A few changes to libdisasm

3rd_party/libdisasm/ia32_reg.cpp

@@ -190,7 +190,7 @@ static struct {
     /* REG_EIPMSR_INDEX : SYSENTER_EIP_MSR : 92 */
     { REG_DWORD_SIZE, reg_sys, 0, "eip_msr" },
     { 0 }
- };
+};
 
 
 static size_t sz_regtable = NUM_X86_REGS + 1;

3rd_party/libdisasm/libdis.h

@@ -210,7 +210,7 @@ enum x86_op_datatype {          /* these use Intel's lame terminology */
         op_fpustate32 = 23,	/* 108 byte FPU state (env & reg stack) */
         op_fpregset = 24,	/* 512 bytes: register set */
         op_fpreg = 25,		/* FPU register */
-    op_none = 0xFF,     /* operand without a datatype (INVLPG) */
+        op_none = 0xFF     /* operand without a datatype (INVLPG) */
 };
 
 enum x86_op_access {    /* ORed together */
@@ -275,24 +275,20 @@ struct x86_op_t{
         unsigned char   fpuenv[28];
         /* offset from segment */
         uint32_t        offset;
-                /* ID of CPU register */
+        x86_reg_t       reg;            /* ID of CPU register */
-                x86_reg_t       reg;
+        char            relative_near; /* offsets from current insn */
-                /* offsets from current insn */
-                char            relative_near;
         int32_t         relative_far;
-                /* segment:offset */
+        x86_absolute_t	absolute;   /* segment:offset */
-                x86_absolute_t	absolute;
+        x86_ea_t        expression; /* effective address [expression] */
-                /* effective address [expression] */
-                x86_ea_t        expression;
     } data;
     /* this is needed to make formatting operands more sane */
     void * insn;		/* pointer to x86_insn_t owning operand */
     size_t size()
     {
-        return x86_operand_size();
+        return operand_size();
     }
     /* get size of operand data in bytes */
-    size_t x86_operand_size();
+    size_t operand_size();
     /* format (sprintf) an operand into 'buf' using specified syntax */
     int x86_format_operand(char *buf, int len, enum x86_asm_format format );
     bool is_address( ) {
@@ -302,9 +298,9 @@ struct x86_op_t{
         return ( type == op_relative_near || type == op_relative_far );
     }
     char * format( enum x86_asm_format format );
-    x86_op_t * copy() {
+    x86_op_t * copy()
+    {
         x86_op_t *op = (x86_op_t *) calloc( sizeof(x86_op_t), 1 );
-
         if ( op ) {
             memcpy( op, this, sizeof(x86_op_t) );
         }
@@ -524,6 +520,7 @@ enum x86_insn_prefix {
 
 
 /* TODO: maybe provide insn_new/free(), and have disasm return new insn_t */
+
 /* FOREACH types: these are used to limit the foreach results to
  * operands which match a certain "type" (implicit or explicit)
  * or which are accessed in certain ways (e.g. read or write). Note
@@ -577,9 +574,9 @@ public:
     /* information about the instruction */
     uint32_t addr;              /* load address */
     uint32_t offset;            /* offset into file/buffer */
-    enum x86_insn_group group;      /* meta-type, e.g. INS_EXEC */
+    x86_insn_group group;       /* meta-type, e.g. INS_EXEC */
-    enum x86_insn_type type;        /* type, e.g. INS_BRANCH */
+    x86_insn_type type;         /* type, e.g. INS_BRANCH */
-    enum x86_insn_note note;	/* note, e.g. RING0 */
+    x86_insn_note note;         /* note, e.g. RING0 */
     unsigned char bytes[MAX_INSN_SIZE];
     unsigned char size;             /* size of insn in bytes */
     /* 16/32-bit mode settings */
@@ -607,58 +604,29 @@ public:
     void *block;                    /* code block containing this insn */
     void *function;                 /* function containing this insn */
     int tag;			/* tag the insn as seen/processed */
-    x86_op_t *x86_operand_new();
+    x86_op_t *  x86_operand_new();
-    /* convenience routine: returns count of operands matching 'type' */
     size_t      x86_operand_count( enum x86_op_foreach_type type );
     /* accessor functions for the operands */
     x86_op_t *  x86_operand_1st( );
     x86_op_t *  x86_operand_2nd( );
     x86_op_t *  x86_operand_3rd( );
-    /* Get Relative Offset: return as a sign-extended int32_t the near or far
+    x86_op_t *  get_dest();
-     * relative offset operand, or 0 if there is none. There can be only one
-     * relaive offset operand in an instruction. */
     int32_t     x86_get_rel_offset( );
-    /* Get Branch Target: return the x86_op_t containing the target of
-     * a jump or call operand, or NULL if there is no branch target.
-     * Internally, a 'branch target' is defined as any operand with
-     * Execute Access set. There can be only one branch target per instruction. */
     x86_op_t *  x86_get_branch_target( );
-    /* Get Immediate: return the x86_op_t containing the immediate operand
-     * for this instruction, or NULL if there is no immediate operand. There
-     * can be only one immediate operand per instruction */
     x86_op_t *  x86_get_imm( );
-    /* Get Raw Immediate Data: returns a pointer to the immediate data encoded
+    uint8_t *   x86_get_raw_imm( );
-     * in the instruction. This is useful for large data types [>32 bits] currently
-     * not supported by libdisasm, or for determining if the disassembler
-     * screwed up the conversion of the immediate data. Note that 'imm' in this
-     * context refers to immediate data encoded at the end of an instruction as
-     * detailed in the Intel Manual Vol II Chapter 2; it does not refer to the
-     * 'op_imm' operand (the third operand in instructions like 'mul' */
-    unsigned char * x86_get_raw_imm( );
     /* More accessor fuctions, this time for user-defined info... */
-    /* set the address (usually RVA) of the insn */
     void    x86_set_insn_addr( uint32_t addr );
-    /* format (sprintf) an instruction mnemonic into 'buf' using specified syntax */
     int     x86_format_mnemonic( char *buf, int len, enum x86_asm_format format);
-    /* format (sprintf) an instruction into 'buf' using specified syntax;
-     * this includes formatting all operands */
     int     x86_format_insn( char *buf, int len, enum x86_asm_format);
-    /* free the operand list associated with an instruction -- useful for
-     * preventing memory leaks when free()ing an x86_insn_t */
     void    x86_oplist_free( );
-    /* returns 0 if an instruction is invalid, 1 if valid */
+    bool    is_valid( );
-    int x86_insn_is_valid( );
-    /* Get Address: return the value of an offset operand, or the offset of
-     * a segment:offset absolute address */
     uint32_t x86_get_address( );
     void    make_invalid(unsigned char *buf);
     /* instruction tagging: these routines allow the programmer to mark
      * instructions as "seen" in a DFS, for example. libdisasm does not use
      * the tag field.*/
-    /* set insn->tag to 1 */
     void x86_tag_insn( );
-
-    /* return insn->tag */
     int x86_insn_is_tagged();
     /* set insn->tag to 0 */
     void x86_untag_insn();

3rd_party/libdisasm/x86_format.cpp

@@ -1077,10 +1077,10 @@ static int format_att_mnemonic( x86_insn_t *insn, char *buf, int len) {
             )) {
             if ( insn->x86_operand_count( op_explicit ) > 0 &&
                  is_memory_op( insn->x86_operand_1st() ) ){
-                size = insn->x86_operand_1st()->x86_operand_size();
+                size = insn->x86_operand_1st()->operand_size();
             } else if ( insn->x86_operand_count( op_explicit ) > 1 &&
                         is_memory_op( insn->x86_operand_2nd() ) ){
-                size = insn->x86_operand_2nd()->x86_operand_size();
+                size = insn->x86_operand_2nd()->operand_size();
             }
         }
 
@@ -1094,6 +1094,7 @@ static int format_att_mnemonic( x86_insn_t *insn, char *buf, int len) {
         return ( strlen( buf ) );
 }
 
+/** format (sprintf) an instruction mnemonic into 'buf' using specified syntax */
 int x86_format_mnemonic(x86_insn_t *insn, char *buf, int len,
                         enum x86_asm_format format){
         char str[MAX_OP_STRING];
@@ -1325,6 +1326,8 @@ int X86_Disasm::x86_format_header( char *buf, int len, enum x86_asm_format forma
         return( strlen(buf) );
 }
 
+/** format (sprintf) an instruction into 'buf' using specified syntax;
+ * this includes formatting all operands */
 int x86_insn_t::x86_format_insn( char *buf, int len,
                      enum x86_asm_format format ){
     char str[MAX_OP_STRING];

3rd_party/libdisasm/x86_insn.cpp

@@ -17,7 +17,8 @@ int x86_insn_is_valid( x86_insn_t *insn ) {
     return 0;
 }
 
-int x86_insn_t::x86_insn_is_valid( )
+/** \returns false if an instruction is invalid, true if valid */
+bool x86_insn_t::is_valid( )
 {
     if ( this && this->type != insn_invalid && this->size > 0 )
     {
@@ -26,10 +27,15 @@ int x86_insn_t::x86_insn_is_valid( )
 
     return 0;
 }
-uint32_t x86_insn_t::x86_get_address() {
+/* Get Address: return the value of an offset operand, or the offset of
+ * a segment:offset absolute address */
+
+uint32_t x86_insn_t::x86_get_address()
+{
     x86_oplist_t *op_lst;
     assert(this);
-    if (! operands ) {
+    if (! operands )
+    {
         return 0;
     }
 
@@ -48,6 +54,9 @@ uint32_t x86_insn_t::x86_get_address() {
     return 0;
 }
 
+/** Get Relative Offset: return as a sign-extended int32_t the near or far
+ * relative offset operand, or 0 if there is none. There can be only one
+ * relaive offset operand in an instruction. */
 int32_t x86_insn_t::x86_get_rel_offset( ) {
     x86_oplist_t *op_lst;
     assert(this);
@@ -66,6 +75,10 @@ int32_t x86_insn_t::x86_get_rel_offset( ) {
     return 0;
 }
 
+/** Get Branch Target: return the x86_op_t containing the target of
+ * a jump or call operand, or NULL if there is no branch target.
+ * Internally, a 'branch target' is defined as any operand with
+ * Execute Access set. There can be only one branch target per instruction. */
 x86_op_t * x86_insn_t::x86_get_branch_target() {
     x86_oplist_t *op_lst;
     assert(this);
@@ -81,6 +94,24 @@ x86_op_t * x86_insn_t::x86_get_branch_target() {
 
     return NULL;
 }
+x86_op_t * x86_insn_t::get_dest() {
+    x86_oplist_t *op_lst;
+    assert(this);
+    if ( ! operands ) {
+        return NULL;
+    }
+    assert(this->x86_operand_count(op_dest)==1);
+    for (op_lst = operands; op_lst; op_lst = op_lst->next ) {
+        if ( op_lst->op.access & op_write)
+            return &(op_lst->op);
+    }
+
+    return NULL;
+}
+
+/** \brief Get Immediate: return the x86_op_t containing the immediate operand
+  for this instruction, or NULL if there is no immediate operand. There
+  can be only one immediate operand per instruction */
 x86_op_t * x86_insn_t::x86_get_imm() {
     x86_oplist_t *op_lst;
     assert(this);
@@ -101,9 +132,15 @@ x86_op_t * x86_insn_t::x86_get_imm() {
     x->op.type == op_immediate && ! (x->op.flags.op_hardcode)
 
 
-/* if there is an immediate value in the instruction, return a pointer to
+/** \brief if there is an immediate value in the instruction, return a pointer to it
- * it */
+ * Get Raw Immediate Data: returns a pointer to the immediate data encoded
-unsigned char * x86_insn_t::x86_get_raw_imm() {
+ * in the instruction. This is useful for large data types [>32 bits] currently
+ * not supported by libdisasm, or for determining if the disassembler
+ * screwed up the conversion of the immediate data. Note that 'imm' in this
+ * context refers to immediate data encoded at the end of an instruction as
+ * detailed in the Intel Manual Vol II Chapter 2; it does not refer to the
+ * 'op_imm' operand (the third operand in instructions like 'mul' */
+uint8_t *x86_insn_t::x86_get_raw_imm() {
     int size, offset;
     x86_op_t *op  = NULL;
     assert(this);
@@ -128,13 +165,13 @@ unsigned char * x86_insn_t::x86_get_raw_imm() {
     }
 
     /* immediate data is at the end of the insn */
-    size = op->x86_operand_size();
+    size = op->operand_size();
     offset = size - size;
     return( &bytes[offset] );
 }
 
 
-size_t x86_op_t::x86_operand_size() {
+size_t x86_op_t::operand_size() {
     switch (datatype ) {
         case op_byte:    return 1;
         case op_word:    return 2;
@@ -166,6 +203,7 @@ size_t x86_op_t::x86_operand_size() {
     return(4);      /* default size */
 }
 
+/** set the address (usually RVA) of the insn */
 void x86_insn_t::x86_set_insn_addr( uint32_t _addr ) {
     addr = _addr;
 }
@@ -182,6 +220,7 @@ void x86_insn_t::x86_set_insn_block( void * _block ){
     block = _block;
 }
 
+/** set insn->tag to 1 */
 void x86_insn_t::x86_tag_insn(){
     tag = 1;
 }
@@ -190,6 +229,7 @@ void x86_insn_t::x86_untag_insn(){
     tag = 0;
 }
 
+/** \return insn->tag */
 int x86_insn_t::x86_insn_is_tagged(){
     return tag;
 }

3rd_party/libdisasm/x86_operand_list.cpp

@@ -64,6 +64,8 @@ x86_op_t * x86_insn_t::x86_operand_new( ) {
     return( &(op->op) );
 }
 
+/** free the operand list associated with an instruction -- useful for
+ * preventing memory leaks when free()ing an x86_insn_t */
 void x86_insn_t::x86_oplist_free( )
 {
     x86_oplist_t *op, *list;
@@ -84,8 +86,8 @@ void x86_insn_t::x86_oplist_free( )
 /* ================================================== LIBDISASM API */
 /* these could probably just be #defines, but that means exposing the
    enum... yet one more confusing thing in the API */
-int x86_insn_t::x86_operand_foreach( x86_operand_fn func, void *arg,
+int x86_insn_t::x86_operand_foreach( x86_operand_fn func, void *arg, enum x86_op_foreach_type type )
-                        enum x86_op_foreach_type type ){
+{
     x86_oplist_t *list;
     char _explicit = 1, implicit = 1;
     assert(this);
@@ -167,6 +169,7 @@ static void count_operand( x86_op_t *op, x86_insn_t *insn, void *arg ) {
     *count = *count + 1;
 }
 
+/** convenience routine: returns count of operands matching 'type' */
 size_t x86_insn_t::x86_operand_count( enum x86_op_foreach_type type ) {
     size_t count = 0;