nfp: bpf: eliminate zero extension code-gen

This patch eliminate zero extension code-gen for instructions including
both alu and load/store. The only exception is for ctx load, because
offload target doesn't go through host ctx convert logic so we do
customized load and ignores zext flag set by verifier.

Cc: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Jiong Wang <jiong.wang@netronome.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

drivers/net/ethernet/netronome/nfp/bpf/main.h

@@ -238,6 +238,8 @@ struct nfp_bpf_reg_state {
 #define FLAG_INSN_SKIP_PREC_DEPENDENT		BIT(4)
 /* Instruction is optimized by the verifier */
 #define FLAG_INSN_SKIP_VERIFIER_OPT		BIT(5)
+/* Instruction needs to zero extend to high 32-bit */
+#define FLAG_INSN_DO_ZEXT			BIT(6)
 
 #define FLAG_INSN_SKIP_MASK		(FLAG_INSN_SKIP_NOOP | \
 					 FLAG_INSN_SKIP_PREC_DEPENDENT | \

drivers/net/ethernet/netronome/nfp/bpf/verifier.c

@@ -744,6 +744,17 @@ static unsigned int nfp_bpf_get_stack_usage(struct nfp_prog *nfp_prog)
 	goto continue_subprog;
 }
 
+static void nfp_bpf_insn_flag_zext(struct nfp_prog *nfp_prog,
+				   struct bpf_insn_aux_data *aux)
+{
+	struct nfp_insn_meta *meta;
+
+	list_for_each_entry(meta, &nfp_prog->insns, l) {
+		if (aux[meta->n].zext_dst)
+			meta->flags |= FLAG_INSN_DO_ZEXT;
+	}
+}
+
 int nfp_bpf_finalize(struct bpf_verifier_env *env)
 {
 	struct bpf_subprog_info *info;
@@ -784,6 +795,7 @@ int nfp_bpf_finalize(struct bpf_verifier_env *env)
 		return -EOPNOTSUPP;
 	}
 
+	nfp_bpf_insn_flag_zext(nfp_prog, env->insn_aux_data);
 	return 0;
 }