ixgbe: Match on multiple headers for cls_u32 offloads

Adds support to set filters with multiple header fields (L3,L4)to match on.
This is achieved in the following order:
1. Create a leaf hash table for the next header.
2. Create a link to the leaf hash table from the base hash table with
   matches on next header type and current header fields.
3. Add filter in leaf hash table with match on next header fields and
   action.

Verified with the following filters :

Match TCP and DIP:
        handle 1: u32 divisor 1
        u32 ht 800: order 1 link 1: \
        offset at 0 mask 0f00 shift 6 plus 0 eat \
        match ip protocol 6 ff match ip dst 10.0.0.1/32
        match tcp src 28 ffff action drop

Delete the filter:

Match on DIP, SIP, UDP (SPort, DPort):
        handle 2: u32 divisor 1
        u32 ht 800: order 2 link 2: \
        offset at 0 mask 0f00 shift 6 plus 0 eat \
        match ip dst 15.0.0.2/32 match ip protocol 17 ff \
        match ip src 15.0.0.1/32
        match udp src 30 ffff match udp dst 32 ffff action drop
Signed-off-by: NAmritha Nambiar <amritha.nambiar@intel.com>
Acked-by: NSridhar Samudrala <sridhar.samudrala@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

drivers/net/ethernet/intel/ixgbe/ixgbe.h

@@ -792,7 +792,7 @@ struct ixgbe_adapter {
 	unsigned long fwd_bitmask; /* Bitmask indicating in use pools */
 
 #define IXGBE_MAX_LINK_HANDLE 10
-	struct ixgbe_mat_field *jump_tables[IXGBE_MAX_LINK_HANDLE];
+	struct ixgbe_jump_table *jump_tables[IXGBE_MAX_LINK_HANDLE];
 	unsigned long tables;
 
 /* maximum number of RETA entries among all devices supported by ixgbe

drivers/net/ethernet/intel/ixgbe/ixgbe_main.c

@@ -5574,6 +5574,7 @@ static int ixgbe_sw_init(struct ixgbe_adapter *adapter)
 	unsigned int rss, fdir;
 	u32 fwsm;
 	u16 device_caps;
+	int i;
 #ifdef CONFIG_IXGBE_DCB
 	int j;
 	struct tc_configuration *tc;
@@ -5609,7 +5610,14 @@ static int ixgbe_sw_init(struct ixgbe_adapter *adapter)
 #endif /* IXGBE_FCOE */
 
 	/* initialize static ixgbe jump table entries */
-	adapter->jump_tables[0] = ixgbe_ipv4_fields;
+	adapter->jump_tables[0] = kzalloc(sizeof(*adapter->jump_tables[0]),
+					  GFP_KERNEL);
+	if (!adapter->jump_tables[0])
+		return -ENOMEM;
+	adapter->jump_tables[0]->mat = ixgbe_ipv4_fields;
+
+	for (i = 1; i < IXGBE_MAX_LINK_HANDLE; i++)
+		adapter->jump_tables[i] = NULL;
 
 	adapter->mac_table = kzalloc(sizeof(struct ixgbe_mac_addr) *
 				     hw->mac.num_rar_entries,
@@ -8399,6 +8407,55 @@ static int parse_tc_actions(struct ixgbe_adapter *adapter,
 }
 #endif /* CONFIG_NET_CLS_ACT */
 
+static int ixgbe_clsu32_build_input(struct ixgbe_fdir_filter *input,
+				    union ixgbe_atr_input *mask,
+				    struct tc_cls_u32_offload *cls,
+				    struct ixgbe_mat_field *field_ptr,
+				    struct ixgbe_nexthdr *nexthdr)
+{
+	int i, j, off;
+	__be32 val, m;
+	bool found_entry = false, found_jump_field = false;
+
+	for (i = 0; i < cls->knode.sel->nkeys; i++) {
+		off = cls->knode.sel->keys[i].off;
+		val = cls->knode.sel->keys[i].val;
+		m = cls->knode.sel->keys[i].mask;
+
+		for (j = 0; field_ptr[j].val; j++) {
+			if (field_ptr[j].off == off) {
+				field_ptr[j].val(input, mask, val, m);
+				input->filter.formatted.flow_type |=
+					field_ptr[j].type;
+				found_entry = true;
+				break;
+			}
+		}
+		if (nexthdr) {
+			if (nexthdr->off == cls->knode.sel->keys[i].off &&
+			    nexthdr->val == cls->knode.sel->keys[i].val &&
+			    nexthdr->mask == cls->knode.sel->keys[i].mask)
+				found_jump_field = true;
+			else
+				continue;
+		}
+	}
+
+	if (nexthdr && !found_jump_field)
+		return -EINVAL;
+
+	if (!found_entry)
+		return 0;
+
+	mask->formatted.flow_type = IXGBE_ATR_L4TYPE_IPV6_MASK |
+				    IXGBE_ATR_L4TYPE_MASK;
+
+	if (input->filter.formatted.flow_type == IXGBE_ATR_FLOW_TYPE_IPV4)
+		mask->formatted.flow_type &= IXGBE_ATR_L4TYPE_IPV6_MASK;
+
+	return 0;
+}
+
 static int ixgbe_configure_clsu32(struct ixgbe_adapter *adapter,
 				  __be16 protocol,
 				  struct tc_cls_u32_offload *cls)
@@ -8406,13 +8463,13 @@ static int ixgbe_configure_clsu32(struct ixgbe_adapter *adapter,
 	u32 loc = cls->knode.handle & 0xfffff;
 	struct ixgbe_hw *hw = &adapter->hw;
 	struct ixgbe_mat_field *field_ptr;
-	struct ixgbe_fdir_filter *input;
-	union ixgbe_atr_input mask;
-	int i, err = 0;
+	struct ixgbe_fdir_filter *input = NULL;
+	union ixgbe_atr_input *mask = NULL;
+	struct ixgbe_jump_table *jump = NULL;
+	int i, err = -EINVAL;
 	u8 queue;
 	u32 uhtid, link_uhtid;
 
-	memset(&mask, 0, sizeof(union ixgbe_atr_input));
 	uhtid = TC_U32_USERHTID(cls->knode.handle);
 	link_uhtid = TC_U32_USERHTID(cls->knode.link_handle);
 
@@ -8424,39 +8481,11 @@ static int ixgbe_configure_clsu32(struct ixgbe_adapter *adapter,
 	 * headers when needed.
 	 */
 	if (protocol != htons(ETH_P_IP))
-		return -EINVAL;
-
-	if (link_uhtid) {
-		struct ixgbe_nexthdr *nexthdr = ixgbe_ipv4_jumps;
-
-		if (link_uhtid >= IXGBE_MAX_LINK_HANDLE)
-			return -EINVAL;
-
-		if (!test_bit(link_uhtid - 1, &adapter->tables))
-			return -EINVAL;
-
-		for (i = 0; nexthdr[i].jump; i++) {
-			if (nexthdr[i].o != cls->knode.sel->offoff ||
-			    nexthdr[i].s != cls->knode.sel->offshift ||
-			    nexthdr[i].m != cls->knode.sel->offmask ||
-			    /* do not support multiple key jumps its just mad */
-			    cls->knode.sel->nkeys > 1)
-				return -EINVAL;
-
-			if (nexthdr[i].off == cls->knode.sel->keys[0].off &&
-			    nexthdr[i].val == cls->knode.sel->keys[0].val &&
-			    nexthdr[i].mask == cls->knode.sel->keys[0].mask) {
-				adapter->jump_tables[link_uhtid] =
-								nexthdr[i].jump;
-				break;
-			}
-		}
-		return 0;
-	}
+		return err;
 
 	if (loc >= ((1024 << adapter->fdir_pballoc) - 2)) {
 		e_err(drv, "Location out of range\n");
-		return -EINVAL;
+		return err;
 	}
 
 	/* cls u32 is a graph starting at root node 0x800. The driver tracks
@@ -8467,47 +8496,85 @@ static int ixgbe_configure_clsu32(struct ixgbe_adapter *adapter,
 	 * this function _should_ be generic try not to hardcode values here.
 	 */
 	if (uhtid == 0x800) {
-		field_ptr = adapter->jump_tables[0];
+		field_ptr = (adapter->jump_tables[0])->mat;
 	} else {
 		if (uhtid >= IXGBE_MAX_LINK_HANDLE)
-			return -EINVAL;
-
-		field_ptr = adapter->jump_tables[uhtid];
+			return err;
+		if (!adapter->jump_tables[uhtid])
+			return err;
+		field_ptr = (adapter->jump_tables[uhtid])->mat;
 	}
 
 	if (!field_ptr)
-		return -EINVAL;
+		return err;
 
-	input = kzalloc(sizeof(*input), GFP_KERNEL);
-	if (!input)
-		return -ENOMEM;
+	/* At this point we know the field_ptr is valid and need to either
+	 * build cls_u32 link or attach filter. Because adding a link to
+	 * a handle that does not exist is invalid and the same for adding
+	 * rules to handles that don't exist.
+	 */
 
-	for (i = 0; i < cls->knode.sel->nkeys; i++) {
-		int off = cls->knode.sel->keys[i].off;
-		__be32 val = cls->knode.sel->keys[i].val;
-		__be32 m = cls->knode.sel->keys[i].mask;
-		bool found_entry = false;
-		int j;
+	if (link_uhtid) {
+		struct ixgbe_nexthdr *nexthdr = ixgbe_ipv4_jumps;
 
-		for (j = 0; field_ptr[j].val; j++) {
-			if (field_ptr[j].off == off) {
-				field_ptr[j].val(input, &mask, val, m);
-				input->filter.formatted.flow_type |=
-					field_ptr[j].type;
-				found_entry = true;
+		if (link_uhtid >= IXGBE_MAX_LINK_HANDLE)
+			return err;
+
+		if (!test_bit(link_uhtid - 1, &adapter->tables))
+			return err;
+
+		for (i = 0; nexthdr[i].jump; i++) {
+			if (nexthdr[i].o != cls->knode.sel->offoff ||
+			    nexthdr[i].s != cls->knode.sel->offshift ||
+			    nexthdr[i].m != cls->knode.sel->offmask)
+				return err;
+
+			jump = kzalloc(sizeof(*jump), GFP_KERNEL);
+			if (!jump)
+				return -ENOMEM;
+			input = kzalloc(sizeof(*input), GFP_KERNEL);
+			if (!input) {
+				err = -ENOMEM;
+				goto free_jump;
+			}
+			mask = kzalloc(sizeof(*mask), GFP_KERNEL);
+			if (!mask) {
+				err = -ENOMEM;
+				goto free_input;
+			}
+			jump->input = input;
+			jump->mask = mask;
+			err = ixgbe_clsu32_build_input(input, mask, cls,
+						       field_ptr, &nexthdr[i]);
+			if (!err) {
+				jump->mat = nexthdr[i].jump;
+				adapter->jump_tables[link_uhtid] = jump;
 				break;
 			}
 		}
-
-		if (!found_entry)
-			goto err_out;
+		return 0;
 	}
 
-	mask.formatted.flow_type = IXGBE_ATR_L4TYPE_IPV6_MASK |
-				   IXGBE_ATR_L4TYPE_MASK;
+	input = kzalloc(sizeof(*input), GFP_KERNEL);
+	if (!input)
+		return -ENOMEM;
+	mask = kzalloc(sizeof(*mask), GFP_KERNEL);
+	if (!mask) {
+		err = -ENOMEM;
+		goto free_input;
+	}
 
-	if (input->filter.formatted.flow_type == IXGBE_ATR_FLOW_TYPE_IPV4)
-		mask.formatted.flow_type &= IXGBE_ATR_L4TYPE_IPV6_MASK;
+	if ((uhtid != 0x800) && (adapter->jump_tables[uhtid])) {
+		if ((adapter->jump_tables[uhtid])->input)
+			memcpy(input, (adapter->jump_tables[uhtid])->input,
+			       sizeof(*input));
+		if ((adapter->jump_tables[uhtid])->mask)
+			memcpy(mask, (adapter->jump_tables[uhtid])->mask,
+			       sizeof(*mask));
+	}
+	err = ixgbe_clsu32_build_input(input, mask, cls, field_ptr, NULL);
+	if (err)
+		goto err_out;
 
 	err = parse_tc_actions(adapter, cls->knode.exts, &input->action,
 			       &queue);
@@ -8519,28 +8586,33 @@ static int ixgbe_configure_clsu32(struct ixgbe_adapter *adapter,
 	spin_lock(&adapter->fdir_perfect_lock);
 
 	if (hlist_empty(&adapter->fdir_filter_list)) {
-		memcpy(&adapter->fdir_mask, &mask, sizeof(mask));
-		err = ixgbe_fdir_set_input_mask_82599(hw, &mask);
+		memcpy(&adapter->fdir_mask, mask, sizeof(*mask));
+		err = ixgbe_fdir_set_input_mask_82599(hw, mask);
 		if (err)
 			goto err_out_w_lock;
-	} else if (memcmp(&adapter->fdir_mask, &mask, sizeof(mask))) {
+	} else if (memcmp(&adapter->fdir_mask, mask, sizeof(*mask))) {
 		err = -EINVAL;
 		goto err_out_w_lock;
 	}
 
-	ixgbe_atr_compute_perfect_hash_82599(&input->filter, &mask);
+	ixgbe_atr_compute_perfect_hash_82599(&input->filter, mask);
 	err = ixgbe_fdir_write_perfect_filter_82599(hw, &input->filter,
 						    input->sw_idx, queue);
 	if (!err)
 		ixgbe_update_ethtool_fdir_entry(adapter, input, input->sw_idx);
 	spin_unlock(&adapter->fdir_perfect_lock);
 
+	kfree(mask);
 	return err;
 err_out_w_lock:
 	spin_unlock(&adapter->fdir_perfect_lock);
 err_out:
+	kfree(mask);
+free_input:
 	kfree(input);
-	return -EINVAL;
+free_jump:
+	kfree(jump);
+	return err;
 }
 
 static int __ixgbe_setup_tc(struct net_device *dev, u32 handle, __be16 proto,
@@ -9612,6 +9684,7 @@ static int ixgbe_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	ixgbe_disable_sriov(adapter);
 	adapter->flags2 &= ~IXGBE_FLAG2_SEARCH_FOR_SFP;
 	iounmap(adapter->io_addr);
+	kfree(adapter->jump_tables[0]);
 	kfree(adapter->mac_table);
 err_ioremap:
 	disable_dev = !test_and_set_bit(__IXGBE_DISABLED, &adapter->state);
@@ -9640,6 +9713,7 @@ static void ixgbe_remove(struct pci_dev *pdev)
 	struct ixgbe_adapter *adapter = pci_get_drvdata(pdev);
 	struct net_device *netdev;
 	bool disable_dev;
+	int i;
 
 	/* if !adapter then we already cleaned up in probe */
 	if (!adapter)
@@ -9689,6 +9763,14 @@ static void ixgbe_remove(struct pci_dev *pdev)
 
 	e_dev_info("complete\n");
 
+	for (i = 0; i < IXGBE_MAX_LINK_HANDLE; i++) {
+		if (adapter->jump_tables[i]) {
+			kfree(adapter->jump_tables[i]->input);
+			kfree(adapter->jump_tables[i]->mask);
+		}
+		kfree(adapter->jump_tables[i]);
+	}
+
 	kfree(adapter->mac_table);
 	disable_dev = !test_and_set_bit(__IXGBE_DISABLED, &adapter->state);
 	free_netdev(netdev);

drivers/net/ethernet/intel/ixgbe/ixgbe_model.h

@@ -38,6 +38,12 @@ struct ixgbe_mat_field {
 	unsigned int type;
 };
 
+struct ixgbe_jump_table {
+	struct ixgbe_mat_field *mat;
+	struct ixgbe_fdir_filter *input;
+	union ixgbe_atr_input *mask;
+};
+
 static inline int ixgbe_mat_prgm_sip(struct ixgbe_fdir_filter *input,
 				     union ixgbe_atr_input *mask,
 				     u32 val, u32 m)