Merge tag 'dmaengine-4.2-rc1' of git://git.infradead.org/users/vkoul/slave-dma

Pull dmaengine updates from Vinod Koul:
 "This time we have support for few new devices, few new features and
  odd fixes spread thru the subsystem.

  New devices added:
   - support for CSRatlas7 dma controller
   - Allwinner H3(sun8i) controller
   - TI DMA crossbar driver on DRA7x
   - new pxa driver

  New features added:
   - memset support is bought back now that we have a user in xdmac controller
   - interleaved transfers support different source and destination strides
   - supporting DMA routers and configuration thru DT
   - support for reusing descriptors
   - xdmac memset and interleaved transfer support
   - hdmac support for interleaved transfers
   - omap-dma support for memcpy

  Others:
   - Constify platform_device_id
   - mv_xor fixes and improvements"

* tag 'dmaengine-4.2-rc1' of git://git.infradead.org/users/vkoul/slave-dma: (46 commits)
  dmaengine: xgene: fix file permission
  dmaengine: fsl-edma: clear pending interrupts on initialization
  dmaengine: xdmac: Add memset support
  Documentation: dmaengine: document DMA_CTRL_ACK
  dmaengine: virt-dma: don't always free descriptor upon completion
  dmaengine: Revert "drivers/dma: remove unused support for MEMSET operations"
  dmaengine: hdmac: Implement interleaved transfers
  dmaengine: Move icg helpers to global header
  dmaengine: mv_xor: improve descriptors list handling and reduce locking
  dmaengine: mv_xor: Enlarge descriptor pool size
  dmaengine: mv_xor: add support for a38x command in descriptor mode
  dmaengine: mv_xor: Rename function for consistent naming
  dmaengine: mv_xor: bug fix for racing condition in descriptors cleanup
  dmaengine: pl330: fix wording in mcbufsz message
  dmaengine: sirf: add CSRatlas7 SoC support
  dmaengine: xgene-dma: Fix "incorrect type in assignement" warnings
  dmaengine: fix kernel-doc documentation
  dmaengine: pxa_dma: add support for legacy transition
  dmaengine: pxa_dma: add debug information
  dmaengine: pxa: add pxa dmaengine driver
  ...

28 files changed, 2958 insertions, 455 deletions

diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig
index bda2cb06dc7a..88d474b78076 100644
--- a/drivers/dma/Kconfig
+++ b/drivers/dma/Kconfig
@@ -162,6 +162,17 @@ config MX3_IPU_IRQS
 	  To avoid bloating the irq_desc[] array we allocate a sufficient
 	  number of IRQ slots and map them dynamically to specific sources.
 
+config PXA_DMA
+	bool "PXA DMA support"
+	depends on (ARCH_MMP || ARCH_PXA)
+	select DMA_ENGINE
+	select DMA_VIRTUAL_CHANNELS
+	help
+	  Support the DMA engine for PXA. It is also compatible with MMP PDMA
+	  platform. The internal DMA IP of all PXA variants is supported, with
+	  16 to 32 channels for peripheral to memory or memory to memory
+	  transfers.
+
 config TXX9_DMAC
 	tristate "Toshiba TXx9 SoC DMA support"
 	depends on MACH_TX49XX || MACH_TX39XX
@@ -245,6 +256,9 @@ config TI_EDMA
 	  Enable support for the TI EDMA controller. This DMA
 	  engine is found on TI DaVinci and AM33xx parts.
 
+config TI_DMA_CROSSBAR
+	bool
+
 config ARCH_HAS_ASYNC_TX_FIND_CHANNEL
 	bool
 
@@ -330,6 +344,7 @@ config DMA_OMAP
 	depends on ARCH_OMAP
 	select DMA_ENGINE
 	select DMA_VIRTUAL_CHANNELS
+	select TI_DMA_CROSSBAR if SOC_DRA7XX
 
 config DMA_BCM2835
 	tristate "BCM2835 DMA engine support"
diff --git a/drivers/dma/Makefile b/drivers/dma/Makefile
index 69f77d5ba53b..6a4d6f2827da 100644
--- a/drivers/dma/Makefile
+++ b/drivers/dma/Makefile
@@ -25,6 +25,7 @@ obj-$(CONFIG_AMCC_PPC440SPE_ADMA) += ppc4xx/
 obj-$(CONFIG_IMX_SDMA) += imx-sdma.o
 obj-$(CONFIG_IMX_DMA) += imx-dma.o
 obj-$(CONFIG_MXS_DMA) += mxs-dma.o
+obj-$(CONFIG_PXA_DMA) += pxa_dma.o
 obj-$(CONFIG_TIMB_DMA) += timb_dma.o
 obj-$(CONFIG_SIRF_DMA) += sirf-dma.o
 obj-$(CONFIG_TI_EDMA) += edma.o
@@ -38,6 +39,7 @@ obj-$(CONFIG_EP93XX_DMA) += ep93xx_dma.o
 obj-$(CONFIG_DMA_SA11X0) += sa11x0-dma.o
 obj-$(CONFIG_MMP_TDMA) += mmp_tdma.o
 obj-$(CONFIG_DMA_OMAP) += omap-dma.o
+obj-$(CONFIG_TI_DMA_CROSSBAR) += ti-dma-crossbar.o
 obj-$(CONFIG_DMA_BCM2835) += bcm2835-dma.o
 obj-$(CONFIG_MMP_PDMA) += mmp_pdma.o
 obj-$(CONFIG_DMA_JZ4740) += dma-jz4740.o
diff --git a/drivers/dma/amba-pl08x.c b/drivers/dma/amba-pl08x.c
index 49d396ec06e5..5de3cf453f35 100644
--- a/drivers/dma/amba-pl08x.c
+++ b/drivers/dma/amba-pl08x.c
@@ -474,7 +474,7 @@ static void pl08x_terminate_phy_chan(struct pl08x_driver_data *pl08x,
 	u32 val = readl(ch->reg_config);
 
 	val &= ~(PL080_CONFIG_ENABLE | PL080_CONFIG_ERR_IRQ_MASK |
-	         PL080_CONFIG_TC_IRQ_MASK);
+		 PL080_CONFIG_TC_IRQ_MASK);
 
 	writel(val, ch->reg_config);
 
diff --git a/drivers/dma/at_hdmac.c b/drivers/dma/at_hdmac.c
index 57b2141ddddc..59892126d175 100644
--- a/drivers/dma/at_hdmac.c
+++ b/drivers/dma/at_hdmac.c
@@ -247,6 +247,10 @@ static void atc_dostart(struct at_dma_chan *atchan, struct at_desc *first)
 	channel_writel(atchan, CTRLA, 0);
 	channel_writel(atchan, CTRLB, 0);
 	channel_writel(atchan, DSCR, first->txd.phys);
+	channel_writel(atchan, SPIP, ATC_SPIP_HOLE(first->src_hole) |
+		       ATC_SPIP_BOUNDARY(first->boundary));
+	channel_writel(atchan, DPIP, ATC_DPIP_HOLE(first->dst_hole) |
+		       ATC_DPIP_BOUNDARY(first->boundary));
 	dma_writel(atdma, CHER, atchan->mask);
 
 	vdbg_dump_regs(atchan);
@@ -635,6 +639,104 @@ static dma_cookie_t atc_tx_submit(struct dma_async_tx_descriptor *tx)
 }
 
 /**
+ * atc_prep_dma_interleaved - prepare memory to memory interleaved operation
+ * @chan: the channel to prepare operation on
+ * @xt: Interleaved transfer template
+ * @flags: tx descriptor status flags
+ */
+static struct dma_async_tx_descriptor *
+atc_prep_dma_interleaved(struct dma_chan *chan,
+			 struct dma_interleaved_template *xt,
+			 unsigned long flags)
+{
+	struct at_dma_chan	*atchan = to_at_dma_chan(chan);
+	struct data_chunk	*first = xt->sgl;
+	struct at_desc		*desc = NULL;
+	size_t			xfer_count;
+	unsigned int		dwidth;
+	u32			ctrla;
+	u32			ctrlb;
+	size_t			len = 0;
+	int			i;
+
+	dev_info(chan2dev(chan),
+		 "%s: src=0x%08x, dest=0x%08x, numf=%d, frame_size=%d, flags=0x%lx\n",
+		__func__, xt->src_start, xt->dst_start, xt->numf,
+		xt->frame_size, flags);
+
+	if (unlikely(!xt || xt->numf != 1 || !xt->frame_size))
+		return NULL;
+
+	/*
+	 * The controller can only "skip" X bytes every Y bytes, so we
+	 * need to make sure we are given a template that fit that
+	 * description, ie a template with chunks that always have the
+	 * same size, with the same ICGs.
+	 */
+	for (i = 0; i < xt->frame_size; i++) {
+		struct data_chunk *chunk = xt->sgl + i;
+
+		if ((chunk->size != xt->sgl->size) ||
+		    (dmaengine_get_dst_icg(xt, chunk) != dmaengine_get_dst_icg(xt, first)) ||
+		    (dmaengine_get_src_icg(xt, chunk) != dmaengine_get_src_icg(xt, first))) {
+			dev_err(chan2dev(chan),
+				"%s: the controller can transfer only identical chunks\n",
+				__func__);
+			return NULL;
+		}
+
+		len += chunk->size;
+	}
+
+	dwidth = atc_get_xfer_width(xt->src_start,
+				    xt->dst_start, len);
+
+	xfer_count = len >> dwidth;
+	if (xfer_count > ATC_BTSIZE_MAX) {
+		dev_err(chan2dev(chan), "%s: buffer is too big\n", __func__);
+		return NULL;
+	}
+
+	ctrla = ATC_SRC_WIDTH(dwidth) |
+		ATC_DST_WIDTH(dwidth);
+
+	ctrlb =   ATC_DEFAULT_CTRLB | ATC_IEN
+		| ATC_SRC_ADDR_MODE_INCR
+		| ATC_DST_ADDR_MODE_INCR
+		| ATC_SRC_PIP
+		| ATC_DST_PIP
+		| ATC_FC_MEM2MEM;
+
+	/* create the transfer */
+	desc = atc_desc_get(atchan);
+	if (!desc) {
+		dev_err(chan2dev(chan),
+			"%s: couldn't allocate our descriptor\n", __func__);
+		return NULL;
+	}
+
+	desc->lli.saddr = xt->src_start;
+	desc->lli.daddr = xt->dst_start;
+	desc->lli.ctrla = ctrla | xfer_count;
+	desc->lli.ctrlb = ctrlb;
+
+	desc->boundary = first->size >> dwidth;
+	desc->dst_hole = (dmaengine_get_dst_icg(xt, first) >> dwidth) + 1;
+	desc->src_hole = (dmaengine_get_src_icg(xt, first) >> dwidth) + 1;
+
+	desc->txd.cookie = -EBUSY;
+	desc->total_len = desc->len = len;
+	desc->tx_width = dwidth;
+
+	/* set end-of-link to the last link descriptor of list*/
+	set_desc_eol(desc);
+
+	desc->txd.flags = flags; /* client is in control of this ack */
+
+	return &desc->txd;
+}
+
+/**
  * atc_prep_dma_memcpy - prepare a memcpy operation
  * @chan: the channel to prepare operation on
  * @dest: operation virtual destination address
@@ -1609,6 +1711,7 @@ static int __init at_dma_probe(struct platform_device *pdev)
 	/* setup platform data for each SoC */
 	dma_cap_set(DMA_MEMCPY, at91sam9rl_config.cap_mask);
 	dma_cap_set(DMA_SG, at91sam9rl_config.cap_mask);
+	dma_cap_set(DMA_INTERLEAVE, at91sam9g45_config.cap_mask);
 	dma_cap_set(DMA_MEMCPY, at91sam9g45_config.cap_mask);
 	dma_cap_set(DMA_SLAVE, at91sam9g45_config.cap_mask);
 	dma_cap_set(DMA_SG, at91sam9g45_config.cap_mask);
@@ -1713,6 +1816,9 @@ static int __init at_dma_probe(struct platform_device *pdev)
 	atdma->dma_common.dev = &pdev->dev;
 
 	/* set prep routines based on capability */
+	if (dma_has_cap(DMA_INTERLEAVE, atdma->dma_common.cap_mask))
+		atdma->dma_common.device_prep_interleaved_dma = atc_prep_dma_interleaved;
+
 	if (dma_has_cap(DMA_MEMCPY, atdma->dma_common.cap_mask))
 		atdma->dma_common.device_prep_dma_memcpy = atc_prep_dma_memcpy;
 
diff --git a/drivers/dma/at_hdmac_regs.h b/drivers/dma/at_hdmac_regs.h
index 2727ca560572..bc8d5ebedd19 100644
--- a/drivers/dma/at_hdmac_regs.h
+++ b/drivers/dma/at_hdmac_regs.h
@@ -196,6 +196,11 @@ struct at_desc {
 	size_t				len;
 	u32				tx_width;
 	size_t				total_len;
+
+	/* Interleaved data */
+	size_t				boundary;
+	size_t				dst_hole;
+	size_t				src_hole;
 };
 
 static inline struct at_desc *
diff --git a/drivers/dma/at_xdmac.c b/drivers/dma/at_xdmac.c
index 7992164ea9ec..cf1213de7865 100644
--- a/drivers/dma/at_xdmac.c
+++ b/drivers/dma/at_xdmac.c
@@ -235,6 +235,10 @@ struct at_xdmac_lld {
 	dma_addr_t	mbr_sa;		/* Source Address Member */
 	dma_addr_t	mbr_da;		/* Destination Address Member */
 	u32		mbr_cfg;	/* Configuration Register */
+	u32		mbr_bc;		/* Block Control Register */
+	u32		mbr_ds;		/* Data Stride Register */
+	u32		mbr_sus;	/* Source Microblock Stride Register */
+	u32		mbr_dus;	/* Destination Microblock Stride Register */
 };
 
 
@@ -358,6 +362,8 @@ static void at_xdmac_start_xfer(struct at_xdmac_chan *atchan,
 	if (at_xdmac_chan_is_cyclic(atchan)) {
 		reg = AT_XDMAC_CNDC_NDVIEW_NDV1;
 		at_xdmac_chan_write(atchan, AT_XDMAC_CC, first->lld.mbr_cfg);
+	} else if (first->lld.mbr_ubc & AT_XDMAC_MBR_UBC_NDV3) {
+		reg = AT_XDMAC_CNDC_NDVIEW_NDV3;
 	} else {
 		/*
 		 * No need to write AT_XDMAC_CC reg, it will be done when the
@@ -465,6 +471,33 @@ static struct at_xdmac_desc *at_xdmac_get_desc(struct at_xdmac_chan *atchan)
 	return desc;
 }
 
+static void at_xdmac_queue_desc(struct dma_chan *chan,
+				struct at_xdmac_desc *prev,
+				struct at_xdmac_desc *desc)
+{
+	if (!prev || !desc)
+		return;
+
+	prev->lld.mbr_nda = desc->tx_dma_desc.phys;
+	prev->lld.mbr_ubc |= AT_XDMAC_MBR_UBC_NDE;
+
+	dev_dbg(chan2dev(chan),	"%s: chain lld: prev=0x%p, mbr_nda=%pad\n",
+		__func__, prev, &prev->lld.mbr_nda);
+}
+
+static inline void at_xdmac_increment_block_count(struct dma_chan *chan,
+						  struct at_xdmac_desc *desc)
+{
+	if (!desc)
+		return;
+
+	desc->lld.mbr_bc++;
+
+	dev_dbg(chan2dev(chan),
+		"%s: incrementing the block count of the desc 0x%p\n",
+		__func__, desc);
+}
+
 static struct dma_chan *at_xdmac_xlate(struct of_phandle_args *dma_spec,
 				       struct of_dma *of_dma)
 {
@@ -656,19 +689,14 @@ at_xdmac_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl,
 		desc->lld.mbr_ubc = AT_XDMAC_MBR_UBC_NDV2			/* next descriptor view */
 			| AT_XDMAC_MBR_UBC_NDEN					/* next descriptor dst parameter update */
 			| AT_XDMAC_MBR_UBC_NSEN					/* next descriptor src parameter update */
-			| (i == sg_len - 1 ? 0 : AT_XDMAC_MBR_UBC_NDE)		/* descriptor fetch */
 			| (len >> fixed_dwidth);				/* microblock length */
 		dev_dbg(chan2dev(chan),
 			 "%s: lld: mbr_sa=%pad, mbr_da=%pad, mbr_ubc=0x%08x\n",
 			 __func__, &desc->lld.mbr_sa, &desc->lld.mbr_da, desc->lld.mbr_ubc);
 
 		/* Chain lld. */
-		if (prev) {
-			prev->lld.mbr_nda = desc->tx_dma_desc.phys;
-			dev_dbg(chan2dev(chan),
-				 "%s: chain lld: prev=0x%p, mbr_nda=%pad\n",
-				 __func__, prev, &prev->lld.mbr_nda);
-		}
+		if (prev)
+			at_xdmac_queue_desc(chan, prev, desc);
 
 		prev = desc;
 		if (!first)
@@ -748,7 +776,6 @@ at_xdmac_prep_dma_cyclic(struct dma_chan *chan, dma_addr_t buf_addr,
 		desc->lld.mbr_ubc = AT_XDMAC_MBR_UBC_NDV1
 			| AT_XDMAC_MBR_UBC_NDEN
 			| AT_XDMAC_MBR_UBC_NSEN
-			| AT_XDMAC_MBR_UBC_NDE
 			| period_len >> at_xdmac_get_dwidth(desc->lld.mbr_cfg);
 
 		dev_dbg(chan2dev(chan),
@@ -756,12 +783,8 @@ at_xdmac_prep_dma_cyclic(struct dma_chan *chan, dma_addr_t buf_addr,
 			 __func__, &desc->lld.mbr_sa, &desc->lld.mbr_da, desc->lld.mbr_ubc);
 
 		/* Chain lld. */
-		if (prev) {
-			prev->lld.mbr_nda = desc->tx_dma_desc.phys;
-			dev_dbg(chan2dev(chan),
-				 "%s: chain lld: prev=0x%p, mbr_nda=%pad\n",
-				 __func__, prev, &prev->lld.mbr_nda);
-		}
+		if (prev)
+			at_xdmac_queue_desc(chan, prev, desc);
 
 		prev = desc;
 		if (!first)
@@ -783,6 +806,215 @@ at_xdmac_prep_dma_cyclic(struct dma_chan *chan, dma_addr_t buf_addr,
 	return &first->tx_dma_desc;
 }
 
+static inline u32 at_xdmac_align_width(struct dma_chan *chan, dma_addr_t addr)
+{
+	u32 width;
+
+	/*
+	 * Check address alignment to select the greater data width we
+	 * can use.
+	 *
+	 * Some XDMAC implementations don't provide dword transfer, in
+	 * this case selecting dword has the same behavior as
+	 * selecting word transfers.
+	 */
+	if (!(addr & 7)) {
+		width = AT_XDMAC_CC_DWIDTH_DWORD;
+		dev_dbg(chan2dev(chan), "%s: dwidth: double word\n", __func__);
+	} else if (!(addr & 3)) {
+		width = AT_XDMAC_CC_DWIDTH_WORD;
+		dev_dbg(chan2dev(chan), "%s: dwidth: word\n", __func__);
+	} else if (!(addr & 1)) {
+		width = AT_XDMAC_CC_DWIDTH_HALFWORD;
+		dev_dbg(chan2dev(chan), "%s: dwidth: half word\n", __func__);
+	} else {
+		width = AT_XDMAC_CC_DWIDTH_BYTE;
+		dev_dbg(chan2dev(chan), "%s: dwidth: byte\n", __func__);
+	}
+
+	return width;
+}
+
+static struct at_xdmac_desc *
+at_xdmac_interleaved_queue_desc(struct dma_chan *chan,
+				struct at_xdmac_chan *atchan,
+				struct at_xdmac_desc *prev,
+				dma_addr_t src, dma_addr_t dst,
+				struct dma_interleaved_template *xt,
+				struct data_chunk *chunk)
+{
+	struct at_xdmac_desc	*desc;
+	u32			dwidth;
+	unsigned long		flags;
+	size_t			ublen;
+	/*
+	 * WARNING: The channel configuration is set here since there is no
+	 * dmaengine_slave_config call in this case. Moreover we don't know the
+	 * direction, it involves we can't dynamically set the source and dest
+	 * interface so we have to use the same one. Only interface 0 allows EBI
+	 * access. Hopefully we can access DDR through both ports (at least on
+	 * SAMA5D4x), so we can use the same interface for source and dest,
+	 * that solves the fact we don't know the direction.
+	 */
+	u32			chan_cc = AT_XDMAC_CC_DIF(0)
+					| AT_XDMAC_CC_SIF(0)
+					| AT_XDMAC_CC_MBSIZE_SIXTEEN
+					| AT_XDMAC_CC_TYPE_MEM_TRAN;
+
+	dwidth = at_xdmac_align_width(chan, src | dst | chunk->size);
+	if (chunk->size >= (AT_XDMAC_MBR_UBC_UBLEN_MAX << dwidth)) {
+		dev_dbg(chan2dev(chan),
+			"%s: chunk too big (%d, max size %lu)...\n",
+			__func__, chunk->size,
+			AT_XDMAC_MBR_UBC_UBLEN_MAX << dwidth);
+		return NULL;
+	}
+
+	if (prev)
+		dev_dbg(chan2dev(chan),
+			"Adding items at the end of desc 0x%p\n", prev);
+
+	if (xt->src_inc) {
+		if (xt->src_sgl)
+			chan_cc |=  AT_XDMAC_CC_SAM_UBS_DS_AM;
+		else
+			chan_cc |=  AT_XDMAC_CC_SAM_INCREMENTED_AM;
+	}
+
+	if (xt->dst_inc) {
+		if (xt->dst_sgl)
+			chan_cc |=  AT_XDMAC_CC_DAM_UBS_DS_AM;
+		else
+			chan_cc |=  AT_XDMAC_CC_DAM_INCREMENTED_AM;
+	}
+
+	spin_lock_irqsave(&atchan->lock, flags);
+	desc = at_xdmac_get_desc(atchan);
+	spin_unlock_irqrestore(&atchan->lock, flags);
+	if (!desc) {
+		dev_err(chan2dev(chan), "can't get descriptor\n");
+		return NULL;
+	}
+
+	chan_cc |= AT_XDMAC_CC_DWIDTH(dwidth);
+
+	ublen = chunk->size >> dwidth;
+
+	desc->lld.mbr_sa = src;
+	desc->lld.mbr_da = dst;
+	desc->lld.mbr_sus = dmaengine_get_src_icg(xt, chunk);
+	desc->lld.mbr_dus = dmaengine_get_dst_icg(xt, chunk);
+
+	desc->lld.mbr_ubc = AT_XDMAC_MBR_UBC_NDV3
+		| AT_XDMAC_MBR_UBC_NDEN
+		| AT_XDMAC_MBR_UBC_NSEN
+		| ublen;
+	desc->lld.mbr_cfg = chan_cc;
+
+	dev_dbg(chan2dev(chan),
+		"%s: lld: mbr_sa=0x%08x, mbr_da=0x%08x, mbr_ubc=0x%08x, mbr_cfg=0x%08x\n",
+		__func__, desc->lld.mbr_sa, desc->lld.mbr_da,
+		desc->lld.mbr_ubc, desc->lld.mbr_cfg);
+
+	/* Chain lld. */
+	if (prev)
+		at_xdmac_queue_desc(chan, prev, desc);
+
+	return desc;
+}
+
+static struct dma_async_tx_descriptor *
+at_xdmac_prep_interleaved(struct dma_chan *chan,
+			  struct dma_interleaved_template *xt,
+			  unsigned long flags)
+{
+	struct at_xdmac_chan	*atchan = to_at_xdmac_chan(chan);
+	struct at_xdmac_desc	*prev = NULL, *first = NULL;
+	struct data_chunk	*chunk, *prev_chunk = NULL;
+	dma_addr_t		dst_addr, src_addr;
+	size_t			dst_skip, src_skip, len = 0;
+	size_t			prev_dst_icg = 0, prev_src_icg = 0;
+	int			i;
+
+	if (!xt || (xt->numf != 1) || (xt->dir != DMA_MEM_TO_MEM))
+		return NULL;
+
+	dev_dbg(chan2dev(chan), "%s: src=0x%08x, dest=0x%08x, numf=%d, frame_size=%d, flags=0x%lx\n",
+		__func__, xt->src_start, xt->dst_start,	xt->numf,
+		xt->frame_size, flags);
+
+	src_addr = xt->src_start;
+	dst_addr = xt->dst_start;
+
+	for (i = 0; i < xt->frame_size; i++) {
+		struct at_xdmac_desc *desc;
+		size_t src_icg, dst_icg;
+
+		chunk = xt->sgl + i;
+
+		dst_icg = dmaengine_get_dst_icg(xt, chunk);
+		src_icg = dmaengine_get_src_icg(xt, chunk);
+
+		src_skip = chunk->size + src_icg;
+		dst_skip = chunk->size + dst_icg;
+
+		dev_dbg(chan2dev(chan),
+			"%s: chunk size=%d, src icg=%d, dst icg=%d\n",
+			__func__, chunk->size, src_icg, dst_icg);
+
+		/*
+		 * Handle the case where we just have the same
+		 * transfer to setup, we can just increase the
+		 * block number and reuse the same descriptor.
+		 */
+		if (prev_chunk && prev &&
+		    (prev_chunk->size == chunk->size) &&
+		    (prev_src_icg == src_icg) &&
+		    (prev_dst_icg == dst_icg)) {
+			dev_dbg(chan2dev(chan),
+				"%s: same configuration that the previous chunk, merging the descriptors...\n",
+				__func__);
+			at_xdmac_increment_block_count(chan, prev);
+			continue;
+		}
+
+		desc = at_xdmac_interleaved_queue_desc(chan, atchan,
+						       prev,
+						       src_addr, dst_addr,
+						       xt, chunk);
+		if (!desc) {
+			list_splice_init(&first->descs_list,
+					 &atchan->free_descs_list);
+			return NULL;
+		}
+
+		if (!first)
+			first = desc;
+
+		dev_dbg(chan2dev(chan), "%s: add desc 0x%p to descs_list 0x%p\n",
+			__func__, desc, first);
+		list_add_tail(&desc->desc_node, &first->descs_list);
+
+		if (xt->src_sgl)
+			src_addr += src_skip;
+
+		if (xt->dst_sgl)
+			dst_addr += dst_skip;
+
+		len += chunk->size;
+		prev_chunk = chunk;
+		prev_dst_icg = dst_icg;
+		prev_src_icg = src_icg;
+		prev = desc;
+	}
+
+	first->tx_dma_desc.cookie = -EBUSY;
+	first->tx_dma_desc.flags = flags;
+	first->xfer_size = len;
+
+	return &first->tx_dma_desc;
+}
+
 static struct dma_async_tx_descriptor *
 at_xdmac_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 			 size_t len, unsigned long flags)
@@ -814,24 +1046,7 @@ at_xdmac_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 	if (unlikely(!len))
 		return NULL;
 
-	/*
-	 * Check address alignment to select the greater data width we can use.
-	 * Some XDMAC implementations don't provide dword transfer, in this
-	 * case selecting dword has the same behavior as selecting word transfers.
-	 */
-	if (!((src_addr | dst_addr) & 7)) {
-		dwidth = AT_XDMAC_CC_DWIDTH_DWORD;
-		dev_dbg(chan2dev(chan), "%s: dwidth: double word\n", __func__);
-	} else if (!((src_addr | dst_addr)  & 3)) {
-		dwidth = AT_XDMAC_CC_DWIDTH_WORD;
-		dev_dbg(chan2dev(chan), "%s: dwidth: word\n", __func__);
-	} else if (!((src_addr | dst_addr) & 1)) {
-		dwidth = AT_XDMAC_CC_DWIDTH_HALFWORD;
-		dev_dbg(chan2dev(chan), "%s: dwidth: half word\n", __func__);
-	} else {
-		dwidth = AT_XDMAC_CC_DWIDTH_BYTE;
-		dev_dbg(chan2dev(chan), "%s: dwidth: byte\n", __func__);
-	}
+	dwidth = at_xdmac_align_width(chan, src_addr | dst_addr);
 
 	/* Prepare descriptors. */
 	while (remaining_size) {
@@ -861,19 +1076,8 @@ at_xdmac_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 		dev_dbg(chan2dev(chan), "%s: xfer_size=%zu\n", __func__, xfer_size);
 
 		/* Check remaining length and change data width if needed. */
-		if (!((src_addr | dst_addr | xfer_size) & 7)) {
-			dwidth = AT_XDMAC_CC_DWIDTH_DWORD;
-			dev_dbg(chan2dev(chan), "%s: dwidth: double word\n", __func__);
-		} else if (!((src_addr | dst_addr | xfer_size)  & 3)) {
-			dwidth = AT_XDMAC_CC_DWIDTH_WORD;
-			dev_dbg(chan2dev(chan), "%s: dwidth: word\n", __func__);
-		} else if (!((src_addr | dst_addr | xfer_size) & 1)) {
-			dwidth = AT_XDMAC_CC_DWIDTH_HALFWORD;
-			dev_dbg(chan2dev(chan), "%s: dwidth: half word\n", __func__);
-		} else if ((src_addr | dst_addr | xfer_size) & 1) {
-			dwidth = AT_XDMAC_CC_DWIDTH_BYTE;
-			dev_dbg(chan2dev(chan), "%s: dwidth: byte\n", __func__);
-		}
+		dwidth = at_xdmac_align_width(chan,
+					      src_addr | dst_addr | xfer_size);
 		chan_cc |= AT_XDMAC_CC_DWIDTH(dwidth);
 
 		ublen = xfer_size >> dwidth;
@@ -884,7 +1088,6 @@ at_xdmac_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 		desc->lld.mbr_ubc = AT_XDMAC_MBR_UBC_NDV2
 			| AT_XDMAC_MBR_UBC_NDEN
 			| AT_XDMAC_MBR_UBC_NSEN
-			| (remaining_size ? AT_XDMAC_MBR_UBC_NDE : 0)
 			| ublen;
 		desc->lld.mbr_cfg = chan_cc;
 
@@ -893,12 +1096,8 @@ at_xdmac_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 			 __func__, &desc->lld.mbr_sa, &desc->lld.mbr_da, desc->lld.mbr_ubc, desc->lld.mbr_cfg);
 
 		/* Chain lld. */
-		if (prev) {
-			prev->lld.mbr_nda = desc->tx_dma_desc.phys;
-			dev_dbg(chan2dev(chan),
-				 "%s: chain lld: prev=0x%p, mbr_nda=0x%08x\n",
-				 __func__, prev, prev->lld.mbr_nda);
-		}
+		if (prev)
+			at_xdmac_queue_desc(chan, prev, desc);
 
 		prev = desc;
 		if (!first)
@@ -915,6 +1114,93 @@ at_xdmac_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 	return &first->tx_dma_desc;
 }
 
+static struct at_xdmac_desc *at_xdmac_memset_create_desc(struct dma_chan *chan,
+							 struct at_xdmac_chan *atchan,
+							 dma_addr_t dst_addr,
+							 size_t len,
+							 int value)
+{
+	struct at_xdmac_desc	*desc;
+	unsigned long		flags;
+	size_t			ublen;
+	u32			dwidth;
+	/*
+	 * WARNING: The channel configuration is set here since there is no
+	 * dmaengine_slave_config call in this case. Moreover we don't know the
+	 * direction, it involves we can't dynamically set the source and dest
+	 * interface so we have to use the same one. Only interface 0 allows EBI
+	 * access. Hopefully we can access DDR through both ports (at least on
+	 * SAMA5D4x), so we can use the same interface for source and dest,
+	 * that solves the fact we don't know the direction.
+	 */
+	u32			chan_cc = AT_XDMAC_CC_DAM_INCREMENTED_AM
+					| AT_XDMAC_CC_SAM_INCREMENTED_AM
+					| AT_XDMAC_CC_DIF(0)
+					| AT_XDMAC_CC_SIF(0)
+					| AT_XDMAC_CC_MBSIZE_SIXTEEN
+					| AT_XDMAC_CC_MEMSET_HW_MODE
+					| AT_XDMAC_CC_TYPE_MEM_TRAN;
+
+	dwidth = at_xdmac_align_width(chan, dst_addr);
+
+	if (len >= (AT_XDMAC_MBR_UBC_UBLEN_MAX << dwidth)) {
+		dev_err(chan2dev(chan),
+			"%s: Transfer too large, aborting...\n",
+			__func__);
+		return NULL;
+	}
+
+	spin_lock_irqsave(&atchan->lock, flags);
+	desc = at_xdmac_get_desc(atchan);
+	spin_unlock_irqrestore(&atchan->lock, flags);
+	if (!desc) {
+		dev_err(chan2dev(chan), "can't get descriptor\n");
+		return NULL;
+	}
+
+	chan_cc |= AT_XDMAC_CC_DWIDTH(dwidth);
+
+	ublen = len >> dwidth;
+
+	desc->lld.mbr_da = dst_addr;
+	desc->lld.mbr_ds = value;
+	desc->lld.mbr_ubc = AT_XDMAC_MBR_UBC_NDV3
+		| AT_XDMAC_MBR_UBC_NDEN
+		| AT_XDMAC_MBR_UBC_NSEN
+		| ublen;
+	desc->lld.mbr_cfg = chan_cc;
+
+	dev_dbg(chan2dev(chan),
+		"%s: lld: mbr_da=0x%08x, mbr_ds=0x%08x, mbr_ubc=0x%08x, mbr_cfg=0x%08x\n",
+		__func__, desc->lld.mbr_da, desc->lld.mbr_ds, desc->lld.mbr_ubc,
+		desc->lld.mbr_cfg);
+
+	return desc;
+}
+
+struct dma_async_tx_descriptor *
+at_xdmac_prep_dma_memset(struct dma_chan *chan, dma_addr_t dest, int value,
+			 size_t len, unsigned long flags)
+{
+	struct at_xdmac_chan	*atchan = to_at_xdmac_chan(chan);
+	struct at_xdmac_desc	*desc;
+
+	dev_dbg(chan2dev(chan), "%s: dest=0x%08x, len=%d, pattern=0x%x, flags=0x%lx\n",
+		__func__, dest, len, value, flags);
+
+	if (unlikely(!len))
+		return NULL;
+
+	desc = at_xdmac_memset_create_desc(chan, atchan, dest, len, value);
+	list_add_tail(&desc->desc_node, &desc->descs_list);
+
+	desc->tx_dma_desc.cookie = -EBUSY;
+	desc->tx_dma_desc.flags = flags;
+	desc->xfer_size = len;
+
+	return &desc->tx_dma_desc;
+}
+
 static enum dma_status
 at_xdmac_tx_status(struct dma_chan *chan, dma_cookie_t cookie,
 		struct dma_tx_state *txstate)
@@ -1445,7 +1731,9 @@ static int at_xdmac_probe(struct platform_device *pdev)
 	}
 
 	dma_cap_set(DMA_CYCLIC, atxdmac->dma.cap_mask);
+	dma_cap_set(DMA_INTERLEAVE, atxdmac->dma.cap_mask);
 	dma_cap_set(DMA_MEMCPY, atxdmac->dma.cap_mask);
+	dma_cap_set(DMA_MEMSET, atxdmac->dma.cap_mask);
 	dma_cap_set(DMA_SLAVE, atxdmac->dma.cap_mask);
 	/*
 	 * Without DMA_PRIVATE the driver is not able to allocate more than
@@ -1458,7 +1746,9 @@ static int at_xdmac_probe(struct platform_device *pdev)
 	atxdmac->dma.device_tx_status			= at_xdmac_tx_status;
 	atxdmac->dma.device_issue_pending		= at_xdmac_issue_pending;
 	atxdmac->dma.device_prep_dma_cyclic		= at_xdmac_prep_dma_cyclic;
+	atxdmac->dma.device_prep_interleaved_dma	= at_xdmac_prep_interleaved;
 	atxdmac->dma.device_prep_dma_memcpy		= at_xdmac_prep_dma_memcpy;
+	atxdmac->dma.device_prep_dma_memset		= at_xdmac_prep_dma_memset;
 	atxdmac->dma.device_prep_slave_sg		= at_xdmac_prep_slave_sg;
 	atxdmac->dma.device_config			= at_xdmac_device_config;
 	atxdmac->dma.device_pause			= at_xdmac_device_pause;
diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c
index 3ddfd1f6c23c..4a4cce15f25d 100644
--- a/drivers/dma/dmaengine.c
+++ b/drivers/dma/dmaengine.c
@@ -267,6 +267,13 @@ static void dma_chan_put(struct dma_chan *chan)
 	/* This channel is not in use anymore, free it */
 	if (!chan->client_count && chan->device->device_free_chan_resources)
 		chan->device->device_free_chan_resources(chan);
+
+	/* If the channel is used via a DMA request router, free the mapping */
+	if (chan->router && chan->router->route_free) {
+		chan->router->route_free(chan->router->dev, chan->route_data);
+		chan->router = NULL;
+		chan->route_data = NULL;
+	}
 }
 
 enum dma_status dma_sync_wait(struct dma_chan *chan, dma_cookie_t cookie)
@@ -536,7 +543,7 @@ static struct dma_chan *private_candidate(const dma_cap_mask_t *mask,
 }
 
 /**
- * dma_request_slave_channel - try to get specific channel exclusively
+ * dma_get_slave_channel - try to get specific channel exclusively
  * @chan: target channel
  */
 struct dma_chan *dma_get_slave_channel(struct dma_chan *chan)
@@ -648,7 +655,7 @@ struct dma_chan *__dma_request_channel(const dma_cap_mask_t *mask,
 EXPORT_SYMBOL_GPL(__dma_request_channel);
 
 /**
- * dma_request_slave_channel - try to allocate an exclusive slave channel
+ * dma_request_slave_channel_reason - try to allocate an exclusive slave channel
  * @dev:	pointer to client device structure
  * @name:	slave channel name
  *
@@ -836,6 +843,8 @@ int dma_async_device_register(struct dma_device *device)
 		!device->device_prep_dma_pq);
 	BUG_ON(dma_has_cap(DMA_PQ_VAL, device->cap_mask) &&
 		!device->device_prep_dma_pq_val);
+	BUG_ON(dma_has_cap(DMA_MEMSET, device->cap_mask) &&
+		!device->device_prep_dma_memset);
 	BUG_ON(dma_has_cap(DMA_INTERRUPT, device->cap_mask) &&
 		!device->device_prep_dma_interrupt);
 	BUG_ON(dma_has_cap(DMA_SG, device->cap_mask) &&
diff --git a/drivers/dma/ep93xx_dma.c b/drivers/dma/ep93xx_dma.c
index 24e5290faa32..57ff46284f15 100644
--- a/drivers/dma/ep93xx_dma.c
+++ b/drivers/dma/ep93xx_dma.c
@@ -1364,7 +1364,7 @@ static int __init ep93xx_dma_probe(struct platform_device *pdev)
 	return ret;
 }
 
-static struct platform_device_id ep93xx_dma_driver_ids[] = {
+static const struct platform_device_id ep93xx_dma_driver_ids[] = {
 	{ "ep93xx-dma-m2p", 0 },
 	{ "ep93xx-dma-m2m", 1 },
 	{ },
diff --git a/drivers/dma/fsl-edma.c b/drivers/dma/fsl-edma.c
index 09e2842d15ec..915eec3cc279 100644
--- a/drivers/dma/fsl-edma.c
+++ b/drivers/dma/fsl-edma.c