ibmvnic: map L2/L3/L4 header descriptors to firmware

Allow the VNIC driver to provide descriptors containing
L2/L3/L4 headers to firmware.  This feature is needed
for greater hardware compatibility and enablement of checksum
and TCP offloading features.

A new function is included for the hypervisor call,
H_SEND_SUBCRQ_INDIRECT, allowing a DMA-mapped array of SCRQ
descriptor elements to be sent to the VNIC server.

These additions will help fully enable checksum offloading as
well as other features as they are included later.

Signed-off-by: Thomas Falcon <tlfalcon@linux.vnet.ibm.com>
Cc: John Allen <jallen@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

2 files changed, 194 insertions, 4 deletions

diff --git a/drivers/net/ethernet/ibm/ibmvnic.c b/drivers/net/ethernet/ibm/ibmvnic.c
index 6e9e16eee5d0..4e97e762e297 100644
--- a/drivers/net/ethernet/ibm/ibmvnic.c
+++ b/drivers/net/ethernet/ibm/ibmvnic.c
@@ -61,6 +61,7 @@
 #include <linux/proc_fs.h>
 #include <linux/in.h>
 #include <linux/ip.h>
+#include <linux/ipv6.h>
 #include <linux/irq.h>
 #include <linux/kthread.h>
 #include <linux/seq_file.h>
@@ -94,6 +95,7 @@ static int ibmvnic_reenable_crq_queue(struct ibmvnic_adapter *);
 static int ibmvnic_send_crq(struct ibmvnic_adapter *, union ibmvnic_crq *);
 static int send_subcrq(struct ibmvnic_adapter *adapter, u64 remote_handle,
 		       union sub_crq *sub_crq);
+static int send_subcrq_indirect(struct ibmvnic_adapter *, u64, u64, u64);
 static irqreturn_t ibmvnic_interrupt_rx(int irq, void *instance);
 static int enable_scrq_irq(struct ibmvnic_adapter *,
 			   struct ibmvnic_sub_crq_queue *);
@@ -561,10 +563,141 @@ static int ibmvnic_close(struct net_device *netdev)
 	return 0;
 }
 
+/**
+ * build_hdr_data - creates L2/L3/L4 header data buffer
+ * @hdr_field - bitfield determining needed headers
+ * @skb - socket buffer
+ * @hdr_len - array of header lengths
+ * @tot_len - total length of data
+ *
+ * Reads hdr_field to determine which headers are needed by firmware.
+ * Builds a buffer containing these headers.  Saves individual header
+ * lengths and total buffer length to be used to build descriptors.
+ */
+static int build_hdr_data(u8 hdr_field, struct sk_buff *skb,
+			  int *hdr_len, u8 *hdr_data)
+{
+	int len = 0;
+	u8 *hdr;
+
+	hdr_len[0] = sizeof(struct ethhdr);
+
+	if (skb->protocol == htons(ETH_P_IP)) {
+		hdr_len[1] = ip_hdr(skb)->ihl * 4;
+		if (ip_hdr(skb)->protocol == IPPROTO_TCP)
+			hdr_len[2] = tcp_hdrlen(skb);
+		else if (ip_hdr(skb)->protocol == IPPROTO_UDP)
+			hdr_len[2] = sizeof(struct udphdr);
+	} else if (skb->protocol == htons(ETH_P_IPV6)) {
+		hdr_len[1] = sizeof(struct ipv6hdr);
+		if (ipv6_hdr(skb)->nexthdr == IPPROTO_TCP)
+			hdr_len[2] = tcp_hdrlen(skb);
+		else if (ipv6_hdr(skb)->nexthdr == IPPROTO_UDP)
+			hdr_len[2] = sizeof(struct udphdr);
+	}
+
+	memset(hdr_data, 0, 120);
+	if ((hdr_field >> 6) & 1) {
+		hdr = skb_mac_header(skb);
+		memcpy(hdr_data, hdr, hdr_len[0]);
+		len += hdr_len[0];
+	}
+
+	if ((hdr_field >> 5) & 1) {
+		hdr = skb_network_header(skb);
+		memcpy(hdr_data + len, hdr, hdr_len[1]);
+		len += hdr_len[1];
+	}
+
+	if ((hdr_field >> 4) & 1) {
+		hdr = skb_transport_header(skb);
+		memcpy(hdr_data + len, hdr, hdr_len[2]);
+		len += hdr_len[2];
+	}
+	return len;
+}
+
+/**
+ * create_hdr_descs - create header and header extension descriptors
+ * @hdr_field - bitfield determining needed headers
+ * @data - buffer containing header data
+ * @len - length of data buffer
+ * @hdr_len - array of individual header lengths
+ * @scrq_arr - descriptor array
+ *
+ * Creates header and, if needed, header extension descriptors and
+ * places them in a descriptor array, scrq_arr
+ */
+
+static void create_hdr_descs(u8 hdr_field, u8 *hdr_data, int len, int *hdr_len,
+			     union sub_crq *scrq_arr)
+{
+	union sub_crq hdr_desc;
+	int tmp_len = len;
+	u8 *data, *cur;
+	int tmp;
+
+	while (tmp_len > 0) {
+		cur = hdr_data + len - tmp_len;
+
+		memset(&hdr_desc, 0, sizeof(hdr_desc));
+		if (cur != hdr_data) {
+			data = hdr_desc.hdr_ext.data;
+			tmp = tmp_len > 29 ? 29 : tmp_len;
+			hdr_desc.hdr_ext.first = IBMVNIC_CRQ_CMD;
+			hdr_desc.hdr_ext.type = IBMVNIC_HDR_EXT_DESC;
+			hdr_desc.hdr_ext.len = tmp;
+		} else {
+			data = hdr_desc.hdr.data;
+			tmp = tmp_len > 24 ? 24 : tmp_len;
+			hdr_desc.hdr.first = IBMVNIC_CRQ_CMD;
+			hdr_desc.hdr.type = IBMVNIC_HDR_DESC;
+			hdr_desc.hdr.len = tmp;
+			hdr_desc.hdr.l2_len = (u8)hdr_len[0];
+			hdr_desc.hdr.l3_len = cpu_to_be16((u16)hdr_len[1]);
+			hdr_desc.hdr.l4_len = (u8)hdr_len[2];
+			hdr_desc.hdr.flag = hdr_field << 1;
+		}
+		memcpy(data, cur, tmp);
+		tmp_len -= tmp;
+		*scrq_arr = hdr_desc;
+		scrq_arr++;
+	}
+}
+
+/**
+ * build_hdr_descs_arr - build a header descriptor array
+ * @skb - socket buffer
+ * @num_entries - number of descriptors to be sent
+ * @subcrq - first TX descriptor
+ * @hdr_field - bit field determining which headers will be sent
+ *
+ * This function will build a TX descriptor array with applicable
+ * L2/L3/L4 packet header descriptors to be sent by send_subcrq_indirect.
+ */
+
+static void build_hdr_descs_arr(struct ibmvnic_tx_buff *txbuff,
+				int *num_entries, u8 hdr_field)
+{
+	int hdr_len[3] = {0, 0, 0};
+	int tot_len, len;
+	u8 *hdr_data = txbuff->hdr_data;
+
+	tot_len = build_hdr_data(hdr_field, txbuff->skb, hdr_len,
+				 txbuff->hdr_data);
+	len = tot_len;
+	len -= 24;
+	if (len > 0)
+		num_entries += len % 29 ? len / 29 + 1 : len / 29;
+	create_hdr_descs(hdr_field, hdr_data, tot_len, hdr_len,
+			 txbuff->indir_arr + 1);
+}
+
 static int ibmvnic_xmit(struct sk_buff *skb, struct net_device *netdev)
 {
 	struct ibmvnic_adapter *adapter = netdev_priv(netdev);
 	int queue_num = skb_get_queue_mapping(skb);
+	u8 *hdrs = (u8 *)&adapter->tx_rx_desc_req;
 	struct device *dev = &adapter->vdev->dev;
 	struct ibmvnic_tx_buff *tx_buff = NULL;
 	struct ibmvnic_tx_pool *tx_pool;
@@ -579,6 +712,7 @@ static int ibmvnic_xmit(struct sk_buff *skb, struct net_device *netdev)
 	unsigned long lpar_rc;
 	union sub_crq tx_crq;
 	unsigned int offset;
+	int num_entries = 1;
 	unsigned char *dst;
 	u64 *handle_array;
 	int index = 0;
@@ -644,11 +778,34 @@ static int ibmvnic_xmit(struct sk_buff *skb, struct net_device *netdev)
 			tx_crq.v1.flags1 |= IBMVNIC_TX_PROT_UDP;
 	}
 
-	if (skb->ip_summed == CHECKSUM_PARTIAL)
+	if (skb->ip_summed == CHECKSUM_PARTIAL) {
 		tx_crq.v1.flags1 |= IBMVNIC_TX_CHKSUM_OFFLOAD;
-
-	lpar_rc = send_subcrq(adapter, handle_array[0], &tx_crq);
-
+		hdrs += 2;
+	}
+	/* determine if l2/3/4 headers are sent to firmware */
+	if ((*hdrs >> 7) & 1 &&
+	    (skb->protocol == htons(ETH_P_IP) ||
+	     skb->protocol == htons(ETH_P_IPV6))) {
+		build_hdr_descs_arr(tx_buff, &num_entries, *hdrs);
+		tx_crq.v1.n_crq_elem = num_entries;
+		tx_buff->indir_arr[0] = tx_crq;
+		tx_buff->indir_dma = dma_map_single(dev, tx_buff->indir_arr,
+						    sizeof(tx_buff->indir_arr),
+						    DMA_TO_DEVICE);
+		if (dma_mapping_error(dev, tx_buff->indir_dma)) {
+			if (!firmware_has_feature(FW_FEATURE_CMO))
+				dev_err(dev, "tx: unable to map descriptor array\n");
+			tx_map_failed++;
+			tx_dropped++;
+			ret = NETDEV_TX_BUSY;
+			goto out;
+		}
+		lpar_rc = send_subcrq_indirect(adapter, handle_array[0],
+					       (u64)tx_buff->indir_dma,
+					       (u64)num_entries);
+	} else {
+		lpar_rc = send_subcrq(adapter, handle_array[0], &tx_crq);
+	}
 	if (lpar_rc != H_SUCCESS) {
 		dev_err(dev, "tx failed with code %ld\n", lpar_rc);
 
@@ -1159,6 +1316,7 @@ static int ibmvnic_complete_tx(struct ibmvnic_adapter *adapter,
 	union sub_crq *next;
 	int index;
 	int i, j;
+	u8 first;
 
 restart_loop:
 	while (pending_scrq(adapter, scrq)) {
@@ -1181,6 +1339,13 @@ restart_loop:
 				txbuff->data_dma[j] = 0;
 				txbuff->used_bounce = false;
 			}
+			/* if sub_crq was sent indirectly */
+			first = txbuff->indir_arr[0].generic.first;
+			if (first == IBMVNIC_CRQ_CMD) {
+				dma_unmap_single(dev, txbuff->indir_dma,
+						 sizeof(txbuff->indir_arr),
+						 DMA_TO_DEVICE);
+			}
 
 			if (txbuff->last_frag)
 				dev_kfree_skb_any(txbuff->skb);
@@ -1494,6 +1659,28 @@ static int send_subcrq(struct ibmvnic_adapter *adapter, u64 remote_handle,
 	return rc;
 }
 
+static int send_subcrq_indirect(struct ibmvnic_adapter *adapter,
+				u64 remote_handle, u64 ioba, u64 num_entries)
+{
+	unsigned int ua = adapter->vdev->unit_address;
+	struct device *dev = &adapter->vdev->dev;
+	int rc;
+
+	/* Make sure the hypervisor sees the complete request */
+	mb();
+	rc = plpar_hcall_norets(H_SEND_SUB_CRQ_INDIRECT, ua,
+				cpu_to_be64(remote_handle),
+				ioba, num_entries);
+
+	if (rc) {
+		if (rc == H_CLOSED)
+			dev_warn(dev, "CRQ Queue closed\n");
+		dev_err(dev, "Send (indirect) error (rc=%d)\n", rc);
+	}
+
+	return rc;
+}
+
 static int ibmvnic_send_crq(struct ibmvnic_adapter *adapter,
 			    union ibmvnic_crq *crq)
 {
diff --git a/drivers/net/ethernet/ibm/ibmvnic.h b/drivers/net/ethernet/ibm/ibmvnic.h
index 1a9993cc79b5..5af8a796e523 100644
--- a/drivers/net/ethernet/ibm/ibmvnic.h
+++ b/drivers/net/ethernet/ibm/ibmvnic.h
@@ -879,6 +879,9 @@ struct ibmvnic_tx_buff {
 	int pool_index;
 	bool last_frag;
 	bool used_bounce;
+	union sub_crq indir_arr[6];
+	u8 hdr_data[140];
+	dma_addr_t indir_dma;
 };
 
 struct ibmvnic_tx_pool {