Merge tag 'ntb-3.13' of git://github.com/jonmason/ntb
[linux-drm-fsl-dcu.git] / drivers / ntb / ntb_transport.c
index 676937b..3217f39 100644 (file)
@@ -1046,10 +1046,9 @@ static void ntb_async_rx(struct ntb_queue_entry *entry, void *offset,
        struct dma_chan *chan = qp->dma_chan;
        struct dma_device *device;
        size_t pay_off, buff_off;
-       dma_addr_t src, dest;
+       struct dmaengine_unmap_data *unmap;
        dma_cookie_t cookie;
        void *buf = entry->buf;
-       unsigned long flags;
 
        entry->len = len;
 
@@ -1057,35 +1056,49 @@ static void ntb_async_rx(struct ntb_queue_entry *entry, void *offset,
                goto err;
 
        if (len < copy_bytes) 
-               goto err1;
+               goto err_wait;
 
        device = chan->device;
        pay_off = (size_t) offset & ~PAGE_MASK;
        buff_off = (size_t) buf & ~PAGE_MASK;
 
        if (!is_dma_copy_aligned(device, pay_off, buff_off, len))
-               goto err1;
+               goto err_wait;
 
-       dest = dma_map_single(device->dev, buf, len, DMA_FROM_DEVICE);
-       if (dma_mapping_error(device->dev, dest))
-               goto err1;
+       unmap = dmaengine_get_unmap_data(device->dev, 2, GFP_NOWAIT);
+       if (!unmap)
+               goto err_wait;
 
-       src = dma_map_single(device->dev, offset, len, DMA_TO_DEVICE);
-       if (dma_mapping_error(device->dev, src))
-               goto err2;
+       unmap->len = len;
+       unmap->addr[0] = dma_map_page(device->dev, virt_to_page(offset),
+                                     pay_off, len, DMA_TO_DEVICE);
+       if (dma_mapping_error(device->dev, unmap->addr[0]))
+               goto err_get_unmap;
+
+       unmap->to_cnt = 1;
 
-       flags = DMA_COMPL_DEST_UNMAP_SINGLE | DMA_COMPL_SRC_UNMAP_SINGLE |
-               DMA_PREP_INTERRUPT;
-       txd = device->device_prep_dma_memcpy(chan, dest, src, len, flags);
+       unmap->addr[1] = dma_map_page(device->dev, virt_to_page(buf),
+                                     buff_off, len, DMA_FROM_DEVICE);
+       if (dma_mapping_error(device->dev, unmap->addr[1]))
+               goto err_get_unmap;
+
+       unmap->from_cnt = 1;
+
+       txd = device->device_prep_dma_memcpy(chan, unmap->addr[1],
+                                            unmap->addr[0], len,
+                                            DMA_PREP_INTERRUPT);
        if (!txd)
-               goto err3;
+               goto err_get_unmap;
 
        txd->callback = ntb_rx_copy_callback;
        txd->callback_param = entry;
+       dma_set_unmap(txd, unmap);
 
        cookie = dmaengine_submit(txd);
        if (dma_submit_error(cookie))
-               goto err3;
+               goto err_set_unmap;
+
+       dmaengine_unmap_put(unmap);
 
        qp->last_cookie = cookie;
 
@@ -1093,11 +1106,11 @@ static void ntb_async_rx(struct ntb_queue_entry *entry, void *offset,
 
        return;
 
-err3:
-       dma_unmap_single(device->dev, src, len, DMA_TO_DEVICE);
-err2:
-       dma_unmap_single(device->dev, dest, len, DMA_FROM_DEVICE);
-err1:
+err_set_unmap:
+       dmaengine_unmap_put(unmap);
+err_get_unmap:
+       dmaengine_unmap_put(unmap);
+err_wait:
        /* If the callbacks come out of order, the writing of the index to the
         * last completed will be out of order.  This may result in the
         * receive stalling forever.
@@ -1252,12 +1265,12 @@ static void ntb_async_tx(struct ntb_transport_qp *qp,
        struct dma_chan *chan = qp->dma_chan;
        struct dma_device *device;
        size_t dest_off, buff_off;
-       dma_addr_t src, dest;
+       struct dmaengine_unmap_data *unmap;
+       dma_addr_t dest;
        dma_cookie_t cookie;
        void __iomem *offset;
        size_t len = entry->len;
        void *buf = entry->buf;
-       unsigned long flags;
 
        offset = qp->tx_mw + qp->tx_max_frame * qp->tx_index;
        hdr = offset + qp->tx_max_frame - sizeof(struct ntb_payload_header);
@@ -1280,28 +1293,41 @@ static void ntb_async_tx(struct ntb_transport_qp *qp,
        if (!is_dma_copy_aligned(device, buff_off, dest_off, len))
                goto err;
 
-       src = dma_map_single(device->dev, buf, len, DMA_TO_DEVICE);
-       if (dma_mapping_error(device->dev, src))
+       unmap = dmaengine_get_unmap_data(device->dev, 1, GFP_NOWAIT);
+       if (!unmap)
                goto err;
 
-       flags = DMA_COMPL_SRC_UNMAP_SINGLE | DMA_PREP_INTERRUPT;
-       txd = device->device_prep_dma_memcpy(chan, dest, src, len, flags);
+       unmap->len = len;
+       unmap->addr[0] = dma_map_page(device->dev, virt_to_page(buf),
+                                     buff_off, len, DMA_TO_DEVICE);
+       if (dma_mapping_error(device->dev, unmap->addr[0]))
+               goto err_get_unmap;
+
+       unmap->to_cnt = 1;
+
+       txd = device->device_prep_dma_memcpy(chan, dest, unmap->addr[0], len,
+                                            DMA_PREP_INTERRUPT);
        if (!txd)
-               goto err1;
+               goto err_get_unmap;
 
        txd->callback = ntb_tx_copy_callback;
        txd->callback_param = entry;
+       dma_set_unmap(txd, unmap);
 
        cookie = dmaengine_submit(txd);
        if (dma_submit_error(cookie))
-               goto err1;
+               goto err_set_unmap;
+
+       dmaengine_unmap_put(unmap);
 
        dma_async_issue_pending(chan);
        qp->tx_async++;
 
        return;
-err1:
-       dma_unmap_single(device->dev, src, len, DMA_TO_DEVICE);
+err_set_unmap:
+       dmaengine_unmap_put(unmap);
+err_get_unmap:
+       dmaengine_unmap_put(unmap);
 err:
        ntb_memcpy_tx(entry, offset);
        qp->tx_memcpy++;