CodeVerify.h revision d862faa2ceae186da5518607505eb942d634ced9 
1/*
2 * Copyright (C) 2008 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 *      http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17/*
18 * Dalvik bytecode verifier.
19 */
20#ifndef _DALVIK_CODEVERIFY
21#define _DALVIK_CODEVERIFY
22
23#include "analysis/VerifySubs.h"
24#include "analysis/VfyBasicBlock.h"
25
26/*
27 * Enumeration for register type values.  The "hi" piece of a 64-bit value
28 * MUST immediately follow the "lo" piece in the enumeration, so we can check
29 * that hi==lo+1.
30 *
31 * Assignment of constants:
32 *   [-MAXINT,-32768)   : integer
33 *   [-32768,-128)      : short
34 *   [-128,0)           : byte
35 *   0                  : zero
36 *   1                  : one
37 *   [2,128)            : posbyte
38 *   [128,32768)        : posshort
39 *   [32768,65536)      : char
40 *   [65536,MAXINT]     : integer
41 *
42 * Allowed "implicit" widening conversions:
43 *   zero -> boolean, posbyte, byte, posshort, short, char, integer, ref (null)
44 *   one -> boolean, posbyte, byte, posshort, short, char, integer
45 *   boolean -> posbyte, byte, posshort, short, char, integer
46 *   posbyte -> posshort, short, integer, char
47 *   byte -> short, integer
48 *   posshort -> integer, char
49 *   short -> integer
50 *   char -> integer
51 *
52 * In addition, all of the above can convert to "float".
53 *
54 * We're more careful with integer values than the spec requires.  The
55 * motivation is to restrict byte/char/short to the correct range of values.
56 * For example, if a method takes a byte argument, we don't want to allow
57 * the code to load the constant "1024" and pass it in.
58 */
59enum {
60    kRegTypeUnknown = 0,    /* initial state; use value=0 so calloc works */
61    kRegTypeUninit = 1,     /* MUST be odd to distinguish from pointer */
62    kRegTypeConflict,       /* merge clash makes this reg's type unknowable */
63
64    /*
65     * Category-1nr types.  The order of these is chiseled into a couple
66     * of tables, so don't add, remove, or reorder if you can avoid it.
67     */
68#define kRegType1nrSTART    kRegTypeFloat
69    kRegTypeFloat,
70    kRegTypeZero,           /* 32-bit 0, could be Boolean, Int, Float, or Ref */
71    kRegTypeOne,            /* 32-bit 1, could be Boolean, Int, Float */
72    kRegTypeBoolean,        /* must be 0 or 1 */
73    kRegTypePosByte,        /* byte, known positive (can become char) */
74    kRegTypeByte,
75    kRegTypePosShort,       /* short, known positive (can become char) */
76    kRegTypeShort,
77    kRegTypeChar,
78    kRegTypeInteger,
79#define kRegType1nrEND      kRegTypeInteger
80
81    kRegTypeLongLo,         /* lower-numbered register; endian-independent */
82    kRegTypeLongHi,
83    kRegTypeDoubleLo,
84    kRegTypeDoubleHi,
85
86    /*
87     * Enumeration max; this is used with "full" (32-bit) RegType values.
88     *
89     * Anything larger than this is a ClassObject or uninit ref.  Mask off
90     * all but the low 8 bits; if you're left with kRegTypeUninit, pull
91     * the uninit index out of the high 24.  Because kRegTypeUninit has an
92     * odd value, there is no risk of a particular ClassObject pointer bit
93     * pattern being confused for it (assuming our class object allocator
94     * uses word alignment).
95     */
96    kRegTypeMAX
97};
98#define kRegTypeUninitMask  0xff
99#define kRegTypeUninitShift 8
100
101/*
102 * RegType holds information about the type of data held in a register.
103 * For most types it's a simple enum.  For reference types it holds a
104 * pointer to the ClassObject, and for uninitialized references it holds
105 * an index into the UninitInstanceMap.
106 */
107typedef u4 RegType;
108
109/*
110 * A bit vector indicating which entries in the monitor stack are
111 * associated with this register.  The low bit corresponds to the stack's
112 * bottom-most entry.
113 */
114typedef u4 MonitorEntries;
115#define kMaxMonitorStackDepth   (sizeof(MonitorEntries) * 8)
116
117/*
118 * During verification, we associate one of these with every "interesting"
119 * instruction.  We track the status of all registers, and (if the method
120 * has any monitor-enter instructions) maintain a stack of entered monitors
121 * (identified by code unit offset).
122 *
123 * If live-precise register maps are enabled, the "liveRegs" vector will
124 * be populated.  Unlike the other lists of registers here, we do not
125 * track the liveness of the method result register (which is not visible
126 * to the GC).
127 */
128struct RegisterLine {
129    RegType*        regTypes;
130    MonitorEntries* monitorEntries;
131    u4*             monitorStack;
132    unsigned int    monitorStackTop;
133    BitVector*      liveRegs;
134};
135
136/*
137 * Table that maps uninitialized instances to classes, based on the
138 * address of the new-instance instruction.  One per method.
139 */
140struct UninitInstanceMap {
141    int numEntries;
142    struct {
143        int             addr;   /* code offset, or -1 for method arg ("this") */
144        ClassObject*    clazz;  /* class created at this address */
145    } map[1];
146};
147#define kUninitThisArgAddr  (-1)
148#define kUninitThisArgSlot  0
149
150/*
151 * Various bits of data used by the verifier and register map generator.
152 */
153struct VerifierData {
154    /*
155     * The method we're working on.
156     */
157    const Method*   method;
158
159    /*
160     * Number of code units of instructions in the method.  A cache of the
161     * value calculated by dvmGetMethodInsnsSize().
162     */
163    u4              insnsSize;
164
165    /*
166     * Number of registers we track for each instruction.  This is equal
167     * to the method's declared "registersSize".  (Does not include the
168     * pending return value.)
169     */
170    u4              insnRegCount;
171
172    /*
173     * Instruction widths and flags, one entry per code unit.
174     */
175    InsnFlags*      insnFlags;
176
177    /*
178     * Uninitialized instance map, used for tracking the movement of
179     * objects that have been allocated but not initialized.
180     */
181    UninitInstanceMap* uninitMap;
182
183    /*
184     * Array of RegisterLine structs, one entry per code unit.  We only need
185     * entries for code units that hold the start of an "interesting"
186     * instruction.  For register map generation, we're only interested
187     * in GC points.
188     */
189    RegisterLine*   registerLines;
190
191    /*
192     * The number of occurrences of specific opcodes.
193     */
194    size_t          newInstanceCount;
195    size_t          monitorEnterCount;
196
197    /*
198     * Array of pointers to basic blocks, one entry per code unit.  Used
199     * for liveness analysis.
200     */
201    VfyBasicBlock** basicBlocks;
202};
203
204
205/* table with static merge logic for primitive types */
206extern const char gDvmMergeTab[kRegTypeMAX][kRegTypeMAX];
207
208
209/*
210 * Returns "true" if the flags indicate that this address holds the start
211 * of an instruction.
212 */
213INLINE bool dvmInsnIsOpcode(const InsnFlags* insnFlags, int addr) {
214    return (insnFlags[addr] & kInsnFlagWidthMask) != 0;
215}
216
217/*
218 * Extract the unsigned 16-bit instruction width from "flags".
219 */
220INLINE int dvmInsnGetWidth(const InsnFlags* insnFlags, int addr) {
221    return insnFlags[addr] & kInsnFlagWidthMask;
222}
223
224/*
225 * Changed?
226 */
227INLINE bool dvmInsnIsChanged(const InsnFlags* insnFlags, int addr) {
228    return (insnFlags[addr] & kInsnFlagChanged) != 0;
229}
230INLINE void dvmInsnSetChanged(InsnFlags* insnFlags, int addr, bool changed)
231{
232    if (changed)
233        insnFlags[addr] |= kInsnFlagChanged;
234    else
235        insnFlags[addr] &= ~kInsnFlagChanged;
236}
237
238/*
239 * Visited?
240 */
241INLINE bool dvmInsnIsVisited(const InsnFlags* insnFlags, int addr) {
242    return (insnFlags[addr] & kInsnFlagVisited) != 0;
243}
244INLINE void dvmInsnSetVisited(InsnFlags* insnFlags, int addr, bool changed)
245{
246    if (changed)
247        insnFlags[addr] |= kInsnFlagVisited;
248    else
249        insnFlags[addr] &= ~kInsnFlagVisited;
250}
251
252/*
253 * Visited or changed?
254 */
255INLINE bool dvmInsnIsVisitedOrChanged(const InsnFlags* insnFlags, int addr) {
256    return (insnFlags[addr] & (kInsnFlagVisited|kInsnFlagChanged)) != 0;
257}
258
259/*
260 * In a "try" block?
261 */
262INLINE bool dvmInsnIsInTry(const InsnFlags* insnFlags, int addr) {
263    return (insnFlags[addr] & kInsnFlagInTry) != 0;
264}
265INLINE void dvmInsnSetInTry(InsnFlags* insnFlags, int addr, bool inTry)
266{
267    assert(inTry);
268    //if (inTry)
269        insnFlags[addr] |= kInsnFlagInTry;
270    //else
271    //    insnFlags[addr] &= ~kInsnFlagInTry;
272}
273
274/*
275 * Instruction is a branch target or exception handler?
276 */
277INLINE bool dvmInsnIsBranchTarget(const InsnFlags* insnFlags, int addr) {
278    return (insnFlags[addr] & kInsnFlagBranchTarget) != 0;
279}
280INLINE void dvmInsnSetBranchTarget(InsnFlags* insnFlags, int addr,
281    bool isBranch)
282{
283    assert(isBranch);
284    //if (isBranch)
285        insnFlags[addr] |= kInsnFlagBranchTarget;
286    //else
287    //    insnFlags[addr] &= ~kInsnFlagBranchTarget;
288}
289
290/*
291 * Instruction is a GC point?
292 */
293INLINE bool dvmInsnIsGcPoint(const InsnFlags* insnFlags, int addr) {
294    return (insnFlags[addr] & kInsnFlagGcPoint) != 0;
295}
296INLINE void dvmInsnSetGcPoint(InsnFlags* insnFlags, int addr,
297    bool isGcPoint)
298{
299    assert(isGcPoint);
300    //if (isGcPoint)
301        insnFlags[addr] |= kInsnFlagGcPoint;
302    //else
303    //    insnFlags[addr] &= ~kInsnFlagGcPoint;
304}
305
306
307/*
308 * Create a new UninitInstanceMap.
309 */
310UninitInstanceMap* dvmCreateUninitInstanceMap(const Method* meth,
311    const InsnFlags* insnFlags, int newInstanceCount);
312
313/*
314 * Release the storage associated with an UninitInstanceMap.
315 */
316void dvmFreeUninitInstanceMap(UninitInstanceMap* uninitMap);
317
318/*
319 * Verify bytecode in "meth".  "insnFlags" should be populated with
320 * instruction widths and "in try" flags.
321 */
322bool dvmVerifyCodeFlow(VerifierData* vdata);
323
324#endif /*_DALVIK_CODEVERIFY*/
325