Initial Commit

thirdparty/capstone/arch/BPF/BPFConstants.h

@@ -0,0 +1,88 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+/* This file defines constants and macros used for parsing a BPF instruction */
+
+#ifndef CS_BPF_CONSTANTS_H
+#define CS_BPF_CONSTANTS_H
+
+#define BPF_CLASS(code) ((code) & 0x7)
+
+///< Instruction classes
+#define BPF_CLASS_LD	0x00
+#define BPF_CLASS_LDX	0x01
+#define BPF_CLASS_ST	0x02
+#define BPF_CLASS_STX	0x03
+#define BPF_CLASS_ALU	0x04
+#define BPF_CLASS_JMP	0x05
+#define BPF_CLASS_RET	0x06	///< cBPF only
+#define BPF_CLASS_MISC	0x07	///< cBPF only
+#define BPF_CLASS_ALU64	0x07	///< eBPF only
+
+#define BPF_OP(code) ((code) & 0xf0)
+
+///< Types of ALU instruction
+#define BPF_ALU_ADD	0x00
+#define BPF_ALU_SUB	0x10
+#define BPF_ALU_MUL	0x20
+#define BPF_ALU_DIV	0x30
+#define BPF_ALU_OR	0x40
+#define BPF_ALU_AND	0x50
+#define BPF_ALU_LSH	0x60
+#define BPF_ALU_RSH	0x70
+#define BPF_ALU_NEG	0x80
+#define BPF_ALU_MOD	0x90
+#define BPF_ALU_XOR	0xa0
+#define BPF_ALU_MOV	0xb0  ///< eBPF only: mov reg to reg
+#define BPF_ALU_ARSH	0xc0  ///< eBPF only: sign extending shift right
+#define BPF_ALU_END	0xd0  ///< eBPF only: endianness conversion
+
+///< Types of jmp instruction
+#define BPF_JUMP_JA	0x00	///< goto
+#define BPF_JUMP_JEQ	0x10	///< '=='
+#define BPF_JUMP_JGT	0x20	///< unsigned '>'
+#define BPF_JUMP_JGE	0x30	///< unsigned '>='
+#define BPF_JUMP_JSET	0x40	///< '&'
+#define BPF_JUMP_JNE	0x50	///< eBPF only: '!=' */
+#define BPF_JUMP_JSGT	0x60	///< eBPF only: signed '>'
+#define BPF_JUMP_JSGE	0x70	///< eBPF only: signed '>='
+#define BPF_JUMP_CALL	0x80	///< eBPF only: function call
+#define BPF_JUMP_EXIT	0x90	///< eBPF only: exit
+#define BPF_JUMP_JLT	0xa0	///< eBPF only: unsigned '<'
+#define BPF_JUMP_JLE	0xb0	///< eBPF only: unsigned '<='
+#define BPF_JUMP_JSLT	0xc0	///< eBPF only: signed '<'
+#define BPF_JUMP_JSLE	0xd0	///< eBPF only: signed '<='
+
+#define BPF_SRC(code) ((code) & 0x08)
+#define BPF_RVAL(code) ((code) & 0x18) /* cBPF only: for return types */
+///< Source operand
+#define BPF_SRC_K	0x00
+#define BPF_SRC_X	0x08
+#define BPF_SRC_A	0x10 /* cBPF only */
+
+#define BPF_SRC_LITTLE	BPF_SRC_K
+#define BPF_SRC_BIG	BPF_SRC_X
+
+#define BPF_SIZE(code) ((code) & 0x18)
+///< Size modifier
+#define BPF_SIZE_W   0x00	///< word
+#define BPF_SIZE_H   0x08	///< half word
+#define BPF_SIZE_B   0x10	///< byte
+#define BPF_SIZE_DW  0x18	///< eBPF only: double word
+
+#define BPF_MODE(code) ((code) & 0xe0)
+///< Mode modifier
+#define BPF_MODE_IMM	0x00	///< used for 32-bit mov in cBPF and 64-bit in eBPF
+#define BPF_MODE_ABS	0x20
+#define BPF_MODE_IND	0x40
+#define BPF_MODE_MEM	0x60
+#define BPF_MODE_LEN	0x80	///< cBPF only, reserved in eBPF
+#define BPF_MODE_MSH	0xa0	///< cBPF only, reserved in eBPF
+#define BPF_MODE_XADD	0xc0	///< eBPF only: exclusive add
+
+#define BPF_MISCOP(code) ((code) & 0x80)
+///< Operation of misc
+#define BPF_MISCOP_TAX	0x00
+#define BPF_MISCOP_TXA	0x80
+
+#endif

thirdparty/capstone/arch/BPF/BPFDisassembler.c

@@ -0,0 +1,463 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+#ifdef CAPSTONE_HAS_BPF
+
+#include <string.h>
+#include <stddef.h> // offsetof macro
+
+#include "BPFConstants.h"
+#include "BPFDisassembler.h"
+#include "BPFMapping.h"
+#include "../../cs_priv.h"
+
+static uint16_t read_u16(cs_struct *ud, const uint8_t *code)
+{
+	if (MODE_IS_BIG_ENDIAN(ud->mode))
+		return (((uint16_t)code[0] << 8) | code[1]);
+	else
+		return (((uint16_t)code[1] << 8) | code[0]);
+}
+
+static uint32_t read_u32(cs_struct *ud, const uint8_t *code)
+{
+	if (MODE_IS_BIG_ENDIAN(ud->mode))
+		return ((uint32_t)read_u16(ud, code) << 16) | read_u16(ud, code + 2);
+	else
+		return ((uint32_t)read_u16(ud, code + 2) << 16) | read_u16(ud, code);
+}
+
+///< Malloc bpf_internal, also checks if code_len is large enough.
+static bpf_internal *alloc_bpf_internal(size_t code_len)
+{
+	bpf_internal *bpf;
+
+	if (code_len < 8)
+		return NULL;
+	bpf = cs_mem_malloc(sizeof(bpf_internal));
+	if (bpf == NULL)
+		return NULL;
+	/* default value */
+	bpf->insn_size = 8;
+	return bpf;
+}
+
+///< Fetch a cBPF structure from code
+static bpf_internal* fetch_cbpf(cs_struct *ud, const uint8_t *code,
+		size_t code_len)
+{
+	bpf_internal *bpf;
+
+	bpf = alloc_bpf_internal(code_len);
+	if (bpf == NULL)
+		return NULL;
+
+	bpf->op = read_u16(ud, code);
+	bpf->jt = code[2];
+	bpf->jf = code[3];
+	bpf->k = read_u32(ud, code + 4);
+	return bpf;
+}
+
+///< Fetch an eBPF structure from code
+static bpf_internal* fetch_ebpf(cs_struct *ud, const uint8_t *code,
+		size_t code_len)
+{
+	bpf_internal *bpf;
+
+	bpf = alloc_bpf_internal(code_len);
+	if (bpf == NULL)
+		return NULL;
+
+	bpf->op = (uint16_t)code[0];
+	bpf->dst = code[1] & 0xf;
+	bpf->src = (code[1] & 0xf0) >> 4;
+
+	// eBPF has one 16-byte instruction: BPF_LD | BPF_DW | BPF_IMM,
+	// in this case imm is combined with the next block's imm.
+	if (bpf->op == (BPF_CLASS_LD | BPF_SIZE_DW | BPF_MODE_IMM)) {
+		if (code_len < 16) {
+			cs_mem_free(bpf);
+			return NULL;
+		}
+		bpf->k = read_u32(ud, code + 4) | (((uint64_t)read_u32(ud, code + 12)) << 32);
+		bpf->insn_size = 16;
+	}
+	else {
+		bpf->offset = read_u16(ud, code + 2);
+		bpf->k = read_u32(ud, code + 4);
+	}
+	return bpf;
+}
+
+#define CHECK_READABLE_REG(ud, reg) do { \
+		if (! ((reg) >= BPF_REG_R0 && (reg) <= BPF_REG_R10)) \
+			return false; \
+	} while (0)
+
+#define CHECK_WRITABLE_REG(ud, reg) do { \
+		if (! ((reg) >= BPF_REG_R0 && (reg) < BPF_REG_R10)) \
+			return false; \
+	} while (0)
+
+#define CHECK_READABLE_AND_PUSH(ud, MI, r) do { \
+		CHECK_READABLE_REG(ud, r + BPF_REG_R0); \
+		MCOperand_CreateReg0(MI, r + BPF_REG_R0); \
+	} while (0)
+
+#define CHECK_WRITABLE_AND_PUSH(ud, MI, r) do { \
+		CHECK_WRITABLE_REG(ud, r + BPF_REG_R0); \
+		MCOperand_CreateReg0(MI, r + BPF_REG_R0); \
+	} while (0)
+
+static bool decodeLoad(cs_struct *ud, MCInst *MI, bpf_internal *bpf)
+{
+	if (!EBPF_MODE(ud)) {
+		/*
+		 *  +-----+-----------+--------------------+
+		 *  | ldb |    [k]    |       [x+k]        |
+		 *  | ldh |    [k]    |       [x+k]        |
+		 *  +-----+-----------+--------------------+
+		 */
+		if (BPF_SIZE(bpf->op) == BPF_SIZE_DW)
+			return false;
+		if (BPF_SIZE(bpf->op) == BPF_SIZE_B || BPF_SIZE(bpf->op) == BPF_SIZE_H) {
+			/* no ldx */
+			if (BPF_CLASS(bpf->op) != BPF_CLASS_LD)
+				return false;
+			/* can only be BPF_ABS and BPF_IND */
+			if (BPF_MODE(bpf->op) == BPF_MODE_ABS) {
+				MCOperand_CreateImm0(MI, bpf->k);
+				return true;
+			}
+			else if (BPF_MODE(bpf->op) == BPF_MODE_IND) {
+				MCOperand_CreateReg0(MI, BPF_REG_X);
+				MCOperand_CreateImm0(MI, bpf->k);
+				return true;
+			}
+			return false;
+		}
+		/*
+		 *  +-----+----+------+------+-----+-------+
+		 *  | ld  | #k | #len | M[k] | [k] | [x+k] |
+		 *  +-----+----+------+------+-----+-------+
+		 *  | ldx | #k | #len | M[k] | 4*([k]&0xf) |
+		 *  +-----+----+------+------+-------------+
+		 */
+		switch (BPF_MODE(bpf->op)) {
+		default:
+			break;
+		case BPF_MODE_IMM:
+			MCOperand_CreateImm0(MI, bpf->k);
+			return true;
+		case BPF_MODE_LEN:
+			return true;
+		case BPF_MODE_MEM:
+			MCOperand_CreateImm0(MI, bpf->k);
+			return true;
+		}
+		if (BPF_CLASS(bpf->op) == BPF_CLASS_LD) {
+			if (BPF_MODE(bpf->op) == BPF_MODE_ABS) {
+				MCOperand_CreateImm0(MI, bpf->k);
+				return true;
+			}
+			else if (BPF_MODE(bpf->op) == BPF_MODE_IND) {
+				MCOperand_CreateReg0(MI, BPF_REG_X);
+				MCOperand_CreateImm0(MI, bpf->k);
+				return true;
+			}
+		}
+		else { /* LDX */
+			if (BPF_MODE(bpf->op) == BPF_MODE_MSH) {
+				MCOperand_CreateImm0(MI, bpf->k);
+				return true;
+			}
+		}
+		return false;
+	}
+
+	/* eBPF mode */
+	/*
+	 * - IMM: lddw dst, imm64
+	 * - ABS: ld{w,h,b,dw} [k]
+	 * - IND: ld{w,h,b,dw} [src+k]
+	 * - MEM: ldx{w,h,b,dw} dst, [src+off]
+	 */
+	if (BPF_CLASS(bpf->op) == BPF_CLASS_LD) {
+		switch (BPF_MODE(bpf->op)) {
+		case BPF_MODE_IMM:
+			if (bpf->op != (BPF_CLASS_LD | BPF_SIZE_DW | BPF_MODE_IMM))
+				return false;
+			CHECK_WRITABLE_AND_PUSH(ud, MI, bpf->dst);
+			MCOperand_CreateImm0(MI, bpf->k);
+			return true;
+		case BPF_MODE_ABS:
+			MCOperand_CreateImm0(MI, bpf->k);
+			return true;
+		case BPF_MODE_IND:
+			CHECK_READABLE_AND_PUSH(ud, MI, bpf->src);
+			MCOperand_CreateImm0(MI, bpf->k);
+			return true;
+		}
+		return false;
+
+	}
+	/* LDX */
+	if (BPF_MODE(bpf->op) == BPF_MODE_MEM) {
+		CHECK_WRITABLE_AND_PUSH(ud, MI, bpf->dst);
+		CHECK_READABLE_AND_PUSH(ud, MI, bpf->src);
+		MCOperand_CreateImm0(MI, bpf->offset);
+		return true;
+	}
+	return false;
+}
+
+static bool decodeStore(cs_struct *ud, MCInst *MI, bpf_internal *bpf)
+{
+	/* in cBPF, only BPF_ST* | BPF_MEM | BPF_W is valid
+	 * while in eBPF:
+	 * - BPF_STX | BPF_XADD | BPF_{W,DW}
+	 * - BPF_ST* | BPF_MEM | BPF_{W,H,B,DW}
+	 * are valid
+	 */
+	if (!EBPF_MODE(ud)) {
+		/* can only store to M[] */
+		if (bpf->op != (BPF_CLASS(bpf->op) | BPF_MODE_MEM | BPF_SIZE_W))
+			return false;
+		MCOperand_CreateImm0(MI, bpf->k);
+		return true;
+	}
+
+	/* eBPF */
+
+	if (BPF_MODE(bpf->op) == BPF_MODE_XADD) {
+		if (BPF_CLASS(bpf->op) != BPF_CLASS_STX)
+			return false;
+		if (BPF_SIZE(bpf->op) != BPF_SIZE_W && BPF_SIZE(bpf->op) != BPF_SIZE_DW)
+			return false;
+		/* xadd [dst + off], src */
+		CHECK_READABLE_AND_PUSH(ud, MI, bpf->dst);
+		MCOperand_CreateImm0(MI, bpf->offset);
+		CHECK_READABLE_AND_PUSH(ud, MI, bpf->src);
+		return true;
+	}
+
+	if (BPF_MODE(bpf->op) != BPF_MODE_MEM)
+		return false;
+
+	/* st [dst + off], src */
+	CHECK_READABLE_AND_PUSH(ud, MI, bpf->dst);
+	MCOperand_CreateImm0(MI, bpf->offset);
+	if (BPF_CLASS(bpf->op) == BPF_CLASS_ST)
+		MCOperand_CreateImm0(MI, bpf->k);
+	else
+		CHECK_READABLE_AND_PUSH(ud, MI, bpf->src);
+	return true;
+}
+
+static bool decodeALU(cs_struct *ud, MCInst *MI, bpf_internal *bpf)
+{
+	/* Set MI->Operands */
+
+	/* cBPF */
+	if (!EBPF_MODE(ud)) {
+		if (BPF_OP(bpf->op) > BPF_ALU_XOR)
+			return false;
+		/* cBPF's NEG has no operands */
+		if (BPF_OP(bpf->op) == BPF_ALU_NEG)
+			return true;
+		if (BPF_SRC(bpf->op) == BPF_SRC_K)
+			MCOperand_CreateImm0(MI, bpf->k);
+		else /* BPF_SRC_X */
+			MCOperand_CreateReg0(MI, BPF_REG_X);
+		return true;
+	}
+
+	/* eBPF */
+
+	if (BPF_OP(bpf->op) > BPF_ALU_END)
+		return false;
+	/* ENDian's imm must be one of 16, 32, 64 */
+	if (BPF_OP(bpf->op) == BPF_ALU_END) {
+		if (bpf->k != 16 && bpf->k != 32 && bpf->k != 64)
+			return false;
+		if (BPF_CLASS(bpf->op) == BPF_CLASS_ALU64 && BPF_SRC(bpf->op) != BPF_SRC_LITTLE)
+			return false;
+	}
+
+	/* - op dst, imm
+	 * - op dst, src
+	 * - neg dst
+	 * - le<imm> dst
+	 */
+	/* every ALU instructions have dst op */
+	CHECK_WRITABLE_AND_PUSH(ud, MI, bpf->dst);
+
+	/* special cases */
+	if (BPF_OP(bpf->op) == BPF_ALU_NEG)
+		return true;
+	if (BPF_OP(bpf->op) == BPF_ALU_END) {
+		/* bpf->k must be one of 16, 32, 64 */
+		MCInst_setOpcode(MI, MCInst_getOpcode(MI) | ((uint32_t)bpf->k << 4));
+		return true;
+	}
+
+	/* normal cases */
+	if (BPF_SRC(bpf->op) == BPF_SRC_K) {
+		MCOperand_CreateImm0(MI, bpf->k);
+	}
+	else { /* BPF_SRC_X */
+		CHECK_READABLE_AND_PUSH(ud, MI, bpf->src);
+	}
+	return true;
+}
+
+static bool decodeJump(cs_struct *ud, MCInst *MI, bpf_internal *bpf)
+{
+	/* cBPF and eBPF are very different in class jump */
+	if (!EBPF_MODE(ud)) {
+		if (BPF_OP(bpf->op) > BPF_JUMP_JSET)
+			return false;
+
+		/* ja is a special case of jumps */
+		if (BPF_OP(bpf->op) == BPF_JUMP_JA) {
+			MCOperand_CreateImm0(MI, bpf->k);
+			return true;
+		}
+
+		if (BPF_SRC(bpf->op) == BPF_SRC_K)
+			MCOperand_CreateImm0(MI, bpf->k);
+		else /* BPF_SRC_X */
+			MCOperand_CreateReg0(MI, BPF_REG_X);
+		MCOperand_CreateImm0(MI, bpf->jt);
+		MCOperand_CreateImm0(MI, bpf->jf);
+	}
+	else {
+		if (BPF_OP(bpf->op) > BPF_JUMP_JSLE)
+			return false;
+
+		/* No operands for exit */
+		if (BPF_OP(bpf->op) == BPF_JUMP_EXIT)
+			return bpf->op == (BPF_CLASS_JMP | BPF_JUMP_EXIT);
+		if (BPF_OP(bpf->op) == BPF_JUMP_CALL) {
+			if (bpf->op == (BPF_CLASS_JMP | BPF_JUMP_CALL)) {
+				MCOperand_CreateImm0(MI, bpf->k);
+				return true;
+			}
+			if (bpf->op == (BPF_CLASS_JMP | BPF_JUMP_CALL | BPF_SRC_X)) {
+				CHECK_READABLE_AND_PUSH(ud, MI, bpf->k);
+				return true;
+			}
+			return false;
+		}
+
+		/* ja is a special case of jumps */
+		if (BPF_OP(bpf->op) == BPF_JUMP_JA) {
+			if (BPF_SRC(bpf->op) != BPF_SRC_K)
+				return false;
+			MCOperand_CreateImm0(MI, bpf->offset);
+			return true;
+		}
+
+		/* <j>  dst, src, +off */
+		CHECK_READABLE_AND_PUSH(ud, MI, bpf->dst);
+		if (BPF_SRC(bpf->op) == BPF_SRC_K)
+			MCOperand_CreateImm0(MI, bpf->k);
+		else
+			CHECK_READABLE_AND_PUSH(ud, MI, bpf->src);
+		MCOperand_CreateImm0(MI, bpf->offset);
+	}
+	return true;
+}
+
+static bool decodeReturn(cs_struct *ud, MCInst *MI, bpf_internal *bpf)
+{
+	/* Here only handles the BPF_RET class in cBPF */
+	switch (BPF_RVAL(bpf->op)) {
+	case BPF_SRC_K:
+		MCOperand_CreateImm0(MI, bpf->k);
+		return true;
+	case BPF_SRC_X:
+		MCOperand_CreateReg0(MI, BPF_REG_X);
+		return true;
+	case BPF_SRC_A:
+		MCOperand_CreateReg0(MI, BPF_REG_A);
+		return true;
+	}
+	return false;
+}
+
+static bool decodeMISC(cs_struct *ud, MCInst *MI, bpf_internal *bpf)
+{
+	uint16_t op = bpf->op ^ BPF_CLASS_MISC;
+	return op == BPF_MISCOP_TAX || op == BPF_MISCOP_TXA;
+}
+
+///< 1. Check if the instruction is valid
+///< 2. Set MI->opcode
+///< 3. Set MI->Operands
+static bool getInstruction(cs_struct *ud, MCInst *MI, bpf_internal *bpf)
+{
+	cs_detail *detail;
+
+	detail = MI->flat_insn->detail;
+	// initialize detail
+	if (detail) {
+		memset(detail, 0, offsetof(cs_detail, bpf) + sizeof(cs_bpf));
+	}
+
+	MCInst_clear(MI);
+	MCInst_setOpcode(MI, bpf->op);
+
+	switch (BPF_CLASS(bpf->op)) {
+	default: /* should never happen */
+		return false;
+	case BPF_CLASS_LD:
+	case BPF_CLASS_LDX:
+		return decodeLoad(ud, MI, bpf);
+	case BPF_CLASS_ST:
+	case BPF_CLASS_STX:
+		return decodeStore(ud, MI, bpf);
+	case BPF_CLASS_ALU:
+		return decodeALU(ud, MI, bpf);
+	case BPF_CLASS_JMP:
+		return decodeJump(ud, MI, bpf);
+	case BPF_CLASS_RET:
+		/* eBPF doesn't have this class */
+		if (EBPF_MODE(ud))
+			return false;
+		return decodeReturn(ud, MI, bpf);
+	case BPF_CLASS_MISC:
+	/* case BPF_CLASS_ALU64: */
+		if (EBPF_MODE(ud))
+			return decodeALU(ud, MI, bpf);
+		else
+			return decodeMISC(ud, MI, bpf);
+	}
+}
+
+bool BPF_getInstruction(csh ud, const uint8_t *code, size_t code_len,
+		MCInst *instr, uint16_t *size, uint64_t address, void *info)
+{
+	cs_struct *cs;
+	bpf_internal *bpf;
+
+	cs = (cs_struct*)ud;
+	if (EBPF_MODE(cs))
+		bpf = fetch_ebpf(cs, code, code_len);
+	else
+		bpf = fetch_cbpf(cs, code, code_len);
+	if (bpf == NULL)
+		return false;
+	if (!getInstruction(cs, instr, bpf)) {
+		cs_mem_free(bpf);
+		return false;
+	}
+
+	*size = bpf->insn_size;
+	cs_mem_free(bpf);
+
+	return true;
+}
+
+#endif

thirdparty/capstone/arch/BPF/BPFDisassembler.h

@@ -0,0 +1,27 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+#ifndef CS_BPF_DISASSEMBLER_H
+#define CS_BPF_DISASSEMBLER_H
+
+#include "../../MCInst.h"
+
+typedef struct bpf_internal {
+	uint16_t op;
+	uint64_t k;
+	/* for cBPF */
+	uint8_t jt;
+	uint8_t jf;
+	/* for eBPF */
+	uint8_t dst;
+	uint8_t src;
+	uint16_t offset;
+
+	/* length of this bpf instruction */
+	uint8_t insn_size;
+} bpf_internal;
+
+bool BPF_getInstruction(csh ud, const uint8_t *code, size_t code_len,
+		MCInst *instr, uint16_t *size, uint64_t address, void *info);
+
+#endif

thirdparty/capstone/arch/BPF/BPFInstPrinter.c

@@ -0,0 +1,285 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+#include <capstone/platform.h>
+
+#include "BPFConstants.h"
+#include "BPFInstPrinter.h"
+#include "BPFMapping.h"
+
+static cs_bpf_op *expand_bpf_operands(cs_bpf *bpf)
+{
+	/* assert(bpf->op_count < 3); */
+	return &bpf->operands[bpf->op_count++];
+}
+
+static void push_op_reg(cs_bpf *bpf, bpf_op_type val, uint8_t ac_mode)
+{
+	cs_bpf_op *op = expand_bpf_operands(bpf);
+
+	op->type = BPF_OP_REG;
+	op->reg = val;
+	op->access = ac_mode;
+}
+
+static void push_op_imm(cs_bpf *bpf, uint64_t val)
+{
+	cs_bpf_op *op = expand_bpf_operands(bpf);
+
+	op->type = BPF_OP_IMM;
+	op->imm = val;
+}
+
+static void push_op_off(cs_bpf *bpf, uint32_t val)
+{
+	cs_bpf_op *op = expand_bpf_operands(bpf);
+
+	op->type = BPF_OP_OFF;
+	op->off = val;
+}
+
+static void push_op_mem(cs_bpf *bpf, bpf_reg reg, uint32_t val)
+{
+	cs_bpf_op *op = expand_bpf_operands(bpf);
+
+	op->type = BPF_OP_MEM;
+	op->mem.base = reg;
+	op->mem.disp = val;
+}
+
+static void push_op_mmem(cs_bpf *bpf, uint32_t val)
+{
+	cs_bpf_op *op = expand_bpf_operands(bpf);
+
+	op->type = BPF_OP_MMEM;
+	op->mmem = val;
+}
+
+static void push_op_msh(cs_bpf *bpf, uint32_t val)
+{
+	cs_bpf_op *op = expand_bpf_operands(bpf);
+
+	op->type = BPF_OP_MSH;
+	op->msh = val;
+}
+
+static void push_op_ext(cs_bpf *bpf, bpf_ext_type val)
+{
+	cs_bpf_op *op = expand_bpf_operands(bpf);
+
+	op->type = BPF_OP_EXT;
+	op->ext = val;
+}
+
+static void convert_operands(MCInst *MI, cs_bpf *bpf)
+{
+	unsigned opcode = MCInst_getOpcode(MI);
+	unsigned mc_op_count = MCInst_getNumOperands(MI);
+	MCOperand *op;
+	MCOperand *op2;
+	unsigned i;
+
+	bpf->op_count = 0;
+	if (BPF_CLASS(opcode) == BPF_CLASS_LD || BPF_CLASS(opcode) == BPF_CLASS_LDX) {
+		switch (BPF_MODE(opcode)) {
+		case BPF_MODE_IMM:
+			if (EBPF_MODE(MI->csh)) {
+				push_op_reg(bpf, MCOperand_getReg(MCInst_getOperand(MI, 0)), CS_AC_WRITE);
+				push_op_imm(bpf, MCOperand_getImm(MCInst_getOperand(MI, 1)));
+			} else {
+				push_op_imm(bpf, MCOperand_getImm(MCInst_getOperand(MI, 0)));
+			}
+			break;
+		case BPF_MODE_ABS:
+			op = MCInst_getOperand(MI, 0);
+			push_op_mem(bpf, BPF_REG_INVALID, (uint32_t)MCOperand_getImm(op));
+			break;
+		case BPF_MODE_IND:
+			op = MCInst_getOperand(MI, 0);
+			op2 = MCInst_getOperand(MI, 1);
+			push_op_mem(bpf, MCOperand_getReg(op), (uint32_t)MCOperand_getImm(op2));
+			break;
+		case BPF_MODE_MEM:
+			if (EBPF_MODE(MI->csh)) {
+				/* ldx{w,h,b,dw} dst, [src+off] */
+				push_op_reg(bpf, MCOperand_getReg(MCInst_getOperand(MI, 0)), CS_AC_WRITE);
+				op = MCInst_getOperand(MI, 1);
+				op2 = MCInst_getOperand(MI, 2);
+				push_op_mem(bpf, MCOperand_getReg(op), (uint32_t)MCOperand_getImm(op2));
+			}
+			else {
+				push_op_mmem(bpf, (uint32_t)MCOperand_getImm(MCInst_getOperand(MI, 0)));
+			}
+			break;
+		case BPF_MODE_LEN:
+			push_op_ext(bpf, BPF_EXT_LEN);
+			break;
+		case BPF_MODE_MSH:
+			op = MCInst_getOperand(MI, 0);
+			push_op_msh(bpf, (uint32_t)MCOperand_getImm(op));
+			break;
+		/* case BPF_MODE_XADD: // not exists */
+		}
+		return;
+	}
+	if (BPF_CLASS(opcode) == BPF_CLASS_ST || BPF_CLASS(opcode) == BPF_CLASS_STX) {
+		if (!EBPF_MODE(MI->csh)) {
+			// cBPF has only one case - st* M[k]
+			push_op_mmem(bpf, (uint32_t)MCOperand_getImm(MCInst_getOperand(MI, 0)));
+			return;
+		}
+		/* eBPF has two cases:
+		 * - st [dst + off], src
+		 * - xadd [dst + off], src
+		 * they have same form of operands.
+		 */
+		op = MCInst_getOperand(MI, 0);
+		op2 = MCInst_getOperand(MI, 1);
+		push_op_mem(bpf, MCOperand_getReg(op), (uint32_t)MCOperand_getImm(op2));
+		op = MCInst_getOperand(MI, 2);
+		if (MCOperand_isImm(op))
+			push_op_imm(bpf, MCOperand_getImm(op));
+		else if (MCOperand_isReg(op))
+			push_op_reg(bpf, MCOperand_getReg(op), CS_AC_READ);
+		return;
+	}
+
+	if (BPF_CLASS(opcode) == BPF_CLASS_JMP) {
+		for (i = 0; i < mc_op_count; i++) {
+			op = MCInst_getOperand(MI, i);
+			if (MCOperand_isImm(op)) {
+				/* decide the imm is BPF_OP_IMM or BPF_OP_OFF type here */
+				/*
+				 * 1. ja +off
+				 * 2. j {x,k}, +jt, +jf // cBPF
+				 * 3. j dst_reg, {src_reg, k}, +off // eBPF
+				 */
+				if (BPF_OP(opcode) == BPF_JUMP_JA ||
+						(!EBPF_MODE(MI->csh) && i >= 1) ||
+						(EBPF_MODE(MI->csh) && i == 2))
+					push_op_off(bpf, (uint32_t)MCOperand_getImm(op));
+				else
+					push_op_imm(bpf, MCOperand_getImm(op));
+			}
+			else if (MCOperand_isReg(op)) {
+				push_op_reg(bpf, MCOperand_getReg(op), CS_AC_READ);
+			}
+		}
+		return;
+	}
+
+	if (!EBPF_MODE(MI->csh)) {
+		/* In cBPF mode, all registers in operands are accessed as read */
+		for (i = 0; i < mc_op_count; i++) {
+			op = MCInst_getOperand(MI, i);
+			if (MCOperand_isImm(op))
+				push_op_imm(bpf, MCOperand_getImm(op));
+			else if (MCOperand_isReg(op))
+				push_op_reg(bpf, MCOperand_getReg(op), CS_AC_READ);
+		}
+		return;
+	}
+
+	/* remain cases are: eBPF mode && ALU */
+	/* if (BPF_CLASS(opcode) == BPF_CLASS_ALU || BPF_CLASS(opcode) == BPF_CLASS_ALU64) */
+
+	/* We have three types:
+	 * 1. {l,b}e dst               // dst = byteswap(dst)
+	 * 2. neg dst                  // dst = -dst
+	 * 3. <op> dst, {src_reg, imm} // dst = dst <op> src
+	 * so we can simply check the number of operands,
+	 * exactly one operand means we are in case 1. and 2.,
+	 * otherwise in case 3.
+	 */
+	if (mc_op_count == 1) {
+		op = MCInst_getOperand(MI, 0);
+		push_op_reg(bpf, MCOperand_getReg(op), CS_AC_READ | CS_AC_WRITE);
+	}
+	else { // if (mc_op_count == 2)
+		op = MCInst_getOperand(MI, 0);
+		push_op_reg(bpf, MCOperand_getReg(op), CS_AC_READ | CS_AC_WRITE);
+
+		op = MCInst_getOperand(MI, 1);
+		if (MCOperand_isImm(op))
+			push_op_imm(bpf, MCOperand_getImm(op));
+		else if (MCOperand_isReg(op))
+			push_op_reg(bpf, MCOperand_getReg(op), CS_AC_READ);
+	}
+}
+
+static void print_operand(MCInst *MI, struct SStream *O, const cs_bpf_op *op)
+{
+	switch (op->type) {
+	case BPF_OP_INVALID:
+		SStream_concat(O, "invalid");
+		break;
+	case BPF_OP_REG:
+		SStream_concat(O, BPF_reg_name((csh)MI->csh, op->reg));
+		break;
+	case BPF_OP_IMM:
+		SStream_concat(O, "0x%" PRIx64, op->imm);
+		break;
+	case BPF_OP_OFF:
+		SStream_concat(O, "+0x%x", op->off);
+		break;
+	case BPF_OP_MEM:
+		SStream_concat(O, "[");
+		if (op->mem.base != BPF_REG_INVALID)
+			SStream_concat(O, BPF_reg_name((csh)MI->csh, op->mem.base));
+		if (op->mem.disp != 0) {
+			if (op->mem.base != BPF_REG_INVALID)
+				SStream_concat(O, "+");
+			SStream_concat(O, "0x%x", op->mem.disp);
+		}
+		if (op->mem.base == BPF_REG_INVALID && op->mem.disp == 0) // special case
+			SStream_concat(O, "0x0");
+		SStream_concat(O, "]");
+		break;
+	case BPF_OP_MMEM:
+		SStream_concat(O, "m[0x%x]", op->mmem);
+		break;
+	case BPF_OP_MSH:
+		SStream_concat(O, "4*([0x%x]&0xf)", op->msh);
+		break;
+	case BPF_OP_EXT:
+		switch (op->ext) {
+		case BPF_EXT_LEN:
+			SStream_concat(O, "#len");
+			break;
+		}
+		break;
+	}
+}
+
+/*
+ * 1. human readable mnemonic
+ * 2. set pubOpcode (BPF_INSN_*)
+ * 3. set detail->bpf.operands
+ * */
+void BPF_printInst(MCInst *MI, struct SStream *O, void *PrinterInfo)
+{
+	int i;
+	cs_insn insn;
+	cs_bpf bpf;
+
+	insn.detail = NULL;
+	/* set pubOpcode as instruction id */
+	BPF_get_insn_id((cs_struct*)MI->csh, &insn, MCInst_getOpcode(MI));
+	MCInst_setOpcodePub(MI, insn.id);
+
+	SStream_concat(O, BPF_insn_name((csh)MI->csh, insn.id));
+	convert_operands(MI, &bpf);
+	for (i = 0; i < bpf.op_count; i++) {
+		if (i == 0)
+			SStream_concat(O, "\t");
+		else
+			SStream_concat(O, ", ");
+		print_operand(MI, O, &bpf.operands[i]);
+	}
+
+#ifndef CAPSTONE_DIET
+	if (MI->flat_insn->detail) {
+		MI->flat_insn->detail->bpf = bpf;
+	}
+#endif
+}

thirdparty/capstone/arch/BPF/BPFInstPrinter.h

@@ -0,0 +1,16 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+#ifndef CS_BPFINSTPRINTER_H
+#define CS_BPFINSTPRINTER_H
+
+#include <capstone/capstone.h>
+
+#include "../../MCInst.h"
+#include "../../SStream.h"
+
+struct SStream;
+
+void BPF_printInst(MCInst *MI, struct SStream *O, void *Info);
+
+#endif

thirdparty/capstone/arch/BPF/BPFMapping.c

@@ -0,0 +1,534 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+#include <string.h>
+
+#include "BPFConstants.h"
+#include "BPFMapping.h"
+#include "../../Mapping.h"
+#include "../../utils.h"
+
+#ifndef CAPSTONE_DIET
+static const name_map group_name_maps[] = {
+	{ BPF_GRP_INVALID, NULL },
+
+	{ BPF_GRP_LOAD, "load" },
+	{ BPF_GRP_STORE, "store" },
+	{ BPF_GRP_ALU, "alu" },
+	{ BPF_GRP_JUMP, "jump" },
+	{ BPF_GRP_CALL, "call" },
+	{ BPF_GRP_RETURN, "return" },
+	{ BPF_GRP_MISC, "misc" },
+};
+#endif
+
+const char *BPF_group_name(csh handle, unsigned int id)
+{
+#ifndef CAPSTONE_DIET
+	return id2name(group_name_maps, ARR_SIZE(group_name_maps), id);
+#else
+	return NULL;
+#endif
+}
+
+#ifndef CAPSTONE_DIET
+static const name_map insn_name_maps[BPF_INS_ENDING] = {
+	{ BPF_INS_INVALID, NULL },
+
+	{ BPF_INS_ADD, "add" },
+	{ BPF_INS_SUB, "sub" },
+	{ BPF_INS_MUL, "mul" },
+	{ BPF_INS_DIV, "div" },
+	{ BPF_INS_OR, "or" },
+	{ BPF_INS_AND, "and" },
+	{ BPF_INS_LSH, "lsh" },
+	{ BPF_INS_RSH, "rsh" },
+	{ BPF_INS_NEG, "neg" },
+	{ BPF_INS_MOD, "mod" },
+	{ BPF_INS_XOR, "xor" },
+	{ BPF_INS_MOV, "mov" },
+	{ BPF_INS_ARSH, "arsh" },
+
+	{ BPF_INS_ADD64, "add64" },
+	{ BPF_INS_SUB64, "sub64" },
+	{ BPF_INS_MUL64, "mul64" },
+	{ BPF_INS_DIV64, "div64" },
+	{ BPF_INS_OR64, "or64" },
+	{ BPF_INS_AND64, "and64" },
+	{ BPF_INS_LSH64, "lsh64" },
+	{ BPF_INS_RSH64, "rsh64" },
+	{ BPF_INS_NEG64, "neg64" },
+	{ BPF_INS_MOD64, "mod64" },
+	{ BPF_INS_XOR64, "xor64" },
+	{ BPF_INS_MOV64, "mov64" },
+	{ BPF_INS_ARSH64, "arsh64" },
+
+	{ BPF_INS_LE16, "le16" },
+	{ BPF_INS_LE32, "le32" },
+	{ BPF_INS_LE64, "le64" },
+	{ BPF_INS_BE16, "be16" },
+	{ BPF_INS_BE32, "be32" },
+	{ BPF_INS_BE64, "be64" },
+	{ BPF_INS_BSWAP16, "bswap16" },
+	{ BPF_INS_BSWAP32, "bswap32" },
+	{ BPF_INS_BSWAP64, "bswap64" },
+
+	{ BPF_INS_LDW, "ldw" },
+	{ BPF_INS_LDH, "ldh" },
+	{ BPF_INS_LDB, "ldb" },
+	{ BPF_INS_LDDW,	"lddw" },
+	{ BPF_INS_LDXW, "ldxw" },
+	{ BPF_INS_LDXH, "ldxh" },
+	{ BPF_INS_LDXB, "ldxb" },
+	{ BPF_INS_LDXDW, "ldxdw" },
+
+	{ BPF_INS_STW, "stw" },
+	{ BPF_INS_STH, "sth" },
+	{ BPF_INS_STB, "stb" },
+	{ BPF_INS_STDW,	"stdw" },
+	{ BPF_INS_STXW, "stxw" },
+	{ BPF_INS_STXH, "stxh" },
+	{ BPF_INS_STXB, "stxb" },
+	{ BPF_INS_STXDW, "stxdw" },
+	{ BPF_INS_XADDW, "xaddw" },
+	{ BPF_INS_XADDDW, "xadddw" },
+
+	{ BPF_INS_JMP, "jmp" },
+	{ BPF_INS_JEQ, "jeq" },
+	{ BPF_INS_JGT, "jgt" },
+	{ BPF_INS_JGE, "jge" },
+	{ BPF_INS_JSET, "jset" },
+	{ BPF_INS_JNE, "jne" },
+	{ BPF_INS_JSGT,	"jsgt" },
+	{ BPF_INS_JSGE,	"jsge" },
+	{ BPF_INS_CALL,	"call" },
+	{ BPF_INS_CALLX, "callx" },
+	{ BPF_INS_EXIT,	"exit" },
+	{ BPF_INS_JLT, "jlt" },
+	{ BPF_INS_JLE, "jle" },
+	{ BPF_INS_JSLT, "jslt" },
+	{ BPF_INS_JSLE,	"jsle" },
+
+	{ BPF_INS_RET, "ret" },
+
+	{ BPF_INS_TAX, "tax" },
+	{ BPF_INS_TXA, "txa" },
+};
+#endif
+
+const char *BPF_insn_name(csh handle, unsigned int id)
+{
+#ifndef CAPSTONE_DIET
+	/* We have some special cases because 'ld' in cBPF is equivalent to 'ldw'
+	 * in eBPF, and we don't want to see 'ldw' appears in cBPF mode.
+	 */
+	if (!EBPF_MODE(handle)) {
+		switch (id) {
+		case BPF_INS_LD: return "ld";
+		case BPF_INS_LDX: return "ldx";
+		case BPF_INS_ST: return "st";
+		case BPF_INS_STX: return "stx";
+		}
+	}
+	return id2name(insn_name_maps, ARR_SIZE(insn_name_maps), id);
+#else
+	return NULL;
+#endif
+}
+
+const char *BPF_reg_name(csh handle, unsigned int reg)
+{
+#ifndef CAPSTONE_DIET
+	if (EBPF_MODE(handle)) {
+		if (reg < BPF_REG_R0 || reg > BPF_REG_R10)
+			return NULL;
+		static const char reg_names[11][4] = {
+			"r0", "r1", "r2", "r3", "r4",
+			"r5", "r6", "r7", "r8", "r9",
+			"r10"
+		};
+		return reg_names[reg - BPF_REG_R0];
+	}
+
+	/* cBPF mode */
+	if (reg == BPF_REG_A)
+		return "a";
+	else if (reg == BPF_REG_X)
+		return "x";
+	else
+		return NULL;
+#else
+	return NULL;
+#endif
+}
+
+static bpf_insn op2insn_ld(unsigned opcode)
+{
+#define CASE(c) case BPF_SIZE_##c: \
+		if (BPF_CLASS(opcode) == BPF_CLASS_LD) \
+			return BPF_INS_LD##c; \
+		else \
+			return BPF_INS_LDX##c;
+
+	switch (BPF_SIZE(opcode)) {
+	CASE(W);
+	CASE(H);
+	CASE(B);
+	CASE(DW);
+	}
+#undef CASE
+
+	return BPF_INS_INVALID;
+}
+
+static bpf_insn op2insn_st(unsigned opcode)
+{
+	/*
+	 * - BPF_STX | BPF_XADD | BPF_{W,DW}
+	 * - BPF_ST* | BPF_MEM | BPF_{W,H,B,DW}
+	 */
+
+	if (opcode == (BPF_CLASS_STX | BPF_MODE_XADD | BPF_SIZE_W))
+		return BPF_INS_XADDW;
+	if (opcode == (BPF_CLASS_STX | BPF_MODE_XADD | BPF_SIZE_DW))
+		return BPF_INS_XADDDW;
+
+	/* should be BPF_MEM */
+#define CASE(c) case BPF_SIZE_##c: \
+		if (BPF_CLASS(opcode) == BPF_CLASS_ST) \
+			return BPF_INS_ST##c; \
+		else \
+			return BPF_INS_STX##c;
+	switch (BPF_SIZE(opcode)) {
+	CASE(W);
+	CASE(H);
+	CASE(B);
+	CASE(DW);
+	}
+#undef CASE
+
+	return BPF_INS_INVALID;
+}
+
+static bpf_insn op2insn_alu(unsigned opcode)
+{
+	/* Endian is a special case */
+	if (BPF_OP(opcode) == BPF_ALU_END) {
+		if (BPF_CLASS(opcode) == BPF_CLASS_ALU64) {
+			switch (opcode ^ BPF_CLASS_ALU64 ^ BPF_ALU_END ^ BPF_SRC_LITTLE) {
+			case (16 << 4):
+				return BPF_INS_BSWAP16;
+			case (32 << 4):
+				return BPF_INS_BSWAP32;
+			case (64 << 4):
+				return BPF_INS_BSWAP64;
+			default:
+				return BPF_INS_INVALID;
+			}
+		}
+
+		switch (opcode ^ BPF_CLASS_ALU ^ BPF_ALU_END) {
+		case BPF_SRC_LITTLE | (16 << 4):
+			return BPF_INS_LE16;
+		case BPF_SRC_LITTLE | (32 << 4):
+			return BPF_INS_LE32;
+		case BPF_SRC_LITTLE | (64 << 4):
+			return BPF_INS_LE64;
+		case BPF_SRC_BIG | (16 << 4):
+			return BPF_INS_BE16;
+		case BPF_SRC_BIG | (32 << 4):
+			return BPF_INS_BE32;
+		case BPF_SRC_BIG | (64 << 4):
+			return BPF_INS_BE64;
+		}
+		return BPF_INS_INVALID;
+	}
+
+#define CASE(c) case BPF_ALU_##c: \
+		if (BPF_CLASS(opcode) == BPF_CLASS_ALU) \
+			return BPF_INS_##c; \
+		else \
+			return BPF_INS_##c##64;
+
+	switch (BPF_OP(opcode)) {
+	CASE(ADD);
+	CASE(SUB);
+	CASE(MUL);
+	CASE(DIV);
+	CASE(OR);
+	CASE(AND);
+	CASE(LSH);
+	CASE(RSH);
+	CASE(NEG);
+	CASE(MOD);
+	CASE(XOR);
+	CASE(MOV);
+	CASE(ARSH);
+	}
+#undef CASE
+
+	return BPF_INS_INVALID;
+}
+
+static bpf_insn op2insn_jmp(unsigned opcode)
+{
+	if (opcode == (BPF_CLASS_JMP | BPF_JUMP_CALL | BPF_SRC_X)) {
+		return BPF_INS_CALLX;
+	}
+
+#define CASE(c) case BPF_JUMP_##c: return BPF_INS_##c
+	switch (BPF_OP(opcode)) {
+	case BPF_JUMP_JA:
+		return BPF_INS_JMP;
+	CASE(JEQ);
+	CASE(JGT);
+	CASE(JGE);
+	CASE(JSET);
+	CASE(JNE);
+	CASE(JSGT);
+	CASE(JSGE);
+	CASE(CALL);
+	CASE(EXIT);
+	CASE(JLT);
+	CASE(JLE);
+	CASE(JSLT);
+	CASE(JSLE);
+	}
+#undef CASE
+
+	return BPF_INS_INVALID;
+}
+
+#ifndef CAPSTONE_DIET
+static void update_regs_access(cs_struct *ud, cs_detail *detail,
+		bpf_insn insn_id, unsigned int opcode)
+{
+	if (insn_id == BPF_INS_INVALID)
+		return;
+#define PUSH_READ(r) do { \
+		detail->regs_read[detail->regs_read_count] = r; \
+		detail->regs_read_count++; \
+	} while (0)
+#define PUSH_WRITE(r) do { \
+		detail->regs_write[detail->regs_write_count] = r; \
+		detail->regs_write_count++; \
+	} while (0)
+	/*
+	 * In eBPF mode, only these instructions have implicit registers access:
+	 * - legacy ld{w,h,b,dw} * // w: r0
+	 * - exit // r: r0
+	 */
+	if (EBPF_MODE(ud)) {
+		switch (insn_id) {
+		default:
+			break;
+		case BPF_INS_LDW:
+		case BPF_INS_LDH:
+		case BPF_INS_LDB:
+		case BPF_INS_LDDW:
+			if (BPF_MODE(opcode) == BPF_MODE_ABS || BPF_MODE(opcode) == BPF_MODE_IND) {
+				PUSH_WRITE(BPF_REG_R0);
+			}
+			break;
+		case BPF_INS_EXIT:
+			PUSH_READ(BPF_REG_R0);
+			break;
+		}
+		return;
+	}
+
+	/* cBPF mode */
+	switch (BPF_CLASS(opcode)) {
+	default:
+		break;
+	case BPF_CLASS_LD:
+		PUSH_WRITE(BPF_REG_A);
+		break;
+	case BPF_CLASS_LDX:
+		PUSH_WRITE(BPF_REG_X);
+		break;
+	case BPF_CLASS_ST:
+		PUSH_READ(BPF_REG_A);
+		break;
+	case BPF_CLASS_STX:
+		PUSH_READ(BPF_REG_X);
+		break;
+	case BPF_CLASS_ALU:
+		PUSH_READ(BPF_REG_A);
+		PUSH_WRITE(BPF_REG_A);
+		break;
+	case BPF_CLASS_JMP:
+		if (insn_id != BPF_INS_JMP) // except the unconditional jump
+			PUSH_READ(BPF_REG_A);
+		break;
+	/* case BPF_CLASS_RET: */
+	case BPF_CLASS_MISC:
+		if (insn_id == BPF_INS_TAX) {
+			PUSH_READ(BPF_REG_A);
+			PUSH_WRITE(BPF_REG_X);
+		}
+		else {
+			PUSH_READ(BPF_REG_X);
+			PUSH_WRITE(BPF_REG_A);
+		}
+		break;
+	}
+}
+#endif
+
+/*
+ * 1. Convert opcode(id) to BPF_INS_*
+ * 2. Set regs_read/regs_write/groups
+ */
+void BPF_get_insn_id(cs_struct *ud, cs_insn *insn, unsigned int opcode)
+{
+	// No need to care the mode (cBPF or eBPF) since all checks has be done in
+	// BPF_getInstruction, we can simply map opcode to BPF_INS_*.
+	bpf_insn id = BPF_INS_INVALID;
+#ifndef CAPSTONE_DIET
+	cs_detail *detail;
+	bpf_insn_group grp;
+
+	detail = insn->detail;
+ #define PUSH_GROUP(grp) do { \
+		if (detail) { \
+			detail->groups[detail->groups_count] = grp; \
+			detail->groups_count++; \
+		} \
+	} while(0)
+#else
+ #define PUSH_GROUP(grp) do {} while(0)
+#endif
+
+	switch (BPF_CLASS(opcode)) {
+	default:	// will never happen
+		break;
+	case BPF_CLASS_LD:
+	case BPF_CLASS_LDX:
+		id = op2insn_ld(opcode);
+		PUSH_GROUP(BPF_GRP_LOAD);
+		break;
+	case BPF_CLASS_ST:
+	case BPF_CLASS_STX:
+		id = op2insn_st(opcode);
+		PUSH_GROUP(BPF_GRP_STORE);
+		break;
+	case BPF_CLASS_ALU:
+		id = op2insn_alu(opcode);
+		PUSH_GROUP(BPF_GRP_ALU);
+		break;
+	case BPF_CLASS_JMP:
+		id = op2insn_jmp(opcode);
+#ifndef CAPSTONE_DIET
+		grp = BPF_GRP_JUMP;
+		if (id == BPF_INS_CALL || id == BPF_INS_CALLX)
+			grp = BPF_GRP_CALL;
+		else if (id == BPF_INS_EXIT)
+			grp = BPF_GRP_RETURN;
+		PUSH_GROUP(grp);
+#endif
+		break;
+	case BPF_CLASS_RET:
+		id = BPF_INS_RET;
+		PUSH_GROUP(BPF_GRP_RETURN);
+		break;
+	// BPF_CLASS_MISC and BPF_CLASS_ALU64 have exactly same value
+	case BPF_CLASS_MISC:
+	/* case BPF_CLASS_ALU64: */
+		if (EBPF_MODE(ud)) {
+			// ALU64 in eBPF
+			id = op2insn_alu(opcode);
+			PUSH_GROUP(BPF_GRP_ALU);
+		}
+		else {
+			if (BPF_MISCOP(opcode) == BPF_MISCOP_TXA)
+				id = BPF_INS_TXA;
+			else
+				id = BPF_INS_TAX;
+			PUSH_GROUP(BPF_GRP_MISC);
+		}
+		break;
+	}
+
+	insn->id = id;
+#undef PUSH_GROUP
+
+#ifndef CAPSTONE_DIET
+	if (detail) {
+		update_regs_access(ud, detail, id, opcode);
+	}
+#endif
+}
+
+static void sort_and_uniq(cs_regs arr, uint8_t n, uint8_t *new_n)
+{
+	/* arr is always a tiny (usually n < 3) array,
+	 * a simple O(n^2) sort is efficient enough. */
+	int i;
+	int j;
+	int iMin;
+	int tmp;
+
+	/* a modified selection sort for sorting and making unique */
+	for (j = 0; j < n; j++) {
+		/* arr[iMin] will be min(arr[j .. n-1]) */
+		iMin = j;
+		for (i = j + 1; i < n; i++) {
+			if (arr[i] < arr[iMin])
+				iMin = i;
+		}
+		if (j != 0 && arr[iMin] == arr[j - 1]) { // duplicate ele found
+			arr[iMin] = arr[n - 1];
+			--n;
+		}
+		else {
+			tmp = arr[iMin];
+			arr[iMin] = arr[j];
+			arr[j] = tmp;
+		}
+	}
+
+	*new_n = n;
+}
+void BPF_reg_access(const cs_insn *insn,
+		cs_regs regs_read, uint8_t *regs_read_count,
+		cs_regs regs_write, uint8_t *regs_write_count)
+{
+	unsigned i;
+	uint8_t read_count, write_count;
+	const cs_bpf *bpf = &(insn->detail->bpf);
+
+	read_count = insn->detail->regs_read_count;
+	write_count = insn->detail->regs_write_count;
+
+	// implicit registers
+	memcpy(regs_read, insn->detail->regs_read, read_count * sizeof(insn->detail->regs_read[0]));
+	memcpy(regs_write, insn->detail->regs_write, write_count * sizeof(insn->detail->regs_write[0]));
+
+	for (i = 0; i < bpf->op_count; i++) {
+		const cs_bpf_op *op = &(bpf->operands[i]);
+		switch (op->type) {
+		default:
+			break;
+		case BPF_OP_REG:
+			if (op->access & CS_AC_READ) {
+				regs_read[read_count] = op->reg;
+				read_count++;
+			}
+			if (op->access & CS_AC_WRITE) {
+				regs_write[write_count] = op->reg;
+				write_count++;
+			}
+			break;
+		case BPF_OP_MEM:
+			if (op->mem.base != BPF_REG_INVALID) {
+				regs_read[read_count] = op->mem.base;
+				read_count++;
+			}
+			break;
+		}
+	}
+
+	sort_and_uniq(regs_read, read_count, regs_read_count);
+	sort_and_uniq(regs_write, write_count, regs_write_count);
+}

thirdparty/capstone/arch/BPF/BPFMapping.h

@@ -0,0 +1,21 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+#ifndef CS_BPFMAPPING_H
+#define CS_BPFMAPPING_H
+
+#include <capstone/capstone.h>
+
+#include "../../cs_priv.h"
+
+#define EBPF_MODE(ud) (((cs_struct*)ud)->mode & CS_MODE_BPF_EXTENDED)
+
+const char *BPF_group_name(csh handle, unsigned int id);
+const char *BPF_insn_name(csh handle, unsigned int id);
+const char *BPF_reg_name(csh handle, unsigned int reg);
+void BPF_get_insn_id(cs_struct *h, cs_insn *insn, unsigned int id);
+void BPF_reg_access(const cs_insn *insn,
+		cs_regs regs_read, uint8_t *regs_read_count,
+		cs_regs regs_write, uint8_t *regs_write_count);
+
+#endif

thirdparty/capstone/arch/BPF/BPFModule.c

@@ -0,0 +1,34 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+#ifdef CAPSTONE_HAS_BPF
+
+#include "BPFDisassembler.h"
+#include "BPFInstPrinter.h"
+#include "BPFMapping.h"
+#include "BPFModule.h"
+
+cs_err BPF_global_init(cs_struct *ud)
+{
+	ud->printer = BPF_printInst;
+	ud->reg_name = BPF_reg_name;
+	ud->insn_id = BPF_get_insn_id;
+	ud->insn_name = BPF_insn_name;
+	ud->group_name = BPF_group_name;
+#ifndef CAPSTONE_DIET
+	ud->reg_access = BPF_reg_access;
+#endif
+	ud->disasm = BPF_getInstruction;
+
+	return CS_ERR_OK;
+}
+
+cs_err BPF_option(cs_struct *handle, cs_opt_type type, size_t value)
+{
+	if (type == CS_OPT_MODE)
+		handle->mode = (cs_mode)value;
+
+	return CS_ERR_OK;
+}
+
+#endif

thirdparty/capstone/arch/BPF/BPFModule.h

@@ -0,0 +1,12 @@
+/* Capstone Disassembly Engine */
+/* BPF Backend by david942j <david942j@gmail.com>, 2019 */
+
+#ifndef CS_BPF_MODULE_H
+#define CS_BPF_MODULE_H
+
+#include "../../utils.h"
+
+cs_err BPF_global_init(cs_struct *ud);
+cs_err BPF_option(cs_struct *handle, cs_opt_type type, size_t value);
+
+#endif