DMAENGINE: Support for ST-Ericssons DMA40 block v3
authorLinus Walleij <linus.walleij@stericsson.com>
Tue, 30 Mar 2010 13:33:42 +0000 (15:33 +0200)
committerDan Williams <dan.j.williams@intel.com>
Wed, 14 Apr 2010 21:49:20 +0000 (14:49 -0700)
This is a straightforward driver for the ST-Ericsson DMA40 DMA
controller found in U8500, implemented akin to the existing
COH 901 318 driver.

Signed-off-by: Linus Walleij <linus.walleij@stericsson.com>
Acked-by: Srinidh Kasagar <srinidhi.kasagar@stericsson.com>
Cc: STEricsson_nomadik_linux@list.st.com
Cc: Alessandro Rubini <rubini@unipv.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
arch/arm/plat-nomadik/include/plat/ste_dma40.h [new file with mode: 0644]
drivers/dma/Kconfig
drivers/dma/Makefile
drivers/dma/ste_dma40.c [new file with mode: 0644]
drivers/dma/ste_dma40_ll.c [new file with mode: 0644]
drivers/dma/ste_dma40_ll.h [new file with mode: 0644]

diff --git a/arch/arm/plat-nomadik/include/plat/ste_dma40.h b/arch/arm/plat-nomadik/include/plat/ste_dma40.h
new file mode 100644 (file)
index 0000000..4d12ea4
--- /dev/null
@@ -0,0 +1,239 @@
+/*
+ * arch/arm/plat-nomadik/include/plat/ste_dma40.h
+ *
+ * Copyright (C) ST-Ericsson 2007-2010
+ * License terms: GNU General Public License (GPL) version 2
+ * Author: Per Friden <per.friden@stericsson.com>
+ * Author: Jonas Aaberg <jonas.aberg@stericsson.com>
+ */
+
+
+#ifndef STE_DMA40_H
+#define STE_DMA40_H
+
+#include <linux/dmaengine.h>
+#include <linux/workqueue.h>
+#include <linux/interrupt.h>
+#include <linux/dmaengine.h>
+
+/* dev types for memcpy */
+#define STEDMA40_DEV_DST_MEMORY (-1)
+#define        STEDMA40_DEV_SRC_MEMORY (-1)
+
+/*
+ * Description of bitfields of channel_type variable is available in
+ * the info structure.
+ */
+
+/* Priority */
+#define STEDMA40_INFO_PRIO_TYPE_POS 2
+#define STEDMA40_HIGH_PRIORITY_CHANNEL (0x1 << STEDMA40_INFO_PRIO_TYPE_POS)
+#define STEDMA40_LOW_PRIORITY_CHANNEL (0x2 << STEDMA40_INFO_PRIO_TYPE_POS)
+
+/* Mode  */
+#define STEDMA40_INFO_CH_MODE_TYPE_POS 6
+#define STEDMA40_CHANNEL_IN_PHY_MODE (0x1 << STEDMA40_INFO_CH_MODE_TYPE_POS)
+#define STEDMA40_CHANNEL_IN_LOG_MODE (0x2 << STEDMA40_INFO_CH_MODE_TYPE_POS)
+#define STEDMA40_CHANNEL_IN_OPER_MODE (0x3 << STEDMA40_INFO_CH_MODE_TYPE_POS)
+
+/* Mode options */
+#define STEDMA40_INFO_CH_MODE_OPT_POS 8
+#define STEDMA40_PCHAN_BASIC_MODE (0x1 << STEDMA40_INFO_CH_MODE_OPT_POS)
+#define STEDMA40_PCHAN_MODULO_MODE (0x2 << STEDMA40_INFO_CH_MODE_OPT_POS)
+#define STEDMA40_PCHAN_DOUBLE_DST_MODE (0x3 << STEDMA40_INFO_CH_MODE_OPT_POS)
+#define STEDMA40_LCHAN_SRC_PHY_DST_LOG (0x1 << STEDMA40_INFO_CH_MODE_OPT_POS)
+#define STEDMA40_LCHAN_SRC_LOG_DST_PHS (0x2 << STEDMA40_INFO_CH_MODE_OPT_POS)
+#define STEDMA40_LCHAN_SRC_LOG_DST_LOG (0x3 << STEDMA40_INFO_CH_MODE_OPT_POS)
+
+/* Interrupt */
+#define STEDMA40_INFO_TIM_POS 10
+#define STEDMA40_NO_TIM_FOR_LINK (0x0 << STEDMA40_INFO_TIM_POS)
+#define STEDMA40_TIM_FOR_LINK (0x1 << STEDMA40_INFO_TIM_POS)
+
+/* End of channel_type configuration */
+
+#define STEDMA40_ESIZE_8_BIT  0x0
+#define STEDMA40_ESIZE_16_BIT 0x1
+#define STEDMA40_ESIZE_32_BIT 0x2
+#define STEDMA40_ESIZE_64_BIT 0x3
+
+/* The value 4 indicates that PEN-reg shall be set to 0 */
+#define STEDMA40_PSIZE_PHY_1  0x4
+#define STEDMA40_PSIZE_PHY_2  0x0
+#define STEDMA40_PSIZE_PHY_4  0x1
+#define STEDMA40_PSIZE_PHY_8  0x2
+#define STEDMA40_PSIZE_PHY_16 0x3
+
+/*
+ * The number of elements differ in logical and
+ * physical mode
+ */
+#define STEDMA40_PSIZE_LOG_1  STEDMA40_PSIZE_PHY_2
+#define STEDMA40_PSIZE_LOG_4  STEDMA40_PSIZE_PHY_4
+#define STEDMA40_PSIZE_LOG_8  STEDMA40_PSIZE_PHY_8
+#define STEDMA40_PSIZE_LOG_16 STEDMA40_PSIZE_PHY_16
+
+enum stedma40_flow_ctrl {
+       STEDMA40_NO_FLOW_CTRL,
+       STEDMA40_FLOW_CTRL,
+};
+
+enum stedma40_endianess {
+       STEDMA40_LITTLE_ENDIAN,
+       STEDMA40_BIG_ENDIAN
+};
+
+enum stedma40_periph_data_width {
+       STEDMA40_BYTE_WIDTH = STEDMA40_ESIZE_8_BIT,
+       STEDMA40_HALFWORD_WIDTH = STEDMA40_ESIZE_16_BIT,
+       STEDMA40_WORD_WIDTH = STEDMA40_ESIZE_32_BIT,
+       STEDMA40_DOUBLEWORD_WIDTH = STEDMA40_ESIZE_64_BIT
+};
+
+struct stedma40_half_channel_info {
+       enum stedma40_endianess endianess;
+       enum stedma40_periph_data_width data_width;
+       int psize;
+       enum stedma40_flow_ctrl flow_ctrl;
+};
+
+enum stedma40_xfer_dir {
+       STEDMA40_MEM_TO_MEM,
+       STEDMA40_MEM_TO_PERIPH,
+       STEDMA40_PERIPH_TO_MEM,
+       STEDMA40_PERIPH_TO_PERIPH
+};
+
+
+/**
+ * struct stedma40_chan_cfg - Structure to be filled by client drivers.
+ *
+ * @dir: MEM 2 MEM, PERIPH 2 MEM , MEM 2 PERIPH, PERIPH 2 PERIPH
+ * @channel_type: priority, mode, mode options and interrupt configuration.
+ * @src_dev_type: Src device type
+ * @dst_dev_type: Dst device type
+ * @src_info: Parameters for dst half channel
+ * @dst_info: Parameters for dst half channel
+ * @pre_transfer_data: Data to be passed on to the pre_transfer() function.
+ * @pre_transfer: Callback used if needed before preparation of transfer.
+ * Only called if device is set. size of bytes to transfer
+ * (in case of multiple element transfer size is size of the first element).
+ *
+ *
+ * This structure has to be filled by the client drivers.
+ * It is recommended to do all dma configurations for clients in the machine.
+ *
+ */
+struct stedma40_chan_cfg {
+       enum stedma40_xfer_dir                   dir;
+       unsigned int                             channel_type;
+       int                                      src_dev_type;
+       int                                      dst_dev_type;
+       struct stedma40_half_channel_info        src_info;
+       struct stedma40_half_channel_info        dst_info;
+       void                                    *pre_transfer_data;
+       int (*pre_transfer)                     (struct dma_chan *chan,
+                                                void *data,
+                                                int size);
+};
+
+/**
+ * struct stedma40_platform_data - Configuration struct for the dma device.
+ *
+ * @dev_len: length of dev_tx and dev_rx
+ * @dev_tx: mapping between destination event line and io address
+ * @dev_rx: mapping between source event line and io address
+ * @memcpy: list of memcpy event lines
+ * @memcpy_len: length of memcpy
+ * @memcpy_conf_phy: default configuration of physical channel memcpy
+ * @memcpy_conf_log: default configuration of logical channel memcpy
+ * @llis_per_log: number of max linked list items per logical channel
+ *
+ */
+struct stedma40_platform_data {
+       u32                              dev_len;
+       const dma_addr_t                *dev_tx;
+       const dma_addr_t                *dev_rx;
+       int                             *memcpy;
+       u32                              memcpy_len;
+       struct stedma40_chan_cfg        *memcpy_conf_phy;
+       struct stedma40_chan_cfg        *memcpy_conf_log;
+       unsigned int                     llis_per_log;
+};
+
+/**
+ * setdma40_set_psize() - Used for changing the package size of an
+ * already configured dma channel.
+ *
+ * @chan: dmaengine handle
+ * @src_psize: new package side for src. (STEDMA40_PSIZE*)
+ * @src_psize: new package side for dst. (STEDMA40_PSIZE*)
+ *
+ * returns 0 on ok, otherwise negative error number.
+ */
+int stedma40_set_psize(struct dma_chan *chan,
+                      int src_psize,
+                      int dst_psize);
+
+/**
+ * stedma40_filter() - Provides stedma40_chan_cfg to the
+ * ste_dma40 dma driver via the dmaengine framework.
+ * does some checking of what's provided.
+ *
+ * Never directly called by client. It used by dmaengine.
+ * @chan: dmaengine handle.
+ * @data: Must be of type: struct stedma40_chan_cfg and is
+ * the configuration of the framework.
+ *
+ *
+ */
+
+bool stedma40_filter(struct dma_chan *chan, void *data);
+
+/**
+ * stedma40_memcpy_sg() - extension of the dma framework, memcpy to/from
+ * scattergatter lists.
+ *
+ * @chan: dmaengine handle
+ * @sgl_dst: Destination scatter list
+ * @sgl_src: Source scatter list
+ * @sgl_len: The length of each scatterlist. Both lists must be of equal length
+ * and each element must match the corresponding element in the other scatter
+ * list.
+ * @flags: is actually enum dma_ctrl_flags. See dmaengine.h
+ */
+
+struct dma_async_tx_descriptor *stedma40_memcpy_sg(struct dma_chan *chan,
+                                                  struct scatterlist *sgl_dst,
+                                                  struct scatterlist *sgl_src,
+                                                  unsigned int sgl_len,
+                                                  unsigned long flags);
+
+/**
+ * stedma40_slave_mem() - Transfers a raw data buffer to or from a slave
+ * (=device)
+ *
+ * @chan: dmaengine handle
+ * @addr: source or destination physicall address.
+ * @size: bytes to transfer
+ * @direction: direction of transfer
+ * @flags: is actually enum dma_ctrl_flags. See dmaengine.h
+ */
+
+static inline struct
+dma_async_tx_descriptor *stedma40_slave_mem(struct dma_chan *chan,
+                                           dma_addr_t addr,
+                                           unsigned int size,
+                                           enum dma_data_direction direction,
+                                           unsigned long flags)
+{
+       struct scatterlist sg;
+       sg_init_table(&sg, 1);
+       sg.dma_address = addr;
+       sg.length = size;
+
+       return chan->device->device_prep_slave_sg(chan, &sg, 1,
+                                                 direction, flags);
+}
+
+#endif
index a2fcb2e..1b88779 100644 (file)
@@ -141,6 +141,13 @@ config COH901318
        help
          Enable support for ST-Ericsson COH 901 318 DMA.
 
+config STE_DMA40
+       bool "ST-Ericsson DMA40 support"
+       depends on ARCH_U8500
+       select DMA_ENGINE
+       help
+         Support for ST-Ericsson DMA40 controller
+
 config AMCC_PPC440SPE_ADMA
        tristate "AMCC PPC440SPe ADMA support"
        depends on 440SPe || 440SP
index 40c627d..2088142 100644 (file)
@@ -21,3 +21,4 @@ obj-$(CONFIG_SH_DMAE) += shdma.o
 obj-$(CONFIG_COH901318) += coh901318.o coh901318_lli.o
 obj-$(CONFIG_AMCC_PPC440SPE_ADMA) += ppc4xx/
 obj-$(CONFIG_TIMB_DMA) += timb_dma.o
+obj-$(CONFIG_STE_DMA40) += ste_dma40.o ste_dma40_ll.o
diff --git a/drivers/dma/ste_dma40.c b/drivers/dma/ste_dma40.c
new file mode 100644 (file)
index 0000000..e4295a2
--- /dev/null
@@ -0,0 +1,2596 @@
+/*
+ * driver/dma/ste_dma40.c
+ *
+ * Copyright (C) ST-Ericsson 2007-2010
+ * License terms: GNU General Public License (GPL) version 2
+ * Author: Per Friden <per.friden@stericsson.com>
+ * Author: Jonas Aaberg <jonas.aberg@stericsson.com>
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/dmaengine.h>
+#include <linux/platform_device.h>
+#include <linux/clk.h>
+#include <linux/delay.h>
+
+#include <plat/ste_dma40.h>
+
+#include "ste_dma40_ll.h"
+
+#define D40_NAME "dma40"
+
+#define D40_PHY_CHAN -1
+
+/* For masking out/in 2 bit channel positions */
+#define D40_CHAN_POS(chan)  (2 * (chan / 2))
+#define D40_CHAN_POS_MASK(chan) (0x3 << D40_CHAN_POS(chan))
+
+/* Maximum iterations taken before giving up suspending a channel */
+#define D40_SUSPEND_MAX_IT 500
+
+#define D40_ALLOC_FREE         (1 << 31)
+#define D40_ALLOC_PHY          (1 << 30)
+#define D40_ALLOC_LOG_FREE     0
+
+/* The number of free d40_desc to keep in memory before starting
+ * to kfree() them */
+#define D40_DESC_CACHE_SIZE 50
+
+/* Hardware designer of the block */
+#define D40_PERIPHID2_DESIGNER 0x8
+
+/**
+ * enum 40_command - The different commands and/or statuses.
+ *
+ * @D40_DMA_STOP: DMA channel command STOP or status STOPPED,
+ * @D40_DMA_RUN: The DMA channel is RUNNING of the command RUN.
+ * @D40_DMA_SUSPEND_REQ: Request the DMA to SUSPEND as soon as possible.
+ * @D40_DMA_SUSPENDED: The DMA channel is SUSPENDED.
+ */
+enum d40_command {
+       D40_DMA_STOP            = 0,
+       D40_DMA_RUN             = 1,
+       D40_DMA_SUSPEND_REQ     = 2,
+       D40_DMA_SUSPENDED       = 3
+};
+
+/**
+ * struct d40_lli_pool - Structure for keeping LLIs in memory
+ *
+ * @base: Pointer to memory area when the pre_alloc_lli's are not large
+ * enough, IE bigger than the most common case, 1 dst and 1 src. NULL if
+ * pre_alloc_lli is used.
+ * @size: The size in bytes of the memory at base or the size of pre_alloc_lli.
+ * @pre_alloc_lli: Pre allocated area for the most common case of transfers,
+ * one buffer to one buffer.
+ */
+struct d40_lli_pool {
+       void    *base;
+       int     size;
+       /* Space for dst and src, plus an extra for padding */
+       u8      pre_alloc_lli[3 * sizeof(struct d40_phy_lli)];
+};
+
+/**
+ * struct d40_desc - A descriptor is one DMA job.
+ *
+ * @lli_phy: LLI settings for physical channel. Both src and dst=
+ * points into the lli_pool, to base if lli_len > 1 or to pre_alloc_lli if
+ * lli_len equals one.
+ * @lli_log: Same as above but for logical channels.
+ * @lli_pool: The pool with two entries pre-allocated.
+ * @lli_len: Number of LLI's in lli_pool
+ * @lli_tcount: Number of LLIs processed in the transfer. When equals lli_len
+ * then this transfer job is done.
+ * @txd: DMA engine struct. Used for among other things for communication
+ * during a transfer.
+ * @node: List entry.
+ * @dir: The transfer direction of this job.
+ * @is_in_client_list: true if the client owns this descriptor.
+ *
+ * This descriptor is used for both logical and physical transfers.
+ */
+
+struct d40_desc {
+       /* LLI physical */
+       struct d40_phy_lli_bidir         lli_phy;
+       /* LLI logical */
+       struct d40_log_lli_bidir         lli_log;
+
+       struct d40_lli_pool              lli_pool;
+       u32                              lli_len;
+       u32                              lli_tcount;
+
+       struct dma_async_tx_descriptor   txd;
+       struct list_head                 node;
+
+       enum dma_data_direction          dir;
+       bool                             is_in_client_list;
+};
+
+/**
+ * struct d40_lcla_pool - LCLA pool settings and data.
+ *
+ * @base: The virtual address of LCLA.
+ * @phy: Physical base address of LCLA.
+ * @base_size: size of lcla.
+ * @lock: Lock to protect the content in this struct.
+ * @alloc_map: Mapping between physical channel and LCLA entries.
+ * @num_blocks: The number of entries of alloc_map. Equals to the
+ * number of physical channels.
+ */
+struct d40_lcla_pool {
+       void            *base;
+       dma_addr_t       phy;
+       resource_size_t  base_size;
+       spinlock_t       lock;
+       u32             *alloc_map;
+       int              num_blocks;
+};
+
+/**
+ * struct d40_phy_res - struct for handling eventlines mapped to physical
+ * channels.
+ *
+ * @lock: A lock protection this entity.
+ * @num: The physical channel number of this entity.
+ * @allocated_src: Bit mapped to show which src event line's are mapped to
+ * this physical channel. Can also be free or physically allocated.
+ * @allocated_dst: Same as for src but is dst.
+ * allocated_dst and allocated_src uses the D40_ALLOC* defines as well as
+ * event line number. Both allocated_src and allocated_dst can not be
+ * allocated to a physical channel, since the interrupt handler has then
+ * no way of figure out which one the interrupt belongs to.
+ */
+struct d40_phy_res {
+       spinlock_t lock;
+       int        num;
+       u32        allocated_src;
+       u32        allocated_dst;
+};
+
+struct d40_base;
+
+/**
+ * struct d40_chan - Struct that describes a channel.
+ *
+ * @lock: A spinlock to protect this struct.
+ * @log_num: The logical number, if any of this channel.
+ * @completed: Starts with 1, after first interrupt it is set to dma engine's
+ * current cookie.
+ * @pending_tx: The number of pending transfers. Used between interrupt handler
+ * and tasklet.
+ * @busy: Set to true when transfer is ongoing on this channel.
+ * @phy_chan: Pointer to physical channel which this instance runs on.
+ * @chan: DMA engine handle.
+ * @tasklet: Tasklet that gets scheduled from interrupt context to complete a
+ * transfer and call client callback.
+ * @client: Cliented owned descriptor list.
+ * @active: Active descriptor.
+ * @queue: Queued jobs.
+ * @free: List of free descripts, ready to be reused.
+ * @free_len: Number of descriptors in the free list.
+ * @dma_cfg: The client configuration of this dma channel.
+ * @base: Pointer to the device instance struct.
+ * @src_def_cfg: Default cfg register setting for src.
+ * @dst_def_cfg: Default cfg register setting for dst.
+ * @log_def: Default logical channel settings.
+ * @lcla: Space for one dst src pair for logical channel transfers.
+ * @lcpa: Pointer to dst and src lcpa settings.
+ *
+ * This struct can either "be" a logical or a physical channel.
+ */
+struct d40_chan {
+       spinlock_t                       lock;
+       int                              log_num;
+       /* ID of the most recent completed transfer */
+       int                              completed;
+       int                              pending_tx;
+       bool                             busy;
+       struct d40_phy_res              *phy_chan;
+       struct dma_chan                  chan;
+       struct tasklet_struct            tasklet;
+       struct list_head                 client;
+       struct list_head                 active;
+       struct list_head                 queue;
+       struct list_head                 free;
+       int                              free_len;
+       struct stedma40_chan_cfg         dma_cfg;
+       struct d40_base                 *base;
+       /* Default register configurations */
+       u32                              src_def_cfg;
+       u32                              dst_def_cfg;
+       struct d40_def_lcsp              log_def;
+       struct d40_lcla_elem             lcla;
+       struct d40_log_lli_full         *lcpa;
+};
+
+/**
+ * struct d40_base - The big global struct, one for each probe'd instance.
+ *
+ * @interrupt_lock: Lock used to make sure one interrupt is handle a time.
+ * @execmd_lock: Lock for execute command usage since several channels share
+ * the same physical register.
+ * @dev: The device structure.
+ * @virtbase: The virtual base address of the DMA's register.
+ * @clk: Pointer to the DMA clock structure.
+ * @phy_start: Physical memory start of the DMA registers.
+ * @phy_size: Size of the DMA register map.
+ * @irq: The IRQ number.
+ * @num_phy_chans: The number of physical channels. Read from HW. This
+ * is the number of available channels for this driver, not counting "Secure
+ * mode" allocated physical channels.
+ * @num_log_chans: The number of logical channels. Calculated from
+ * num_phy_chans.
+ * @dma_both: dma_device channels that can do both memcpy and slave transfers.
+ * @dma_slave: dma_device channels that can do only do slave transfers.
+ * @dma_memcpy: dma_device channels that can do only do memcpy transfers.
+ * @phy_chans: Room for all possible physical channels in system.
+ * @log_chans: Room for all possible logical channels in system.
+ * @lookup_log_chans: Used to map interrupt number to logical channel. Points
+ * to log_chans entries.
+ * @lookup_phy_chans: Used to map interrupt number to physical channel. Points
+ * to phy_chans entries.
+ * @plat_data: Pointer to provided platform_data which is the driver
+ * configuration.
+ * @phy_res: Vector containing all physical channels.
+ * @lcla_pool: lcla pool settings and data.
+ * @lcpa_base: The virtual mapped address of LCPA.
+ * @phy_lcpa: The physical address of the LCPA.
+ * @lcpa_size: The size of the LCPA area.
+ */
+struct d40_base {
+       spinlock_t                       interrupt_lock;
+       spinlock_t                       execmd_lock;
+       struct device                    *dev;
+       void __iomem                     *virtbase;
+       struct clk                       *clk;
+       phys_addr_t                       phy_start;
+       resource_size_t                   phy_size;
+       int                               irq;
+       int                               num_phy_chans;
+       int                               num_log_chans;
+       struct dma_device                 dma_both;
+       struct dma_device                 dma_slave;
+       struct dma_device                 dma_memcpy;
+       struct d40_chan                  *phy_chans;
+       struct d40_chan                  *log_chans;
+       struct d40_chan                 **lookup_log_chans;
+       struct d40_chan                 **lookup_phy_chans;
+       struct stedma40_platform_data    *plat_data;
+       /* Physical half channels */
+       struct d40_phy_res               *phy_res;
+       struct d40_lcla_pool              lcla_pool;
+       void                             *lcpa_base;
+       dma_addr_t                        phy_lcpa;
+       resource_size_t                   lcpa_size;
+};
+
+/**
+ * struct d40_interrupt_lookup - lookup table for interrupt handler
+ *
+ * @src: Interrupt mask register.
+ * @clr: Interrupt clear register.
+ * @is_error: true if this is an error interrupt.
+ * @offset: start delta in the lookup_log_chans in d40_base. If equals to
+ * D40_PHY_CHAN, the lookup_phy_chans shall be used instead.
+ */
+struct d40_interrupt_lookup {
+       u32 src;
+       u32 clr;
+       bool is_error;
+       int offset;
+};
+
+/**
+ * struct d40_reg_val - simple lookup struct
+ *
+ * @reg: The register.
+ * @val: The value that belongs to the register in reg.
+ */
+struct d40_reg_val {
+       unsigned int reg;
+       unsigned int val;
+};
+
+static int d40_pool_lli_alloc(struct d40_desc *d40d,
+                             int lli_len, bool is_log)
+{
+       u32 align;
+       void *base;
+
+       if (is_log)
+               align = sizeof(struct d40_log_lli);
+       else
+               align = sizeof(struct d40_phy_lli);
+
+       if (lli_len == 1) {
+               base = d40d->lli_pool.pre_alloc_lli;
+               d40d->lli_pool.size = sizeof(d40d->lli_pool.pre_alloc_lli);
+               d40d->lli_pool.base = NULL;
+       } else {
+               d40d->lli_pool.size = ALIGN(lli_len * 2 * align, align);
+
+               base = kmalloc(d40d->lli_pool.size + align, GFP_NOWAIT);
+               d40d->lli_pool.base = base;
+
+               if (d40d->lli_pool.base == NULL)
+                       return -ENOMEM;
+       }
+
+       if (is_log) {
+               d40d->lli_log.src = PTR_ALIGN((struct d40_log_lli *) base,
+                                             align);
+               d40d->lli_log.dst = PTR_ALIGN(d40d->lli_log.src + lli_len,
+                                             align);
+       } else {
+               d40d->lli_phy.src = PTR_ALIGN((struct d40_phy_lli *)base,
+                                             align);
+               d40d->lli_phy.dst = PTR_ALIGN(d40d->lli_phy.src + lli_len,
+                                             align);
+
+               d40d->lli_phy.src_addr = virt_to_phys(d40d->lli_phy.src);
+               d40d->lli_phy.dst_addr = virt_to_phys(d40d->lli_phy.dst);
+       }
+
+       return 0;
+}
+
+static void d40_pool_lli_free(struct d40_desc *d40d)
+{
+       kfree(d40d->lli_pool.base);
+       d40d->lli_pool.base = NULL;
+       d40d->lli_pool.size = 0;
+       d40d->lli_log.src = NULL;
+       d40d->lli_log.dst = NULL;
+       d40d->lli_phy.src = NULL;
+       d40d->lli_phy.dst = NULL;
+       d40d->lli_phy.src_addr = 0;
+       d40d->lli_phy.dst_addr = 0;
+}
+
+static dma_cookie_t d40_assign_cookie(struct d40_chan *d40c,
+                                     struct d40_desc *desc)
+{
+       dma_cookie_t cookie = d40c->chan.cookie;
+
+       if (++cookie < 0)
+               cookie = 1;
+
+       d40c->chan.cookie = cookie;
+       desc->txd.cookie = cookie;
+
+       return cookie;
+}
+
+static void d40_desc_reset(struct d40_desc *d40d)
+{
+       d40d->lli_tcount = 0;
+}
+
+static void d40_desc_remove(struct d40_desc *d40d)
+{
+       list_del(&d40d->node);
+}
+
+static struct d40_desc *d40_desc_get(struct d40_chan *d40c)
+{
+       struct d40_desc *desc;
+       struct d40_desc *d;
+       struct d40_desc *_d;
+
+       if (!list_empty(&d40c->client)) {
+               list_for_each_entry_safe(d, _d, &d40c->client, node)
+                       if (async_tx_test_ack(&d->txd)) {
+                               d40_pool_lli_free(d);
+                               d40_desc_remove(d);
+                               desc = d;
+                               goto out;
+                       }
+       }
+
+       if (list_empty(&d40c->free)) {
+               /* Alloc new desc because we're out of used ones */
+               desc = kzalloc(sizeof(struct d40_desc), GFP_NOWAIT);
+               if (desc == NULL)
+                       goto out;
+               INIT_LIST_HEAD(&desc->node);
+       } else {
+               /* Reuse an old desc. */
+               desc = list_first_entry(&d40c->free,
+                                       struct d40_desc,
+                                       node);
+               list_del(&desc->node);
+               d40c->free_len--;
+       }
+out:
+       return desc;
+}
+
+static void d40_desc_free(struct d40_chan *d40c, struct d40_desc *d40d)
+{
+       if (d40c->free_len < D40_DESC_CACHE_SIZE) {
+               list_add_tail(&d40d->node, &d40c->free);
+               d40c->free_len++;
+       } else
+               kfree(d40d);
+}
+
+static void d40_desc_submit(struct d40_chan *d40c, struct d40_desc *desc)
+{
+       list_add_tail(&desc->node, &d40c->active);
+}
+
+static struct d40_desc *d40_first_active_get(struct d40_chan *d40c)
+{
+       struct d40_desc *d;
+
+       if (list_empty(&d40c->active))
+               return NULL;
+
+       d = list_first_entry(&d40c->active,
+                            struct d40_desc,
+                            node);
+       return d;
+}
+
+static void d40_desc_queue(struct d40_chan *d40c, struct d40_desc *desc)
+{
+       list_add_tail(&desc->node, &d40c->queue);
+}
+
+static struct d40_desc *d40_first_queued(struct d40_chan *d40c)
+{
+       struct d40_desc *d;
+
+       if (list_empty(&d40c->queue))
+               return NULL;
+
+       d = list_first_entry(&d40c->queue,
+                            struct d40_desc,
+                            node);
+       return d;
+}
+
+/* Support functions for logical channels */
+
+static int d40_lcla_id_get(struct d40_chan *d40c,
+                          struct d40_lcla_pool *pool)
+{
+       int src_id = 0;
+       int dst_id = 0;
+       struct d40_log_lli *lcla_lidx_base =
+               pool->base + d40c->phy_chan->num * 1024;
+       int i;
+       int lli_per_log = d40c->base->plat_data->llis_per_log;
+
+       if (d40c->lcla.src_id >= 0 && d40c->lcla.dst_id >= 0)
+               return 0;
+
+       if (pool->num_blocks > 32)
+               return -EINVAL;
+
+       spin_lock(&pool->lock);
+
+       for (i = 0; i < pool->num_blocks; i++) {
+               if (!(pool->alloc_map[d40c->phy_chan->num] & (0x1 << i))) {
+                       pool->alloc_map[d40c->phy_chan->num] |= (0x1 << i);
+                       break;
+               }
+       }
+       src_id = i;
+       if (src_id >= pool->num_blocks)
+               goto err;
+
+       for (; i < pool->num_blocks; i++) {
+               if (!(pool->alloc_map[d40c->phy_chan->num] & (0x1 << i))) {
+                       pool->alloc_map[d40c->phy_chan->num] |= (0x1 << i);
+                       break;
+               }
+       }
+
+       dst_id = i;
+       if (dst_id == src_id)
+               goto err;
+
+       d40c->lcla.src_id = src_id;
+       d40c->lcla.dst_id = dst_id;
+       d40c->lcla.dst = lcla_lidx_base + dst_id * lli_per_log + 1;
+       d40c->lcla.src = lcla_lidx_base + src_id * lli_per_log + 1;
+
+
+       spin_unlock(&pool->lock);
+       return 0;
+err:
+       spin_unlock(&pool->lock);
+       return -EINVAL;
+}
+
+static void d40_lcla_id_put(struct d40_chan *d40c,
+                           struct d40_lcla_pool *pool,
+                           int id)
+{
+       if (id < 0)
+               return;
+
+       d40c->lcla.src_id = -1;
+       d40c->lcla.dst_id = -1;
+
+       spin_lock(&pool->lock);
+       pool->alloc_map[d40c->phy_chan->num] &= (~(0x1 << id));
+       spin_unlock(&pool->lock);
+}
+
+static int d40_channel_execute_command(struct d40_chan *d40c,
+                                      enum d40_command command)
+{
+       int status, i;
+       void __iomem *active_reg;
+       int ret = 0;
+       unsigned long flags;
+
+       spin_lock_irqsave(&d40c->base->execmd_lock, flags);
+
+       if (d40c->phy_chan->num % 2 == 0)
+               active_reg = d40c->base->virtbase + D40_DREG_ACTIVE;
+       else
+               active_reg = d40c->base->virtbase + D40_DREG_ACTIVO;
+
+       if (command == D40_DMA_SUSPEND_REQ) {
+               status = (readl(active_reg) &
+                         D40_CHAN_POS_MASK(d40c->phy_chan->num)) >>
+                       D40_CHAN_POS(d40c->phy_chan->num);
+
+               if (status == D40_DMA_SUSPENDED || status == D40_DMA_STOP)
+                       goto done;
+       }
+
+       writel(command << D40_CHAN_POS(d40c->phy_chan->num), active_reg);
+
+       if (command == D40_DMA_SUSPEND_REQ) {
+
+               for (i = 0 ; i < D40_SUSPEND_MAX_IT; i++) {
+                       status = (readl(active_reg) &
+                                 D40_CHAN_POS_MASK(d40c->phy_chan->num)) >>
+                               D40_CHAN_POS(d40c->phy_chan->num);
+
+                       cpu_relax();
+                       /*
+                        * Reduce the number of bus accesses while
+                        * waiting for the DMA to suspend.
+                        */
+                       udelay(3);
+
+                       if (status == D40_DMA_STOP ||
+                           status == D40_DMA_SUSPENDED)
+                               break;
+               }
+
+               if (i == D40_SUSPEND_MAX_IT) {
+                       dev_err(&d40c->chan.dev->device,
+                               "[%s]: unable to suspend the chl %d (log: %d) status %x\n",
+                               __func__, d40c->phy_chan->num, d40c->log_num,
+                               status);
+                       dump_stack();
+                       ret = -EBUSY;
+               }
+
+       }
+done:
+       spin_unlock_irqrestore(&d40c->base->execmd_lock, flags);
+       return ret;
+}
+
+static void d40_term_all(struct d40_chan *d40c)
+{
+       struct d40_desc *d40d;
+       struct d40_desc *d;
+       struct d40_desc *_d;
+
+       /* Release active descriptors */
+       while ((d40d = d40_first_active_get(d40c))) {
+               d40_desc_remove(d40d);
+
+               /* Return desc to free-list */
+               d40_desc_free(d40c, d40d);
+       }
+
+       /* Release queued descriptors waiting for transfer */
+       while ((d40d = d40_first_queued(d40c))) {
+               d40_desc_remove(d40d);
+
+               /* Return desc to free-list */
+               d40_desc_free(d40c, d40d);
+       }
+
+       /* Release client owned descriptors */
+       if (!list_empty(&d40c->client))
+               list_for_each_entry_safe(d, _d, &d40c->client, node) {
+                       d40_pool_lli_free(d);
+                       d40_desc_remove(d);
+                       /* Return desc to free-list */
+                       d40_desc_free(d40c, d40d);
+               }
+
+       d40_lcla_id_put(d40c, &d40c->base->lcla_pool,
+                       d40c->lcla.src_id);
+       d40_lcla_id_put(d40c, &d40c->base->lcla_pool,
+                       d40c->lcla.dst_id);
+
+       d40c->pending_tx = 0;
+       d40c->busy = false;
+}
+
+static void d40_config_set_event(struct d40_chan *d40c, bool do_enable)
+{
+       u32 val;
+       unsigned long flags;
+
+       if (do_enable)
+               val = D40_ACTIVATE_EVENTLINE;
+       else
+               val = D40_DEACTIVATE_EVENTLINE;
+
+       spin_lock_irqsave(&d40c->phy_chan->lock, flags);
+
+       /* Enable event line connected to device (or memcpy) */
+       if ((d40c->dma_cfg.dir ==  STEDMA40_PERIPH_TO_MEM) ||
+           (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_PERIPH)) {
+               u32 event = D40_TYPE_TO_EVENT(d40c->dma_cfg.src_dev_type);
+
+               writel((val << D40_EVENTLINE_POS(event)) |
+                      ~D40_EVENTLINE_MASK(event),
+                      d40c->base->virtbase + D40_DREG_PCBASE +
+                      d40c->phy_chan->num * D40_DREG_PCDELTA +
+                      D40_CHAN_REG_SSLNK);
+       }
+       if (d40c->dma_cfg.dir !=  STEDMA40_PERIPH_TO_MEM) {
+               u32 event = D40_TYPE_TO_EVENT(d40c->dma_cfg.dst_dev_type);
+
+               writel((val << D40_EVENTLINE_POS(event)) |
+                      ~D40_EVENTLINE_MASK(event),
+                      d40c->base->virtbase + D40_DREG_PCBASE +
+                      d40c->phy_chan->num * D40_DREG_PCDELTA +
+                      D40_CHAN_REG_SDLNK);
+       }
+
+       spin_unlock_irqrestore(&d40c->phy_chan->lock, flags);
+}
+
+static bool d40_chan_has_events(struct d40_chan *d40c)
+{
+       u32 val = 0;
+
+       /* If SSLNK or SDLNK is zero all events are disabled */
+       if ((d40c->dma_cfg.dir ==  STEDMA40_PERIPH_TO_MEM) ||
+           (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_PERIPH))
+               val = readl(d40c->base->virtbase + D40_DREG_PCBASE +
+                           d40c->phy_chan->num * D40_DREG_PCDELTA +
+                           D40_CHAN_REG_SSLNK);
+
+       if (d40c->dma_cfg.dir !=  STEDMA40_PERIPH_TO_MEM)
+               val = readl(d40c->base->virtbase + D40_DREG_PCBASE +
+                           d40c->phy_chan->num * D40_DREG_PCDELTA +
+                           D40_CHAN_REG_SDLNK);
+       return (bool) val;
+}
+
+static void d40_config_enable_lidx(struct d40_chan *d40c)
+{
+       /* Set LIDX for lcla */
+       writel((d40c->phy_chan->num << D40_SREG_ELEM_LOG_LIDX_POS) &
+              D40_SREG_ELEM_LOG_LIDX_MASK,
+              d40c->base->virtbase + D40_DREG_PCBASE +
+              d40c->phy_chan->num * D40_DREG_PCDELTA + D40_CHAN_REG_SDELT);
+
+       writel((d40c->phy_chan->num << D40_SREG_ELEM_LOG_LIDX_POS) &
+              D40_SREG_ELEM_LOG_LIDX_MASK,
+              d40c->base->virtbase + D40_DREG_PCBASE +
+              d40c->phy_chan->num * D40_DREG_PCDELTA + D40_CHAN_REG_SSELT);
+}
+
+static int d40_config_write(struct d40_chan *d40c)
+{
+       u32 addr_base;
+       u32 var;
+       int res;
+
+       res = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ);
+       if (res)
+               return res;
+
+       /* Odd addresses are even addresses + 4 */
+       addr_base = (d40c->phy_chan->num % 2) * 4;
+       /* Setup channel mode to logical or physical */
+       var = ((u32)(d40c->log_num != D40_PHY_CHAN) + 1) <<
+               D40_CHAN_POS(d40c->phy_chan->num);
+       writel(var, d40c->base->virtbase + D40_DREG_PRMSE + addr_base);
+
+       /* Setup operational mode option register */
+       var = ((d40c->dma_cfg.channel_type >> STEDMA40_INFO_CH_MODE_OPT_POS) &
+              0x3) << D40_CHAN_POS(d40c->phy_chan->num);
+
+       writel(var, d40c->base->virtbase + D40_DREG_PRMOE + addr_base);
+
+       if (d40c->log_num != D40_PHY_CHAN) {
+               /* Set default config for CFG reg */
+               writel(d40c->src_def_cfg,
+                      d40c->base->virtbase + D40_DREG_PCBASE +
+                      d40c->phy_chan->num * D40_DREG_PCDELTA +
+                      D40_CHAN_REG_SSCFG);
+               writel(d40c->dst_def_cfg,
+                      d40c->base->virtbase + D40_DREG_PCBASE +
+                      d40c->phy_chan->num * D40_DREG_PCDELTA +
+                      D40_CHAN_REG_SDCFG);
+
+               d40_config_enable_lidx(d40c);
+       }
+       return res;
+}
+
+static void d40_desc_load(struct d40_chan *d40c, struct d40_desc *d40d)
+{
+
+       if (d40d->lli_phy.dst && d40d->lli_phy.src) {
+               d40_phy_lli_write(d40c->base->virtbase,
+                                 d40c->phy_chan->num,
+                                 d40d->lli_phy.dst,
+                                 d40d->lli_phy.src);
+               d40d->lli_tcount = d40d->lli_len;
+       } else if (d40d->lli_log.dst && d40d->lli_log.src) {
+               u32 lli_len;
+               struct d40_log_lli *src = d40d->lli_log.src;
+               struct d40_log_lli *dst = d40d->lli_log.dst;
+
+               src += d40d->lli_tcount;
+               dst += d40d->lli_tcount;
+
+               if (d40d->lli_len <= d40c->base->plat_data->llis_per_log)
+                       lli_len = d40d->lli_len;
+               else
+                       lli_len = d40c->base->plat_data->llis_per_log;
+               d40d->lli_tcount += lli_len;
+               d40_log_lli_write(d40c->lcpa, d40c->lcla.src,
+                                 d40c->lcla.dst,
+                                 dst, src,
+                                 d40c->base->plat_data->llis_per_log);
+       }
+}
+
+static dma_cookie_t d40_tx_submit(struct dma_async_tx_descriptor *tx)
+{
+       struct d40_chan *d40c = container_of(tx->chan,
+                                            struct d40_chan,
+                                            chan);
+       struct d40_desc *d40d = container_of(tx, struct d40_desc, txd);
+       unsigned long flags;
+
+       spin_lock_irqsave(&d40c->lock, flags);
+
+       tx->cookie = d40_assign_cookie(d40c, d40d);
+
+       d40_desc_queue(d40c, d40d);
+
+       spin_unlock_irqrestore(&d40c->lock, flags);
+
+       return tx->cookie;
+}
+
+static int d40_start(struct d40_chan *d40c)
+{
+       int err;
+
+       if (d40c->log_num != D40_PHY_CHAN) {
+               err = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ);
+               if (err)
+                       return err;
+               d40_config_set_event(d40c, true);
+       }
+
+       err = d40_channel_execute_command(d40c, D40_DMA_RUN);
+
+       return err;
+}
+
+static struct d40_desc *d40_queue_start(struct d40_chan *d40c)
+{
+       struct d40_desc *d40d;
+       int err;
+
+       /* Start queued jobs, if any */
+       d40d = d40_first_queued(d40c);
+
+       if (d40d != NULL) {
+               d40c->busy = true;
+
+               /* Remove from queue */
+               d40_desc_remove(d40d);
+
+               /* Add to active queue */
+               d40_desc_submit(d40c, d40d);
+
+               /* Initiate DMA job */
+               d40_desc_load(d40c, d40d);
+
+               /* Start dma job */
+               err = d40_start(d40c);
+
+               if (err)
+                       return NULL;
+       }
+
+       return d40d;
+}
+
+/* called from interrupt context */
+static void dma_tc_handle(struct d40_chan *d40c)
+{
+       struct d40_desc *d40d;
+
+       if (!d40c->phy_chan)
+               return;
+
+       /* Get first active entry from list */
+       d40d = d40_first_active_get(d40c);
+
+       if (d40d == NULL)
+               return;
+
+       if (d40d->lli_tcount < d40d->lli_len) {
+
+               d40_desc_load(d40c, d40d);
+               /* Start dma job */
+               (void) d40_start(d40c);
+               return;
+       }
+
+       if (d40_queue_start(d40c) == NULL)
+               d40c->busy = false;
+
+       d40c->pending_tx++;
+       tasklet_schedule(&d40c->tasklet);
+
+}
+
+static void dma_tasklet(unsigned long data)
+{
+       struct d40_chan *d40c = (struct d40_chan *) data;
+       struct d40_desc *d40d_fin;
+       unsigned long flags;
+       dma_async_tx_callback callback;
+       void *callback_param;
+
+       spin_lock_irqsave(&d40c->lock, flags);
+
+       /* Get first active entry from list */
+       d40d_fin = d40_first_active_get(d40c);
+
+       if (d40d_fin == NULL)
+               goto err;
+
+       d40c->completed = d40d_fin->txd.cookie;
+
+       /*
+        * If terminating a channel pending_tx is set to zero.
+        * This prevents any finished active jobs to return to the client.
+        */
+       if (d40c->pending_tx == 0) {
+               spin_unlock_irqrestore(&d40c->lock, flags);
+               return;
+       }
+
+       /* Callback to client */
+       callback = d40d_fin->txd.callback;
+       callback_param = d40d_fin->txd.callback_param;
+
+       if (async_tx_test_ack(&d40d_fin->txd)) {
+               d40_pool_lli_free(d40d_fin);
+               d40_desc_remove(d40d_fin);
+               /* Return desc to free-list */
+               d40_desc_free(d40c, d40d_fin);
+       } else {
+               d40_desc_reset(d40d_fin);
+               if (!d40d_fin->is_in_client_list) {
+                       d40_desc_remove(d40d_fin);
+                       list_add_tail(&d40d_fin->node, &d40c->client);
+                       d40d_fin->is_in_client_list = true;
+               }
+       }
+
+       d40c->pending_tx--;
+
+       if (d40c->pending_tx)
+               tasklet_schedule(&d40c->tasklet);
+
+       spin_unlock_irqrestore(&d40c->lock, flags);
+
+       if (callback)
+               callback(callback_param);
+
+       return;
+
+ err:
+       /* Rescue manouver if receiving double interrupts */
+       if (d40c->pending_tx > 0)
+               d40c->pending_tx--;
+       spin_unlock_irqrestore(&d40c->lock, flags);
+}
+
+static irqreturn_t d40_handle_interrupt(int irq, void *data)
+{
+       static const struct d40_interrupt_lookup il[] = {
+               {D40_DREG_LCTIS0, D40_DREG_LCICR0, false,  0},
+               {D40_DREG_LCTIS1, D40_DREG_LCICR1, false, 32},
+               {D40_DREG_LCTIS2, D40_DREG_LCICR2, false, 64},
+               {D40_DREG_LCTIS3, D40_DREG_LCICR3, false, 96},
+               {D40_DREG_LCEIS0, D40_DREG_LCICR0, true,   0},
+               {D40_DREG_LCEIS1, D40_DREG_LCICR1, true,  32},
+               {D40_DREG_LCEIS2, D40_DREG_LCICR2, true,  64},
+               {D40_DREG_LCEIS3, D40_DREG_LCICR3, true,  96},
+               {D40_DREG_PCTIS,  D40_DREG_PCICR,  false, D40_PHY_CHAN},
+               {D40_DREG_PCEIS,  D40_DREG_PCICR,  true,  D40_PHY_CHAN},
+       };
+
+       int i;
+       u32 regs[ARRAY_SIZE(il)];
+       u32 tmp;
+       u32 idx;
+       u32 row;
+       long chan = -1;
+       struct d40_chan *d40c;
+       unsigned long flags;
+       struct d40_base *base = data;
+
+       spin_lock_irqsave(&base->interrupt_lock, flags);
+
+       /* Read interrupt status of both logical and physical channels */
+       for (i = 0; i < ARRAY_SIZE(il); i++)
+               regs[i] = readl(base->virtbase + il[i].src);
+
+       for (;;) {
+
+               chan = find_next_bit((unsigned long *)regs,
+                                    BITS_PER_LONG * ARRAY_SIZE(il), chan + 1);
+
+               /* No more set bits found? */
+               if (chan == BITS_PER_LONG * ARRAY_SIZE(il))
+                       break;
+
+               row = chan / BITS_PER_LONG;
+               idx = chan & (BITS_PER_LONG - 1);
+
+               /* ACK interrupt */
+               tmp = readl(base->virtbase + il[row].clr);
+               tmp |= 1 << idx;
+               writel(tmp, base->virtbase + il[row].clr);
+
+               if (il[row].offset == D40_PHY_CHAN)
+                       d40c = base->lookup_phy_chans[idx];
+               else
+                       d40c = base->lookup_log_chans[il[row].offset + idx];
+               spin_lock(&d40c->lock);
+
+               if (!il[row].is_error)
+                       dma_tc_handle(d40c);
+               else
+                       dev_err(base->dev, "[%s] IRQ chan: %ld offset %d idx %d\n",
+                               __func__, chan, il[row].offset, idx);
+
+               spin_unlock(&d40c->lock);
+       }
+
+       spin_unlock_irqrestore(&base->interrupt_lock, flags);
+
+       return IRQ_HANDLED;
+}
+
+
+static int d40_validate_conf(struct d40_chan *d40c,
+                            struct stedma40_chan_cfg *conf)
+{
+       int res = 0;
+       u32 dst_event_group = D40_TYPE_TO_GROUP(conf->dst_dev_type);
+       u32 src_event_group = D40_TYPE_TO_GROUP(conf->src_dev_type);
+       bool is_log = (conf->channel_type & STEDMA40_CHANNEL_IN_OPER_MODE)
+               == STEDMA40_CHANNEL_IN_LOG_MODE;
+
+       if (d40c->dma_cfg.dir == STEDMA40_MEM_TO_PERIPH &&
+           dst_event_group == STEDMA40_DEV_DST_MEMORY) {
+               dev_err(&d40c->chan.dev->device, "[%s] Invalid dst\n",
+                       __func__);
+               res = -EINVAL;
+       }
+
+       if (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_MEM &&
+           src_event_group == STEDMA40_DEV_SRC_MEMORY) {
+               dev_err(&d40c->chan.dev->device, "[%s] Invalid src\n",
+                       __func__);
+               res = -EINVAL;
+       }
+
+       if (src_event_group == STEDMA40_DEV_SRC_MEMORY &&
+           dst_event_group == STEDMA40_DEV_DST_MEMORY && is_log) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] No event line\n", __func__);
+               res = -EINVAL;
+       }
+
+       if (conf->dir == STEDMA40_PERIPH_TO_PERIPH &&
+           (src_event_group != dst_event_group)) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Invalid event group\n", __func__);
+               res = -EINVAL;
+       }
+
+       if (conf->dir == STEDMA40_PERIPH_TO_PERIPH) {
+               /*
+                * DMAC HW supports it. Will be added to this driver,
+                * in case any dma client requires it.
+                */
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] periph to periph not supported\n",
+                       __func__);
+               res = -EINVAL;
+       }
+
+       return res;
+}
+
+static bool d40_alloc_mask_set(struct d40_phy_res *phy, bool is_src,
+                              int log_event_line)
+{
+       unsigned long flags;
+       spin_lock_irqsave(&phy->lock, flags);
+       if (!log_event_line) {
+               /* Physical interrupts are masked per physical full channel */
+               if (phy->allocated_src == D40_ALLOC_FREE &&
+                   phy->allocated_dst == D40_ALLOC_FREE) {
+                       phy->allocated_dst = D40_ALLOC_PHY;
+                       phy->allocated_src = D40_ALLOC_PHY;
+                       goto found;
+               } else
+                       goto not_found;
+       }
+
+       /* Logical channel */
+       if (is_src) {
+               if (phy->allocated_src == D40_ALLOC_PHY)
+                       goto not_found;
+
+               if (phy->allocated_src == D40_ALLOC_FREE)
+                       phy->allocated_src = D40_ALLOC_LOG_FREE;
+
+               if (!(phy->allocated_src & (1 << log_event_line))) {
+                       phy->allocated_src |= 1 << log_event_line;
+                       goto found;
+               } else
+                       goto not_found;
+       } else {
+               if (phy->allocated_dst == D40_ALLOC_PHY)
+                       goto not_found;
+
+               if (phy->allocated_dst == D40_ALLOC_FREE)
+                       phy->allocated_dst = D40_ALLOC_LOG_FREE;
+
+               if (!(phy->allocated_dst & (1 << log_event_line))) {
+                       phy->allocated_dst |= 1 << log_event_line;
+                       goto found;
+               } else
+                       goto not_found;
+       }
+
+not_found:
+       spin_unlock_irqrestore(&phy->lock, flags);
+       return false;
+found:
+       spin_unlock_irqrestore(&phy->lock, flags);
+       return true;
+}
+
+static bool d40_alloc_mask_free(struct d40_phy_res *phy, bool is_src,
+                              int log_event_line)
+{
+       unsigned long flags;
+       bool is_free = false;
+
+       spin_lock_irqsave(&phy->lock, flags);
+       if (!log_event_line) {
+               /* Physical interrupts are masked per physical full channel */
+               phy->allocated_dst = D40_ALLOC_FREE;
+               phy->allocated_src = D40_ALLOC_FREE;
+               is_free = true;
+               goto out;
+       }
+
+       /* Logical channel */
+       if (is_src) {
+               phy->allocated_src &= ~(1 << log_event_line);
+               if (phy->allocated_src == D40_ALLOC_LOG_FREE)
+                       phy->allocated_src = D40_ALLOC_FREE;
+       } else {
+               phy->allocated_dst &= ~(1 << log_event_line);
+               if (phy->allocated_dst == D40_ALLOC_LOG_FREE)
+                       phy->allocated_dst = D40_ALLOC_FREE;
+       }
+
+       is_free = ((phy->allocated_src | phy->allocated_dst) ==
+                  D40_ALLOC_FREE);
+
+out:
+       spin_unlock_irqrestore(&phy->lock, flags);
+
+       return is_free;
+}
+
+static int d40_allocate_channel(struct d40_chan *d40c)
+{
+       int dev_type;
+       int event_group;
+       int event_line;
+       struct d40_phy_res *phys;
+       int i;
+       int j;
+       int log_num;
+       bool is_src;
+       bool is_log = (d40c->dma_cfg.channel_type & STEDMA40_CHANNEL_IN_OPER_MODE)
+               == STEDMA40_CHANNEL_IN_LOG_MODE;
+
+
+       phys = d40c->base->phy_res;
+
+       if (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_MEM) {
+               dev_type = d40c->dma_cfg.src_dev_type;
+               log_num = 2 * dev_type;
+               is_src = true;
+       } else if (d40c->dma_cfg.dir == STEDMA40_MEM_TO_PERIPH ||
+                  d40c->dma_cfg.dir == STEDMA40_MEM_TO_MEM) {
+               /* dst event lines are used for logical memcpy */
+               dev_type = d40c->dma_cfg.dst_dev_type;
+               log_num = 2 * dev_type + 1;
+               is_src = false;
+       } else
+               return -EINVAL;
+
+       event_group = D40_TYPE_TO_GROUP(dev_type);
+       event_line = D40_TYPE_TO_EVENT(dev_type);
+
+       if (!is_log) {
+               if (d40c->dma_cfg.dir == STEDMA40_MEM_TO_MEM) {
+                       /* Find physical half channel */
+                       for (i = 0; i < d40c->base->num_phy_chans; i++) {
+
+                               if (d40_alloc_mask_set(&phys[i], is_src, 0))
+                                       goto found_phy;
+                       }
+               } else
+                       for (j = 0; j < d40c->base->num_phy_chans; j += 8) {
+                               int phy_num = j  + event_group * 2;
+                               for (i = phy_num; i < phy_num + 2; i++) {
+                                       if (d40_alloc_mask_set(&phys[i],
+                                                              is_src, 0))
+                                               goto found_phy;
+                               }
+                       }
+               return -EINVAL;
+found_phy:
+               d40c->phy_chan = &phys[i];
+               d40c->log_num = D40_PHY_CHAN;
+               goto out;
+       }
+       if (dev_type == -1)
+               return -EINVAL;
+
+       /* Find logical channel */
+       for (j = 0; j < d40c->base->num_phy_chans; j += 8) {
+               int phy_num = j + event_group * 2;
+               /*
+                * Spread logical channels across all available physical rather
+                * than pack every logical channel at the first available phy
+                * channels.
+                */
+               if (is_src) {
+                       for (i = phy_num; i < phy_num + 2; i++) {
+                               if (d40_alloc_mask_set(&phys[i], is_src,
+                                                      event_line))
+                                       goto found_log;
+                       }
+               } else {
+                       for (i = phy_num + 1; i >= phy_num; i--) {
+                               if (d40_alloc_mask_set(&phys[i], is_src,
+                                                      event_line))
+                                       goto found_log;
+                       }
+               }
+       }
+       return -EINVAL;
+
+found_log:
+       d40c->phy_chan = &phys[i];
+       d40c->log_num = log_num;
+out:
+
+       if (is_log)
+               d40c->base->lookup_log_chans[d40c->log_num] = d40c;
+       else
+               d40c->base->lookup_phy_chans[d40c->phy_chan->num] = d40c;
+
+       return 0;
+
+}
+
+static int d40_config_chan(struct d40_chan *d40c,
+                          struct stedma40_chan_cfg *info)
+{
+
+       /* Fill in basic CFG register values */
+       d40_phy_cfg(&d40c->dma_cfg, &d40c->src_def_cfg,
+                   &d40c->dst_def_cfg, d40c->log_num != D40_PHY_CHAN);
+
+       if (d40c->log_num != D40_PHY_CHAN) {
+               d40_log_cfg(&d40c->dma_cfg,
+                           &d40c->log_def.lcsp1, &d40c->log_def.lcsp3);
+
+               if (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_MEM)
+                       d40c->lcpa = d40c->base->lcpa_base +
+                               d40c->dma_cfg.src_dev_type * 32;
+               else
+                       d40c->lcpa = d40c->base->lcpa_base +
+                               d40c->dma_cfg.dst_dev_type * 32 + 16;
+       }
+
+       /* Write channel configuration to the DMA */
+       return d40_config_write(d40c);
+}
+
+static int d40_config_memcpy(struct d40_chan *d40c)
+{
+       dma_cap_mask_t cap = d40c->chan.device->cap_mask;
+
+       if (dma_has_cap(DMA_MEMCPY, cap) && !dma_has_cap(DMA_SLAVE, cap)) {
+               d40c->dma_cfg = *d40c->base->plat_data->memcpy_conf_log;
+               d40c->dma_cfg.src_dev_type = STEDMA40_DEV_SRC_MEMORY;
+               d40c->dma_cfg.dst_dev_type = d40c->base->plat_data->
+                       memcpy[d40c->chan.chan_id];
+
+       } else if (dma_has_cap(DMA_MEMCPY, cap) &&
+                  dma_has_cap(DMA_SLAVE, cap)) {
+               d40c->dma_cfg = *d40c->base->plat_data->memcpy_conf_phy;
+       } else {
+               dev_err(&d40c->chan.dev->device, "[%s] No memcpy\n",
+                       __func__);
+               return -EINVAL;
+       }
+
+       return 0;
+}
+
+
+static int d40_free_dma(struct d40_chan *d40c)
+{
+
+       int res = 0;
+       u32 event, dir;
+       struct d40_phy_res *phy = d40c->phy_chan;
+       bool is_src;
+
+       /* Terminate all queued and active transfers */
+       d40_term_all(d40c);
+
+       if (phy == NULL) {
+               dev_err(&d40c->chan.dev->device, "[%s] phy == null\n",
+                       __func__);
+               return -EINVAL;
+       }
+
+       if (phy->allocated_src == D40_ALLOC_FREE &&
+           phy->allocated_dst == D40_ALLOC_FREE) {
+               dev_err(&d40c->chan.dev->device, "[%s] channel already free\n",
+                       __func__);
+               return -EINVAL;
+       }
+
+
+       res = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ);
+       if (res) {
+               dev_err(&d40c->chan.dev->device, "[%s] suspend\n",
+                       __func__);
+               return res;
+       }
+
+       if (d40c->dma_cfg.dir == STEDMA40_MEM_TO_PERIPH ||
+           d40c->dma_cfg.dir == STEDMA40_MEM_TO_MEM) {
+               event = D40_TYPE_TO_EVENT(d40c->dma_cfg.dst_dev_type);
+               dir = D40_CHAN_REG_SDLNK;
+               is_src = false;
+       } else if (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_MEM) {
+               event = D40_TYPE_TO_EVENT(d40c->dma_cfg.src_dev_type);
+               dir = D40_CHAN_REG_SSLNK;
+               is_src = true;
+       } else {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Unknown direction\n", __func__);
+               return -EINVAL;
+       }
+
+       if (d40c->log_num != D40_PHY_CHAN) {
+               /*
+                * Release logical channel, deactivate the event line during
+                * the time physical res is suspended.
+                */
+               writel((D40_DEACTIVATE_EVENTLINE << D40_EVENTLINE_POS(event)) &
+                      D40_EVENTLINE_MASK(event),
+                      d40c->base->virtbase + D40_DREG_PCBASE +
+                      phy->num * D40_DREG_PCDELTA + dir);
+
+               d40c->base->lookup_log_chans[d40c->log_num] = NULL;
+
+               /*
+                * Check if there are more logical allocation
+                * on this phy channel.
+                */
+               if (!d40_alloc_mask_free(phy, is_src, event)) {
+                       /* Resume the other logical channels if any */
+                       if (d40_chan_has_events(d40c)) {
+                               res = d40_channel_execute_command(d40c,
+                                                                 D40_DMA_RUN);
+                               if (res) {
+                                       dev_err(&d40c->chan.dev->device,
+                                               "[%s] Executing RUN command\n",
+                                               __func__);
+                                       return res;
+                               }
+                       }
+                       return 0;
+               }
+       } else
+               d40_alloc_mask_free(phy, is_src, 0);
+
+       /* Release physical channel */
+       res = d40_channel_execute_command(d40c, D40_DMA_STOP);
+       if (res) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Failed to stop channel\n", __func__);
+               return res;
+       }
+       d40c->phy_chan = NULL;
+       /* Invalidate channel type */
+       d40c->dma_cfg.channel_type = 0;
+       d40c->base->lookup_phy_chans[phy->num] = NULL;
+
+       return 0;
+
+
+}
+
+static int d40_pause(struct dma_chan *chan)
+{
+       struct d40_chan *d40c =
+               container_of(chan, struct d40_chan, chan);
+       int res;
+
+       unsigned long flags;
+
+       spin_lock_irqsave(&d40c->lock, flags);
+
+       res = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ);
+       if (res == 0) {
+               if (d40c->log_num != D40_PHY_CHAN) {
+                       d40_config_set_event(d40c, false);
+                       /* Resume the other logical channels if any */
+                       if (d40_chan_has_events(d40c))
+                               res = d40_channel_execute_command(d40c,
+                                                                 D40_DMA_RUN);
+               }
+       }
+
+       spin_unlock_irqrestore(&d40c->lock, flags);
+       return res;
+}
+
+static bool d40_tx_is_linked(struct d40_chan *d40c)
+{
+       bool is_link;
+
+       if (d40c->log_num != D40_PHY_CHAN)
+               is_link = readl(&d40c->lcpa->lcsp3) &  D40_MEM_LCSP3_DLOS_MASK;
+       else
+               is_link = readl(d40c->base->virtbase + D40_DREG_PCBASE +
+                               d40c->phy_chan->num * D40_DREG_PCDELTA +
+                               D40_CHAN_REG_SDLNK) &
+                       D40_SREG_LNK_PHYS_LNK_MASK;
+       return is_link;
+}
+
+static u32 d40_residue(struct d40_chan *d40c)
+{
+       u32 num_elt;
+
+       if (d40c->log_num != D40_PHY_CHAN)
+               num_elt = (readl(&d40c->lcpa->lcsp2) &  D40_MEM_LCSP2_ECNT_MASK)
+                       >> D40_MEM_LCSP2_ECNT_POS;
+       else
+               num_elt = (readl(d40c->base->virtbase + D40_DREG_PCBASE +
+                                d40c->phy_chan->num * D40_DREG_PCDELTA +
+                                D40_CHAN_REG_SDELT) &
+                          D40_SREG_ELEM_PHY_ECNT_MASK) >> D40_SREG_ELEM_PHY_ECNT_POS;
+       return num_elt * (1 << d40c->dma_cfg.dst_info.data_width);
+}
+
+static int d40_resume(struct dma_chan *chan)
+{
+       struct d40_chan *d40c =
+               container_of(chan, struct d40_chan, chan);
+       int res = 0;
+       unsigned long flags;
+
+       spin_lock_irqsave(&d40c->lock, flags);
+
+       if (d40c->log_num != D40_PHY_CHAN) {
+               res = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ);
+               if (res)
+                       goto out;
+
+               /* If bytes left to transfer or linked tx resume job */
+               if (d40_residue(d40c) || d40_tx_is_linked(d40c)) {
+                       d40_config_set_event(d40c, true);
+                       res = d40_channel_execute_command(d40c, D40_DMA_RUN);
+               }
+       } else if (d40_residue(d40c) || d40_tx_is_linked(d40c))
+               res = d40_channel_execute_command(d40c, D40_DMA_RUN);
+
+out:
+       spin_unlock_irqrestore(&d40c->lock, flags);
+       return res;
+}
+
+static u32 stedma40_residue(struct dma_chan *chan)
+{
+       struct d40_chan *d40c =
+               container_of(chan, struct d40_chan, chan);
+       u32 bytes_left;
+       unsigned long flags;
+
+       spin_lock_irqsave(&d40c->lock, flags);
+       bytes_left = d40_residue(d40c);
+       spin_unlock_irqrestore(&d40c->lock, flags);
+
+       return bytes_left;
+}
+
+/* Public DMA functions in addition to the DMA engine framework */
+
+int stedma40_set_psize(struct dma_chan *chan,
+                      int src_psize,
+                      int dst_psize)
+{
+       struct d40_chan *d40c =
+               container_of(chan, struct d40_chan, chan);
+       unsigned long flags;
+
+       spin_lock_irqsave(&d40c->lock, flags);
+
+       if (d40c->log_num != D40_PHY_CHAN) {
+               d40c->log_def.lcsp1 &= ~D40_MEM_LCSP1_SCFG_PSIZE_MASK;
+               d40c->log_def.lcsp3 &= ~D40_MEM_LCSP1_SCFG_PSIZE_MASK;
+               d40c->log_def.lcsp1 |= src_psize << D40_MEM_LCSP1_SCFG_PSIZE_POS;
+               d40c->log_def.lcsp3 |= dst_psize << D40_MEM_LCSP1_SCFG_PSIZE_POS;
+               goto out;
+       }
+
+       if (src_psize == STEDMA40_PSIZE_PHY_1)
+               d40c->src_def_cfg &= ~(1 << D40_SREG_CFG_PHY_PEN_POS);
+       else {
+               d40c->src_def_cfg |= 1 << D40_SREG_CFG_PHY_PEN_POS;
+               d40c->src_def_cfg &= ~(STEDMA40_PSIZE_PHY_16 <<
+                                      D40_SREG_CFG_PSIZE_POS);
+               d40c->src_def_cfg |= src_psize << D40_SREG_CFG_PSIZE_POS;
+       }
+
+       if (dst_psize == STEDMA40_PSIZE_PHY_1)
+               d40c->dst_def_cfg &= ~(1 << D40_SREG_CFG_PHY_PEN_POS);
+       else {
+               d40c->dst_def_cfg |= 1 << D40_SREG_CFG_PHY_PEN_POS;
+               d40c->dst_def_cfg &= ~(STEDMA40_PSIZE_PHY_16 <<
+                                      D40_SREG_CFG_PSIZE_POS);
+               d40c->dst_def_cfg |= dst_psize << D40_SREG_CFG_PSIZE_POS;
+       }
+out:
+       spin_unlock_irqrestore(&d40c->lock, flags);
+       return 0;
+}
+EXPORT_SYMBOL(stedma40_set_psize);
+
+struct dma_async_tx_descriptor *stedma40_memcpy_sg(struct dma_chan *chan,
+                                                  struct scatterlist *sgl_dst,
+                                                  struct scatterlist *sgl_src,
+                                                  unsigned int sgl_len,
+                                                  unsigned long flags)
+{
+       int res;
+       struct d40_desc *d40d;
+       struct d40_chan *d40c = container_of(chan, struct d40_chan,
+                                            chan);
+       unsigned long flg;
+       int lli_max = d40c->base->plat_data->llis_per_log;
+
+
+       spin_lock_irqsave(&d40c->lock, flg);
+       d40d = d40_desc_get(d40c);
+
+       if (d40d == NULL)
+               goto err;
+
+       memset(d40d, 0, sizeof(struct d40_desc));
+       d40d->lli_len = sgl_len;
+
+       d40d->txd.flags = flags;
+
+       if (d40c->log_num != D40_PHY_CHAN) {
+               if (sgl_len > 1)
+                       /*
+                        * Check if there is space available in lcla. If not,
+                        * split list into 1-length and run only in lcpa
+                        * space.
+                        */
+                       if (d40_lcla_id_get(d40c,
+                                           &d40c->base->lcla_pool) != 0)
+                               lli_max = 1;
+
+               if (d40_pool_lli_alloc(d40d, sgl_len, true) < 0) {
+                       dev_err(&d40c->chan.dev->device,
+                               "[%s] Out of memory\n", __func__);
+                       goto err;
+               }
+
+               (void) d40_log_sg_to_lli(d40c->lcla.src_id,
+                                        sgl_src,
+                                        sgl_len,
+                                        d40d->lli_log.src,
+                                        d40c->log_def.lcsp1,
+                                        d40c->dma_cfg.src_info.data_width,
+                                        flags & DMA_PREP_INTERRUPT, lli_max,
+                                        d40c->base->plat_data->llis_per_log);
+
+               (void) d40_log_sg_to_lli(d40c->lcla.dst_id,
+                                        sgl_dst,
+                                        sgl_len,
+                                        d40d->lli_log.dst,
+                                        d40c->log_def.lcsp3,
+                                        d40c->dma_cfg.dst_info.data_width,
+                                        flags & DMA_PREP_INTERRUPT, lli_max,
+                                        d40c->base->plat_data->llis_per_log);
+
+
+       } else {
+               if (d40_pool_lli_alloc(d40d, sgl_len, false) < 0) {
+                       dev_err(&d40c->chan.dev->device,
+                               "[%s] Out of memory\n", __func__);
+                       goto err;
+               }
+
+               res = d40_phy_sg_to_lli(sgl_src,
+                                       sgl_len,
+                                       0,
+                                       d40d->lli_phy.src,
+                                       d40d->lli_phy.src_addr,
+                                       d40c->src_def_cfg,
+                                       d40c->dma_cfg.src_info.data_width,
+                                       d40c->dma_cfg.src_info.psize,
+                                       true);
+
+               if (res < 0)
+                       goto err;
+
+               res = d40_phy_sg_to_lli(sgl_dst,
+                                       sgl_len,
+                                       0,
+                                       d40d->lli_phy.dst,
+                                       d40d->lli_phy.dst_addr,
+                                       d40c->dst_def_cfg,
+                                       d40c->dma_cfg.dst_info.data_width,
+                                       d40c->dma_cfg.dst_info.psize,
+                                       true);
+
+               if (res < 0)
+                       goto err;
+
+               (void) dma_map_single(d40c->base->dev, d40d->lli_phy.src,
+                                     d40d->lli_pool.size, DMA_TO_DEVICE);
+       }
+
+       dma_async_tx_descriptor_init(&d40d->txd, chan);
+
+       d40d->txd.tx_submit = d40_tx_submit;
+
+       spin_unlock_irqrestore(&d40c->lock, flg);
+
+       return &d40d->txd;
+err:
+       spin_unlock_irqrestore(&d40c->lock, flg);
+       return NULL;
+}
+EXPORT_SYMBOL(stedma40_memcpy_sg);
+
+bool stedma40_filter(struct dma_chan *chan, void *data)
+{
+       struct stedma40_chan_cfg *info = data;
+       struct d40_chan *d40c =
+               container_of(chan, struct d40_chan, chan);
+       int err;
+
+       if (data) {
+               err = d40_validate_conf(d40c, info);
+               if (!err)
+                       d40c->dma_cfg = *info;
+       } else
+               err = d40_config_memcpy(d40c);
+
+       return err == 0;
+}
+EXPORT_SYMBOL(stedma40_filter);
+
+/* DMA ENGINE functions */
+static int d40_alloc_chan_resources(struct dma_chan *chan)
+{
+       int err;
+       unsigned long flags;
+       struct d40_chan *d40c =
+               container_of(chan, struct d40_chan, chan);
+
+       spin_lock_irqsave(&d40c->lock, flags);
+
+       d40c->completed = chan->cookie = 1;
+
+       /*
+        * If no dma configuration is set (channel_type == 0)
+        * use default configuration
+        */
+       if (d40c->dma_cfg.channel_type == 0) {
+               err = d40_config_memcpy(d40c);
+               if (err)
+                       goto err_alloc;
+       }
+
+       err = d40_allocate_channel(d40c);
+       if (err) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Failed to allocate channel\n", __func__);
+               goto err_alloc;
+       }
+
+       err = d40_config_chan(d40c, &d40c->dma_cfg);
+       if (err) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Failed to configure channel\n",
+                       __func__);
+               goto err_config;
+       }
+
+       spin_unlock_irqrestore(&d40c->lock, flags);
+       return 0;
+
+ err_config:
+       (void) d40_free_dma(d40c);
+ err_alloc:
+       spin_unlock_irqrestore(&d40c->lock, flags);
+       dev_err(&d40c->chan.dev->device,
+               "[%s] Channel allocation failed\n", __func__);
+       return -EINVAL;
+}
+
+static void d40_free_chan_resources(struct dma_chan *chan)
+{
+       struct d40_chan *d40c =
+               container_of(chan, struct d40_chan, chan);
+       int err;
+       unsigned long flags;
+
+       spin_lock_irqsave(&d40c->lock, flags);
+
+       err = d40_free_dma(d40c);
+
+       if (err)
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Failed to free channel\n", __func__);
+       spin_unlock_irqrestore(&d40c->lock, flags);
+}
+
+static struct dma_async_tx_descriptor *d40_prep_memcpy(struct dma_chan *chan,
+                                                      dma_addr_t dst,
+                                                      dma_addr_t src,
+                                                      size_t size,
+                                                      unsigned long flags)
+{
+       struct d40_desc *d40d;
+       struct d40_chan *d40c = container_of(chan, struct d40_chan,
+                                            chan);
+       unsigned long flg;
+       int err = 0;
+
+       spin_lock_irqsave(&d40c->lock, flg);
+       d40d = d40_desc_get(d40c);
+
+       if (d40d == NULL) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Descriptor is NULL\n", __func__);
+               goto err;
+       }
+
+       memset(d40d, 0, sizeof(struct d40_desc));
+
+       d40d->txd.flags = flags;
+
+       dma_async_tx_descriptor_init(&d40d->txd, chan);
+
+       d40d->txd.tx_submit = d40_tx_submit;
+
+       if (d40c->log_num != D40_PHY_CHAN) {
+
+               if (d40_pool_lli_alloc(d40d, 1, true) < 0) {
+                       dev_err(&d40c->chan.dev->device,
+                               "[%s] Out of memory\n", __func__);
+                       goto err;
+               }
+               d40d->lli_len = 1;
+
+               d40_log_fill_lli(d40d->lli_log.src,
+                                src,
+                                size,
+                                0,
+                                d40c->log_def.lcsp1,
+                                d40c->dma_cfg.src_info.data_width,
+                                true, true);
+
+               d40_log_fill_lli(d40d->lli_log.dst,
+                                dst,
+                                size,
+                                0,
+                                d40c->log_def.lcsp3,
+                                d40c->dma_cfg.dst_info.data_width,
+                                true, true);
+
+       } else {
+
+               if (d40_pool_lli_alloc(d40d, 1, false) < 0) {
+                       dev_err(&d40c->chan.dev->device,
+                               "[%s] Out of memory\n", __func__);
+                       goto err;
+               }
+
+               err = d40_phy_fill_lli(d40d->lli_phy.src,
+                                      src,
+                                      size,
+                                      d40c->dma_cfg.src_info.psize,
+                                      0,
+                                      d40c->src_def_cfg,
+                                      true,
+                                      d40c->dma_cfg.src_info.data_width,
+                                      false);
+               if (err)
+                       goto err_fill_lli;
+
+               err = d40_phy_fill_lli(d40d->lli_phy.dst,
+                                      dst,
+                                      size,
+                                      d40c->dma_cfg.dst_info.psize,
+                                      0,
+                                      d40c->dst_def_cfg,
+                                      true,
+                                      d40c->dma_cfg.dst_info.data_width,
+                                      false);
+
+               if (err)
+                       goto err_fill_lli;
+
+               (void) dma_map_single(d40c->base->dev, d40d->lli_phy.src,
+                                     d40d->lli_pool.size, DMA_TO_DEVICE);
+       }
+
+       spin_unlock_irqrestore(&d40c->lock, flg);
+       return &d40d->txd;
+
+err_fill_lli:
+       dev_err(&d40c->chan.dev->device,
+               "[%s] Failed filling in PHY LLI\n", __func__);
+       d40_pool_lli_free(d40d);
+err:
+       spin_unlock_irqrestore(&d40c->lock, flg);
+       return NULL;
+}
+
+static int d40_prep_slave_sg_log(struct d40_desc *d40d,
+                                struct d40_chan *d40c,
+                                struct scatterlist *sgl,
+                                unsigned int sg_len,
+                                enum dma_data_direction direction,
+                                unsigned long flags)
+{
+       dma_addr_t dev_addr = 0;
+       int total_size;
+       int lli_max = d40c->base->plat_data->llis_per_log;
+
+       if (d40_pool_lli_alloc(d40d, sg_len, true) < 0) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Out of memory\n", __func__);
+               return -ENOMEM;
+       }
+
+       d40d->lli_len = sg_len;
+       d40d->lli_tcount = 0;
+
+       if (sg_len > 1)
+               /*
+                * Check if there is space available in lcla.
+                * If not, split list into 1-length and run only
+                * in lcpa space.
+                */
+               if (d40_lcla_id_get(d40c, &d40c->base->lcla_pool) != 0)
+                       lli_max = 1;
+
+       if (direction == DMA_FROM_DEVICE) {
+               dev_addr = d40c->base->plat_data->dev_rx[d40c->dma_cfg.src_dev_type];
+               total_size = d40_log_sg_to_dev(&d40c->lcla,
+                                              sgl, sg_len,
+                                              &d40d->lli_log,
+                                              &d40c->log_def,
+                                              d40c->dma_cfg.src_info.data_width,
+                                              d40c->dma_cfg.dst_info.data_width,
+                                              direction,
+                                              flags & DMA_PREP_INTERRUPT,
+                                              dev_addr, lli_max,
+                                              d40c->base->plat_data->llis_per_log);
+       } else if (direction == DMA_TO_DEVICE) {
+               dev_addr = d40c->base->plat_data->dev_tx[d40c->dma_cfg.dst_dev_type];
+               total_size = d40_log_sg_to_dev(&d40c->lcla,
+                                              sgl, sg_len,
+                                              &d40d->lli_log,
+                                              &d40c->log_def,
+                                              d40c->dma_cfg.src_info.data_width,
+                                              d40c->dma_cfg.dst_info.data_width,
+                                              direction,
+                                              flags & DMA_PREP_INTERRUPT,
+                                              dev_addr, lli_max,
+                                              d40c->base->plat_data->llis_per_log);
+       } else
+               return -EINVAL;
+       if (total_size < 0)
+               return -EINVAL;
+
+       return 0;
+}
+
+static int d40_prep_slave_sg_phy(struct d40_desc *d40d,
+                                struct d40_chan *d40c,
+                                struct scatterlist *sgl,
+                                unsigned int sgl_len,
+                                enum dma_data_direction direction,
+                                unsigned long flags)
+{
+       dma_addr_t src_dev_addr;
+       dma_addr_t dst_dev_addr;
+       int res;
+
+       if (d40_pool_lli_alloc(d40d, sgl_len, false) < 0) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Out of memory\n", __func__);
+               return -ENOMEM;
+       }
+
+       d40d->lli_len = sgl_len;
+       d40d->lli_tcount = 0;
+
+       if (direction == DMA_FROM_DEVICE) {
+               dst_dev_addr = 0;
+               src_dev_addr = d40c->base->plat_data->dev_rx[d40c->dma_cfg.src_dev_type];
+       } else if (direction == DMA_TO_DEVICE) {
+               dst_dev_addr = d40c->base->plat_data->dev_tx[d40c->dma_cfg.dst_dev_type];
+               src_dev_addr = 0;
+       } else
+               return -EINVAL;
+
+       res = d40_phy_sg_to_lli(sgl,
+                               sgl_len,
+                               src_dev_addr,
+                               d40d->lli_phy.src,
+                               d40d->lli_phy.src_addr,
+                               d40c->src_def_cfg,
+                               d40c->dma_cfg.src_info.data_width,
+                               d40c->dma_cfg.src_info.psize,
+                               true);
+       if (res < 0)
+               return res;
+
+       res = d40_phy_sg_to_lli(sgl,
+                               sgl_len,
+                               dst_dev_addr,
+                               d40d->lli_phy.dst,
+                               d40d->lli_phy.dst_addr,
+                               d40c->dst_def_cfg,
+                               d40c->dma_cfg.dst_info.data_width,
+                               d40c->dma_cfg.dst_info.psize,
+                                true);
+       if (res < 0)
+               return res;
+
+       (void) dma_map_single(d40c->base->dev, d40d->lli_phy.src,
+                             d40d->lli_pool.size, DMA_TO_DEVICE);
+       return 0;
+}
+
+static struct dma_async_tx_descriptor *d40_prep_slave_sg(struct dma_chan *chan,
+                                                        struct scatterlist *sgl,
+                                                        unsigned int sg_len,
+                                                        enum dma_data_direction direction,
+                                                        unsigned long flags)
+{
+       struct d40_desc *d40d;
+       struct d40_chan *d40c = container_of(chan, struct d40_chan,
+                                            chan);
+       unsigned long flg;
+       int err;
+
+       if (d40c->dma_cfg.pre_transfer)
+               d40c->dma_cfg.pre_transfer(chan,
+                                          d40c->dma_cfg.pre_transfer_data,
+                                          sg_dma_len(sgl));
+
+       spin_lock_irqsave(&d40c->lock, flg);
+       d40d = d40_desc_get(d40c);
+       spin_unlock_irqrestore(&d40c->lock, flg);
+
+       if (d40d == NULL)
+               return NULL;
+
+       memset(d40d, 0, sizeof(struct d40_desc));
+
+       if (d40c->log_num != D40_PHY_CHAN)
+               err = d40_prep_slave_sg_log(d40d, d40c, sgl, sg_len,
+                                           direction, flags);
+       else
+               err = d40_prep_slave_sg_phy(d40d, d40c, sgl, sg_len,
+                                           direction, flags);
+       if (err) {
+               dev_err(&d40c->chan.dev->device,
+                       "[%s] Failed to prepare %s slave sg job: %d\n",
+                       __func__,
+                       d40c->log_num != D40_PHY_CHAN ? "log" : "phy", err);
+               return NULL;
+       }
+
+       d40d->txd.flags = flags;
+
+       dma_async_tx_descriptor_init(&d40d->txd, chan);
+
+       d40d->txd.tx_submit = d40_tx_submit;
+
+       return &d40d->txd;
+}
+
+static enum dma_status d40_tx_status(struct dma_chan *chan,
+                                    dma_cookie_t cookie,
+                                    struct dma_tx_state *txstate)
+{
+       struct d40_chan *d40c = container_of(chan, struct d40_chan, chan);
+       dma_cookie_t last_used;
+       dma_cookie_t last_complete;
+       int ret;
+
+       last_complete = d40c->completed;
+       last_used = chan->cookie;
+
+       ret = dma_async_is_complete(cookie, last_complete, last_used);
+
+       if (txstate) {
+               txstate->last = last_complete;
+               txstate->used = last_used;
+               txstate->residue = stedma40_residue(chan);
+       }
+
+       return ret;
+}
+
+static void d40_issue_pending(struct dma_chan *chan)
+{
+       struct d40_chan *d40c = container_of(chan, struct d40_chan, chan);
+       unsigned long flags;
+
+       spin_lock_irqsave(&d40c->lock, flags);
+
+       /* Busy means that pending jobs are already being processed */
+       if (!d40c->busy)
+               (void) d40_queue_start(d40c);
+
+       spin_unlock_irqrestore(&d40c->lock, flags);
+}
+
+static int d40_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd)
+{
+       unsigned long flags;
+       struct d40_chan *d40c = container_of(chan, struct d40_chan, chan);
+
+       switch (cmd) {
+       case DMA_TERMINATE_ALL:
+               spin_lock_irqsave(&d40c->lock, flags);
+               d40_term_all(d40c);
+               spin_unlock_irqrestore(&d40c->lock, flags);
+               return 0;
+       case DMA_PAUSE:
+               return d40_pause(chan);
+       case DMA_RESUME:
+               return d40_resume(chan);
+       }
+
+       /* Other commands are unimplemented */
+       return -ENXIO;
+}
+
+/* Initialization functions */
+
+static void __init d40_chan_init(struct d40_base *base, struct dma_device *dma,
+                                struct d40_chan *chans, int offset,
+                                int num_chans)
+{
+       int i = 0;
+       struct d40_chan *d40c;
+
+       INIT_LIST_HEAD(&dma->channels);
+
+       for (i = offset; i < offset + num_chans; i++) {
+               d40c = &chans[i];
+               d40c->base = base;
+               d40c->chan.device = dma;
+
+               /* Invalidate lcla element */
+               d40c->lcla.src_id = -1;
+               d40c->lcla.dst_id = -1;
+
+               spin_lock_init(&d40c->lock);
+
+               d40c->log_num = D40_PHY_CHAN;
+
+               INIT_LIST_HEAD(&d40c->free);
+               INIT_LIST_HEAD(&d40c->active);
+               INIT_LIST_HEAD(&d40c->queue);
+               INIT_LIST_HEAD(&d40c->client);
+
+               d40c->free_len = 0;
+
+               tasklet_init(&d40c->tasklet, dma_tasklet,
+                            (unsigned long) d40c);
+
+               list_add_tail(&d40c->chan.device_node,
+                             &dma->channels);
+       }
+}
+
+static int __init d40_dmaengine_init(struct d40_base *base,
+                                    int num_reserved_chans)
+{
+       int err ;
+
+       d40_chan_init(base, &base->dma_slave, base->log_chans,
+                     0, base->num_log_chans);
+
+       dma_cap_zero(base->dma_slave.cap_mask);
+       dma_cap_set(DMA_SLAVE, base->dma_slave.cap_mask);
+
+       base->dma_slave.device_alloc_chan_resources = d40_alloc_chan_resources;
+       base->dma_slave.device_free_chan_resources = d40_free_chan_resources;
+       base->dma_slave.device_prep_dma_memcpy = d40_prep_memcpy;
+       base->dma_slave.device_prep_slave_sg = d40_prep_slave_sg;
+       base->dma_slave.device_tx_status = d40_tx_status;
+       base->dma_slave.device_issue_pending = d40_issue_pending;
+       base->dma_slave.device_control = d40_control;
+       base->dma_slave.dev = base->dev;
+
+       err = dma_async_device_register(&base->dma_slave);
+
+       if (err) {
+               dev_err(base->dev,
+                       "[%s] Failed to register slave channels\n",
+                       __func__);
+               goto failure1;
+       }
+
+       d40_chan_init(base, &base->dma_memcpy, base->log_chans,
+                     base->num_log_chans, base->plat_data->memcpy_len);
+
+       dma_cap_zero(base->dma_memcpy.cap_mask);
+       dma_cap_set(DMA_MEMCPY, base->dma_memcpy.cap_mask);
+
+       base->dma_memcpy.device_alloc_chan_resources = d40_alloc_chan_resources;
+       base->dma_memcpy.device_free_chan_resources = d40_free_chan_resources;
+       base->dma_memcpy.device_prep_dma_memcpy = d40_prep_memcpy;
+       base->dma_memcpy.device_prep_slave_sg = d40_prep_slave_sg;
+       base->dma_memcpy.device_tx_status = d40_tx_status;
+       base->dma_memcpy.device_issue_pending = d40_issue_pending;
+       base->dma_memcpy.device_control = d40_control;
+       base->dma_memcpy.dev = base->dev;
+       /*
+        * This controller can only access address at even
+        * 32bit boundaries, i.e. 2^2
+        */
+       base->dma_memcpy.copy_align = 2;
+
+       err = dma_async_device_register(&base->dma_memcpy);
+
+       if (err) {
+               dev_err(base->dev,
+                       "[%s] Failed to regsiter memcpy only channels\n",
+                       __func__);
+               goto failure2;
+       }
+
+       d40_chan_init(base, &base->dma_both, base->phy_chans,
+                     0, num_reserved_chans);
+
+       dma_cap_zero(base->dma_both.cap_mask);
+       dma_cap_set(DMA_SLAVE, base->dma_both.cap_mask);
+       dma_cap_set(DMA_MEMCPY, base->dma_both.cap_mask);
+
+       base->dma_both.device_alloc_chan_resources = d40_alloc_chan_resources;
+       base->dma_both.device_free_chan_resources = d40_free_chan_resources;
+       base->dma_both.device_prep_dma_memcpy = d40_prep_memcpy;
+       base->dma_both.device_prep_slave_sg = d40_prep_slave_sg;
+       base->dma_both.device_tx_status = d40_tx_status;
+       base->dma_both.device_issue_pending = d40_issue_pending;
+       base->dma_both.device_control = d40_control;
+       base->dma_both.dev = base->dev;
+       base->dma_both.copy_align = 2;
+       err = dma_async_device_register(&base->dma_both);
+
+       if (err) {
+               dev_err(base->dev,
+                       "[%s] Failed to register logical and physical capable channels\n",
+                       __func__);
+               goto failure3;
+       }
+       return 0;
+failure3:
+       dma_async_device_unregister(&base->dma_memcpy);
+failure2:
+       dma_async_device_unregister(&base->dma_slave);
+failure1:
+       return err;
+}
+
+/* Initialization functions. */
+
+static int __init d40_phy_res_init(struct d40_base *base)
+{
+       int i;
+       int num_phy_chans_avail = 0;
+       u32 val[2];
+       int odd_even_bit = -2;
+
+       val[0] = readl(base->virtbase + D40_DREG_PRSME);
+       val[1] = readl(base->virtbase + D40_DREG_PRSMO);
+
+       for (i = 0; i < base->num_phy_chans; i++) {
+               base->phy_res[i].num = i;
+               odd_even_bit += 2 * ((i % 2) == 0);
+               if (((val[i % 2] >> odd_even_bit) & 3) == 1) {
+                       /* Mark security only channels as occupied */
+                       base->phy_res[i].allocated_src = D40_ALLOC_PHY;
+                       base->phy_res[i].allocated_dst = D40_ALLOC_PHY;
+               } else {
+                       base->phy_res[i].allocated_src = D40_ALLOC_FREE;
+                       base->phy_res[i].allocated_dst = D40_ALLOC_FREE;
+                       num_phy_chans_avail++;
+               }
+               spin_lock_init(&base->phy_res[i].lock);
+       }
+       dev_info(base->dev, "%d of %d physical DMA channels available\n",
+                num_phy_chans_avail, base->num_phy_chans);
+
+       /* Verify settings extended vs standard */
+       val[0] = readl(base->virtbase + D40_DREG_PRTYP);
+
+       for (i = 0; i < base->num_phy_chans; i++) {
+
+               if (base->phy_res[i].allocated_src == D40_ALLOC_FREE &&
+                   (val[0] & 0x3) != 1)
+                       dev_info(base->dev,
+                                "[%s] INFO: channel %d is misconfigured (%d)\n",
+                                __func__, i, val[0] & 0x3);
+
+               val[0] = val[0] >> 2;
+       }
+
+       return num_phy_chans_avail;
+}
+
+static struct d40_base * __init d40_hw_detect_init(struct platform_device *pdev)
+{
+       static const struct d40_reg_val dma_id_regs[] = {
+               /* Peripheral Id */
+               { .reg = D40_DREG_PERIPHID0, .val = 0x0040},
+               { .reg = D40_DREG_PERIPHID1, .val = 0x0000},
+               /*
+                * D40_DREG_PERIPHID2 Depends on HW revision:
+                *  MOP500/HREF ED has 0x0008,
+                *  ? has 0x0018,
+                *  HREF V1 has 0x0028
+                */
+               { .reg = D40_DREG_PERIPHID3, .val = 0x0000},
+
+               /* PCell Id */
+               { .reg = D40_DREG_CELLID0, .val = 0x000d},
+               { .reg = D40_DREG_CELLID1, .val = 0x00f0},
+               { .reg = D40_DREG_CELLID2, .val = 0x0005},
+               { .reg = D40_DREG_CELLID3, .val = 0x00b1}
+       };
+       struct stedma40_platform_data *plat_data;
+       struct clk *clk = NULL;
+       void __iomem *virtbase = NULL;
+       struct resource *res = NULL;
+       struct d40_base *base = NULL;
+       int num_log_chans = 0;
+       int num_phy_chans;
+       int i;
+
+       clk = clk_get(&pdev->dev, NULL);
+
+       if (IS_ERR(clk)) {
+               dev_err(&pdev->dev, "[%s] No matching clock found\n",
+                       __func__);
+               goto failure;
+       }
+
+       clk_enable(clk);
+
+       /* Get IO for DMAC base address */
+       res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "base");
+       if (!res)
+               goto failure;
+
+       if (request_mem_region(res->start, resource_size(res),
+                              D40_NAME " I/O base") == NULL)
+               goto failure;
+
+       virtbase = ioremap(res->start, resource_size(res));
+       if (!virtbase)
+               goto failure;
+
+       /* HW version check */
+       for (i = 0; i < ARRAY_SIZE(dma_id_regs); i++) {
+               if (dma_id_regs[i].val !=
+                   readl(virtbase + dma_id_regs[i].reg)) {
+                       dev_err(&pdev->dev,
+                               "[%s] Unknown hardware! Expected 0x%x at 0x%x but got 0x%x\n",
+                               __func__,
+                               dma_id_regs[i].val,
+                               dma_id_regs[i].reg,
+                               readl(virtbase + dma_id_regs[i].reg));
+                       goto failure;
+               }
+       }
+
+       i = readl(virtbase + D40_DREG_PERIPHID2);
+
+       if ((i & 0xf) != D40_PERIPHID2_DESIGNER) {
+               dev_err(&pdev->dev,
+                       "[%s] Unknown designer! Got %x wanted %x\n",
+                       __func__, i & 0xf, D40_PERIPHID2_DESIGNER);
+               goto failure;
+       }
+
+       /* The number of physical channels on this HW */
+       num_phy_chans = 4 * (readl(virtbase + D40_DREG_ICFG) & 0x7) + 4;
+
+       dev_info(&pdev->dev, "hardware revision: %d @ 0x%x\n",
+                (i >> 4) & 0xf, res->start);
+
+       plat_data = pdev->dev.platform_data;
+
+       /* Count the number of logical channels in use */
+       for (i = 0; i < plat_data->dev_len; i++)
+               if (plat_data->dev_rx[i] != 0)
+                       num_log_chans++;
+
+       for (i = 0; i < plat_data->dev_len; i++)
+               if (plat_data->dev_tx[i] != 0)
+                       num_log_chans++;
+
+       base = kzalloc(ALIGN(sizeof(struct d40_base), 4) +
+                      (num_phy_chans + num_log_chans + plat_data->memcpy_len) *
+                      sizeof(struct d40_chan), GFP_KERNEL);
+
+       if (base == NULL) {
+               dev_err(&pdev->dev, "[%s] Out of memory\n", __func__);
+               goto failure;
+       }
+
+       base->clk = clk;
+       base->num_phy_chans = num_phy_chans;
+       base->num_log_chans = num_log_chans;
+       base->phy_start = res->start;
+       base->phy_size = resource_size(res);
+       base->virtbase = virtbase;
+       base->plat_data = plat_data;
+       base->dev = &pdev->dev;
+       base->phy_chans = ((void *)base) + ALIGN(sizeof(struct d40_base), 4);
+       base->log_chans = &base->phy_chans[num_phy_chans];
+
+       base->phy_res = kzalloc(num_phy_chans * sizeof(struct d40_phy_res),
+                               GFP_KERNEL);
+       if (!base->phy_res)
+               goto failure;
+
+       base->lookup_phy_chans = kzalloc(num_phy_chans *
+                                        sizeof(struct d40_chan *),
+                                        GFP_KERNEL);
+       if (!base->lookup_phy_chans)
+               goto failure;
+
+       if (num_log_chans + plat_data->memcpy_len) {
+               /*
+                * The max number of logical channels are event lines for all
+                * src devices and dst devices
+                */
+               base->lookup_log_chans = kzalloc(plat_data->dev_len * 2 *
+                                                sizeof(struct d40_chan *),
+                                                GFP_KERNEL);
+               if (!base->lookup_log_chans)
+                       goto failure;
+       }
+       base->lcla_pool.alloc_map = kzalloc(num_phy_chans * sizeof(u32),
+                                           GFP_KERNEL);
+       if (!base->lcla_pool.alloc_map)
+               goto failure;
+
+       return base;
+
+failure:
+       if (clk) {
+               clk_disable(clk);
+               clk_put(clk);
+       }
+       if (virtbase)
+               iounmap(virtbase);
+       if (res)
+               release_mem_region(res->start,
+                                  resource_size(res));
+       if (virtbase)
+               iounmap(virtbase);
+
+       if (base) {
+               kfree(base->lcla_pool.alloc_map);
+               kfree(base->lookup_log_chans);
+               kfree(base->lookup_phy_chans);
+               kfree(base->phy_res);
+               kfree(base);
+       }
+
+       return NULL;
+}
+
+static void __init d40_hw_init(struct d40_base *base)
+{
+
+       static const struct d40_reg_val dma_init_reg[] = {
+               /* Clock every part of the DMA block from start */
+               { .reg = D40_DREG_GCC,    .val = 0x0000ff01},
+
+               /* Interrupts on all logical channels */
+               { .reg = D40_DREG_LCMIS0, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCMIS1, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCMIS2, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCMIS3, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCICR0, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCICR1, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCICR2, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCICR3, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCTIS0, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCTIS1, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCTIS2, .val = 0xFFFFFFFF},
+               { .reg = D40_DREG_LCTIS3, .val = 0xFFFFFFFF}
+       };
+       int i;
+       u32 prmseo[2] = {0, 0};
+       u32 activeo[2] = {0xFFFFFFFF, 0xFFFFFFFF};
+       u32 pcmis = 0;
+       u32 pcicr = 0;
+
+       for (i = 0; i < ARRAY_SIZE(dma_init_reg); i++)
+               writel(dma_init_reg[i].val,
+                      base->virtbase + dma_init_reg[i].reg);
+
+       /* Configure all our dma channels to default settings */
+       for (i = 0; i < base->num_phy_chans; i++) {
+
+               activeo[i % 2] = activeo[i % 2] << 2;
+
+               if (base->phy_res[base->num_phy_chans - i - 1].allocated_src
+                   == D40_ALLOC_PHY) {
+                       activeo[i % 2] |= 3;
+                       continue;
+               }
+
+               /* Enable interrupt # */
+               pcmis = (pcmis << 1) | 1;
+
+               /* Clear interrupt # */
+               pcicr = (pcicr << 1) | 1;
+
+               /* Set channel to physical mode */
+               prmseo[i % 2] = prmseo[i % 2] << 2;
+               prmseo[i % 2] |= 1;
+
+       }
+
+       writel(prmseo[1], base->virtbase + D40_DREG_PRMSE);
+       writel(prmseo[0], base->virtbase + D40_DREG_PRMSO);
+       writel(activeo[1], base->virtbase + D40_DREG_ACTIVE);
+       writel(activeo[0], base->virtbase + D40_DREG_ACTIVO);
+
+       /* Write which interrupt to enable */
+       writel(pcmis, base->virtbase + D40_DREG_PCMIS);
+
+       /* Write which interrupt to clear */
+       writel(pcicr, base->virtbase + D40_DREG_PCICR);
+
+}
+
+static int __init d40_probe(struct platform_device *pdev)
+{
+       int err;
+       int ret = -ENOENT;
+       struct d40_base *base;
+       struct resource *res = NULL;
+       int num_reserved_chans;
+       u32 val;
+
+       base = d40_hw_detect_init(pdev);
+
+       if (!base)
+               goto failure;
+
+       num_reserved_chans = d40_phy_res_init(base);
+
+       platform_set_drvdata(pdev, base);
+
+       spin_lock_init(&base->interrupt_lock);
+       spin_lock_init(&base->execmd_lock);
+
+       /* Get IO for logical channel parameter address */
+       res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "lcpa");
+       if (!res) {
+               ret = -ENOENT;
+               dev_err(&pdev->dev,
+                       "[%s] No \"lcpa\" memory resource\n",
+                       __func__);
+               goto failure;
+       }
+       base->lcpa_size = resource_size(res);
+       base->phy_lcpa = res->start;
+
+       if (request_mem_region(res->start, resource_size(res),
+                              D40_NAME " I/O lcpa") == NULL) {
+               ret = -EBUSY;
+               dev_err(&pdev->dev,
+                       "[%s] Failed to request LCPA region 0x%x-0x%x\n",
+                       __func__, res->start, res->end);
+               goto failure;
+       }
+
+       /* We make use of ESRAM memory for this. */
+       val = readl(base->virtbase + D40_DREG_LCPA);
+       if (res->start != val && val != 0) {
+               dev_warn(&pdev->dev,
+                        "[%s] Mismatch LCPA dma 0x%x, def 0x%x\n",
+                        __func__, val, res->start);
+       } else
+               writel(res->start, base->virtbase + D40_DREG_LCPA);
+
+       base->lcpa_base = ioremap(res->start, resource_size(res));
+       if (!base->lcpa_base) {
+               ret = -ENOMEM;
+               dev_err(&pdev->dev,
+                       "[%s] Failed to ioremap LCPA region\n",
+                       __func__);
+               goto failure;
+       }
+       /* Get IO for logical channel link address */
+       res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "lcla");
+       if (!res) {
+               ret = -ENOENT;
+               dev_err(&pdev->dev,
+                       "[%s] No \"lcla\" resource defined\n",
+                       __func__);
+               goto failure;
+       }
+
+       base->lcla_pool.base_size = resource_size(res);
+       base->lcla_pool.phy = res->start;
+
+       if (request_mem_region(res->start, resource_size(res),
+                              D40_NAME " I/O lcla") == NULL) {
+               ret = -EBUSY;
+               dev_err(&pdev->dev,
+                       "[%s] Failed to request LCLA region 0x%x-0x%x\n",
+                       __func__, res->start, res->end);
+               goto failure;
+       }
+       val = readl(base->virtbase + D40_DREG_LCLA);
+       if (res->start != val && val != 0) {
+               dev_warn(&pdev->dev,
+                        "[%s] Mismatch LCLA dma 0x%x, def 0x%x\n",
+                        __func__, val, res->start);
+       } else
+               writel(res->start, base->virtbase + D40_DREG_LCLA);
+
+       base->lcla_pool.base = ioremap(res->start, resource_size(res));
+       if (!base->lcla_pool.base) {
+               ret = -ENOMEM;
+               dev_err(&pdev->dev,
+                       "[%s] Failed to ioremap LCLA 0x%x-0x%x\n",
+                       __func__, res->start, res->end);
+               goto failure;
+       }
+
+       spin_lock_init(&base->lcla_pool.lock);
+
+       base->lcla_pool.num_blocks = base->num_phy_chans;
+
+       base->irq = platform_get_irq(pdev, 0);
+
+       ret = request_irq(base->irq, d40_handle_interrupt, 0, D40_NAME, base);
+
+       if (ret) {
+               dev_err(&pdev->dev, "[%s] No IRQ defined\n", __func__);
+               goto failure;
+       }
+
+       err = d40_dmaengine_init(base, num_reserved_chans);
+       if (err)
+               goto failure;
+
+       d40_hw_init(base);
+
+       dev_info(base->dev, "initialized\n");
+       return 0;
+
+failure:
+       if (base) {
+               if (base->virtbase)
+                       iounmap(base->virtbase);
+               if (base->lcla_pool.phy)
+                       release_mem_region(base->lcla_pool.phy,
+                                          base->lcla_pool.base_size);
+               if (base->phy_lcpa)
+                       release_mem_region(base->phy_lcpa,
+                                          base->lcpa_size);
+               if (base->phy_start)
+                       release_mem_region(base->phy_start,
+                                          base->phy_size);
+               if (base->clk) {
+                       clk_disable(base->clk);
+                       clk_put(base->clk);
+               }
+
+               kfree(base->lcla_pool.alloc_map);
+               kfree(base->lookup_log_chans);
+               kfree(base->lookup_phy_chans);
+               kfree(base->phy_res);
+               kfree(base);
+       }
+
+       dev_err(&pdev->dev, "[%s] probe failed\n", __func__);
+       return ret;
+}
+
+static struct platform_driver d40_driver = {
+       .driver = {
+               .owner = THIS_MODULE,
+               .name  = D40_NAME,
+       },
+};
+
+int __init stedma40_init(void)
+{
+       return platform_driver_probe(&d40_driver, d40_probe);
+}
+arch_initcall(stedma40_init);
diff --git a/drivers/dma/ste_dma40_ll.c b/drivers/dma/ste_dma40_ll.c
new file mode 100644 (file)
index 0000000..561fdd8
--- /dev/null
@@ -0,0 +1,454 @@
+/*
+ * driver/dma/ste_dma40_ll.c
+ *
+ * Copyright (C) ST-Ericsson 2007-2010
+ * License terms: GNU General Public License (GPL) version 2
+ * Author: Per Friden <per.friden@stericsson.com>
+ * Author: Jonas Aaberg <jonas.aberg@stericsson.com>
+ */
+
+#include <linux/kernel.h>
+#include <plat/ste_dma40.h>
+
+#include "ste_dma40_ll.h"
+
+/* Sets up proper LCSP1 and LCSP3 register for a logical channel */
+void d40_log_cfg(struct stedma40_chan_cfg *cfg,
+                u32 *lcsp1, u32 *lcsp3)
+{
+       u32 l3 = 0; /* dst */
+       u32 l1 = 0; /* src */
+
+       /* src is mem? -> increase address pos */
+       if (cfg->dir ==  STEDMA40_MEM_TO_PERIPH ||
+           cfg->dir ==  STEDMA40_MEM_TO_MEM)
+               l1 |= 1 << D40_MEM_LCSP1_SCFG_INCR_POS;
+
+       /* dst is mem? -> increase address pos */
+       if (cfg->dir ==  STEDMA40_PERIPH_TO_MEM ||
+           cfg->dir ==  STEDMA40_MEM_TO_MEM)
+               l3 |= 1 << D40_MEM_LCSP3_DCFG_INCR_POS;
+
+       /* src is hw? -> master port 1 */
+       if (cfg->dir ==  STEDMA40_PERIPH_TO_MEM ||
+           cfg->dir ==  STEDMA40_PERIPH_TO_PERIPH)
+               l1 |= 1 << D40_MEM_LCSP1_SCFG_MST_POS;
+
+       /* dst is hw? -> master port 1 */
+       if (cfg->dir ==  STEDMA40_MEM_TO_PERIPH ||
+           cfg->dir ==  STEDMA40_PERIPH_TO_PERIPH)
+               l3 |= 1 << D40_MEM_LCSP3_DCFG_MST_POS;
+
+       l3 |= 1 << D40_MEM_LCSP3_DCFG_TIM_POS;
+       l3 |= 1 << D40_MEM_LCSP3_DCFG_EIM_POS;
+       l3 |= cfg->dst_info.psize << D40_MEM_LCSP3_DCFG_PSIZE_POS;
+       l3 |= cfg->dst_info.data_width << D40_MEM_LCSP3_DCFG_ESIZE_POS;
+       l3 |= 1 << D40_MEM_LCSP3_DTCP_POS;
+
+       l1 |= 1 << D40_MEM_LCSP1_SCFG_EIM_POS;
+       l1 |= cfg->src_info.psize << D40_MEM_LCSP1_SCFG_PSIZE_POS;
+       l1 |= cfg->src_info.data_width << D40_MEM_LCSP1_SCFG_ESIZE_POS;
+       l1 |= 1 << D40_MEM_LCSP1_STCP_POS;
+
+       *lcsp1 = l1;
+       *lcsp3 = l3;
+
+}
+
+/* Sets up SRC and DST CFG register for both logical and physical channels */
+void d40_phy_cfg(struct stedma40_chan_cfg *cfg,
+                u32 *src_cfg, u32 *dst_cfg, bool is_log)
+{
+       u32 src = 0;
+       u32 dst = 0;
+
+       if (!is_log) {
+               /* Physical channel */
+               if ((cfg->dir ==  STEDMA40_PERIPH_TO_MEM) ||
+                   (cfg->dir == STEDMA40_PERIPH_TO_PERIPH)) {
+                       /* Set master port to 1 */
+                       src |= 1 << D40_SREG_CFG_MST_POS;
+                       src |= D40_TYPE_TO_EVENT(cfg->src_dev_type);
+
+                       if (cfg->src_info.flow_ctrl == STEDMA40_NO_FLOW_CTRL)
+                               src |= 1 << D40_SREG_CFG_PHY_TM_POS;
+                       else
+                               src |= 3 << D40_SREG_CFG_PHY_TM_POS;
+               }
+               if ((cfg->dir ==  STEDMA40_MEM_TO_PERIPH) ||
+                   (cfg->dir == STEDMA40_PERIPH_TO_PERIPH)) {
+                       /* Set master port to 1 */
+                       dst |= 1 << D40_SREG_CFG_MST_POS;
+                       dst |= D40_TYPE_TO_EVENT(cfg->dst_dev_type);
+
+                       if (cfg->dst_info.flow_ctrl == STEDMA40_NO_FLOW_CTRL)
+                               dst |= 1 << D40_SREG_CFG_PHY_TM_POS;
+                       else
+                               dst |= 3 << D40_SREG_CFG_PHY_TM_POS;
+               }
+               /* Interrupt on end of transfer for destination */
+               dst |= 1 << D40_SREG_CFG_TIM_POS;
+
+               /* Generate interrupt on error */
+               src |= 1 << D40_SREG_CFG_EIM_POS;
+               dst |= 1 << D40_SREG_CFG_EIM_POS;
+
+               /* PSIZE */
+               if (cfg->src_info.psize != STEDMA40_PSIZE_PHY_1) {
+                       src |= 1 << D40_SREG_CFG_PHY_PEN_POS;
+                       src |= cfg->src_info.psize << D40_SREG_CFG_PSIZE_POS;
+               }
+               if (cfg->dst_info.psize != STEDMA40_PSIZE_PHY_1) {
+                       dst |= 1 << D40_SREG_CFG_PHY_PEN_POS;
+                       dst |= cfg->dst_info.psize << D40_SREG_CFG_PSIZE_POS;
+               }
+
+               /* Element size */
+               src |= cfg->src_info.data_width << D40_SREG_CFG_ESIZE_POS;
+               dst |= cfg->dst_info.data_width << D40_SREG_CFG_ESIZE_POS;
+
+       } else {
+               /* Logical channel */
+               dst |= 1 << D40_SREG_CFG_LOG_GIM_POS;
+               src |= 1 << D40_SREG_CFG_LOG_GIM_POS;
+       }
+
+       if (cfg->channel_type & STEDMA40_HIGH_PRIORITY_CHANNEL) {
+               src |= 1 << D40_SREG_CFG_PRI_POS;
+               dst |= 1 << D40_SREG_CFG_PRI_POS;
+       }
+
+       src |= cfg->src_info.endianess << D40_SREG_CFG_LBE_POS;
+       dst |= cfg->dst_info.endianess << D40_SREG_CFG_LBE_POS;
+
+       *src_cfg = src;
+       *dst_cfg = dst;
+}
+
+int d40_phy_fill_lli(struct d40_phy_lli *lli,
+                    dma_addr_t data,
+                    u32 data_size,
+                    int psize,
+                    dma_addr_t next_lli,
+                    u32 reg_cfg,
+                    bool term_int,
+                    u32 data_width,
+                    bool is_device)
+{
+       int num_elems;
+
+       if (psize == STEDMA40_PSIZE_PHY_1)
+               num_elems = 1;
+       else
+               num_elems = 2 << psize;
+
+       /*
+        * Size is 16bit. data_width is 8, 16, 32 or 64 bit
+        * Block large than 64 KiB must be split.
+        */
+       if (data_size > (0xffff << data_width))
+               return -EINVAL;
+
+       /* Must be aligned */
+       if (!IS_ALIGNED(data, 0x1 << data_width))
+               return -EINVAL;
+
+       /* Transfer size can't be smaller than (num_elms * elem_size) */
+       if (data_size < num_elems * (0x1 << data_width))
+               return -EINVAL;
+
+       /* The number of elements. IE now many chunks */
+       lli->reg_elt = (data_size >> data_width) << D40_SREG_ELEM_PHY_ECNT_POS;
+
+       /*
+        * Distance to next element sized entry.
+        * Usually the size of the element unless you want gaps.
+        */
+       if (!is_device)
+               lli->reg_elt |= (0x1 << data_width) <<
+                       D40_SREG_ELEM_PHY_EIDX_POS;
+
+       /* Where the data is */
+       lli->reg_ptr = data;
+       lli->reg_cfg = reg_cfg;
+
+       /* If this scatter list entry is the last one, no next link */
+       if (next_lli == 0)
+               lli->reg_lnk = 0x1 << D40_SREG_LNK_PHY_TCP_POS;
+       else
+               lli->reg_lnk = next_lli;
+
+       /* Set/clear interrupt generation on this link item.*/
+       if (term_int)
+               lli->reg_cfg |= 0x1 << D40_SREG_CFG_TIM_POS;
+       else
+               lli->reg_cfg &= ~(0x1 << D40_SREG_CFG_TIM_POS);
+
+       /* Post link */
+       lli->reg_lnk |= 0 << D40_SREG_LNK_PHY_PRE_POS;
+
+       return 0;
+}
+
+int d40_phy_sg_to_lli(struct scatterlist *sg,
+                     int sg_len,
+                     dma_addr_t target,
+                     struct d40_phy_lli *lli,
+                     dma_addr_t lli_phys,
+                     u32 reg_cfg,
+                     u32 data_width,
+                     int psize,
+                     bool term_int)
+{
+       int total_size = 0;
+       int i;
+       struct scatterlist *current_sg = sg;
+       dma_addr_t next_lli_phys;
+       dma_addr_t dst;
+       int err = 0;
+
+       for_each_sg(sg, current_sg, sg_len, i) {
+
+               total_size += sg_dma_len(current_sg);
+
+               /* If this scatter list entry is the last one, no next link */
+               if (sg_len - 1 == i)
+                       next_lli_phys = 0;
+               else
+                       next_lli_phys = ALIGN(lli_phys + (i + 1) *
+                                             sizeof(struct d40_phy_lli),
+                                             D40_LLI_ALIGN);
+
+               if (target)
+                       dst = target;
+               else
+                       dst = sg_phys(current_sg);
+
+               err = d40_phy_fill_lli(&lli[i],
+                                      dst,
+                                      sg_dma_len(current_sg),
+                                      psize,
+                                      next_lli_phys,
+                                      reg_cfg,
+                                      !next_lli_phys,
+                                      data_width,
+                                      target == dst);
+               if (err)
+                       goto err;
+       }
+
+       return total_size;
+ err:
+       return err;
+}
+
+
+void d40_phy_lli_write(void __iomem *virtbase,
+                      u32 phy_chan_num,
+                      struct d40_phy_lli *lli_dst,
+                      struct d40_phy_lli *lli_src)
+{
+
+       writel(lli_src->reg_cfg, virtbase + D40_DREG_PCBASE +
+              phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SSCFG);
+       writel(lli_src->reg_elt, virtbase + D40_DREG_PCBASE +
+              phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SSELT);
+       writel(lli_src->reg_ptr, virtbase + D40_DREG_PCBASE +
+              phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SSPTR);
+       writel(lli_src->reg_lnk, virtbase + D40_DREG_PCBASE +
+              phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SSLNK);
+
+       writel(lli_dst->reg_cfg, virtbase + D40_DREG_PCBASE +
+              phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SDCFG);
+       writel(lli_dst->reg_elt, virtbase + D40_DREG_PCBASE +
+              phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SDELT);
+       writel(lli_dst->reg_ptr, virtbase + D40_DREG_PCBASE +
+              phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SDPTR);
+       writel(lli_dst->reg_lnk, virtbase + D40_DREG_PCBASE +
+              phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SDLNK);
+
+}
+
+/* DMA logical lli operations */
+
+void d40_log_fill_lli(struct d40_log_lli *lli,
+                     dma_addr_t data, u32 data_size,
+                     u32 lli_next_off, u32 reg_cfg,
+                     u32 data_width,
+                     bool term_int, bool addr_inc)
+{
+       lli->lcsp13 = reg_cfg;
+
+       /* The number of elements to transfer */
+       lli->lcsp02 = ((data_size >> data_width) <<
+                      D40_MEM_LCSP0_ECNT_POS) & D40_MEM_LCSP0_ECNT_MASK;
+       /* 16 LSBs address of the current element */
+       lli->lcsp02 |= data & D40_MEM_LCSP0_SPTR_MASK;
+       /* 16 MSBs address of the current element */
+       lli->lcsp13 |= data & D40_MEM_LCSP1_SPTR_MASK;
+
+       if (addr_inc)
+               lli->lcsp13 |= D40_MEM_LCSP1_SCFG_INCR_MASK;
+
+       lli->lcsp13 |= D40_MEM_LCSP3_DTCP_MASK;
+       /* If this scatter list entry is the last one, no next link */
+       lli->lcsp13 |= (lli_next_off << D40_MEM_LCSP1_SLOS_POS) &
+               D40_MEM_LCSP1_SLOS_MASK;
+
+       if (term_int)
+               lli->lcsp13 |= D40_MEM_LCSP1_SCFG_TIM_MASK;
+       else
+               lli->lcsp13 &= ~D40_MEM_LCSP1_SCFG_TIM_MASK;
+}
+
+int d40_log_sg_to_dev(struct d40_lcla_elem *lcla,
+                     struct scatterlist *sg,
+                     int sg_len,
+                     struct d40_log_lli_bidir *lli,
+                     struct d40_def_lcsp *lcsp,
+                     u32 src_data_width,
+                     u32 dst_data_width,
+                     enum dma_data_direction direction,
+                     bool term_int, dma_addr_t dev_addr, int max_len,
+                     int llis_per_log)
+{
+       int total_size = 0;
+       struct scatterlist *current_sg = sg;
+       int i;
+       u32 next_lli_off_dst;
+       u32 next_lli_off_src;
+
+       next_lli_off_src = 0;
+       next_lli_off_dst = 0;
+
+       for_each_sg(sg, current_sg, sg_len, i) {
+               total_size += sg_dma_len(current_sg);
+
+               /*
+                * If this scatter list entry is the last one or
+                * max length, terminate link.
+                */
+               if (sg_len - 1 == i || ((i+1) % max_len == 0)) {
+                       next_lli_off_src = 0;
+                       next_lli_off_dst = 0;
+               } else {
+                       if (next_lli_off_dst == 0 &&
+                           next_lli_off_src == 0) {
+                               /* The first lli will be at next_lli_off */
+                               next_lli_off_dst = (lcla->dst_id *
+                                                   llis_per_log + 1);
+                               next_lli_off_src = (lcla->src_id *
+                                                   llis_per_log + 1);
+                       } else {
+                               next_lli_off_dst++;
+                               next_lli_off_src++;
+                       }
+               }
+
+               if (direction == DMA_TO_DEVICE) {
+                       d40_log_fill_lli(&lli->src[i],
+                                        sg_phys(current_sg),
+                                        sg_dma_len(current_sg),
+                                        next_lli_off_src,
+                                        lcsp->lcsp1, src_data_width,
+                                        term_int && !next_lli_off_src,
+                                        true);
+                       d40_log_fill_lli(&lli->dst[i],
+                                        dev_addr,
+                                        sg_dma_len(current_sg),
+                                        next_lli_off_dst,
+                                        lcsp->lcsp3, dst_data_width,
+                                        /* No next == terminal interrupt */
+                                        term_int && !next_lli_off_dst,
+                                        false);
+               } else {
+                       d40_log_fill_lli(&lli->dst[i],
+                                        sg_phys(current_sg),
+                                        sg_dma_len(current_sg),
+                                        next_lli_off_dst,
+                                        lcsp->lcsp3, dst_data_width,
+                                        /* No next == terminal interrupt */
+                                        term_int && !next_lli_off_dst,
+                                        true);
+                       d40_log_fill_lli(&lli->src[i],
+                                        dev_addr,
+                                        sg_dma_len(current_sg),
+                                        next_lli_off_src,
+                                        lcsp->lcsp1, src_data_width,
+                                        term_int && !next_lli_off_src,
+                                        false);
+               }
+       }
+       return total_size;
+}
+
+int d40_log_sg_to_lli(int lcla_id,
+                     struct scatterlist *sg,
+                     int sg_len,
+                     struct d40_log_lli *lli_sg,
+                     u32 lcsp13, /* src or dst*/
+                     u32 data_width,
+                     bool term_int, int max_len, int llis_per_log)
+{
+       int total_size = 0;
+       struct scatterlist *current_sg = sg;
+       int i;
+       u32 next_lli_off = 0;
+
+       for_each_sg(sg, current_sg, sg_len, i) {
+               total_size += sg_dma_len(current_sg);
+
+               /*
+                * If this scatter list entry is the last one or
+                * max length, terminate link.
+                */
+               if (sg_len - 1 == i || ((i+1) % max_len == 0))
+                       next_lli_off = 0;
+               else {
+                       if (next_lli_off == 0)
+                               /* The first lli will be at next_lli_off */
+                               next_lli_off = lcla_id * llis_per_log + 1;
+                       else
+                               next_lli_off++;
+               }
+
+               d40_log_fill_lli(&lli_sg[i],
+                                sg_phys(current_sg),
+                                sg_dma_len(current_sg),
+                                next_lli_off,
+                                lcsp13, data_width,
+                                term_int && !next_lli_off,
+                                true);
+       }
+       return total_size;
+}
+
+void d40_log_lli_write(struct d40_log_lli_full *lcpa,
+                      struct d40_log_lli *lcla_src,
+                      struct d40_log_lli *lcla_dst,
+                      struct d40_log_lli *lli_dst,
+                      struct d40_log_lli *lli_src,
+                      int llis_per_log)
+{
+       u32 slos = 0;
+       u32 dlos = 0;
+       int i;
+
+       lcpa->lcsp0 = lli_src->lcsp02;
+       lcpa->lcsp1 = lli_src->lcsp13;
+       lcpa->lcsp2 = lli_dst->lcsp02;
+       lcpa->lcsp3 = lli_dst->lcsp13;
+
+       slos = lli_src->lcsp13 & D40_MEM_LCSP1_SLOS_MASK;
+       dlos = lli_dst->lcsp13 & D40_MEM_LCSP3_DLOS_MASK;
+
+       for (i = 0; (i < llis_per_log) && slos && dlos; i++) {
+               writel(lli_src[i+1].lcsp02, &lcla_src[i].lcsp02);
+               writel(lli_src[i+1].lcsp13, &lcla_src[i].lcsp13);
+               writel(lli_dst[i+1].lcsp02, &lcla_dst[i].lcsp02);
+               writel(lli_dst[i+1].lcsp13, &lcla_dst[i].lcsp13);
+
+               slos = lli_src[i+1].lcsp13 & D40_MEM_LCSP1_SLOS_MASK;
+               dlos = lli_dst[i+1].lcsp13 & D40_MEM_LCSP3_DLOS_MASK;
+       }
+}
diff --git a/drivers/dma/ste_dma40_ll.h b/drivers/dma/ste_dma40_ll.h
new file mode 100644 (file)
index 0000000..2029280
--- /dev/null
@@ -0,0 +1,354 @@
+/*
+ * driver/dma/ste_dma40_ll.h
+ *
+ * Copyright (C) ST-Ericsson 2007-2010
+ * License terms: GNU General Public License (GPL) version 2
+ * Author: Per Friden <per.friden@stericsson.com>
+ * Author: Jonas Aaberg <jonas.aberg@stericsson.com>
+ */
+#ifndef STE_DMA40_LL_H
+#define STE_DMA40_LL_H
+
+#define D40_DREG_PCBASE                0x400
+#define D40_DREG_PCDELTA       (8 * 4)
+#define D40_LLI_ALIGN          16 /* LLI alignment must be 16 bytes. */
+
+#define D40_TYPE_TO_GROUP(type) (type / 16)
+#define D40_TYPE_TO_EVENT(type) (type % 16)
+
+/* Most bits of the CFG register are the same in log as in phy mode */
+#define D40_SREG_CFG_MST_POS           15
+#define D40_SREG_CFG_TIM_POS           14
+#define D40_SREG_CFG_EIM_POS           13
+#define D40_SREG_CFG_LOG_INCR_POS      12
+#define D40_SREG_CFG_PHY_PEN_POS       12
+#define D40_SREG_CFG_PSIZE_POS         10
+#define D40_SREG_CFG_ESIZE_POS          8
+#define D40_SREG_CFG_PRI_POS            7
+#define D40_SREG_CFG_LBE_POS            6
+#define D40_SREG_CFG_LOG_GIM_POS        5
+#define D40_SREG_CFG_LOG_MFU_POS        4
+#define D40_SREG_CFG_PHY_TM_POS                 4
+#define D40_SREG_CFG_PHY_EVTL_POS       0
+
+
+/* Standard channel parameters - basic mode (element register) */
+#define D40_SREG_ELEM_PHY_ECNT_POS     16
+#define D40_SREG_ELEM_PHY_EIDX_POS      0
+
+#define D40_SREG_ELEM_PHY_ECNT_MASK    (0xFFFF << D40_SREG_ELEM_PHY_ECNT_POS)
+
+/* Standard channel parameters - basic mode (Link register) */
+#define D40_SREG_LNK_PHY_TCP_POS       0
+#define D40_SREG_LNK_PHY_LMP_POS       1
+#define D40_SREG_LNK_PHY_PRE_POS       2
+/*
+ * Source  destination link address. Contains the
+ * 29-bit byte word aligned address of the reload area.
+ */
+#define D40_SREG_LNK_PHYS_LNK_MASK     0xFFFFFFF8UL
+
+/* Standard basic channel logical mode */
+
+/* Element register */
+#define D40_SREG_ELEM_LOG_ECNT_POS     16
+#define D40_SREG_ELEM_LOG_LIDX_POS      8
+#define D40_SREG_ELEM_LOG_LOS_POS       1
+#define D40_SREG_ELEM_LOG_TCP_POS       0
+
+#define D40_SREG_ELEM_LOG_LIDX_MASK    (0xFF << D40_SREG_ELEM_LOG_LIDX_POS)
+
+/* Link register */
+#define D40_DEACTIVATE_EVENTLINE       0x0
+#define D40_ACTIVATE_EVENTLINE         0x1
+#define D40_EVENTLINE_POS(i)           (2 * i)
+#define D40_EVENTLINE_MASK(i)          (0x3 << D40_EVENTLINE_POS(i))
+
+/* Standard basic channel logical params in memory */
+
+/* LCSP0 */
+#define D40_MEM_LCSP0_ECNT_POS         16
+#define D40_MEM_LCSP0_SPTR_POS          0
+
+#define D40_MEM_LCSP0_ECNT_MASK                (0xFFFF << D40_MEM_LCSP0_ECNT_POS)
+#define D40_MEM_LCSP0_SPTR_MASK                (0xFFFF << D40_MEM_LCSP0_SPTR_POS)
+
+/* LCSP1 */
+#define D40_MEM_LCSP1_SPTR_POS         16
+#define D40_MEM_LCSP1_SCFG_MST_POS     15
+#define D40_MEM_LCSP1_SCFG_TIM_POS     14
+#define D40_MEM_LCSP1_SCFG_EIM_POS     13
+#define D40_MEM_LCSP1_SCFG_INCR_POS    12
+#define D40_MEM_LCSP1_SCFG_PSIZE_POS   10
+#define D40_MEM_LCSP1_SCFG_ESIZE_POS    8
+#define D40_MEM_LCSP1_SLOS_POS          1
+#define D40_MEM_LCSP1_STCP_POS          0
+
+#define D40_MEM_LCSP1_SPTR_MASK                (0xFFFF << D40_MEM_LCSP1_SPTR_POS)
+#define D40_MEM_LCSP1_SCFG_TIM_MASK    (0x1 << D40_MEM_LCSP1_SCFG_TIM_POS)
+#define D40_MEM_LCSP1_SCFG_INCR_MASK   (0x1 << D40_MEM_LCSP1_SCFG_INCR_POS)
+#define D40_MEM_LCSP1_SCFG_PSIZE_MASK  (0x3 << D40_MEM_LCSP1_SCFG_PSIZE_POS)
+#define D40_MEM_LCSP1_SLOS_MASK                (0x7F << D40_MEM_LCSP1_SLOS_POS)
+#define D40_MEM_LCSP1_STCP_MASK                (0x1 << D40_MEM_LCSP1_STCP_POS)
+
+/* LCSP2 */
+#define D40_MEM_LCSP2_ECNT_POS         16
+
+#define D40_MEM_LCSP2_ECNT_MASK                (0xFFFF << D40_MEM_LCSP2_ECNT_POS)
+
+/* LCSP3 */
+#define D40_MEM_LCSP3_DCFG_MST_POS     15
+#define D40_MEM_LCSP3_DCFG_TIM_POS     14
+#define D40_MEM_LCSP3_DCFG_EIM_POS     13
+#define D40_MEM_LCSP3_DCFG_INCR_POS    12
+#define D40_MEM_LCSP3_DCFG_PSIZE_POS   10
+#define D40_MEM_LCSP3_DCFG_ESIZE_POS    8
+#define D40_MEM_LCSP3_DLOS_POS          1
+#define D40_MEM_LCSP3_DTCP_POS          0
+
+#define D40_MEM_LCSP3_DLOS_MASK                (0x7F << D40_MEM_LCSP3_DLOS_POS)
+#define D40_MEM_LCSP3_DTCP_MASK                (0x1 << D40_MEM_LCSP3_DTCP_POS)
+
+
+/* Standard channel parameter register offsets */
+#define D40_CHAN_REG_SSCFG     0x00
+#define D40_CHAN_REG_SSELT     0x04
+#define D40_CHAN_REG_SSPTR     0x08
+#define D40_CHAN_REG_SSLNK     0x0C
+#define D40_CHAN_REG_SDCFG     0x10
+#define D40_CHAN_REG_SDELT     0x14
+#define D40_CHAN_REG_SDPTR     0x18
+#define D40_CHAN_REG_SDLNK     0x1C
+
+/* DMA Register Offsets */
+#define D40_DREG_GCC           0x000
+#define D40_DREG_PRTYP         0x004
+#define D40_DREG_PRSME         0x008
+#define D40_DREG_PRSMO         0x00C
+#define D40_DREG_PRMSE         0x010
+#define D40_DREG_PRMSO         0x014
+#define D40_DREG_PRMOE         0x018
+#define D40_DREG_PRMOO         0x01C
+#define D40_DREG_LCPA          0x020
+#define D40_DREG_LCLA          0x024
+#define D40_DREG_ACTIVE                0x050
+#define D40_DREG_ACTIVO                0x054
+#define D40_DREG_FSEB1         0x058
+#define D40_DREG_FSEB2         0x05C
+#define D40_DREG_PCMIS         0x060
+#define D40_DREG_PCICR         0x064
+#define D40_DREG_PCTIS         0x068
+#define D40_DREG_PCEIS         0x06C
+#define D40_DREG_LCMIS0                0x080
+#define D40_DREG_LCMIS1                0x084
+#define D40_DREG_LCMIS2                0x088
+#define D40_DREG_LCMIS3                0x08C
+#define D40_DREG_LCICR0                0x090
+#define D40_DREG_LCICR1                0x094
+#define D40_DREG_LCICR2                0x098
+#define D40_DREG_LCICR3                0x09C
+#define D40_DREG_LCTIS0                0x0A0
+#define D40_DREG_LCTIS1                0x0A4
+#define D40_DREG_LCTIS2                0x0A8
+#define D40_DREG_LCTIS3                0x0AC
+#define D40_DREG_LCEIS0                0x0B0
+#define D40_DREG_LCEIS1                0x0B4
+#define D40_DREG_LCEIS2                0x0B8
+#define D40_DREG_LCEIS3                0x0BC
+#define D40_DREG_STFU          0xFC8
+#define D40_DREG_ICFG          0xFCC
+#define D40_DREG_PERIPHID0     0xFE0
+#define D40_DREG_PERIPHID1     0xFE4
+#define D40_DREG_PERIPHID2     0xFE8
+#define D40_DREG_PERIPHID3     0xFEC
+#define D40_DREG_CELLID0       0xFF0
+#define D40_DREG_CELLID1       0xFF4
+#define D40_DREG_CELLID2       0xFF8
+#define D40_DREG_CELLID3       0xFFC
+
+/* LLI related structures */
+
+/**
+ * struct d40_phy_lli - The basic configration register for each physical
+ * channel.
+ *
+ * @reg_cfg: The configuration register.
+ * @reg_elt: The element register.
+ * @reg_ptr: The pointer register.
+ * @reg_lnk: The link register.
+ *
+ * These registers are set up for both physical and logical transfers
+ * Note that the bit in each register means differently in logical and
+ * physical(standard) mode.
+ *
+ * This struct must be 16 bytes aligned, and only contain physical registers
+ * since it will be directly accessed by the DMA.
+ */
+struct d40_phy_lli {
+       u32 reg_cfg;
+       u32 reg_elt;
+       u32 reg_ptr;
+       u32 reg_lnk;
+};
+
+/**
+ * struct d40_phy_lli_bidir - struct for a transfer.
+ *
+ * @src: Register settings for src channel.
+ * @dst: Register settings for dst channel.
+ * @dst_addr: Physical destination address.
+ * @src_addr: Physical source address.
+ *
+ * All DMA transfers have a source and a destination.
+ */
+
+struct d40_phy_lli_bidir {
+       struct d40_phy_lli      *src;
+       struct d40_phy_lli      *dst;
+       dma_addr_t               dst_addr;
+       dma_addr_t               src_addr;
+};
+
+
+/**
+ * struct d40_log_lli - logical lli configuration
+ *
+ * @lcsp02: Either maps to register lcsp0 if src or lcsp2 if dst.
+ * @lcsp13: Either maps to register lcsp1 if src or lcsp3 if dst.
+ *
+ * This struct must be 8 bytes aligned since it will be accessed directy by
+ * the DMA. Never add any none hw mapped registers to this struct.
+ */
+
+struct d40_log_lli {
+       u32 lcsp02;
+       u32 lcsp13;
+};
+
+/**
+ * struct d40_log_lli_bidir - For both src and dst
+ *
+ * @src: pointer to src lli configuration.
+ * @dst: pointer to dst lli configuration.
+ *
+ * You always have a src and a dst when doing DMA transfers.
+ */
+
+struct d40_log_lli_bidir {
+       struct d40_log_lli *src;
+       struct d40_log_lli *dst;
+};
+
+/**
+ * struct d40_log_lli_full - LCPA layout
+ *
+ * @lcsp0: Logical Channel Standard Param 0 - Src.
+ * @lcsp1: Logical Channel Standard Param 1 - Src.
+ * @lcsp2: Logical Channel Standard Param 2 - Dst.
+ * @lcsp3: Logical Channel Standard Param 3 - Dst.
+ *
+ * This struct maps to LCPA physical memory layout. Must map to
+ * the hw.
+ */
+struct d40_log_lli_full {
+       u32 lcsp0;
+       u32 lcsp1;
+       u32 lcsp2;
+       u32 lcsp3;
+};
+
+/**
+ * struct d40_def_lcsp - Default LCSP1 and LCSP3 settings
+ *
+ * @lcsp3: The default configuration for dst.
+ * @lcsp1: The default configuration for src.
+ */
+struct d40_def_lcsp {
+       u32 lcsp3;
+       u32 lcsp1;
+};
+
+/**
+ * struct d40_lcla_elem - Info for one LCA element.
+ *
+ * @src_id: logical channel src id
+ * @dst_id: logical channel dst id
+ * @src: LCPA formated src parameters
+ * @dst: LCPA formated dst parameters
+ *
+ */
+struct d40_lcla_elem {
+       int                     src_id;
+       int                     dst_id;
+       struct d40_log_lli     *src;
+       struct d40_log_lli     *dst;
+};
+
+/* Physical channels */
+
+void d40_phy_cfg(struct stedma40_chan_cfg *cfg,
+                u32 *src_cfg, u32 *dst_cfg, bool is_log);
+
+void d40_log_cfg(struct stedma40_chan_cfg *cfg,
+                u32 *lcsp1, u32 *lcsp2);
+
+int d40_phy_sg_to_lli(struct scatterlist *sg,
+                     int sg_len,
+                     dma_addr_t target,
+                     struct d40_phy_lli *lli,
+                     dma_addr_t lli_phys,
+                     u32 reg_cfg,
+                     u32 data_width,
+                     int psize,
+                     bool term_int);
+
+int d40_phy_fill_lli(struct d40_phy_lli *lli,
+                    dma_addr_t data,
+                    u32 data_size,
+                    int psize,
+                    dma_addr_t next_lli,
+                    u32 reg_cfg,
+                    bool term_int,
+                    u32 data_width,
+                    bool is_device);
+
+void d40_phy_lli_write(void __iomem *virtbase,
+                      u32 phy_chan_num,
+                      struct d40_phy_lli *lli_dst,
+                      struct d40_phy_lli *lli_src);
+
+/* Logical channels */
+
+void d40_log_fill_lli(struct d40_log_lli *lli,
+                     dma_addr_t data, u32 data_size,
+                     u32 lli_next_off, u32 reg_cfg,
+                     u32 data_width,
+                     bool term_int, bool addr_inc);
+
+int d40_log_sg_to_dev(struct d40_lcla_elem *lcla,
+                     struct scatterlist *sg,
+                     int sg_len,
+                     struct d40_log_lli_bidir *lli,
+                     struct d40_def_lcsp *lcsp,
+                     u32 src_data_width,
+                     u32 dst_data_width,
+                     enum dma_data_direction direction,
+                     bool term_int, dma_addr_t dev_addr, int max_len,
+                     int llis_per_log);
+
+void d40_log_lli_write(struct d40_log_lli_full *lcpa,
+                      struct d40_log_lli *lcla_src,
+                      struct d40_log_lli *lcla_dst,
+                      struct d40_log_lli *lli_dst,
+                      struct d40_log_lli *lli_src,
+                      int llis_per_log);
+
+int d40_log_sg_to_lli(int lcla_id,
+                     struct scatterlist *sg,
+                     int sg_len,
+                     struct d40_log_lli *lli_sg,
+                     u32 lcsp13, /* src or dst*/
+                     u32 data_width,
+                     bool term_int, int max_len, int llis_per_log);
+
+#endif /* STE_DMA40_LLI_H */