Merge branch 'bpf-nfp-shift-insns'

Jiong Wang says:

====================
NFP eBPF JIT is missing logic indirect shifts (both left and right) and
arithmetic right shift (both indirect shift and shift by constant).

This patch adds support for them.

For indirect shifts, shift amount is not specified as constant, NFP needs
to get the shift amount through the low 5 bits of source A operand in
PREV_ALU, therefore extra instructions are needed compared with shifts by
constants.

Because NFP is 32-bit, so we are using register pair for 64-bit shifts and
therefore would need different instruction sequences depending on whether
shift amount is less than 32 or not.

NFP branch-on-bit-test instruction emitter is added by this patch set and
is used for efficient runtime check on shift amount. We'd think the shift
amount is less than 32 if bit 5 is clear and greater or equal then 32
otherwise. Shift amount is greater than or equal to 64 will result in
undefined behavior.

This patch also use range info to avoid generating unnecessary runtime code
if we are certain shift amount is less than 32 or not.
====================
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

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

@@ -263,6 +263,8 @@ struct nfp_bpf_reg_state {
  * @func_id: function id for call instructions
  * @arg1: arg1 for call instructions
  * @arg2: arg2 for call instructions
+ * @umin: copy of core verifier umin_value.
+ * @umax: copy of core verifier umax_value.
  * @off: index of first generated machine instruction (in nfp_prog.prog)
  * @n: eBPF instruction number
  * @flags: eBPF instruction extra optimization flags
@@ -298,6 +300,13 @@ struct nfp_insn_meta {
 			struct bpf_reg_state arg1;
 			struct nfp_bpf_reg_state arg2;
 		};
+		/* We are interested in range info for some operands,
+		 * for example, the shift amount.
+		 */
+		struct {
+			u64 umin;
+			u64 umax;
+		};
 	};
 	unsigned int off;
 	unsigned short n;
@@ -375,6 +384,25 @@ static inline bool is_mbpf_xadd(const struct nfp_insn_meta *meta)
 	return (meta->insn.code & ~BPF_SIZE_MASK) == (BPF_STX | BPF_XADD);
 }
 
+static inline bool is_mbpf_indir_shift(const struct nfp_insn_meta *meta)
+{
+	u8 code = meta->insn.code;
+	bool is_alu, is_shift;
+	u8 opclass, opcode;
+
+	opclass = BPF_CLASS(code);
+	is_alu = opclass == BPF_ALU64 || opclass == BPF_ALU;
+	if (!is_alu)
+		return false;
+
+	opcode = BPF_OP(code);
+	is_shift = opcode == BPF_LSH || opcode == BPF_RSH || opcode == BPF_ARSH;
+	if (!is_shift)
+		return false;
+
+	return BPF_SRC(code) == BPF_X;
+}
+
 /**
  * struct nfp_prog - nfp BPF program
  * @bpf: backpointer to the bpf app priv structure

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

@@ -190,6 +190,8 @@ nfp_prog_prepare(struct nfp_prog *nfp_prog, const struct bpf_insn *prog,
 
 		meta->insn = prog[i];
 		meta->n = i;
+		if (is_mbpf_indir_shift(meta))
+			meta->umin = U64_MAX;
 
 		list_add_tail(&meta->l, &nfp_prog->insns);
 	}

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

@@ -551,6 +551,14 @@ nfp_verify_insn(struct bpf_verifier_env *env, int insn_idx, int prev_insn_idx)
 	if (is_mbpf_xadd(meta))
 		return nfp_bpf_check_xadd(nfp_prog, meta, env);
 
+	if (is_mbpf_indir_shift(meta)) {
+		const struct bpf_reg_state *sreg =
+			cur_regs(env) + meta->insn.src_reg;
+
+		meta->umin = min(meta->umin, sreg->umin_value);
+		meta->umax = max(meta->umax, sreg->umax_value);
+	}
+
 	return 0;
 }
 

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

@@ -72,8 +72,21 @@
 #define OP_BR_ADDR_LO		0x007ffc00000ULL
 #define OP_BR_ADDR_HI		0x10000000000ULL
 
-#define nfp_is_br(_insn)				\
-	(((_insn) & OP_BR_BASE_MASK) == OP_BR_BASE)
+#define OP_BR_BIT_BASE		0x0d000000000ULL
+#define OP_BR_BIT_BASE_MASK	0x0f800080300ULL
+#define OP_BR_BIT_A_SRC		0x000000000ffULL
+#define OP_BR_BIT_B_SRC		0x0000003fc00ULL
+#define OP_BR_BIT_BV		0x00000040000ULL
+#define OP_BR_BIT_SRC_LMEXTN	0x40000000000ULL
+#define OP_BR_BIT_DEFBR		OP_BR_DEFBR
+#define OP_BR_BIT_ADDR_LO	OP_BR_ADDR_LO
+#define OP_BR_BIT_ADDR_HI	OP_BR_ADDR_HI
+
+static inline bool nfp_is_br(u64 insn)
+{
+	return (insn & OP_BR_BASE_MASK) == OP_BR_BASE ||
+	       (insn & OP_BR_BIT_BASE_MASK) == OP_BR_BIT_BASE;
+}
 
 enum br_mask {
 	BR_BEQ = 0x00,
@@ -161,6 +174,7 @@ enum shf_op {
 	SHF_OP_NONE = 0,
 	SHF_OP_AND = 2,
 	SHF_OP_OR = 5,
+	SHF_OP_ASHR = 6,
 };
 
 enum shf_sc {