Merge branch 'nfp-XDP_TX-optimizations'

Jakub Kicinski says:

====================
nfp: optimize XDP TX and small fixes

This series optimizes the nfp XDP TX performance a little bit.
I run quick tests on an Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz.
Single core/queue performance for both touch and drop and touch and
forward is above 20Mpps @64B packets, drop being 2Mpps faster.
I think this is max for a single queue on the low power NFPs.

There are also a few minor fixes included for code in net-next.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/netronome/nfp/nfp_bpf_jit.c

@@ -798,7 +798,7 @@ wrp_test_reg(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta,
 	const struct bpf_insn *insn = &meta->insn;
 
 	if (insn->off < 0) /* TODO */
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	wrp_test_reg_one(nfp_prog, insn->dst_reg * 2, alu_op,
 			 insn->src_reg * 2, br_mask, insn->off);
@@ -818,7 +818,7 @@ wrp_cmp_imm(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta,
 	u32 tmp_reg;
 
 	if (insn->off < 0) /* TODO */
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	tmp_reg = ur_load_imm_any(nfp_prog, imm & ~0U, imm_b(nfp_prog));
 	if (!swap)
@@ -847,7 +847,7 @@ wrp_cmp_reg(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta,
 	u8 areg = insn->src_reg * 2, breg = insn->dst_reg * 2;
 
 	if (insn->off < 0) /* TODO */
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	if (swap) {
 		areg ^= breg;
@@ -1132,7 +1132,7 @@ static int mem_ldx4_skb(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 		emit_alu(nfp_prog, reg_both(meta->insn.dst_reg * 2),
 			 reg_none(), ALU_OP_NONE, NFP_BPF_ABI_LEN);
 	else
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	return 0;
 }
@@ -1143,7 +1143,7 @@ static int mem_ldx4_xdp(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 
 	if (meta->insn.off != offsetof(struct xdp_md, data) &&
 	    meta->insn.off != offsetof(struct xdp_md, data_end))
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	emit_alu(nfp_prog, dst, reg_none(), ALU_OP_NONE, NFP_BPF_ABI_PKT);
 
@@ -1174,12 +1174,12 @@ static int mem_stx4_skb(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 	if (meta->insn.off == offsetof(struct sk_buff, mark))
 		return wrp_set_mark(nfp_prog, meta->insn.src_reg * 2);
 
-	return -ENOTSUPP;
+	return -EOPNOTSUPP;
 }
 
 static int mem_stx4_xdp(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 {
-	return -ENOTSUPP;
+	return -EOPNOTSUPP;
 }
 
 static int mem_stx4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
@@ -1192,7 +1192,7 @@ static int mem_stx4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 static int jump(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 {
 	if (meta->insn.off < 0) /* TODO */
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 	emit_br(nfp_prog, BR_UNC, meta->insn.off, 0);
 
 	return 0;
@@ -1206,7 +1206,7 @@ static int jeq_imm(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 	u32 tmp_reg;
 
 	if (insn->off < 0) /* TODO */
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	if (imm & ~0U) {
 		tmp_reg = ur_load_imm_any(nfp_prog, imm & ~0U, imm_b(nfp_prog));
@@ -1245,7 +1245,7 @@ static int jset_imm(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 	u32 tmp_reg;
 
 	if (insn->off < 0) /* TODO */
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	if (!imm) {
 		meta->skip = true;
@@ -1276,7 +1276,7 @@ static int jne_imm(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 	u32 tmp_reg;
 
 	if (insn->off < 0) /* TODO */
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	if (!imm) {
 		emit_alu(nfp_prog, reg_none(), reg_a(insn->dst_reg * 2),
@@ -1302,7 +1302,7 @@ static int jeq_reg(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 	const struct bpf_insn *insn = &meta->insn;
 
 	if (insn->off < 0) /* TODO */
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	emit_alu(nfp_prog, imm_a(nfp_prog), reg_a(insn->dst_reg * 2),
 		 ALU_OP_XOR, reg_b(insn->src_reg * 2));

drivers/net/ethernet/netronome/nfp/nfp_net.h

@@ -201,6 +201,7 @@ struct nfp_net_tx_buf {
  * @txds:       Virtual address of TX ring in host memory
  * @dma:        DMA address of the TX ring
  * @size:       Size, in bytes, of the TX ring (needed to free)
+ * @is_xdp:	Is this a XDP TX ring?
  */
 struct nfp_net_tx_ring {
 	struct nfp_net_r_vector *r_vec;
@@ -221,6 +222,7 @@ struct nfp_net_tx_ring {
 
 	dma_addr_t dma;
 	unsigned int size;
+	bool is_xdp;
 } ____cacheline_aligned;
 
 /* RX and freelist descriptor format */
@@ -468,10 +470,10 @@ struct nfp_net_dp {
 	u8 chained_metadata_format:1;
 
 	u8 rx_dma_dir;
-	u8 rx_dma_off;
-
 	u8 rx_offset;
 
+	u32 rx_dma_off;
+
 	u32 ctrl;
 	u32 fl_bufsz;
 

drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c

@@ -791,7 +791,7 @@ static int nfp_net_set_coalesce(struct net_device *netdev,
 	    ec->tx_coalesce_usecs_high ||
 	    ec->tx_max_coalesced_frames_high ||
 	    ec->rate_sample_interval)
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	/* Compute factor used to convert coalesce '_usecs' parameters to
 	 * ME timestamp ticks.  There are 16 ME clock cycles for each timestamp

drivers/net/ethernet/netronome/nfp/nfp_net_offload.c

@@ -119,12 +119,12 @@ nfp_net_bpf_get_act(struct nfp_net *nn, struct tc_cls_bpf_offload *cls_bpf)
 		if (tc_no_actions(cls_bpf->exts))
 			return NN_ACT_DIRECT;
 
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 	}
 
 	/* TC legacy mode */
 	if (!tc_single_action(cls_bpf->exts))
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	tcf_exts_to_list(cls_bpf->exts, &actions);
 	list_for_each_entry(a, &actions, list) {
@@ -136,7 +136,7 @@ nfp_net_bpf_get_act(struct nfp_net *nn, struct tc_cls_bpf_offload *cls_bpf)
 			return NN_ACT_TC_REDIR;
 	}
 
-	return -ENOTSUPP;
+	return -EOPNOTSUPP;
 }
 
 static int
@@ -152,7 +152,7 @@ nfp_net_bpf_offload_prepare(struct nfp_net *nn,
 	int ret;
 
 	if (!IS_ENABLED(CONFIG_BPF_SYSCALL))
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 
 	ret = nfp_net_bpf_get_act(nn, cls_bpf);
 	if (ret < 0)
@@ -162,7 +162,7 @@ nfp_net_bpf_offload_prepare(struct nfp_net *nn,
 	max_mtu = nn_readb(nn, NFP_NET_CFG_BPF_INL_MTU) * 64 - 32;
 	if (max_mtu < nn->dp.netdev->mtu) {
 		nn_info(nn, "BPF offload not supported with MTU larger than HW packet split boundary\n");
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 	}
 
 	start_off = nn_readw(nn, NFP_NET_CFG_BPF_START);
@@ -289,6 +289,6 @@ int nfp_net_bpf_offload(struct nfp_net *nn, struct tc_cls_bpf_offload *cls_bpf)
 		return nfp_net_bpf_stats_update(nn, cls_bpf);
 
 	default:
-		return -ENOTSUPP;
+		return -EOPNOTSUPP;
 	}
 }

drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.c

@@ -78,7 +78,7 @@
 
 #define NSP_MAGIC		0xab10
 #define NSP_MAJOR		0
-#define NSP_MINOR		(__MAX_SPCODE - 1)
+#define NSP_MINOR		8
 
 #define NSP_CODE_MAJOR		GENMASK(15, 12)
 #define NSP_CODE_MINOR		GENMASK(11, 0)
@@ -94,8 +94,6 @@ enum nfp_nsp_cmd {
 	SPCODE_ETH_RESCAN	= 7, /* Rescan ETHs, write ETH_TABLE to buf */
 	SPCODE_ETH_CONTROL	= 8, /* Update media config from buffer */
 	SPCODE_NSP_IDENTIFY	= 13, /* Read NSP version */
-
-	__MAX_SPCODE,
 };
 
 static const struct {