async_pq: convert to dmaengine_unmap_data
authorDan Williams <dan.j.williams@intel.com>
Fri, 18 Oct 2013 17:35:29 +0000 (19:35 +0200)
committerDan Williams <dan.j.williams@intel.com>
Thu, 14 Nov 2013 19:01:31 +0000 (11:01 -0800)
Use the generic unmap object to unmap dma buffers.

Cc: Vinod Koul <vinod.koul@intel.com>
Cc: Tomasz Figa <t.figa@samsung.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Reported-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
[bzolnier: keep temporary dma_dest array in do_async_gen_syndrome()]
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
crypto/async_tx/async_pq.c
drivers/dma/dmaengine.c

index 91d5d385899ee06c507cee1aa496330405d6276b..8cdbf33bd0464901e0c84b6d246809c2460ff4c7 100644 (file)
@@ -46,49 +46,25 @@ static struct page *pq_scribble_page;
  * do_async_gen_syndrome - asynchronously calculate P and/or Q
  */
 static __async_inline struct dma_async_tx_descriptor *
-do_async_gen_syndrome(struct dma_chan *chan, struct page **blocks,
-                     const unsigned char *scfs, unsigned int offset, int disks,
-                     size_t len, dma_addr_t *dma_src,
+do_async_gen_syndrome(struct dma_chan *chan,
+                     const unsigned char *scfs, int disks,
+                     struct dmaengine_unmap_data *unmap,
+                     enum dma_ctrl_flags dma_flags,
                      struct async_submit_ctl *submit)
 {
        struct dma_async_tx_descriptor *tx = NULL;
        struct dma_device *dma = chan->device;
-       enum dma_ctrl_flags dma_flags = 0;
        enum async_tx_flags flags_orig = submit->flags;
        dma_async_tx_callback cb_fn_orig = submit->cb_fn;
        dma_async_tx_callback cb_param_orig = submit->cb_param;
        int src_cnt = disks - 2;
-       unsigned char coefs[src_cnt];
        unsigned short pq_src_cnt;
        dma_addr_t dma_dest[2];
        int src_off = 0;
-       int idx;
-       int i;
 
-       /* DMAs use destinations as sources, so use BIDIRECTIONAL mapping */
-       if (P(blocks, disks))
-               dma_dest[0] = dma_map_page(dma->dev, P(blocks, disks), offset,
-                                          len, DMA_BIDIRECTIONAL);
-       else
-               dma_flags |= DMA_PREP_PQ_DISABLE_P;
-       if (Q(blocks, disks))
-               dma_dest[1] = dma_map_page(dma->dev, Q(blocks, disks), offset,
-                                          len, DMA_BIDIRECTIONAL);
-       else
-               dma_flags |= DMA_PREP_PQ_DISABLE_Q;
-
-       /* convert source addresses being careful to collapse 'empty'
-        * sources and update the coefficients accordingly
-        */
-       for (i = 0, idx = 0; i < src_cnt; i++) {
-               if (blocks[i] == NULL)
-                       continue;
-               dma_src[idx] = dma_map_page(dma->dev, blocks[i], offset, len,
-                                           DMA_TO_DEVICE);
-               coefs[idx] = scfs[i];
-               idx++;
-       }
-       src_cnt = idx;
+       dma_flags |= DMA_COMPL_SKIP_SRC_UNMAP | DMA_COMPL_SKIP_DEST_UNMAP;
+       if (submit->flags & ASYNC_TX_FENCE)
+               dma_flags |= DMA_PREP_FENCE;
 
        while (src_cnt > 0) {
                submit->flags = flags_orig;
@@ -100,28 +76,25 @@ do_async_gen_syndrome(struct dma_chan *chan, struct page **blocks,
                if (src_cnt > pq_src_cnt) {
                        submit->flags &= ~ASYNC_TX_ACK;
                        submit->flags |= ASYNC_TX_FENCE;
-                       dma_flags |= DMA_COMPL_SKIP_DEST_UNMAP;
                        submit->cb_fn = NULL;
                        submit->cb_param = NULL;
                } else {
-                       dma_flags &= ~DMA_COMPL_SKIP_DEST_UNMAP;
                        submit->cb_fn = cb_fn_orig;
                        submit->cb_param = cb_param_orig;
                        if (cb_fn_orig)
                                dma_flags |= DMA_PREP_INTERRUPT;
                }
-               if (submit->flags & ASYNC_TX_FENCE)
-                       dma_flags |= DMA_PREP_FENCE;
 
-               /* Since we have clobbered the src_list we are committed
-                * to doing this asynchronously.  Drivers force forward
-                * progress in case they can not provide a descriptor
+               /* Drivers force forward progress in case they can not provide
+                * a descriptor
                 */
                for (;;) {
+                       dma_dest[0] = unmap->addr[disks - 2];
+                       dma_dest[1] = unmap->addr[disks - 1];
                        tx = dma->device_prep_dma_pq(chan, dma_dest,
-                                                    &dma_src[src_off],
+                                                    &unmap->addr[src_off],
                                                     pq_src_cnt,
-                                                    &coefs[src_off], len,
+                                                    &scfs[src_off], unmap->len,
                                                     dma_flags);
                        if (likely(tx))
                                break;
@@ -129,6 +102,7 @@ do_async_gen_syndrome(struct dma_chan *chan, struct page **blocks,
                        dma_async_issue_pending(chan);
                }
 
+               dma_set_unmap(tx, unmap);
                async_tx_submit(chan, tx, submit);
                submit->depend_tx = tx;
 
@@ -188,10 +162,6 @@ do_sync_gen_syndrome(struct page **blocks, unsigned int offset, int disks,
  * set to NULL those buffers will be replaced with the raid6_zero_page
  * in the synchronous path and omitted in the hardware-asynchronous
  * path.
- *
- * 'blocks' note: if submit->scribble is NULL then the contents of
- * 'blocks' may be overwritten to perform address conversions
- * (dma_map_page() or page_address()).
  */
 struct dma_async_tx_descriptor *
 async_gen_syndrome(struct page **blocks, unsigned int offset, int disks,
@@ -202,26 +172,69 @@ async_gen_syndrome(struct page **blocks, unsigned int offset, int disks,
                                                      &P(blocks, disks), 2,
                                                      blocks, src_cnt, len);
        struct dma_device *device = chan ? chan->device : NULL;
-       dma_addr_t *dma_src = NULL;
+       struct dmaengine_unmap_data *unmap = NULL;
 
        BUG_ON(disks > 255 || !(P(blocks, disks) || Q(blocks, disks)));
 
-       if (submit->scribble)
-               dma_src = submit->scribble;
-       else if (sizeof(dma_addr_t) <= sizeof(struct page *))
-               dma_src = (dma_addr_t *) blocks;
+       if (device)
+               unmap = dmaengine_get_unmap_data(device->dev, disks, GFP_NOIO);
 
-       if (dma_src && device &&
+       if (unmap &&
            (src_cnt <= dma_maxpq(device, 0) ||
             dma_maxpq(device, DMA_PREP_CONTINUE) > 0) &&
            is_dma_pq_aligned(device, offset, 0, len)) {
+               struct dma_async_tx_descriptor *tx;
+               enum dma_ctrl_flags dma_flags = 0;
+               unsigned char coefs[src_cnt];
+               int i, j;
+
                /* run the p+q asynchronously */
                pr_debug("%s: (async) disks: %d len: %zu\n",
                         __func__, disks, len);
-               return do_async_gen_syndrome(chan, blocks, raid6_gfexp, offset,
-                                            disks, len, dma_src, submit);
+
+               /* convert source addresses being careful to collapse 'empty'
+                * sources and update the coefficients accordingly
+                */
+               unmap->len = len;
+               for (i = 0, j = 0; i < src_cnt; i++) {
+                       if (blocks[i] == NULL)
+                               continue;
+                       unmap->addr[j] = dma_map_page(device->dev, blocks[i], offset,
+                                                     len, DMA_TO_DEVICE);
+                       coefs[j] = raid6_gfexp[i];
+                       unmap->to_cnt++;
+                       j++;
+               }
+
+               /*
+                * DMAs use destinations as sources,
+                * so use BIDIRECTIONAL mapping
+                */
+               unmap->bidi_cnt++;
+               if (P(blocks, disks))
+                       unmap->addr[j++] = dma_map_page(device->dev, P(blocks, disks),
+                                                       offset, len, DMA_BIDIRECTIONAL);
+               else {
+                       unmap->addr[j++] = 0;
+                       dma_flags |= DMA_PREP_PQ_DISABLE_P;
+               }
+
+               unmap->bidi_cnt++;
+               if (Q(blocks, disks))
+                       unmap->addr[j++] = dma_map_page(device->dev, Q(blocks, disks),
+                                                      offset, len, DMA_BIDIRECTIONAL);
+               else {
+                       unmap->addr[j++] = 0;
+                       dma_flags |= DMA_PREP_PQ_DISABLE_Q;
+               }
+
+               tx = do_async_gen_syndrome(chan, coefs, j, unmap, dma_flags, submit);
+               dmaengine_unmap_put(unmap);
+               return tx;
        }
 
+       dmaengine_unmap_put(unmap);
+
        /* run the pq synchronously */
        pr_debug("%s: (sync) disks: %d len: %zu\n", __func__, disks, len);
 
index 54138b57b37c868f84ba1a8c94f7ddc60cdc3086..f878c808466e0b38c4c3347ab3ba6aaebd88c08e 100644 (file)
@@ -953,9 +953,12 @@ static void dmaengine_unmap(struct kref *kref)
                dma_unmap_page(dev, unmap->addr[i], unmap->len,
                               DMA_FROM_DEVICE);
        cnt += unmap->bidi_cnt;
-       for (; i < cnt; i++)
+       for (; i < cnt; i++) {
+               if (unmap->addr[i] == 0)
+                       continue;
                dma_unmap_page(dev, unmap->addr[i], unmap->len,
                               DMA_BIDIRECTIONAL);
+       }
        mempool_free(unmap, __get_unmap_pool(cnt)->pool);
 }