1*9712c20fSFrederick Mayle // Copyright 2010 Google LLC
2*9712c20fSFrederick Mayle //
3*9712c20fSFrederick Mayle // Redistribution and use in source and binary forms, with or without
4*9712c20fSFrederick Mayle // modification, are permitted provided that the following conditions are
5*9712c20fSFrederick Mayle // met:
6*9712c20fSFrederick Mayle //
7*9712c20fSFrederick Mayle // * Redistributions of source code must retain the above copyright
8*9712c20fSFrederick Mayle // notice, this list of conditions and the following disclaimer.
9*9712c20fSFrederick Mayle // * Redistributions in binary form must reproduce the above
10*9712c20fSFrederick Mayle // copyright notice, this list of conditions and the following disclaimer
11*9712c20fSFrederick Mayle // in the documentation and/or other materials provided with the
12*9712c20fSFrederick Mayle // distribution.
13*9712c20fSFrederick Mayle // * Neither the name of Google LLC nor the names of its
14*9712c20fSFrederick Mayle // contributors may be used to endorse or promote products derived from
15*9712c20fSFrederick Mayle // this software without specific prior written permission.
16*9712c20fSFrederick Mayle //
17*9712c20fSFrederick Mayle // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
18*9712c20fSFrederick Mayle // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
19*9712c20fSFrederick Mayle // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
20*9712c20fSFrederick Mayle // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
21*9712c20fSFrederick Mayle // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
22*9712c20fSFrederick Mayle // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
23*9712c20fSFrederick Mayle // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
24*9712c20fSFrederick Mayle // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
25*9712c20fSFrederick Mayle // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
26*9712c20fSFrederick Mayle // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
27*9712c20fSFrederick Mayle // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
28*9712c20fSFrederick Mayle
29*9712c20fSFrederick Mayle // disassembler_x86.cc: simple x86 disassembler.
30*9712c20fSFrederick Mayle //
31*9712c20fSFrederick Mayle // Provides single step disassembly of x86 bytecode and flags instructions
32*9712c20fSFrederick Mayle // that utilize known bad register values.
33*9712c20fSFrederick Mayle //
34*9712c20fSFrederick Mayle // Author: Cris Neckar
35*9712c20fSFrederick Mayle
36*9712c20fSFrederick Mayle #ifdef HAVE_CONFIG_H
37*9712c20fSFrederick Mayle #include <config.h> // Must come first
38*9712c20fSFrederick Mayle #endif
39*9712c20fSFrederick Mayle
40*9712c20fSFrederick Mayle #include "processor/disassembler_x86.h"
41*9712c20fSFrederick Mayle
42*9712c20fSFrederick Mayle #include <string.h>
43*9712c20fSFrederick Mayle
44*9712c20fSFrederick Mayle namespace google_breakpad {
45*9712c20fSFrederick Mayle
DisassemblerX86(const uint8_t * bytecode,uint32_t size,uint32_t virtual_address)46*9712c20fSFrederick Mayle DisassemblerX86::DisassemblerX86(const uint8_t* bytecode,
47*9712c20fSFrederick Mayle uint32_t size,
48*9712c20fSFrederick Mayle uint32_t virtual_address) :
49*9712c20fSFrederick Mayle bytecode_(bytecode),
50*9712c20fSFrederick Mayle size_(size),
51*9712c20fSFrederick Mayle virtual_address_(virtual_address),
52*9712c20fSFrederick Mayle current_byte_offset_(0),
53*9712c20fSFrederick Mayle current_inst_offset_(0),
54*9712c20fSFrederick Mayle instr_valid_(false),
55*9712c20fSFrederick Mayle register_valid_(false),
56*9712c20fSFrederick Mayle pushed_bad_value_(false),
57*9712c20fSFrederick Mayle end_of_block_(false),
58*9712c20fSFrederick Mayle flags_(0) {
59*9712c20fSFrederick Mayle libdis::x86_init(libdis::opt_none, NULL, NULL);
60*9712c20fSFrederick Mayle }
61*9712c20fSFrederick Mayle
~DisassemblerX86()62*9712c20fSFrederick Mayle DisassemblerX86::~DisassemblerX86() {
63*9712c20fSFrederick Mayle if (instr_valid_)
64*9712c20fSFrederick Mayle libdis::x86_oplist_free(¤t_instr_);
65*9712c20fSFrederick Mayle
66*9712c20fSFrederick Mayle libdis::x86_cleanup();
67*9712c20fSFrederick Mayle }
68*9712c20fSFrederick Mayle
NextInstruction()69*9712c20fSFrederick Mayle uint32_t DisassemblerX86::NextInstruction() {
70*9712c20fSFrederick Mayle if (instr_valid_)
71*9712c20fSFrederick Mayle libdis::x86_oplist_free(¤t_instr_);
72*9712c20fSFrederick Mayle
73*9712c20fSFrederick Mayle if (current_byte_offset_ >= size_) {
74*9712c20fSFrederick Mayle instr_valid_ = false;
75*9712c20fSFrederick Mayle return 0;
76*9712c20fSFrederick Mayle }
77*9712c20fSFrederick Mayle uint32_t instr_size = 0;
78*9712c20fSFrederick Mayle instr_size = libdis::x86_disasm((unsigned char*)bytecode_, size_,
79*9712c20fSFrederick Mayle virtual_address_, current_byte_offset_,
80*9712c20fSFrederick Mayle ¤t_instr_);
81*9712c20fSFrederick Mayle if (instr_size == 0) {
82*9712c20fSFrederick Mayle instr_valid_ = false;
83*9712c20fSFrederick Mayle return 0;
84*9712c20fSFrederick Mayle }
85*9712c20fSFrederick Mayle
86*9712c20fSFrederick Mayle current_byte_offset_ += instr_size;
87*9712c20fSFrederick Mayle current_inst_offset_++;
88*9712c20fSFrederick Mayle instr_valid_ = libdis::x86_insn_is_valid(¤t_instr_);
89*9712c20fSFrederick Mayle if (!instr_valid_)
90*9712c20fSFrederick Mayle return 0;
91*9712c20fSFrederick Mayle
92*9712c20fSFrederick Mayle if (current_instr_.type == libdis::insn_return)
93*9712c20fSFrederick Mayle end_of_block_ = true;
94*9712c20fSFrederick Mayle libdis::x86_op_t* src = libdis::x86_get_src_operand(¤t_instr_);
95*9712c20fSFrederick Mayle libdis::x86_op_t* dest = libdis::x86_get_dest_operand(¤t_instr_);
96*9712c20fSFrederick Mayle
97*9712c20fSFrederick Mayle if (register_valid_) {
98*9712c20fSFrederick Mayle switch (current_instr_.group) {
99*9712c20fSFrederick Mayle // Flag branches based off of bad registers and calls that occur
100*9712c20fSFrederick Mayle // after pushing bad values.
101*9712c20fSFrederick Mayle case libdis::insn_controlflow:
102*9712c20fSFrederick Mayle switch (current_instr_.type) {
103*9712c20fSFrederick Mayle case libdis::insn_jmp:
104*9712c20fSFrederick Mayle case libdis::insn_jcc:
105*9712c20fSFrederick Mayle case libdis::insn_call:
106*9712c20fSFrederick Mayle case libdis::insn_callcc:
107*9712c20fSFrederick Mayle if (dest) {
108*9712c20fSFrederick Mayle switch (dest->type) {
109*9712c20fSFrederick Mayle case libdis::op_expression:
110*9712c20fSFrederick Mayle if (dest->data.expression.base.id == bad_register_.id)
111*9712c20fSFrederick Mayle flags_ |= DISX86_BAD_BRANCH_TARGET;
112*9712c20fSFrederick Mayle break;
113*9712c20fSFrederick Mayle case libdis::op_register:
114*9712c20fSFrederick Mayle if (dest->data.reg.id == bad_register_.id)
115*9712c20fSFrederick Mayle flags_ |= DISX86_BAD_BRANCH_TARGET;
116*9712c20fSFrederick Mayle break;
117*9712c20fSFrederick Mayle default:
118*9712c20fSFrederick Mayle if (pushed_bad_value_ &&
119*9712c20fSFrederick Mayle (current_instr_.type == libdis::insn_call ||
120*9712c20fSFrederick Mayle current_instr_.type == libdis::insn_callcc))
121*9712c20fSFrederick Mayle flags_ |= DISX86_BAD_ARGUMENT_PASSED;
122*9712c20fSFrederick Mayle break;
123*9712c20fSFrederick Mayle }
124*9712c20fSFrederick Mayle }
125*9712c20fSFrederick Mayle break;
126*9712c20fSFrederick Mayle default:
127*9712c20fSFrederick Mayle break;
128*9712c20fSFrederick Mayle }
129*9712c20fSFrederick Mayle break;
130*9712c20fSFrederick Mayle
131*9712c20fSFrederick Mayle // Flag block data operations that use bad registers for src or dest.
132*9712c20fSFrederick Mayle case libdis::insn_string:
133*9712c20fSFrederick Mayle if (dest && dest->type == libdis::op_expression &&
134*9712c20fSFrederick Mayle dest->data.expression.base.id == bad_register_.id)
135*9712c20fSFrederick Mayle flags_ |= DISX86_BAD_BLOCK_WRITE;
136*9712c20fSFrederick Mayle if (src && src->type == libdis::op_expression &&
137*9712c20fSFrederick Mayle src->data.expression.base.id == bad_register_.id)
138*9712c20fSFrederick Mayle flags_ |= DISX86_BAD_BLOCK_READ;
139*9712c20fSFrederick Mayle break;
140*9712c20fSFrederick Mayle
141*9712c20fSFrederick Mayle // Flag comparisons based on bad data.
142*9712c20fSFrederick Mayle case libdis::insn_comparison:
143*9712c20fSFrederick Mayle if ((dest && dest->type == libdis::op_expression &&
144*9712c20fSFrederick Mayle dest->data.expression.base.id == bad_register_.id) ||
145*9712c20fSFrederick Mayle (src && src->type == libdis::op_expression &&
146*9712c20fSFrederick Mayle src->data.expression.base.id == bad_register_.id) ||
147*9712c20fSFrederick Mayle (dest && dest->type == libdis::op_register &&
148*9712c20fSFrederick Mayle dest->data.reg.id == bad_register_.id) ||
149*9712c20fSFrederick Mayle (src && src->type == libdis::op_register &&
150*9712c20fSFrederick Mayle src->data.reg.id == bad_register_.id))
151*9712c20fSFrederick Mayle flags_ |= DISX86_BAD_COMPARISON;
152*9712c20fSFrederick Mayle break;
153*9712c20fSFrederick Mayle
154*9712c20fSFrederick Mayle // Flag any other instruction which derefs a bad register for
155*9712c20fSFrederick Mayle // src or dest.
156*9712c20fSFrederick Mayle default:
157*9712c20fSFrederick Mayle if (dest && dest->type == libdis::op_expression &&
158*9712c20fSFrederick Mayle dest->data.expression.base.id == bad_register_.id)
159*9712c20fSFrederick Mayle flags_ |= DISX86_BAD_WRITE;
160*9712c20fSFrederick Mayle if (src && src->type == libdis::op_expression &&
161*9712c20fSFrederick Mayle src->data.expression.base.id == bad_register_.id)
162*9712c20fSFrederick Mayle flags_ |= DISX86_BAD_READ;
163*9712c20fSFrederick Mayle break;
164*9712c20fSFrederick Mayle }
165*9712c20fSFrederick Mayle }
166*9712c20fSFrederick Mayle
167*9712c20fSFrederick Mayle // When a register is marked as tainted check if it is pushed.
168*9712c20fSFrederick Mayle // TODO(cdn): may also want to check for MOVs into EBP offsets.
169*9712c20fSFrederick Mayle if (register_valid_ && dest && current_instr_.type == libdis::insn_push) {
170*9712c20fSFrederick Mayle switch (dest->type) {
171*9712c20fSFrederick Mayle case libdis::op_expression:
172*9712c20fSFrederick Mayle if (dest->data.expression.base.id == bad_register_.id ||
173*9712c20fSFrederick Mayle dest->data.expression.index.id == bad_register_.id)
174*9712c20fSFrederick Mayle pushed_bad_value_ = true;
175*9712c20fSFrederick Mayle break;
176*9712c20fSFrederick Mayle case libdis::op_register:
177*9712c20fSFrederick Mayle if (dest->data.reg.id == bad_register_.id)
178*9712c20fSFrederick Mayle pushed_bad_value_ = true;
179*9712c20fSFrederick Mayle break;
180*9712c20fSFrederick Mayle default:
181*9712c20fSFrederick Mayle break;
182*9712c20fSFrederick Mayle }
183*9712c20fSFrederick Mayle }
184*9712c20fSFrederick Mayle
185*9712c20fSFrederick Mayle // Check if a tainted register value is clobbered.
186*9712c20fSFrederick Mayle // For conditional MOVs and XCHGs assume that
187*9712c20fSFrederick Mayle // there is a hit.
188*9712c20fSFrederick Mayle if (register_valid_) {
189*9712c20fSFrederick Mayle switch (current_instr_.type) {
190*9712c20fSFrederick Mayle case libdis::insn_xor:
191*9712c20fSFrederick Mayle if (src && src->type == libdis::op_register &&
192*9712c20fSFrederick Mayle dest && dest->type == libdis::op_register &&
193*9712c20fSFrederick Mayle src->data.reg.id == bad_register_.id &&
194*9712c20fSFrederick Mayle src->data.reg.id == dest->data.reg.id)
195*9712c20fSFrederick Mayle register_valid_ = false;
196*9712c20fSFrederick Mayle break;
197*9712c20fSFrederick Mayle case libdis::insn_pop:
198*9712c20fSFrederick Mayle case libdis::insn_mov:
199*9712c20fSFrederick Mayle case libdis::insn_movcc:
200*9712c20fSFrederick Mayle if (dest && dest->type == libdis::op_register &&
201*9712c20fSFrederick Mayle dest->data.reg.id == bad_register_.id)
202*9712c20fSFrederick Mayle register_valid_ = false;
203*9712c20fSFrederick Mayle break;
204*9712c20fSFrederick Mayle case libdis::insn_popregs:
205*9712c20fSFrederick Mayle register_valid_ = false;
206*9712c20fSFrederick Mayle break;
207*9712c20fSFrederick Mayle case libdis::insn_xchg:
208*9712c20fSFrederick Mayle case libdis::insn_xchgcc:
209*9712c20fSFrederick Mayle if (dest && dest->type == libdis::op_register &&
210*9712c20fSFrederick Mayle src && src->type == libdis::op_register) {
211*9712c20fSFrederick Mayle if (dest->data.reg.id == bad_register_.id)
212*9712c20fSFrederick Mayle memcpy(&bad_register_, &src->data.reg, sizeof(libdis::x86_reg_t));
213*9712c20fSFrederick Mayle else if (src->data.reg.id == bad_register_.id)
214*9712c20fSFrederick Mayle memcpy(&bad_register_, &dest->data.reg, sizeof(libdis::x86_reg_t));
215*9712c20fSFrederick Mayle }
216*9712c20fSFrederick Mayle break;
217*9712c20fSFrederick Mayle default:
218*9712c20fSFrederick Mayle break;
219*9712c20fSFrederick Mayle }
220*9712c20fSFrederick Mayle }
221*9712c20fSFrederick Mayle
222*9712c20fSFrederick Mayle return instr_size;
223*9712c20fSFrederick Mayle }
224*9712c20fSFrederick Mayle
setBadRead()225*9712c20fSFrederick Mayle bool DisassemblerX86::setBadRead() {
226*9712c20fSFrederick Mayle if (!instr_valid_)
227*9712c20fSFrederick Mayle return false;
228*9712c20fSFrederick Mayle
229*9712c20fSFrederick Mayle libdis::x86_op_t* operand = libdis::x86_get_src_operand(¤t_instr_);
230*9712c20fSFrederick Mayle if (!operand || operand->type != libdis::op_expression)
231*9712c20fSFrederick Mayle return false;
232*9712c20fSFrederick Mayle
233*9712c20fSFrederick Mayle memcpy(&bad_register_, &operand->data.expression.base,
234*9712c20fSFrederick Mayle sizeof(libdis::x86_reg_t));
235*9712c20fSFrederick Mayle register_valid_ = true;
236*9712c20fSFrederick Mayle return true;
237*9712c20fSFrederick Mayle }
238*9712c20fSFrederick Mayle
setBadWrite()239*9712c20fSFrederick Mayle bool DisassemblerX86::setBadWrite() {
240*9712c20fSFrederick Mayle if (!instr_valid_)
241*9712c20fSFrederick Mayle return false;
242*9712c20fSFrederick Mayle
243*9712c20fSFrederick Mayle libdis::x86_op_t* operand = libdis::x86_get_dest_operand(¤t_instr_);
244*9712c20fSFrederick Mayle if (!operand || operand->type != libdis::op_expression)
245*9712c20fSFrederick Mayle return false;
246*9712c20fSFrederick Mayle
247*9712c20fSFrederick Mayle memcpy(&bad_register_, &operand->data.expression.base,
248*9712c20fSFrederick Mayle sizeof(libdis::x86_reg_t));
249*9712c20fSFrederick Mayle register_valid_ = true;
250*9712c20fSFrederick Mayle return true;
251*9712c20fSFrederick Mayle }
252*9712c20fSFrederick Mayle
253*9712c20fSFrederick Mayle } // namespace google_breakpad
254