aboutsummaryrefslogtreecommitdiffstats
path: root/libevmasm/Assembly.cpp
blob: 597fdae19dd7b328967c0e2bb7690795661a3eb3 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
/*
    This file is part of solidity.

    solidity is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.

    solidity is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with solidity.  If not, see <http://www.gnu.org/licenses/>.
*/
/** @file Assembly.cpp
 * @author Gav Wood <i@gavwood.com>
 * @date 2014
 */

#include "Assembly.h"

#include <libevmasm/CommonSubexpressionEliminator.h>
#include <libevmasm/ControlFlowGraph.h>
#include <libevmasm/PeepholeOptimiser.h>
#include <libevmasm/BlockDeduplicator.h>
#include <libevmasm/ConstantOptimiser.h>
#include <libevmasm/GasMeter.h>

#include <fstream>
#include <json/json.h>

using namespace std;
using namespace dev;
using namespace dev::eth;

void Assembly::append(Assembly const& _a)
{
    auto newDeposit = m_deposit + _a.deposit();
    for (AssemblyItem i: _a.m_items)
    {
        if (i.type() == Tag || i.type() == PushTag)
            i.setData(i.data() + m_usedTags);
        else if (i.type() == PushSub || i.type() == PushSubSize)
            i.setData(i.data() + m_subs.size());
        append(i);
    }
    m_deposit = newDeposit;
    m_usedTags += _a.m_usedTags;
    for (auto const& i: _a.m_data)
        m_data.insert(i);
    for (auto const& i: _a.m_strings)
        m_strings.insert(i);
    m_subs += _a.m_subs;
    for (auto const& lib: _a.m_libraries)
        m_libraries.insert(lib);
}

void Assembly::append(Assembly const& _a, int _deposit)
{
    assertThrow(_deposit <= _a.m_deposit, InvalidDeposit, "");

    append(_a);
    while (_deposit++ < _a.m_deposit)
        append(Instruction::POP);
}

unsigned Assembly::bytesRequired(unsigned subTagSize) const
{
    for (unsigned tagSize = subTagSize; true; ++tagSize)
    {
        unsigned ret = 1;
        for (auto const& i: m_data)
            ret += i.second.size();

        for (AssemblyItem const& i: m_items)
            ret += i.bytesRequired(tagSize);
        if (dev::bytesRequired(ret) <= tagSize)
            return ret;
    }
}

namespace
{

string locationFromSources(StringMap const& _sourceCodes, SourceLocation const& _location)
{
    if (_location.isEmpty() || _sourceCodes.empty() || _location.start >= _location.end || _location.start < 0)
        return "";

    auto it = _sourceCodes.find(*_location.sourceName);
    if (it == _sourceCodes.end())
        return "";

    string const& source = it->second;
    if (size_t(_location.start) >= source.size())
        return "";

    string cut = source.substr(_location.start, _location.end - _location.start);
    auto newLinePos = cut.find_first_of("\n");
    if (newLinePos != string::npos)
        cut = cut.substr(0, newLinePos) + "...";

    return cut;
}

class Functionalizer
{
public:
    Functionalizer (ostream& _out, string const& _prefix, StringMap const& _sourceCodes):
        m_out(_out), m_prefix(_prefix), m_sourceCodes(_sourceCodes)
    {}

    void feed(AssemblyItem const& _item)
    {
        if (!_item.location().isEmpty() && _item.location() != m_location)
        {
            flush();
            m_location = _item.location();
            printLocation();
        }
        if (!(
            _item.canBeFunctional() &&
            _item.returnValues() <= 1 &&
            _item.arguments() <= int(m_pending.size())
        ))
        {
            flush();
            m_out << m_prefix << (_item.type() == Tag ? "" : "  ") << _item.toAssemblyText() << endl;
            return;
        }
        string expression = _item.toAssemblyText();
        if (_item.arguments() > 0)
        {
            expression += "(";
            for (int i = 0; i < _item.arguments(); ++i)
            {
                expression += m_pending.back();
                m_pending.pop_back();
                if (i + 1 < _item.arguments())
                    expression += ", ";
            }
            expression += ")";
        }

        m_pending.push_back(expression);
        if (_item.returnValues() != 1)
            flush();
    }

    void flush()
    {
        for (string const& expression: m_pending)
            m_out << m_prefix << "  " << expression << endl;
        m_pending.clear();
    }

    void printLocation()
    {
        if (!m_location.sourceName && m_location.isEmpty())
            return;
        m_out << m_prefix << "    /*";
        if (m_location.sourceName)
            m_out << " \"" + *m_location.sourceName + "\"";
        if (!m_location.isEmpty())
            m_out << ":" << to_string(m_location.start) + ":" + to_string(m_location.end);
        m_out << "  " << locationFromSources(m_sourceCodes, m_location);
        m_out << " */" << endl;
    }

private:
    strings m_pending;
    SourceLocation m_location;

    ostream& m_out;
    string const& m_prefix;
    StringMap const& m_sourceCodes;
};

}

ostream& Assembly::streamAsm(ostream& _out, string const& _prefix, StringMap const& _sourceCodes) const
{
    Functionalizer f(_out, _prefix, _sourceCodes);

    for (auto const& i: m_items)
        f.feed(i);
    f.flush();

    if (!m_data.empty() || !m_subs.empty())
    {
        _out << _prefix << "stop" << endl;
        for (auto const& i: m_data)
            if (u256(i.first) >= m_subs.size())
                _out << _prefix << "data_" << toHex(u256(i.first)) << " " << toHex(i.second) << endl;

        for (size_t i = 0; i < m_subs.size(); ++i)
        {
            _out << endl << _prefix << "sub_" << i << ": assembly {\n";
            m_subs[i]->streamAsm(_out, _prefix + "    ", _sourceCodes);
            _out << _prefix << "}" << endl;
        }
    }

    if (m_auxiliaryData.size() > 0)
        _out << endl << _prefix << "auxdata: 0x" << toHex(m_auxiliaryData) << endl;

    return _out;
}

Json::Value Assembly::createJsonValue(string _name, int _begin, int _end, string _value, string _jumpType) const
{
    Json::Value value;
    value["name"] = _name;
    value["begin"] = _begin;
    value["end"] = _end;
    if (!_value.empty())
        value["value"] = _value;
    if (!_jumpType.empty())
        value["jumpType"] = _jumpType;
    return value;
}

string toStringInHex(u256 _value)
{
    std::stringstream hexStr;
    hexStr << hex << _value;
    return hexStr.str();
}

Json::Value Assembly::streamAsmJson(ostream& _out, StringMap const& _sourceCodes) const
{
    Json::Value root;

    Json::Value collection(Json::arrayValue);
    for (AssemblyItem const& i: m_items)
    {
        switch (i.type())
        {
        case Operation:
            collection.append(
                createJsonValue(instructionInfo(i.instruction()).name, i.location().start, i.location().end, i.getJumpTypeAsString()));
            break;
        case Push:
            collection.append(
                createJsonValue("PUSH", i.location().start, i.location().end, toStringInHex(i.data()), i.getJumpTypeAsString()));
            break;
        case PushString:
            collection.append(
                createJsonValue("PUSH tag", i.location().start, i.location().end, m_strings.at((h256)i.data())));
            break;
        case PushTag:
            if (i.data() == 0)
                collection.append(
                    createJsonValue("PUSH [ErrorTag]", i.location().start, i.location().end, ""));
            else
                collection.append(
                    createJsonValue("PUSH [tag]", i.location().start, i.location().end, string(i.data())));
            break;
        case PushSub:
            collection.append(
                createJsonValue("PUSH [$]", i.location().start, i.location().end, dev::toString(h256(i.data()))));
            break;
        case PushSubSize:
            collection.append(
                createJsonValue("PUSH #[$]", i.location().start, i.location().end, dev::toString(h256(i.data()))));
            break;
        case PushProgramSize:
            collection.append(
                createJsonValue("PUSHSIZE", i.location().start, i.location().end));
            break;
        case PushLibraryAddress:
            collection.append(
                createJsonValue("PUSHLIB", i.location().start, i.location().end, m_libraries.at(h256(i.data())))
            );
            break;
        case Tag:
            collection.append(
                createJsonValue("tag", i.location().start, i.location().end, string(i.data())));
            collection.append(
                createJsonValue("JUMPDEST", i.location().start, i.location().end));
            break;
        case PushData:
            collection.append(createJsonValue("PUSH data", i.location().start, i.location().end, toStringInHex(i.data())));
            break;
        default:
            BOOST_THROW_EXCEPTION(InvalidOpcode());
        }
    }

    root[".code"] = collection;

    if (!m_data.empty() || !m_subs.empty())
    {
        Json::Value data;
        for (auto const& i: m_data)
            if (u256(i.first) >= m_subs.size())
                data[toStringInHex((u256)i.first)] = toHex(i.second);

        for (size_t i = 0; i < m_subs.size(); ++i)
        {
            std::stringstream hexStr;
            hexStr << hex << i;
            data[hexStr.str()] = m_subs[i]->stream(_out, "", _sourceCodes, true);
        }
        root[".data"] = data;
    }

    if (m_auxiliaryData.size() > 0)
        root[".auxdata"] = toHex(m_auxiliaryData);

    _out << root;

    return root;
}

Json::Value Assembly::stream(ostream& _out, string const& _prefix, StringMap const& _sourceCodes, bool _inJsonFormat) const
{
    if (_inJsonFormat)
        return streamAsmJson(_out, _sourceCodes);
    else
    {
        streamAsm(_out, _prefix, _sourceCodes);
        return Json::Value();
    }
}

AssemblyItem const& Assembly::append(AssemblyItem const& _i)
{
    assertThrow(m_deposit >= 0, AssemblyException, "");
    m_deposit += _i.deposit();
    m_items.push_back(_i);
    if (m_items.back().location().isEmpty() && !m_currentSourceLocation.isEmpty())
        m_items.back().setLocation(m_currentSourceLocation);
    return back();
}

AssemblyItem Assembly::newPushLibraryAddress(string const& _identifier)
{
    h256 h(dev::keccak256(_identifier));
    m_libraries[h] = _identifier;
    return AssemblyItem(PushLibraryAddress, h);
}

void Assembly::injectStart(AssemblyItem const& _i)
{
    m_items.insert(m_items.begin(), _i);
}

Assembly& Assembly::optimise(bool _enable, bool _isCreation, size_t _runs)
{
    optimiseInternal(_enable, _isCreation, _runs);
    return *this;
}

map<u256, u256> Assembly::optimiseInternal(bool _enable, bool _isCreation, size_t _runs)
{
    for (size_t subId = 0; subId < m_subs.size(); ++subId)
    {
        map<u256, u256> subTagReplacements = m_subs[subId]->optimiseInternal(_enable, false, _runs);
        BlockDeduplicator::applyTagReplacement(m_items, subTagReplacements, subId);
    }

    map<u256, u256> tagReplacements;
    for (unsigned count = 1; count > 0;)
    {
        count = 0;

        PeepholeOptimiser peepOpt(m_items);
        while (peepOpt.optimise())
            count++;

        if (!_enable)
            continue;

        // This only modifies PushTags, we have to run again to actually remove code.
        BlockDeduplicator dedup(m_items);
        if (dedup.deduplicate())
        {
            tagReplacements.insert(dedup.replacedTags().begin(), dedup.replacedTags().end());
            count++;
        }

        {
            // Control flow graph optimization has been here before but is disabled because it
            // assumes we only jump to tags that are pushed. This is not the case anymore with
            // function types that can be stored in storage.
            AssemblyItems optimisedItems;

            auto iter = m_items.begin();
            while (iter != m_items.end())
            {
                KnownState emptyState;
                CommonSubexpressionEliminator eliminator(emptyState);
                auto orig = iter;
                iter = eliminator.feedItems(iter, m_items.end());
                bool shouldReplace = false;
                AssemblyItems optimisedChunk;
                try
                {
                    optimisedChunk = eliminator.getOptimizedItems();
                    shouldReplace = (optimisedChunk.size() < size_t(iter - orig));
                }
                catch (StackTooDeepException const&)
                {
                    // This might happen if the opcode reconstruction is not as efficient
                    // as the hand-crafted code.
                }
                catch (ItemNotAvailableException const&)
                {
                    // This might happen if e.g. associativity and commutativity rules
                    // reorganise the expression tree, but not all leaves are available.
                }

                if (shouldReplace)
                {
                    count++;
                    optimisedItems += optimisedChunk;
                }
                else
                    copy(orig, iter, back_inserter(optimisedItems));
            }
            if (optimisedItems.size() < m_items.size())
            {
                m_items = move(optimisedItems);
                count++;
            }
        }
    }

    if (_enable)
        ConstantOptimisationMethod::optimiseConstants(
            _isCreation,
            _isCreation ? 1 : _runs,
            *this,
            m_items
        );

    return tagReplacements;
}

LinkerObject const& Assembly::assemble() const
{
    if (!m_assembledObject.bytecode.empty())
        return m_assembledObject;

    size_t subTagSize = 1;
    for (auto const& sub: m_subs)
    {
        sub->assemble();
        if (!sub->m_tagPositionsInBytecode.empty())
            subTagSize = max(subTagSize, *max_element(sub->m_tagPositionsInBytecode.begin(), sub->m_tagPositionsInBytecode.end()));
    }

    LinkerObject& ret = m_assembledObject;

    size_t bytesRequiredForCode = bytesRequired(subTagSize);
    m_tagPositionsInBytecode = vector<size_t>(m_usedTags, -1);
    map<size_t, pair<size_t, size_t>> tagRef;
    multimap<h256, unsigned> dataRef;
    multimap<size_t, size_t> subRef;
    vector<unsigned> sizeRef; ///< Pointers to code locations where the size of the program is inserted
    unsigned bytesPerTag = dev::bytesRequired(bytesRequiredForCode);
    byte tagPush = (byte)Instruction::PUSH1 - 1 + bytesPerTag;

    unsigned bytesRequiredIncludingData = bytesRequiredForCode + 1 + m_auxiliaryData.size();
    for (auto const& sub: m_subs)
        bytesRequiredIncludingData += sub->assemble().bytecode.size();

    unsigned bytesPerDataRef = dev::bytesRequired(bytesRequiredIncludingData);
    byte dataRefPush = (byte)Instruction::PUSH1 - 1 + bytesPerDataRef;
    ret.bytecode.reserve(bytesRequiredIncludingData);

    for (AssemblyItem const& i: m_items)
    {
        // store position of the invalid jump destination
        if (i.type() != Tag && m_tagPositionsInBytecode[0] == size_t(-1))
            m_tagPositionsInBytecode[0] = ret.bytecode.size();

        switch (i.type())
        {
        case Operation:
            ret.bytecode.push_back((byte)i.instruction());
            break;
        case PushString:
        {
            ret.bytecode.push_back((byte)Instruction::PUSH32);
            unsigned ii = 0;
            for (auto j: m_strings.at((h256)i.data()))
                if (++ii > 32)
                    break;
                else
                    ret.bytecode.push_back((byte)j);
            while (ii++ < 32)
                ret.bytecode.push_back(0);
            break;
        }
        case Push:
        {
            byte b = max<unsigned>(1, dev::bytesRequired(i.data()));
            ret.bytecode.push_back((byte)Instruction::PUSH1 - 1 + b);
            ret.bytecode.resize(ret.bytecode.size() + b);
            bytesRef byr(&ret.bytecode.back() + 1 - b, b);
            toBigEndian(i.data(), byr);
            break;
        }
        case PushTag:
        {
            ret.bytecode.push_back(tagPush);
            tagRef[ret.bytecode.size()] = i.splitForeignPushTag();
            ret.bytecode.resize(ret.bytecode.size() + bytesPerTag);
            break;
        }
        case PushData:
            ret.bytecode.push_back(dataRefPush);
            dataRef.insert(make_pair((h256)i.data(), ret.bytecode.size()));
            ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
            break;
        case PushSub:
            ret.bytecode.push_back(dataRefPush);
            subRef.insert(make_pair(size_t(i.data()), ret.bytecode.size()));
            ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
            break;
        case PushSubSize:
        {
            auto s = m_subs.at(size_t(i.data()))->assemble().bytecode.size();
            i.setPushedValue(u256(s));
            byte b = max<unsigned>(1, dev::bytesRequired(s));
            ret.bytecode.push_back((byte)Instruction::PUSH1 - 1 + b);
            ret.bytecode.resize(ret.bytecode.size() + b);
            bytesRef byr(&ret.bytecode.back() + 1 - b, b);
            toBigEndian(s, byr);
            break;
        }
        case PushProgramSize:
        {
            ret.bytecode.push_back(dataRefPush);
            sizeRef.push_back(ret.bytecode.size());
            ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
            break;
        }
        case PushLibraryAddress:
            ret.bytecode.push_back(byte(Instruction::PUSH20));
            ret.linkReferences[ret.bytecode.size()] = m_libraries.at(i.data());
            ret.bytecode.resize(ret.bytecode.size() + 20);
            break;
        case Tag:
            assertThrow(i.data() != 0, AssemblyException, "");
            assertThrow(i.splitForeignPushTag().first == size_t(-1), AssemblyException, "Foreign tag.");
            assertThrow(ret.bytecode.size() < 0xffffffffL, AssemblyException, "Tag too large.");
            m_tagPositionsInBytecode[size_t(i.data())] = ret.bytecode.size();
            ret.bytecode.push_back((byte)Instruction::JUMPDEST);
            break;
        default:
            BOOST_THROW_EXCEPTION(InvalidOpcode());
        }
    }

    if (!m_subs.empty() || !m_data.empty() || !m_auxiliaryData.empty())
        // Append a STOP just to be sure.
        ret.bytecode.push_back(0);

    for (size_t i = 0; i < m_subs.size(); ++i)
    {
        auto references = subRef.equal_range(i);
        if (references.first == references.second)
            continue;
        for (auto ref = references.first; ref != references.second; ++ref)
        {
            bytesRef r(ret.bytecode.data() + ref->second, bytesPerDataRef);
            toBigEndian(ret.bytecode.size(), r);
        }
        ret.append(m_subs[i]->assemble());
    }
    for (auto const& i: tagRef)
    {
        size_t subId;
        size_t tagId;
        tie(subId, tagId) = i.second;
        assertThrow(subId == size_t(-1) || subId < m_subs.size(), AssemblyException, "Invalid sub id");
        std::vector<size_t> const& tagPositions =
            subId == size_t(-1) ?
            m_tagPositionsInBytecode :
            m_subs[subId]->m_tagPositionsInBytecode;
        assertThrow(tagId < tagPositions.size(), AssemblyException, "Reference to non-existing tag.");
        size_t pos = tagPositions[tagId];
        assertThrow(pos != size_t(-1), AssemblyException, "Reference to tag without position.");
        assertThrow(dev::bytesRequired(pos) <= bytesPerTag, AssemblyException, "Tag too large for reserved space.");
        bytesRef r(ret.bytecode.data() + i.first, bytesPerTag);
        toBigEndian(pos, r);
    }
    for (auto const& dataItem: m_data)
    {
        auto references = dataRef.equal_range(dataItem.first);
        if (references.first == references.second)
            continue;
        for (auto ref = references.first; ref != references.second; ++ref)
        {
            bytesRef r(ret.bytecode.data() + ref->second, bytesPerDataRef);
            toBigEndian(ret.bytecode.size(), r);
        }
        ret.bytecode += dataItem.second;
    }

    ret.bytecode += m_auxiliaryData;

    for (unsigned pos: sizeRef)
    {
        bytesRef r(ret.bytecode.data() + pos, bytesPerDataRef);
        toBigEndian(ret.bytecode.size(), r);
    }
    return ret;
}