma-common.o
+fsl-edma-objs := fsl-edma-main.o fsl-edma-common.o ${fsl-edma-trace-y}
 obj-$(CONFIG_FSL_EDMA) += fsl-edma.o
-mcf-edma-objs := mcf-edma-main.o fsl-edma-common.o
+mcf-edma-objs := mcf-edma-main.o fsl-edma-common.o ${fsl-edma-trace-y}
 obj-$(CONFIG_MCF_EDMA) += mcf-edma.o
 obj-$(CONFIG_FSL_QDMA) += fsl-qdma.o
 obj-$(CONFIG_FSL_RAID) += fsl_raid.o
diff --git a/drivers/dma/amba-pl08x.c b/drivers/dma/amba-pl08x.c
index fbf048f432bf..73a5cfb4da8a 100644
--- a/drivers/dma/amba-pl08x.c
+++ b/drivers/dma/amba-pl08x.c
@@ -2855,8 +2855,8 @@ static int pl08x_probe(struct amba_device *adev, const struct amba_id *id)
 	}
 
 	/* Initialize physical channels */
-	pl08x->phy_chans = kzalloc((vd->channels * sizeof(*pl08x->phy_chans)),
-			GFP_KERNEL);
+	pl08x->phy_chans = kcalloc(vd->channels, sizeof(*pl08x->phy_chans),
+				   GFP_KERNEL);
 	if (!pl08x->phy_chans) {
 		ret = -ENOMEM;
 		goto out_no_phychans;
diff --git a/drivers/dma/dma-axi-dmac.c b/drivers/dma/dma-axi-dmac.c
index 4e339c04fc1e..bdb752f11869 100644
--- a/drivers/dma/dma-axi-dmac.c
+++ b/drivers/dma/dma-axi-dmac.c
@@ -1002,6 +1002,16 @@ static int axi_dmac_detect_caps(struct axi_dmac *dmac, unsigned int version)
 	return 0;
 }
 
+static void axi_dmac_tasklet_kill(void *task)
+{
+	tasklet_kill(task);
+}
+
+static void axi_dmac_free_dma_controller(void *of_node)
+{
+	of_dma_controller_free(of_node);
+}
+
 static int axi_dmac_probe(struct platform_device *pdev)
 {
 	struct dma_device *dma_dev;
@@ -1025,14 +1035,10 @@ static int axi_dmac_probe(struct platform_device *pdev)
 	if (IS_ERR(dmac->base))
 		return PTR_ERR(dmac->base);
 
-	dmac->clk = devm_clk_get(&pdev->dev, NULL);
+	dmac->clk = devm_clk_get_enabled(&pdev->dev, NULL);
 	if (IS_ERR(dmac->clk))
 		return PTR_ERR(dmac->clk);
 
-	ret = clk_prepare_enable(dmac->clk);
-	if (ret < 0)
-		return ret;
-
 	version = axi_dmac_read(dmac, ADI_AXI_REG_VERSION);
 
 	if (version >= ADI_AXI_PCORE_VER(4, 3, 'a'))
@@ -1041,7 +1047,7 @@ static int axi_dmac_probe(struct platform_device *pdev)
 		ret = axi_dmac_parse_dt(&pdev->dev, dmac);
 
 	if (ret < 0)
-		goto err_clk_disable;
+		return ret;
 
 	INIT_LIST_HEAD(&dmac->chan.active_descs);
 
@@ -1072,7 +1078,7 @@ static int axi_dmac_probe(struct platform_device *pdev)
 
 	ret = axi_dmac_detect_caps(dmac, version);
 	if (ret)
-		goto err_clk_disable;
+		return ret;
 
 	dma_dev->copy_align = (dmac->chan.address_align_mask + 1);
 
@@ -1088,57 +1094,42 @@ static int axi_dmac_probe(struct platform_device *pdev)
 		    !AXI_DMAC_DST_COHERENT_GET(ret)) {
 			dev_err(dmac->dma_dev.dev,
 				"Coherent DMA not supported in hardware");
-			ret = -EINVAL;
-			goto err_clk_disable;
+			return -EINVAL;
 		}
 	}
 
-	ret = dma_async_device_register(dma_dev);
+	ret = dmaenginem_async_device_register(dma_dev);
+	if (ret)
+		return ret;
+
+	/*
+	 * Put the action in here so it get's done before unregistering the DMA
+	 * device.
+	 */
+	ret = devm_add_action_or_reset(&pdev->dev, axi_dmac_tasklet_kill,
+				       &dmac->chan.vchan.task);
 	if (ret)
-		goto err_clk_disable;
+		return ret;
 
 	ret = of_dma_controller_register(pdev->dev.of_node,
 		of_dma_xlate_by_chan_id, dma_dev);
 	if (ret)
-		goto err_unregister_device;
+		return ret;
 
-	ret = request_irq(dmac->irq, axi_dmac_interrupt_handler, IRQF_SHARED,
-		dev_name(&pdev->dev), dmac);
+	ret = devm_add_action_or_reset(&pdev->dev, axi_dmac_free_dma_controller,
+				       pdev->dev.of_node);
 	if (ret)
-		goto err_unregister_of;
+		return ret;
 
-	platform_set_drvdata(pdev, dmac);
+	ret = devm_request_irq(&pdev->dev, dmac->irq, axi_dmac_interrupt_handler,
+			       IRQF_SHARED, dev_name(&pdev->dev), dmac);
+	if (ret)
+		return ret;
 
 	regmap = devm_regmap_init_mmio(&pdev->dev, dmac->base,
 		 &axi_dmac_regmap_config);
-	if (IS_ERR(regmap)) {
-		ret = PTR_ERR(regmap);
-		goto err_free_irq;
-	}
-
-	return 0;
-
-err_free_irq:
-	free_irq(dmac->irq, dmac);
-err_unregister_of:
-	of_dma_controller_free(pdev->dev.of_node);
-err_unregister_device:
-	dma_async_device_unregister(&dmac->dma_dev);
-err_clk_disable:
-	clk_disable_unprepare(dmac->clk);
-
-	return ret;
-}
-
-static void axi_dmac_remove(struct platform_device *pdev)
-{
-	struct axi_dmac *dmac = platform_get_drvdata(pdev);
 
-	of_dma_controller_free(pdev->dev.of_node);
-	free_irq(dmac->irq, dmac);
-	tasklet_kill(&dmac->chan.vchan.task);
-	dma_async_device_unregister(&dmac->dma_dev);
-	clk_disable_unprepare(dmac->clk);
+	return PTR_ERR_OR_ZERO(regmap);
 }
 
 static const struct of_device_id axi_dmac_of_match_table[] = {
@@ -1153,7 +1144,6 @@ static struct platform_driver axi_dmac_driver = {
 		.of_match_table = axi_dmac_of_match_table,
 	},
 	.probe = axi_dmac_probe,
-	.remove_new = axi_dmac_remove,
 };
 module_platform_driver(axi_dmac_driver);
 
diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
index a86a81ff0caa..fffafa86d964 100644
--- a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
+++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
@@ -302,6 +302,7 @@ static struct axi_dma_desc *axi_desc_alloc(u32 num)
 		kfree(desc);
 		return NULL;
 	}
+	desc->nr_hw_descs = num;
 
 	return desc;
 }
@@ -328,7 +329,7 @@ static struct axi_dma_lli *axi_desc_get(struct axi_dma_chan *chan,
 static void axi_desc_put(struct axi_dma_desc *desc)
 {
 	struct axi_dma_chan *chan = desc->chan;
-	int count = atomic_read(&chan->descs_allocated);
+	int count = desc->nr_hw_descs;
 	struct axi_dma_hw_desc *hw_desc;
 	int descs_put;
 
@@ -1139,9 +1140,6 @@ static void axi_chan_block_xfer_complete(struct axi_dma_chan *chan)
 		/* Remove the completed descriptor from issued list before completing */
 		list_del(&vd->node);
 		vchan_cookie_complete(vd);
-
-		/* Submit queued descriptors after processing the completed ones */
-		axi_chan_start_first_queued(chan);
 	}
 
 out:
@@ -1445,6 +1443,24 @@ static int parse_device_properties(struct axi_dma_chip *chip)
 	return 0;
 }
 
+static int axi_req_irqs(struct platform_device *pdev, struct axi_dma_chip *chip)
+{
+	int irq_count = platform_irq_count(pdev);
+	int ret;
+
+	for (int i = 0; i < irq_count; i++) {
+		chip->irq[i] = platform_get_irq(pdev, i);
+		if (chip->irq[i] < 0)
+			return chip->irq[i];
+		ret = devm_request_irq(chip->dev, chip->irq[i], dw_axi_dma_interrupt,
+				IRQF_SHARED, KBUILD_MODNAME, chip);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
 static int dw_probe(struct platform_device *pdev)
 {
 	struct axi_dma_chip *chip;
@@ -1471,10 +1487,6 @@ static int dw_probe(struct platform_device *pdev)
 	chip->dev = &pdev->dev;
 	chip->dw->hdata = hdata;
 
-	chip->irq = platform_get_irq(pdev, 0);
-	if (chip->irq < 0)
-		return chip->irq;
-
 	chip->regs = devm_platform_ioremap_resource(pdev, 0);
 	if (IS_ERR(chip->regs))
 		return PTR_ERR(chip->regs);
@@ -1515,8 +1527,7 @@ static int dw_probe(struct platform_device *pdev)
 	if (!dw->chan)
 		return -ENOMEM;
 
-	ret = devm_request_irq(chip->dev, chip->irq, dw_axi_dma_interrupt,
-			       IRQF_SHARED, KBUILD_MODNAME, chip);
+	ret = axi_req_irqs(pdev, chip);
 	if (ret)
 		return ret;
 
@@ -1629,7 +1640,9 @@ static void dw_remove(struct platform_device *pdev)
 	pm_runtime_disable(chip->dev);
 	axi_dma_suspend(chip);
 
-	devm_free_irq(chip->dev, chip->irq, chip);
+	for (i = 0; i < DMAC_MAX_CHANNELS; i++)
+		if (chip->irq[i] > 0)
+			devm_free_irq(chip->dev, chip->irq[i], chip);
 
 	of_dma_controller_free(chip->dev->of_node);
 
@@ -1653,6 +1666,9 @@ static const struct of_device_id dw_dma_of_id_table[] = {
 	}, {
 		.compatible = "starfive,jh7110-axi-dma",
 		.data = (void *)(AXI_DMA_FLAG_HAS_RESETS | AXI_DMA_FLAG_USE_CFG2),
+	}, {
+		.compatible = "starfive,jh8100-axi-dma",
+		.data = (void *)AXI_DMA_FLAG_HAS_RESETS,
 	},
 	{}
 };
diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
index 454904d99654..b842e6a8d90d 100644
--- a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
+++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
@@ -65,7 +65,7 @@ struct dw_axi_dma {
 
 struct axi_dma_chip {
 	struct device		*dev;
-	int			irq;
+	int			irq[DMAC_MAX_CHANNELS];
 	void __iomem		*regs;
 	void __iomem		*apb_regs;
 	struct clk		*core_clk;
@@ -104,6 +104,7 @@ struct axi_dma_desc {
 	u32				completed_blocks;
 	u32				length;
 	u32				period_len;
+	u32				nr_hw_descs;
 };
 
 struct axi_dma_chan_config {
diff --git a/drivers/dma/fsl-dpaa2-qdma/dpaa2-qdma.c b/drivers/dma/fsl-dpaa2-qdma/dpaa2-qdma.c
index 5a8061a307cd..36384d019263 100644
--- a/drivers/dma/fsl-dpaa2-qdma/dpaa2-qdma.c
+++ b/drivers/dma/fsl-dpaa2-qdma/dpaa2-qdma.c
@@ -362,7 +362,7 @@ static int __cold dpaa2_qdma_setup(struct fsl_mc_device *ls_dev)
 
 	for (i = 0; i < priv->num_pairs; i++) {
 		err = dpdmai_get_rx_queue(priv->mc_io, 0, ls_dev->mc_handle,
-					  i, &priv->rx_queue_attr[i]);
+					  i, 0, &priv->rx_queue_attr[i]);
 		if (err) {
 			dev_err(dev, "dpdmai_get_rx_queue() failed\n");
 			goto exit;
@@ -370,13 +370,13 @@ static int __cold dpaa2_qdma_setup(struct fsl_mc_device *ls_dev)
 		ppriv->rsp_fqid = priv->rx_queue_attr[i].fqid;
 
 		err = dpdmai_get_tx_queue(priv->mc_io, 0, ls_dev->mc_handle,
-					  i, &priv->tx_fqid[i]);
+					  i, 0, &priv->tx_queue_attr[i]);
 		if (err) {
 			dev_err(dev, "dpdmai_get_tx_queue() failed\n");
 			goto exit;
 		}
-		ppriv->req_fqid = priv->tx_fqid[i];
-		ppriv->prio = i;
+		ppriv->req_fqid = priv->tx_queue_attr[i].fqid;
+		ppriv->prio = DPAA2_QDMA_DEFAULT_PRIORITY;
 		ppriv->priv = priv;
 		ppriv++;
 	}
@@ -542,7 +542,7 @@ static int __cold dpaa2_dpdmai_bind(struct dpaa2_qdma_priv *priv)
 		rx_queue_cfg.dest_cfg.dest_id = ppriv->nctx.dpio_id;
 		rx_queue_cfg.dest_cfg.priority = ppriv->prio;
 		err = dpdmai_set_rx_queue(priv->mc_io, 0, ls_dev->mc_handle,
-					  rx_queue_cfg.dest_cfg.priority,
+					  rx_queue_cfg.dest_cfg.priority, 0,
 					  &rx_queue_cfg);
 		if (err) {
 			dev_err(dev, "dpdmai_set_rx_queue() failed\n");
@@ -642,7 +642,7 @@ static int dpaa2_dpdmai_init_channels(struct dpaa2_qdma_engine *dpaa2_qdma)
 	for (i = 0; i < dpaa2_qdma->n_chans; i++) {
 		dpaa2_chan = &dpaa2_qdma->chans[i];
 		dpaa2_chan->qdma = dpaa2_qdma;
-		dpaa2_chan->fqid = priv->tx_fqid[i % num];
+		dpaa2_chan->fqid = priv->tx_queue_attr[i % num].fqid;
 		dpaa2_chan->vchan.desc_free = dpaa2_qdma_free_desc;
 		vchan_init(&dpaa2_chan->vchan, &dpaa2_qdma->dma_dev);
 		spin_lock_init(&dpaa2_chan->queue_lock);
@@ -802,7 +802,7 @@ static void dpaa2_qdma_shutdown(struct fsl_mc_device *ls_dev)
 	dpdmai_disable(priv->mc_io, 0, ls_dev->mc_handle);
 	dpaa2_dpdmai_dpio_unbind(priv);
 	dpdmai_close(priv->mc_io, 0, ls_dev->mc_handle);
-	dpdmai_destroy(priv->mc_io, 0, ls_dev->mc_handle);
+	dpdmai_destroy(priv->mc_io, 0, priv->dpqdma_id, ls_dev->mc_handle);
 }
 
 static const struct fsl_mc_device_id dpaa2_qdma_id_table[] = {
diff --git a/drivers/dma/fsl-dpaa2-qdma/dpaa2-qdma.h b/drivers/dma/fsl-dpaa2-qdma/dpaa2-qdma.h
index 03e2f4e0baca..2c80077cb7c0 100644
--- a/drivers/dma/fsl-dpaa2-qdma/dpaa2-qdma.h
+++ b/drivers/dma/fsl-dpaa2-qdma/dpaa2-qdma.h
@@ -6,6 +6,7 @@
 
 #define DPAA2_QDMA_STORE_SIZE 16
 #define NUM_CH 8
+#define DPAA2_QDMA_DEFAULT_PRIORITY 0
 
 struct dpaa2_qdma_sd_d {
 	u32 rsv:32;
@@ -122,8 +123,8 @@ struct dpaa2_qdma_priv {
 	struct dpaa2_qdma_engine	*dpaa2_qdma;
 	struct dpaa2_qdma_priv_per_prio	*ppriv;
 
-	struct dpdmai_rx_queue_attr rx_queue_attr[DPDMAI_PRIO_NUM];
-	u32 tx_fqid[DPDMAI_PRIO_NUM];
+	struct dpdmai_rx_queue_attr rx_queue_attr[DPDMAI_MAX_QUEUE_NUM];
+	struct dpdmai_tx_queue_attr tx_queue_attr[DPDMAI_MAX_QUEUE_NUM];
 };
 
 struct dpaa2_qdma_priv_per_prio {
diff --git a/drivers/dma/fsl-dpaa2-qdma/dpdmai.c b/drivers/dma/fsl-dpaa2-qdma/dpdmai.c
index 878662aaa1c2..36897b41ee7e 100644
--- a/drivers/dma/fsl-dpaa2-qdma/dpdmai.c
+++ b/drivers/dma/fsl-dpaa2-qdma/dpdmai.c
@@ -1,47 +1,52 @@
 // SPDX-License-Identifier: GPL-2.0
 // Copyright 2019 NXP
 
+#include <linux/bitfield.h>
 #include <linux/module.h>
 #include <linux/types.h>
 #include <linux/io.h>
 #include <linux/fsl/mc.h>
 #include "dpdmai.h"
 
+#define DEST_TYPE_MASK 0xF
+
 struct dpdmai_rsp_get_attributes {
 	__le32 id;
 	u8 num_of_priorities;
-	u8 pad0[3];
+	u8 num_of_queues;
+	u8 pad0[2];
 	__le16 major;
 	__le16 minor;
 };
 
 struct dpdmai_cmd_queue {
 	__le32 dest_id;
-	u8 priority;
-	u8 queue;
+	u8 dest_priority;
+	union {
+		u8 queue;
+		u8 pri;
+	};
 	u8 dest_type;
-	u8 pad;
+	u8 queue_idx;
 	__le64 user_ctx;
 	union {
 		__le32 options;
 		__le32 fqid;
 	};
-};
+} __packed;
 
 struct dpdmai_rsp_get_tx_queue {
 	__le64 pad;
 	__le32 fqid;
 };
 
-#define MC_CMD_OP(_cmd, _param, _offset, _width, _type, _arg) \
-	((_cmd).params[_param] |= mc_enc((_offset), (_width), _arg))
+struct dpdmai_cmd_open {
+	__le32 dpdmai_id;
+} __packed;
 
-/* cmd, param, offset, width, type, arg_name */
-#define DPDMAI_CMD_CREATE(cmd, cfg) \
-do { \
-	MC_CMD_OP(cmd, 0, 8,  8,  u8,  (cfg)->priorities[0]);\
-	MC_CMD_OP(cmd, 0, 16, 8,  u8,  (cfg)->priorities[1]);\
-} while (0)
+struct dpdmai_cmd_destroy {
+	__le32 dpdmai_id;
+} __packed;
 
 static inline u64 mc_enc(int lsoffset, int width, u64 val)
 {
@@ -68,16 +73,16 @@ static inline u64 mc_enc(int lsoffset, int width, u64 val)
 int dpdmai_open(struct fsl_mc_io *mc_io, u32 cmd_flags,
 		int dpdmai_id, u16 *token)
 {
+	struct dpdmai_cmd_open *cmd_params;
 	struct fsl_mc_command cmd = { 0 };
-	__le64 *cmd_dpdmai_id;
 	int err;
 
 	/* prepare command */
 	cmd.header = mc_encode_cmd_header(DPDMAI_CMDID_OPEN,
 					  cmd_flags, 0);
 
-	cmd_dpdmai_id = cmd.params;
-	*cmd_dpdmai_id = cpu_to_le32(dpdmai_id);
+	cmd_params = (struct dpdmai_cmd_open *)&cmd.params;
+	cmd_params->dpdmai_id = cpu_to_le32(dpdmai_id);
 
 	/* send command to mc*/
 	err = mc_send_command(mc_io, &cmd);
@@ -116,65 +121,26 @@ int dpdmai_close(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token)
 EXPORT_SYMBOL_GPL(dpdmai_close);
 
 /**
- * dpdmai_create() - Create the DPDMAI object
- * @mc_io:	Pointer to MC portal's I/O object
- * @cmd_flags:	Command flags; one or more of 'MC_CMD_FLAG_'
- * @cfg:	Configuration structure
- * @token:	Returned token; use in subsequent API calls
- *
- * Create the DPDMAI object, allocate required resources and
- * perform required initialization.
- *
- * The object can be created either by declaring it in the
- * DPL file, or by calling this function.
- *
- * This function returns a unique authentication token,
- * associated with the specific object ID and the specific MC
- * portal; this token must be used in all subsequent calls to
- * this specific object. For objects that are created using the
- * DPL file, call dpdmai_open() function to get an authentication
- * token first.
- *
- * Return:	'0' on Success; Error code otherwise.
- */
-int dpdmai_create(struct fsl_mc_io *mc_io, u32 cmd_flags,
-		  const struct dpdmai_cfg *cfg, u16 *token)
-{
-	struct fsl_mc_command cmd = { 0 };
-	int err;
-
-	/* prepare command */
-	cmd.header = mc_encode_cmd_header(DPDMAI_CMDID_CREATE,
-					  cmd_flags, 0);
-	DPDMAI_CMD_CREATE(cmd, cfg);
-
-	/* send command to mc*/
-	err = mc_send_command(mc_io, &cmd);
-	if (err)
-		return err;
-
-	/* retrieve response parameters */
-	*token = mc_cmd_hdr_read_token(&cmd);
-
-	return 0;
-}
-
-/**
  * dpdmai_destroy() - Destroy the DPDMAI object and release all its resources.
  * @mc_io:      Pointer to MC portal's I/O object
  * @cmd_flags:  Command flags; one or more of 'MC_CMD_FLAG_'
+ * @dpdmai_id:	The object id; it must be a valid id within the container that created this object;
  * @token:      Token of DPDMAI object
  *
  * Return:      '0' on Success; error code otherwise.
  */
-int dpdmai_destroy(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token)
+int dpdmai_destroy(struct fsl_mc_io *mc_io, u32 cmd_flags, u32 dpdmai_id, u16 token)
 {
+	struct dpdmai_cmd_destroy *cmd_params;
 	struct fsl_mc_command cmd = { 0 };
 
 	/* prepare command */
 	cmd.header = mc_encode_cmd_header(DPDMAI_CMDID_DESTROY,
 					  cmd_flags, token);
 
+	cmd_params = (struct dpdmai_cmd_destroy *)&cmd.params;
+	cmd_params->dpdmai_id = cpu_to_le32(dpdmai_id);
+
 	/* send command to mc*/
 	return mc_send_command(mc_io, &cmd);
 }
@@ -274,6 +240,7 @@ int dpdmai_get_attributes(struct fsl_mc_io *mc_io, u32 cmd_flags,
 	attr->version.major = le16_to_cpu(rsp_params->major);
 	attr->version.minor = le16_to_cpu(rsp_params->minor);
 	attr->num_of_priorities = rsp_params->num_of_priorities;
+	attr->num_of_queues = rsp_params->num_of_queues;
 
 	return 0;
 }
@@ -284,13 +251,14 @@ EXPORT_SYMBOL_GPL(dpdmai_get_attributes);
  * @mc_io:	Pointer to MC portal's I/O object
  * @cmd_flags:	Command flags; one or more of 'MC_CMD_FLAG_'
  * @token:	Token of DPDMAI object
+ * @queue_idx:	DMA queue index
  * @priority:	Select the queue relative to number of
  *		priorities configured at DPDMAI creation
  * @cfg:	Rx queue configuration
  *
  * Return:	'0' on Success; Error code otherwise.
  */
-int dpdmai_set_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token,
+int dpdmai_set_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token, u8 queue_idx,
 			u8 priority, const struct dpdmai_rx_queue_cfg *cfg)
 {
 	struct dpdmai_cmd_queue *cmd_params;
@@ -302,11 +270,12 @@ int dpdmai_set_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token,
 
 	cmd_params = (struct dpdmai_cmd_queue *)cmd.params;
 	cmd_params->dest_id = cpu_to_le32(cfg->dest_cfg.dest_id);
-	cmd_params->priority = cfg->dest_cfg.priority;
-	cmd_params->queue = priority;
+	cmd_params->dest_priority = cfg->dest_cfg.priority;
+	cmd_params->pri = priority;
 	cmd_params->dest_type = cfg->dest_cfg.dest_type;
 	cmd_params->user_ctx = cpu_to_le64(cfg->user_ctx);
 	cmd_params->options = cpu_to_le32(cfg->options);
+	cmd_params->queue_idx = queue_idx;
 
 	/* send command to mc*/
 	return mc_send_command(mc_io, &cmd);
@@ -318,13 +287,14 @@ EXPORT_SYMBOL_GPL(dpdmai_set_rx_queue);
  * @mc_io:	Pointer to MC portal's I/O object
  * @cmd_flags:	Command flags; one or more of 'MC_CMD_FLAG_'
  * @token:	Token of DPDMAI object
+ * @queue_idx:	DMA Queue index
  * @priority:	Select the queue relative to number of
  *				priorities configured at DPDMAI creation
  * @attr:	Returned Rx queue attributes
  *
  * Return:	'0' on Success; Error code otherwise.
  */
-int dpdmai_get_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token,
+int dpdmai_get_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token, u8 queue_idx,
 			u8 priority, struct dpdmai_rx_queue_attr *attr)
 {
 	struct dpdmai_cmd_queue *cmd_params;
@@ -337,6 +307,7 @@ int dpdmai_get_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token,
 
 	cmd_params = (struct dpdmai_cmd_queue *)cmd.params;
 	cmd_params->queue = priority;
+	cmd_params->queue_idx = queue_idx;
 
 	/* send command to mc*/
 	err = mc_send_command(mc_io, &cmd);
@@ -345,8 +316,8 @@ int dpdmai_get_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token,
 
 	/* retrieve response parameters */
 	attr->dest_cfg.dest_id = le32_to_cpu(cmd_params->dest_id);
-	attr->dest_cfg.priority = cmd_params->priority;
-	attr->dest_cfg.dest_type = cmd_params->dest_type;
+	attr->dest_cfg.priority = cmd_params->dest_priority;
+	attr->dest_cfg.dest_type = FIELD_GET(DEST_TYPE_MASK, cmd_params->dest_type);
 	attr->user_ctx = le64_to_cpu(cmd_params->user_ctx);
 	attr->fqid = le32_to_cpu(cmd_params->fqid);
 
@@ -359,14 +330,15 @@ EXPORT_SYMBOL_GPL(dpdmai_get_rx_queue);
  * @mc_io:	Pointer to MC portal's I/O object
  * @cmd_flags:	Command flags; one or more of 'MC_CMD_FLAG_'
  * @token:	Token of DPDMAI object
+ * @queue_idx:	DMA queue index
  * @priority:	Select the queue relative to number of
  *			priorities configured at DPDMAI creation
- * @fqid:	Returned Tx queue
+ * @attr:	Returned DMA Tx queue attributes
  *
  * Return:	'0' on Success; Error code otherwise.
  */
 int dpdmai_get_tx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags,
-			u16 token, u8 priority, u32 *fqid)
+			u16 token, u8 queue_idx, u8 priority, struct dpdmai_tx_queue_attr *attr)
 {
 	struct dpdmai_rsp_get_tx_queue *rsp_params;
 	struct dpdmai_cmd_queue *cmd_params;
@@ -379,6 +351,7 @@ int dpdmai_get_tx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags,
 
 	cmd_params = (struct dpdmai_cmd_queue *)cmd.params;
 	cmd_params->queue = priority;
+	cmd_params->queue_idx = queue_idx;
 
 	/* send command to mc*/
 	err = mc_send_command(mc_io, &cmd);
@@ -388,7 +361,7 @@ int dpdmai_get_tx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags,
 	/* retrieve response parameters */
 
 	rsp_params = (struct dpdmai_rsp_get_tx_queue *)cmd.params;
-	*fqid = le32_to_cpu(rsp_params->fqid);
+	attr->fqid = le32_to_cpu(rsp_params->fqid);
 
 	return 0;
 }
diff --git a/drivers/dma/fsl-dpaa2-qdma/dpdmai.h b/drivers/dma/fsl-dpaa2-qdma/dpdmai.h
index b13b9bf0c003..3fe7d8327366 100644
--- a/drivers/dma/fsl-dpaa2-qdma/dpdmai.h
+++ b/drivers/dma/fsl-dpaa2-qdma/dpdmai.h
@@ -5,14 +5,19 @@
 #define __FSL_DPDMAI_H
 
 /* DPDMAI Version */
-#define DPDMAI_VER_MAJOR	2
-#define DPDMAI_VER_MINOR	2
+#define DPDMAI_VER_MAJOR	3
+#define DPDMAI_VER_MINOR	3
 
-#define DPDMAI_CMD_BASE_VERSION	0
+#define DPDMAI_CMD_BASE_VERSION	1
 #define DPDMAI_CMD_ID_OFFSET	4
 
-#define DPDMAI_CMDID_FORMAT(x)	(((x) << DPDMAI_CMD_ID_OFFSET) | \
-				DPDMAI_CMD_BASE_VERSION)
+/*
+ * Maximum number of Tx/Rx queues per DPDMAI object
+ */
+#define DPDMAI_MAX_QUEUE_NUM	8
+
+#define DPDMAI_CMDID_FORMAT_V(x, v)	(((x) << DPDMAI_CMD_ID_OFFSET) | (v))
+#define DPDMAI_CMDID_FORMAT(x)		DPDMAI_CMDID_FORMAT_V(x, DPDMAI_CMD_BASE_VERSION)
 
 /* Command IDs */
 #define DPDMAI_CMDID_CLOSE		DPDMAI_CMDID_FORMAT(0x800)
@@ -26,18 +31,9 @@
 #define DPDMAI_CMDID_RESET              DPDMAI_CMDID_FORMAT(0x005)
 #define DPDMAI_CMDID_IS_ENABLED         DPDMAI_CMDID_FORMAT(0x006)
 
-#define DPDMAI_CMDID_SET_IRQ            DPDMAI_CMDID_FORMAT(0x010)
-#define DPDMAI_CMDID_GET_IRQ            DPDMAI_CMDID_FORMAT(0x011)
-#define DPDMAI_CMDID_SET_IRQ_ENABLE     DPDMAI_CMDID_FORMAT(0x012)
-#define DPDMAI_CMDID_GET_IRQ_ENABLE     DPDMAI_CMDID_FORMAT(0x013)
-#define DPDMAI_CMDID_SET_IRQ_MASK       DPDMAI_CMDID_FORMAT(0x014)
-#define DPDMAI_CMDID_GET_IRQ_MASK       DPDMAI_CMDID_FORMAT(0x015)
-#define DPDMAI_CMDID_GET_IRQ_STATUS     DPDMAI_CMDID_FORMAT(0x016)
-#define DPDMAI_CMDID_CLEAR_IRQ_STATUS	DPDMAI_CMDID_FORMAT(0x017)
-
-#define DPDMAI_CMDID_SET_RX_QUEUE	DPDMAI_CMDID_FORMAT(0x1A0)
-#define DPDMAI_CMDID_GET_RX_QUEUE       DPDMAI_CMDID_FORMAT(0x1A1)
-#define DPDMAI_CMDID_GET_TX_QUEUE       DPDMAI_CMDID_FORMAT(0x1A2)
+#define DPDMAI_CMDID_SET_RX_QUEUE	DPDMAI_CMDID_FORMAT_V(0x1A0, 2)
+#define DPDMAI_CMDID_GET_RX_QUEUE       DPDMAI_CMDID_FORMAT_V(0x1A1, 2)
+#define DPDMAI_CMDID_GET_TX_QUEUE       DPDMAI_CMDID_FORMAT_V(0x1A2, 2)
 
 #define MC_CMD_HDR_TOKEN_O 32  /* Token field offset */
 #define MC_CMD_HDR_TOKEN_S 16  /* Token field size */
@@ -49,30 +45,32 @@
  * Contains initialization APIs and runtime control APIs for DPDMAI
  */
 
-/**
+/*
  * Maximum number of Tx/Rx priorities per DPDMAI object
  */
 #define DPDMAI_PRIO_NUM		2
 
 /* DPDMAI queue modification options */
 
-/**
+/*
  * Select to modify the user's context associated with the queue
  */
 #define DPDMAI_QUEUE_OPT_USER_CTX	0x1
 
-/**
+/*
  * Select to modify the queue's destination
  */
 #define DPDMAI_QUEUE_OPT_DEST		0x2
 
 /**
  * struct dpdmai_cfg - Structure representing DPDMAI configuration
+ * @num_queues:	Number of the DMA queues
  * @priorities: Priorities for the DMA hardware processing; valid priorities are
  *	configured with values 1-8; the entry following last valid entry
  *	should be configured with 0
  */
 struct dpdmai_cfg {
+	u8 num_queues;
 	u8 priorities[DPDMAI_PRIO_NUM];
 };
 
@@ -80,20 +78,19 @@ struct dpdmai_cfg {
  * struct dpdmai_attr - Structure representing DPDMAI attributes
  * @id: DPDMAI object ID
  * @version: DPDMAI version
+ * @version.major: DPDMAI major version
+ * @version.minor: DPDMAI minor version
  * @num_of_priorities: number of priorities
+ * @num_of_queues: number of the DMA queues
  */
 struct dpdmai_attr {
 	int	id;
-	/**
-	 * struct version - DPDMAI version
-	 * @major: DPDMAI major version
-	 * @minor: DPDMAI minor version
-	 */
 	struct {
 		u16 major;
 		u16 minor;
 	} version;
 	u8 num_of_priorities;
+	u8 num_of_queues;
 };
 
 /**
@@ -158,22 +155,24 @@ struct dpdmai_rx_queue_attr {
 	u32 fqid;
 };
 
+struct dpdmai_tx_queue_attr {
+	u32 fqid;
+};
+
 int dpdmai_open(struct fsl_mc_io *mc_io, u32 cmd_flags,
 		int dpdmai_id, u16 *token);
 int dpdmai_close(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token);
-int dpdmai_destroy(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token);
-int dpdmai_create(struct fsl_mc_io *mc_io, u32 cmd_flags,
-		  const struct dpdmai_cfg *cfg, u16 *token);
+int dpdmai_destroy(struct fsl_mc_io *mc_io, u32 cmd_flags, u32 dpdmai_id, u16 token);
 int dpdmai_enable(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token);
 int dpdmai_disable(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token);
 int dpdmai_reset(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token);
 int dpdmai_get_attributes(struct fsl_mc_io *mc_io, u32 cmd_flags,
 			  u16 token, struct dpdmai_attr	*attr);
 int dpdmai_set_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token,
-			u8 priority, const struct dpdmai_rx_queue_cfg *cfg);
+			u8 queue_idx, u8 priority, const struct dpdmai_rx_queue_cfg *cfg);
 int dpdmai_get_rx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags, u16 token,
-			u8 priority, struct dpdmai_rx_queue_attr *attr);
+			u8 queue_idx, u8 priority, struct dpdmai_rx_queue_attr *attr);
 int dpdmai_get_tx_queue(struct fsl_mc_io *mc_io, u32 cmd_flags,
-			u16 token, u8 priority, u32 *fqid);
+			u16 token, u8 queue_idx, u8 priority, struct dpdmai_tx_queue_attr *attr);
 
 #endif /* __FSL_DPDMAI_H */
diff --git a/drivers/dma/fsl-edma-common.c b/drivers/dma/fsl-edma-common.c
index b18faa7cfedb..3af430787315 100644
--- a/drivers/dma/fsl-edma-common.c
+++ b/drivers/dma/fsl-edma-common.c
@@ -3,6 +3,8 @@
 // Copyright (c) 2013-2014 Freescale Semiconductor, Inc
 // Copyright (c) 2017 Sysam, Angelo Dureghello  <angelo@sysam.it>
 
+#include <linux/cleanup.h>
+#include <linux/clk.h>
 #include <linux/dmapool.h>
 #include <linux/module.h>
 #include <linux/slab.h>
@@ -74,18 +76,10 @@ static void fsl_edma3_enable_request(struct fsl_edma_chan *fsl_chan)
 
 	flags = fsl_edma_drvflags(fsl_chan);
 	val = edma_readl_chreg(fsl_chan, ch_sbr);
-	/* Remote/local swapped wrongly on iMX8 QM Audio edma */
-	if (flags & FSL_EDMA_DRV_QUIRK_SWAPPED) {
-		if (!fsl_chan->is_rxchan)
-			val |= EDMA_V3_CH_SBR_RD;
-		else
-			val |= EDMA_V3_CH_SBR_WR;
-	} else {
-		if (fsl_chan->is_rxchan)
-			val |= EDMA_V3_CH_SBR_RD;
-		else
-			val |= EDMA_V3_CH_SBR_WR;
-	}
+	if (fsl_chan->is_rxchan)
+		val |= EDMA_V3_CH_SBR_RD;
+	else
+		val |= EDMA_V3_CH_SBR_WR;
 
 	if (fsl_chan->is_remote)
 		val &= ~(EDMA_V3_CH_SBR_RD | EDMA_V3_CH_SBR_WR);
@@ -546,6 +540,8 @@ void fsl_edma_fill_tcd(struct fsl_edma_chan *fsl_chan,
 		csr |= EDMA_TCD_CSR_START;
 
 	fsl_edma_set_tcd_to_le(fsl_chan, tcd, csr, csr);
+
+	trace_edma_fill_tcd(fsl_chan, tcd);
 }
 
 static struct fsl_edma_desc *fsl_edma_alloc_desc(struct fsl_edma_chan *fsl_chan,
@@ -810,6 +806,9 @@ int fsl_edma_alloc_chan_resources(struct dma_chan *chan)
 {
 	struct fsl_edma_chan *fsl_chan = to_fsl_edma_chan(chan);
 
+	if (fsl_edma_drvflags(fsl_chan) & FSL_EDMA_DRV_HAS_CHCLK)
+		clk_prepare_enable(fsl_chan->clk);
+
 	fsl_chan->tcd_pool = dma_pool_create("tcd_pool", chan->device->dev,
 				fsl_edma_drvflags(fsl_chan) & FSL_EDMA_DRV_TCD64 ?
 				sizeof(struct fsl_edma_hw_tcd64) : sizeof(struct fsl_edma_hw_tcd),
@@ -838,6 +837,8 @@ void fsl_edma_free_chan_resources(struct dma_chan *chan)
 	fsl_chan->tcd_pool = NULL;
 	fsl_chan->is_sw = false;
 	fsl_chan->srcid = 0;
+	if (fsl_edma_drvflags(fsl_chan) & FSL_EDMA_DRV_HAS_CHCLK)
+		clk_disable_unprepare(fsl_chan->clk);
 }
 
 void fsl_edma_cleanup_vchan(struct dma_device *dmadev)
diff --git a/drivers/dma/fsl-edma-common.h b/drivers/dma/fsl-edma-common.h
index 7bf0aba471a8..ac66222c1604 100644
--- a/drivers/dma/fsl-edma-common.h
+++ b/drivers/dma/fsl-edma-common.h
@@ -151,7 +151,6 @@ struct fsl_edma_chan {
 	enum dma_status			status;
 	enum fsl_edma_pm_state		pm_state;
 	bool				idle;
-	u32				slave_id;
 	struct fsl_edma_engine		*edma;
 	struct fsl_edma_desc		*edesc;
 	struct dma_slave_config		cfg;
@@ -195,8 +194,6 @@ struct fsl_edma_desc {
 #define FSL_EDMA_DRV_HAS_PD		BIT(5)
 #define FSL_EDMA_DRV_HAS_CHCLK		BIT(6)
 #define FSL_EDMA_DRV_HAS_CHMUX		BIT(7)
-/* imx8 QM audio edma remote local swapped */
-#define FSL_EDMA_DRV_QUIRK_SWAPPED	BIT(8)
 /* control and status register is in tcd address space, edma3 reg layout */
 #define FSL_EDMA_DRV_SPLIT_REG		BIT(9)
 #define FSL_EDMA_DRV_BUS_8BYTE		BIT(10)
@@ -238,7 +235,6 @@ struct fsl_edma_engine {
 	void __iomem		*muxbase[DMAMUX_NR];
 	struct clk		*muxclk[DMAMUX_NR];
 	struct clk		*dmaclk;
-	struct clk		*chclk;
 	struct mutex		fsl_edma_mutex;
 	const struct fsl_edma_drvdata *drvdata;
 	u32			n_chans;
@@ -250,13 +246,17 @@ struct fsl_edma_engine {
 	struct fsl_edma_chan	chans[] __counted_by(n_chans);
 };
 
+static inline u32 fsl_edma_drvflags(struct fsl_edma_chan *fsl_chan)
+{
+	return fsl_chan->edma->drvdata->flags;
+}
+
 #define edma_read_tcdreg_c(chan, _tcd,  __name)				\
-(sizeof((_tcd)->__name) == sizeof(u64) ?				\
-	edma_readq(chan->edma, &(_tcd)->__name) :			\
-		((sizeof((_tcd)->__name) == sizeof(u32)) ?		\
-			edma_readl(chan->edma, &(_tcd)->__name) :	\
-			edma_readw(chan->edma, &(_tcd)->__name)		\
-		))
+_Generic(((_tcd)->__name),						\
+	__iomem __le64 : edma_readq(chan->edma, &(_tcd)->__name),		\
+	__iomem __le32 : edma_readl(chan->edma, &(_tcd)->__name),		\
+	__iomem __le16 : edma_readw(chan->edma, &(_tcd)->__name)		\
+	)
 
 #define edma_read_tcdreg(chan, __name)								\
 ((fsl_edma_drvflags(chan) & FSL_EDMA_DRV_TCD64) ?						\
@@ -264,23 +264,13 @@ struct fsl_edma_engine {
 	edma_read_tcdreg_c(chan, ((struct fsl_edma_hw_tcd __iomem *)chan->tcd), __name)		\
 )
 
-#define edma_write_tcdreg_c(chan, _tcd, _val, __name)				\
-do {										\
-	switch (sizeof(_tcd->__name)) {						\
-	case sizeof(u64):							\
-		edma_writeq(chan->edma, (u64 __force)_val, &_tcd->__name);	\
-		break;								\
-	case sizeof(u32):							\
-		edma_writel(chan->edma, (u32 __force)_val, &_tcd->__name);	\
-		break;								\
-	case sizeof(u16):							\
-		edma_writew(chan->edma, (u16 __force)_val, &_tcd->__name);	\
-		break;								\
-	case sizeof(u8):							\
-		edma_writeb(chan->edma, (u8 __force)_val, &_tcd->__name);	\
-		break;								\
-	}									\
-} while (0)
+#define edma_write_tcdreg_c(chan, _tcd, _val, __name)					\
+_Generic((_tcd->__name),								\
+	__iomem __le64 : edma_writeq(chan->edma, (u64 __force)(_val), &_tcd->__name),	\
+	__iomem __le32 : edma_writel(chan->edma, (u32 __force)(_val), &_tcd->__name),	\
+	__iomem __le16 : edma_writew(chan->edma, (u16 __force)(_val), &_tcd->__name),	\
+	__iomem u8 : edma_writeb(chan->edma, _val, &_tcd->__name)			\
+	)
 
 #define edma_write_tcdreg(chan, val, __name)							   \
 do {												   \
@@ -321,9 +311,11 @@ do {	\
 						 (((struct fsl_edma_hw_tcd *)_tcd)->_field))
 
 #define fsl_edma_le_to_cpu(x)						\
-(sizeof(x) == sizeof(u64) ? le64_to_cpu((__force __le64)(x)) :		\
-	(sizeof(x) == sizeof(u32) ? le32_to_cpu((__force __le32)(x)) :	\
-				    le16_to_cpu((__force __le16)(x))))
+_Generic((x),								\
+	__le64 : le64_to_cpu((x)),					\
+	__le32 : le32_to_cpu((x)),					\
+	__le16 : le16_to_cpu((x))					\
+)
 
 #define fsl_edma_get_tcd_to_cpu(_chan, _tcd, _field)				\
 (fsl_edma_drvflags(_chan) & FSL_EDMA_DRV_TCD64 ?				\
@@ -331,19 +323,11 @@ do {	\
 	fsl_edma_le_to_cpu(((struct fsl_edma_hw_tcd *)_tcd)->_field))
 
 #define fsl_edma_set_tcd_to_le_c(_tcd, _val, _field)					\
-do {											\
-	switch (sizeof((_tcd)->_field)) {						\
-	case sizeof(u64):								\
-		*(__force __le64 *)(&((_tcd)->_field)) = cpu_to_le64(_val);		\
-		break;									\
-	case sizeof(u32):								\
-		*(__force __le32 *)(&((_tcd)->_field)) = cpu_to_le32(_val);		\
-		break;									\
-	case sizeof(u16):								\
-		*(__force __le16 *)(&((_tcd)->_field)) = cpu_to_le16(_val);		\
-		break;									\
-	}										\
-} while (0)
+_Generic(((_tcd)->_field),								\
+	__le64 : (_tcd)->_field = cpu_to_le64(_val),					\
+	__le32 : (_tcd)->_field = cpu_to_le32(_val),					\
+	__le16 : (_tcd)->_field = cpu_to_le16(_val)					\
+)
 
 #define fsl_edma_set_tcd_to_le(_chan, _tcd, _val, _field)	\
 do {								\
@@ -353,6 +337,9 @@ do {								\
 		fsl_edma_set_tcd_to_le_c((struct fsl_edma_hw_tcd *)_tcd, _val, _field);		\
 } while (0)
 
+/* Need after struct defination */
+#include "fsl-edma-trace.h"
+
 /*
  * R/W functions for big- or little-endian registers:
  * The eDMA controller's endian is independent of the CPU core's endian.
@@ -371,23 +358,38 @@ static inline u64 edma_readq(struct fsl_edma_engine *edma, void __iomem *addr)
 		h = ioread32(addr + 4);
 	}
 
+	trace_edma_readl(edma, addr, l);
+	trace_edma_readl(edma, addr + 4, h);
+
 	return (h << 32) | l;
 }
 
 static inline u32 edma_readl(struct fsl_edma_engine *edma, void __iomem *addr)
 {
+	u32 val;
+
 	if (edma->big_endian)
-		return ioread32be(addr);
+		val = ioread32be(addr);
 	else
-		return ioread32(addr);
+		val = ioread32(addr);
+
+	trace_edma_readl(edma, addr, val);
+
+	return val;
 }
 
 static inline u16 edma_readw(struct fsl_edma_engine *edma, void __iomem *addr)
 {
+	u16 val;
+
 	if (edma->big_endian)
-		return ioread16be(addr);
+		val = ioread16be(addr);
 	else
-		return ioread16(addr);
+		val = ioread16(addr);
+
+	trace_edma_readw(edma, addr, val);
+
+	return val;
 }
 
 static inline void edma_writeb(struct fsl_edma_engine *edma,
@@ -398,6 +400,8 @@ static inline void edma_writeb(struct fsl_edma_engine *edma,
 		iowrite8(val, (void __iomem *)((unsigned long)addr ^ 0x3));
 	else
 		iowrite8(val, addr);
+
+	trace_edma_writeb(edma, addr, val);
 }
 
 static inline void edma_writew(struct fsl_edma_engine *edma,
@@ -408,6 +412,8 @@ static inline void edma_writew(struct fsl_edma_engine *edma,
 		iowrite16be(val, (void __iomem *)((unsigned long)addr ^ 0x2));
 	else
 		iowrite16(val, addr);
+
+	trace_edma_writew(edma, addr, val);
 }
 
 static inline void edma_writel(struct fsl_edma_engine *edma,
@@ -417,6 +423,8 @@ static inline void edma_writel(struct fsl_edma_engine *edma,
 		iowrite32be(val, addr);
 	else
 		iowrite32(val, addr);
+
+	trace_edma_writel(edma, addr, val);
 }
 
 static inline void edma_writeq(struct fsl_edma_engine *edma,
@@ -429,6 +437,9 @@ static inline void edma_writeq(struct fsl_edma_engine *edma,
 		iowrite32(val & 0xFFFFFFFF, addr);
 		iowrite32(val >> 32, addr + 4);
 	}
+
+	trace_edma_writel(edma, addr, val & 0xFFFFFFFF);
+	trace_edma_writel(edma, addr + 4, val >> 32);
 }
 
 static inline struct fsl_edma_chan *to_fsl_edma_chan(struct dma_chan *chan)
@@ -436,11 +447,6 @@ static inline struct fsl_edma_chan *to_fsl_edma_chan(struct dma_chan *chan)
 	return container_of(chan, struct fsl_edma_chan, vchan.chan);
 }
 
-static inline u32 fsl_edma_drvflags(struct fsl_edma_chan *fsl_chan)
-{
-	return fsl_chan->edma->drvdata->flags;
-}
-
 static inline struct fsl_edma_desc *to_fsl_edma_desc(struct virt_dma_desc *vd)
 {
 	return container_of(vd, struct fsl_edma_desc, vdesc);
diff --git a/drivers/dma/fsl-edma-main.c b/drivers/dma/fsl-edma-main.c
index 402f0058a180..391e4f13dfeb 100644
--- a/drivers/dma/fsl-edma-main.c
+++ b/drivers/dma/fsl-edma-main.c
@@ -105,7 +105,8 @@ static struct dma_chan *fsl_edma_xlate(struct of_phandle_args *dma_spec,
 	if (dma_spec->args_count != 2)
 		return NULL;
 
-	mutex_lock(&fsl_edma->fsl_edma_mutex);
+	guard(mutex)(&fsl_edma->fsl_edma_mutex);
+
 	list_for_each_entry_safe(chan, _chan, &fsl_edma->dma_dev.channels, device_node) {
 		if (chan->client_count)
 			continue;
@@ -114,15 +115,20 @@ static struct dma_chan *fsl_edma_xlate(struct of_phandle_args *dma_spec,
 			if (chan) {
 				chan->device->privatecnt++;
 				fsl_chan = to_fsl_edma_chan(chan);
-				fsl_chan->slave_id = dma_spec->args[1];
-				fsl_edma_chan_mux(fsl_chan, fsl_chan->slave_id,
+				fsl_chan->srcid = dma_spec->args[1];
+
+				if (!fsl_chan->srcid) {
+					dev_err(&fsl_chan->pdev->dev, "Invalidate srcid %d\n",
+						fsl_chan->srcid);
+					return NULL;
+				}
+
+				fsl_edma_chan_mux(fsl_chan, fsl_chan->srcid,
 						true);
-				mutex_unlock(&fsl_edma->fsl_edma_mutex);
 				return chan;
 			}
 		}
 	}
-	mutex_unlock(&fsl_edma->fsl_edma_mutex);
 	return NULL;
 }
 
@@ -342,10 +348,13 @@ static struct fsl_edma_drvdata imx8qm_data = {
 	.setup_irq = fsl_edma3_irq_init,
 };
 
-static struct fsl_edma_drvdata imx8qm_audio_data = {
-	.flags = FSL_EDMA_DRV_QUIRK_SWAPPED | FSL_EDMA_DRV_HAS_PD | FSL_EDMA_DRV_EDMA3,
+static struct fsl_edma_drvdata imx8ulp_data = {
+	.flags = FSL_EDMA_DRV_HAS_CHMUX | FSL_EDMA_DRV_HAS_CHCLK | FSL_EDMA_DRV_HAS_DMACLK |
+		 FSL_EDMA_DRV_EDMA3,
 	.chreg_space_sz = 0x10000,
 	.chreg_off = 0x10000,
+	.mux_off = 0x10000 + offsetof(struct fsl_edma3_ch_reg, ch_mux),
+	.mux_skip = 0x10000,
 	.setup_irq = fsl_edma3_irq_init,
 };
 
@@ -380,7 +389,7 @@ static const struct of_device_id fsl_edma_dt_ids[] = {
 	{ .compatible = "fsl,ls1028a-edma", .data = &ls1028a_data},
 	{ .compatible = "fsl,imx7ulp-edma", .data = &imx7ulp_data},
 	{ .compatible = "fsl,imx8qm-edma", .data = &imx8qm_data},
-	{ .compatible = "fsl,imx8qm-adma", .data = &imx8qm_audio_data},
+	{ .compatible = "fsl,imx8ulp-edma", .data = &imx8ulp_data},
 	{ .compatible = "fsl,imx93-edma3", .data = &imx93_data3},
 	{ .compatible = "fsl,imx93-edma4", .data = &imx93_data4},
 	{ .compatible = "fsl,imx95-edma5", .data = &imx95_data5},
@@ -434,6 +443,7 @@ static int fsl_edma_probe(struct platform_device *pdev)
 	struct fsl_edma_engine *fsl_edma;
 	const struct fsl_edma_drvdata *drvdata = NULL;
 	u32 chan_mask[2] = {0, 0};
+	char clk_name[36];
 	struct edma_regs *regs;
 	int chans;
 	int ret, i;
@@ -476,14 +486,6 @@ static int fsl_edma_probe(struct platform_device *pdev)
 		}
 	}
 
-	if (drvdata->flags & FSL_EDMA_DRV_HAS_CHCLK) {
-		fsl_edma->chclk = devm_clk_get_enabled(&pdev->dev, "mp");
-		if (IS_ERR(fsl_edma->chclk)) {
-			dev_err(&pdev->dev, "Missing MP block clock.\n");
-			return PTR_ERR(fsl_edma->chclk);
-		}
-	}
-
 	ret = of_property_read_variable_u32_array(np, "dma-channel-mask", chan_mask, 1, 2);
 
 	if (ret > 0) {
@@ -540,7 +542,7 @@ static int fsl_edma_probe(struct platform_device *pdev)
 
 		fsl_chan->edma = fsl_edma;
 		fsl_chan->pm_state = RUNNING;
-		fsl_chan->slave_id = 0;
+		fsl_chan->srcid = 0;
 		fsl_chan->idle = true;
 		fsl_chan->dma_dir = DMA_NONE;
 		fsl_chan->vchan.desc_free = fsl_edma_free_desc;
@@ -551,11 +553,21 @@ static int fsl_edma_probe(struct platform_device *pdev)
 				+ i * drvdata->chreg_space_sz + drvdata->chreg_off + len;
 		fsl_chan->mux_addr = fsl_edma->membase + drvdata->mux_off + i * drvdata->mux_skip;
 
+		if (drvdata->flags & FSL_EDMA_DRV_HAS_CHCLK) {
+			snprintf(clk_name, sizeof(clk_name), "ch%02d", i);
+			fsl_chan->clk = devm_clk_get_enabled(&pdev->dev,
+							     (const char *)clk_name);
+
+			if (IS_ERR(fsl_chan->clk))
+				return PTR_ERR(fsl_chan->clk);
+		}
 		fsl_chan->pdev = pdev;
 		vchan_init(&fsl_chan->vchan, &fsl_edma->dma_dev);
 
 		edma_write_tcdreg(fsl_chan, cpu_to_le32(0), csr);
 		fsl_edma_chan_mux(fsl_chan, 0, false);
+		if (fsl_chan->edma->drvdata->flags & FSL_EDMA_DRV_HAS_CHCLK)
+			clk_disable_unprepare(fsl_chan->clk);
 	}
 
 	ret = fsl_edma->drvdata->setup_irq(pdev, fsl_edma);
@@ -682,8 +694,8 @@ static int fsl_edma_resume_early(struct device *dev)
 			continue;
 		fsl_chan->pm_state = RUNNING;
 		edma_write_tcdreg(fsl_chan, 0, csr);
-		if (fsl_chan->slave_id != 0)
-			fsl_edma_chan_mux(fsl_chan, fsl_chan->slave_id, true);
+		if (fsl_chan->srcid != 0)
+			fsl_edma_chan_mux(fsl_chan, fsl_chan->srcid, true);
 	}
 
 	if (!(fsl_edma->drvdata->flags & FSL_EDMA_DRV_SPLIT_REG))
diff --git a/drivers/dma/fsl-edma-trace.c b/drivers/dma/fsl-edma-trace.c
new file mode 100644
index 000000000000..28300ad80bb7
--- /dev/null
+++ b/drivers/dma/fsl-edma-trace.c
@@ -0,0 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#define CREATE_TRACE_POINTS
+#include "fsl-edma-common.h"
diff --git a/drivers/dma/fsl-edma-trace.h b/drivers/dma/fsl-edma-trace.h
new file mode 100644
index 000000000000..d3541301a247
--- /dev/null
+++ b/drivers/dma/fsl-edma-trace.h
@@ -0,0 +1,132 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright 2023 NXP.
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM fsl_edma
+
+#if !defined(__LINUX_FSL_EDMA_TRACE) || defined(TRACE_HEADER_MULTI_READ)
+#define __LINUX_FSL_EDMA_TRACE
+
+#include <linux/types.h>
+#include <linux/tracepoint.h>
+
+DECLARE_EVENT_CLASS(edma_log_io,
+	TP_PROTO(struct fsl_edma_engine *edma, void __iomem *addr, u32 value),
+	TP_ARGS(edma, addr, value),
+	TP_STRUCT__entry(
+		__field(struct fsl_edma_engine *, edma)
+		__field(void __iomem *, addr)
+		__field(u32, value)
+	),
+	TP_fast_assign(
+		__entry->edma = edma;
+		__entry->addr = addr;
+		__entry->value = value;
+	),
+	TP_printk("offset %08x: value %08x",
+		(u32)(__entry->addr - __entry->edma->membase), __entry->value)
+);
+
+DEFINE_EVENT(edma_log_io, edma_readl,
+	TP_PROTO(struct fsl_edma_engine *edma, void __iomem *addr, u32 value),
+	TP_ARGS(edma, addr, value)
+);
+
+DEFINE_EVENT(edma_log_io, edma_writel,
+	TP_PROTO(struct fsl_edma_engine *edma, void __iomem *addr,  u32 value),
+	TP_ARGS(edma, addr, value)
+);
+
+DEFINE_EVENT(edma_log_io, edma_readw,
+	TP_PROTO(struct fsl_edma_engine *edma, void __iomem *addr, u32 value),
+	TP_ARGS(edma, addr, value)
+);
+
+DEFINE_EVENT(edma_log_io, edma_writew,
+	TP_PROTO(struct fsl_edma_engine *edma, void __iomem *addr,  u32 value),
+	TP_ARGS(edma, addr, value)
+);
+
+DEFINE_EVENT(edma_log_io, edma_readb,
+	TP_PROTO(struct fsl_edma_engine *edma, void __iomem *addr, u32 value),
+	TP_ARGS(edma, addr, value)
+);
+
+DEFINE_EVENT(edma_log_io, edma_writeb,
+	TP_PROTO(struct fsl_edma_engine *edma, void __iomem *addr,  u32 value),
+	TP_ARGS(edma, addr, value)
+);
+
+DECLARE_EVENT_CLASS(edma_log_tcd,
+	TP_PROTO(struct fsl_edma_chan *chan, void *tcd),
+	TP_ARGS(chan, tcd),
+	TP_STRUCT__entry(
+		__field(u64, saddr)
+		__field(u16, soff)
+		__field(u16, attr)
+		__field(u32, nbytes)
+		__field(u64, slast)
+		__field(u64, daddr)
+		__field(u16, doff)
+		__field(u16, citer)
+		__field(u64, dlast_sga)
+		__field(u16, csr)
+		__field(u16, biter)
+
+	),
+	TP_fast_assign(
+		__entry->saddr = fsl_edma_get_tcd_to_cpu(chan, tcd, saddr),
+		__entry->soff = fsl_edma_get_tcd_to_cpu(chan, tcd, soff),
+		__entry->attr = fsl_edma_get_tcd_to_cpu(chan, tcd, attr),
+		__entry->nbytes = fsl_edma_get_tcd_to_cpu(chan, tcd, nbytes),
+		__entry->slast = fsl_edma_get_tcd_to_cpu(chan, tcd, slast),
+		__entry->daddr = fsl_edma_get_tcd_to_cpu(chan, tcd, daddr),
+		__entry->doff = fsl_edma_get_tcd_to_cpu(chan, tcd, doff),
+		__entry->citer = fsl_edma_get_tcd_to_cpu(chan, tcd, citer),
+		__entry->dlast_sga = fsl_edma_get_tcd_to_cpu(chan, tcd, dlast_sga),
+		__entry->csr = fsl_edma_get_tcd_to_cpu(chan, tcd, csr),
+		__entry->biter = fsl_edma_get_tcd_to_cpu(chan, tcd, biter);
+	),
+	TP_printk("\n==== TCD =====\n"
+		  "  saddr:  0x%016llx\n"
+		  "  soff:               0x%04x\n"
+		  "  attr:               0x%04x\n"
+		  "  nbytes:         0x%08x\n"
+		  "  slast:  0x%016llx\n"
+		  "  daddr:  0x%016llx\n"
+		  "  doff:               0x%04x\n"
+		  "  citer:              0x%04x\n"
+		  "  dlast:  0x%016llx\n"
+		  "  csr:                0x%04x\n"
+		  "  biter:              0x%04x\n",
+		__entry->saddr,
+		__entry->soff,
+		__entry->attr,
+		__entry->nbytes,
+		__entry->slast,
+		__entry->daddr,
+		__entry->doff,
+		__entry->citer,
+		__entry->dlast_sga,
+		__entry->csr,
+		__entry->biter)
+);
+
+DEFINE_EVENT(edma_log_tcd, edma_fill_tcd,
+	TP_PROTO(struct fsl_edma_chan *chan, void *tcd),
+	TP_ARGS(chan, tcd)
+);
+
+#endif
+
+/* this part must be outside header guard */
+
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE fsl-edma-trace
+
+#include <trace/define_trace.h>
diff --git a/drivers/dma/idma64.c b/drivers/dma/idma64.c
index 1398814d8fbb..e3505e56784b 100644
--- a/drivers/dma/idma64.c
+++ b/drivers/dma/idma64.c
@@ -598,7 +598,9 @@ static int idma64_probe(struct idma64_chip *chip)
 
 	idma64->dma.dev = chip->sysdev;
 
-	dma_set_max_seg_size(idma64->dma.dev, IDMA64C_CTLH_BLOCK_TS_MASK);
+	ret = dma_set_max_seg_size(idma64->dma.dev, IDMA64C_CTLH_BLOCK_TS_MASK);
+	if (ret)
+		return ret;
 
 	ret = dma_async_device_register(&idma64->dma);
 	if (ret)
diff --git a/drivers/dma/idxd/cdev.c b/drivers/dma/idxd/cdev.c
index 39935071174a..57f1bf2ab20b 100644
--- a/drivers/dma/idxd/cdev.c
+++ b/drivers/dma/idxd/cdev.c
@@ -577,7 +577,6 @@ void idxd_wq_del_cdev(struct idxd_wq *wq)
 	struct idxd_cdev *idxd_cdev;
 
 	idxd_cdev = wq->idxd_cdev;
-	ida_destroy(&file_ida);
 	wq->idxd_cdev = NULL;
 	cdev_device_del(&idxd_cdev->cdev, cdev_dev(idxd_cdev));
 	put_device(cdev_dev(idxd_cdev));
@@ -593,6 +592,14 @@ static int idxd_user_drv_probe(struct idxd_dev *idxd_dev)
 	if (idxd->state != IDXD_DEV_ENABLED)
 		return -ENXIO;
 
+	mutex_lock(&wq->wq_lock);
+
+	if (!idxd_wq_driver_name_match(wq, dev)) {
+		idxd->cmd_status = IDXD_SCMD_WQ_NO_DRV_NAME;
+		rc = -ENODEV;
+		goto wq_err;
+	}
+
 	/*
 	 * User type WQ is enabled only when SVA is enabled for two reasons:
 	 *   - If no IOMMU or IOMMU Passthrough without SVA, userspace
@@ -608,14 +615,7 @@ static int idxd_user_drv_probe(struct idxd_dev *idxd_dev)
 		dev_dbg(&idxd->pdev->dev,
 			"User type WQ cannot be enabled without SVA.\n");
 
-		return -EOPNOTSUPP;
-	}
-
-	mutex_lock(&wq->wq_lock);
-
-	if (!idxd_wq_driver_name_match(wq, dev)) {
-		idxd->cmd_status = IDXD_SCMD_WQ_NO_DRV_NAME;
-		rc = -ENODEV;
+		rc = -EOPNOTSUPP;
 		goto wq_err;
 	}
 
diff --git a/drivers/dma/imx-sdma.c b/drivers/dma/imx-sdma.c
index 9b42f5e96b1e..003e1580b902 100644
--- a/drivers/dma/imx-sdma.c
+++ b/drivers/dma/imx-sdma.c
@@ -24,6 +24,7 @@
 #include <linux/semaphore.h>
 #include <linux/spinlock.h>
 #include <linux/device.h>
+#include <linux/genalloc.h>
 #include <linux/dma-mapping.h>
 #include <linux/firmware.h>
 #include <linux/slab.h>
@@ -137,7 +138,11 @@
  *						0: Source on AIPS
  *	12		Destination Bit(DP)	1: Destination on SPBA
  *						0: Destination on AIPS
- *	13-15		---------		MUST BE 0
+ *	13		Source FIFO		1: Source is dual FIFO
+ *						0: Source is single FIFO
+ *	14		Destination FIFO	1: Destination is dual FIFO
+ *						0: Destination is single FIFO
+ *	15		---------		MUST BE 0
  *	16-23		Higher WML		HWML
  *	24-27		N			Total number of samples after
  *						which Pad adding/Swallowing
@@ -168,6 +173,8 @@
 #define SDMA_WATERMARK_LEVEL_SPDIF	BIT(10)
 #define SDMA_WATERMARK_LEVEL_SP		BIT(11)
 #define SDMA_WATERMARK_LEVEL_DP		BIT(12)
+#define SDMA_WATERMARK_LEVEL_SD		BIT(13)
+#define SDMA_WATERMARK_LEVEL_DD		BIT(14)
 #define SDMA_WATERMARK_LEVEL_HWML	(0xFF << 16)
 #define SDMA_WATERMARK_LEVEL_LWE	BIT(28)
 #define SDMA_WATERMARK_LEVEL_HWE	BIT(29)
@@ -175,6 +182,7 @@
 
 #define SDMA_DMA_BUSWIDTHS	(BIT(DMA_SLAVE_BUSWIDTH_1_BYTE) | \
 				 BIT(DMA_SLAVE_BUSWIDTH_2_BYTES) | \
+				 BIT(DMA_SLAVE_BUSWIDTH_3_BYTES) | \
 				 BIT(DMA_SLAVE_BUSWIDTH_4_BYTES))
 
 #define SDMA_DMA_DIRECTIONS	(BIT(DMA_DEV_TO_MEM) | \
@@ -232,20 +240,23 @@ struct sdma_script_start_addrs {
 	s32 utra_addr;
 	s32 ram_code_start_addr;
 	/* End of v1 array */
-	s32 mcu_2_ssish_addr;
+	union {	s32 v1_end; s32 mcu_2_ssish_addr; };
 	s32 ssish_2_mcu_addr;
 	s32 hdmi_dma_addr;
 	/* End of v2 array */
-	s32 zcanfd_2_mcu_addr;
+	union { s32 v2_end; s32 zcanfd_2_mcu_addr; };
 	s32 zqspi_2_mcu_addr;
 	s32 mcu_2_ecspi_addr;
 	s32 mcu_2_sai_addr;
 	s32 sai_2_mcu_addr;
 	s32 uart_2_mcu_rom_addr;
 	s32 uartsh_2_mcu_rom_addr;
+	s32 i2c_2_mcu_addr;
+	s32 mcu_2_i2c_addr;
 	/* End of v3 array */
-	s32 mcu_2_zqspi_addr;
+	union { s32 v3_end; s32 mcu_2_zqspi_addr; };
 	/* End of v4 array */
+	s32 v4_end[0];
 };
 
 /*
@@ -531,6 +542,7 @@ struct sdma_engine {
 	/* clock ratio for AHB:SDMA core. 1:1 is 1, 2:1 is 0*/
 	bool				clk_ratio;
 	bool                            fw_loaded;
+	struct gen_pool			*iram_pool;
 };
 
 static int sdma_config_write(struct dma_chan *chan,
@@ -1072,6 +1084,11 @@ static int sdma_get_pc(struct sdma_channel *sdmac,
 		per_2_emi = sdma->script_addrs->sai_2_mcu_addr;
 		emi_2_per = sdma->script_addrs->mcu_2_sai_addr;
 		break;
+	case IMX_DMATYPE_I2C:
+		per_2_emi = sdma->script_addrs->i2c_2_mcu_addr;
+		emi_2_per = sdma->script_addrs->mcu_2_i2c_addr;
+		sdmac->is_ram_script = true;
+		break;
 	case IMX_DMATYPE_HDMI:
 		emi_2_per = sdma->script_addrs->hdmi_dma_addr;
 		sdmac->is_ram_script = true;
@@ -1255,6 +1272,16 @@ static void sdma_set_watermarklevel_for_p2p(struct sdma_channel *sdmac)
 		sdmac->watermark_level |= SDMA_WATERMARK_LEVEL_DP;
 
 	sdmac->watermark_level |= SDMA_WATERMARK_LEVEL_CONT;
+
+	/*
+	 * Limitation: The p2p script support dual fifos in maximum,
+	 * So when fifo number is larger than 1, force enable dual
+	 * fifos.
+	 */
+	if (sdmac->n_fifos_src > 1)
+		sdmac->watermark_level |= SDMA_WATERMARK_LEVEL_SD;
+	if (sdmac->n_fifos_dst > 1)
+		sdmac->watermark_level |= SDMA_WATERMARK_LEVEL_DD;
 }
 
 static void sdma_set_watermarklevel_for_sais(struct sdma_channel *sdmac)
@@ -1358,8 +1385,14 @@ static int sdma_request_channel0(struct sdma_engine *sdma)
 {
 	int ret = -EBUSY;
 
-	sdma->bd0 = dma_alloc_coherent(sdma->dev, PAGE_SIZE, &sdma->bd0_phys,
-				       GFP_NOWAIT);
+	if (sdma->iram_pool)
+		sdma->bd0 = gen_pool_dma_alloc(sdma->iram_pool,
+					sizeof(struct sdma_buffer_descriptor),
+					&sdma->bd0_phys);
+	else
+		sdma->bd0 = dma_alloc_coherent(sdma->dev,
+					sizeof(struct sdma_buffer_descriptor),
+					&sdma->bd0_phys, GFP_NOWAIT);
 	if (!sdma->bd0) {
 		ret = -ENOMEM;
 		goto out;
@@ -1379,10 +1412,14 @@ out:
 static int sdma_alloc_bd(struct sdma_desc *desc)
 {
 	u32 bd_size = desc->num_bd * sizeof(struct sdma_buffer_descriptor);
+	struct sdma_engine *sdma = desc->sdmac->sdma;
 	int ret = 0;
 
-	desc->bd = dma_alloc_coherent(desc->sdmac->sdma->dev, bd_size,
-				      &desc->bd_phys, GFP_NOWAIT);
+	if (sdma->iram_pool)
+		desc->bd = gen_pool_dma_alloc(sdma->iram_pool, bd_size, &desc->bd_phys);
+	else
+		desc->bd = dma_alloc_coherent(sdma->dev, bd_size, &desc->bd_phys, GFP_NOWAIT);
+
 	if (!desc->bd) {
 		ret = -ENOMEM;
 		goto out;
@@ -1394,9 +1431,12 @@ out:
 static void sdma_free_bd(struct sdma_desc *desc)
 {
 	u32 bd_size = desc->num_bd * sizeof(struct sdma_buffer_descriptor);
+	struct sdma_engine *sdma = desc->sdmac->sdma;
 
-	dma_free_coherent(desc->sdmac->sdma->dev, bd_size, desc->bd,
-			  desc->bd_phys);
+	if (sdma->iram_pool)
+		gen_pool_free(sdma->iram_pool, (unsigned long)desc->bd, bd_size);
+	else
+		dma_free_coherent(desc->sdmac->sdma->dev, bd_size, desc->bd, desc->bd_phys);
 }
 
 static void sdma_desc_free(struct virt_dma_desc *vd)
@@ -1643,6 +1683,9 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg(
 			if (count & 3 || sg->dma_address & 3)
 				goto err_bd_out;
 			break;
+		case DMA_SLAVE_BUSWIDTH_3_BYTES:
+			bd->mode.command = 3;
+			break;
 		case DMA_SLAVE_BUSWIDTH_2_BYTES:
 			bd->mode.command = 2;
 			if (count & 1 || sg->dma_address & 1)
@@ -1880,10 +1923,17 @@ static void sdma_issue_pending(struct dma_chan *chan)
 	spin_unlock_irqrestore(&sdmac->vc.lock, flags);
 }
 
-#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1	34
-#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V2	38
-#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V3	45
-#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V4	46
+#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1	\
+(offsetof(struct sdma_script_start_addrs, v1_end) / sizeof(s32))
+
+#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V2 \
+(offsetof(struct sdma_script_start_addrs, v2_end) / sizeof(s32))
+
+#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V3 \
+(offsetof(struct sdma_script_start_addrs, v3_end) / sizeof(s32))
+
+#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V4 \
+(offsetof(struct sdma_script_start_addrs, v4_end) / sizeof(s32))
 
 static void sdma_add_scripts(struct sdma_engine *sdma,
 			     const struct sdma_script_start_addrs *addr)
@@ -2068,6 +2118,7 @@ static int sdma_init(struct sdma_engine *sdma)
 {
 	int i, ret;
 	dma_addr_t ccb_phys;
+	int ccbsize;
 
 	ret = clk_enable(sdma->clk_ipg);
 	if (ret)
@@ -2083,10 +2134,14 @@ static int sdma_init(struct sdma_engine *sdma)
 	/* Be sure SDMA has not started yet */
 	writel_relaxed(0, sdma->regs + SDMA_H_C0PTR);
 
-	sdma->channel_control = dma_alloc_coherent(sdma->dev,
-			MAX_DMA_CHANNELS * sizeof(struct sdma_channel_control) +
-			sizeof(struct sdma_context_data),
-			&ccb_phys, GFP_KERNEL);
+	ccbsize = MAX_DMA_CHANNELS * (sizeof(struct sdma_channel_control)
+		  + sizeof(struct sdma_context_data));
+
+	if (sdma->iram_pool)
+		sdma->channel_control = gen_pool_dma_alloc(sdma->iram_pool, ccbsize, &ccb_phys);
+	else
+		sdma->channel_control = dma_alloc_coherent(sdma->dev, ccbsize, &ccb_phys,
+							   GFP_KERNEL);
 
 	if (!sdma->channel_control) {
 		ret = -ENOMEM;
@@ -2272,6 +2327,12 @@ static int sdma_probe(struct platform_device *pdev)
 			vchan_init(&sdmac->vc, &sdma->dma_device);
 	}
 
+	if (np) {
+		sdma->iram_pool = of_gen_pool_get(np, "iram", 0);
+		if (sdma->iram_pool)
+			dev_info(&pdev->dev, "alloc bd from iram.\n");
+	}
+
 	ret = sdma_init(sdma);
 	if (ret)
 		goto err_init;
diff --git a/drivers/dma/mcf-edma-main.c b/drivers/dma/mcf-edma-main.c
index dba631783876..78c606f6d002 100644
--- a/drivers/dma/mcf-edma-main.c
+++ b/drivers/dma/mcf-edma-main.c
@@ -195,7 +195,7 @@ static int mcf_edma_probe(struct platform_device *pdev)
 		struct fsl_edma_chan *mcf_chan = &mcf_edma->chans[i];
 
 		mcf_chan->edma = mcf_edma;
-		mcf_chan->slave_id = i;
+		mcf_chan->srcid = i;
 		mcf_chan->idle = true;
 		mcf_chan->dma_dir = DMA_NONE;
 		mcf_chan->vchan.desc_free = fsl_edma_free_desc;
@@ -277,7 +277,7 @@ bool mcf_edma_filter_fn(struct dma_chan *chan, void *param)
 	if (chan->device->dev->driver == &mcf_edma_driver.driver) {
 		struct fsl_edma_chan *mcf_chan = to_fsl_edma_chan(chan);
 
-		return (mcf_chan->slave_id == (uintptr_t)param);
+		return (mcf_chan->srcid == (uintptr_t)param);
 	}
 
 	return false;
diff --git a/drivers/dma/pch_dma.c b/drivers/dma/pch_dma.c
index c359decc07a3..6b2793b07694 100644
--- a/drivers/dma/pch_dma.c
+++ b/drivers/dma/pch_dma.c
@@ -155,11 +155,6 @@ static inline struct device *chan2dev(struct dma_chan *chan)
 	return &chan->dev->device;
 }
 
-static inline struct device *chan2parent(struct dma_chan *chan)
-{
-	return chan->dev->device.parent;
-}
-
 static inline
 struct pch_dma_desc *pdc_first_active(struct pch_dma_chan *pd_chan)
 {
diff --git a/drivers/dma/qcom/hidma.c b/drivers/dma/qcom/hidma.c
index 202ac95227cb..721b4ac0857a 100644
--- a/drivers/dma/qcom/hidma.c
+++ b/drivers/dma/qcom/hidma.c
@@ -50,7 +50,6 @@
 #include <linux/platform_device.h>
 #include <linux/slab.h>
 #include <linux/spinlock.h>
-#include <linux/of_dma.h>
 #include <linux/property.h>
 #include <linux/delay.h>
 #include <linux/acpi.h>
@@ -947,22 +946,12 @@ static const struct acpi_device_id hidma_acpi_ids[] = {
 MODULE_DEVICE_TABLE(acpi, hidma_acpi_ids);
 #endif
 
-static const struct of_device_id hidma_match[] = {
-	{.compatible = "qcom,hidma-1.0",},
-	{.compatible = "qcom,hidma-1.1", .data = (void *)(HIDMA_MSI_CAP),},
-	{.compatible = "qcom,hidma-1.2",
-	 .data = (void *)(HIDMA_MSI_CAP | HIDMA_IDENTITY_CAP),},
-	{},
-};
-MODULE_DEVICE_TABLE(of, hidma_match);
-
 static struct platform_driver hidma_driver = {
 	.probe = hidma_probe,
 	.remove_new = hidma_remove,
 	.shutdown = hidma_shutdown,
 	.driver = {
 		   .name = "hidma",
-		   .of_match_table = hidma_match,
 		   .acpi_match_table = ACPI_PTR(hidma_acpi_ids),
 	},
 };
diff --git a/drivers/dma/qcom/hidma_mgmt.c b/drivers/dma/qcom/hidma_mgmt.c
index 1d675f31252b..bb883e138ebf 100644
--- a/drivers/dma/qcom/hidma_mgmt.c
+++ b/drivers/dma/qcom/hidma_mgmt.c
@@ -7,12 +7,7 @@
 
 #include <linux/dmaengine.h>
 #include <linux/acpi.h>
-#include <linux/of.h>
 #include <linux/property.h>
-#include <linux/of_address.h>
-#include <linux/of_irq.h>
-#include <linux/of_platform.h>
-#include <linux/of_device.h>
 #include <linux/platform_device.h>
 #include <linux/module.h>
 #include <linux/uaccess.h>
@@ -327,115 +322,13 @@ static const struct acpi_device_id hidma_mgmt_acpi_ids[] = {
 MODULE_DEVICE_TABLE(acpi, hidma_mgmt_acpi_ids);
 #endif
 
-static const struct of_device_id hidma_mgmt_match[] = {
-	{.compatible = "qcom,hidma-mgmt-1.0",},
-	{},
-};
-MODULE_DEVICE_TABLE(of, hidma_mgmt_match);
-
 static struct platform_driver hidma_mgmt_driver = {
 	.probe = hidma_mgmt_probe,
 	.driver = {
 		   .name = "hidma-mgmt",
-		   .of_match_table = hidma_mgmt_match,
 		   .acpi_match_table = ACPI_PTR(hidma_mgmt_acpi_ids),
 	},
 };
 
-#if defined(CONFIG_OF) && defined(CONFIG_OF_IRQ)
-static int object_counter;
-
-static int __init hidma_mgmt_of_populate_channels(struct device_node *np)
-{
-	struct platform_device *pdev_parent = of_find_device_by_node(np);
-	struct platform_device_info pdevinfo;
-	struct device_node *child;
-	struct resource *res;
-	int ret = 0;
-
-	/* allocate a resource array */
-	res = kcalloc(3, sizeof(*res), GFP_KERNEL);
-	if (!res)
-		return -ENOMEM;
-
-	for_each_available_child_of_node(np, child) {
-		struct platform_device *new_pdev;
-
-		ret = of_address_to_resource(child, 0, &res[0]);
-		if (!ret)
-			goto out;
-
-		ret = of_address_to_resource(child, 1, &res[1]);
-		if (!ret)
-			goto out;
-
-		ret = of_irq_to_resource(child, 0, &res[2]);
-		if (ret <= 0)
-			goto out;
-
-		memset(&pdevinfo, 0, sizeof(pdevinfo));
-		pdevinfo.fwnode = &child->fwnode;
-		pdevinfo.parent = pdev_parent ? &pdev_parent->dev : NULL;
-		pdevinfo.name = child->name;
-		pdevinfo.id = object_counter++;
-		pdevinfo.res = res;
-		pdevinfo.num_res = 3;
-		pdevinfo.data = NULL;
-		pdevinfo.size_data = 0;
-		pdevinfo.dma_mask = DMA_BIT_MASK(64);
-		new_pdev = platform_device_register_full(&pdevinfo);
-		if (IS_ERR(new_pdev)) {
-			ret = PTR_ERR(new_pdev);
-			goto out;
-		}
-		new_pdev->dev.of_node = child;
-		of_dma_configure(&new_pdev->dev, child, true);
-		/*
-		 * It is assumed that calling of_msi_configure is safe on
-		 * platforms with or without MSI support.
-		 */
-		of_msi_configure(&new_pdev->dev, child);
-	}
-
-	kfree(res);
-
-	return ret;
-
-out:
-	of_node_put(child);
-	kfree(res);
-
-	return ret;
-}
-#endif
-
-static int __init hidma_mgmt_init(void)
-{
-#if defined(CONFIG_OF) && defined(CONFIG_OF_IRQ)
-	struct device_node *child;
-
-	for_each_matching_node(child, hidma_mgmt_match) {
-		/* device tree based firmware here */
-		hidma_mgmt_of_populate_channels(child);
-	}
-#endif
-	/*
-	 * We do not check for return value here, as it is assumed that
-	 * platform_driver_register must not fail. The reason for this is that
-	 * the (potential) hidma_mgmt_of_populate_channels calls above are not
-	 * cleaned up if it does fail, and to do this work is quite
-	 * complicated. In particular, various calls of of_address_to_resource,
-	 * of_irq_to_resource, platform_device_register_full, of_dma_configure,
-	 * and of_msi_configure which then call other functions and so on, must
-	 * be cleaned up - this is not a trivial exercise.
-	 *
-	 * Currently, this module is not intended to be unloaded, and there is
-	 * no module_exit function defined which does the needed cleanup. For
-	 * this reason, we have to assume success here.
-	 */
-	platform_driver_register(&hidma_mgmt_driver);
-
-	return 0;
-}
-module_init(hidma_mgmt_init);
+module_platform_driver(hidma_mgmt_driver);
 MODULE_LICENSE("GPL v2");
diff --git a/drivers/dma/virt-dma.h b/drivers/dma/virt-dma.h
index e9f5250fbe4d..59d9eabc8b67 100644
--- a/drivers/dma/virt-dma.h
+++ b/drivers/dma/virt-dma.h
@@ -81,6 +81,8 @@ static inline struct dma_async_tx_descriptor *vchan_tx_prep(struct virt_dma_chan
  */
 static inline bool vchan_issue_pending(struct virt_dma_chan *vc)
 {
+	lockdep_assert_held(&vc->lock);
+
 	list_splice_tail_init(&vc->desc_submitted, &vc->desc_issued);
 	return !list_empty(&vc->desc_issued);
 }
@@ -96,6 +98,8 @@ static inline void vchan_cookie_complete(struct virt_dma_desc *vd)
 	struct virt_dma_chan *vc = to_virt_chan(vd->tx.chan);
 	dma_cookie_t cookie;
 
+	lockdep_assert_held(&vc->lock);
+
 	cookie = vd->tx.cookie;
 	dma_cookie_complete(&vd->tx);
 	dev_vdbg(vc->chan.device->dev, "txd %p[%x]: marked complete\n",
@@ -146,6 +150,8 @@ static inline void vchan_terminate_vdesc(struct virt_dma_desc *vd)
 {
 	struct virt_dma_chan *vc = to_virt_chan(vd->tx.chan);
 
+	lockdep_assert_held(&vc->lock);
+
 	list_add_tail(&vd->node, &vc->desc_terminated);
 
 	if (vc->cyclic == vd)
@@ -160,6 +166,8 @@ static inline void vchan_terminate_vdesc(struct virt_dma_desc *vd)
  */
 static inline struct virt_dma_desc *vchan_next_desc(struct virt_dma_chan *vc)
 {
+	lockdep_assert_held(&vc->lock);
+
 	return list_first_entry_or_null(&vc->desc_issued,
 					struct virt_dma_desc, node);
 }
@@ -177,6 +185,8 @@ static inline struct virt_dma_desc *vchan_next_desc(struct virt_dma_chan *vc)
 static inline void vchan_get_all_descriptors(struct virt_dma_chan *vc,
 	struct list_head *head)
 {
+	lockdep_assert_held(&vc->lock);
+
 	list_splice_tail_init(&vc->desc_allocated, head);
 	list_splice_tail_init(&vc->desc_submitted, head);
 	list_splice_tail_init(&vc->desc_issued, head);
diff --git a/drivers/dma/xilinx/xdma.c b/drivers/dma/xilinx/xdma.c
index 313b217388fe..e143a7330816 100644
--- a/drivers/dma/xilinx/xdma.c
+++ b/drivers/dma/xilinx/xdma.c
@@ -1307,6 +1307,7 @@ static const struct platform_device_id xdma_id_table[] = {
 	{ "xdma", 0},
 	{ },
 };
+MODULE_DEVICE_TABLE(platform, xdma_id_table);
 
 static struct platform_driver xdma_driver = {
 	.driver		= {
diff --git a/drivers/dma/xilinx/xilinx_dpdma.c b/drivers/dma/xilinx/xilinx_dpdma.c
index eb0637d90342..36bd4825d389 100644
--- a/drivers/dma/xilinx/xilinx_dpdma.c
+++ b/drivers/dma/xilinx/xilinx_dpdma.c
@@ -1043,9 +1043,8 @@ static int xilinx_dpdma_chan_stop(struct xilinx_dpdma_chan *chan)
 static void xilinx_dpdma_chan_done_irq(struct xilinx_dpdma_chan *chan)
 {
 	struct xilinx_dpdma_tx_desc *active;
-	unsigned long flags;
 
-	spin_lock_irqsave(&chan->lock, flags);
+	spin_lock(&chan->lock);
 
 	xilinx_dpdma_debugfs_desc_done_irq(chan);
 
@@ -1057,7 +1056,7 @@ static void xilinx_dpdma_chan_done_irq(struct xilinx_dpdma_chan *chan)
 			 "chan%u: DONE IRQ with no active descriptor!\n",
 			 chan->id);
 
-	spin_unlock_irqrestore(&chan->lock, flags);
+	spin_unlock(&chan->lock);
 }
 
 /**
@@ -1072,10 +1071,9 @@ static void xilinx_dpdma_chan_vsync_irq(struct  xilinx_dpdma_chan *chan)
 {
 	struct xilinx_dpdma_tx_desc *pending;
 	struct xilinx_dpdma_sw_desc *sw_desc;
-	unsigned long flags;
 	u32 desc_id;
 
-	spin_lock_irqsave(&chan->lock, flags);
+	spin_lock(&chan->lock);
 
 	pending = chan->desc.pending;
 	if (!chan->running || !pending)
@@ -1108,7 +1106,7 @@ static void xilinx_dpdma_chan_vsync_irq(struct  xilinx_dpdma_chan *chan)
 	spin_unlock(&chan->vchan.lock);
 
 out:
-	spin_unlock_irqrestore(&chan->lock, flags);
+	spin_unlock(&chan->lock);
 }
 
 /**