net: axienet: Upgrade descriptors to hold 64-bit addresses

Newer revisions of the AXI DMA IP (>= v7.1) support 64-bit addresses,
both for the descriptors itself, as well as for the buffers they are
pointing to.
This is realised by adding "MSB" words for the next and phys pointer
right behind the existing address word, now named "LSB". These MSB words
live in formerly reserved areas of the descriptor.

If the hardware supports it, write both words when setting an address.
The buffer address is handled by two wrapper functions, the two
occasions where we set the next pointers are open coded.

For now this is guarded by a flag which we don't set yet.

Signed-off-by: Andre Przywara <[e-mail]>
Signed-off-by: David S. Miller <[e-mail]>

drivers/net/ethernet/xilinx/xilinx_axienet.h

···
 #define XAE_FEATURE_PARTIAL_TX_CSUM	(1 << 1)
 #define XAE_FEATURE_FULL_RX_CSUM	(1 << 2)
 #define XAE_FEATURE_FULL_TX_CSUM	(1 << 3)
+#define XAE_FEATURE_DMA_64BIT		(1 << 4)
 
 #define XAE_NO_CSUM_OFFLOAD		0
 
···
 /**
  * struct axidma_bd - Axi Dma buffer descriptor layout
  * @next:         MM2S/S2MM Next Descriptor Pointer
- * @reserved1:    Reserved and not used
+ * @next_msb:     MM2S/S2MM Next Descriptor Pointer (high 32 bits)
  * @phys:         MM2S/S2MM Buffer Address
- * @reserved2:    Reserved and not used
+ * @phys_msb:     MM2S/S2MM Buffer Address (high 32 bits)
  * @reserved3:    Reserved and not used
  * @reserved4:    Reserved and not used
  * @cntrl:        MM2S/S2MM Control value
···
  */
 struct axidma_bd {
 	u32 next;	/* Physical address of next buffer descriptor */
-	u32 reserved1;
+	u32 next_msb;	/* high 32 bits for IP >= v7.1, reserved on older IP */
 	u32 phys;
-	u32 reserved2;
+	u32 phys_msb;	/* for IP >= v7.1, reserved for older IP */
 	u32 reserved3;
 	u32 reserved4;
 	u32 cntrl;

drivers/net/ethernet/xilinx/xilinx_axienet_main.c

···
 	axienet_dma_out32(lp, reg, lower_32_bits(addr));
 }
 
+static void desc_set_phys_addr(struct axienet_local *lp, dma_addr_t addr,
+			       struct axidma_bd *desc)
+{
+	desc->phys = lower_32_bits(addr);
+	if (lp->features & XAE_FEATURE_DMA_64BIT)
+		desc->phys_msb = upper_32_bits(addr);
+}
+
+static dma_addr_t desc_get_phys_addr(struct axienet_local *lp,
+				     struct axidma_bd *desc)
+{
+	dma_addr_t ret = desc->phys;
+
+	if (lp->features & XAE_FEATURE_DMA_64BIT)
+		ret |= ((dma_addr_t)desc->phys_msb << 16) << 16;
+
+	return ret;
+}
+
 /**
  * axienet_dma_bd_release - Release buffer descriptor rings
  * @ndev:	Pointer to the net_device structure
···
 		return;
 
 	for (i = 0; i < lp->rx_bd_num; i++) {
+		dma_addr_t phys;
+
 		/* A NULL skb means this descriptor has not been initialised
 		 * at all.
 		 */
···
 		 * descriptor size, after it had been successfully allocated.
 		 * So a non-zero value in there means we need to unmap it.
 		 */
-		if (lp->rx_bd_v[i].cntrl)
-			dma_unmap_single(ndev->dev.parent, lp->rx_bd_v[i].phys,
+		if (lp->rx_bd_v[i].cntrl) {
+			phys = desc_get_phys_addr(lp, &lp->rx_bd_v[i]);
+			dma_unmap_single(ndev->dev.parent, phys,
 					 lp->max_frm_size, DMA_FROM_DEVICE);
+		}
 	}
 
 	dma_free_coherent(ndev->dev.parent,
···
 		goto out;
 
 	for (i = 0; i < lp->tx_bd_num; i++) {
-		lp->tx_bd_v[i].next = lp->tx_bd_p +
-				      sizeof(*lp->tx_bd_v) *
-				      ((i + 1) % lp->tx_bd_num);
+		dma_addr_t addr = lp->tx_bd_p +
+				  sizeof(*lp->tx_bd_v) *
+				  ((i + 1) % lp->tx_bd_num);
+
+		lp->tx_bd_v[i].next = lower_32_bits(addr);
+		if (lp->features & XAE_FEATURE_DMA_64BIT)
+			lp->tx_bd_v[i].next_msb = upper_32_bits(addr);
 	}
 
 	for (i = 0; i < lp->rx_bd_num; i++) {
-		lp->rx_bd_v[i].next = lp->rx_bd_p +
-				      sizeof(*lp->rx_bd_v) *
-				      ((i + 1) % lp->rx_bd_num);
+		dma_addr_t addr;
+
+		addr = lp->rx_bd_p + sizeof(*lp->rx_bd_v) *
+			((i + 1) % lp->rx_bd_num);
+		lp->rx_bd_v[i].next = lower_32_bits(addr);
+		if (lp->features & XAE_FEATURE_DMA_64BIT)
+			lp->rx_bd_v[i].next_msb = upper_32_bits(addr);
 
 		skb = netdev_alloc_skb_ip_align(ndev, lp->max_frm_size);
 		if (!skb)
 			goto out;
 
 		lp->rx_bd_v[i].skb = skb;
-		lp->rx_bd_v[i].phys = dma_map_single(ndev->dev.parent,
-						     skb->data,
-						     lp->max_frm_size,
-						     DMA_FROM_DEVICE);
-		if (dma_mapping_error(ndev->dev.parent, lp->rx_bd_v[i].phys)) {
+		addr = dma_map_single(ndev->dev.parent, skb->data,
+				      lp->max_frm_size, DMA_FROM_DEVICE);
+		if (dma_mapping_error(ndev->dev.parent, addr)) {
 			netdev_err(ndev, "DMA mapping error\n");
 			goto out;
 		}
+		desc_set_phys_addr(lp, addr, &lp->rx_bd_v[i]);
 
 		lp->rx_bd_v[i].cntrl = lp->max_frm_size;
 	}
···
 	struct axidma_bd *cur_p;
 	int max_bds = nr_bds;
 	unsigned int status;
+	dma_addr_t phys;
 	int i;
 
 	if (max_bds == -1)
···
 		if (nr_bds == -1 && !(status & XAXIDMA_BD_STS_COMPLETE_MASK))
 			break;
 
-		dma_unmap_single(ndev->dev.parent, cur_p->phys,
-				(cur_p->cntrl & XAXIDMA_BD_CTRL_LENGTH_MASK),
-				DMA_TO_DEVICE);
+		phys = desc_get_phys_addr(lp, cur_p);
+		dma_unmap_single(ndev->dev.parent, phys,
+				 (cur_p->cntrl & XAXIDMA_BD_CTRL_LENGTH_MASK),
+				 DMA_TO_DEVICE);
 
 		if (cur_p->skb && (status & XAXIDMA_BD_STS_COMPLETE_MASK))
 			dev_consume_skb_irq(cur_p->skb);
···
 	u32 csum_start_off;
 	u32 csum_index_off;
 	skb_frag_t *frag;
-	dma_addr_t tail_p;
+	dma_addr_t tail_p, phys;
 	struct axienet_local *lp = netdev_priv(ndev);
 	struct axidma_bd *cur_p;
 	u32 orig_tail_ptr = lp->tx_bd_tail;
···
 		cur_p->app0 |= 2; /* Tx Full Checksum Offload Enabled */
 	}
 
-	cur_p->phys = dma_map_single(ndev->dev.parent, skb->data,
-				     skb_headlen(skb), DMA_TO_DEVICE);
-	if (unlikely(dma_mapping_error(ndev->dev.parent, cur_p->phys))) {
+	phys = dma_map_single(ndev->dev.parent, skb->data,
+			      skb_headlen(skb), DMA_TO_DEVICE);
+	if (unlikely(dma_mapping_error(ndev->dev.parent, phys))) {
 		if (net_ratelimit())
 			netdev_err(ndev, "TX DMA mapping error\n");
 		ndev->stats.tx_dropped++;
 		return NETDEV_TX_OK;
 	}
+	desc_set_phys_addr(lp, phys, cur_p);
 	cur_p->cntrl = skb_headlen(skb) | XAXIDMA_BD_CTRL_TXSOF_MASK;
 
 	for (ii = 0; ii < num_frag; ii++) {
···
 			lp->tx_bd_tail = 0;
 		cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
 		frag = &skb_shinfo(skb)->frags[ii];
-		cur_p->phys = dma_map_single(ndev->dev.parent,
-					     skb_frag_address(frag),
-					     skb_frag_size(frag),
-					     DMA_TO_DEVICE);
-		if (unlikely(dma_mapping_error(ndev->dev.parent, cur_p->phys))) {
+		phys = dma_map_single(ndev->dev.parent,
+				      skb_frag_address(frag),
+				      skb_frag_size(frag),
+				      DMA_TO_DEVICE);
+		if (unlikely(dma_mapping_error(ndev->dev.parent, phys))) {
 			if (net_ratelimit())
 				netdev_err(ndev, "TX DMA mapping error\n");
 			ndev->stats.tx_dropped++;
···
 
 			return NETDEV_TX_OK;
 		}
+		desc_set_phys_addr(lp, phys, cur_p);
 		cur_p->cntrl = skb_frag_size(frag);
 	}
 
···
 	cur_p = &lp->rx_bd_v[lp->rx_bd_ci];
 
 	while ((cur_p->status & XAXIDMA_BD_STS_COMPLETE_MASK)) {
+		dma_addr_t phys;
+
 		tail_p = lp->rx_bd_p + sizeof(*lp->rx_bd_v) * lp->rx_bd_ci;
 
-		dma_unmap_single(ndev->dev.parent, cur_p->phys,
-				 lp->max_frm_size,
+		phys = desc_get_phys_addr(lp, cur_p);
+		dma_unmap_single(ndev->dev.parent, phys, lp->max_frm_size,
 				 DMA_FROM_DEVICE);
 
 		skb = cur_p->skb;
···
 		if (!new_skb)
 			return;
 
-		cur_p->phys = dma_map_single(ndev->dev.parent, new_skb->data,
-					     lp->max_frm_size,
-					     DMA_FROM_DEVICE);
-		if (unlikely(dma_mapping_error(ndev->dev.parent, cur_p->phys))) {
+		phys = dma_map_single(ndev->dev.parent, new_skb->data,
+				      lp->max_frm_size,
+				      DMA_FROM_DEVICE);
+		if (unlikely(dma_mapping_error(ndev->dev.parent, phys))) {
 			if (net_ratelimit())
 				netdev_err(ndev, "RX DMA mapping error\n");
 			dev_kfree_skb(new_skb);
 			return;
 		}
+		desc_set_phys_addr(lp, phys, cur_p);
 
 		cur_p->cntrl = lp->max_frm_size;
 		cur_p->status = 0;
···
 		return IRQ_NONE;
 	if (status & XAXIDMA_IRQ_ERROR_MASK) {
 		dev_err(&ndev->dev, "DMA Tx error 0x%x\n", status);
-		dev_err(&ndev->dev, "Current BD is at: 0x%x\n",
+		dev_err(&ndev->dev, "Current BD is at: 0x%x%08x\n",
+			(lp->tx_bd_v[lp->tx_bd_ci]).phys_msb,
 			(lp->tx_bd_v[lp->tx_bd_ci]).phys);
 
 		cr = axienet_dma_in32(lp, XAXIDMA_TX_CR_OFFSET);
···
 		return IRQ_NONE;
 	if (status & XAXIDMA_IRQ_ERROR_MASK) {
 		dev_err(&ndev->dev, "DMA Rx error 0x%x\n", status);
-		dev_err(&ndev->dev, "Current BD is at: 0x%x\n",
+		dev_err(&ndev->dev, "Current BD is at: 0x%x%08x\n",
+			(lp->rx_bd_v[lp->rx_bd_ci]).phys_msb,
 			(lp->rx_bd_v[lp->rx_bd_ci]).phys);
 
 		cr = axienet_dma_in32(lp, XAXIDMA_TX_CR_OFFSET);
···
 
 	for (i = 0; i < lp->tx_bd_num; i++) {
 		cur_p = &lp->tx_bd_v[i];
-		if (cur_p->cntrl)
-			dma_unmap_single(ndev->dev.parent, cur_p->phys,
+		if (cur_p->cntrl) {
+			dma_addr_t addr = desc_get_phys_addr(lp, cur_p);
+
+			dma_unmap_single(ndev->dev.parent, addr,
 					 (cur_p->cntrl &
 					  XAXIDMA_BD_CTRL_LENGTH_MASK),
 					 DMA_TO_DEVICE);
+		}
 		if (cur_p->skb)
 			dev_kfree_skb_irq(cur_p->skb);
 		cur_p->phys = 0;
+		cur_p->phys_msb = 0;
 		cur_p->cntrl = 0;
 		cur_p->status = 0;
 		cur_p->app0 = 0;