i40e/i40evf: RSS changes for X722

X722 uses the admin queue to configure RSS. This patch adds the necessary
flow changes to configure RSS through AQ. It also adds the separate VMDQ2
lookup tables and hash key programming for X722.

X722 also exposes a different set of PCTYPES for RSS, this patch
accommodates those changes.

Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Tested-by: Jim Young <james.m.young@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

drivers/net/ethernet/intel/i40e/i40e.h

@@ -79,10 +79,13 @@
 #define I40E_MIN_MSIX                 2
 #define I40E_DEFAULT_NUM_VMDQ_VSI     8 /* max 256 VSIs */
 #define I40E_MIN_VSI_ALLOC            51 /* LAN, ATR, FCOE, 32 VF, 16 VMDQ */
-#define I40E_DEFAULT_QUEUES_PER_VMDQ  2 /* max 16 qps */
+/* max 16 qps */
+#define i40e_default_queues_per_vmdq(pf) \
+		(((pf)->flags & I40E_FLAG_RSS_AQ_CAPABLE) ? 4 : 1)
 #define I40E_DEFAULT_QUEUES_PER_VF    4
 #define I40E_DEFAULT_QUEUES_PER_TC    1 /* should be a power of 2 */
-#define I40E_MAX_QUEUES_PER_TC        64 /* should be a power of 2 */
+#define i40e_pf_get_max_q_per_tc(pf) \
+		(((pf)->flags & I40E_FLAG_128_QP_RSS_CAPABLE) ? 128 : 64)
 #define I40E_FDIR_RING                0
 #define I40E_FDIR_RING_COUNT          32
 #ifdef I40E_FCOE

drivers/net/ethernet/intel/i40e/i40e_main.c

@@ -1550,7 +1550,7 @@ static void i40e_vsi_setup_queue_map(struct i40e_vsi *vsi,
 	 */
 	qcount = min_t(int, vsi->alloc_queue_pairs, pf->num_lan_msix);
 	num_tc_qps = qcount / numtc;
-	num_tc_qps = min_t(int, num_tc_qps, I40E_MAX_QUEUES_PER_TC);
+	num_tc_qps = min_t(int, num_tc_qps, i40e_pf_get_max_q_per_tc(pf));
 
 	/* Setup queue offset/count for all TCs for given VSI */
 	for (i = 0; i < I40E_MAX_TRAFFIC_CLASS; i++) {
@@ -7469,62 +7469,139 @@ static int i40e_setup_misc_vector(struct i40e_pf *pf)
 }
 
 /**
- * i40e_config_rss - Prepare for RSS if used
+ * i40e_config_rss_aq - Prepare for RSS using AQ commands
+ * @vsi: vsi structure
+ * @seed: RSS hash seed
+ **/
+static int i40e_config_rss_aq(struct i40e_vsi *vsi, const u8 *seed)
+{
+	struct i40e_aqc_get_set_rss_key_data rss_key;
+	struct i40e_pf *pf = vsi->back;
+	struct i40e_hw *hw = &pf->hw;
+	bool pf_lut = false;
+	u8 *rss_lut;
+	int ret, i;
+
+	memset(&rss_key, 0, sizeof(rss_key));
+	memcpy(&rss_key, seed, sizeof(rss_key));
+
+	rss_lut = kzalloc(pf->rss_table_size, GFP_KERNEL);
+	if (!rss_lut)
+		return -ENOMEM;
+
+	/* Populate the LUT with max no. of queues in round robin fashion */
+	for (i = 0; i < vsi->rss_table_size; i++)
+		rss_lut[i] = i % vsi->rss_size;
+
+	ret = i40e_aq_set_rss_key(hw, vsi->id, &rss_key);
+	if (ret) {
+		dev_info(&pf->pdev->dev,
+			 "Cannot set RSS key, err %s aq_err %s\n",
+			 i40e_stat_str(&pf->hw, ret),
+			 i40e_aq_str(&pf->hw, pf->hw.aq.asq_last_status));
+		return ret;
+	}
+
+	if (vsi->type == I40E_VSI_MAIN)
+		pf_lut = true;
+
+	ret = i40e_aq_set_rss_lut(hw, vsi->id, pf_lut, rss_lut,
+				  vsi->rss_table_size);
+	if (ret)
+		dev_info(&pf->pdev->dev,
+			 "Cannot set RSS lut, err %s aq_err %s\n",
+			 i40e_stat_str(&pf->hw, ret),
+			 i40e_aq_str(&pf->hw, pf->hw.aq.asq_last_status));
+
+	return ret;
+}
+
+/**
+ * i40e_vsi_config_rss - Prepare for VSI(VMDq) RSS if used
+ * @vsi: VSI structure
+ **/
+static int i40e_vsi_config_rss(struct i40e_vsi *vsi)
+{
+	u8 seed[I40E_HKEY_ARRAY_SIZE];
+	struct i40e_pf *pf = vsi->back;
+
+	netdev_rss_key_fill((void *)seed, I40E_HKEY_ARRAY_SIZE);
+	vsi->rss_size = min_t(int, pf->rss_size, vsi->num_queue_pairs);
+
+	if (pf->flags & I40E_FLAG_RSS_AQ_CAPABLE)
+		return i40e_config_rss_aq(vsi, seed);
+
+	return 0;
+}
+
+/**
+ * i40e_config_rss_reg - Prepare for RSS if used
  * @pf: board private structure
+ * @seed: RSS hash seed
  **/
-static int i40e_config_rss(struct i40e_pf *pf)
+static int i40e_config_rss_reg(struct i40e_pf *pf, const u8 *seed)
 {
-	u32 rss_key[I40E_PFQF_HKEY_MAX_INDEX + 1];
 	struct i40e_vsi *vsi = pf->vsi[pf->lan_vsi];
 	struct i40e_hw *hw = &pf->hw;
+	u32 *seed_dw = (u32 *)seed;
+	u32 current_queue = 0;
 	u32 lut = 0;
 	int i, j;
-	u64 hena;
-	u32 reg_val;
 
-	netdev_rss_key_fill(rss_key, sizeof(rss_key));
+	/* Fill out hash function seed */
 	for (i = 0; i <= I40E_PFQF_HKEY_MAX_INDEX; i++)
-		wr32(hw, I40E_PFQF_HKEY(i), rss_key[i]);
+		wr32(hw, I40E_PFQF_HKEY(i), seed_dw[i]);
+
+	for (i = 0; i <= I40E_PFQF_HLUT_MAX_INDEX; i++) {
+		lut = 0;
+		for (j = 0; j < 4; j++) {
+			if (current_queue == vsi->rss_size)
+				current_queue = 0;
+			lut |= ((current_queue) << (8 * j));
+			current_queue++;
+		}
+		wr32(&pf->hw, I40E_PFQF_HLUT(i), lut);
+	}
+	i40e_flush(hw);
+
+	return 0;
+}
+
+/**
+ * i40e_config_rss - Prepare for RSS if used
+ * @pf: board private structure
+ **/
+static int i40e_config_rss(struct i40e_pf *pf)
+{
+	struct i40e_vsi *vsi = pf->vsi[pf->lan_vsi];
+	u8 seed[I40E_HKEY_ARRAY_SIZE];
+	struct i40e_hw *hw = &pf->hw;
+	u32 reg_val;
+	u64 hena;
+
+	netdev_rss_key_fill((void *)seed, I40E_HKEY_ARRAY_SIZE);
 
 	/* By default we enable TCP/UDP with IPv4/IPv6 ptypes */
 	hena = (u64)rd32(hw, I40E_PFQF_HENA(0)) |
 		((u64)rd32(hw, I40E_PFQF_HENA(1)) << 32);
-	hena |= I40E_DEFAULT_RSS_HENA;
+	hena |= i40e_pf_get_default_rss_hena(pf);
+
 	wr32(hw, I40E_PFQF_HENA(0), (u32)hena);
 	wr32(hw, I40E_PFQF_HENA(1), (u32)(hena >> 32));
 
 	vsi->rss_size = min_t(int, pf->rss_size, vsi->num_queue_pairs);
 
-	/* Check capability and Set table size and register per hw expectation*/
+	/* Determine the RSS table size based on the hardware capabilities */
 	reg_val = rd32(hw, I40E_PFQF_CTL_0);
-	if (pf->rss_table_size == 512)
-		reg_val |= I40E_PFQF_CTL_0_HASHLUTSIZE_512;
-	else
-		reg_val &= ~I40E_PFQF_CTL_0_HASHLUTSIZE_512;
+	reg_val = (pf->rss_table_size == 512) ?
+			(reg_val | I40E_PFQF_CTL_0_HASHLUTSIZE_512) :
+			(reg_val & ~I40E_PFQF_CTL_0_HASHLUTSIZE_512);
 	wr32(hw, I40E_PFQF_CTL_0, reg_val);
 
-	/* Populate the LUT with max no. of queues in round robin fashion */
-	for (i = 0, j = 0; i < pf->rss_table_size; i++, j++) {
-
-		/* The assumption is that lan qp count will be the highest
-		 * qp count for any PF VSI that needs RSS.
-		 * If multiple VSIs need RSS support, all the qp counts
-		 * for those VSIs should be a power of 2 for RSS to work.
-		 * If LAN VSI is the only consumer for RSS then this requirement
-		 * is not necessary.
-		 */
-		if (j == vsi->rss_size)
-			j = 0;
-		/* lut = 4-byte sliding window of 4 lut entries */
-		lut = (lut << 8) | (j &
-			 (BIT(pf->hw.func_caps.rss_table_entry_width) - 1));
-		/* On i = 3, we have 4 entries in lut; write to the register */
-		if ((i & 3) == 3)
-			wr32(hw, I40E_PFQF_HLUT(i >> 2), lut);
-	}
-	i40e_flush(hw);
-
-	return 0;
+	if (pf->flags & I40E_FLAG_RSS_AQ_CAPABLE)
+		return i40e_config_rss_aq(pf->vsi[pf->lan_vsi], seed);
+	else
+		return i40e_config_rss_reg(pf, seed);
 }
 
 /**
@@ -7765,9 +7842,8 @@ static int i40e_sw_init(struct i40e_pf *pf)
 	}
 
 	if (pf->hw.func_caps.vmdq) {
-		pf->flags |= I40E_FLAG_VMDQ_ENABLED;
 		pf->num_vmdq_vsis = I40E_DEFAULT_NUM_VMDQ_VSI;
-		pf->num_vmdq_qps = I40E_DEFAULT_QUEUES_PER_VMDQ;
+		pf->flags |= I40E_FLAG_VMDQ_ENABLED;
 	}
 
 #ifdef I40E_FCOE
@@ -8948,6 +9024,10 @@ struct i40e_vsi *i40e_vsi_setup(struct i40e_pf *pf, u8 type,
 		break;
 	}
 
+	if ((pf->flags & I40E_FLAG_RSS_AQ_CAPABLE) &&
+	    (vsi->type == I40E_VSI_VMDQ2)) {
+		ret = i40e_vsi_config_rss(vsi);
+	}
 	return vsi;
 
 err_rings:

drivers/net/ethernet/intel/i40e/i40e_txrx.h

@@ -78,6 +78,18 @@ enum i40e_dyn_idx_t {
 	BIT_ULL(I40E_FILTER_PCTYPE_FRAG_IPV6) | \
 	BIT_ULL(I40E_FILTER_PCTYPE_L2_PAYLOAD))
 
+#define I40E_DEFAULT_RSS_HENA_EXPANDED (I40E_DEFAULT_RSS_HENA | \
+	BIT(I40E_FILTER_PCTYPE_NONF_IPV4_TCP_SYN_NO_ACK) | \
+	BIT(I40E_FILTER_PCTYPE_NONF_UNICAST_IPV4_UDP) | \
+	BIT(I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV4_UDP) | \
+	BIT(I40E_FILTER_PCTYPE_NONF_IPV6_TCP_SYN_NO_ACK) | \
+	BIT(I40E_FILTER_PCTYPE_NONF_UNICAST_IPV6_UDP) | \
+	BIT(I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV6_UDP))
+
+#define i40e_pf_get_default_rss_hena(pf) \
+	(((pf)->flags & I40E_FLAG_MULTIPLE_TCP_UDP_RSS_PCTYPE) ? \
+	  I40E_DEFAULT_RSS_HENA_EXPANDED : I40E_DEFAULT_RSS_HENA)
+
 /* Supported Rx Buffer Sizes */
 #define I40E_RXBUFFER_512   512    /* Used for packet split */
 #define I40E_RXBUFFER_2048  2048

drivers/net/ethernet/intel/i40e/i40e_type.h

@@ -973,15 +973,24 @@ struct i40e_filter_program_desc {
 
 /* Packet Classifier Types for filters */
 enum i40e_filter_pctype {
-	/* Note: Values 0-30 are reserved for future use */
+	/* Note: Values 0-28 are reserved for future use.
+	 * Value 29, 30, 32 are not supported on XL710 and X710.
+	 */
+	I40E_FILTER_PCTYPE_NONF_UNICAST_IPV4_UDP	= 29,
+	I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV4_UDP	= 30,
 	I40E_FILTER_PCTYPE_NONF_IPV4_UDP		= 31,
-	/* Note: Value 32 is reserved for future use */
+	I40E_FILTER_PCTYPE_NONF_IPV4_TCP_SYN_NO_ACK	= 32,
 	I40E_FILTER_PCTYPE_NONF_IPV4_TCP		= 33,
 	I40E_FILTER_PCTYPE_NONF_IPV4_SCTP		= 34,
 	I40E_FILTER_PCTYPE_NONF_IPV4_OTHER		= 35,
 	I40E_FILTER_PCTYPE_FRAG_IPV4			= 36,
-	/* Note: Values 37-40 are reserved for future use */
+	/* Note: Values 37-38 are reserved for future use.
+	 * Value 39, 40, 42 are not supported on XL710 and X710.
+	 */
+	I40E_FILTER_PCTYPE_NONF_UNICAST_IPV6_UDP	= 39,
+	I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV6_UDP	= 40,
 	I40E_FILTER_PCTYPE_NONF_IPV6_UDP		= 41,
+	I40E_FILTER_PCTYPE_NONF_IPV6_TCP_SYN_NO_ACK	= 42,
 	I40E_FILTER_PCTYPE_NONF_IPV6_TCP		= 43,
 	I40E_FILTER_PCTYPE_NONF_IPV6_SCTP		= 44,
 	I40E_FILTER_PCTYPE_NONF_IPV6_OTHER		= 45,

drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c

@@ -1177,9 +1177,14 @@ static int i40e_vc_get_vf_resources_msg(struct i40e_vf *vf, u8 *msg)
 	vfres->vf_offload_flags = I40E_VIRTCHNL_VF_OFFLOAD_L2;
 	vsi = pf->vsi[vf->lan_vsi_idx];
 	if (!vsi->info.pvid)
-		vfres->vf_offload_flags |= I40E_VIRTCHNL_VF_OFFLOAD_VLAN |
-					   I40E_VIRTCHNL_VF_OFFLOAD_RSS_REG;
-
+		vfres->vf_offload_flags |= I40E_VIRTCHNL_VF_OFFLOAD_VLAN;
+	if (pf->flags & I40E_FLAG_RSS_AQ_CAPABLE) {
+		if (vf->driver_caps & I40E_VIRTCHNL_VF_OFFLOAD_RSS_AQ)
+			vfres->vf_offload_flags |=
+				I40E_VIRTCHNL_VF_OFFLOAD_RSS_AQ;
+	} else {
+		vfres->vf_offload_flags |= I40E_VIRTCHNL_VF_OFFLOAD_RSS_REG;
+	}
 	vfres->num_vsis = num_vsis;
 	vfres->num_queue_pairs = vf->num_queue_pairs;
 	vfres->max_vectors = pf->hw.func_caps.num_msix_vectors_vf;

drivers/net/ethernet/intel/i40evf/i40e_txrx.h

@@ -78,6 +78,18 @@ enum i40e_dyn_idx_t {
 	BIT_ULL(I40E_FILTER_PCTYPE_FRAG_IPV6) | \
 	BIT_ULL(I40E_FILTER_PCTYPE_L2_PAYLOAD))
 
+#define I40E_DEFAULT_RSS_HENA_EXPANDED (I40E_DEFAULT_RSS_HENA | \
+		BIT(I40E_FILTER_PCTYPE_NONF_IPV4_TCP_SYN_NO_ACK) | \
+		BIT(I40E_FILTER_PCTYPE_NONF_UNICAST_IPV4_UDP) | \
+		BIT(I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV4_UDP) | \
+		BIT(I40E_FILTER_PCTYPE_NONF_IPV6_TCP_SYN_NO_ACK) | \
+		BIT(I40E_FILTER_PCTYPE_NONF_UNICAST_IPV6_UDP) | \
+		BIT(I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV6_UDP))
+
+#define i40e_pf_get_default_rss_hena(pf) \
+	(((pf)->flags & I40E_FLAG_MULTIPLE_TCP_UDP_RSS_PCTYPE) ? \
+		I40E_DEFAULT_RSS_HENA_EXPANDED : I40E_DEFAULT_RSS_HENA)
+
 /* Supported Rx Buffer Sizes */
 #define I40E_RXBUFFER_512   512    /* Used for packet split */
 #define I40E_RXBUFFER_2048  2048

drivers/net/ethernet/intel/i40evf/i40e_type.h

@@ -967,15 +967,24 @@ struct i40e_filter_program_desc {
 
 /* Packet Classifier Types for filters */
 enum i40e_filter_pctype {
-	/* Note: Values 0-30 are reserved for future use */
+	/* Note: Values 0-28 are reserved for future use.
+	 * Value 29, 30, 32 are not supported on XL710 and X710.
+	 */
+	I40E_FILTER_PCTYPE_NONF_UNICAST_IPV4_UDP	= 29,
+	I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV4_UDP	= 30,
 	I40E_FILTER_PCTYPE_NONF_IPV4_UDP		= 31,
-	/* Note: Value 32 is reserved for future use */
+	I40E_FILTER_PCTYPE_NONF_IPV4_TCP_SYN_NO_ACK	= 32,
 	I40E_FILTER_PCTYPE_NONF_IPV4_TCP		= 33,
 	I40E_FILTER_PCTYPE_NONF_IPV4_SCTP		= 34,
 	I40E_FILTER_PCTYPE_NONF_IPV4_OTHER		= 35,
 	I40E_FILTER_PCTYPE_FRAG_IPV4			= 36,
-	/* Note: Values 37-40 are reserved for future use */
+	/* Note: Values 37-38 are reserved for future use.
+	 * Value 39, 40, 42 are not supported on XL710 and X710.
+	 */
+	I40E_FILTER_PCTYPE_NONF_UNICAST_IPV6_UDP	= 39,
+	I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV6_UDP	= 40,
 	I40E_FILTER_PCTYPE_NONF_IPV6_UDP		= 41,
+	I40E_FILTER_PCTYPE_NONF_IPV6_TCP_SYN_NO_ACK	= 42,
 	I40E_FILTER_PCTYPE_NONF_IPV6_TCP		= 43,
 	I40E_FILTER_PCTYPE_NONF_IPV6_SCTP		= 44,
 	I40E_FILTER_PCTYPE_NONF_IPV6_OTHER		= 45,

drivers/net/ethernet/intel/i40evf/i40evf.h

@@ -101,6 +101,8 @@ struct i40e_vsi {
 #define MAX_RX_QUEUES 8
 #define MAX_TX_QUEUES MAX_RX_QUEUES
 
+#define I40EVF_HKEY_ARRAY_SIZE ((I40E_VFQF_HKEY_MAX_INDEX + 1) * 4)
+
 /* MAX_MSIX_Q_VECTORS of these are allocated,
  * but we only use one per queue-specific vector.
  */