Merge tag 'dmaengine-4.17-rc1' of git://git.infradead.org/users/vkoul/slave-dma

Pull dmaengine updates from Vinod Koul:
 "This time we have couple of new drivers along with updates to drivers:

   - new drivers for the DesignWare AXI DMAC and MediaTek High-Speed DMA
     controllers

   - stm32 dma and qcom bam dma driver updates

   - norandom test option for dmatest"

* tag 'dmaengine-4.17-rc1' of git://git.infradead.org/users/vkoul/slave-dma: (30 commits)
  dmaengine: stm32-dma: properly mask irq bits
  dmaengine: stm32-dma: fix max items per transfer
  dmaengine: stm32-dma: fix DMA IRQ status handling
  dmaengine: stm32-dma: Improve memory burst management
  dmaengine: stm32-dma: fix typo and reported checkpatch warnings
  dmaengine: stm32-dma: fix incomplete configuration in cyclic mode
  dmaengine: stm32-dma: threshold manages with bitfield feature
  dt-bindings: stm32-dma: introduce DMA features bitfield
  dt-bindings: rcar-dmac: Document r8a77470 support
  dmaengine: rcar-dmac: Fix too early/late system suspend/resume callbacks
  dmaengine: dw-axi-dmac: fix spelling mistake: "catched" -> "caught"
  dmaengine: edma: Check the memory allocation for the memcpy dma device
  dmaengine: at_xdmac: fix rare residue corruption
  dmaengine: mediatek: update MAINTAINERS entry with MediaTek DMA driver
  dmaengine: mediatek: Add MediaTek High-Speed DMA controller for MT7622 and MT7623 SoC
  dt-bindings: dmaengine: Add MediaTek High-Speed DMA controller bindings
  dt-bindings: Document the Synopsys DW AXI DMA bindings
  dmaengine: Introduce DW AXI DMAC driver
  dmaengine: pl330: fix a race condition in case of threaded irqs
  dmaengine: imx-sdma: fix pagefault when channel is disabled during interrupt
  ...

16 files changed, 2768 insertions, 71 deletions

diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig
index 27df3e2837fd..6d61cd023633 100644
--- a/drivers/dma/Kconfig
+++ b/drivers/dma/Kconfig
@@ -187,6 +187,16 @@ config DMA_SUN6I
 	help
 	  Support for the DMA engine first found in Allwinner A31 SoCs.
 
+config DW_AXI_DMAC
+	tristate "Synopsys DesignWare AXI DMA support"
+	depends on OF || COMPILE_TEST
+	select DMA_ENGINE
+	select DMA_VIRTUAL_CHANNELS
+	help
+	  Enable support for Synopsys DesignWare AXI DMA controller.
+	  NOTE: This driver wasn't tested on 64 bit platform because
+	  of lack 64 bit platform with Synopsys DW AXI DMAC.
+
 config EP93XX_DMA
 	bool "Cirrus Logic EP93xx DMA support"
 	depends on ARCH_EP93XX || COMPILE_TEST
@@ -633,6 +643,8 @@ config ZX_DMA
 # driver files
 source "drivers/dma/bestcomm/Kconfig"
 
+source "drivers/dma/mediatek/Kconfig"
+
 source "drivers/dma/qcom/Kconfig"
 
 source "drivers/dma/dw/Kconfig"
diff --git a/drivers/dma/Makefile b/drivers/dma/Makefile
index b9dca8a0e142..0f62a4d49aab 100644
--- a/drivers/dma/Makefile
+++ b/drivers/dma/Makefile
@@ -28,6 +28,7 @@ obj-$(CONFIG_DMA_OMAP) += omap-dma.o
 obj-$(CONFIG_DMA_SA11X0) += sa11x0-dma.o
 obj-$(CONFIG_DMA_SUN4I) += sun4i-dma.o
 obj-$(CONFIG_DMA_SUN6I) += sun6i-dma.o
+obj-$(CONFIG_DW_AXI_DMAC) += dw-axi-dmac/
 obj-$(CONFIG_DW_DMAC_CORE) += dw/
 obj-$(CONFIG_EP93XX_DMA) += ep93xx_dma.o
 obj-$(CONFIG_FSL_DMA) += fsldma.o
@@ -75,5 +76,6 @@ obj-$(CONFIG_XGENE_DMA) += xgene-dma.o
 obj-$(CONFIG_ZX_DMA) += zx_dma.o
 obj-$(CONFIG_ST_FDMA) += st_fdma.o
 
+obj-y += mediatek/
 obj-y += qcom/
 obj-y += xilinx/
diff --git a/drivers/dma/at_xdmac.c b/drivers/dma/at_xdmac.c
index c00e3923d7d8..94236ec9d410 100644
--- a/drivers/dma/at_xdmac.c
+++ b/drivers/dma/at_xdmac.c
@@ -1471,10 +1471,10 @@ at_xdmac_tx_status(struct dma_chan *chan, dma_cookie_t cookie,
 	for (retry = 0; retry < AT_XDMAC_RESIDUE_MAX_RETRIES; retry++) {
 		check_nda = at_xdmac_chan_read(atchan, AT_XDMAC_CNDA) & 0xfffffffc;
 		rmb();
-		initd = !!(at_xdmac_chan_read(atchan, AT_XDMAC_CC) & AT_XDMAC_CC_INITD);
-		rmb();
 		cur_ubc = at_xdmac_chan_read(atchan, AT_XDMAC_CUBC);
 		rmb();
+		initd = !!(at_xdmac_chan_read(atchan, AT_XDMAC_CC) & AT_XDMAC_CC_INITD);
+		rmb();
 		cur_nda = at_xdmac_chan_read(atchan, AT_XDMAC_CNDA) & 0xfffffffc;
 		rmb();
 
diff --git a/drivers/dma/dmatest.c b/drivers/dma/dmatest.c
index 80cc2be6483c..b9339524d5bd 100644
--- a/drivers/dma/dmatest.c
+++ b/drivers/dma/dmatest.c
@@ -74,7 +74,11 @@ MODULE_PARM_DESC(timeout, "Transfer Timeout in msec (default: 3000), "
 
 static bool noverify;
 module_param(noverify, bool, S_IRUGO | S_IWUSR);
-MODULE_PARM_DESC(noverify, "Disable random data setup and verification");
+MODULE_PARM_DESC(noverify, "Disable data verification (default: verify)");
+
+static bool norandom;
+module_param(norandom, bool, 0644);
+MODULE_PARM_DESC(norandom, "Disable random offset setup (default: random)");
 
 static bool verbose;
 module_param(verbose, bool, S_IRUGO | S_IWUSR);
@@ -103,6 +107,7 @@ struct dmatest_params {
 	unsigned int	pq_sources;
 	int		timeout;
 	bool		noverify;
+	bool		norandom;
 };
 
 /**
@@ -575,7 +580,7 @@ static int dmatest_func(void *data)
 			break;
 		}
 
-		if (params->noverify)
+		if (params->norandom)
 			len = params->buf_size;
 		else
 			len = dmatest_random() % params->buf_size + 1;
@@ -586,17 +591,19 @@ static int dmatest_func(void *data)
 
 		total_len += len;
 
-		if (params->noverify) {
+		if (params->norandom) {
 			src_off = 0;
 			dst_off = 0;
 		} else {
-			start = ktime_get();
 			src_off = dmatest_random() % (params->buf_size - len + 1);
 			dst_off = dmatest_random() % (params->buf_size - len + 1);
 
 			src_off = (src_off >> align) << align;
 			dst_off = (dst_off >> align) << align;
+		}
 
+		if (!params->noverify) {
+			start = ktime_get();
 			dmatest_init_srcs(thread->srcs, src_off, len,
 					  params->buf_size, is_memset);
 			dmatest_init_dsts(thread->dsts, dst_off, len,
@@ -975,6 +982,7 @@ static void run_threaded_test(struct dmatest_info *info)
 	params->pq_sources = pq_sources;
 	params->timeout = timeout;
 	params->noverify = noverify;
+	params->norandom = norandom;
 
 	request_channels(info, DMA_MEMCPY);
 	request_channels(info, DMA_MEMSET);
diff --git a/drivers/dma/dw-axi-dmac/Makefile b/drivers/dma/dw-axi-dmac/Makefile
new file mode 100644
index 000000000000..4bfa462005be
--- /dev/null
+++ b/drivers/dma/dw-axi-dmac/Makefile
@@ -0,0 +1 @@
+obj-$(CONFIG_DW_AXI_DMAC) += dw-axi-dmac-platform.o
diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
new file mode 100644
index 000000000000..c4eb55e3011c
--- /dev/null
+++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
@@ -0,0 +1,1008 @@
+// SPDX-License-Identifier:  GPL-2.0
+// (C) 2017-2018 Synopsys, Inc. (www.synopsys.com)
+
+/*
+ * Synopsys DesignWare AXI DMA Controller driver.
+ *
+ * Author: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>
+ */
+
+#include <linux/bitops.h>
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/dmaengine.h>
+#include <linux/dmapool.h>
+#include <linux/err.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+#include <linux/property.h>
+#include <linux/types.h>
+
+#include "dw-axi-dmac.h"
+#include "../dmaengine.h"
+#include "../virt-dma.h"
+
+/*
+ * The set of bus widths supported by the DMA controller. DW AXI DMAC supports
+ * master data bus width up to 512 bits (for both AXI master interfaces), but
+ * it depends on IP block configurarion.
+ */
+#define AXI_DMA_BUSWIDTHS		  \
+	(DMA_SLAVE_BUSWIDTH_1_BYTE	| \
+	DMA_SLAVE_BUSWIDTH_2_BYTES	| \
+	DMA_SLAVE_BUSWIDTH_4_BYTES	| \
+	DMA_SLAVE_BUSWIDTH_8_BYTES	| \
+	DMA_SLAVE_BUSWIDTH_16_BYTES	| \
+	DMA_SLAVE_BUSWIDTH_32_BYTES	| \
+	DMA_SLAVE_BUSWIDTH_64_BYTES)
+
+static inline void
+axi_dma_iowrite32(struct axi_dma_chip *chip, u32 reg, u32 val)
+{
+	iowrite32(val, chip->regs + reg);
+}
+
+static inline u32 axi_dma_ioread32(struct axi_dma_chip *chip, u32 reg)
+{
+	return ioread32(chip->regs + reg);
+}
+
+static inline void
+axi_chan_iowrite32(struct axi_dma_chan *chan, u32 reg, u32 val)
+{
+	iowrite32(val, chan->chan_regs + reg);
+}
+
+static inline u32 axi_chan_ioread32(struct axi_dma_chan *chan, u32 reg)
+{
+	return ioread32(chan->chan_regs + reg);
+}
+
+static inline void
+axi_chan_iowrite64(struct axi_dma_chan *chan, u32 reg, u64 val)
+{
+	/*
+	 * We split one 64 bit write for two 32 bit write as some HW doesn't
+	 * support 64 bit access.
+	 */
+	iowrite32(lower_32_bits(val), chan->chan_regs + reg);
+	iowrite32(upper_32_bits(val), chan->chan_regs + reg + 4);
+}
+
+static inline void axi_dma_disable(struct axi_dma_chip *chip)
+{
+	u32 val;
+
+	val = axi_dma_ioread32(chip, DMAC_CFG);
+	val &= ~DMAC_EN_MASK;
+	axi_dma_iowrite32(chip, DMAC_CFG, val);
+}
+
+static inline void axi_dma_enable(struct axi_dma_chip *chip)
+{
+	u32 val;
+
+	val = axi_dma_ioread32(chip, DMAC_CFG);
+	val |= DMAC_EN_MASK;
+	axi_dma_iowrite32(chip, DMAC_CFG, val);
+}
+
+static inline void axi_dma_irq_disable(struct axi_dma_chip *chip)
+{
+	u32 val;
+
+	val = axi_dma_ioread32(chip, DMAC_CFG);
+	val &= ~INT_EN_MASK;
+	axi_dma_iowrite32(chip, DMAC_CFG, val);
+}
+
+static inline void axi_dma_irq_enable(struct axi_dma_chip *chip)
+{
+	u32 val;
+
+	val = axi_dma_ioread32(chip, DMAC_CFG);
+	val |= INT_EN_MASK;
+	axi_dma_iowrite32(chip, DMAC_CFG, val);
+}
+
+static inline void axi_chan_irq_disable(struct axi_dma_chan *chan, u32 irq_mask)
+{
+	u32 val;
+
+	if (likely(irq_mask == DWAXIDMAC_IRQ_ALL)) {
+		axi_chan_iowrite32(chan, CH_INTSTATUS_ENA, DWAXIDMAC_IRQ_NONE);
+	} else {
+		val = axi_chan_ioread32(chan, CH_INTSTATUS_ENA);
+		val &= ~irq_mask;
+		axi_chan_iowrite32(chan, CH_INTSTATUS_ENA, val);
+	}
+}
+
+static inline void axi_chan_irq_set(struct axi_dma_chan *chan, u32 irq_mask)
+{
+	axi_chan_iowrite32(chan, CH_INTSTATUS_ENA, irq_mask);
+}
+
+static inline void axi_chan_irq_sig_set(struct axi_dma_chan *chan, u32 irq_mask)
+{
+	axi_chan_iowrite32(chan, CH_INTSIGNAL_ENA, irq_mask);
+}
+
+static inline void axi_chan_irq_clear(struct axi_dma_chan *chan, u32 irq_mask)
+{
+	axi_chan_iowrite32(chan, CH_INTCLEAR, irq_mask);
+}
+
+static inline u32 axi_chan_irq_read(struct axi_dma_chan *chan)
+{
+	return axi_chan_ioread32(chan, CH_INTSTATUS);
+}
+
+static inline void axi_chan_disable(struct axi_dma_chan *chan)
+{
+	u32 val;
+
+	val = axi_dma_ioread32(chan->chip, DMAC_CHEN);
+	val &= ~(BIT(chan->id) << DMAC_CHAN_EN_SHIFT);
+	val |=   BIT(chan->id) << DMAC_CHAN_EN_WE_SHIFT;
+	axi_dma_iowrite32(chan->chip, DMAC_CHEN, val);
+}
+
+static inline void axi_chan_enable(struct axi_dma_chan *chan)
+{
+	u32 val;
+
+	val = axi_dma_ioread32(chan->chip, DMAC_CHEN);
+	val |= BIT(chan->id) << DMAC_CHAN_EN_SHIFT |
+	       BIT(chan->id) << DMAC_CHAN_EN_WE_SHIFT;
+	axi_dma_iowrite32(chan->chip, DMAC_CHEN, val);
+}
+
+static inline bool axi_chan_is_hw_enable(struct axi_dma_chan *chan)
+{
+	u32 val;
+
+	val = axi_dma_ioread32(chan->chip, DMAC_CHEN);
+
+	return !!(val & (BIT(chan->id) << DMAC_CHAN_EN_SHIFT));
+}
+
+static void axi_dma_hw_init(struct axi_dma_chip *chip)
+{
+	u32 i;
+
+	for (i = 0; i < chip->dw->hdata->nr_channels; i++) {
+		axi_chan_irq_disable(&chip->dw->chan[i], DWAXIDMAC_IRQ_ALL);
+		axi_chan_disable(&chip->dw->chan[i]);
+	}
+}
+
+static u32 axi_chan_get_xfer_width(struct axi_dma_chan *chan, dma_addr_t src,
+				   dma_addr_t dst, size_t len)
+{
+	u32 max_width = chan->chip->dw->hdata->m_data_width;
+
+	return __ffs(src | dst | len | BIT(max_width));
+}
+
+static inline const char *axi_chan_name(struct axi_dma_chan *chan)
+{
+	return dma_chan_name(&chan->vc.chan);
+}
+
+static struct axi_dma_desc *axi_desc_get(struct axi_dma_chan *chan)
+{
+	struct dw_axi_dma *dw = chan->chip->dw;
+	struct axi_dma_desc *desc;
+	dma_addr_t phys;
+
+	desc = dma_pool_zalloc(dw->desc_pool, GFP_NOWAIT, &phys);
+	if (unlikely(!desc)) {
+		dev_err(chan2dev(chan), "%s: not enough descriptors available\n",
+			axi_chan_name(chan));
+		return NULL;
+	}
+
+	atomic_inc(&chan->descs_allocated);
+	INIT_LIST_HEAD(&desc->xfer_list);
+	desc->vd.tx.phys = phys;
+	desc->chan = chan;
+
+	return desc;
+}
+
+static void axi_desc_put(struct axi_dma_desc *desc)
+{
+	struct axi_dma_chan *chan = desc->chan;
+	struct dw_axi_dma *dw = chan->chip->dw;
+	struct axi_dma_desc *child, *_next;
+	unsigned int descs_put = 0;
+
+	list_for_each_entry_safe(child, _next, &desc->xfer_list, xfer_list) {
+		list_del(&child->xfer_list);
+		dma_pool_free(dw->desc_pool, child, child->vd.tx.phys);
+		descs_put++;
+	}
+
+	dma_pool_free(dw->desc_pool, desc, desc->vd.tx.phys);
+	descs_put++;
+
+	atomic_sub(descs_put, &chan->descs_allocated);
+	dev_vdbg(chan2dev(chan), "%s: %d descs put, %d still allocated\n",
+		axi_chan_name(chan), descs_put,
+		atomic_read(&chan->descs_allocated));
+}
+
+static void vchan_desc_put(struct virt_dma_desc *vdesc)
+{
+	axi_desc_put(vd_to_axi_desc(vdesc));
+}
+
+static enum dma_status
+dma_chan_tx_status(struct dma_chan *dchan, dma_cookie_t cookie,
+		  struct dma_tx_state *txstate)
+{
+	struct axi_dma_chan *chan = dchan_to_axi_dma_chan(dchan);
+	enum dma_status ret;
+
+	ret = dma_cookie_status(dchan, cookie, txstate);
+
+	if (chan->is_paused && ret == DMA_IN_PROGRESS)
+		ret = DMA_PAUSED;
+
+	return ret;
+}
+
+static void write_desc_llp(struct axi_dma_desc *desc, dma_addr_t adr)
+{
+	desc->lli.llp = cpu_to_le64(adr);
+}
+
+static void write_chan_llp(struct axi_dma_chan *chan, dma_addr_t adr)
+{
+	axi_chan_iowrite64(chan, CH_LLP, adr);
+}
+
+/* Called in chan locked context */
+static void axi_chan_block_xfer_start(struct axi_dma_chan *chan,
+				      struct axi_dma_desc *first)
+{
+	u32 priority = chan->chip->dw->hdata->priority[chan->id];
+	u32 reg, irq_mask;
+	u8 lms = 0; /* Select AXI0 master for LLI fetching */
+
+	if (unlikely(axi_chan_is_hw_enable(chan))) {
+		dev_err(chan2dev(chan), "%s is non-idle!\n",
+			axi_chan_name(chan));
+
+		return;
+	}
+
+	axi_dma_enable(chan->chip);
+
+	reg = (DWAXIDMAC_MBLK_TYPE_LL << CH_CFG_L_DST_MULTBLK_TYPE_POS |
+	       DWAXIDMAC_MBLK_TYPE_LL << CH_CFG_L_SRC_MULTBLK_TYPE_POS);
+	axi_chan_iowrite32(chan, CH_CFG_L, reg);
+
+	reg = (DWAXIDMAC_TT_FC_MEM_TO_MEM_DMAC << CH_CFG_H_TT_FC_POS |
+	       priority << CH_CFG_H_PRIORITY_POS |
+	       DWAXIDMAC_HS_SEL_HW << CH_CFG_H_HS_SEL_DST_POS |
+	       DWAXIDMAC_HS_SEL_HW << CH_CFG_H_HS_SEL_SRC_POS);
+	axi_chan_iowrite32(chan, CH_CFG_H, reg);
+
+	write_chan_llp(chan, first->vd.tx.phys | lms);
+
+	irq_mask = DWAXIDMAC_IRQ_DMA_TRF | DWAXIDMAC_IRQ_ALL_ERR;
+	axi_chan_irq_sig_set(chan, irq_mask);
+
+	/* Generate 'suspend' status but don't generate interrupt */
+	irq_mask |= DWAXIDMAC_IRQ_SUSPENDED;
+	axi_chan_irq_set(chan, irq_mask);
+
+	axi_chan_enable(chan);
+}
+
+static void axi_chan_start_first_queued(struct axi_dma_chan *chan)
+{
+	struct axi_dma_desc *desc;
+	struct virt_dma_desc *vd;
+
+	vd = vchan_next_desc(&chan->vc);
+	if (!vd)
+		return;
+
+	desc = vd_to_axi_desc(vd);
+	dev_vdbg(chan2dev(chan), "%s: started %u\n", axi_chan_name(chan),
+		vd->tx.cookie);
+	axi_chan_block_xfer_start(chan, desc);
+}
+
+static void dma_chan_issue_pending(struct dma_chan *dchan)
+{
+	struct axi_dma_chan *chan = dchan_to_axi_dma_chan(dchan);
+	unsigned long flags;
+
+	spin_lock_irqsave(&chan->vc.lock, flags);
+	if (vchan_issue_pending(&chan->vc))
+		axi_chan_start_first_queued(chan);
+	spin_unlock_irqrestore(&chan->vc.lock, flags);
+}
+
+static int dma_chan_alloc_chan_resources(struct dma_chan *dchan)
+{
+	struct axi_dma_chan *chan = dchan_to_axi_dma_chan(dchan);
+
+	/* ASSERT: channel is idle */
+	if (axi_chan_is_hw_enable(chan)) {
+		dev_err(chan2dev(chan), "%s is non-idle!\n",
+			axi_chan_name(chan));
+		return -EBUSY;
+	}
+
+	dev_vdbg(dchan2dev(dchan), "%s: allocating\n", axi_chan_name(chan));
+
+	pm_runtime_get(chan->chip->dev);
+
+	return 0;
+}
+
+static void dma_chan_free_chan_resources(struct dma_chan *dchan)
+{
+	struct axi_dma_chan *chan = dchan_to_axi_dma_chan(dchan);
+
+	/* ASSERT: channel is idle */
+	if (axi_chan_is_hw_enable(chan))
+		dev_err(dchan2dev(dchan), "%s is non-idle!\n",
+			axi_chan_name(chan));
+
+	axi_chan_disable(chan);
+	axi_chan_irq_disable(chan, DWAXIDMAC_IRQ_ALL);
+
+	vchan_free_chan_resources(&chan->vc);
+
+	dev_vdbg(dchan2dev(dchan),
+		 "%s: free resources, descriptor still allocated: %u\n",
+		 axi_chan_name(chan), atomic_read(&chan->descs_allocated));
+
+	pm_runtime_put(chan->chip->dev);
+}
+
+/*
+ * If DW_axi_dmac sees CHx_CTL.ShadowReg_Or_LLI_Last bit of the fetched LLI
+ * as 1, it understands that the current block is the final block in the
+ * transfer and completes the DMA transfer operation at the end of current
+ * block transfer.
+ */
+static void set_desc_last(struct axi_dma_desc *desc)
+{
+	u32 val;
+
+	val = le32_to_cpu(desc->lli.ctl_hi);
+	val |= CH_CTL_H_LLI_LAST;
+	desc->lli.ctl_hi = cpu_to_le32(val);
+}
+
+static void write_desc_sar(struct axi_dma_desc *desc, dma_addr_t adr)
+{
+	desc->lli.sar = cpu_to_le64(adr);
+}
+
+static void write_desc_dar(struct axi_dma_desc *desc, dma_addr_t adr)
+{
+	desc->lli.dar = cpu_to_le64(adr);
+}
+
+static void set_desc_src_master(struct axi_dma_desc *desc)
+{
+	u32 val;
+
+	/* Select AXI0 for source master */
+	val = le32_to_cpu(desc->lli.ctl_lo);
+	val &= ~CH_CTL_L_SRC_MAST;
+	desc->lli.ctl_lo = cpu_to_le32(val);
+}
+
+static void set_desc_dest_master(struct axi_dma_desc *desc)
+{
+	u32 val;
+
+	/* Select AXI1 for source master if available */
+	val = le32_to_cpu(desc->lli.ctl_lo);
+	if (desc->chan->chip->dw->hdata->nr_masters > 1)
+		val |= CH_CTL_L_DST_MAST;
+	else
+		val &= ~CH_CTL_L_DST_MAST;
+
+	desc->lli.ctl_lo = cpu_to_le32(val);
+}
+
+static struct dma_async_tx_descriptor *
+dma_chan_prep_dma_memcpy(struct dma_chan *dchan, dma_addr_t dst_adr,
+			 dma_addr_t src_adr, size_t len, unsigned long flags)
+{
+	struct axi_dma_desc *first = NULL, *desc = NULL, *prev = NULL;
+	struct axi_dma_chan *chan = dchan_to_axi_dma_chan(dchan);
+	size_t block_ts, max_block_ts, xfer_len;
+	u32 xfer_width, reg;
+	u8 lms = 0; /* Select AXI0 master for LLI fetching */
+
+	dev_dbg(chan2dev(chan), "%s: memcpy: src: %pad dst: %pad length: %zd flags: %#lx",
+		axi_chan_name(chan), &src_adr, &dst_adr, len, flags);
+
+	max_block_ts = chan->chip->dw->hdata->block_size[chan->id];
+
+	while (len) {
+		xfer_len = len;
+
+		/*
+		 * Take care for the alignment.
+		 * Actually source and destination widths can be different, but
+		 * make them same to be simpler.
+		 */
+		xfer_width = axi_chan_get_xfer_width(chan, src_adr, dst_adr, xfer_len);
+
+		/*
+		 * block_ts indicates the total number of data of width
+		 * to be transferred in a DMA block transfer.
+		 * BLOCK_TS register should be set to block_ts - 1
+		 */
+		block_ts = xfer_len >> xfer_width;
+		if (block_ts > max_block_ts) {
+			block_ts = max_block_ts;
+			xfer_len = max_block_ts << xfer_width;
+		}
+
+		desc = axi_desc_get(chan);
+		if (unlikely(!desc))
+			goto err_desc_get;
+
+		write_desc_sar(desc, src_adr);
+		write_desc_dar(desc, dst_adr);
+		desc->lli.block_ts_lo = cpu_to_le32(block_ts - 1);
+
+		reg = CH_CTL_H_LLI_VALID;
+		if (chan->chip->dw->hdata->restrict_axi_burst_len) {
+			u32 burst_len = chan->chip->dw->hdata->axi_rw_burst_len;
+
+			reg |= (CH_CTL_H_ARLEN_EN |
+				burst_len << CH_CTL_H_ARLEN_POS |
+				CH_CTL_H_AWLEN_EN |
+				burst_len << CH_CTL_H_AWLEN_POS);
+		}
+		desc->lli.ctl_hi = cpu_to_le32(reg);
+
+		reg = (DWAXIDMAC_BURST_TRANS_LEN_4 << CH_CTL_L_DST_MSIZE_POS |
+		       DWAXIDMAC_BURST_TRANS_LEN_4 << CH_CTL_L_SRC_MSIZE_POS |
+		       xfer_width << CH_CTL_L_DST_WIDTH_POS |
+		       xfer_width << CH_CTL_L_SRC_WIDTH_POS |
+		       DWAXIDMAC_CH_CTL_L_INC << CH_CTL_L_DST_INC_POS |
+		       DWAXIDMAC_CH_CTL_L_INC << CH_CTL_L_SRC_INC_POS);
+		desc->lli.ctl_lo = cpu_to_le32(reg);
+
+		set_desc_src_master(desc);
+		set_desc_dest_master(desc);
+
+		/* Manage transfer list (xfer_list) */
+		if (!first) {
+			first = desc;
+		} else {
+			list_add_tail(&desc->xfer_list, &first->xfer_list);
+			write_desc_llp(prev, desc->vd.tx.phys | lms);
+		}
+		prev = desc;
+
+		/* update the length and addresses for the next loop cycle */
+		len -= xfer_len;
+		dst_adr += xfer_len;
+		src_adr += xfer_len;
+	}
+
+	/* Total len of src/dest sg == 0, so no descriptor were allocated */
+	if (unlikely(!first))
+		return NULL;
+
+	/* Set end-of-link to the last link descriptor of list */
+	set_desc_last(desc);
+
+	return vchan_tx_prep(&chan->vc, &first->vd, flags);
+
+err_desc_get:
+	axi_desc_put(first);
+	return NULL;
+}
+
+static void axi_chan_dump_lli(struct axi_dma_chan *chan,
+			      struct axi_dma_desc *desc)
+{
+	dev_err(dchan2dev(&chan->vc.chan),
+		"SAR: 0x%llx DAR: 0x%llx LLP: 0x%llx BTS 0x%x CTL: 0x%x:%08x",
+		le64_to_cpu(desc->lli.sar),
+		le64_to_cpu(desc->lli.dar),
+		le64_to_cpu(desc->lli.llp),
+		le32_to_cpu(desc->lli.block_ts_lo),
+		le32_to_cpu(desc->lli.ctl_hi),
+		le32_to_cpu(desc->lli.ctl_lo));
+}
+
+static void axi_chan_list_dump_lli(struct axi_dma_chan *chan,
+				   struct axi_dma_desc *desc_head)
+{
+	struct axi_dma_desc *desc;
+
+	axi_chan_dump_lli(chan, desc_head);
+	list_for_each_entry(desc, &desc_head->xfer_list, xfer_list)
+		axi_chan_dump_lli(chan, desc);
+}
+
+static noinline void axi_chan_handle_err(struct axi_dma_chan *chan, u32 status)
+{
+	struct virt_dma_desc *vd;
+	unsigned long flags;
+
+	spin_lock_irqsave(&chan->vc.lock, flags);
+
+	axi_chan_disable(chan);
+
+	/* The bad descriptor currently is in the head of vc list */
+	vd = vchan_next_desc(&chan->vc);
+	/* Remove the completed descriptor from issued list */
+	list_del(&vd->node);
+
+	/* WARN about bad descriptor */
+	dev_err(chan2dev(chan),
+		"Bad descriptor submitted for %s, cookie: %d, irq: 0x%08x\n",
+		axi_chan_name(chan), vd->tx.cookie, status);
+	axi_chan_list_dump_lli(chan, vd_to_axi_desc(vd));
+
+	vchan_cookie_complete(vd);
+
+	/* Try to restart the controller */
+	axi_chan_start_first_queued(chan);
+
+	spin_unlock_irqrestore(&chan->vc.lock, flags);
+}
+
+static void axi_chan_block_xfer_complete(struct axi_dma_chan *chan)
+{
+	struct virt_dma_desc *vd;
+	unsigned long flags;
+
+	spin_lock_irqsave(&chan->vc.lock, flags);
+	if (unlikely(axi_chan_is_hw_enable(chan))) {
+		dev_err(chan2dev(chan), "BUG: %s caught DWAXIDMAC_IRQ_DMA_TRF, but channel not idle!\n",
+			axi_chan_name(chan));
+		axi_chan_disable(chan);
+	}
+
+	/* The completed descriptor currently is in the head of vc list */
+	vd = vchan_next_desc(&chan->vc);
+	/* 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);
+
+	spin_unlock_irqrestore(&chan->vc.lock, flags);
+}
+
+static irqreturn_t dw_axi_dma_interrupt(int irq, void *dev_id)
+{
+	struct axi_dma_chip *chip = dev_id;
+	struct dw_axi_dma *dw = chip->dw;
+	struct axi_dma_chan *chan;
+
+	u32 status, i;
+
+	/* Disable DMAC inerrupts. We'll enable them after processing chanels */
+	axi_dma_irq_disable(chip);
+
+	/* Poll, clear and process every chanel interrupt status */
+	for (i = 0; i < dw->hdata->nr_channels; i++) {
+		chan = &dw->chan[i];
+		status = axi_chan_irq_read(chan);
+		axi_chan_irq_clear(chan, status);
+
+		dev_vdbg(chip->dev, "%s %u IRQ status: 0x%08x\n",
+			axi_chan_name(chan), i, status);
+
+		if (status & DWAXIDMAC_IRQ_ALL_ERR)
+			axi_chan_handle_err(chan, status);
+		else if (status & DWAXIDMAC_IRQ_DMA_TRF)
+			axi_chan_block_xfer_complete(chan);
+	}
+
+	/* Re-enable interrupts */
+	axi_dma_irq_enable(chip);
+
+	return IRQ_HANDLED;
+}
+
+static int dma_chan_terminate_all(struct dma_chan *dchan)
+{
+	struct axi_dma_chan *chan = dchan_to_axi_dma_chan(dchan);
+	unsigned long flags;
+	LIST_HEAD(head);
+
+	spin_lock_irqsave(&chan->vc.lock, flags);
+
+	axi_chan_disable(chan);
+
+	vchan_get_all_descriptors(&chan->vc, &head);
+
+	/*
+	 * As vchan_dma_desc_free_list can access to desc_allocated list
+	 * we need to call it in vc.lock context.
+	 */
+	vchan_dma_desc_free_list(&chan->vc, &head);
+
+	spin_unlock_irqrestore(&chan->vc.lock, flags);
+
+	dev_vdbg(dchan2dev(dchan), "terminated: %s\n", axi_chan_name(chan));
+
+	return 0;
+}
+
+static int dma_chan_pause(struct dma_chan *dchan)
+{
+	struct axi_dma_chan *chan = dchan_to_axi_dma_chan(dchan);
+	unsigned long flags;
+	unsigned int timeout = 20; /* timeout iterations */
+	u32 val;
+
+	spin_lock_irqsave(&chan->vc.lock, flags);
+
+	val = axi_dma_ioread32(chan->chip, DMAC_CHEN);
+	val |= BIT(chan->id) << DMAC_CHAN_SUSP_SHIFT |
+	       BIT(chan->id) << DMAC_CHAN_SUSP_WE_SHIFT;
+	axi_dma_iowrite32(chan->chip, DMAC_CHEN, val);
+
+	do  {
+		if (axi_chan_irq_read(chan) & DWAXIDMAC_IRQ_SUSPENDED)
+			break;
+
+		udelay(2);
+	} while (--timeout);
+
+	axi_chan_irq_clear(chan, DWAXIDMAC_IRQ_SUSPENDED);
+
+	chan->is_paused = true;
+
+	spin_unlock_irqrestore(&chan->vc.lock, flags);
+
+	return timeout ? 0 : -EAGAIN;
+}
+
+/* Called in chan locked context */
+static inline void axi_chan_resume(struct axi_dma_chan *chan)
+{
+	u32 val;
+
+	val = axi_dma_ioread32(chan->chip, DMAC_CHEN);
+	val &= ~(BIT(chan->id) << DMAC_CHAN_SUSP_SHIFT);
+	val |=  (BIT(chan->id) << DMAC_CHAN_SUSP_WE_SHIFT);
+	axi_dma_iowrite32(chan->chip, DMAC_CHEN, val);
+
+	chan->is_paused = false;
+}
+
+static int dma_chan_resume(struct dma_chan *dchan)
+{
+	struct axi_dma_chan *chan = dchan_to_axi_dma_chan(dchan);
+	unsigned long flags;
+
+	spin_lock_irqsave(&chan->vc.lock, flags);
+
+	if (chan->is_paused)
+		axi_chan_resume(chan);
+
+	spin_unlock_irqrestore(&chan->vc.lock, flags);
+
+	return 0;
+}
+
+static int axi_dma_suspend(struct axi_dma_chip *chip)
+{
+	axi_dma_irq_disable(chip);
+	axi_dma_disable(chip);
+
+	clk_disable_unprepare(chip->core_clk);
+	clk_disable_unprepare(chip->cfgr_clk);
+
+	return 0;
+}
+
+static int axi_dma_resume(struct axi_dma_chip *chip)
+{
+	int ret;
+
+	ret = clk_prepare_enable(chip->cfgr_clk);
+	if (ret < 0)
+		return ret;
+
+	ret = clk_prepare_enable(chip->core_clk);
+	if (ret < 0)
+		return ret;
+
+	axi_dma_enable(chip);
+	axi_dma_irq_enable(chip);
+
+	return 0;
+}
+
+static int __maybe_unused axi_dma_runtime_suspend(struct device *dev)
+{
+	struct axi_dma_chip *chip = dev_get_drvdata(dev);
+
+	return axi_dma_suspend(chip);
+}
+
+static int __maybe_unused axi_dma_runtime_resume(struct device *dev)
+{
+	struct axi_dma_chip *chip = dev_get_drvdata(dev);
+
+	return axi_dma_resume(chip);
+}
+
+static int parse_device_properties(struct axi_dma_chip *chip)
+{
+	struct device *dev = chip->dev;
+	u32 tmp, carr[DMAC_MAX_CHANNELS];
+	int ret;
+
+	ret = device_property_read_u32(dev, "dma-channels", &tmp);
+	if (ret)
+		return ret;
+	if (tmp == 0 || tmp > DMAC_MAX_CHANNELS)
+		return -EINVAL;
+
+	chip->dw->hdata->nr_channels = tmp;
+
+	ret = device_property_read_u32(dev, "snps,dma-masters", &tmp);
+	if (ret)
+		return ret;
+	if (tmp == 0 || tmp > DMAC_MAX_MASTERS)
+		return -EINVAL;
+
+	chip->dw->hdata->nr_masters = tmp;
+
+	ret = device_property_read_u32(dev, "snps,data-width", &tmp);
+	if (ret)
+		return ret;
+	if (tmp > DWAXIDMAC_TRANS_WIDTH_MAX)
+		return -EINVAL;
+
+	chip->dw->hdata->m_data_width = tmp;
+
+	ret = device_property_read_u32_array(dev, "snps,block-size", carr,
+					     chip->dw->hdata->nr_channels);
+	if (ret)
+		return ret;
+	for (tmp = 0; tmp < chip->dw->hdata->nr_channels; tmp++) {
+		if (carr[tmp] == 0 || carr[tmp] > DMAC_MAX_BLK_SIZE)
+			return -EINVAL;
+
+		chip->dw->hdata->block_size[tmp] = carr[tmp];
+	}
+
+	ret = device_property_read_u32_array(dev, "snps,priority", carr,
+					     chip->dw->hdata->nr_channels);
+	if (ret)
+		return ret;
+	/* Priority value must be programmed within [0:nr_channels-1] range */
+	for (tmp = 0; tmp < chip->dw->hdata->nr_channels; tmp++) {
+		if (carr[tmp] >= chip->dw->hdata->nr_channels)
+			return -EINVAL;
+
+		chip->dw->hdata->priority[tmp] = carr[tmp];
+	}
+
+	/* axi-max-burst-len is optional property */
+	ret = device_property_read_u32(dev, "snps,axi-max-burst-len", &tmp);
+	if (!ret) {
+		if (tmp > DWAXIDMAC_ARWLEN_MAX + 1)
+			return -EINVAL;
+		if (tmp < DWAXIDMAC_ARWLEN_MIN + 1)
+			return -EINVAL;
+
+		chip->dw->hdata->restrict_axi_burst_len = true;
+		chip->dw->hdata->axi_rw_burst_len = tmp - 1;
+	}
+
+	return 0;
+}
+
+static int dw_probe(struct platform_device *pdev)
+{
+	struct axi_dma_chip *chip;
+	struct resource *mem;
+	struct dw_axi_dma *dw;
+	struct dw_axi_dma_hcfg *hdata;
+	u32 i;
+	int ret;
+
+	chip = devm_kzalloc(&pdev->dev, sizeof(*chip), GFP_KERNEL);
+	if (!chip)
+		return -ENOMEM;
+
+	dw = devm_kzalloc(&pdev->dev, sizeof(*dw), GFP_KERNEL);
+	if (!dw)
+		return -ENOMEM;
+
+	hdata = devm_kzalloc(&pdev->dev, sizeof(*hdata), GFP_KERNEL);
+	if (!hdata)
+		return -ENOMEM;
+
+	chip->dw = dw;
+	chip->dev = &pdev->dev;
+	chip->dw->hdata = hdata;
+
+	chip->irq = platform_get_irq(pdev, 0);
+	if (chip->irq < 0)
+		return chip->irq;
+
+	mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	chip->regs = devm_ioremap_resource(chip->dev, mem);
+	if (IS_ERR(chip->regs))
+		return PTR_ERR(chip->regs);
+
+	chip->core_clk = devm_clk_get(chip->dev, "core-clk");
+	if (IS_ERR(chip->core_clk))
+		return PTR_ERR(chip->core_clk);
+
+	chip->cfgr_clk = devm_clk_get(chip->dev, "cfgr-clk");
+	if (IS_ERR(chip->cfgr_clk))
+		return PTR_ERR(chip->cfgr_clk);
+
+	ret = parse_device_properties(chip);
+	if (ret)
+		return ret;
+
+	dw->chan = devm_kcalloc(chip->dev, hdata->nr_channels,
+				sizeof(*dw->chan), GFP_KERNEL);
+	if (!dw->chan)
+		return -ENOMEM;
+
+	ret = devm_request_irq(chip->dev, chip->irq, dw_axi_dma_interrupt,
+			       IRQF_SHARED, KBUILD_MODNAME, chip);
+	if (ret)
+		return ret;
+
+	/* Lli address must be aligned to a 64-byte boundary */
+	dw->desc_pool = dmam_pool_create(KBUILD_MODNAME, chip->dev,
+					 sizeof(struct axi_dma_desc), 64, 0);
+	if (!dw->desc_pool) {
+		dev_err(chip->dev, "No memory for descriptors dma pool\n");
+		return -ENOMEM;
+	}
+
+	INIT_LIST_HEAD(&dw->dma.channels);
+	for (i = 0; i < hdata->nr_channels; i++) {
+		struct axi_dma_chan *chan = &dw->chan[i];
+
+		chan->chip = chip;
+		chan->id = i;
+		chan->chan_regs = chip->regs + COMMON_REG_LEN + i * CHAN_REG_LEN;
+		atomic_set(&chan->descs_allocated, 0);
+
+		chan->vc.desc_free = vchan_desc_put;
+		vchan_init(&chan->vc, &dw->dma);
+	}
+
+	/* Set capabilities */
+	dma_cap_set(DMA_MEMCPY, dw->dma.cap_mask);
+
+	/* DMA capabilities */
+	dw->dma.chancnt = hdata->nr_channels;
+	dw->dma.src_addr_widths = AXI_DMA_BUSWIDTHS;
+	dw->dma.dst_addr_widths = AXI_DMA_BUSWIDTHS;
+	dw->dma.directions = BIT(DMA_MEM_TO_MEM);
+	dw->dma.residue_granularity = DMA_RESIDUE_GRANULARITY_DESCRIPTOR;
+
+	dw->dma.dev = chip->dev;
+	dw->dma.device_tx_status = dma_chan_tx_status;
+	dw->dma.device_issue_pending = dma_chan_issue_pending;
+	dw->dma.device_terminate_all = dma_chan_terminate_all;
+	dw->dma.device_pause = dma_chan_pause;
+	dw->dma.device_resume = dma_chan_resume;
+
+	dw->dma.device_alloc_chan_resources = dma_chan_alloc_chan_resources;
+	dw->dma.device_free_chan_resources = dma_chan_free_chan_resources;
+
+	dw->dma.device_prep_dma_memcpy = dma_chan_prep_dma_memcpy;
+
+	platform_set_drvdata(pdev, chip);
+
+	pm_runtime_enable(chip->dev);
+
+	/*
+	 * We can't just call pm_runtime_get here instead of
+	 * pm_runtime_get_noresume + axi_dma_resume because we need
+	 * driver to work also without Runtime PM.
+	 */
+	pm_runtime_get_noresume(chip->dev);
+	ret = axi_dma_resume(chip);
+	if (ret < 0)
+		goto err_pm_disable;
+
+	axi_dma_hw_init(chip);
+
+	pm_runtime_put(chip->dev);
+
+	ret = dma_async_device_register(&dw->dma);
+	if (ret)
+		goto err_pm_disable;
+
+	dev_info(chip->dev, "DesignWare AXI DMA Controller, %d channels\n",
+		 dw->hdata->nr_channels);
+
+	return 0;
+
+err_pm_disable:
+	pm_runtime_disable(chip->dev);
+
+	return ret;
+}
+
+static int dw_remove(struct platform_device *pdev)
+{
+	struct axi_dma_chip *chip = platform_get_drvdata(pdev);
+	struct dw_axi_dma *dw = chip->dw;
+	struct axi_dma_chan *chan, *_chan;
+	u32 i;
+
+	/* Enable clk before accessing to registers */
+	clk_prepare_enable(chip->cfgr_clk);
+	clk_prepare_enable(chip->core_clk);
+	axi_dma_irq_disable(chip);
+	for (i = 0; i < dw->hdata->nr_channels; i++) {
+		axi_chan_disable(&chip->dw->chan[i]);
+		axi_chan_irq_disable(&chip->dw->chan[i], DWAXIDMAC_IRQ_ALL);
+	}
+	axi_dma_disable(chip);
+
+	pm_runtime_disable(chip->dev);
+	axi_dma_suspend(chip);
+
+	devm_free_irq(chip->dev, chip->irq, chip);
+
+	list_for_each_entry_safe(chan, _chan, &dw->dma.channels,
+			vc.chan.device_node) {
+		list_del(&chan->vc.chan.device_node);
+		tasklet_kill(&chan->vc.task);
+	}
+
+	dma_async_device_unregister(&dw->dma);
+
+	return 0;
+}
+
+static const struct dev_pm_ops dw_axi_dma_pm_ops = {
+	SET_RUNTIME_PM_OPS(axi_dma_runtime_suspend, axi_dma_runtime_resume, NULL)
+};
+
+static const struct of_device_id dw_dma_of_id_table[] = {
+	{ .compatible = "snps,axi-dma-1.01a" },
+	{}
+};
+MODULE_DEVICE_TABLE(of, dw_dma_of_id_table);
+
+static struct platform_driver dw_driver = {
+	.probe		= dw_probe,
+	.remove		= dw_remove,
+	.driver = {
+		.name	= KBUILD_MODNAME,
+		.of_match_table = of_match_ptr(dw_dma_of_id_table),
+		.pm = &dw_axi_dma_pm_ops,
+	},
+};
+module_platform_driver(dw_driver);
+
+MODULE_LICENSE("GPL v2");
+MODULE_DESCRIPTION("Synopsys DesignWare AXI DMA Controller platform driver");
+MODULE_AUTHOR("Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>");
diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac.h b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
new file mode 100644
index 000000000000..f8888dc0b8dc
--- /dev/null
+++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac.h
@@ -0,0 +1,334 @@
+// SPDX-License-Identifier:  GPL-2.0
+// (C) 2017-2018 Synopsys, Inc. (www.synopsys.com)
+
+/*
+ * Synopsys DesignWare AXI DMA Controller driver.
+ *
+ * Author: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>
+ */
+
+#ifndef _AXI_DMA_PLATFORM_H
+#define _AXI_DMA_PLATFORM_H
+
+#include <linux/bitops.h>
+#include <linux/clk.h>
+#include <linux/device.h>
+#include <linux/dmaengine.h>
+#include <linux/types.h>
+
+#include "../virt-dma.h"
+
+#define DMAC_MAX_CHANNELS	8
+#define DMAC_MAX_MASTERS	2
+#define DMAC_MAX_BLK_SIZE	0x200000
+
+struct dw_axi_dma_hcfg {
+	u32	nr_channels;
+	u32	nr_masters;
+	u32	m_data_width;
+	u32	block_size[DMAC_MAX_CHANNELS];
+	u32	priority[DMAC_MAX_CHANNELS];
+	/* maximum supported axi burst length */
+	u32	axi_rw_burst_len;
+	bool	restrict_axi_burst_len;
+};
+
+struct axi_dma_chan {
+	struct axi_dma_chip		*chip;
+	void __iomem			*chan_regs;
+	u8				id;
+	atomic_t			descs_allocated;
+
+	struct virt_dma_chan		vc;
+
+	/* these other elements are all protected by vc.lock */
+	bool				is_paused;
+};
+
+struct dw_axi_dma {
+	struct dma_device	dma;
+	struct dw_axi_dma_hcfg	*hdata;
+	struct dma_pool		*desc_pool;
+
+	/* channels */
+	struct axi_dma_chan	*chan;
+};
+
+struct axi_dma_chip {
+	struct device		*dev;
+	int			irq;
+	void __iomem		*regs;
+	struct clk		*core_clk;
+	struct clk		*cfgr_clk;
+	struct dw_axi_dma	*dw;
+};
+
+/* LLI == Linked List Item */
+struct __packed axi_dma_lli {
+	__le64		sar;
+	__le64		dar;
+	__le32		block_ts_lo;
+	__le32		block_ts_hi;
+	__le64		llp;
+	__le32		ctl_lo;
+	__le32		ctl_hi;
+	__le32		sstat;
+	__le32		dstat;
+	__le32		status_lo;
+	__le32		ststus_hi;
+	__le32		reserved_lo;
+	__le32		reserved_hi;
+};
+
+struct axi_dma_desc {
+	struct axi_dma_lli		lli;
+
+	struct virt_dma_desc		vd;
+	struct axi_dma_chan		*chan;
+	struct list_head		xfer_list;
+};
+
+static inline struct device *dchan2dev(struct dma_chan *dchan)
+{
+	return &dchan->dev->device;
+}
+
+static inline struct device *chan2dev(struct axi_dma_chan *chan)
+{
+	return &chan->vc.chan.dev->device;
+}
+
+static inline struct axi_dma_desc *vd_to_axi_desc(struct virt_dma_desc *vd)
+{
+	return container_of(vd, struct axi_dma_desc, vd);
+}
+
+static inline struct axi_dma_chan *vc_to_axi_dma_chan(struct virt_dma_chan *vc)
+{
+	return container_of(vc, struct axi_dma_chan, vc);
+}
+
+static inline struct axi_dma_chan *dchan_to_axi_dma_chan(struct dma_chan *dchan)
+{
+	return vc_to_axi_dma_chan(to_virt_chan(dchan));
+}
+
+
+#define COMMON_REG_LEN		0x100
+#define CHAN_REG_LEN		0x100
+
+/* Common registers offset */
+#define DMAC_ID			0x000 /* R DMAC ID */
+#define DMAC_COMPVER		0x008 /* R DMAC Component Version */
+#define DMAC_CFG		0x010 /* R/W DMAC Configuration */
+#define DMAC_CHEN		0x018 /* R/W DMAC Channel Enable */
+#define DMAC_CHEN_L		0x018 /* R/W DMAC Channel Enable 00-31 */
+#define DMAC_CHEN_H		0x01C /* R/W DMAC Channel Enable 32-63 */
+#define DMAC_INTSTATUS		0x030 /* R DMAC Interrupt Status */
+#define DMAC_COMMON_INTCLEAR	0x038 /* W DMAC Interrupt Clear */
+#define DMAC_COMMON_INTSTATUS_ENA 0x040 /* R DMAC Interrupt Status Enable */
+#define DMAC_COMMON_INTSIGNAL_ENA 0x048 /* R/W DMAC Interrupt Signal Enable */
+#define DMAC_COMMON_INTSTATUS	0x050 /* R DMAC Interrupt Status */
+#define DMAC_RESET		0x058 /* R DMAC Reset Register1 */
+
+/* DMA channel registers offset */
+#define CH_SAR			0x000 /* R/W Chan Source Address */
+#define CH_DAR			0x008 /* R/W Chan Destination Address */
+#define CH_BLOCK_TS		0x010 /* R/W Chan Block Transfer Size */
+#define CH_CTL			0x018 /* R/W Chan Control */
+#define CH_CTL_L		0x018 /* R/W Chan Control 00-31 */
+#define CH_CTL_H		0x01C /* R/W Chan Control 32-63 */
+#define CH_CFG			0x020 /* R/W Chan Configuration */
+#define CH_CFG_L		0x020 /* R/W Chan Configuration 00-31 */
+#define CH_CFG_H		0x024 /* R/W Chan Configuration 32-63 */
+#define CH_LLP			0x028 /* R/W Chan Linked List Pointer */
+#define CH_STATUS		0x030 /* R Chan Status */
+#define CH_SWHSSRC		0x038 /* R/W Chan SW Handshake Source */
+#define CH_SWHSDST		0x040 /* R/W Chan SW Handshake Destination */
+#define CH_BLK_TFR_RESUMEREQ	0x048 /* W Chan Block Transfer Resume Req */
+#define CH_AXI_ID		0x050 /* R/W Chan AXI ID */
+#define CH_AXI_QOS		0x058 /* R/W Chan AXI QOS */
+#define CH_SSTAT		0x060 /* R Chan Source Status */
+#define CH_DSTAT		0x068 /* R Chan Destination Status */
+#define CH_SSTATAR		0x070 /* R/W Chan Source Status Fetch Addr */
+#define CH_DSTATAR		0x078 /* R/W Chan Destination Status Fetch Addr */
+#define CH_INTSTATUS_ENA	0x080 /* R/W Chan Interrupt Status Enable */
+#define CH_INTSTATUS		0x088 /* R/W Chan Interrupt Status */
+#define CH_INTSIGNAL_ENA	0x090 /* R/W Chan Interrupt Signal Enable */
+#define CH_INTCLEAR		0x098 /* W Chan Interrupt Clear */
+
+
+/* DMAC_CFG */
+#define DMAC_EN_POS			0
+#define DMAC_EN_MASK			BIT(DMAC_EN_POS)
+
+#define INT_EN_POS			1
+#define INT_EN_MASK			BIT(INT_EN_POS)
+
+#define DMAC_CHAN_EN_SHIFT		0
+#define DMAC_CHAN_EN_WE_SHIFT		8
+
+#define DMAC_CHAN_SUSP_SHIFT		16
+#define DMAC_CHAN_SUSP_WE_SHIFT		24
+
+/* CH_CTL_H */
+#define CH_CTL_H_ARLEN_EN		BIT(6)
+#define CH_CTL_H_ARLEN_POS		7
+#define CH_CTL_H_AWLEN_EN		BIT(15)
+#define CH_CTL_H_AWLEN_POS		16
+
+enum {
+	DWAXIDMAC_ARWLEN_1		= 0,
+	DWAXIDMAC_ARWLEN_2		= 1,
+	DWAXIDMAC_ARWLEN_4		= 3,
+	DWAXIDMAC_ARWLEN_8		= 7,
+	DWAXIDMAC_ARWLEN_16		= 15,
+	DWAXIDMAC_ARWLEN_32		= 31,
+	DWAXIDMAC_ARWLEN_64		= 63,
+	DWAXIDMAC_ARWLEN_128		= 127,
+	DWAXIDMAC_ARWLEN_256		= 255,
+	DWAXIDMAC_ARWLEN_MIN		= DWAXIDMAC_ARWLEN_1,
+	DWAXIDMAC_ARWLEN_MAX		= DWAXIDMAC_ARWLEN_256
+};
+
+#define CH_CTL_H_LLI_LAST		BIT(30)
+#define CH_CTL_H_LLI_VALID		BIT(31)
+
+/* CH_CTL_L */
+#define CH_CTL_L_LAST_WRITE_EN		BIT(30)
+
+#define CH_CTL_L_DST_MSIZE_POS		18
+#define CH_CTL_L_SRC_MSIZE_POS		14
+
+enum {
+	DWAXIDMAC_BURST_TRANS_LEN_1	= 0,
+	DWAXIDMAC_BURST_TRANS_LEN_4,
+	DWAXIDMAC_BURST_TRANS_LEN_8,
+	DWAXIDMAC_BURST_TRANS_LEN_16,
+	DWAXIDMAC_BURST_TRANS_LEN_32,
+	DWAXIDMAC_BURST_TRANS_LEN_64,
+	DWAXIDMAC_BURST_TRANS_LEN_128,
+	DWAXIDMAC_BURST_TRANS_LEN_256,
+	DWAXIDMAC_BURST_TRANS_LEN_512,
+	DWAXIDMAC_BURST_TRANS_LEN_1024
+};
+
+#define CH_CTL_L_DST_WIDTH_POS		11
+#define CH_CTL_L_SRC_WIDTH_POS		8
+
+#define CH_CTL_L_DST_INC_POS		6
+#define CH_CTL_L_SRC_INC_POS		4
+enum {
+	DWAXIDMAC_CH_CTL_L_INC		= 0,
+	DWAXIDMAC_CH_CTL_L_NOINC
+};
+
+#define CH_CTL_L_DST_MAST		BIT(2)
+#define CH_CTL_L_SRC_MAST		BIT(0)
+
+/* CH_CFG_H */
+#define CH_CFG_H_PRIORITY_POS		17
+#define CH_CFG_H_HS_SEL_DST_POS		4
+#define CH_CFG_H_HS_SEL_SRC_POS		3
+enum {
+	DWAXIDMAC_HS_SEL_HW		= 0,
+	DWAXIDMAC_HS_SEL_SW
+};
+
+#define CH_CFG_H_TT_FC_POS		0
+enum {
+	DWAXIDMAC_TT_FC_MEM_TO_MEM_DMAC	= 0,
+	DWAXIDMAC_TT_FC_MEM_TO_PER_DMAC,
+	DWAXIDMAC_TT_FC_PER_TO_MEM_DMAC,
+	DWAXIDMAC_TT_FC_PER_TO_PER_DMAC,
+	DWAXIDMAC_TT_FC_PER_TO_MEM_SRC,
+	DWAXIDMAC_TT_FC_PER_TO_PER_SRC,
+	DWAXIDMAC_TT_FC_MEM_TO_PER_DST,
+	DWAXIDMAC_TT_FC_PER_TO_PER_DST
+};
+
+/* CH_CFG_L */
+#define CH_CFG_L_DST_MULTBLK_TYPE_POS	2
+#define CH_CFG_L_SRC_MULTBLK_TYPE_POS	0
+enum {
+	DWAXIDMAC_MBLK_TYPE_CONTIGUOUS	= 0,
+	DWAXIDMAC_MBLK_TYPE_RELOAD,
+	DWAXIDMAC_MBLK_TYPE_SHADOW_REG,
+	DWAXIDMAC_MBLK_TYPE_LL
+};
+
+/**
+ * DW AXI DMA channel interrupts
+ *
+ * @DWAXIDMAC_IRQ_NONE: Bitmask of no one interrupt
+ * @DWAXIDMAC_IRQ_BLOCK_TRF: Block transfer complete
+ * @DWAXIDMAC_IRQ_DMA_TRF: Dma transfer complete
+ * @DWAXIDMAC_IRQ_SRC_TRAN: Source transaction complete
+ * @DWAXIDMAC_IRQ_DST_TRAN: Destination transaction complete
+ * @DWAXIDMAC_IRQ_SRC_DEC_ERR: Source decode error
+ * @DWAXIDMAC_IRQ_DST_DEC_ERR: Destination decode error
+ * @DWAXIDMAC_IRQ_SRC_SLV_ERR: Source slave error
+ * @DWAXIDMAC_IRQ_DST_SLV_ERR: Destination slave error
+ * @DWAXIDMAC_IRQ_LLI_RD_DEC_ERR: LLI read decode error
+ * @DWAXIDMAC_IRQ_LLI_WR_DEC_ERR: LLI write decode error
+ * @DWAXIDMAC_IRQ_LLI_RD_SLV_ERR: LLI read slave error
+ * @DWAXIDMAC_IRQ_LLI_WR_SLV_ERR: LLI write slave error
+ * @DWAXIDMAC_IRQ_INVALID_ERR: LLI invalid error or Shadow register error
+ * @DWAXIDMAC_IRQ_MULTIBLKTYPE_ERR: Slave Interface Multiblock type error
+ * @DWAXIDMAC_IRQ_DEC_ERR: Slave Interface decode error
+ * @DWAXIDMAC_IRQ_WR2RO_ERR: Slave Interface write to read only error
+ * @DWAXIDMAC_IRQ_RD2RWO_ERR: Slave Interface read to write only error
+ * @DWAXIDMAC_IRQ_WRONCHEN_ERR: Slave Interface write to channel error
+ * @DWAXIDMAC_IRQ_SHADOWREG_ERR: Slave Interface shadow reg error
+ * @DWAXIDMAC_IRQ_WRONHOLD_ERR: Slave Interface hold error
+ * @DWAXIDMAC_IRQ_LOCK_CLEARED: Lock Cleared Status
+ * @DWAXIDMAC_IRQ_SRC_SUSPENDED: Source Suspended Status
+ * @DWAXIDMAC_IRQ_SUSPENDED: Channel Suspended Status
+ * @DWAXIDMAC_IRQ_DISABLED: Channel Disabled Status
+ * @DWAXIDMAC_IRQ_ABORTED: Channel Aborted Status
+ * @DWAXIDMAC_IRQ_ALL_ERR: Bitmask of all error interrupts
+ * @DWAXIDMAC_IRQ_ALL: Bitmask of all interrupts
+ */
+enum {
+	DWAXIDMAC_IRQ_NONE		= 0,
+	DWAXIDMAC_IRQ_BLOCK_TRF		= BIT(0),
+	DWAXIDMAC_IRQ_DMA_TRF		= BIT(1),
+	DWAXIDMAC_IRQ_SRC_TRAN		= BIT(3),
+	DWAXIDMAC_IRQ_DST_TRAN		= BIT(4),
+	DWAXIDMAC_IRQ_SRC_DEC_ERR	= BIT(5),
+	DWAXIDMAC_IRQ_DST_DEC_ERR	= BIT(6),
+	DWAXIDMAC_IRQ_SRC_SLV_ERR	= BIT(7),
+	DWAXIDMAC_IRQ_DST_SLV_ERR	= BIT(8),
+	DWAXIDMAC_IRQ_LLI_RD_DEC_ERR	= BIT(9),
+	DWAXIDMAC_IRQ_LLI_WR_DEC_ERR	= BIT(10),
+	DWAXIDMAC_IRQ_LLI_RD_SLV_ERR	= BIT(11),
+	DWAXIDMAC_IRQ_LLI_WR_SLV_ERR	= BIT(12),
+	DWAXIDMAC_IRQ_INVALID_ERR	= BIT(13),
+	DWAXIDMAC_IRQ_MULTIBLKTYPE_ERR	= BIT(14),
+	DWAXIDMAC_IRQ_DEC_ERR		= BIT(16),
+	DWAXIDMAC_IRQ_WR2RO_ERR		= BIT(17),
+	DWAXIDMAC_IRQ_RD2RWO_ERR	= BIT(18),
+	DWAXIDMAC_IRQ_WRONCHEN_ERR	= BIT(19),
+	DWAXIDMAC_IRQ_SHADOWREG_ERR	= BIT(20),
+	DWAXIDMAC_IRQ_WRONHOLD_ERR	= BIT(21),
+	DWAXIDMAC_IRQ_LOCK_CLEARED	= BIT(27),
+	DWAXIDMAC_IRQ_SRC_SUSPENDED	= BIT(28),
+	DWAXIDMAC_IRQ_SUSPENDED		= BIT(29),
+	DWAXIDMAC_IRQ_DISABLED		= BIT(30),
+	DWAXIDMAC_IRQ_ABORTED		= BIT(31),
+	DWAXIDMAC_IRQ_ALL_ERR		= (GENMASK(21, 16) | GENMASK(14, 5)),
+	DWAXIDMAC_IRQ_ALL		= GENMASK(31, 0)
+};
+
+enum {
+	DWAXIDMAC_TRANS_WIDTH_8		= 0,
+	DWAXIDMAC_TRANS_WIDTH_16,
+	DWAXIDMAC_TRANS_WIDTH_32,
+	DWAXIDMAC_TRANS_WIDTH_64,
+	DWAXIDMAC_TRANS_WIDTH_128,
+	DWAXIDMAC_TRANS_WIDTH_256,
+	DWAXIDMAC_TRANS_WIDTH_512,
+	DWAXIDMAC_TRANS_WIDTH_MAX	= DWAXIDMAC_TRANS_WIDTH_512
+};
+
+#endif /* _AXI_DMA_PLATFORM_H */
diff --git a/drivers/dma/edma.c b/drivers/dma/edma.c
index 948df1ab5f1a..85ea92fcea54 100644
--- a/drivers/dma/edma.c
+++ b/drivers/dma/edma.c
@@ -1876,6 +1876,11 @@ static void edma_dma_init(struct edma_cc *ecc, bool legacy_mode)
 
 	if (memcpy_channels) {
 		m_ddev = devm_kzalloc(ecc->dev, sizeof(*m_ddev), GFP_KERNEL);
+		if (!m_ddev) {
+			dev_warn(ecc->dev, "memcpy is disabled due to OoM\n");
+			memcpy_channels = NULL;
+			goto ch_setup;
+		}
 		ecc->dma_memcpy = m_ddev;
 
 		dma_cap_zero(m_ddev->cap_mask);
@@ -1903,6 +1908,7 @@ static void edma_dma_init(struct edma_cc *ecc, bool legacy_mode)
 		dev_info(ecc->dev, "memcpy is disabled\n");
 	}
 
+ch_setup:
 	for (i = 0; i < ecc->num_channels; i++) {
 		struct edma_chan *echan = &ecc->slave_chans[i];
 		echan->ch_num = EDMA_CTLR_CHAN(ecc->id, i);
diff --git a/drivers/dma/imx-sdma.c b/drivers/dma/imx-sdma.c
index e7db24c67030..ccd03c3cedfe 100644
--- a/drivers/dma/imx-sdma.c
+++ b/drivers/dma/imx-sdma.c
@@ -338,6 +338,7 @@ struct sdma_channel {
 	unsigned int			chn_real_count;
 	struct tasklet_struct		tasklet;
 	struct imx_dma_data		data;
+	bool				enabled;
 };
 
 #define IMX_DMA_SG_LOOP		BIT(0)
@@ -596,7 +597,14 @@ static int sdma_config_ownership(struct sdma_channel *sdmac,
 
 static void sdma_enable_channel(struct sdma_engine *sdma, int channel)
 {
+	unsigned long flags;
+	struct sdma_channel *sdmac = &sdma->channel[channel];
+
 	writel(BIT(channel), sdma->regs + SDMA_H_START);
+
+	spin_lock_irqsave(&sdmac->lock, flags);
+	sdmac->enabled = true;
+	spin_unlock_irqrestore(&sdmac->lock, flags);
 }
 
 /*
@@ -685,6 +693,14 @@ static void sdma_update_channel_loop(struct sdma_channel *sdmac)
 	struct sdma_buffer_descriptor *bd;
 	int error = 0;
 	enum dma_status	old_status = sdmac->status;
+	unsigned long flags;
+
+	spin_lock_irqsave(&sdmac->lock, flags);
+	if (!sdmac->enabled) {
+		spin_unlock_irqrestore(&sdmac->lock, flags);
+		return;
+	}
+	spin_unlock_irqrestore(&sdmac->lock, flags);
 
 	/*
 	 * loop mode. Iterate over descriptors, re-setup them and
@@ -938,10 +954,15 @@ static int sdma_disable_channel(struct dma_chan *chan)
 	struct sdma_channel *sdmac = to_sdma_chan(chan);
 	struct sdma_engine *sdma = sdmac->sdma;
 	int channel = sdmac->channel;
+	unsigned long flags;
 
 	writel_relaxed(BIT(channel), sdma->regs + SDMA_H_STATSTOP);
 	sdmac->status = DMA_ERROR;
 
+	spin_lock_irqsave(&sdmac->lock, flags);
+	sdmac->enabled = false;
+	spin_unlock_irqrestore(&sdmac->lock, flags);
+
 	return 0;
 }
 
diff --git a/drivers/dma/mediatek/Kconfig b/drivers/dma/mediatek/Kconfig
new file mode 100644
index 000000000000..27bac0bba09e
--- /dev/null
+++ b/drivers/dma/mediatek/Kconfig
@@ -0,0 +1,13 @@
+
+config MTK_HSDMA
+	tristate "MediaTek High-Speed DMA controller support"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	select DMA_ENGINE
+	select DMA_VIRTUAL_CHANNELS
+	---help---
+	  Enable support for High-Speed DMA controller on MediaTek
+	  SoCs.
+
+	  This controller provides the channels which is dedicated to
+	  memory-to-memory transfer to offload from CPU through ring-
+	  based descriptor management.
diff --git a/drivers/dma/mediatek/Makefile b/drivers/dma/mediatek/Makefile
new file mode 100644
index 000000000000..6e778f842f01
--- /dev/null
+++ b/drivers/dma/mediatek/Makefile
@@ -0,0 +1 @@
+obj-$(CONFIG_MTK_HSDMA) += mtk-hsdma.o
diff --git a/drivers/dma/mediatek/mtk-hsdma.c b/drivers/dma/mediatek/mtk-hsdma.c
new file mode 100644
index 000000000000..b7ec56ae02a6
--- /dev/null
+++ b/drivers/dma/mediatek/mtk-hsdma.c
@@ -0,0 +1,1056 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2017-2018 MediaTek Inc.
+
+/*
+ * Driver for MediaTek High-Speed DMA Controller
+ *
+ * Author: Sean Wang <sean.wang@mediatek.com>
+ *
+ */
+
+#include <linux/bitops.h>
+#include <linux/clk.h>
+#include <linux/dmaengine.h>
+#include <linux/dma-mapping.h>
+#include <linux/err.h>
+#include <linux/iopoll.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/of_dma.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+#include <linux/refcount.h>
+#include <linux/slab.h>
+
+#include "../virt-dma.h"
+
+#define MTK_HSDMA_USEC_POLL		20
+#define MTK_HSDMA_TIMEOUT_POLL		200000
+#define MTK_HSDMA_DMA_BUSWIDTHS		BIT(DMA_SLAVE_BUSWIDTH_4_BYTES)
+
+/* The default number of virtual channel */
+#define MTK_HSDMA_NR_VCHANS		3
+
+/* Only one physical channel supported */
+#define MTK_HSDMA_NR_MAX_PCHANS		1
+
+/* Macro for physical descriptor (PD) manipulation */
+/* The number of PD which must be 2 of power */
+#define MTK_DMA_SIZE			64
+#define MTK_HSDMA_NEXT_DESP_IDX(x, y)	(((x) + 1) & ((y) - 1))
+#define MTK_HSDMA_LAST_DESP_IDX(x, y)	(((x) - 1) & ((y) - 1))
+#define MTK_HSDMA_MAX_LEN		0x3f80
+#define MTK_HSDMA_ALIGN_SIZE		4
+#define MTK_HSDMA_PLEN_MASK		0x3fff
+#define MTK_HSDMA_DESC_PLEN(x)		(((x) & MTK_HSDMA_PLEN_MASK) << 16)
+#define MTK_HSDMA_DESC_PLEN_GET(x)	(((x) >> 16) & MTK_HSDMA_PLEN_MASK)
+
+/* Registers for underlying ring manipulation */
+#define MTK_HSDMA_TX_BASE		0x0
+#define MTK_HSDMA_TX_CNT		0x4
+#define MTK_HSDMA_TX_CPU		0x8
+#define MTK_HSDMA_TX_DMA		0xc
+#define MTK_HSDMA_RX_BASE		0x100
+#define MTK_HSDMA_RX_CNT		0x104
+#define MTK_HSDMA_RX_CPU		0x108
+#define MTK_HSDMA_RX_DMA		0x10c
+
+/* Registers for global setup */
+#define MTK_HSDMA_GLO			0x204
+#define MTK_HSDMA_GLO_MULTI_DMA		BIT(10)
+#define MTK_HSDMA_TX_WB_DDONE		BIT(6)
+#define MTK_HSDMA_BURST_64BYTES		(0x2 << 4)
+#define MTK_HSDMA_GLO_RX_BUSY		BIT(3)
+#define MTK_HSDMA_GLO_RX_DMA		BIT(2)
+#define MTK_HSDMA_GLO_TX_BUSY		BIT(1)
+#define MTK_HSDMA_GLO_TX_DMA		BIT(0)
+#define MTK_HSDMA_GLO_DMA		(MTK_HSDMA_GLO_TX_DMA |	\
+					 MTK_HSDMA_GLO_RX_DMA)
+#define MTK_HSDMA_GLO_BUSY		(MTK_HSDMA_GLO_RX_BUSY | \
+					 MTK_HSDMA_GLO_TX_BUSY)
+#define MTK_HSDMA_GLO_DEFAULT		(MTK_HSDMA_GLO_TX_DMA | \
+					 MTK_HSDMA_GLO_RX_DMA | \
+					 MTK_HSDMA_TX_WB_DDONE | \
+					 MTK_HSDMA_BURST_64BYTES | \
+					 MTK_HSDMA_GLO_MULTI_DMA)
+
+/* Registers for reset */
+#define MTK_HSDMA_RESET			0x208
+#define MTK_HSDMA_RST_TX		BIT(0)
+#define MTK_HSDMA_RST_RX		BIT(16)
+
+/* Registers for interrupt control */
+#define MTK_HSDMA_DLYINT		0x20c
+#define MTK_HSDMA_RXDLY_INT_EN		BIT(15)
+
+/* Interrupt fires when the pending number's more than the specified */
+#define MTK_HSDMA_RXMAX_PINT(x)		(((x) & 0x7f) << 8)
+
+/* Interrupt fires when the pending time's more than the specified in 20 us */
+#define MTK_HSDMA_RXMAX_PTIME(x)	((x) & 0x7f)
+#define MTK_HSDMA_DLYINT_DEFAULT	(MTK_HSDMA_RXDLY_INT_EN | \
+					 MTK_HSDMA_RXMAX_PINT(20) | \
+					 MTK_HSDMA_RXMAX_PTIME(20))
+#define MTK_HSDMA_INT_STATUS		0x220
+#define MTK_HSDMA_INT_ENABLE		0x228
+#define MTK_HSDMA_INT_RXDONE		BIT(16)
+
+enum mtk_hsdma_vdesc_flag {
+	MTK_HSDMA_VDESC_FINISHED	= 0x01,
+};
+
+#define IS_MTK_HSDMA_VDESC_FINISHED(x) ((x) == MTK_HSDMA_VDESC_FINISHED)
+
+/**
+ * struct mtk_hsdma_pdesc - This is the struct holding info describing physical
+ *			    descriptor (PD) and its placement must be kept at
+ *			    4-bytes alignment in little endian order.
+ * @desc[1-4]:		    The control pad used to indicate hardware how to
+ *			    deal with the descriptor such as source and
+ *			    destination address and data length. The maximum
+ *			    data length each pdesc can handle is 0x3f80 bytes
+ */
+struct mtk_hsdma_pdesc {
+	__le32 desc1;
+	__le32 desc2;
+	__le32 desc3;
+	__le32 desc4;
+} __packed __aligned(4);
+
+/**
+ * struct mtk_hsdma_vdesc - This is the struct holding info describing virtual
+ *			    descriptor (VD)
+ * @vd:			    An instance for struct virt_dma_desc
+ * @len:		    The total data size device wants to move
+ * @residue:		    The remaining data size device will move
+ * @dest:		    The destination address device wants to move to
+ * @src:		    The source address device wants to move from
+ */
+struct mtk_hsdma_vdesc {
+	struct virt_dma_desc vd;
+	size_t len;
+	size_t residue;
+	dma_addr_t dest;
+	dma_addr_t src;
+};
+
+/**
+ * struct mtk_hsdma_cb - This is the struct holding extra info required for RX
+ *			 ring to know what relevant VD the the PD is being
+ *			 mapped to.
+ * @vd:			 Pointer to the relevant VD.
+ * @flag:		 Flag indicating what action should be taken when VD
+ *			 is completed.
+ */
+struct mtk_hsdma_cb {
+	struct virt_dma_desc *vd;
+	enum mtk_hsdma_vdesc_flag flag;
+};
+
+/**
+ * struct mtk_hsdma_ring - This struct holds info describing underlying ring
+ *			   space
+ * @txd:		   The descriptor TX ring which describes DMA source
+ *			   information
+ * @rxd:		   The descriptor RX ring which describes DMA
+ *			   destination information
+ * @cb:			   The extra information pointed at by RX ring
+ * @tphys:		   The physical addr of TX ring
+ * @rphys:		   The physical addr of RX ring
+ * @cur_tptr:		   Pointer to the next free descriptor used by the host
+ * @cur_rptr:		   Pointer to the last done descriptor by the device
+ */
+struct mtk_hsdma_ring {
+	struct mtk_hsdma_pdesc *txd;
+	struct mtk_hsdma_pdesc *rxd;
+	struct mtk_hsdma_cb *cb;
+	dma_addr_t tphys;
+	dma_addr_t rphys;
+	u16 cur_tptr;
+	u16 cur_rptr;
+};
+
+/**
+ * struct mtk_hsdma_pchan - This is the struct holding info describing physical
+ *			   channel (PC)
+ * @ring:		   An instance for the underlying ring
+ * @sz_ring:		   Total size allocated for the ring
+ * @nr_free:		   Total number of free rooms in the ring. It would
+ *			   be accessed and updated frequently between IRQ
+ *			   context and user context to reflect whether ring
+ *			   can accept requests from VD.
+ */
+struct mtk_hsdma_pchan {
+	struct mtk_hsdma_ring ring;
+	size_t sz_ring;
+	atomic_t nr_free;
+};
+
+/**
+ * struct mtk_hsdma_vchan - This is the struct holding info describing virtual
+ *			   channel (VC)
+ * @vc:			   An instance for struct virt_dma_chan
+ * @issue_completion:	   The wait for all issued descriptors completited
+ * @issue_synchronize:	   Bool indicating channel synchronization starts
+ * @desc_hw_processing:	   List those descriptors the hardware is processing,
+ *			   which is protected by vc.lock
+ */
+struct mtk_hsdma_vchan {
+	struct virt_dma_chan vc;
+	struct completion issue_completion;
+	bool issue_synchronize;
+	struct list_head desc_hw_processing;
+};
+
+/**
+ * struct mtk_hsdma_soc - This is the struct holding differences among SoCs
+ * @ddone:		  Bit mask for DDONE
+ * @ls0:		  Bit mask for LS0
+ */
+struct mtk_hsdma_soc {
+	__le32 ddone;
+	__le32 ls0;
+};
+
+/**
+ * struct mtk_hsdma_device - This is the struct holding info describing HSDMA
+ *			     device
+ * @ddev:		     An instance for struct dma_device
+ * @base:		     The mapped register I/O base
+ * @clk:		     The clock that device internal is using
+ * @irq:		     The IRQ that device are using
+ * @dma_requests:	     The number of VCs the device supports to
+ * @vc:			     The pointer to all available VCs
+ * @pc:			     The pointer to the underlying PC
+ * @pc_refcnt:		     Track how many VCs are using the PC
+ * @lock:		     Lock protect agaisting multiple VCs access PC
+ * @soc:		     The pointer to area holding differences among
+ *			     vaious platform
+ */
+struct mtk_hsdma_device {
+	struct dma_device ddev;
+	void __iomem *base;
+	struct clk *clk;
+	u32 irq;
+
+	u32 dma_requests;
+	struct mtk_hsdma_vchan *vc;
+	struct mtk_hsdma_pchan *pc;
+	refcount_t pc_refcnt;
+
+	/* Lock used to protect against multiple VCs access PC */
+	spinlock_t lock;
+
+	const struct mtk_hsdma_soc *soc;
+};
+
+static struct mtk_hsdma_device *to_hsdma_dev(struct dma_chan *chan)
+{
+	return container_of(chan->device, struct mtk_hsdma_device, ddev);
+}
+
+static inline struct mtk_hsdma_vchan *to_hsdma_vchan(struct dma_chan *chan)
+{
+	return container_of(chan, struct mtk_hsdma_vchan, vc.chan);
+}
+
+static struct mtk_hsdma_vdesc *to_hsdma_vdesc(struct virt_dma_desc *vd)
+{
+	return container_of(vd, struct mtk_hsdma_vdesc, vd);
+}
+
+static struct device *hsdma2dev(struct mtk_hsdma_device *hsdma)
+{
+	return hsdma->ddev.dev;
+}
+
+static u32 mtk_dma_read(struct mtk_hsdma_device *hsdma, u32 reg)
+{
+	return readl(hsdma->base + reg);
+}
+
+static void mtk_dma_write(struct mtk_hsdma_device *hsdma, u32 reg, u32 val)
+{
+	writel(val, hsdma->base + reg);
+}
+
+static void mtk_dma_rmw(struct mtk_hsdma_device *hsdma, u32 reg,
+			u32 mask, u32 set)
+{
+	u32 val;
+
+	val = mtk_dma_read(hsdma, reg);
+	val &= ~mask;
+	val |= set;
+	mtk_dma_write(hsdma, reg, val);
+}
+
+static void mtk_dma_set(struct mtk_hsdma_device *hsdma, u32 reg, u32 val)
+{
+	mtk_dma_rmw(hsdma, reg, 0, val);
+}
+
+static void mtk_dma_clr(struct mtk_hsdma_device *hsdma, u32 reg, u32 val)
+{
+	mtk_dma_rmw(hsdma, reg, val, 0);
+}
+
+static void mtk_hsdma_vdesc_free(struct virt_dma_desc *vd)
+{
+	kfree(container_of(vd, struct mtk_hsdma_vdesc, vd));
+}
+
+static int mtk_hsdma_busy_wait(struct mtk_hsdma_device *hsdma)
+{
+	u32 status = 0;
+
+	return readl_poll_timeout(hsdma->base + MTK_HSDMA_GLO, status,
+				  !(status & MTK_HSDMA_GLO_BUSY),
+				  MTK_HSDMA_USEC_POLL,
+				  MTK_HSDMA_TIMEOUT_POLL);
+}
+
+static int mtk_hsdma_alloc_pchan(struct mtk_hsdma_device *hsdma,
+				 struct mtk_hsdma_pchan *pc)
+{
+	struct mtk_hsdma_ring *ring = &pc->ring;
+	int err;
+
+	memset(pc, 0, sizeof(*pc));
+
+	/*
+	 * Allocate ring space where [0 ... MTK_DMA_SIZE - 1] is for TX ring
+	 * and [MTK_DMA_SIZE ... 2 * MTK_DMA_SIZE - 1] is for RX ring.
+	 */
+	pc->sz_ring = 2 * MTK_DMA_SIZE * sizeof(*ring->txd);
+	ring->txd = dma_zalloc_coherent(hsdma2dev(hsdma), pc->sz_ring,
+					&ring->tphys, GFP_NOWAIT);
+	if (!ring->txd)
+		return -ENOMEM;
+
+	ring->rxd = &ring->txd[MTK_DMA_SIZE];
+	ring->rphys = ring->tphys + MTK_DMA_SIZE * sizeof(*ring->txd);
+	ring->cur_tptr = 0;
+	ring->cur_rptr = MTK_DMA_SIZE - 1;
+
+	ring->cb = kcalloc(MTK_DMA_SIZE, sizeof(*ring->cb), GFP_NOWAIT);
+	if (!ring->cb) {
+		err = -ENOMEM;
+		goto err_free_dma;
+	}
+
+	atomic_set(&pc->nr_free, MTK_DMA_SIZE - 1);
+
+	/* Disable HSDMA and wait for the completion */
+	mtk_dma_clr(hsdma, MTK_HSDMA_GLO, MTK_HSDMA_GLO_DMA);
+	err = mtk_hsdma_busy_wait(hsdma);
+	if (err)
+		goto err_free_cb;
+
+	/* Reset */
+	mtk_dma_set(hsdma, MTK_HSDMA_RESET,
+		    MTK_HSDMA_RST_TX | MTK_HSDMA_RST_RX);
+	mtk_dma_clr(hsdma, MTK_HSDMA_RESET,
+		    MTK_HSDMA_RST_TX | MTK_HSDMA_RST_RX);
+
+	/* Setup HSDMA initial pointer in the ring */
+	mtk_dma_write(hsdma, MTK_HSDMA_TX_BASE, ring->tphys);
+	mtk_dma_write(hsdma, MTK_HSDMA_TX_CNT, MTK_DMA_SIZE);
+	mtk_dma_write(hsdma, MTK_HSDMA_TX_CPU, ring->cur_tptr);
+	mtk_dma_write(hsdma, MTK_HSDMA_TX_DMA, 0);
+	mtk_dma_write(hsdma, MTK_HSDMA_RX_BASE, ring->rphys);
+	mtk_dma_write(hsdma, MTK_HSDMA_RX_CNT, MTK_DMA_SIZE);
+	mtk_dma_write(hsdma, MTK_HSDMA_RX_CPU, ring->cur_rptr);
+	mtk_dma_write(hsdma, MTK_HSDMA_RX_DMA, 0);
+
+	/* Enable HSDMA */
+	mtk_dma_set(hsdma, MTK_HSDMA_GLO, MTK_HSDMA_GLO_DMA);
+
+	/* Setup delayed interrupt */
+	mtk_dma_write(hsdma, MTK_HSDMA_DLYINT, MTK_HSDMA_DLYINT_DEFAULT);
+
+	/* Enable interrupt */
+	mtk_dma_set(hsdma, MTK_HSDMA_INT_ENABLE, MTK_HSDMA_INT_RXDONE);
+
+	return 0;
+
+err_free_cb:
+	kfree(ring->cb);
+
+err_free_dma:
+	dma_free_coherent(hsdma2dev(hsdma),
+			  pc->sz_ring, ring->txd, ring->tphys);
+	return err;
+}
+
+static void mtk_hsdma_free_pchan(struct mtk_hsdma_device *hsdma,
+				 struct mtk_hsdma_pchan *pc)
+{
+	struct mtk_hsdma_ring *ring = &pc->ring;
+
+	/* Disable HSDMA and then wait for the completion */
+	mtk_dma_clr(hsdma, MTK_HSDMA_GLO, MTK_HSDMA_GLO_DMA);
+	mtk_hsdma_busy_wait(hsdma);
+
+	/* Reset pointer in the ring */
+	mtk_dma_clr(hsdma, MTK_HSDMA_INT_ENABLE, MTK_HSDMA_INT_RXDONE);
+	mtk_dma_write(hsdma, MTK_HSDMA_TX_BASE, 0);
+	mtk_dma_write(hsdma, MTK_HSDMA_TX_CNT, 0);
+	mtk_dma_write(hsdma, MTK_HSDMA_TX_CPU, 0);
+	mtk_dma_write(hsdma, MTK_HSDMA_RX_BASE, 0);
+	mtk_dma_write(hsdma, MTK_HSDMA_RX_CNT, 0);
+	mtk_dma_write(hsdma, MTK_HSDMA_RX_CPU, MTK_DMA_SIZE - 1);
+
+	kfree(ring->cb);
+
+	dma_free_coherent(hsdma2dev(hsdma),
+			  pc->sz_ring, ring->txd, ring->tphys);
+}
+
+static int mtk_hsdma_issue_pending_vdesc(struct mtk_hsdma_device *hsdma,
+					 struct mtk_hsdma_pchan *pc,
+					 struct mtk_hsdma_vdesc *hvd)
+{
+	struct mtk_hsdma_ring *ring = &pc->ring;
+	struct mtk_hsdma_pdesc *txd, *rxd;
+	u16 reserved, prev, tlen, num_sgs;
+	unsigned long flags;
+
+	/* Protect against PC is accessed by multiple VCs simultaneously */
+	spin_lock_irqsave(&hsdma->lock, flags);
+
+	/*
+	 * Reserve rooms, where pc->nr_free is used to track how many free
+	 * rooms in the ring being updated in user and IRQ context.
+	 */
+	num_sgs = DIV_ROUND_UP(hvd->len, MTK_HSDMA_MAX_LEN);
+	reserved = min_t(u16, num_sgs, atomic_read(&pc->nr_free));
+
+	if (!reserved) {
+		spin_unlock_irqrestore(&hsdma->lock, flags);
+		return -ENOSPC;
+	}
+
+	atomic_sub(reserved, &pc->nr_free);
+
+	while (reserved--) {
+		/* Limit size by PD capability for valid data moving */
+		tlen = (hvd->len > MTK_HSDMA_MAX_LEN) ?
+		       MTK_HSDMA_MAX_LEN : hvd->len;
+
+		/*
+		 * Setup PDs using the remaining VD info mapped on those
+		 * reserved rooms. And since RXD is shared memory between the
+		 * host and the device allocated by dma_alloc_coherent call,
+		 * the helper macro WRITE_ONCE can ensure the data written to
+		 * RAM would really happens.
+		 */
+		txd = &ring->txd[ring->cur_tptr];
+		WRITE_ONCE(txd->desc1, hvd->src);
+		WRITE_ONCE(txd->desc2,
+			   hsdma->soc->ls0 | MTK_HSDMA_DESC_PLEN(tlen));
+
+		rxd = &ring->rxd[ring->cur_tptr];
+		WRITE_ONCE(rxd->desc1, hvd->dest);
+		WRITE_ONCE(rxd->desc2, MTK_HSDMA_DESC_PLEN(tlen));
+
+		/* Associate VD, the PD belonged to */
+		ring->cb[ring->cur_tptr].vd = &hvd->vd;
+
+		/* Move forward the pointer of TX ring */
+		ring->cur_tptr = MTK_HSDMA_NEXT_DESP_IDX(ring->cur_tptr,
+							 MTK_DMA_SIZE);
+
+		/* Update VD with remaining data */
+		hvd->src  += tlen;
+		hvd->dest += tlen;
+		hvd->len  -= tlen;
+	}
+
+	/*
+	 * Tagging flag for the last PD for VD will be responsible for
+	 * completing VD.
+	 */
+	if (!hvd->len) {
+		prev = MTK_HSDMA_LAST_DESP_IDX(ring->cur_tptr, MTK_DMA_SIZE);
+		ring->cb[prev].flag = MTK_HSDMA_VDESC_FINISHED;
+	}
+
+	/* Ensure all changes indeed done before we're going on */
+	wmb();
+
+	/*
+	 * Updating into hardware the pointer of TX ring lets HSDMA to take
+	 * action for those pending PDs.
+	 */
+	mtk_dma_write(hsdma, MTK_HSDMA_TX_CPU, ring->cur_tptr);
+
+	spin_unlock_irqrestore(&hsdma->lock, flags);
+
+	return 0;
+}
+
+static void mtk_hsdma_issue_vchan_pending(struct mtk_hsdma_device *hsdma,
+					  struct mtk_hsdma_vchan *hvc)
+{
+	struct virt_dma_desc *vd, *vd2;
+	int err;
+
+	lockdep_assert_held(&hvc->vc.lock);
+
+	list_for_each_entry_safe(vd, vd2, &hvc->vc.desc_issued, node) {
+		struct mtk_hsdma_vdesc *hvd;
+
+		hvd = to_hsdma_vdesc(vd);
+
+		/* Map VD into PC and all VCs shares a single PC */
+		err = mtk_hsdma_issue_pending_vdesc(hsdma, hsdma->pc, hvd);
+
+		/*
+		 * Move VD from desc_issued to desc_hw_processing when entire
+		 * VD is fit into available PDs. Otherwise, the uncompleted
+		 * VDs would stay in list desc_issued and then restart the
+		 * processing as soon as possible once underlying ring space
+		 * got freed.
+		 */
+		if (err == -ENOSPC || hvd->len > 0)
+			break;
+
+		/*
+		 * The extra list desc_hw_processing is used because
+		 * hardware can't provide sufficient information allowing us
+		 * to know what VDs are still working on the underlying ring.
+		 * Through the additional list, it can help us to implement
+		 * terminate_all, residue calculation and such thing needed
+		 * to know detail descriptor status on the hardware.
+		 */
+		list_move_tail(&vd->node, &hvc->desc_hw_processing);
+	}
+}
+
+static void mtk_hsdma_free_rooms_in_ring(struct mtk_hsdma_device *hsdma)
+{
+	struct mtk_hsdma_vchan *hvc;
+	struct mtk_hsdma_pdesc *rxd;
+	struct mtk_hsdma_vdesc *hvd;
+	struct mtk_hsdma_pchan *pc;
+	struct mtk_hsdma_cb *cb;
+	int i = MTK_DMA_SIZE;
+	__le32 desc2;
+	u32 status;
+	u16 next;
+
+	/* Read IRQ status */
+	status = mtk_dma_read(hsdma, MTK_HSDMA_INT_STATUS);
+	if (unlikely(!(status & MTK_HSDMA_INT_RXDONE)))
+		goto rx_done;
+
+	pc = hsdma->pc;
+
+	/*
+	 * Using a fail-safe loop with iterations of up to MTK_DMA_SIZE to
+	 * reclaim these finished descriptors: The most number of PDs the ISR
+	 * can handle at one time shouldn't be more than MTK_DMA_SIZE so we
+	 * take it as limited count instead of just using a dangerous infinite
+	 * poll.
+	 */
+	while (i--) {
+		next = MTK_HSDMA_NEXT_DESP_IDX(pc->ring.cur_rptr,
+					       MTK_DMA_SIZE);
+		rxd = &pc->ring.rxd[next];
+
+		/*
+		 * If MTK_HSDMA_DESC_DDONE is no specified, that means data
+		 * moving for the PD is still under going.
+		 */
+		desc2 = READ_ONCE(rxd->desc2);
+		if (!(desc2 & hsdma->soc->ddone))
+			break;
+
+		cb = &pc->ring.cb[next];
+		if (unlikely(!cb->vd)) {
+			dev_err(hsdma2dev(hsdma), "cb->vd cannot be null\n");
+			break;
+		}
+
+		/* Update residue of VD the associated PD belonged to */
+		hvd = to_hsdma_vdesc(cb->vd);
+		hvd->residue -= MTK_HSDMA_DESC_PLEN_GET(rxd->desc2);
+
+		/* Complete VD until the relevant last PD is finished */
+		if (IS_MTK_HSDMA_VDESC_FINISHED(cb->flag)) {
+			hvc = to_hsdma_vchan(cb->vd->tx.chan);
+
+			spin_lock(&hvc->vc.lock);
+
+			/* Remove VD from list desc_hw_processing */
+			list_del(&cb->vd->node);
+
+			/* Add VD into list desc_completed */
+			vchan_cookie_complete(cb->vd);
+
+			if (hvc->issue_synchronize &&
+			    list_empty(&hvc->desc_hw_processing)) {
+				complete(&hvc->issue_completion);
+				hvc->issue_synchronize = false;
+			}
+			spin_unlock(&hvc->vc.lock);
+
+			cb->flag = 0;
+		}
+
+		cb->vd = 0;
+
+		/*
+		 * Recycle the RXD with the helper WRITE_ONCE that can ensure
+		 * data written into RAM would really happens.
+		 */
+		WRITE_ONCE(rxd->desc1, 0);
+		WRITE_ONCE(rxd->desc2, 0);
+		pc->ring.cur_rptr = next;
+
+		/* Release rooms */
+		atomic_inc(&pc->nr_free);
+	}
+
+	/* Ensure all changes indeed done before we're going on */
+	wmb();
+
+	/* Update CPU pointer for those completed PDs */
+	mtk_dma_write(hsdma, MTK_HSDMA_RX_CPU, pc->ring.cur_rptr);
+
+	/*
+	 * Acking the pending IRQ allows hardware no longer to keep the used
+	 * IRQ line in certain trigger state when software has completed all
+	 * the finished physical descriptors.
+	 */
+	if (atomic_read(&pc->nr_free) >= MTK_DMA_SIZE - 1)
+		mtk_dma_write(hsdma, MTK_HSDMA_INT_STATUS, status);
+
+	/* ASAP handles pending VDs in all VCs after freeing some rooms */
+	for (i = 0; i < hsdma->dma_requests; i++) {
+		hvc = &hsdma->vc[i];
+		spin_lock(&hvc->vc.lock);
+		mtk_hsdma_issue_vchan_pending(hsdma, hvc);
+		spin_unlock(&hvc->vc.lock);
+	}
+
+rx_done:
+	/* All completed PDs are cleaned up, so enable interrupt again */
+	mtk_dma_set(hsdma, MTK_HSDMA_INT_ENABLE, MTK_HSDMA_INT_RXDONE);
+}
+
+static irqreturn_t mtk_hsdma_irq(int irq, void *devid)
+{
+	struct mtk_hsdma_device *hsdma = devid;
+
+	/*
+	 * Disable interrupt until all completed PDs are cleaned up in
+	 * mtk_hsdma_free_rooms call.
+	 */
+	mtk_dma_clr(hsdma, MTK_HSDMA_INT_ENABLE, MTK_HSDMA_INT_RXDONE);
+
+	mtk_hsdma_free_rooms_in_ring(hsdma);
+
+	return IRQ_HANDLED;
+}
+
+static struct virt_dma_desc *mtk_hsdma_find_active_desc(struct dma_chan *c,
+							dma_cookie_t cookie)
+{
+	struct mtk_hsdma_vchan *hvc = to_hsdma_vchan(c);
+	struct virt_dma_desc *vd;
+
+	list_for_each_entry(vd, &hvc->desc_hw_processing, node)
+		if (vd->tx.cookie == cookie)
+			return vd;
+
+	list_for_each_entry(vd, &hvc->vc.desc_issued, node)
+		if (vd->tx.cookie == cookie)
+			return vd;
+
+	return NULL;
+}
+
+static enum dma_status mtk_hsdma_tx_status(struct dma_chan *c,
+					   dma_cookie_t cookie,
+					   struct dma_tx_state *txstate)
+{
+	struct mtk_hsdma_vchan *hvc = to_hsdma_vchan(c);
+	struct mtk_hsdma_vdesc *hvd;
+	struct virt_dma_desc *vd;
+	enum dma_status ret;
+	unsigned long flags;
+	size_t bytes = 0;
+
+	ret = dma_cookie_status(c, cookie, txstate);
+	if (ret == DMA_COMPLETE || !txstate)
+		return ret;
+
+	spin_lock_irqsave(&hvc->vc.lock, flags);
+	vd = mtk_hsdma_find_active_desc(c, cookie);
+	spin_unlock_irqrestore(&hvc->vc.lock, flags);
+
+	if (vd) {
+		hvd = to_hsdma_vdesc(vd);
+		bytes = hvd->residue;
+	}
+
+	dma_set_residue(txstate, bytes);
+
+	return ret;
+}
+
+static void mtk_hsdma_issue_pending(struct dma_chan *c)
+{
+	struct mtk_hsdma_device *hsdma = to_hsdma_dev(c);
+	struct mtk_hsdma_vchan *hvc = to_hsdma_vchan(c);
+	unsigned long flags;
+
+	spin_lock_irqsave(&hvc->vc.lock, flags);
+
+	if (vchan_issue_pending(&hvc->vc))
+		mtk_hsdma_issue_vchan_pending(hsdma, hvc);
+
+	spin_unlock_irqrestore(&hvc->vc.lock, flags);
+}
+
+static struct dma_async_tx_descriptor *
+mtk_hsdma_prep_dma_memcpy(struct dma_chan *c, dma_addr_t dest,
+			  dma_addr_t src, size_t len, unsigned long flags)
+{
+	struct mtk_hsdma_vdesc *hvd;
+
+	hvd = kzalloc(sizeof(*hvd), GFP_NOWAIT);
+	if (!hvd)
+		return NULL;
+
+	hvd->len = len;
+	hvd->residue = len;
+	hvd->src = src;
+	hvd->dest = dest;
+
+	return vchan_tx_prep(to_virt_chan(c), &hvd->vd, flags);
+}
+
+static int mtk_hsdma_free_inactive_desc(struct dma_chan *c)
+{
+	struct virt_dma_chan *vc = to_virt_chan(c);
+	unsigned long flags;
+	LIST_HEAD(head);
+
+	spin_lock_irqsave(&vc->lock, flags);
+	list_splice_tail_init(&vc->desc_allocated, &head);
+	list_splice_tail_init(&vc->desc_submitted, &head);
+	list_splice_tail_init(&vc->desc_issued, &head);
+	spin_unlock_irqrestore(&vc->lock, flags);
+
+	/* At the point, we don't expect users put descriptor into VC again */
+	vchan_dma_desc_free_list(vc, &head);
+
+	return 0;
+}
+
+static void mtk_hsdma_free_active_desc(struct dma_chan *c)
+{
+	struct mtk_hsdma_vchan *hvc = to_hsdma_vchan(c);
+	bool sync_needed = false;
+
+	/*
+	 * Once issue_synchronize is being set, which means once the hardware
+	 * consumes all descriptors for the channel in the ring, the
+	 * synchronization must be be notified immediately it is completed.
+	 */
+	spin_lock(&hvc->vc.lock);
+	if (!list_empty(&hvc->desc_hw_processing)) {
+		hvc->issue_synchronize = true;
+		sync_needed = true;
+	}
+	spin_unlock(&hvc->vc.lock);
+
+	if (sync_needed)
+		wait_for_completion(&hvc->issue_completion);
+	/*
+	 * At the point, we expect that all remaining descriptors in the ring
+	 * for the channel should be all processing done.
+	 */
+	WARN_ONCE(!list_empty(&hvc->desc_hw_processing),
+		  "Desc pending still in list desc_hw_processing\n");
+
+	/* Free all descriptors in list desc_completed */
+	vchan_synchronize(&hvc->vc);
+
+	WARN_ONCE(!list_empty(&hvc->vc.desc_completed),
+		  "Desc pending still in list desc_completed\n");
+}
+
+static int mtk_hsdma_terminate_all(struct dma_chan *c)
+{
+	/*
+	 * Free pending descriptors not processed yet by hardware that have
+	 * previously been submitted to the channel.
+	 */
+	mtk_hsdma_free_inactive_desc(c);
+
+	/*
+	 * However, the DMA engine doesn't provide any way to stop these
+	 * descriptors being processed currently by hardware. The only way is
+	 * to just waiting until these descriptors are all processed completely
+	 * through mtk_hsdma_free_active_desc call.
+	 */
+	mtk_hsdma_free_active_desc(c);
+
+	return 0;
+}
+
+static int mtk_hsdma_alloc_chan_resources(struct dma_chan *c)
+{
+	struct mtk_hsdma_device *hsdma = to_hsdma_dev(c);
+	int err;
+
+	/*
+	 * Since HSDMA has only one PC, the resource for PC is being allocated
+	 * when the first VC is being created and the other VCs would run on
+	 * the same PC.
+	 */
+	if (!refcount_read(&hsdma->pc_refcnt)) {
+		err = mtk_hsdma_alloc_pchan(hsdma, hsdma->pc);
+		if (err)
+			return err;
+		/*
+		 * refcount_inc would complain increment on 0; use-after-free.
+		 * Thus, we need to explicitly set it as 1 initially.
+		 */
+		refcount_set(&hsdma->pc_refcnt, 1);
+	} else {
+		refcount_inc(&hsdma->pc_refcnt);
+	}
+
+	return 0;
+}
+
+static void mtk_hsdma_free_chan_resources(struct dma_chan *c)
+{
+	struct mtk_hsdma_device *hsdma = to_hsdma_dev(c);
+
+	/* Free all descriptors in all lists on the VC */
+	mtk_hsdma_terminate_all(c);
+
+	/* The resource for PC is not freed until all the VCs are destroyed */
+	if (!refcount_dec_and_test(&hsdma->pc_refcnt))
+		return;
+
+	mtk_hsdma_free_pchan(hsdma, hsdma->pc);
+}
+
+static int mtk_hsdma_hw_init(struct mtk_hsdma_device *hsdma)
+{
+	int err;
+
+	pm_runtime_enable(hsdma2dev(hsdma));
+	pm_runtime_get_sync(hsdma2dev(hsdma));
+
+	err = clk_prepare_enable(hsdma->clk);
+	if (err)
+		return err;
+
+	mtk_dma_write(hsdma, MTK_HSDMA_INT_ENABLE, 0);
+	mtk_dma_write(hsdma, MTK_HSDMA_GLO, MTK_HSDMA_GLO_DEFAULT);
+
+	return 0;
+}
+
+static int mtk_hsdma_hw_deinit(struct mtk_hsdma_device *hsdma)
+{
+	mtk_dma_write(hsdma, MTK_HSDMA_GLO, 0);
+
+	clk_disable_unprepare(hsdma->clk);
+
+	pm_runtime_put_sync(hsdma2dev(hsdma));
+	pm_runtime_disable(hsdma2dev(hsdma));
+
+	return 0;
+}
+
+static const struct mtk_hsdma_soc mt7623_soc = {
+	.ddone = BIT(31),
+	.ls0 = BIT(30),
+};
+
+static const struct mtk_hsdma_soc mt7622_soc = {
+	.ddone = BIT(15),
+	.ls0 = BIT(14),
+};
+
+static const struct of_device_id mtk_hsdma_match[] = {
+	{ .compatible = "mediatek,mt7623-hsdma", .data = &mt7623_soc},
+	{ .compatible = "mediatek,mt7622-hsdma", .data = &mt7622_soc},
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, mtk_hsdma_match);
+
+static int mtk_hsdma_probe(struct platform_device *pdev)
+{
+	struct mtk_hsdma_device *hsdma;
+	struct mtk_hsdma_vchan *vc;
+	struct dma_device *dd;
+	struct resource *res;
+	int i, err;
+
+	hsdma = devm_kzalloc(&pdev->dev, sizeof(*hsdma), GFP_KERNEL);
+	if (!hsdma)
+		return -ENOMEM;
+
+	dd = &hsdma->ddev;
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	hsdma->base = devm_ioremap_resource(&pdev->dev, res);
+	if (IS_ERR(hsdma->base))
+		return PTR_ERR(hsdma->base);
+
+	hsdma->soc = of_device_get_match_data(&pdev->dev);
+	if (!hsdma->soc) {
+		dev_err(&pdev->dev, "No device match found\n");
+		return -ENODEV;
+	}
+
+	hsdma->clk = devm_clk_get(&pdev->dev, "hsdma");
+	if (IS_ERR(hsdma->clk)) {
+		dev_err(&pdev->dev, "No clock for %s\n",
+			dev_name(&pdev->dev));
+		return PTR_ERR(hsdma->clk);
+	}
+
+	res = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
+	if (!res) {
+		dev_err(&pdev->dev, "No irq resource for %s\n",
+			dev_name(&pdev->dev));
+		return -EINVAL;
+	}
+	hsdma->irq = res->start;
+
+	refcount_set(&hsdma->pc_refcnt, 0);
+	spin_lock_init(&hsdma->lock);
+
+	dma_cap_set(DMA_MEMCPY, dd->cap_mask);
+
+	dd->copy_align = MTK_HSDMA_ALIGN_SIZE;
+	dd->device_alloc_chan_resources = mtk_hsdma_alloc_chan_resources;
+	dd->device_free_chan_resources = mtk_hsdma_free_chan_resources;
+	dd->device_tx_status = mtk_hsdma_tx_status;
+	dd->device_issue_pending = mtk_hsdma_issue_pending;
+	dd->device_prep_dma_memcpy = mtk_hsdma_prep_dma_memcpy;
+	dd->device_terminate_all = mtk_hsdma_terminate_all;
+	dd->src_addr_widths = MTK_HSDMA_DMA_BUSWIDTHS;
+	dd->dst_addr_widths = MTK_HSDMA_DMA_BUSWIDTHS;
+	dd->directions = BIT(DMA_MEM_TO_MEM);
+	dd->residue_granularity = DMA_RESIDUE_GRANULARITY_SEGMENT;
+	dd->dev = &pdev->dev;
+	INIT_LIST_HEAD(&dd->channels);
+
+	hsdma->dma_requests = MTK_HSDMA_NR_VCHANS;
+	if (pdev->dev.of_node && of_property_read_u32(pdev->dev.of_node,
+						      "dma-requests",
+						      &hsdma->dma_requests)) {
+		dev_info(&pdev->dev,
+			 "Using %u as missing dma-requests property\n",
+			 MTK_HSDMA_NR_VCHANS);
+	}
+
+	hsdma->pc = devm_kcalloc(&pdev->dev, MTK_HSDMA_NR_MAX_PCHANS,
+				 sizeof(*hsdma->pc), GFP_KERNEL);
+	if (!hsdma->pc)
+		return -ENOMEM;
+
+	hsdma->vc = devm_kcalloc(&pdev->dev, hsdma->dma_requests,
+				 sizeof(*hsdma->vc), GFP_KERNEL);
+	if (!hsdma->vc)
+		return -ENOMEM;
+
+	for (i = 0; i < hsdma->dma_requests; i++) {
+		vc = &hsdma->vc[i];
+		vc->vc.desc_free = mtk_hsdma_vdesc_free;
+		vchan_init(&vc->vc, dd);
+		init_completion(&vc->issue_completion);
+		INIT_LIST_HEAD(&vc->desc_hw_processing);
+	}
+
+	err = dma_async_device_register(dd);
+	if (err)
+		return err;
+
+	err = of_dma_controller_register(pdev->dev.of_node,
+					 of_dma_xlate_by_chan_id, hsdma);
+	if (err) {
+		dev_err(&pdev->dev,
+			"MediaTek HSDMA OF registration failed %d\n", err);
+		goto err_unregister;
+	}
+
+	mtk_hsdma_hw_init(hsdma);
+
+	err = devm_request_irq(&pdev->dev, hsdma->irq,
+			       mtk_hsdma_irq, 0,
+			       dev_name(&pdev->dev), hsdma);
+	if (err) {
+		dev_err(&pdev->dev,
+			"request_irq failed with err %d\n", err);
+		goto err_unregister;
+	}
+
+	platform_set_drvdata(pdev, hsdma);
+
+	dev_info(&pdev->dev, "MediaTek HSDMA driver registered\n");
+
+	return 0;
+
+err_unregister:
+	dma_async_device_unregister(dd);
+
+	return err;
+}
+
+static int mtk_hsdma_remove(struct platform_device *pdev)
+{
+	struct mtk_hsdma_device *hsdma = platform_get_drvdata(pdev);
+	struct mtk_hsdma_vchan *vc;
+	int i;
+
+	/* Kill VC task */
+	for (i = 0; i < hsdma->dma_requests; i++) {
+		vc = &hsdma->vc[i];
+
+		list_del(&vc->vc.chan.device_node);
+		tasklet_kill(&vc->vc.task);
+	}
+
+	/* Disable DMA interrupt */
+	mtk_dma_write(hsdma, MTK_HSDMA_INT_ENABLE, 0);
+
+	/* Waits for any pending IRQ handlers to complete */
+	synchronize_irq(hsdma->irq);
+
+	/* Disable hardware */
+	mtk_hsdma_hw_deinit(hsdma);
+
+	dma_async_device_unregister(&hsdma->ddev);
+	of_dma_controller_free(pdev->dev.of_node);
+
+	return 0;
+}
+
+static struct platform_driver mtk_hsdma_driver = {
+	.probe		= mtk_hsdma_probe,
+	.remove		= mtk_hsdma_remove,
+	.driver = {
+		.name		= KBUILD_MODNAME,
+		.of_match_table	= mtk_hsdma_match,
+	},
+};
+module_platform_driver(mtk_hsdma_driver);
+
+MODULE_DESCRIPTION("MediaTek High-Speed DMA Controller Driver");
+MODULE_AUTHOR("Sean Wang <sean.wang@mediatek.com>");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/dma/pl330.c b/drivers/dma/pl330.c
index d7327fd5f445..de1fd59fe136 100644
--- a/drivers/dma/pl330.c
+++ b/drivers/dma/pl330.c
@@ -1510,7 +1510,7 @@ static void pl330_dotask(unsigned long data)
 /* Returns 1 if state was updated, 0 otherwise */
 static int pl330_update(struct pl330_dmac *pl330)
 {
-	struct dma_pl330_desc *descdone, *tmp;
+	struct dma_pl330_desc *descdone;
 	unsigned long flags;
 	void __iomem *regs;
 	u32 val;
@@ -1588,7 +1588,9 @@ static int pl330_update(struct pl330_dmac *pl330)
 	}
 
 	/* Now that we are in no hurry, do the callbacks */
-	list_for_each_entry_safe(descdone, tmp, &pl330->req_done, rqd) {
+	while (!list_empty(&pl330->req_done)) {
+		descdone = list_first_entry(&pl330->req_done,
+					    struct dma_pl330_desc, rqd);
 		list_del(&descdone->rqd);
 		spin_unlock_irqrestore(&pl330->lock, flags);
 		dma_pl330_rqcb(descdone, PL330_ERR_NONE);
diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c
index d076940e0c69..d29275b97e84 100644
--- a/drivers/dma/qcom/bam_dma.c
+++ b/drivers/dma/qcom/bam_dma.c
@@ -393,6 +393,7 @@ struct bam_device {
 	struct device_dma_parameters dma_parms;
 	struct bam_chan *channels;
 	u32 num_channels;
+	u32 num_ees;
 
 	/* execution environment ID, from DT */
 	u32 ee;
@@ -934,12 +935,15 @@ static void bam_apply_new_config(struct bam_chan *bchan,
 	struct bam_device *bdev = bchan->bdev;
 	u32 maxburst;
 
-	if (dir == DMA_DEV_TO_MEM)
-		maxburst = bchan->slave.src_maxburst;
-	else
-		maxburst = bchan->slave.dst_maxburst;
+	if (!bdev->controlled_remotely) {
+		if (dir == DMA_DEV_TO_MEM)
+			maxburst = bchan->slave.src_maxburst;
+		else
+			maxburst = bchan->slave.dst_maxburst;
 
-	writel_relaxed(maxburst, bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
+		writel_relaxed(maxburst,
+			       bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
+	}
 
 	bchan->reconfigure = 0;
 }
@@ -1128,15 +1132,19 @@ static int bam_init(struct bam_device *bdev)
 	u32 val;
 
 	/* read revision and configuration information */
-	val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION)) >> NUM_EES_SHIFT;
-	val &= NUM_EES_MASK;
+	if (!bdev->num_ees) {
+		val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
+		bdev->num_ees = (val >> NUM_EES_SHIFT) & NUM_EES_MASK;
+	}
 
 	/* check that configured EE is within range */
-	if (bdev->ee >= val)
+	if (bdev->ee >= bdev->num_ees)
 		return -EINVAL;
 
-	val = readl_relaxed(bam_addr(bdev, 0, BAM_NUM_PIPES));
-	bdev->num_channels = val & BAM_NUM_PIPES_MASK;
+	if (!bdev->num_channels) {
+		val = readl_relaxed(bam_addr(bdev, 0, BAM_NUM_PIPES));
+		bdev->num_channels = val & BAM_NUM_PIPES_MASK;
+	}
 
 	if (bdev->controlled_remotely)
 		return 0;
@@ -1232,9 +1240,25 @@ static int bam_dma_probe(struct platform_device *pdev)
 	bdev->controlled_remotely = of_property_read_bool(pdev->dev.of_node,
 						"qcom,controlled-remotely");
 
+	if (bdev->controlled_remotely) {
+		ret = of_property_read_u32(pdev->dev.of_node, "num-channels",
+					   &bdev->num_channels);
+		if (ret)
+			dev_err(bdev->dev, "num-channels unspecified in dt\n");
+
+		ret = of_property_read_u32(pdev->dev.of_node, "qcom,num-ees",
+					   &bdev->num_ees);
+		if (ret)
+			dev_err(bdev->dev, "num-ees unspecified in dt\n");
+	}
+
 	bdev->bamclk = devm_clk_get(bdev->dev, "bam_clk");
-	if (IS_ERR(bdev->bamclk))
-		return PTR_ERR(bdev->bamclk);
+	if (IS_ERR(bdev->bamclk)) {
+		if (!bdev->controlled_remotely)
+			return PTR_ERR(bdev->bamclk);
+
+		bdev->bamclk = NULL;
+	}
 
 	ret = clk_prepare_enable(bdev->bamclk);
 	if (ret) {
@@ -1309,6 +1333,11 @@ static int bam_dma_probe(struct platform_device *pdev)
 	if (ret)
 		goto err_unregister_dma;
 
+	if (bdev->controlled_remotely) {
+		pm_runtime_disable(&pdev->dev);
+		return 0;
+	}
+
 	pm_runtime_irq_safe(&pdev->dev);
 	pm_runtime_set_autosuspend_delay(&pdev->dev, BAM_DMA_AUTOSUSPEND_DELAY);
 	pm_runtime_use_autosuspend(&pdev->dev);
@@ -1392,7 +1421,8 @@ static int __maybe_unused bam_dma_suspend(struct device *dev)
 {
 	struct bam_device *bdev = dev_get_drvdata(dev);
 
-	pm_runtime_force_suspend(dev);
+	if (!bdev->controlled_remotely)
+		pm_runtime_force_suspend(dev);
 
 	clk_unprepare(bdev->bamclk);
 
@@ -1408,7 +1438,8 @@ static int __maybe_unused bam_dma_resume(struct device *dev)
 	if (ret)
 		return ret;
 
-	pm_runtime_force_resume(dev);
+	if (!bdev->controlled_remotely)
+		pm_runtime_force_resume(dev);
 
 	return 0;
 }
diff --git a/drivers/dma/sh/rcar-dmac.c b/drivers/dma/sh/rcar-dmac.c
index d0cacdb0713e..2a2ccd9c78e4 100644
--- a/drivers/dma/sh/rcar-dmac.c
+++ b/drivers/dma/sh/rcar-dmac.c
@@ -1301,8 +1301,17 @@ static unsigned int rcar_dmac_chan_get_residue(struct rcar_dmac_chan *chan,
 	 * If the cookie doesn't correspond to the currently running transfer
 	 * then the descriptor hasn't been processed yet, and the residue is
 	 * equal to the full descriptor size.
+	 * Also, a client driver is possible to call this function before
+	 * rcar_dmac_isr_channel_thread() runs. In this case, the "desc.running"
+	 * will be the next descriptor, and the done list will appear. So, if
+	 * the argument cookie matches the done list's cookie, we can assume
+	 * the residue is zero.
 	 */
 	if (cookie != desc->async_tx.cookie) {
+		list_for_each_entry(desc, &chan->desc.done, node) {
+			if (cookie == desc->async_tx.cookie)
+				return 0;
+		}
 		list_for_each_entry(desc, &chan->desc.pending, node) {
 			if (cookie == desc->async_tx.cookie)
 				return desc->size;
@@ -1677,8 +1686,8 @@ static const struct dev_pm_ops rcar_dmac_pm = {
 	 *   - Wait for the current transfer to complete and stop the device,
 	 *   - Resume transfers, if any.
 	 */
-	SET_LATE_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
-				     pm_runtime_force_resume)
+	SET_NOIRQ_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
+				      pm_runtime_force_resume)
 	SET_RUNTIME_PM_OPS(rcar_dmac_runtime_suspend, rcar_dmac_runtime_resume,
 			   NULL)
 };
diff --git a/drivers/dma/stm32-dma.c b/drivers/dma/stm32-dma.c
index 786fc8fcc38e..8c5807362a25 100644
--- a/drivers/dma/stm32-dma.c
+++ b/drivers/dma/stm32-dma.c
@@ -5,6 +5,7 @@
  *
  * Copyright (C) M'boumba Cedric Madianga 2015
  * Author: M'boumba Cedric Madianga <cedric.madianga@gmail.com>
+ *         Pierre-Yves Mordret <pierre-yves.mordret@st.com>
  *
  * License terms:  GNU General Public License (GPL), version 2
  */
@@ -33,9 +34,14 @@
 #define STM32_DMA_LIFCR			0x0008 /* DMA Low Int Flag Clear Reg */
 #define STM32_DMA_HIFCR			0x000c /* DMA High Int Flag Clear Reg */
 #define STM32_DMA_TCI			BIT(5) /* Transfer Complete Interrupt */
+#define STM32_DMA_HTI			BIT(4) /* Half Transfer Interrupt */
 #define STM32_DMA_TEI			BIT(3) /* Transfer Error Interrupt */
 #define STM32_DMA_DMEI			BIT(2) /* Direct Mode Error Interrupt */
 #define STM32_DMA_FEI			BIT(0) /* FIFO Error Interrupt */
+#define STM32_DMA_MASKI			(STM32_DMA_TCI \
+					 | STM32_DMA_TEI \
+					 | STM32_DMA_DMEI \
+					 | STM32_DMA_FEI)
 
 /* DMA Stream x Configuration Register */
 #define STM32_DMA_SCR(x)		(0x0010 + 0x18 * (x)) /* x = 0..7 */
@@ -60,7 +66,8 @@
 #define STM32_DMA_SCR_PINC		BIT(9) /* Peripheral increment mode */
 #define STM32_DMA_SCR_CIRC		BIT(8) /* Circular mode */
 #define STM32_DMA_SCR_PFCTRL		BIT(5) /* Peripheral Flow Controller */
-#define STM32_DMA_SCR_TCIE		BIT(4) /* Transfer Cplete Int Enable*/
+#define STM32_DMA_SCR_TCIE		BIT(4) /* Transfer Complete Int Enable
+						*/
 #define STM32_DMA_SCR_TEIE		BIT(2) /* Transfer Error Int Enable */
 #define STM32_DMA_SCR_DMEIE		BIT(1) /* Direct Mode Err Int Enable */
 #define STM32_DMA_SCR_EN		BIT(0) /* Stream Enable */
@@ -111,11 +118,24 @@
 #define STM32_DMA_FIFO_THRESHOLD_FULL			0x03
 
 #define STM32_DMA_MAX_DATA_ITEMS	0xffff
+/*
+ * Valid transfer starts from @0 to @0xFFFE leading to unaligned scatter
+ * gather at boundary. Thus it's safer to round down this value on FIFO
+ * size (16 Bytes)
+ */
+#define STM32_DMA_ALIGNED_MAX_DATA_ITEMS	\
+	ALIGN_DOWN(STM32_DMA_MAX_DATA_ITEMS, 16)
 #define STM32_DMA_MAX_CHANNELS		0x08
 #define STM32_DMA_MAX_REQUEST_ID	0x08
 #define STM32_DMA_MAX_DATA_PARAM	0x03
+#define STM32_DMA_FIFO_SIZE		16	/* FIFO is 16 bytes */
+#define STM32_DMA_MIN_BURST		4
 #define STM32_DMA_MAX_BURST		16
 
+/* DMA Features */
+#define STM32_DMA_THRESHOLD_FTR_MASK	GENMASK(1, 0)
+#define STM32_DMA_THRESHOLD_FTR_GET(n)	((n) & STM32_DMA_THRESHOLD_FTR_MASK)
+
 enum stm32_dma_width {
 	STM32_DMA_BYTE,
 	STM32_DMA_HALF_WORD,
@@ -129,11 +149,18 @@ enum stm32_dma_burst_size {
 	STM32_DMA_BURST_INCR16,
 };
 
+/**
+ * struct stm32_dma_cfg - STM32 DMA custom configuration
+ * @channel_id: channel ID
+ * @request_line: DMA request
+ * @stream_config: 32bit mask specifying the DMA channel configuration
+ * @features: 32bit mask specifying the DMA Feature list
+ */
 struct stm32_dma_cfg {
 	u32 channel_id;
 	u32 request_line;
 	u32 stream_config;
-	u32 threshold;
+	u32 features;
 };
 
 struct stm32_dma_chan_reg {
@@ -171,6 +198,9 @@ struct stm32_dma_chan {
 	u32 next_sg;
 	struct dma_slave_config	dma_sconfig;
 	struct stm32_dma_chan_reg chan_reg;
+	u32 threshold;
+	u32 mem_burst;
+	u32 mem_width;
 };
 
 struct stm32_dma_device {
@@ -235,6 +265,85 @@ static int stm32_dma_get_width(struct stm32_dma_chan *chan,
 	}
 }
 
+static enum dma_slave_buswidth stm32_dma_get_max_width(u32 buf_len,
+						       u32 threshold)
+{
+	enum dma_slave_buswidth max_width;
+
+	if (threshold == STM32_DMA_FIFO_THRESHOLD_FULL)
+		max_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
+	else
+		max_width = DMA_SLAVE_BUSWIDTH_2_BYTES;
+
+	while ((buf_len < max_width  || buf_len % max_width) &&
+	       max_width > DMA_SLAVE_BUSWIDTH_1_BYTE)
+		max_width = max_width >> 1;
+
+	return max_width;
+}
+
+static bool stm32_dma_fifo_threshold_is_allowed(u32 burst, u32 threshold,
+						enum dma_slave_buswidth width)
+{
+	u32 remaining;
+
+	if (width != DMA_SLAVE_BUSWIDTH_UNDEFINED) {
+		if (burst != 0) {
+			/*
+			 * If number of beats fit in several whole bursts
+			 * this configuration is allowed.
+			 */
+			remaining = ((STM32_DMA_FIFO_SIZE / width) *
+				     (threshold + 1) / 4) % burst;
+
+			if (remaining == 0)
+				return true;
+		} else {
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static bool stm32_dma_is_burst_possible(u32 buf_len, u32 threshold)
+{
+	switch (threshold) {
+	case STM32_DMA_FIFO_THRESHOLD_FULL:
+		if (buf_len >= STM32_DMA_MAX_BURST)
+			return true;
+		else
+			return false;
+	case STM32_DMA_FIFO_THRESHOLD_HALFFULL:
+		if (buf_len >= STM32_DMA_MAX_BURST / 2)
+			return true;
+		else
+			return false;
+	default:
+		return false;
+	}
+}
+
+static u32 stm32_dma_get_best_burst(u32 buf_len, u32 max_burst, u32 threshold,
+				    enum dma_slave_buswidth width)
+{
+	u32 best_burst = max_burst;
+
+	if (best_burst == 1 || !stm32_dma_is_burst_possible(buf_len, threshold))
+		return 0;
+
+	while ((buf_len < best_burst * width && best_burst > 1) ||
+	       !stm32_dma_fifo_threshold_is_allowed(best_burst, threshold,
+						    width)) {
+		if (best_burst > STM32_DMA_MIN_BURST)
+			best_burst = best_burst >> 1;
+		else
+			best_burst = 0;
+	}
+
+	return best_burst;
+}
+
 static int stm32_dma_get_burst(struct stm32_dma_chan *chan, u32 maxburst)
 {
 	switch (maxburst) {
@@ -254,12 +363,12 @@ static int stm32_dma_get_burst(struct stm32_dma_chan *chan, u32 maxburst)
 }
 
 static void stm32_dma_set_fifo_config(struct stm32_dma_chan *chan,
-				      u32 src_maxburst, u32 dst_maxburst)
+				      u32 src_burst, u32 dst_burst)
 {
 	chan->chan_reg.dma_sfcr &= ~STM32_DMA_SFCR_MASK;
 	chan->chan_reg.dma_scr &= ~STM32_DMA_SCR_DMEIE;
 
-	if ((!src_maxburst) && (!dst_maxburst)) {
+	if (!src_burst && !dst_burst) {
 		/* Using direct mode */
 		chan->chan_reg.dma_scr |= STM32_DMA_SCR_DMEIE;
 	} else {
@@ -300,7 +409,7 @@ static u32 stm32_dma_irq_status(struct stm32_dma_chan *chan)
 
 	flags = dma_isr >> (((chan->id & 2) << 3) | ((chan->id & 1) * 6));
 
-	return flags;
+	return flags & STM32_DMA_MASKI;
 }
 
 static void stm32_dma_irq_clear(struct stm32_dma_chan *chan, u32 flags)
@@ -315,6 +424,7 @@ static void stm32_dma_irq_clear(struct stm32_dma_chan *chan, u32 flags)
 	 * If (ch % 4) is 2 or 3, left shift the mask by 16 bits.
 	 * If (ch % 4) is 1 or 3, additionally left shift the mask by 6 bits.
 	 */
+	flags &= STM32_DMA_MASKI;
 	dma_ifcr = flags << (((chan->id & 2) << 3) | ((chan->id & 1) * 6));
 
 	if (chan->id & 4)
@@ -429,6 +539,8 @@ static void stm32_dma_dump_reg(struct stm32_dma_chan *chan)
 	dev_dbg(chan2dev(chan), "SFCR:  0x%08x\n", sfcr);
 }
 
+static void stm32_dma_configure_next_sg(struct stm32_dma_chan *chan);
+
 static void stm32_dma_start_transfer(struct stm32_dma_chan *chan)
 {
 	struct stm32_dma_device *dmadev = stm32_dma_get_dev(chan);
@@ -471,6 +583,9 @@ static void stm32_dma_start_transfer(struct stm32_dma_chan *chan)
 	if (status)
 		stm32_dma_irq_clear(chan, status);
 
+	if (chan->desc->cyclic)
+		stm32_dma_configure_next_sg(chan);
+
 	stm32_dma_dump_reg(chan);
 
 	/* Start DMA */
@@ -541,13 +656,29 @@ static irqreturn_t stm32_dma_chan_irq(int irq, void *devid)
 	status = stm32_dma_irq_status(chan);
 	scr = stm32_dma_read(dmadev, STM32_DMA_SCR(chan->id));
 
-	if ((status & STM32_DMA_TCI) && (scr & STM32_DMA_SCR_TCIE)) {
+	if (status & STM32_DMA_TCI) {
 		stm32_dma_irq_clear(chan, STM32_DMA_TCI);
-		stm32_dma_handle_chan_done(chan);
-
-	} else {
+		if (scr & STM32_DMA_SCR_TCIE)
+			stm32_dma_handle_chan_done(chan);
+		status &= ~STM32_DMA_TCI;
+	}
+	if (status & STM32_DMA_HTI) {
+		stm32_dma_irq_clear(chan, STM32_DMA_HTI);
+		status &= ~STM32_DMA_HTI;
+	}
+	if (status & STM32_DMA_FEI) {
+		stm32_dma_irq_clear(chan, STM32_DMA_FEI);
+		status &= ~STM32_DMA_FEI;
+		if (!(scr & STM32_DMA_SCR_EN))
+			dev_err(chan2dev(chan), "FIFO Error\n");
+		else
+			dev_dbg(chan2dev(chan), "FIFO over/underrun\n");
+	}
+	if (status) {
 		stm32_dma_irq_clear(chan, status);
 		dev_err(chan2dev(chan), "DMA error: status=0x%08x\n", status);
+		if (!(scr & STM32_DMA_SCR_EN))
+			dev_err(chan2dev(chan), "chan disabled by HW\n");
 	}
 
 	spin_unlock(&chan->vchan.lock);
@@ -564,45 +695,59 @@ static void stm32_dma_issue_pending(struct dma_chan *c)
 	if (vchan_issue_pending(&chan->vchan) && !chan->desc && !chan->busy) {
 		dev_dbg(chan2dev(chan), "vchan %p: issued\n", &chan->vchan);
 		stm32_dma_start_transfer(chan);
-		if (chan->desc->cyclic)
-			stm32_dma_configure_next_sg(chan);
+
 	}
 	spin_unlock_irqrestore(&chan->vchan.lock, flags);
 }
 
 static int stm32_dma_set_xfer_param(struct stm32_dma_chan *chan,
 				    enum dma_transfer_direction direction,
-				    enum dma_slave_buswidth *buswidth)
+				    enum dma_slave_buswidth *buswidth,
+				    u32 buf_len)
 {
 	enum dma_slave_buswidth src_addr_width, dst_addr_width;
 	int src_bus_width, dst_bus_width;
 	int src_burst_size, dst_burst_size;
-	u32 src_maxburst, dst_maxburst;
-	u32 dma_scr = 0;
+	u32 src_maxburst, dst_maxburst, src_best_burst, dst_best_burst;
+	u32 dma_scr, threshold;
 
 	src_addr_width = chan->dma_sconfig.src_addr_width;
 	dst_addr_width = chan->dma_sconfig.dst_addr_width;
 	src_maxburst = chan->dma_sconfig.src_maxburst;
 	dst_maxburst = chan->dma_sconfig.dst_maxburst;
+	threshold = chan->threshold;
 
 	switch (direction) {
 	case DMA_MEM_TO_DEV:
+		/* Set device data size */
 		dst_bus_width = stm32_dma_get_width(chan, dst_addr_width);
 		if (dst_bus_width < 0)
 			return dst_bus_width;
 
-		dst_burst_size = stm32_dma_get_burst(chan, dst_maxburst);
+		/* Set device burst size */
+		dst_best_burst = stm32_dma_get_best_burst(buf_len,
+							  dst_maxburst,
+							  threshold,
+							  dst_addr_width);
+
+		dst_burst_size = stm32_dma_get_burst(chan, dst_best_burst);
 		if (dst_burst_size < 0)
 			return dst_burst_size;
 
-		if (!src_addr_width)
-			src_addr_width = dst_addr_width;
-
+		/* Set memory data size */
+		src_addr_width = stm32_dma_get_max_width(buf_len, threshold);
+		chan->mem_width = src_addr_width;
 		src_bus_width = stm32_dma_get_width(chan, src_addr_width);
 		if (src_bus_width < 0)
 			return src_bus_width;
 
-		src_burst_size = stm32_dma_get_burst(chan, src_maxburst);
+		/* Set memory burst size */
+		src_maxburst = STM32_DMA_MAX_BURST;
+		src_best_burst = stm32_dma_get_best_burst(buf_len,
+							  src_maxburst,
+							  threshold,
+							  src_addr_width);
+		src_burst_size = stm32_dma_get_burst(chan, src_best_burst);
 		if (src_burst_size < 0)
 			return src_burst_size;
 
@@ -612,27 +757,46 @@ static int stm32_dma_set_xfer_param(struct stm32_dma_chan *chan,
 			STM32_DMA_SCR_PBURST(dst_burst_size) |
 			STM32_DMA_SCR_MBURST(src_burst_size);
 
+		/* Set FIFO threshold */
+		chan->chan_reg.dma_sfcr &= ~STM32_DMA_SFCR_FTH_MASK;
+		chan->chan_reg.dma_sfcr |= STM32_DMA_SFCR_FTH(threshold);
+
+		/* Set peripheral address */
 		chan->chan_reg.dma_spar = chan->dma_sconfig.dst_addr;
 		*buswidth = dst_addr_width;
 		break;
 
 	case DMA_DEV_TO_MEM:
+		/* Set device data size */
 		src_bus_width = stm32_dma_get_width(chan, src_addr_width);
 		if (src_bus_width < 0)
 			return src_bus_width;
 
-		src_burst_size = stm32_dma_get_burst(chan, src_maxburst);
+		/* Set device burst size */
+		src_best_burst = stm32_dma_get_best_burst(buf_len,
+							  src_maxburst,
+							  threshold,
+							  src_addr_width);
+		chan->mem_burst = src_best_burst;
+		src_burst_size = stm32_dma_get_burst(chan, src_best_burst);
 		if (src_burst_size < 0)
 			return src_burst_size;
 
-		if (!dst_addr_width)
-			dst_addr_width = src_addr_width;
-
+		/* Set memory data size */
+		dst_addr_width = stm32_dma_get_max_width(buf_len, threshold);
+		chan->mem_width = dst_addr_width;
 		dst_bus_width = stm32_dma_get_width(chan, dst_addr_width);
 		if (dst_bus_width < 0)
 			return dst_bus_width;
 
-		dst_burst_size = stm32_dma_get_burst(chan, dst_maxburst);
+		/* Set memory burst size */
+		dst_maxburst = STM32_DMA_MAX_BURST;
+		dst_best_burst = stm32_dma_get_best_burst(buf_len,
+							  dst_maxburst,
+							  threshold,
+							  dst_addr_width);
+		chan->mem_burst = dst_best_burst;
+		dst_burst_size = stm32_dma_get_burst(chan, dst_best_burst);
 		if (dst_burst_size < 0)
 			return dst_burst_size;
 
@@ -642,6 +806,11 @@ static int stm32_dma_set_xfer_param(struct stm32_dma_chan *chan,
 			STM32_DMA_SCR_PBURST(src_burst_size) |
 			STM32_DMA_SCR_MBURST(dst_burst_size);
 
+		/* Set FIFO threshold */
+		chan->chan_reg.dma_sfcr &= ~STM32_DMA_SFCR_FTH_MASK;
+		chan->chan_reg.dma_sfcr |= STM32_DMA_SFCR_FTH(threshold);
+
+		/* Set peripheral address */
 		chan->chan_reg.dma_spar = chan->dma_sconfig.src_addr;
 		*buswidth = chan->dma_sconfig.src_addr_width;
 		break;
@@ -651,8 +820,9 @@ static int stm32_dma_set_xfer_param(struct stm32_dma_chan *chan,
 		return -EINVAL;
 	}
 
-	stm32_dma_set_fifo_config(chan, src_maxburst, dst_maxburst);
+	stm32_dma_set_fifo_config(chan, src_best_burst, dst_best_burst);
 
+	/* Set DMA control register */
 	chan->chan_reg.dma_scr &= ~(STM32_DMA_SCR_DIR_MASK |
 			STM32_DMA_SCR_PSIZE_MASK | STM32_DMA_SCR_MSIZE_MASK |
 			STM32_DMA_SCR_PBURST_MASK | STM32_DMA_SCR_MBURST_MASK);
@@ -692,10 +862,6 @@ static struct dma_async_tx_descriptor *stm32_dma_prep_slave_sg(
 	if (!desc)
 		return NULL;
 
-	ret = stm32_dma_set_xfer_param(chan, direction, &buswidth);
-	if (ret < 0)
-		goto err;
-
 	/* Set peripheral flow controller */
 	if (chan->dma_sconfig.device_fc)
 		chan->chan_reg.dma_scr |= STM32_DMA_SCR_PFCTRL;
@@ -703,10 +869,15 @@ static struct dma_async_tx_descriptor *stm32_dma_prep_slave_sg(
 		chan->chan_reg.dma_scr &= ~STM32_DMA_SCR_PFCTRL;
 
 	for_each_sg(sgl, sg, sg_len, i) {
+		ret = stm32_dma_set_xfer_param(chan, direction, &buswidth,
+					       sg_dma_len(sg));
+		if (ret < 0)
+			goto err;
+
 		desc->sg_req[i].len = sg_dma_len(sg);
 
 		nb_data_items = desc->sg_req[i].len / buswidth;
-		if (nb_data_items > STM32_DMA_MAX_DATA_ITEMS) {
+		if (nb_data_items > STM32_DMA_ALIGNED_MAX_DATA_ITEMS) {
 			dev_err(chan2dev(chan), "nb items not supported\n");
 			goto err;
 		}
@@ -767,12 +938,12 @@ static struct dma_async_tx_descriptor *stm32_dma_prep_dma_cyclic(
 		return NULL;
 	}
 
-	ret = stm32_dma_set_xfer_param(chan, direction, &buswidth);
+	ret = stm32_dma_set_xfer_param(chan, direction, &buswidth, period_len);
 	if (ret < 0)
 		return NULL;
 
 	nb_data_items = period_len / buswidth;
-	if (nb_data_items > STM32_DMA_MAX_DATA_ITEMS) {
+	if (nb_data_items > STM32_DMA_ALIGNED_MAX_DATA_ITEMS) {
 		dev_err(chan2dev(chan), "number of items not supported\n");
 		return NULL;
 	}
@@ -816,35 +987,45 @@ static struct dma_async_tx_descriptor *stm32_dma_prep_dma_memcpy(
 	dma_addr_t src, size_t len, unsigned long flags)
 {
 	struct stm32_dma_chan *chan = to_stm32_dma_chan(c);
-	u32 num_sgs;
+	enum dma_slave_buswidth max_width;
 	struct stm32_dma_desc *desc;
 	size_t xfer_count, offset;
+	u32 num_sgs, best_burst, dma_burst, threshold;
 	int i;
 
-	num_sgs = DIV_ROUND_UP(len, STM32_DMA_MAX_DATA_ITEMS);
+	num_sgs = DIV_ROUND_UP(len, STM32_DMA_ALIGNED_MAX_DATA_ITEMS);
 	desc = stm32_dma_alloc_desc(num_sgs);
 	if (!desc)
 		return NULL;
 
+	threshold = chan->threshold;
+
 	for (offset = 0, i = 0; offset < len; offset += xfer_count, i++) {
 		xfer_count = min_t(size_t, len - offset,
-				   STM32_DMA_MAX_DATA_ITEMS);
+				   STM32_DMA_ALIGNED_MAX_DATA_ITEMS);
 
-		desc->sg_req[i].len = xfer_count;
+		/* Compute best burst size */
+		max_width = DMA_SLAVE_BUSWIDTH_1_BYTE;
+		best_burst = stm32_dma_get_best_burst(len, STM32_DMA_MAX_BURST,
+						      threshold, max_width);
+		dma_burst = stm32_dma_get_burst(chan, best_burst);
 
 		stm32_dma_clear_reg(&desc->sg_req[i].chan_reg);
 		desc->sg_req[i].chan_reg.dma_scr =
 			STM32_DMA_SCR_DIR(STM32_DMA_MEM_TO_MEM) |
+			STM32_DMA_SCR_PBURST(dma_burst) |
+			STM32_DMA_SCR_MBURST(dma_burst) |
 			STM32_DMA_SCR_MINC |
 			STM32_DMA_SCR_PINC |
 			STM32_DMA_SCR_TCIE |
 			STM32_DMA_SCR_TEIE;
-		desc->sg_req[i].chan_reg.dma_sfcr = STM32_DMA_SFCR_DMDIS |
-			STM32_DMA_SFCR_FTH(STM32_DMA_FIFO_THRESHOLD_FULL) |
-			STM32_DMA_SFCR_FEIE;
+		desc->sg_req[i].chan_reg.dma_sfcr |= STM32_DMA_SFCR_MASK;
+		desc->sg_req[i].chan_reg.dma_sfcr |=
+			STM32_DMA_SFCR_FTH(threshold);
 		desc->sg_req[i].chan_reg.dma_spar = src + offset;
 		desc->sg_req[i].chan_reg.dma_sm0ar = dest + offset;
 		desc->sg_req[i].chan_reg.dma_sndtr = xfer_count;
+		desc->sg_req[i].len = xfer_count;
 	}
 
 	desc->num_sgs = num_sgs;
@@ -869,6 +1050,7 @@ static size_t stm32_dma_desc_residue(struct stm32_dma_chan *chan,
 				     struct stm32_dma_desc *desc,
 				     u32 next_sg)
 {
+	u32 modulo, burst_size;
 	u32 residue = 0;
 	int i;
 
@@ -876,8 +1058,10 @@ static size_t stm32_dma_desc_residue(struct stm32_dma_chan *chan,
 	 * In cyclic mode, for the last period, residue = remaining bytes from
 	 * NDTR
 	 */
-	if (chan->desc->cyclic && next_sg == 0)
-		return stm32_dma_get_remaining_bytes(chan);
+	if (chan->desc->cyclic && next_sg == 0) {
+		residue = stm32_dma_get_remaining_bytes(chan);
+		goto end;
+	}
 
 	/*
 	 * For all other periods in cyclic mode, and in sg mode,
@@ -888,6 +1072,15 @@ static size_t stm32_dma_desc_residue(struct stm32_dma_chan *chan,
 		residue += desc->sg_req[i].len;
 	residue += stm32_dma_get_remaining_bytes(chan);
 
+end:
+	if (!chan->mem_burst)
+		return residue;
+
+	burst_size = chan->mem_burst * chan->mem_width;
+	modulo = residue % burst_size;
+	if (modulo)
+		residue = residue - modulo + burst_size;
+
 	return residue;
 }
 
@@ -902,7 +1095,7 @@ static enum dma_status stm32_dma_tx_status(struct dma_chan *c,
 	u32 residue = 0;
 
 	status = dma_cookie_status(c, cookie, state);
-	if ((status == DMA_COMPLETE) || (!state))
+	if (status == DMA_COMPLETE || !state)
 		return status;
 
 	spin_lock_irqsave(&chan->vchan.lock, flags);
@@ -966,7 +1159,7 @@ static void stm32_dma_desc_free(struct virt_dma_desc *vdesc)
 }
 
 static void stm32_dma_set_config(struct stm32_dma_chan *chan,
-			  struct stm32_dma_cfg *cfg)
+				 struct stm32_dma_cfg *cfg)
 {
 	stm32_dma_clear_reg(&chan->chan_reg);
 
@@ -976,7 +1169,7 @@ static void stm32_dma_set_config(struct stm32_dma_chan *chan,
 	/* Enable Interrupts  */
 	chan->chan_reg.dma_scr |= STM32_DMA_SCR_TEIE | STM32_DMA_SCR_TCIE;
 
-	chan->chan_reg.dma_sfcr = cfg->threshold & STM32_DMA_SFCR_FTH_MASK;
+	chan->threshold = STM32_DMA_THRESHOLD_FTR_GET(cfg->features);
 }
 
 static struct dma_chan *stm32_dma_of_xlate(struct of_phandle_args *dma_spec,
@@ -996,10 +1189,10 @@ static struct dma_chan *stm32_dma_of_xlate(struct of_phandle_args *dma_spec,
 	cfg.channel_id = dma_spec->args[0];
 	cfg.request_line = dma_spec->args[1];
 	cfg.stream_config = dma_spec->args[2];
-	cfg.threshold = dma_spec->args[3];
+	cfg.features = dma_spec->args[3];
 
-	if ((cfg.channel_id >= STM32_DMA_MAX_CHANNELS) ||
-	    (cfg.request_line >= STM32_DMA_MAX_REQUEST_ID)) {
+	if (cfg.channel_id >= STM32_DMA_MAX_CHANNELS ||
+	    cfg.request_line >= STM32_DMA_MAX_REQUEST_ID) {
 		dev_err(dev, "Bad channel and/or request id\n");
 		return NULL;
 	}