Merge branch 'topic/pxa' into for-linus

6 files changed, 1660 insertions, 0 deletions

diff --git a/Documentation/dmaengine/pxa_dma.txt b/Documentation/dmaengine/pxa_dma.txt
new file mode 100644
index 000000000000..413ef9cfaa4d
--- /dev/null
+++ b/Documentation/dmaengine/pxa_dma.txt
@@ -0,0 +1,153 @@
+PXA/MMP - DMA Slave controller
+==============================
+
+Constraints
+-----------
+  a) Transfers hot queuing
+     A driver submitting a transfer and issuing it should be granted the transfer
+     is queued even on a running DMA channel.
+     This implies that the queuing doesn't wait for the previous transfer end,
+     and that the descriptor chaining is not only done in the irq/tasklet code
+     triggered by the end of the transfer.
+     A transfer which is submitted and issued on a phy doesn't wait for a phy to
+     stop and restart, but is submitted on a "running channel". The other
+     drivers, especially mmp_pdma waited for the phy to stop before relaunching
+     a new transfer.
+
+  b) All transfers having asked for confirmation should be signaled
+     Any issued transfer with DMA_PREP_INTERRUPT should trigger a callback call.
+     This implies that even if an irq/tasklet is triggered by end of tx1, but
+     at the time of irq/dma tx2 is already finished, tx1->complete() and
+     tx2->complete() should be called.
+
+  c) Channel running state
+     A driver should be able to query if a channel is running or not. For the
+     multimedia case, such as video capture, if a transfer is submitted and then
+     a check of the DMA channel reports a "stopped channel", the transfer should
+     not be issued until the next "start of frame interrupt", hence the need to
+     know if a channel is in running or stopped state.
+
+  d) Bandwidth guarantee
+     The PXA architecture has 4 levels of DMAs priorities : high, normal, low.
+     The high prorities get twice as much bandwidth as the normal, which get twice
+     as much as the low priorities.
+     A driver should be able to request a priority, especially the real-time
+     ones such as pxa_camera with (big) throughputs.
+
+Design
+------
+  a) Virtual channels
+     Same concept as in sa11x0 driver, ie. a driver was assigned a "virtual
+     channel" linked to the requestor line, and the physical DMA channel is
+     assigned on the fly when the transfer is issued.
+
+  b) Transfer anatomy for a scatter-gather transfer
+     +------------+-----+---------------+----------------+-----------------+
+     | desc-sg[0] | ... | desc-sg[last] | status updater | finisher/linker |
+     +------------+-----+---------------+----------------+-----------------+
+
+     This structure is pointed by dma->sg_cpu.
+     The descriptors are used as follows :
+      - desc-sg[i]: i-th descriptor, transferring the i-th sg
+        element to the video buffer scatter gather
+      - status updater
+        Transfers a single u32 to a well known dma coherent memory to leave
+        a trace that this transfer is done. The "well known" is unique per
+        physical channel, meaning that a read of this value will tell which
+        is the last finished transfer at that point in time.
+      - finisher: has ddadr=DADDR_STOP, dcmd=ENDIRQEN
+      - linker: has ddadr= desc-sg[0] of next transfer, dcmd=0
+
+  c) Transfers hot-chaining
+     Suppose the running chain is :
+         Buffer 1         Buffer 2
+     +---------+----+---+  +----+----+----+---+
+     | d0 | .. | dN | l |  | d0 | .. | dN | f |
+     +---------+----+-|-+  ^----+----+----+---+
+                      |    |
+                      +----+
+
+     After a call to dmaengine_submit(b3), the chain will look like :
+          Buffer 1              Buffer 2             Buffer 3
+     +---------+----+---+  +----+----+----+---+  +----+----+----+---+
+     | d0 | .. | dN | l |  | d0 | .. | dN | l |  | d0 | .. | dN | f |
+     +---------+----+-|-+  ^----+----+----+-|-+  ^----+----+----+---+
+                      |    |                |    |
+                      +----+                +----+
+                                           new_link
+
+     If while new_link was created the DMA channel stopped, it is _not_
+     restarted. Hot-chaining doesn't break the assumption that
+     dma_async_issue_pending() is to be used to ensure the transfer is actually started.
+
+     One exception to this rule :
+       - if Buffer1 and Buffer2 had all their addresses 8 bytes aligned
+       - and if Buffer3 has at least one address not 4 bytes aligned
+       - then hot-chaining cannot happen, as the channel must be stopped, the
+         "align bit" must be set, and the channel restarted As a consequence,
+         such a transfer tx_submit() will be queued on the submitted queue, and
+         this specific case if the DMA is already running in aligned mode.
+
+  d) Transfers completion updater
+     Each time a transfer is completed on a channel, an interrupt might be
+     generated or not, up to the client's request. But in each case, the last
+     descriptor of a transfer, the "status updater", will write the latest
+     transfer being completed into the physical channel's completion mark.
+
+     This will speed up residue calculation, for large transfers such as video
+     buffers which hold around 6k descriptors or more. This also allows without
+     any lock to find out what is the latest completed transfer in a running
+     DMA chain.
+
+  e) Transfers completion, irq and tasklet
+     When a transfer flagged as "DMA_PREP_INTERRUPT" is finished, the dma irq
+     is raised. Upon this interrupt, a tasklet is scheduled for the physical
+     channel.
+     The tasklet is responsible for :
+      - reading the physical channel last updater mark
+      - calling all the transfer callbacks of finished transfers, based on
+        that mark, and each transfer flags.
+     If a transfer is completed while this handling is done, a dma irq will
+     be raised, and the tasklet will be scheduled once again, having a new
+     updater mark.
+
+  f) Residue
+     Residue granularity will be descriptor based. The issued but not completed
+     transfers will be scanned for all of their descriptors against the
+     currently running descriptor.
+
+  g) Most complicated case of driver's tx queues
+     The most tricky situation is when :
+       - there are not "acked" transfers (tx0)
+       - a driver submitted an aligned tx1, not chained
+       - a driver submitted an aligned tx2 => tx2 is cold chained to tx1
+       - a driver issued tx1+tx2 => channel is running in aligned mode
+       - a driver submitted an aligned tx3 => tx3 is hot-chained
+       - a driver submitted an unaligned tx4 => tx4 is put in submitted queue,
+         not chained
+       - a driver issued tx4 => tx4 is put in issued queue, not chained
+       - a driver submitted an aligned tx5 => tx5 is put in submitted queue, not
+         chained
+       - a driver submitted an aligned tx6 => tx6 is put in submitted queue,
+         cold chained to tx5
+
+     This translates into (after tx4 is issued) :
+       - issued queue
+     +-----+ +-----+ +-----+ +-----+
+     | tx1 | | tx2 | | tx3 | | tx4 |
+     +---|-+ ^---|-+ ^-----+ +-----+
+         |   |   |   |
+         +---+   +---+
+       - submitted queue
+     +-----+ +-----+
+     | tx5 | | tx6 |
+     +---|-+ ^-----+
+         |   |
+         +---+
+       - completed queue : empty
+       - allocated queue : tx0
+
+     It should be noted that after tx3 is completed, the channel is stopped, and
+     restarted in "unaligned mode" to handle tx4.
+
+Author: Robert Jarzmik <robert.jarzmik@free.fr>
diff --git a/MAINTAINERS b/MAINTAINERS
index 781e099495d3..238ea990ed29 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -7937,6 +7937,7 @@ T:	git git://github.com/hzhuang1/linux.git
 T:	git git://github.com/rjarzmik/linux.git
 S:	Maintained
 F:	arch/arm/mach-pxa/
+F:	drivers/dma/pxa*
 F:	drivers/pcmcia/pxa2xx*
 F:	drivers/spi/spi-pxa2xx*
 F:	drivers/usb/gadget/udc/pxa2*
diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig
index 0205ade0ba32..88d474b78076 100644
--- a/drivers/dma/Kconfig
+++ b/drivers/dma/Kconfig
@@ -162,6 +162,17 @@ config MX3_IPU_IRQS
 	  To avoid bloating the irq_desc[] array we allocate a sufficient
 	  number of IRQ slots and map them dynamically to specific sources.
 
+config PXA_DMA
+	bool "PXA DMA support"
+	depends on (ARCH_MMP || ARCH_PXA)
+	select DMA_ENGINE
+	select DMA_VIRTUAL_CHANNELS
+	help
+	  Support the DMA engine for PXA. It is also compatible with MMP PDMA
+	  platform. The internal DMA IP of all PXA variants is supported, with
+	  16 to 32 channels for peripheral to memory or memory to memory
+	  transfers.
+
 config TXX9_DMAC
 	tristate "Toshiba TXx9 SoC DMA support"
 	depends on MACH_TX49XX || MACH_TX39XX
diff --git a/drivers/dma/Makefile b/drivers/dma/Makefile
index 535919559f12..6a4d6f2827da 100644
--- a/drivers/dma/Makefile
+++ b/drivers/dma/Makefile
@@ -25,6 +25,7 @@ obj-$(CONFIG_AMCC_PPC440SPE_ADMA) += ppc4xx/
 obj-$(CONFIG_IMX_SDMA) += imx-sdma.o
 obj-$(CONFIG_IMX_DMA) += imx-dma.o
 obj-$(CONFIG_MXS_DMA) += mxs-dma.o
+obj-$(CONFIG_PXA_DMA) += pxa_dma.o
 obj-$(CONFIG_TIMB_DMA) += timb_dma.o
 obj-$(CONFIG_SIRF_DMA) += sirf-dma.o
 obj-$(CONFIG_TI_EDMA) += edma.o
diff --git a/drivers/dma/pxa_dma.c b/drivers/dma/pxa_dma.c
new file mode 100644
index 000000000000..ddcbbf5cd9e9
--- /dev/null
+++ b/drivers/dma/pxa_dma.c
@@ -0,0 +1,1467 @@
+/*
+ * Copyright 2015 Robert Jarzmik <robert.jarzmik@free.fr>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/err.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/types.h>
+#include <linux/interrupt.h>
+#include <linux/dma-mapping.h>
+#include <linux/slab.h>
+#include <linux/dmaengine.h>
+#include <linux/platform_device.h>
+#include <linux/device.h>
+#include <linux/platform_data/mmp_dma.h>
+#include <linux/dmapool.h>
+#include <linux/of_device.h>
+#include <linux/of_dma.h>
+#include <linux/of.h>
+#include <linux/dma/pxa-dma.h>
+
+#include "dmaengine.h"
+#include "virt-dma.h"
+
+#define DCSR(n)		(0x0000 + ((n) << 2))
+#define DALGN(n)	0x00a0
+#define DINT		0x00f0
+#define DDADR(n)	(0x0200 + ((n) << 4))
+#define DSADR(n)	(0x0204 + ((n) << 4))
+#define DTADR(n)	(0x0208 + ((n) << 4))
+#define DCMD(n)		(0x020c + ((n) << 4))
+
+#define PXA_DCSR_RUN		BIT(31)	/* Run Bit (read / write) */
+#define PXA_DCSR_NODESC		BIT(30)	/* No-Descriptor Fetch (read / write) */
+#define PXA_DCSR_STOPIRQEN	BIT(29)	/* Stop Interrupt Enable (R/W) */
+#define PXA_DCSR_REQPEND	BIT(8)	/* Request Pending (read-only) */
+#define PXA_DCSR_STOPSTATE	BIT(3)	/* Stop State (read-only) */
+#define PXA_DCSR_ENDINTR	BIT(2)	/* End Interrupt (read / write) */
+#define PXA_DCSR_STARTINTR	BIT(1)	/* Start Interrupt (read / write) */
+#define PXA_DCSR_BUSERR		BIT(0)	/* Bus Error Interrupt (read / write) */
+
+#define PXA_DCSR_EORIRQEN	BIT(28)	/* End of Receive IRQ Enable (R/W) */
+#define PXA_DCSR_EORJMPEN	BIT(27)	/* Jump to next descriptor on EOR */
+#define PXA_DCSR_EORSTOPEN	BIT(26)	/* STOP on an EOR */
+#define PXA_DCSR_SETCMPST	BIT(25)	/* Set Descriptor Compare Status */
+#define PXA_DCSR_CLRCMPST	BIT(24)	/* Clear Descriptor Compare Status */
+#define PXA_DCSR_CMPST		BIT(10)	/* The Descriptor Compare Status */
+#define PXA_DCSR_EORINTR	BIT(9)	/* The end of Receive */
+
+#define DRCMR_MAPVLD	BIT(7)	/* Map Valid (read / write) */
+#define DRCMR_CHLNUM	0x1f	/* mask for Channel Number (read / write) */
+
+#define DDADR_DESCADDR	0xfffffff0	/* Address of next descriptor (mask) */
+#define DDADR_STOP	BIT(0)	/* Stop (read / write) */
+
+#define PXA_DCMD_INCSRCADDR	BIT(31)	/* Source Address Increment Setting. */
+#define PXA_DCMD_INCTRGADDR	BIT(30)	/* Target Address Increment Setting. */
+#define PXA_DCMD_FLOWSRC	BIT(29)	/* Flow Control by the source. */
+#define PXA_DCMD_FLOWTRG	BIT(28)	/* Flow Control by the target. */
+#define PXA_DCMD_STARTIRQEN	BIT(22)	/* Start Interrupt Enable */
+#define PXA_DCMD_ENDIRQEN	BIT(21)	/* End Interrupt Enable */
+#define PXA_DCMD_ENDIAN		BIT(18)	/* Device Endian-ness. */
+#define PXA_DCMD_BURST8		(1 << 16)	/* 8 byte burst */
+#define PXA_DCMD_BURST16	(2 << 16)	/* 16 byte burst */
+#define PXA_DCMD_BURST32	(3 << 16)	/* 32 byte burst */
+#define PXA_DCMD_WIDTH1		(1 << 14)	/* 1 byte width */
+#define PXA_DCMD_WIDTH2		(2 << 14)	/* 2 byte width (HalfWord) */
+#define PXA_DCMD_WIDTH4		(3 << 14)	/* 4 byte width (Word) */
+#define PXA_DCMD_LENGTH		0x01fff		/* length mask (max = 8K - 1) */
+
+#define PDMA_ALIGNMENT		3
+#define PDMA_MAX_DESC_BYTES	(PXA_DCMD_LENGTH & ~((1 << PDMA_ALIGNMENT) - 1))
+
+struct pxad_desc_hw {
+	u32 ddadr;	/* Points to the next descriptor + flags */
+	u32 dsadr;	/* DSADR value for the current transfer */
+	u32 dtadr;	/* DTADR value for the current transfer */
+	u32 dcmd;	/* DCMD value for the current transfer */
+} __aligned(16);
+
+struct pxad_desc_sw {
+	struct virt_dma_desc	vd;		/* Virtual descriptor */
+	int			nb_desc;	/* Number of hw. descriptors */
+	size_t			len;		/* Number of bytes xfered */
+	dma_addr_t		first;		/* First descriptor's addr */
+
+	/* At least one descriptor has an src/dst address not multiple of 8 */
+	bool			misaligned;
+	bool			cyclic;
+	struct dma_pool		*desc_pool;	/* Channel's used allocator */
+
+	struct pxad_desc_hw	*hw_desc[];	/* DMA coherent descriptors */
+};
+
+struct pxad_phy {
+	int			idx;
+	void __iomem		*base;
+	struct pxad_chan	*vchan;
+};
+
+struct pxad_chan {
+	struct virt_dma_chan	vc;		/* Virtual channel */
+	u32			drcmr;		/* Requestor of the channel */
+	enum pxad_chan_prio	prio;		/* Required priority of phy */
+	/*
+	 * At least one desc_sw in submitted or issued transfers on this channel
+	 * has one address such as: addr % 8 != 0. This implies the DALGN
+	 * setting on the phy.
+	 */
+	bool			misaligned;
+	struct dma_slave_config	cfg;		/* Runtime config */
+
+	/* protected by vc->lock */
+	struct pxad_phy		*phy;
+	struct dma_pool		*desc_pool;	/* Descriptors pool */
+};
+
+struct pxad_device {
+	struct dma_device		slave;
+	int				nr_chans;
+	void __iomem			*base;
+	struct pxad_phy			*phys;
+	spinlock_t			phy_lock;	/* Phy association */
+#ifdef CONFIG_DEBUG_FS
+	struct dentry			*dbgfs_root;
+	struct dentry			*dbgfs_state;
+	struct dentry			**dbgfs_chan;
+#endif
+};
+
+#define tx_to_pxad_desc(tx)					\
+	container_of(tx, struct pxad_desc_sw, async_tx)
+#define to_pxad_chan(dchan)					\
+	container_of(dchan, struct pxad_chan, vc.chan)
+#define to_pxad_dev(dmadev)					\
+	container_of(dmadev, struct pxad_device, slave)
+#define to_pxad_sw_desc(_vd)				\
+	container_of((_vd), struct pxad_desc_sw, vd)
+
+#define _phy_readl_relaxed(phy, _reg)					\
+	readl_relaxed((phy)->base + _reg((phy)->idx))
+#define phy_readl_relaxed(phy, _reg)					\
+	({								\
+		u32 _v;							\
+		_v = readl_relaxed((phy)->base + _reg((phy)->idx));	\
+		dev_vdbg(&phy->vchan->vc.chan.dev->device,		\
+			 "%s(): readl(%s): 0x%08x\n", __func__, #_reg,	\
+			  _v);						\
+		_v;							\
+	})
+#define phy_writel(phy, val, _reg)					\
+	do {								\
+		writel((val), (phy)->base + _reg((phy)->idx));		\
+		dev_vdbg(&phy->vchan->vc.chan.dev->device,		\
+			 "%s(): writel(0x%08x, %s)\n",			\
+			 __func__, (u32)(val), #_reg);			\
+	} while (0)
+#define phy_writel_relaxed(phy, val, _reg)				\
+	do {								\
+		writel_relaxed((val), (phy)->base + _reg((phy)->idx));	\
+		dev_vdbg(&phy->vchan->vc.chan.dev->device,		\
+			 "%s(): writel_relaxed(0x%08x, %s)\n",		\
+			 __func__, (u32)(val), #_reg);			\
+	} while (0)
+
+static unsigned int pxad_drcmr(unsigned int line)
+{
+	if (line < 64)
+		return 0x100 + line * 4;
+	return 0x1000 + line * 4;
+}
+
+/*
+ * Debug fs
+ */
+#ifdef CONFIG_DEBUG_FS
+#include <linux/debugfs.h>
+#include <linux/uaccess.h>
+#include <linux/seq_file.h>
+
+static int dbg_show_requester_chan(struct seq_file *s, void *p)
+{
+	int pos = 0;
+	struct pxad_phy *phy = s->private;
+	int i;
+	u32 drcmr;
+
+	pos += seq_printf(s, "DMA channel %d requester :\n", phy->idx);
+	for (i = 0; i < 70; i++) {
+		drcmr = readl_relaxed(phy->base + pxad_drcmr(i));
+		if ((drcmr & DRCMR_CHLNUM) == phy->idx)
+			pos += seq_printf(s, "\tRequester %d (MAPVLD=%d)\n", i,
+					  !!(drcmr & DRCMR_MAPVLD));
+	}
+	return pos;
+}
+
+static inline int dbg_burst_from_dcmd(u32 dcmd)
+{
+	int burst = (dcmd >> 16) & 0x3;
+
+	return burst ? 4 << burst : 0;
+}
+
+static int is_phys_valid(unsigned long addr)
+{
+	return pfn_valid(__phys_to_pfn(addr));
+}
+
+#define PXA_DCSR_STR(flag) (dcsr & PXA_DCSR_##flag ? #flag" " : "")
+#define PXA_DCMD_STR(flag) (dcmd & PXA_DCMD_##flag ? #flag" " : "")
+
+static int dbg_show_descriptors(struct seq_file *s, void *p)
+{
+	struct pxad_phy *phy = s->private;
+	int i, max_show = 20, burst, width;
+	u32 dcmd;
+	unsigned long phys_desc, ddadr;
+	struct pxad_desc_hw *desc;
+
+	phys_desc = ddadr = _phy_readl_relaxed(phy, DDADR);
+
+	seq_printf(s, "DMA channel %d descriptors :\n", phy->idx);
+	seq_printf(s, "[%03d] First descriptor unknown\n", 0);
+	for (i = 1; i < max_show && is_phys_valid(phys_desc); i++) {
+		desc = phys_to_virt(phys_desc);
+		dcmd = desc->dcmd;
+		burst = dbg_burst_from_dcmd(dcmd);
+		width = (1 << ((dcmd >> 14) & 0x3)) >> 1;
+
+		seq_printf(s, "[%03d] Desc at %08lx(virt %p)\n",
+			   i, phys_desc, desc);
+		seq_printf(s, "\tDDADR = %08x\n", desc->ddadr);
+		seq_printf(s, "\tDSADR = %08x\n", desc->dsadr);
+		seq_printf(s, "\tDTADR = %08x\n", desc->dtadr);
+		seq_printf(s, "\tDCMD  = %08x (%s%s%s%s%s%s%sburst=%d width=%d len=%d)\n",
+			   dcmd,
+			   PXA_DCMD_STR(INCSRCADDR), PXA_DCMD_STR(INCTRGADDR),
+			   PXA_DCMD_STR(FLOWSRC), PXA_DCMD_STR(FLOWTRG),
+			   PXA_DCMD_STR(STARTIRQEN), PXA_DCMD_STR(ENDIRQEN),
+			   PXA_DCMD_STR(ENDIAN), burst, width,
+			   dcmd & PXA_DCMD_LENGTH);
+		phys_desc = desc->ddadr;
+	}
+	if (i == max_show)
+		seq_printf(s, "[%03d] Desc at %08lx ... max display reached\n",
+			   i, phys_desc);
+	else
+		seq_printf(s, "[%03d] Desc at %08lx is %s\n",
+			   i, phys_desc, phys_desc == DDADR_STOP ?
+			   "DDADR_STOP" : "invalid");
+
+	return 0;
+}
+
+static int dbg_show_chan_state(struct seq_file *s, void *p)
+{
+	struct pxad_phy *phy = s->private;
+	u32 dcsr, dcmd;
+	int burst, width;
+	static const char * const str_prio[] = {
+		"high", "normal", "low", "invalid"
+	};
+
+	dcsr = _phy_readl_relaxed(phy, DCSR);
+	dcmd = _phy_readl_relaxed(phy, DCMD);
+	burst = dbg_burst_from_dcmd(dcmd);
+	width = (1 << ((dcmd >> 14) & 0x3)) >> 1;
+
+	seq_printf(s, "DMA channel %d\n", phy->idx);
+	seq_printf(s, "\tPriority : %s\n",
+			  str_prio[(phy->idx & 0xf) / 4]);
+	seq_printf(s, "\tUnaligned transfer bit: %s\n",
+			  _phy_readl_relaxed(phy, DALGN) & BIT(phy->idx) ?
+			  "yes" : "no");
+	seq_printf(s, "\tDCSR  = %08x (%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s)\n",
+		   dcsr, PXA_DCSR_STR(RUN), PXA_DCSR_STR(NODESC),
+		   PXA_DCSR_STR(STOPIRQEN), PXA_DCSR_STR(EORIRQEN),
+		   PXA_DCSR_STR(EORJMPEN), PXA_DCSR_STR(EORSTOPEN),
+		   PXA_DCSR_STR(SETCMPST), PXA_DCSR_STR(CLRCMPST),
+		   PXA_DCSR_STR(CMPST), PXA_DCSR_STR(EORINTR),
+		   PXA_DCSR_STR(REQPEND), PXA_DCSR_STR(STOPSTATE),
+		   PXA_DCSR_STR(ENDINTR), PXA_DCSR_STR(STARTINTR),
+		   PXA_DCSR_STR(BUSERR));
+
+	seq_printf(s, "\tDCMD  = %08x (%s%s%s%s%s%s%sburst=%d width=%d len=%d)\n",
+		   dcmd,
+		   PXA_DCMD_STR(INCSRCADDR), PXA_DCMD_STR(INCTRGADDR),
+		   PXA_DCMD_STR(FLOWSRC), PXA_DCMD_STR(FLOWTRG),
+		   PXA_DCMD_STR(STARTIRQEN), PXA_DCMD_STR(ENDIRQEN),
+		   PXA_DCMD_STR(ENDIAN), burst, width, dcmd & PXA_DCMD_LENGTH);
+	seq_printf(s, "\tDSADR = %08x\n", _phy_readl_relaxed(phy, DSADR));
+	seq_printf(s, "\tDTADR = %08x\n", _phy_readl_relaxed(phy, DTADR));
+	seq_printf(s, "\tDDADR = %08x\n", _phy_readl_relaxed(phy, DDADR));
+
+	return 0;
+}
+
+static int dbg_show_state(struct seq_file *s, void *p)
+{
+	struct pxad_device *pdev = s->private;
+
+	/* basic device status */
+	seq_puts(s, "DMA engine status\n");
+	seq_printf(s, "\tChannel number: %d\n", pdev->nr_chans);
+
+	return 0;
+}
+
+#define DBGFS_FUNC_DECL(name) \
+static int dbg_open_##name(struct inode *inode, struct file *file) \
+{ \
+	return single_open(file, dbg_show_##name, inode->i_private); \
+} \
+static const struct file_operations dbg_fops_##name = { \
+	.owner		= THIS_MODULE, \
+	.open		= dbg_open_##name, \
+	.llseek		= seq_lseek, \
+	.read		= seq_read, \
+	.release	= single_release, \
+}
+
+DBGFS_FUNC_DECL(state);
+DBGFS_FUNC_DECL(chan_state);
+DBGFS_FUNC_DECL(descriptors);
+DBGFS_FUNC_DECL(requester_chan);
+
+static struct dentry *pxad_dbg_alloc_chan(struct pxad_device *pdev,
+					     int ch, struct dentry *chandir)
+{
+	char chan_name[11];
+	struct dentry *chan, *chan_state = NULL, *chan_descr = NULL;
+	struct dentry *chan_reqs = NULL;
+	void *dt;
+
+	scnprintf(chan_name, sizeof(chan_name), "%d", ch);
+	chan = debugfs_create_dir(chan_name, chandir);
+	dt = (void *)&pdev->phys[ch];
+
+	if (chan)
+		chan_state = debugfs_create_file("state", 0400, chan, dt,
+						 &dbg_fops_chan_state);
+	if (chan_state)
+		chan_descr = debugfs_create_file("descriptors", 0400, chan, dt,
+						 &dbg_fops_descriptors);
+	if (chan_descr)
+		chan_reqs = debugfs_create_file("requesters", 0400, chan, dt,
+						&dbg_fops_requester_chan);
+	if (!chan_reqs)
+		goto err_state;
+
+	return chan;
+
+err_state:
+	debugfs_remove_recursive(chan);
+	return NULL;
+}
+
+static void pxad_init_debugfs(struct pxad_device *pdev)
+{
+	int i;
+	struct dentry *chandir;
+
+	pdev->dbgfs_root = debugfs_create_dir(dev_name(pdev->slave.dev), NULL);
+	if (IS_ERR(pdev->dbgfs_root) || !pdev->dbgfs_root)
+		goto err_root;
+
+	pdev->dbgfs_state = debugfs_create_file("state", 0400, pdev->dbgfs_root,
+						pdev, &dbg_fops_state);
+	if (!pdev->dbgfs_state)
+		goto err_state;
+
+	pdev->dbgfs_chan =
+		kmalloc_array(pdev->nr_chans, sizeof(*pdev->dbgfs_state),
+			      GFP_KERNEL);
+	if (!pdev->dbgfs_chan)
+		goto err_alloc;
+
+	chandir = debugfs_create_dir("channels", pdev->dbgfs_root);
+	if (!chandir)
+		goto err_chandir;
+
+	for (i = 0; i < pdev->nr_chans; i++) {
+		pdev->dbgfs_chan[i] = pxad_dbg_alloc_chan(pdev, i, chandir);
+		if (!pdev->dbgfs_chan[i])
+			goto err_chans;
+	}
+
+	return;
+err_chans:
+err_chandir:
+	kfree(pdev->dbgfs_chan);
+err_alloc:
+err_state:
+	debugfs_remove_recursive(pdev->dbgfs_root);
+err_root:
+	pr_err("pxad: debugfs is not available\n");
+}
+
+static void pxad_cleanup_debugfs(struct pxad_device *pdev)
+{
+	debugfs_remove_recursive(pdev->dbgfs_root);
+}
+#else
+static inline void pxad_init_debugfs(struct pxad_device *pdev) {}
+static inline void pxad_cleanup_debugfs(struct pxad_device *pdev) {}
+#endif
+
+/*
+ * In the transition phase where legacy pxa handling is done at the same time as
+ * mmp_dma, the DMA physical channel split between the 2 DMA providers is done
+ * through legacy_reserved. Legacy code reserves DMA channels by settings
+ * corresponding bits in legacy_reserved.
+ */
+static u32 legacy_reserved;
+static u32 legacy_unavailable;
+
+static struct pxad_phy *lookup_phy(struct pxad_chan *pchan)
+{
+	int prio, i;
+	struct pxad_device *pdev = to_pxad_dev(pchan->vc.chan.device);
+	struct pxad_phy *phy, *found = NULL;
+	unsigned long flags;
+
+	/*
+	 * dma channel priorities
+	 * ch 0 - 3,  16 - 19  <--> (0)
+	 * ch 4 - 7,  20 - 23  <--> (1)
+	 * ch 8 - 11, 24 - 27  <--> (2)
+	 * ch 12 - 15, 28 - 31  <--> (3)
+	 */
+
+	spin_lock_irqsave(&pdev->phy_lock, flags);
+	for (prio = pchan->prio; prio >= PXAD_PRIO_HIGHEST; prio--) {
+		for (i = 0; i < pdev->nr_chans; i++) {
+			if (prio != (i & 0xf) >> 2)
+				continue;
+			if ((i < 32) && (legacy_reserved & BIT(i)))
+				continue;
+			phy = &pdev->phys[i];
+			if (!phy->vchan) {
+				phy->vchan = pchan;
+				found = phy;
+				if (i < 32)
+					legacy_unavailable |= BIT(i);
+				goto out_unlock;
+			}
+		}
+	}
+
+out_unlock:
+	spin_unlock_irqrestore(&pdev->phy_lock, flags);
+	dev_dbg(&pchan->vc.chan.dev->device,
+		"%s(): phy=%p(%d)\n", __func__, found,
+		found ? found->idx : -1);
+
+	return found;
+}
+
+static void pxad_free_phy(struct pxad_chan *chan)
+{
+	struct pxad_device *pdev = to_pxad_dev(chan->vc.chan.device);
+	unsigned long flags;
+	u32 reg;
+	int i;
+
+	dev_dbg(&chan->vc.chan.dev->device,
+		"%s(): freeing\n", __func__);
+	if (!chan->phy)
+		return;
+
+	/* clear the channel mapping in DRCMR */
+	reg = pxad_drcmr(chan->drcmr);
+	writel_relaxed(0, chan->phy->base + reg);
+
+	spin_lock_irqsave(&pdev->phy_lock, flags);
+	for (i = 0; i < 32; i++)
+		if (chan->phy == &pdev->phys[i])
+			legacy_unavailable &= ~BIT(i);
+	chan->phy->vchan = NULL;
+	chan->phy = NULL;
+	spin_unlock_irqrestore(&pdev->phy_lock, flags);
+}
+
+static bool is_chan_running(struct pxad_chan *chan)
+{
+	u32 dcsr;
+	struct pxad_phy *phy = chan->phy;
+
+	if (!phy)
+		return false;
+	dcsr = phy_readl_relaxed(phy, DCSR);
+	return dcsr & PXA_DCSR_RUN;
+}
+
+static bool is_running_chan_misaligned(struct pxad_chan *chan)
+{
+	u32 dalgn;
+
+	BUG_ON(!chan->phy);
+	dalgn = phy_readl_relaxed(chan->phy, DALGN);
+	return dalgn & (BIT(chan->phy->idx));
+}
+
+static void phy_enable(struct pxad_phy *phy, bool misaligned)
+{
+	u32 reg, dalgn;
+
+	if (!phy->vchan)
+		return;
+
+	dev_dbg(&phy->vchan->vc.chan.dev->device,
+		"%s(); phy=%p(%d) misaligned=%d\n", __func__,
+		phy, phy->idx, misaligned);
+
+	reg = pxad_drcmr(phy->vchan->drcmr);
+	writel_relaxed(DRCMR_MAPVLD | phy->idx, phy->base + reg);
+
+	dalgn = phy_readl_relaxed(phy, DALGN);
+	if (misaligned)
+		dalgn |= BIT(phy->idx);
+	else
+		dalgn &= ~BIT(phy->idx);
+	phy_writel_relaxed(phy, dalgn, DALGN);
+
+	phy_writel(phy, PXA_DCSR_STOPIRQEN | PXA_DCSR_ENDINTR |
+		   PXA_DCSR_BUSERR | PXA_DCSR_RUN, DCSR);
+}
+
+static void phy_disable(struct pxad_phy *phy)
+{
+	u32 dcsr;
+
+	if (!phy)
+		return;
+
+	dcsr = phy_readl_relaxed(phy, DCSR);
+	dev_dbg(&phy->vchan->vc.chan.dev->device,
+		"%s(): phy=%p(%d)\n", __func__, phy, phy->idx);
+	phy_writel(phy, dcsr & ~PXA_DCSR_RUN & ~PXA_DCSR_STOPIRQEN, DCSR);
+}
+
+static void pxad_launch_chan(struct pxad_chan *chan,
+				 struct pxad_desc_sw *desc)
+{
+	dev_dbg(&chan->vc.chan.dev->device,
+		"%s(): desc=%p\n", __func__, desc);
+	if (!chan->phy) {
+		chan->phy = lookup_phy(chan);
+		if (!chan->phy) {
+			dev_dbg(&chan->vc.chan.dev->device,
+				"%s(): no free dma channel\n", __func__);
+			return;
+		}
+	}
+
+	/*
+	 * Program the descriptor's address into the DMA controller,
+	 * then start the DMA transaction
+	 */
+	phy_writel(chan->phy, desc->first, DDADR);
+	phy_enable(chan->phy, chan->misaligned);
+}
+
+static void set_updater_desc(struct pxad_desc_sw *sw_desc,
+			     unsigned long flags)
+{
+	struct pxad_desc_hw *updater =
+		sw_desc->hw_desc[sw_desc->nb_desc - 1];
+	dma_addr_t dma = sw_desc->hw_desc[sw_desc->nb_desc - 2]->ddadr;
+
+	updater->ddadr = DDADR_STOP;
+	updater->dsadr = dma;
+	updater->dtadr = dma + 8;
+	updater->dcmd = PXA_DCMD_WIDTH4 | PXA_DCMD_BURST32 |
+		(PXA_DCMD_LENGTH & sizeof(u32));
+	if (flags & DMA_PREP_INTERRUPT)
+		updater->dcmd |= PXA_DCMD_ENDIRQEN;
+}
+
+static bool is_desc_completed(struct virt_dma_desc *vd)
+{
+	struct pxad_desc_sw *sw_desc = to_pxad_sw_desc(vd);
+	struct pxad_desc_hw *updater =
+		sw_desc->hw_desc[sw_desc->nb_desc - 1];
+
+	return updater->dtadr != (updater->dsadr + 8);
+}
+
+static void pxad_desc_chain(struct virt_dma_desc *vd1,
+				struct virt_dma_desc *vd2)
+{
+	struct pxad_desc_sw *desc1 = to_pxad_sw_desc(vd1);
+	struct pxad_desc_sw *desc2 = to_pxad_sw_desc(vd2);
+	dma_addr_t dma_to_chain;
+
+	dma_to_chain = desc2->first;
+	desc1->hw_desc[desc1->nb_desc - 1]->ddadr = dma_to_chain;
+}
+
+static bool pxad_try_hotchain(struct virt_dma_chan *vc,
+				  struct virt_dma_desc *vd)
+{
+	struct virt_dma_desc *vd_last_issued = NULL;
+	struct pxad_chan *chan = to_pxad_chan(&vc->chan);
+
+	/*
+	 * Attempt to hot chain the tx if the phy is still running. This is
+	 * considered successful only if either the channel is still running
+	 * after the chaining, or if the chained transfer is completed after
+	 * having been hot chained.
+	 * A change of alignment is not allowed, and forbids hotchaining.
+	 */
+	if (is_chan_running(chan)) {
+		BUG_ON(list_empty(&vc->desc_issued));
+
+		if (!is_running_chan_misaligned(chan) &&
+		    to_pxad_sw_desc(vd)->misaligned)
+			return false;
+
+		vd_last_issued = list_entry(vc->desc_issued.prev,
+					    struct virt_dma_desc, node);
+		pxad_desc_chain(vd_last_issued, vd);
+		if (is_chan_running(chan) || is_desc_completed(vd_last_issued))
+			return true;
+	}
+
+	return false;
+}
+
+static unsigned int clear_chan_irq(struct pxad_phy *phy)
+{
+	u32 dcsr;
+	u32 dint = readl(phy->base + DINT);
+
+	if (!(dint & BIT(phy->idx)))
+		return PXA_DCSR_RUN;
+
+	/* clear irq */
+	dcsr = phy_readl_relaxed(phy, DCSR);
+	phy_writel(phy, dcsr, DCSR);
+	if ((dcsr & PXA_DCSR_BUSERR) && (phy->vchan))
+		dev_warn(&phy->vchan->vc.chan.dev->device,
+			 "%s(chan=%p): PXA_DCSR_BUSERR\n",
+			 __func__, &phy->vchan);
+
+	return dcsr & ~PXA_DCSR_RUN;
+}
+
+static irqreturn_t pxad_chan_handler(int irq, void *dev_id)
+{
+	struct pxad_phy *phy = dev_id;
+	struct pxad_chan *chan = phy->vchan;
+	struct virt_dma_desc *vd, *tmp;
+	unsigned int dcsr;
+	unsigned long flags;
+
+	BUG_ON(!chan);
+
+	dcsr = clear_chan_irq(phy);
+	if (dcsr & PXA_DCSR_RUN)
+		return IRQ_NONE;
+
+	spin_lock_irqsave(&chan->vc.lock, flags);
+	list_for_each_entry_safe(vd, tmp, &chan->vc.desc_issued, node) {
+		dev_dbg(&chan->vc.chan.dev->device,
+			"%s(): checking txd %p[%x]: completed=%d\n",
+			__func__, vd, vd->tx.cookie, is_desc_completed(vd));
+		if (is_desc_completed(vd)) {
+			list_del(&vd->node);
+			vchan_cookie_complete(vd);
+		} else {
+			break;
+		}
+	}
+
+	if (dcsr & PXA_DCSR_STOPSTATE) {
+		dev_dbg(&chan->vc.chan.dev->device,
+		"%s(): channel stopped, submitted_empty=%d issued_empty=%d",
+			__func__,
+			list_empty(&chan->vc.desc_submitted),
+			list_empty(&chan->vc.desc_issued));
+		phy_writel_relaxed(phy, dcsr & ~PXA_DCSR_STOPIRQEN, DCSR);
+
+		if (list_empty(&chan->vc.desc_issued)) {
+			chan->misaligned =
+				!list_empty(&chan->vc.desc_submitted);
+		} else {
+			vd = list_first_entry(&chan->vc.desc_issued,
+					      struct virt_dma_desc, node);
+			pxad_launch_chan(chan, to_pxad_sw_desc(vd));
+		}
+	}
+	spin_unlock_irqrestore(&chan->vc.lock, flags);
+
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t pxad_int_handler(int irq, void *dev_id)
+{
+	struct pxad_device *pdev = dev_id;
+	struct pxad_phy *phy;
+	u32 dint = readl(pdev->base + DINT);
+	int i, ret = IRQ_NONE;
+
+	while (dint) {
+		i = __ffs(dint);
+		dint &= (dint - 1);
+		phy = &pdev->phys[i];
+		if ((i < 32) && (legacy_reserved & BIT(i)))
+			continue;
+		if (pxad_chan_handler(irq, phy) == IRQ_HANDLED)
+			ret = IRQ_HANDLED;
+	}
+
+	return ret;
+}
+
+static int pxad_alloc_chan_resources(struct dma_chan *dchan)
+{
+	struct pxad_chan *chan = to_pxad_chan(dchan);
+	struct pxad_device *pdev = to_pxad_dev(chan->vc.chan.device);
+
+	if (chan->desc_pool)
+		return 1;
+
+	chan->desc_pool = dma_pool_create(dma_chan_name(dchan),
+					  pdev->slave.dev,
+					  sizeof(struct pxad_desc_hw),
+					  __alignof__(struct pxad_desc_hw),
+					  0);
+	if (!chan->desc_pool) {
+		dev_err(&chan->vc.chan.dev->device,
+			"%s(): unable to allocate descriptor pool\n",
+			__func__);
+		return -ENOMEM;
+	}
+
+	return 1;
+}
+
+static void pxad_free_chan_resources(struct dma_chan *dchan)
+{
+	struct pxad_chan *chan = to_pxad_chan(dchan);
+
+	vchan_free_chan_resources(&chan->vc);
+	dma_pool_destroy(chan->desc_pool);
+	chan->desc_pool = NULL;
+
+}
+
+static void pxad_free_desc(struct virt_dma_desc *vd)
+{
+	int i;
+	dma_addr_t dma;
+	struct pxad_desc_sw *sw_desc = to_pxad_sw_desc(vd);
+
+	BUG_ON(sw_desc->nb_desc == 0);
+	for (i = sw_desc->nb_desc - 1; i >= 0; i--) {
+		if (i > 0)
+			dma = sw_desc->hw_desc[i - 1]->ddadr;
+		else
+			dma = sw_desc->first;
+		dma_pool_free(sw_desc->desc_pool,
+			      sw_desc->hw_desc[i], dma);
+	}
+	sw_desc->nb_desc = 0;
+	kfree(sw_desc);
+}
+
+static struct pxad_desc_sw *
+pxad_alloc_desc(struct pxad_chan *chan, unsigned int nb_hw_desc)
+{
+	struct pxad_desc_sw *sw_desc;
+	dma_addr_t dma;
+	int i;
+
+	sw_desc = kzalloc(sizeof(*sw_desc) +
+			  nb_hw_desc