Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / dma / omap-dma.c
diff --git a/kernel/drivers/dma/omap-dma.c b/kernel/drivers/dma/omap-dma.c
new file mode 100644 (file)
index 0000000..167dbaf
--- /dev/null
@@ -0,0 +1,1233 @@
+/*
+ * OMAP DMAengine support
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+#include <linux/delay.h>
+#include <linux/dmaengine.h>
+#include <linux/dma-mapping.h>
+#include <linux/err.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/omap-dma.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+#include <linux/of_dma.h>
+#include <linux/of_device.h>
+
+#include "virt-dma.h"
+
+struct omap_dmadev {
+       struct dma_device ddev;
+       spinlock_t lock;
+       struct tasklet_struct task;
+       struct list_head pending;
+       void __iomem *base;
+       const struct omap_dma_reg *reg_map;
+       struct omap_system_dma_plat_info *plat;
+       bool legacy;
+       spinlock_t irq_lock;
+       uint32_t irq_enable_mask;
+       struct omap_chan *lch_map[32];
+};
+
+struct omap_chan {
+       struct virt_dma_chan vc;
+       struct list_head node;
+       void __iomem *channel_base;
+       const struct omap_dma_reg *reg_map;
+       uint32_t ccr;
+
+       struct dma_slave_config cfg;
+       unsigned dma_sig;
+       bool cyclic;
+       bool paused;
+
+       int dma_ch;
+       struct omap_desc *desc;
+       unsigned sgidx;
+};
+
+struct omap_sg {
+       dma_addr_t addr;
+       uint32_t en;            /* number of elements (24-bit) */
+       uint32_t fn;            /* number of frames (16-bit) */
+};
+
+struct omap_desc {
+       struct virt_dma_desc vd;
+       enum dma_transfer_direction dir;
+       dma_addr_t dev_addr;
+
+       int16_t fi;             /* for OMAP_DMA_SYNC_PACKET */
+       uint8_t es;             /* CSDP_DATA_TYPE_xxx */
+       uint32_t ccr;           /* CCR value */
+       uint16_t clnk_ctrl;     /* CLNK_CTRL value */
+       uint16_t cicr;          /* CICR value */
+       uint32_t csdp;          /* CSDP value */
+
+       unsigned sglen;
+       struct omap_sg sg[0];
+};
+
+enum {
+       CCR_FS                  = BIT(5),
+       CCR_READ_PRIORITY       = BIT(6),
+       CCR_ENABLE              = BIT(7),
+       CCR_AUTO_INIT           = BIT(8),       /* OMAP1 only */
+       CCR_REPEAT              = BIT(9),       /* OMAP1 only */
+       CCR_OMAP31_DISABLE      = BIT(10),      /* OMAP1 only */
+       CCR_SUSPEND_SENSITIVE   = BIT(8),       /* OMAP2+ only */
+       CCR_RD_ACTIVE           = BIT(9),       /* OMAP2+ only */
+       CCR_WR_ACTIVE           = BIT(10),      /* OMAP2+ only */
+       CCR_SRC_AMODE_CONSTANT  = 0 << 12,
+       CCR_SRC_AMODE_POSTINC   = 1 << 12,
+       CCR_SRC_AMODE_SGLIDX    = 2 << 12,
+       CCR_SRC_AMODE_DBLIDX    = 3 << 12,
+       CCR_DST_AMODE_CONSTANT  = 0 << 14,
+       CCR_DST_AMODE_POSTINC   = 1 << 14,
+       CCR_DST_AMODE_SGLIDX    = 2 << 14,
+       CCR_DST_AMODE_DBLIDX    = 3 << 14,
+       CCR_CONSTANT_FILL       = BIT(16),
+       CCR_TRANSPARENT_COPY    = BIT(17),
+       CCR_BS                  = BIT(18),
+       CCR_SUPERVISOR          = BIT(22),
+       CCR_PREFETCH            = BIT(23),
+       CCR_TRIGGER_SRC         = BIT(24),
+       CCR_BUFFERING_DISABLE   = BIT(25),
+       CCR_WRITE_PRIORITY      = BIT(26),
+       CCR_SYNC_ELEMENT        = 0,
+       CCR_SYNC_FRAME          = CCR_FS,
+       CCR_SYNC_BLOCK          = CCR_BS,
+       CCR_SYNC_PACKET         = CCR_BS | CCR_FS,
+
+       CSDP_DATA_TYPE_8        = 0,
+       CSDP_DATA_TYPE_16       = 1,
+       CSDP_DATA_TYPE_32       = 2,
+       CSDP_SRC_PORT_EMIFF     = 0 << 2, /* OMAP1 only */
+       CSDP_SRC_PORT_EMIFS     = 1 << 2, /* OMAP1 only */
+       CSDP_SRC_PORT_OCP_T1    = 2 << 2, /* OMAP1 only */
+       CSDP_SRC_PORT_TIPB      = 3 << 2, /* OMAP1 only */
+       CSDP_SRC_PORT_OCP_T2    = 4 << 2, /* OMAP1 only */
+       CSDP_SRC_PORT_MPUI      = 5 << 2, /* OMAP1 only */
+       CSDP_SRC_PACKED         = BIT(6),
+       CSDP_SRC_BURST_1        = 0 << 7,
+       CSDP_SRC_BURST_16       = 1 << 7,
+       CSDP_SRC_BURST_32       = 2 << 7,
+       CSDP_SRC_BURST_64       = 3 << 7,
+       CSDP_DST_PORT_EMIFF     = 0 << 9, /* OMAP1 only */
+       CSDP_DST_PORT_EMIFS     = 1 << 9, /* OMAP1 only */
+       CSDP_DST_PORT_OCP_T1    = 2 << 9, /* OMAP1 only */
+       CSDP_DST_PORT_TIPB      = 3 << 9, /* OMAP1 only */
+       CSDP_DST_PORT_OCP_T2    = 4 << 9, /* OMAP1 only */
+       CSDP_DST_PORT_MPUI      = 5 << 9, /* OMAP1 only */
+       CSDP_DST_PACKED         = BIT(13),
+       CSDP_DST_BURST_1        = 0 << 14,
+       CSDP_DST_BURST_16       = 1 << 14,
+       CSDP_DST_BURST_32       = 2 << 14,
+       CSDP_DST_BURST_64       = 3 << 14,
+
+       CICR_TOUT_IE            = BIT(0),       /* OMAP1 only */
+       CICR_DROP_IE            = BIT(1),
+       CICR_HALF_IE            = BIT(2),
+       CICR_FRAME_IE           = BIT(3),
+       CICR_LAST_IE            = BIT(4),
+       CICR_BLOCK_IE           = BIT(5),
+       CICR_PKT_IE             = BIT(7),       /* OMAP2+ only */
+       CICR_TRANS_ERR_IE       = BIT(8),       /* OMAP2+ only */
+       CICR_SUPERVISOR_ERR_IE  = BIT(10),      /* OMAP2+ only */
+       CICR_MISALIGNED_ERR_IE  = BIT(11),      /* OMAP2+ only */
+       CICR_DRAIN_IE           = BIT(12),      /* OMAP2+ only */
+       CICR_SUPER_BLOCK_IE     = BIT(14),      /* OMAP2+ only */
+
+       CLNK_CTRL_ENABLE_LNK    = BIT(15),
+};
+
+static const unsigned es_bytes[] = {
+       [CSDP_DATA_TYPE_8] = 1,
+       [CSDP_DATA_TYPE_16] = 2,
+       [CSDP_DATA_TYPE_32] = 4,
+};
+
+static struct of_dma_filter_info omap_dma_info = {
+       .filter_fn = omap_dma_filter_fn,
+};
+
+static inline struct omap_dmadev *to_omap_dma_dev(struct dma_device *d)
+{
+       return container_of(d, struct omap_dmadev, ddev);
+}
+
+static inline struct omap_chan *to_omap_dma_chan(struct dma_chan *c)
+{
+       return container_of(c, struct omap_chan, vc.chan);
+}
+
+static inline struct omap_desc *to_omap_dma_desc(struct dma_async_tx_descriptor *t)
+{
+       return container_of(t, struct omap_desc, vd.tx);
+}
+
+static void omap_dma_desc_free(struct virt_dma_desc *vd)
+{
+       kfree(container_of(vd, struct omap_desc, vd));
+}
+
+static void omap_dma_write(uint32_t val, unsigned type, void __iomem *addr)
+{
+       switch (type) {
+       case OMAP_DMA_REG_16BIT:
+               writew_relaxed(val, addr);
+               break;
+       case OMAP_DMA_REG_2X16BIT:
+               writew_relaxed(val, addr);
+               writew_relaxed(val >> 16, addr + 2);
+               break;
+       case OMAP_DMA_REG_32BIT:
+               writel_relaxed(val, addr);
+               break;
+       default:
+               WARN_ON(1);
+       }
+}
+
+static unsigned omap_dma_read(unsigned type, void __iomem *addr)
+{
+       unsigned val;
+
+       switch (type) {
+       case OMAP_DMA_REG_16BIT:
+               val = readw_relaxed(addr);
+               break;
+       case OMAP_DMA_REG_2X16BIT:
+               val = readw_relaxed(addr);
+               val |= readw_relaxed(addr + 2) << 16;
+               break;
+       case OMAP_DMA_REG_32BIT:
+               val = readl_relaxed(addr);
+               break;
+       default:
+               WARN_ON(1);
+               val = 0;
+       }
+
+       return val;
+}
+
+static void omap_dma_glbl_write(struct omap_dmadev *od, unsigned reg, unsigned val)
+{
+       const struct omap_dma_reg *r = od->reg_map + reg;
+
+       WARN_ON(r->stride);
+
+       omap_dma_write(val, r->type, od->base + r->offset);
+}
+
+static unsigned omap_dma_glbl_read(struct omap_dmadev *od, unsigned reg)
+{
+       const struct omap_dma_reg *r = od->reg_map + reg;
+
+       WARN_ON(r->stride);
+
+       return omap_dma_read(r->type, od->base + r->offset);
+}
+
+static void omap_dma_chan_write(struct omap_chan *c, unsigned reg, unsigned val)
+{
+       const struct omap_dma_reg *r = c->reg_map + reg;
+
+       omap_dma_write(val, r->type, c->channel_base + r->offset);
+}
+
+static unsigned omap_dma_chan_read(struct omap_chan *c, unsigned reg)
+{
+       const struct omap_dma_reg *r = c->reg_map + reg;
+
+       return omap_dma_read(r->type, c->channel_base + r->offset);
+}
+
+static void omap_dma_clear_csr(struct omap_chan *c)
+{
+       if (dma_omap1())
+               omap_dma_chan_read(c, CSR);
+       else
+               omap_dma_chan_write(c, CSR, ~0);
+}
+
+static unsigned omap_dma_get_csr(struct omap_chan *c)
+{
+       unsigned val = omap_dma_chan_read(c, CSR);
+
+       if (!dma_omap1())
+               omap_dma_chan_write(c, CSR, val);
+
+       return val;
+}
+
+static void omap_dma_assign(struct omap_dmadev *od, struct omap_chan *c,
+       unsigned lch)
+{
+       c->channel_base = od->base + od->plat->channel_stride * lch;
+
+       od->lch_map[lch] = c;
+}
+
+static void omap_dma_start(struct omap_chan *c, struct omap_desc *d)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(c->vc.chan.device);
+
+       if (__dma_omap15xx(od->plat->dma_attr))
+               omap_dma_chan_write(c, CPC, 0);
+       else
+               omap_dma_chan_write(c, CDAC, 0);
+
+       omap_dma_clear_csr(c);
+
+       /* Enable interrupts */
+       omap_dma_chan_write(c, CICR, d->cicr);
+
+       /* Enable channel */
+       omap_dma_chan_write(c, CCR, d->ccr | CCR_ENABLE);
+}
+
+static void omap_dma_stop(struct omap_chan *c)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(c->vc.chan.device);
+       uint32_t val;
+
+       /* disable irq */
+       omap_dma_chan_write(c, CICR, 0);
+
+       omap_dma_clear_csr(c);
+
+       val = omap_dma_chan_read(c, CCR);
+       if (od->plat->errata & DMA_ERRATA_i541 && val & CCR_TRIGGER_SRC) {
+               uint32_t sysconfig;
+               unsigned i;
+
+               sysconfig = omap_dma_glbl_read(od, OCP_SYSCONFIG);
+               val = sysconfig & ~DMA_SYSCONFIG_MIDLEMODE_MASK;
+               val |= DMA_SYSCONFIG_MIDLEMODE(DMA_IDLEMODE_NO_IDLE);
+               omap_dma_glbl_write(od, OCP_SYSCONFIG, val);
+
+               val = omap_dma_chan_read(c, CCR);
+               val &= ~CCR_ENABLE;
+               omap_dma_chan_write(c, CCR, val);
+
+               /* Wait for sDMA FIFO to drain */
+               for (i = 0; ; i++) {
+                       val = omap_dma_chan_read(c, CCR);
+                       if (!(val & (CCR_RD_ACTIVE | CCR_WR_ACTIVE)))
+                               break;
+
+                       if (i > 100)
+                               break;
+
+                       udelay(5);
+               }
+
+               if (val & (CCR_RD_ACTIVE | CCR_WR_ACTIVE))
+                       dev_err(c->vc.chan.device->dev,
+                               "DMA drain did not complete on lch %d\n",
+                               c->dma_ch);
+
+               omap_dma_glbl_write(od, OCP_SYSCONFIG, sysconfig);
+       } else {
+               val &= ~CCR_ENABLE;
+               omap_dma_chan_write(c, CCR, val);
+       }
+
+       mb();
+
+       if (!__dma_omap15xx(od->plat->dma_attr) && c->cyclic) {
+               val = omap_dma_chan_read(c, CLNK_CTRL);
+
+               if (dma_omap1())
+                       val |= 1 << 14; /* set the STOP_LNK bit */
+               else
+                       val &= ~CLNK_CTRL_ENABLE_LNK;
+
+               omap_dma_chan_write(c, CLNK_CTRL, val);
+       }
+}
+
+static void omap_dma_start_sg(struct omap_chan *c, struct omap_desc *d,
+       unsigned idx)
+{
+       struct omap_sg *sg = d->sg + idx;
+       unsigned cxsa, cxei, cxfi;
+
+       if (d->dir == DMA_DEV_TO_MEM) {
+               cxsa = CDSA;
+               cxei = CDEI;
+               cxfi = CDFI;
+       } else {
+               cxsa = CSSA;
+               cxei = CSEI;
+               cxfi = CSFI;
+       }
+
+       omap_dma_chan_write(c, cxsa, sg->addr);
+       omap_dma_chan_write(c, cxei, 0);
+       omap_dma_chan_write(c, cxfi, 0);
+       omap_dma_chan_write(c, CEN, sg->en);
+       omap_dma_chan_write(c, CFN, sg->fn);
+
+       omap_dma_start(c, d);
+}
+
+static void omap_dma_start_desc(struct omap_chan *c)
+{
+       struct virt_dma_desc *vd = vchan_next_desc(&c->vc);
+       struct omap_desc *d;
+       unsigned cxsa, cxei, cxfi;
+
+       if (!vd) {
+               c->desc = NULL;
+               return;
+       }
+
+       list_del(&vd->node);
+
+       c->desc = d = to_omap_dma_desc(&vd->tx);
+       c->sgidx = 0;
+
+       /*
+        * This provides the necessary barrier to ensure data held in
+        * DMA coherent memory is visible to the DMA engine prior to
+        * the transfer starting.
+        */
+       mb();
+
+       omap_dma_chan_write(c, CCR, d->ccr);
+       if (dma_omap1())
+               omap_dma_chan_write(c, CCR2, d->ccr >> 16);
+
+       if (d->dir == DMA_DEV_TO_MEM) {
+               cxsa = CSSA;
+               cxei = CSEI;
+               cxfi = CSFI;
+       } else {
+               cxsa = CDSA;
+               cxei = CDEI;
+               cxfi = CDFI;
+       }
+
+       omap_dma_chan_write(c, cxsa, d->dev_addr);
+       omap_dma_chan_write(c, cxei, 0);
+       omap_dma_chan_write(c, cxfi, d->fi);
+       omap_dma_chan_write(c, CSDP, d->csdp);
+       omap_dma_chan_write(c, CLNK_CTRL, d->clnk_ctrl);
+
+       omap_dma_start_sg(c, d, 0);
+}
+
+static void omap_dma_callback(int ch, u16 status, void *data)
+{
+       struct omap_chan *c = data;
+       struct omap_desc *d;
+       unsigned long flags;
+
+       spin_lock_irqsave(&c->vc.lock, flags);
+       d = c->desc;
+       if (d) {
+               if (!c->cyclic) {
+                       if (++c->sgidx < d->sglen) {
+                               omap_dma_start_sg(c, d, c->sgidx);
+                       } else {
+                               omap_dma_start_desc(c);
+                               vchan_cookie_complete(&d->vd);
+                       }
+               } else {
+                       vchan_cyclic_callback(&d->vd);
+               }
+       }
+       spin_unlock_irqrestore(&c->vc.lock, flags);
+}
+
+/*
+ * This callback schedules all pending channels.  We could be more
+ * clever here by postponing allocation of the real DMA channels to
+ * this point, and freeing them when our virtual channel becomes idle.
+ *
+ * We would then need to deal with 'all channels in-use'
+ */
+static void omap_dma_sched(unsigned long data)
+{
+       struct omap_dmadev *d = (struct omap_dmadev *)data;
+       LIST_HEAD(head);
+
+       spin_lock_irq(&d->lock);
+       list_splice_tail_init(&d->pending, &head);
+       spin_unlock_irq(&d->lock);
+
+       while (!list_empty(&head)) {
+               struct omap_chan *c = list_first_entry(&head,
+                       struct omap_chan, node);
+
+               spin_lock_irq(&c->vc.lock);
+               list_del_init(&c->node);
+               omap_dma_start_desc(c);
+               spin_unlock_irq(&c->vc.lock);
+       }
+}
+
+static irqreturn_t omap_dma_irq(int irq, void *devid)
+{
+       struct omap_dmadev *od = devid;
+       unsigned status, channel;
+
+       spin_lock(&od->irq_lock);
+
+       status = omap_dma_glbl_read(od, IRQSTATUS_L1);
+       status &= od->irq_enable_mask;
+       if (status == 0) {
+               spin_unlock(&od->irq_lock);
+               return IRQ_NONE;
+       }
+
+       while ((channel = ffs(status)) != 0) {
+               unsigned mask, csr;
+               struct omap_chan *c;
+
+               channel -= 1;
+               mask = BIT(channel);
+               status &= ~mask;
+
+               c = od->lch_map[channel];
+               if (c == NULL) {
+                       /* This should never happen */
+                       dev_err(od->ddev.dev, "invalid channel %u\n", channel);
+                       continue;
+               }
+
+               csr = omap_dma_get_csr(c);
+               omap_dma_glbl_write(od, IRQSTATUS_L1, mask);
+
+               omap_dma_callback(channel, csr, c);
+       }
+
+       spin_unlock(&od->irq_lock);
+
+       return IRQ_HANDLED;
+}
+
+static int omap_dma_alloc_chan_resources(struct dma_chan *chan)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(chan->device);
+       struct omap_chan *c = to_omap_dma_chan(chan);
+       int ret;
+
+       if (od->legacy) {
+               ret = omap_request_dma(c->dma_sig, "DMA engine",
+                                      omap_dma_callback, c, &c->dma_ch);
+       } else {
+               ret = omap_request_dma(c->dma_sig, "DMA engine", NULL, NULL,
+                                      &c->dma_ch);
+       }
+
+       dev_dbg(od->ddev.dev, "allocating channel %u for %u\n",
+               c->dma_ch, c->dma_sig);
+
+       if (ret >= 0) {
+               omap_dma_assign(od, c, c->dma_ch);
+
+               if (!od->legacy) {
+                       unsigned val;
+
+                       spin_lock_irq(&od->irq_lock);
+                       val = BIT(c->dma_ch);
+                       omap_dma_glbl_write(od, IRQSTATUS_L1, val);
+                       od->irq_enable_mask |= val;
+                       omap_dma_glbl_write(od, IRQENABLE_L1, od->irq_enable_mask);
+
+                       val = omap_dma_glbl_read(od, IRQENABLE_L0);
+                       val &= ~BIT(c->dma_ch);
+                       omap_dma_glbl_write(od, IRQENABLE_L0, val);
+                       spin_unlock_irq(&od->irq_lock);
+               }
+       }
+
+       if (dma_omap1()) {
+               if (__dma_omap16xx(od->plat->dma_attr)) {
+                       c->ccr = CCR_OMAP31_DISABLE;
+                       /* Duplicate what plat-omap/dma.c does */
+                       c->ccr |= c->dma_ch + 1;
+               } else {
+                       c->ccr = c->dma_sig & 0x1f;
+               }
+       } else {
+               c->ccr = c->dma_sig & 0x1f;
+               c->ccr |= (c->dma_sig & ~0x1f) << 14;
+       }
+       if (od->plat->errata & DMA_ERRATA_IFRAME_BUFFERING)
+               c->ccr |= CCR_BUFFERING_DISABLE;
+
+       return ret;
+}
+
+static void omap_dma_free_chan_resources(struct dma_chan *chan)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(chan->device);
+       struct omap_chan *c = to_omap_dma_chan(chan);
+
+       if (!od->legacy) {
+               spin_lock_irq(&od->irq_lock);
+               od->irq_enable_mask &= ~BIT(c->dma_ch);
+               omap_dma_glbl_write(od, IRQENABLE_L1, od->irq_enable_mask);
+               spin_unlock_irq(&od->irq_lock);
+       }
+
+       c->channel_base = NULL;
+       od->lch_map[c->dma_ch] = NULL;
+       vchan_free_chan_resources(&c->vc);
+       omap_free_dma(c->dma_ch);
+
+       dev_dbg(od->ddev.dev, "freeing channel for %u\n", c->dma_sig);
+}
+
+static size_t omap_dma_sg_size(struct omap_sg *sg)
+{
+       return sg->en * sg->fn;
+}
+
+static size_t omap_dma_desc_size(struct omap_desc *d)
+{
+       unsigned i;
+       size_t size;
+
+       for (size = i = 0; i < d->sglen; i++)
+               size += omap_dma_sg_size(&d->sg[i]);
+
+       return size * es_bytes[d->es];
+}
+
+static size_t omap_dma_desc_size_pos(struct omap_desc *d, dma_addr_t addr)
+{
+       unsigned i;
+       size_t size, es_size = es_bytes[d->es];
+
+       for (size = i = 0; i < d->sglen; i++) {
+               size_t this_size = omap_dma_sg_size(&d->sg[i]) * es_size;
+
+               if (size)
+                       size += this_size;
+               else if (addr >= d->sg[i].addr &&
+                        addr < d->sg[i].addr + this_size)
+                       size += d->sg[i].addr + this_size - addr;
+       }
+       return size;
+}
+
+/*
+ * OMAP 3.2/3.3 erratum: sometimes 0 is returned if CSAC/CDAC is
+ * read before the DMA controller finished disabling the channel.
+ */
+static uint32_t omap_dma_chan_read_3_3(struct omap_chan *c, unsigned reg)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(c->vc.chan.device);
+       uint32_t val;
+
+       val = omap_dma_chan_read(c, reg);
+       if (val == 0 && od->plat->errata & DMA_ERRATA_3_3)
+               val = omap_dma_chan_read(c, reg);
+
+       return val;
+}
+
+static dma_addr_t omap_dma_get_src_pos(struct omap_chan *c)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(c->vc.chan.device);
+       dma_addr_t addr, cdac;
+
+       if (__dma_omap15xx(od->plat->dma_attr)) {
+               addr = omap_dma_chan_read(c, CPC);
+       } else {
+               addr = omap_dma_chan_read_3_3(c, CSAC);
+               cdac = omap_dma_chan_read_3_3(c, CDAC);
+
+               /*
+                * CDAC == 0 indicates that the DMA transfer on the channel has
+                * not been started (no data has been transferred so far).
+                * Return the programmed source start address in this case.
+                */
+               if (cdac == 0)
+                       addr = omap_dma_chan_read(c, CSSA);
+       }
+
+       if (dma_omap1())
+               addr |= omap_dma_chan_read(c, CSSA) & 0xffff0000;
+
+       return addr;
+}
+
+static dma_addr_t omap_dma_get_dst_pos(struct omap_chan *c)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(c->vc.chan.device);
+       dma_addr_t addr;
+
+       if (__dma_omap15xx(od->plat->dma_attr)) {
+               addr = omap_dma_chan_read(c, CPC);
+       } else {
+               addr = omap_dma_chan_read_3_3(c, CDAC);
+
+               /*
+                * CDAC == 0 indicates that the DMA transfer on the channel
+                * has not been started (no data has been transferred so
+                * far).  Return the programmed destination start address in
+                * this case.
+                */
+               if (addr == 0)
+                       addr = omap_dma_chan_read(c, CDSA);
+       }
+
+       if (dma_omap1())
+               addr |= omap_dma_chan_read(c, CDSA) & 0xffff0000;
+
+       return addr;
+}
+
+static enum dma_status omap_dma_tx_status(struct dma_chan *chan,
+       dma_cookie_t cookie, struct dma_tx_state *txstate)
+{
+       struct omap_chan *c = to_omap_dma_chan(chan);
+       struct virt_dma_desc *vd;
+       enum dma_status ret;
+       unsigned long flags;
+
+       ret = dma_cookie_status(chan, cookie, txstate);
+       if (ret == DMA_COMPLETE || !txstate)
+               return ret;
+
+       spin_lock_irqsave(&c->vc.lock, flags);
+       vd = vchan_find_desc(&c->vc, cookie);
+       if (vd) {
+               txstate->residue = omap_dma_desc_size(to_omap_dma_desc(&vd->tx));
+       } else if (c->desc && c->desc->vd.tx.cookie == cookie) {
+               struct omap_desc *d = c->desc;
+               dma_addr_t pos;
+
+               if (d->dir == DMA_MEM_TO_DEV)
+                       pos = omap_dma_get_src_pos(c);
+               else if (d->dir == DMA_DEV_TO_MEM)
+                       pos = omap_dma_get_dst_pos(c);
+               else
+                       pos = 0;
+
+               txstate->residue = omap_dma_desc_size_pos(d, pos);
+       } else {
+               txstate->residue = 0;
+       }
+       spin_unlock_irqrestore(&c->vc.lock, flags);
+
+       return ret;
+}
+
+static void omap_dma_issue_pending(struct dma_chan *chan)
+{
+       struct omap_chan *c = to_omap_dma_chan(chan);
+       unsigned long flags;
+
+       spin_lock_irqsave(&c->vc.lock, flags);
+       if (vchan_issue_pending(&c->vc) && !c->desc) {
+               /*
+                * c->cyclic is used only by audio and in this case the DMA need
+                * to be started without delay.
+                */
+               if (!c->cyclic) {
+                       struct omap_dmadev *d = to_omap_dma_dev(chan->device);
+                       spin_lock(&d->lock);
+                       if (list_empty(&c->node))
+                               list_add_tail(&c->node, &d->pending);
+                       spin_unlock(&d->lock);
+                       tasklet_schedule(&d->task);
+               } else {
+                       omap_dma_start_desc(c);
+               }
+       }
+       spin_unlock_irqrestore(&c->vc.lock, flags);
+}
+
+static struct dma_async_tx_descriptor *omap_dma_prep_slave_sg(
+       struct dma_chan *chan, struct scatterlist *sgl, unsigned sglen,
+       enum dma_transfer_direction dir, unsigned long tx_flags, void *context)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(chan->device);
+       struct omap_chan *c = to_omap_dma_chan(chan);
+       enum dma_slave_buswidth dev_width;
+       struct scatterlist *sgent;
+       struct omap_desc *d;
+       dma_addr_t dev_addr;
+       unsigned i, j = 0, es, en, frame_bytes;
+       u32 burst;
+
+       if (dir == DMA_DEV_TO_MEM) {
+               dev_addr = c->cfg.src_addr;
+               dev_width = c->cfg.src_addr_width;
+               burst = c->cfg.src_maxburst;
+       } else if (dir == DMA_MEM_TO_DEV) {
+               dev_addr = c->cfg.dst_addr;
+               dev_width = c->cfg.dst_addr_width;
+               burst = c->cfg.dst_maxburst;
+       } else {
+               dev_err(chan->device->dev, "%s: bad direction?\n", __func__);
+               return NULL;
+       }
+
+       /* Bus width translates to the element size (ES) */
+       switch (dev_width) {
+       case DMA_SLAVE_BUSWIDTH_1_BYTE:
+               es = CSDP_DATA_TYPE_8;
+               break;
+       case DMA_SLAVE_BUSWIDTH_2_BYTES:
+               es = CSDP_DATA_TYPE_16;
+               break;
+       case DMA_SLAVE_BUSWIDTH_4_BYTES:
+               es = CSDP_DATA_TYPE_32;
+               break;
+       default: /* not reached */
+               return NULL;
+       }
+
+       /* Now allocate and setup the descriptor. */
+       d = kzalloc(sizeof(*d) + sglen * sizeof(d->sg[0]), GFP_ATOMIC);
+       if (!d)
+               return NULL;
+
+       d->dir = dir;
+       d->dev_addr = dev_addr;
+       d->es = es;
+
+       d->ccr = c->ccr | CCR_SYNC_FRAME;
+       if (dir == DMA_DEV_TO_MEM)
+               d->ccr |= CCR_DST_AMODE_POSTINC | CCR_SRC_AMODE_CONSTANT;
+       else
+               d->ccr |= CCR_DST_AMODE_CONSTANT | CCR_SRC_AMODE_POSTINC;
+
+       d->cicr = CICR_DROP_IE | CICR_BLOCK_IE;
+       d->csdp = es;
+
+       if (dma_omap1()) {
+               d->cicr |= CICR_TOUT_IE;
+
+               if (dir == DMA_DEV_TO_MEM)
+                       d->csdp |= CSDP_DST_PORT_EMIFF | CSDP_SRC_PORT_TIPB;
+               else
+                       d->csdp |= CSDP_DST_PORT_TIPB | CSDP_SRC_PORT_EMIFF;
+       } else {
+               if (dir == DMA_DEV_TO_MEM)
+                       d->ccr |= CCR_TRIGGER_SRC;
+
+               d->cicr |= CICR_MISALIGNED_ERR_IE | CICR_TRANS_ERR_IE;
+       }
+       if (od->plat->errata & DMA_ERRATA_PARALLEL_CHANNELS)
+               d->clnk_ctrl = c->dma_ch;
+
+       /*
+        * Build our scatterlist entries: each contains the address,
+        * the number of elements (EN) in each frame, and the number of
+        * frames (FN).  Number of bytes for this entry = ES * EN * FN.
+        *
+        * Burst size translates to number of elements with frame sync.
+        * Note: DMA engine defines burst to be the number of dev-width
+        * transfers.
+        */
+       en = burst;
+       frame_bytes = es_bytes[es] * en;
+       for_each_sg(sgl, sgent, sglen, i) {
+               d->sg[j].addr = sg_dma_address(sgent);
+               d->sg[j].en = en;
+               d->sg[j].fn = sg_dma_len(sgent) / frame_bytes;
+               j++;
+       }
+
+       d->sglen = j;
+
+       return vchan_tx_prep(&c->vc, &d->vd, tx_flags);
+}
+
+static struct dma_async_tx_descriptor *omap_dma_prep_dma_cyclic(
+       struct dma_chan *chan, dma_addr_t buf_addr, size_t buf_len,
+       size_t period_len, enum dma_transfer_direction dir, unsigned long flags)
+{
+       struct omap_dmadev *od = to_omap_dma_dev(chan->device);
+       struct omap_chan *c = to_omap_dma_chan(chan);
+       enum dma_slave_buswidth dev_width;
+       struct omap_desc *d;
+       dma_addr_t dev_addr;
+       unsigned es;
+       u32 burst;
+
+       if (dir == DMA_DEV_TO_MEM) {
+               dev_addr = c->cfg.src_addr;
+               dev_width = c->cfg.src_addr_width;
+               burst = c->cfg.src_maxburst;
+       } else if (dir == DMA_MEM_TO_DEV) {
+               dev_addr = c->cfg.dst_addr;
+               dev_width = c->cfg.dst_addr_width;
+               burst = c->cfg.dst_maxburst;
+       } else {
+               dev_err(chan->device->dev, "%s: bad direction?\n", __func__);
+               return NULL;
+       }
+
+       /* Bus width translates to the element size (ES) */
+       switch (dev_width) {
+       case DMA_SLAVE_BUSWIDTH_1_BYTE:
+               es = CSDP_DATA_TYPE_8;
+               break;
+       case DMA_SLAVE_BUSWIDTH_2_BYTES:
+               es = CSDP_DATA_TYPE_16;
+               break;
+       case DMA_SLAVE_BUSWIDTH_4_BYTES:
+               es = CSDP_DATA_TYPE_32;
+               break;
+       default: /* not reached */
+               return NULL;
+       }
+
+       /* Now allocate and setup the descriptor. */
+       d = kzalloc(sizeof(*d) + sizeof(d->sg[0]), GFP_ATOMIC);
+       if (!d)
+               return NULL;
+
+       d->dir = dir;
+       d->dev_addr = dev_addr;
+       d->fi = burst;
+       d->es = es;
+       d->sg[0].addr = buf_addr;
+       d->sg[0].en = period_len / es_bytes[es];
+       d->sg[0].fn = buf_len / period_len;
+       d->sglen = 1;
+
+       d->ccr = c->ccr;
+       if (dir == DMA_DEV_TO_MEM)
+               d->ccr |= CCR_DST_AMODE_POSTINC | CCR_SRC_AMODE_CONSTANT;
+       else
+               d->ccr |= CCR_DST_AMODE_CONSTANT | CCR_SRC_AMODE_POSTINC;
+
+       d->cicr = CICR_DROP_IE;
+       if (flags & DMA_PREP_INTERRUPT)
+               d->cicr |= CICR_FRAME_IE;
+
+       d->csdp = es;
+
+       if (dma_omap1()) {
+               d->cicr |= CICR_TOUT_IE;
+
+               if (dir == DMA_DEV_TO_MEM)
+                       d->csdp |= CSDP_DST_PORT_EMIFF | CSDP_SRC_PORT_MPUI;
+               else
+                       d->csdp |= CSDP_DST_PORT_MPUI | CSDP_SRC_PORT_EMIFF;
+       } else {
+               if (burst)
+                       d->ccr |= CCR_SYNC_PACKET;
+               else
+                       d->ccr |= CCR_SYNC_ELEMENT;
+
+               if (dir == DMA_DEV_TO_MEM)
+                       d->ccr |= CCR_TRIGGER_SRC;
+
+               d->cicr |= CICR_MISALIGNED_ERR_IE | CICR_TRANS_ERR_IE;
+
+               d->csdp |= CSDP_DST_BURST_64 | CSDP_SRC_BURST_64;
+       }
+
+       if (__dma_omap15xx(od->plat->dma_attr))
+               d->ccr |= CCR_AUTO_INIT | CCR_REPEAT;
+       else
+               d->clnk_ctrl = c->dma_ch | CLNK_CTRL_ENABLE_LNK;
+
+       c->cyclic = true;
+
+       return vchan_tx_prep(&c->vc, &d->vd, flags);
+}
+
+static int omap_dma_slave_config(struct dma_chan *chan, struct dma_slave_config *cfg)
+{
+       struct omap_chan *c = to_omap_dma_chan(chan);
+
+       if (cfg->src_addr_width == DMA_SLAVE_BUSWIDTH_8_BYTES ||
+           cfg->dst_addr_width == DMA_SLAVE_BUSWIDTH_8_BYTES)
+               return -EINVAL;
+
+       memcpy(&c->cfg, cfg, sizeof(c->cfg));
+
+       return 0;
+}
+
+static int omap_dma_terminate_all(struct dma_chan *chan)
+{
+       struct omap_chan *c = to_omap_dma_chan(chan);
+       struct omap_dmadev *d = to_omap_dma_dev(c->vc.chan.device);
+       unsigned long flags;
+       LIST_HEAD(head);
+
+       spin_lock_irqsave(&c->vc.lock, flags);
+
+       /* Prevent this channel being scheduled */
+       spin_lock(&d->lock);
+       list_del_init(&c->node);
+       spin_unlock(&d->lock);
+
+       /*
+        * Stop DMA activity: we assume the callback will not be called
+        * after omap_dma_stop() returns (even if it does, it will see
+        * c->desc is NULL and exit.)
+        */
+       if (c->desc) {
+               omap_dma_desc_free(&c->desc->vd);
+               c->desc = NULL;
+               /* Avoid stopping the dma twice */
+               if (!c->paused)
+                       omap_dma_stop(c);
+       }
+
+       if (c->cyclic) {
+               c->cyclic = false;
+               c->paused = false;
+       }
+
+       vchan_get_all_descriptors(&c->vc, &head);
+       spin_unlock_irqrestore(&c->vc.lock, flags);
+       vchan_dma_desc_free_list(&c->vc, &head);
+
+       return 0;
+}
+
+static int omap_dma_pause(struct dma_chan *chan)
+{
+       struct omap_chan *c = to_omap_dma_chan(chan);
+
+       /* Pause/Resume only allowed with cyclic mode */
+       if (!c->cyclic)
+               return -EINVAL;
+
+       if (!c->paused) {
+               omap_dma_stop(c);
+               c->paused = true;
+       }
+
+       return 0;
+}
+
+static int omap_dma_resume(struct dma_chan *chan)
+{
+       struct omap_chan *c = to_omap_dma_chan(chan);
+
+       /* Pause/Resume only allowed with cyclic mode */
+       if (!c->cyclic)
+               return -EINVAL;
+
+       if (c->paused) {
+               mb();
+
+               /* Restore channel link register */
+               omap_dma_chan_write(c, CLNK_CTRL, c->desc->clnk_ctrl);
+
+               omap_dma_start(c, c->desc);
+               c->paused = false;
+       }
+
+       return 0;
+}
+
+static int omap_dma_chan_init(struct omap_dmadev *od, int dma_sig)
+{
+       struct omap_chan *c;
+
+       c = kzalloc(sizeof(*c), GFP_KERNEL);
+       if (!c)
+               return -ENOMEM;
+
+       c->reg_map = od->reg_map;
+       c->dma_sig = dma_sig;
+       c->vc.desc_free = omap_dma_desc_free;
+       vchan_init(&c->vc, &od->ddev);
+       INIT_LIST_HEAD(&c->node);
+
+       return 0;
+}
+
+static void omap_dma_free(struct omap_dmadev *od)
+{
+       tasklet_kill(&od->task);
+       while (!list_empty(&od->ddev.channels)) {
+               struct omap_chan *c = list_first_entry(&od->ddev.channels,
+                       struct omap_chan, vc.chan.device_node);
+
+               list_del(&c->vc.chan.device_node);
+               tasklet_kill(&c->vc.task);
+               kfree(c);
+       }
+}
+
+#define OMAP_DMA_BUSWIDTHS     (BIT(DMA_SLAVE_BUSWIDTH_1_BYTE) | \
+                                BIT(DMA_SLAVE_BUSWIDTH_2_BYTES) | \
+                                BIT(DMA_SLAVE_BUSWIDTH_4_BYTES))
+
+static int omap_dma_probe(struct platform_device *pdev)
+{
+       struct omap_dmadev *od;
+       struct resource *res;
+       int rc, i, irq;
+
+       od = devm_kzalloc(&pdev->dev, sizeof(*od), GFP_KERNEL);
+       if (!od)
+               return -ENOMEM;
+
+       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+       od->base = devm_ioremap_resource(&pdev->dev, res);
+       if (IS_ERR(od->base))
+               return PTR_ERR(od->base);
+
+       od->plat = omap_get_plat_info();
+       if (!od->plat)
+               return -EPROBE_DEFER;
+
+       od->reg_map = od->plat->reg_map;
+
+       dma_cap_set(DMA_SLAVE, od->ddev.cap_mask);
+       dma_cap_set(DMA_CYCLIC, od->ddev.cap_mask);
+       od->ddev.device_alloc_chan_resources = omap_dma_alloc_chan_resources;
+       od->ddev.device_free_chan_resources = omap_dma_free_chan_resources;
+       od->ddev.device_tx_status = omap_dma_tx_status;
+       od->ddev.device_issue_pending = omap_dma_issue_pending;
+       od->ddev.device_prep_slave_sg = omap_dma_prep_slave_sg;
+       od->ddev.device_prep_dma_cyclic = omap_dma_prep_dma_cyclic;
+       od->ddev.device_config = omap_dma_slave_config;
+       od->ddev.device_pause = omap_dma_pause;
+       od->ddev.device_resume = omap_dma_resume;
+       od->ddev.device_terminate_all = omap_dma_terminate_all;
+       od->ddev.src_addr_widths = OMAP_DMA_BUSWIDTHS;
+       od->ddev.dst_addr_widths = OMAP_DMA_BUSWIDTHS;
+       od->ddev.directions = BIT(DMA_DEV_TO_MEM) | BIT(DMA_MEM_TO_DEV);
+       od->ddev.residue_granularity = DMA_RESIDUE_GRANULARITY_BURST;
+       od->ddev.dev = &pdev->dev;
+       INIT_LIST_HEAD(&od->ddev.channels);
+       INIT_LIST_HEAD(&od->pending);
+       spin_lock_init(&od->lock);
+       spin_lock_init(&od->irq_lock);
+
+       tasklet_init(&od->task, omap_dma_sched, (unsigned long)od);
+
+       for (i = 0; i < 127; i++) {
+               rc = omap_dma_chan_init(od, i);
+               if (rc) {
+                       omap_dma_free(od);
+                       return rc;
+               }
+       }
+
+       irq = platform_get_irq(pdev, 1);
+       if (irq <= 0) {
+               dev_info(&pdev->dev, "failed to get L1 IRQ: %d\n", irq);
+               od->legacy = true;
+       } else {
+               /* Disable all interrupts */
+               od->irq_enable_mask = 0;
+               omap_dma_glbl_write(od, IRQENABLE_L1, 0);
+
+               rc = devm_request_irq(&pdev->dev, irq, omap_dma_irq,
+                                     IRQF_SHARED, "omap-dma-engine", od);
+               if (rc)
+                       return rc;
+       }
+
+       rc = dma_async_device_register(&od->ddev);
+       if (rc) {
+               pr_warn("OMAP-DMA: failed to register slave DMA engine device: %d\n",
+                       rc);
+               omap_dma_free(od);
+               return rc;
+       }
+
+       platform_set_drvdata(pdev, od);
+
+       if (pdev->dev.of_node) {
+               omap_dma_info.dma_cap = od->ddev.cap_mask;
+
+               /* Device-tree DMA controller registration */
+               rc = of_dma_controller_register(pdev->dev.of_node,
+                               of_dma_simple_xlate, &omap_dma_info);
+               if (rc) {
+                       pr_warn("OMAP-DMA: failed to register DMA controller\n");
+                       dma_async_device_unregister(&od->ddev);
+                       omap_dma_free(od);
+               }
+       }
+
+       dev_info(&pdev->dev, "OMAP DMA engine driver\n");
+
+       return rc;
+}
+
+static int omap_dma_remove(struct platform_device *pdev)
+{
+       struct omap_dmadev *od = platform_get_drvdata(pdev);
+
+       if (pdev->dev.of_node)
+               of_dma_controller_free(pdev->dev.of_node);
+
+       dma_async_device_unregister(&od->ddev);
+
+       if (!od->legacy) {
+               /* Disable all interrupts */
+               omap_dma_glbl_write(od, IRQENABLE_L0, 0);
+       }
+
+       omap_dma_free(od);
+
+       return 0;
+}
+
+static const struct of_device_id omap_dma_match[] = {
+       { .compatible = "ti,omap2420-sdma", },
+       { .compatible = "ti,omap2430-sdma", },
+       { .compatible = "ti,omap3430-sdma", },
+       { .compatible = "ti,omap3630-sdma", },
+       { .compatible = "ti,omap4430-sdma", },
+       {},
+};
+MODULE_DEVICE_TABLE(of, omap_dma_match);
+
+static struct platform_driver omap_dma_driver = {
+       .probe  = omap_dma_probe,
+       .remove = omap_dma_remove,
+       .driver = {
+               .name = "omap-dma-engine",
+               .of_match_table = of_match_ptr(omap_dma_match),
+       },
+};
+
+bool omap_dma_filter_fn(struct dma_chan *chan, void *param)
+{
+       if (chan->device->dev->driver == &omap_dma_driver.driver) {
+               struct omap_chan *c = to_omap_dma_chan(chan);
+               unsigned req = *(unsigned *)param;
+
+               return req == c->dma_sig;
+       }
+       return false;
+}
+EXPORT_SYMBOL_GPL(omap_dma_filter_fn);
+
+static int omap_dma_init(void)
+{
+       return platform_driver_register(&omap_dma_driver);
+}
+subsys_initcall(omap_dma_init);
+
+static void __exit omap_dma_exit(void)
+{
+       platform_driver_unregister(&omap_dma_driver);
+}
+module_exit(omap_dma_exit);
+
+MODULE_AUTHOR("Russell King");
+MODULE_LICENSE("GPL");