/src/wasm3/source/m3_parse.c

Source
//
//  m3_parse.c
//
//  Created by Steven Massey on 4/19/19.
//  Copyright © 2019 Steven Massey. All rights reserved.
//

#include "m3_env.h"
#include "m3_compile.h"
#include "m3_exception.h"
#include "m3_info.h"


M3Result  ParseType_Table  (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    return result;
}


M3Result  ParseType_Memory  (M3MemoryInfo * o_memory, bytes_t * io_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    u8 flag;

_   (ReadLEB_u7 (& flag, io_bytes, i_end));                   // really a u1
_   (ReadLEB_u32 (& o_memory->initPages, io_bytes, i_end));

    o_memory->maxPages = 0;
    if (flag & (1u << 0))
_       (ReadLEB_u32 (& o_memory->maxPages, io_bytes, i_end));

    o_memory->pageSize = 0;
    if (flag & (1u << 3)) {
        u32 logPageSize;
_       (ReadLEB_u32 (& logPageSize, io_bytes, i_end));
        o_memory->pageSize = 1u << logPageSize;
    }

    _catch: return result;
}


M3Result  ParseSection_Type  (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
{
    IM3FuncType ftype = NULL;

_try {
    u32 numTypes;
_   (ReadLEB_u32 (& numTypes, & i_bytes, i_end));                                   m3log (parse, "** Type [%d]", numTypes);

    _throwif("too many types", numTypes > d_m3MaxSaneTypesCount);

    if (numTypes)
    {
        // table of IM3FuncType (that point to the actual M3FuncType struct in the Environment)
        io_module->funcTypes = m3_AllocArray (IM3FuncType, numTypes);
        _throwifnull (io_module->funcTypes);
        io_module->numFuncTypes = numTypes;

        for (u32 i = 0; i < numTypes; ++i)
        {
            i8 form;
_           (ReadLEB_i7 (& form, & i_bytes, i_end));
            _throwif (m3Err_wasmMalformed, form != -32); // for Wasm MVP

            u32 numArgs;
_           (ReadLEB_u32 (& numArgs, & i_bytes, i_end));

            _throwif (m3Err_tooManyArgsRets, numArgs > d_m3MaxSaneFunctionArgRetCount);
#if defined(M3_COMPILER_MSVC)
            u8 argTypes [d_m3MaxSaneFunctionArgRetCount];
#else
            u8 argTypes[numArgs+1]; // make ubsan happy
#endif
            for (u32 a = 0; a < numArgs; ++a)
            {
                i8 wasmType;
                u8 argType;
_               (ReadLEB_i7 (& wasmType, & i_bytes, i_end));
_               (NormalizeType (& argType, wasmType));

                argTypes[a] = argType;
            }

            u32 numRets;
_           (ReadLEB_u32 (& numRets, & i_bytes, i_end));
            _throwif (m3Err_tooManyArgsRets, (u64)(numRets) + numArgs > d_m3MaxSaneFunctionArgRetCount);

_           (AllocFuncType (& ftype, numRets + numArgs));
            ftype->numArgs = numArgs;
            ftype->numRets = numRets;

            for (u32 r = 0; r < numRets; ++r)
            {
                i8 wasmType;
                u8 retType;
_               (ReadLEB_i7 (& wasmType, & i_bytes, i_end));
_               (NormalizeType (& retType, wasmType));

                ftype->types[r] = retType;
            }
            memcpy (ftype->types + numRets, argTypes, numArgs);                                 m3log (parse, "    type %2d: %s", i, SPrintFuncTypeSignature (ftype));

            Environment_AddFuncType (io_module->environment, & ftype);
            io_module->funcTypes [i] = ftype;
            ftype = NULL; // ownership transferred to environment
        }
    }

} _catch:

    if (result)
    {
        m3_Free (ftype);
        // FIX: M3FuncTypes in the table are leaked
        m3_Free (io_module->funcTypes);
        io_module->numFuncTypes = 0;
    }

    return result;
}


M3Result  ParseSection_Function  (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    u32 numFunctions;
_   (ReadLEB_u32 (& numFunctions, & i_bytes, i_end));                               m3log (parse, "** Function [%d]", numFunctions);

    _throwif("too many functions", numFunctions > d_m3MaxSaneFunctionsCount);

_   (Module_PreallocFunctions(io_module, io_module->numFunctions + numFunctions));

    for (u32 i = 0; i < numFunctions; ++i)
    {
        u32 funcTypeIndex;
_       (ReadLEB_u32 (& funcTypeIndex, & i_bytes, i_end));

_       (Module_AddFunction (io_module, funcTypeIndex, NULL /* import info */));
    }

    _catch: return result;
}


M3Result  ParseSection_Import  (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    M3ImportInfo import = { NULL, NULL }, clearImport = { NULL, NULL };

    u32 numImports;
_   (ReadLEB_u32 (& numImports, & i_bytes, i_end));                                 m3log (parse, "** Import [%d]", numImports);

    _throwif("too many imports", numImports > d_m3MaxSaneImportsCount);

    // Most imports are functions, so we won't waste much space anyway (if any)
_   (Module_PreallocFunctions(io_module, numImports));

    for (u32 i = 0; i < numImports; ++i)
    {
        u8 importKind;

_       (Read_utf8 (& import.moduleUtf8, & i_bytes, i_end));
_       (Read_utf8 (& import.fieldUtf8, & i_bytes, i_end));
_       (Read_u8 (& importKind, & i_bytes, i_end));                                 m3log (parse, "    kind: %d '%s.%s' ",
                                                                                                (u32) importKind, import.moduleUtf8, import.fieldUtf8);
        switch (importKind)
        {
            case d_externalKind_function:
            {
                u32 typeIndex;
_               (ReadLEB_u32 (& typeIndex, & i_bytes, i_end))

_               (Module_AddFunction (io_module, typeIndex, & import))
                import = clearImport;

                io_module->numFuncImports++;
            }
            break;

            case d_externalKind_table:
//                  result = ParseType_Table (& i_bytes, i_end);
                break;

            case d_externalKind_memory:
            {
_               (ParseType_Memory (& io_module->memoryInfo, & i_bytes, i_end));
                io_module->memoryImported = true;
                io_module->memoryImport = import;
                import = clearImport;
            }
            break;

            case d_externalKind_global:
            {
                i8 waType;
                u8 type, isMutable;

_               (ReadLEB_i7 (& waType, & i_bytes, i_end));
_               (NormalizeType (& type, waType));
_               (ReadLEB_u7 (& isMutable, & i_bytes, i_end));                     m3log (parse, "     global: %s mutable=%d", c_waTypes [type], (u32) isMutable);

                IM3Global global;
_               (Module_AddGlobal (io_module, & global, type, isMutable, true /* isImport */));
                global->import = import;
                import = clearImport;
            }
            break;

            default:
                _throw (m3Err_wasmMalformed);
        }

        FreeImportInfo (& import);
    }

    _catch:

    FreeImportInfo (& import);

    return result;
}


M3Result  ParseSection_Export  (IM3Module io_module, bytes_t i_bytes, cbytes_t  i_end)
{
    M3Result result = m3Err_none;
    const char * utf8 = NULL;

    u32 numExports;
_   (ReadLEB_u32 (& numExports, & i_bytes, i_end));                                 m3log (parse, "** Export [%d]", numExports);

    _throwif("too many exports", numExports > d_m3MaxSaneExportsCount);

    for (u32 i = 0; i < numExports; ++i)
    {
        u8 exportKind;
        u32 index;

_       (Read_utf8 (& utf8, & i_bytes, i_end));
_       (Read_u8 (& exportKind, & i_bytes, i_end));
_       (ReadLEB_u32 (& index, & i_bytes, i_end));                                  m3log (parse, "    index: %3d; kind: %d; export: '%s'; ", index, (u32) exportKind, utf8);

        if (exportKind == d_externalKind_function)
        {
            _throwif(m3Err_wasmMalformed, index >= io_module->numFunctions);
            IM3Function func = &(io_module->functions [index]);
            if (func->numNames < d_m3MaxDuplicateFunctionImpl)
            {
                func->names[func->numNames++] = utf8;
                func->export_name = utf8;
                utf8 = NULL; // ownership transferred to M3Function
            }
        }
        else if (exportKind == d_externalKind_global)
        {
            _throwif(m3Err_wasmMalformed, index >= io_module->numGlobals);
            IM3Global global = &(io_module->globals [index]);
            m3_Free (global->name);
            global->name = utf8;
            utf8 = NULL; // ownership transferred to M3Global
        }
        else if (exportKind == d_externalKind_memory)
        {
            m3_Free (io_module->memoryExportName);
            io_module->memoryExportName = utf8;
            utf8 = NULL; // ownership transferred to M3Module
        }
        else if (exportKind == d_externalKind_table)
        {
            m3_Free (io_module->table0ExportName);
            io_module->table0ExportName = utf8;
            utf8 = NULL; // ownership transferred to M3Module
        }

        m3_Free (utf8);
    }

_catch:
    m3_Free (utf8);
    return result;
}


M3Result  ParseSection_Start  (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    u32 startFuncIndex;
_   (ReadLEB_u32 (& startFuncIndex, & i_bytes, i_end));                               m3log (parse, "** Start Function: %d", startFuncIndex);

    if (startFuncIndex < io_module->numFunctions)
    {
        io_module->startFunction = startFuncIndex;
    }
    else result = "start function index out of bounds";

    _catch: return result;
}


M3Result  Parse_InitExpr  (M3Module * io_module, bytes_t * io_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    // this doesn't generate code pages. just walks the wasm bytecode to find the end

#if defined(d_m3PreferStaticAlloc)
    static M3Compilation compilation;
#else
    M3Compilation compilation;
#endif
    compilation = (M3Compilation){ .runtime = NULL, .module = io_module, .wasm = * io_bytes, .wasmEnd = i_end };

    result = CompileBlockStatements (& compilation);

    * io_bytes = compilation.wasm;

    return result;
}


M3Result  ParseSection_Element  (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    u32 numSegments;
_   (ReadLEB_u32 (& numSegments, & i_bytes, i_end));                         m3log (parse, "** Element [%d]", numSegments);

    _throwif ("too many element segments", numSegments > d_m3MaxSaneElementSegments);

    io_module->elementSection = i_bytes;
    io_module->elementSectionEnd = i_end;
    io_module->numElementSegments = numSegments;

    _catch: return result;
}


M3Result  ParseSection_Code  (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result;

    u32 numFunctions;
_   (ReadLEB_u32 (& numFunctions, & i_bytes, i_end));                               m3log (parse, "** Code [%d]", numFunctions);

    if (numFunctions != io_module->numFunctions - io_module->numFuncImports)
    {
        _throw ("mismatched function count in code section");
    }

    for (u32 f = 0; f < numFunctions; ++f)
    {
        const u8 * start = i_bytes;

        u32 size;
_       (ReadLEB_u32 (& size, & i_bytes, i_end));

        if (size)
        {
            const u8 * ptr = i_bytes;
            i_bytes += size;

            if (i_bytes <= i_end)
            {
                /*
                u32 numLocalBlocks;
_               (ReadLEB_u32 (& numLocalBlocks, & ptr, i_end));                                      m3log (parse, "    code size: %-4d", size);

                u32 numLocals = 0;

                for (u32 l = 0; l < numLocalBlocks; ++l)
                {
                    u32 varCount;
                    i8 wasmType;
                    u8 normalType;

_                   (ReadLEB_u32 (& varCount, & ptr, i_end));
_                   (ReadLEB_i7 (& wasmType, & ptr, i_end));
_                   (NormalizeType (& normalType, wasmType));

                    numLocals += varCount;                                                      m3log (parse, "      %2d locals; type: '%s'", varCount, c_waTypes [normalType]);
                }
                 */

                IM3Function func = Module_GetFunction (io_module, f + io_module->numFuncImports);

                func->module = io_module;
                func->wasm = start;
                func->wasmEnd = i_bytes;
                //func->ownsWasmCode = io_module->hasWasmCodeCopy;
//                func->numLocals = numLocals;
            }
            else _throw (m3Err_wasmSectionOverrun);
        }
    }

    _catch:

    if (not result and i_bytes != i_end)
        result = m3Err_wasmSectionUnderrun;

    return result;
}


M3Result  ParseSection_Data  (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    u32 numDataSegments;
_   (ReadLEB_u32 (& numDataSegments, & i_bytes, i_end));                            m3log (parse, "** Data [%d]", numDataSegments);

    _throwif("too many data segments", numDataSegments > d_m3MaxSaneDataSegments);

    io_module->dataSegments = m3_AllocArray (M3DataSegment, numDataSegments);
    _throwifnull(io_module->dataSegments);
    io_module->numDataSegments = numDataSegments;

    for (u32 i = 0; i < numDataSegments; ++i)
    {
        M3DataSegment * segment = & io_module->dataSegments [i];

_       (ReadLEB_u32 (& segment->memoryRegion, & i_bytes, i_end));

        segment->initExpr = i_bytes;
_       (Parse_InitExpr (io_module, & i_bytes, i_end));
        segment->initExprSize = (u32) (i_bytes - segment->initExpr);

        _throwif (m3Err_wasmMissingInitExpr, segment->initExprSize <= 1);

_       (ReadLEB_u32 (& segment->size, & i_bytes, i_end));
        segment->data = i_bytes;                                                    m3log (parse, "    segment [%u]  memory: %u;  expr-size: %d;  size: %d",
                                                                                       i, segment->memoryRegion, segment->initExprSize, segment->size);
        i_bytes += segment->size;

        _throwif("data segment underflow", i_bytes > i_end);
    }

    _catch:

    return result;
}


M3Result  ParseSection_Memory  (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    // TODO: MVP; assert no memory imported

    u32 numMemories;
_   (ReadLEB_u32 (& numMemories, & i_bytes, i_end));                             m3log (parse, "** Memory [%d]", numMemories);

    _throwif (m3Err_tooManyMemorySections, numMemories != 1);

    ParseType_Memory (& io_module->memoryInfo, & i_bytes, i_end);

    _catch: return result;
}


M3Result  ParseSection_Global  (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result = m3Err_none;

    u32 numGlobals;
_   (ReadLEB_u32 (& numGlobals, & i_bytes, i_end));                                 m3log (parse, "** Global [%d]", numGlobals);

    _throwif("too many globals", numGlobals > d_m3MaxSaneGlobalsCount);

    for (u32 i = 0; i < numGlobals; ++i)
    {
        i8 waType;
        u8 type, isMutable;

_       (ReadLEB_i7 (& waType, & i_bytes, i_end));
_       (NormalizeType (& type, waType));
_       (ReadLEB_u7 (& isMutable, & i_bytes, i_end));                                 m3log (parse, "    global: [%d] %s mutable: %d", i, c_waTypes [type],   (u32) isMutable);

        IM3Global global;
_       (Module_AddGlobal (io_module, & global, type, isMutable, false /* isImport */));

        global->initExpr = i_bytes;
_       (Parse_InitExpr (io_module, & i_bytes, i_end));
        global->initExprSize = (u32) (i_bytes - global->initExpr);

        _throwif (m3Err_wasmMissingInitExpr, global->initExprSize <= 1);
    }

    _catch: return result;
}


M3Result  ParseSection_Name  (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result;

    cstr_t name;

    while (i_bytes < i_end)
    {
        u8 nameType;
        u32 payloadLength;

_       (ReadLEB_u7 (& nameType, & i_bytes, i_end));
_       (ReadLEB_u32 (& payloadLength, & i_bytes, i_end));

        bytes_t start = i_bytes;
        if (nameType == 1)
        {
            u32 numNames;
_           (ReadLEB_u32 (& numNames, & i_bytes, i_end));

            _throwif("too many names", numNames > d_m3MaxSaneFunctionsCount);

            for (u32 i = 0; i < numNames; ++i)
            {
                u32 index;
_               (ReadLEB_u32 (& index, & i_bytes, i_end));
_               (Read_utf8 (& name, & i_bytes, i_end));

                if (index < io_module->numFunctions)
                {
                    IM3Function func = &(io_module->functions [index]);
                    if (func->numNames == 0)
                    {
                        func->names[0] = name;        m3log (parse, "    naming function%5d:  %s", index, name);
                        func->numNames = 1;
                        name = NULL; // transfer ownership
                    }
//                          else m3log (parse, "prenamed: %s", io_module->functions [index].name);
                }

                m3_Free (name);
            }
        }

        i_bytes = start + payloadLength;
    }

    _catch: return result;
}


M3Result  ParseSection_Custom  (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
{
    M3Result result;

    cstr_t name;
_   (Read_utf8 (& name, & i_bytes, i_end));
                                                                                    m3log (parse, "** Custom: '%s'", name);
    if (strcmp (name, "name") == 0) {
_       (ParseSection_Name(io_module, i_bytes, i_end));
    } else if (io_module->environment->customSectionHandler) {
_       (io_module->environment->customSectionHandler(io_module, name, i_bytes, i_end));
    }

    m3_Free (name);

    _catch: return result;
}


M3Result  ParseModuleSection  (M3Module * o_module, u8 i_sectionType, bytes_t i_bytes, u32 i_numBytes)
{
    M3Result result = m3Err_none;

    typedef M3Result (* M3Parser) (M3Module *, bytes_t, cbytes_t);

    static M3Parser s_parsers [] =
    {
        ParseSection_Custom,    // 0
        ParseSection_Type,      // 1
        ParseSection_Import,    // 2
        ParseSection_Function,  // 3
        NULL,                   // 4: TODO Table
        ParseSection_Memory,    // 5
        ParseSection_Global,    // 6
        ParseSection_Export,    // 7
        ParseSection_Start,     // 8
        ParseSection_Element,   // 9
        ParseSection_Code,      // 10
        ParseSection_Data,      // 11
        NULL,                   // 12: TODO DataCount
    };

    M3Parser parser = NULL;

    if (i_sectionType <= 12)
        parser = s_parsers [i_sectionType];

    if (parser)
    {
        cbytes_t end = i_bytes + i_numBytes;
        result = parser (o_module, i_bytes, end);
    }
    else
    {
        m3log (parse, " skipped section type: %d", (u32) i_sectionType);
    }

    return result;
}


M3Result  m3_ParseModule  (IM3Environment i_environment, IM3Module * o_module, cbytes_t i_bytes, u32 i_numBytes)
{
    IM3Module module;                                                               m3log (parse, "load module: %d bytes", i_numBytes);
_try {
    module = m3_AllocStruct (M3Module);
    _throwifnull (module);
    module->name = ".unnamed";                                                      m3log (parse, "load module: %d bytes", i_numBytes);
    module->startFunction = -1;
    //module->hasWasmCodeCopy = false;
    module->environment = i_environment;

    const u8 * pos = i_bytes;
    const u8 * end = pos + i_numBytes;

    module->wasmStart = pos;
    module->wasmEnd = end;

    u32 magic, version;
_   (Read_u32 (& magic, & pos, end));
_   (Read_u32 (& version, & pos, end));

    _throwif (m3Err_wasmMalformed, magic != 0x6d736100);
    _throwif (m3Err_incompatibleWasmVersion, version != 1);

    static const u8 sectionsOrder[] = { 1, 2, 3, 4, 5, 6, 7, 8, 9, 12, 10, 11, 0 }; // 0 is a placeholder
    u8 expectedSection = 0;

    while (pos < end)
    {
        u8 section;
_       (ReadLEB_u7 (& section, & pos, end));

        if (section != 0) {
            // Ensure sections appear only once and in order
            while (sectionsOrder[expectedSection++] != section) {
                _throwif(m3Err_misorderedWasmSection, expectedSection >= 12);
            }
        }

        u32 sectionLength;
_       (ReadLEB_u32 (& sectionLength, & pos, end));
        _throwif(m3Err_wasmMalformed, pos + sectionLength > end);

_       (ParseModuleSection (module, section, pos, sectionLength));

        pos += sectionLength;
    }

} _catch:

    if (result)
    {
        m3_FreeModule (module);
        module = NULL;
    }

    * o_module = module;

    return result;
}

Coverage Report

Created: 2026-01-17 06:48

Line	Count	Source
1		//
2		// m3_parse.c
3		//
4		// Created by Steven Massey on 4/19/19.
5		// Copyright © 2019 Steven Massey. All rights reserved.
6		//
7
8		#include "m3_env.h"
9		#include "m3_compile.h"
10		#include "m3_exception.h"
11		#include "m3_info.h"
12
13
14		M3Result ParseType_Table (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
15	0	{
16	0	M3Result result = m3Err_none;
17
18	0	return result;
19	0	}
20
21
22		M3Result ParseType_Memory (M3MemoryInfo * o_memory, bytes_t * io_bytes, cbytes_t i_end)
23	538	{
24	538	M3Result result = m3Err_none;
25
26	538	u8 flag;
27
28	538	_ (ReadLEB_u7 (& flag, io_bytes, i_end)); // really a u1
29	538	_ (ReadLEB_u32 (& o_memory->initPages, io_bytes, i_end));
30
31	538	o_memory->maxPages = 0;
32	538	if (flag & (1u << 0))
33	538	_ (ReadLEB_u32 (& o_memory->maxPages, io_bytes, i_end));
34
35	538	o_memory->pageSize = 0;
36	538	if (flag & (1u << 3)) {
37	43	u32 logPageSize;
38	43	_ (ReadLEB_u32 (& logPageSize, io_bytes, i_end));
39	43	o_memory->pageSize = 1u << logPageSize;
40	43	}
41
42	538	_catch: return result;
43	538	}
44
45
46		M3Result ParseSection_Type (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
47	107	{
48	107	IM3FuncType ftype = NULL;
49
50	107	_try {
51	107	u32 numTypes;
52	107	_ (ReadLEB_u32 (& numTypes, & i_bytes, i_end)); m3log (parse, "** Type [%d]", numTypes);
53
54	107	_throwif("too many types", numTypes > d_m3MaxSaneTypesCount);
55
56	107	if (numTypes)
57	106	{
58		// table of IM3FuncType (that point to the actual M3FuncType struct in the Environment)
59	106	io_module->funcTypes = m3_AllocArray (IM3FuncType, numTypes);
60	106	_throwifnull (io_module->funcTypes);
61	106	io_module->numFuncTypes = numTypes;
62
63	236	for (u32 i = 0; i < numTypes; ++i)
64	157	{
65	157	i8 form;
66	157	_ (ReadLEB_i7 (& form, & i_bytes, i_end));
67	157	_throwif (m3Err_wasmMalformed, form != -32); // for Wasm MVP
68
69	132	u32 numArgs;
70	132	_ (ReadLEB_u32 (& numArgs, & i_bytes, i_end));
71
72	132	_throwif (m3Err_tooManyArgsRets, numArgs > d_m3MaxSaneFunctionArgRetCount);
73		#if defined(M3_COMPILER_MSVC)
74		u8 argTypes [d_m3MaxSaneFunctionArgRetCount];
75		#else
76	132	u8 argTypes[numArgs+1]; // make ubsan happy
77	132	#endif
78	6.69k	for (u32 a = 0; a < numArgs; ++a)
79	6.56k	{
80	6.56k	i8 wasmType;
81	6.56k	u8 argType;
82	6.56k	_ (ReadLEB_i7 (& wasmType, & i_bytes, i_end));
83	6.56k	_ (NormalizeType (& argType, wasmType));
84
85	6.56k	argTypes[a] = argType;
86	6.56k	}
87
88	131	u32 numRets;
89	131	_ (ReadLEB_u32 (& numRets, & i_bytes, i_end));
90	131	_throwif (m3Err_tooManyArgsRets, (u64)(numRets) + numArgs > d_m3MaxSaneFunctionArgRetCount);
91
92	131	_ (AllocFuncType (& ftype, numRets + numArgs));
93	131	ftype->numArgs = numArgs;
94	131	ftype->numRets = numRets;
95
96	5.47k	for (u32 r = 0; r < numRets; ++r)
97	5.34k	{
98	5.34k	i8 wasmType;
99	5.34k	u8 retType;
100	5.34k	_ (ReadLEB_i7 (& wasmType, & i_bytes, i_end));
101	5.34k	_ (NormalizeType (& retType, wasmType));
102
103	5.34k	ftype->types[r] = retType;
104	5.34k	}
105	130	memcpy (ftype->types + numRets, argTypes, numArgs); m3log (parse, " type %2d: %s", i, SPrintFuncTypeSignature (ftype));
106
107	130	Environment_AddFuncType (io_module->environment, & ftype);
108	130	io_module->funcTypes [i] = ftype;
109	130	ftype = NULL; // ownership transferred to environment
110	130	}
111	106	}
112
113	107	} _catch:
114
115	107	if (result)
116	27	{
117	27	m3_Free (ftype);
118		// FIX: M3FuncTypes in the table are leaked
119	27	m3_Free (io_module->funcTypes);
120	27	io_module->numFuncTypes = 0;
121	27	}
122
123	107	return result;
124	107	}
125
126
127		M3Result ParseSection_Function (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
128	72	{
129	72	M3Result result = m3Err_none;
130
131	72	u32 numFunctions;
132	72	_ (ReadLEB_u32 (& numFunctions, & i_bytes, i_end)); m3log (parse, "** Function [%d]", numFunctions);
133
134	72	_throwif("too many functions", numFunctions > d_m3MaxSaneFunctionsCount);
135
136	72	_ (Module_PreallocFunctions(io_module, io_module->numFunctions + numFunctions));
137
138	1.51k	for (u32 i = 0; i < numFunctions; ++i)
139	1.44k	{
140	1.44k	u32 funcTypeIndex;
141	1.44k	_ (ReadLEB_u32 (& funcTypeIndex, & i_bytes, i_end));
142
143	1.44k	_ (Module_AddFunction (io_module, funcTypeIndex, NULL /* import info */));
144	1.44k	}
145
146	72	_catch: return result;
147	72	}
148
149
150		M3Result ParseSection_Import (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
151	16	{
152	16	M3Result result = m3Err_none;
153
154	16	M3ImportInfo import = { NULL, NULL }, clearImport = { NULL, NULL };
155
156	16	u32 numImports;
157	16	_ (ReadLEB_u32 (& numImports, & i_bytes, i_end)); m3log (parse, "** Import [%d]", numImports);
158
159	16	_throwif("too many imports", numImports > d_m3MaxSaneImportsCount);
160
161		// Most imports are functions, so we won't waste much space anyway (if any)
162	16	_ (Module_PreallocFunctions(io_module, numImports));
163
164	3.50k	for (u32 i = 0; i < numImports; ++i)
165	3.50k	{
166	3.50k	u8 importKind;
167
168	3.50k	_ (Read_utf8 (& import.moduleUtf8, & i_bytes, i_end));
169	3.49k	_ (Read_utf8 (& import.fieldUtf8, & i_bytes, i_end));
170	3.49k	_ (Read_u8 (& importKind, & i_bytes, i_end)); m3log (parse, " kind: %d '%s.%s' ",
171	3.49k	(u32) importKind, import.moduleUtf8, import.fieldUtf8);
172	3.49k	switch (importKind)
173	3.49k	{
174	2.42k	case d_externalKind_function:
175	2.42k	{
176	2.42k	u32 typeIndex;
177	2.42k	_ (ReadLEB_u32 (& typeIndex, & i_bytes, i_end))
178
179	2.42k	_ (Module_AddFunction (io_module, typeIndex, & import))
180	2.41k	import = clearImport;
181
182	2.41k	io_module->numFuncImports++;
183	2.41k	}
184	0	break;
185
186	418	case d_externalKind_table:
187		// result = ParseType_Table (& i_bytes, i_end);
188	418	break;
189
190	538	case d_externalKind_memory:
191	538	{
192	538	_ (ParseType_Memory (& io_module->memoryInfo, & i_bytes, i_end));
193	538	io_module->memoryImported = true;
194	538	io_module->memoryImport = import;
195	538	import = clearImport;
196	538	}
197	0	break;
198
199	119	case d_externalKind_global:
200	119	{
201	119	i8 waType;
202	119	u8 type, isMutable;
203
204	119	_ (ReadLEB_i7 (& waType, & i_bytes, i_end));
205	119	_ (NormalizeType (& type, waType));
206	119	_ (ReadLEB_u7 (& isMutable, & i_bytes, i_end)); m3log (parse, " global: %s mutable=%d", c_waTypes [type], (u32) isMutable);
207
208	119	IM3Global global;
209	119	_ (Module_AddGlobal (io_module, & global, type, isMutable, true /* isImport */));
210	119	global->import = import;
211	119	import = clearImport;
212	119	}
213	0	break;
214
215	2	default:
216	2	_throw (m3Err_wasmMalformed);
217	3.49k	}
218
219	3.49k	FreeImportInfo (& import);
220	3.49k	}
221
222	16	_catch:
223
224	16	FreeImportInfo (& import);
225
226	16	return result;
227	16	}
228
229
230		M3Result ParseSection_Export (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
231	19	{
232	19	M3Result result = m3Err_none;
233	19	const char * utf8 = NULL;
234
235	19	u32 numExports;
236	19	_ (ReadLEB_u32 (& numExports, & i_bytes, i_end)); m3log (parse, "** Export [%d]", numExports);
237
238	19	_throwif("too many exports", numExports > d_m3MaxSaneExportsCount);
239
240	2.21k	for (u32 i = 0; i < numExports; ++i)
241	2.20k	{
242	2.20k	u8 exportKind;
243	2.20k	u32 index;
244
245	2.20k	_ (Read_utf8 (& utf8, & i_bytes, i_end));
246	2.20k	_ (Read_u8 (& exportKind, & i_bytes, i_end));
247	2.20k	_ (ReadLEB_u32 (& index, & i_bytes, i_end)); m3log (parse, " index: %3d; kind: %d; export: '%s'; ", index, (u32) exportKind, utf8);
248
249	2.19k	if (exportKind == d_externalKind_function)
250	503	{
251	503	_throwif(m3Err_wasmMalformed, index >= io_module->numFunctions);
252	502	IM3Function func = &(io_module->functions [index]);
253	502	if (func->numNames < d_m3MaxDuplicateFunctionImpl)
254	135	{
255	135	func->names[func->numNames++] = utf8;
256	135	func->export_name = utf8;
257	135	utf8 = NULL; // ownership transferred to M3Function
258	135	}
259	502	}
260	1.69k	else if (exportKind == d_externalKind_global)
261	0	{
262	0	_throwif(m3Err_wasmMalformed, index >= io_module->numGlobals);
263	0	IM3Global global = &(io_module->globals [index]);
264	0	m3_Free (global->name);
265	0	global->name = utf8;
266	0	utf8 = NULL; // ownership transferred to M3Global
267	0	}
268	1.69k	else if (exportKind == d_externalKind_memory)
269	508	{
270	508	m3_Free (io_module->memoryExportName);
271	508	io_module->memoryExportName = utf8;
272	508	utf8 = NULL; // ownership transferred to M3Module
273	508	}
274	1.18k	else if (exportKind == d_externalKind_table)
275	446	{
276	446	m3_Free (io_module->table0ExportName);
277	446	io_module->table0ExportName = utf8;
278	446	utf8 = NULL; // ownership transferred to M3Module
279	446	}
280
281	2.19k	m3_Free (utf8);
282	2.19k	}
283
284	19	_catch:
285	19	m3_Free (utf8);
286	19	return result;
287	19	}
288
289
290		M3Result ParseSection_Start (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
291	1	{
292	1	M3Result result = m3Err_none;
293
294	1	u32 startFuncIndex;
295	1	_ (ReadLEB_u32 (& startFuncIndex, & i_bytes, i_end)); m3log (parse, "** Start Function: %d", startFuncIndex);
296
297	1	if (startFuncIndex < io_module->numFunctions)
298	1	{
299	1	io_module->startFunction = startFuncIndex;
300	1	}
301	0	else result = "start function index out of bounds";
302
303	1	_catch: return result;
304	1	}
305
306
307		M3Result Parse_InitExpr (M3Module * io_module, bytes_t * io_bytes, cbytes_t i_end)
308	291	{
309	291	M3Result result = m3Err_none;
310
311		// this doesn't generate code pages. just walks the wasm bytecode to find the end
312
313		#if defined(d_m3PreferStaticAlloc)
314		static M3Compilation compilation;
315		#else
316	291	M3Compilation compilation;
317	291	#endif
318	291	compilation = (M3Compilation){ .runtime = NULL, .module = io_module, .wasm = * io_bytes, .wasmEnd = i_end };
319
320	291	result = CompileBlockStatements (& compilation);
321
322	291	* io_bytes = compilation.wasm;
323
324	291	return result;
325	291	}
326
327
328		M3Result ParseSection_Element (IM3Module io_module, bytes_t i_bytes, cbytes_t i_end)
329	3	{
330	3	M3Result result = m3Err_none;
331
332	3	u32 numSegments;
333	3	_ (ReadLEB_u32 (& numSegments, & i_bytes, i_end)); m3log (parse, "** Element [%d]", numSegments);
334
335	3	_throwif ("too many element segments", numSegments > d_m3MaxSaneElementSegments);
336
337	3	io_module->elementSection = i_bytes;
338	3	io_module->elementSectionEnd = i_end;
339	3	io_module->numElementSegments = numSegments;
340
341	3	_catch: return result;
342	3	}
343
344
345		M3Result ParseSection_Code (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
346	48	{
347	48	M3Result result;
348
349	48	u32 numFunctions;
350	48	_ (ReadLEB_u32 (& numFunctions, & i_bytes, i_end)); m3log (parse, "** Code [%d]", numFunctions);
351
352	48	if (numFunctions != io_module->numFunctions - io_module->numFuncImports)
353	7	{
354	7	_throw ("mismatched function count in code section");
355	0	}
356
357	197	for (u32 f = 0; f < numFunctions; ++f)
358	157	{
359	157	const u8 * start = i_bytes;
360
361	157	u32 size;
362	157	_ (ReadLEB_u32 (& size, & i_bytes, i_end));
363
364	157	if (size)
365	41	{
366	41	const u8 * ptr = i_bytes;
367	41	i_bytes += size;
368
369	41	if (i_bytes <= i_end)
370	40	{
371		/*
372		u32 numLocalBlocks;
373		_ (ReadLEB_u32 (& numLocalBlocks, & ptr, i_end)); m3log (parse, " code size: %-4d", size);
374
375		u32 numLocals = 0;
376
377		for (u32 l = 0; l < numLocalBlocks; ++l)
378		{
379		u32 varCount;
380		i8 wasmType;
381		u8 normalType;
382
383		_ (ReadLEB_u32 (& varCount, & ptr, i_end));
384		_ (ReadLEB_i7 (& wasmType, & ptr, i_end));
385		_ (NormalizeType (& normalType, wasmType));
386
387		numLocals += varCount; m3log (parse, " %2d locals; type: '%s'", varCount, c_waTypes [normalType]);
388		}
389		*/
390
391	40	IM3Function func = Module_GetFunction (io_module, f + io_module->numFuncImports);
392
393	40	func->module = io_module;
394	40	func->wasm = start;
395	40	func->wasmEnd = i_bytes;
396		//func->ownsWasmCode = io_module->hasWasmCodeCopy;
397		// func->numLocals = numLocals;
398	40	}
399	40	else _throw (m3Err_wasmSectionOverrun);
400	40	}
401	157	}
402
403	48	_catch:
404
405	48	if (not result and i_bytes != i_end)
406	0	result = m3Err_wasmSectionUnderrun;
407
408	48	return result;
409	41	}
410
411
412		M3Result ParseSection_Data (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
413	40	{
414	40	M3Result result = m3Err_none;
415
416	40	u32 numDataSegments;
417	40	_ (ReadLEB_u32 (& numDataSegments, & i_bytes, i_end)); m3log (parse, "** Data [%d]", numDataSegments);
418
419	40	_throwif("too many data segments", numDataSegments > d_m3MaxSaneDataSegments);
420
421	40	io_module->dataSegments = m3_AllocArray (M3DataSegment, numDataSegments);
422	40	_throwifnull(io_module->dataSegments);
423	40	io_module->numDataSegments = numDataSegments;
424
425	126	for (u32 i = 0; i < numDataSegments; ++i)
426	98	{
427	98	M3DataSegment * segment = & io_module->dataSegments [i];
428
429	98	_ (ReadLEB_u32 (& segment->memoryRegion, & i_bytes, i_end));
430
431	98	segment->initExpr = i_bytes;
432	98	_ (Parse_InitExpr (io_module, & i_bytes, i_end));
433	86	segment->initExprSize = (u32) (i_bytes - segment->initExpr);
434
435	86	_throwif (m3Err_wasmMissingInitExpr, segment->initExprSize <= 1);
436
437	86	_ (ReadLEB_u32 (& segment->size, & i_bytes, i_end));
438	86	segment->data = i_bytes; m3log (parse, " segment [%u] memory: %u; expr-size: %d; size: %d",
439	86	i, segment->memoryRegion, segment->initExprSize, segment->size);
440	86	i_bytes += segment->size;
441
442	86	_throwif("data segment underflow", i_bytes > i_end);
443	86	}
444
445	40	_catch:
446
447	40	return result;
448	40	}
449
450
451		M3Result ParseSection_Memory (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
452	0	{
453	0	M3Result result = m3Err_none;
454
455		// TODO: MVP; assert no memory imported
456
457	0	u32 numMemories;
458	0	_ (ReadLEB_u32 (& numMemories, & i_bytes, i_end)); m3log (parse, "** Memory [%d]", numMemories);
459
460	0	_throwif (m3Err_tooManyMemorySections, numMemories != 1);
461
462	0	ParseType_Memory (& io_module->memoryInfo, & i_bytes, i_end);
463
464	0	_catch: return result;
465	0	}
466
467
468		M3Result ParseSection_Global (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
469	14	{
470	14	M3Result result = m3Err_none;
471
472	14	u32 numGlobals;
473	14	_ (ReadLEB_u32 (& numGlobals, & i_bytes, i_end)); m3log (parse, "** Global [%d]", numGlobals);
474
475	14	_throwif("too many globals", numGlobals > d_m3MaxSaneGlobalsCount);
476
477	204	for (u32 i = 0; i < numGlobals; ++i)
478	194	{
479	194	i8 waType;
480	194	u8 type, isMutable;
481
482	194	_ (ReadLEB_i7 (& waType, & i_bytes, i_end));
483	194	_ (NormalizeType (& type, waType));
484	193	_ (ReadLEB_u7 (& isMutable, & i_bytes, i_end)); m3log (parse, " global: [%d] %s mutable: %d", i, c_waTypes [type], (u32) isMutable);
485
486	193	IM3Global global;
487	193	_ (Module_AddGlobal (io_module, & global, type, isMutable, false /* isImport */));
488
489	193	global->initExpr = i_bytes;
490	193	_ (Parse_InitExpr (io_module, & i_bytes, i_end));
491	190	global->initExprSize = (u32) (i_bytes - global->initExpr);
492
493	190	_throwif (m3Err_wasmMissingInitExpr, global->initExprSize <= 1);
494	190	}
495
496	14	_catch: return result;
497	14	}
498
499
500		M3Result ParseSection_Name (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
501	377	{
502	377	M3Result result;
503
504	377	cstr_t name;
505
506	972	while (i_bytes < i_end)
507	598	{
508	598	u8 nameType;
509	598	u32 payloadLength;
510
511	598	_ (ReadLEB_u7 (& nameType, & i_bytes, i_end));
512	598	_ (ReadLEB_u32 (& payloadLength, & i_bytes, i_end));
513
514	597	bytes_t start = i_bytes;
515	597	if (nameType == 1)
516	216	{
517	216	u32 numNames;
518	216	_ (ReadLEB_u32 (& numNames, & i_bytes, i_end));
519
520	216	_throwif("too many names", numNames > d_m3MaxSaneFunctionsCount);
521
522	972	for (u32 i = 0; i < numNames; ++i)
523	758	{
524	758	u32 index;
525	758	_ (ReadLEB_u32 (& index, & i_bytes, i_end));
526	758	_ (Read_utf8 (& name, & i_bytes, i_end));
527
528	757	if (index < io_module->numFunctions)
529	723	{
530	723	IM3Function func = &(io_module->functions [index]);
531	723	if (func->numNames == 0)
532	115	{
533	115	func->names[0] = name; m3log (parse, " naming function%5d: %s", index, name);
534	115	func->numNames = 1;
535	115	name = NULL; // transfer ownership
536	115	}
537		// else m3log (parse, "prenamed: %s", io_module->functions [index].name);
538	723	}
539
540	757	m3_Free (name);
541	757	}
542	215	}
543
544	595	i_bytes = start + payloadLength;
545	595	}
546
547	377	_catch: return result;
548	377	}
549
550
551		M3Result ParseSection_Custom (M3Module * io_module, bytes_t i_bytes, cbytes_t i_end)
552	971	{
553	971	M3Result result;
554
555	971	cstr_t name;
556	971	_ (Read_utf8 (& name, & i_bytes, i_end));
557	969	m3log (parse, "** Custom: '%s'", name);
558	969	if (strcmp (name, "name") == 0) {
559	377	_ (ParseSection_Name(io_module, i_bytes, i_end));
560	592	} else if (io_module->environment->customSectionHandler) {
561	0	_ (io_module->environment->customSectionHandler(io_module, name, i_bytes, i_end));
562	0	}
563
564	966	m3_Free (name);
565
566	971	_catch: return result;
567	966	}
568
569
570		M3Result ParseModuleSection (M3Module * o_module, u8 i_sectionType, bytes_t i_bytes, u32 i_numBytes)
571	1.29k	{
572	1.29k	M3Result result = m3Err_none;
573
574	1.29k	typedef M3Result (* M3Parser) (M3Module *, bytes_t, cbytes_t);
575
576	1.29k	static M3Parser s_parsers [] =
577	1.29k	{
578	1.29k	ParseSection_Custom, // 0
579	1.29k	ParseSection_Type, // 1
580	1.29k	ParseSection_Import, // 2
581	1.29k	ParseSection_Function, // 3
582	1.29k	NULL, // 4: TODO Table
583	1.29k	ParseSection_Memory, // 5
584	1.29k	ParseSection_Global, // 6
585	1.29k	ParseSection_Export, // 7
586	1.29k	ParseSection_Start, // 8
587	1.29k	ParseSection_Element, // 9
588	1.29k	ParseSection_Code, // 10
589	1.29k	ParseSection_Data, // 11
590	1.29k	NULL, // 12: TODO DataCount
591	1.29k	};
592
593	1.29k	M3Parser parser = NULL;
594
595	1.29k	if (i_sectionType <= 12)
596	1.29k	parser = s_parsers [i_sectionType];
597
598	1.29k	if (parser)
599	1.29k	{
600	1.29k	cbytes_t end = i_bytes + i_numBytes;
601	1.29k	result = parser (o_module, i_bytes, end);
602	1.29k	}
603	1	else
604	1	{
605	1	m3log (parse, " skipped section type: %d", (u32) i_sectionType);
606	1	}
607
608	1.29k	return result;
609	1.29k	}
610
611
612		M3Result m3_ParseModule (IM3Environment i_environment, IM3Module * o_module, cbytes_t i_bytes, u32 i_numBytes)
613	184	{
614	184	IM3Module module; m3log (parse, "load module: %d bytes", i_numBytes);
615	184	_try {
616	184	module = m3_AllocStruct (M3Module);
617	184	_throwifnull (module);
618	184	module->name = ".unnamed"; m3log (parse, "load module: %d bytes", i_numBytes);
619	184	module->startFunction = -1;
620		//module->hasWasmCodeCopy = false;
621	184	module->environment = i_environment;
622
623	184	const u8 * pos = i_bytes;
624	184	const u8 * end = pos + i_numBytes;
625
626	184	module->wasmStart = pos;
627	184	module->wasmEnd = end;
628
629	184	u32 magic, version;
630	184	_ (Read_u32 (& magic, & pos, end));
631	184	_ (Read_u32 (& version, & pos, end));
632
633	184	_throwif (m3Err_wasmMalformed, magic != 0x6d736100);
634	182	_throwif (m3Err_incompatibleWasmVersion, version != 1);
635
636	180	static const u8 sectionsOrder[] = { 1, 2, 3, 4, 5, 6, 7, 8, 9, 12, 10, 11, 0 }; // 0 is a placeholder
637	180	u8 expectedSection = 0;
638
639	1.39k	while (pos < end)
640	1.30k	{
641	1.30k	u8 section;
642	1.30k	_ (ReadLEB_u7 (& section, & pos, end));
643
644	1.30k	if (section != 0) {
645		// Ensure sections appear only once and in order
646	1.33k	while (sectionsOrder[expectedSection++] != section) {
647	1.01k	_throwif(m3Err_misorderedWasmSection, expectedSection >= 12);
648	1.00k	}
649	327	}
650
651	1.29k	u32 sectionLength;
652	1.29k	_ (ReadLEB_u32 (& sectionLength, & pos, end));
653	1.29k	_throwif(m3Err_wasmMalformed, pos + sectionLength > end);
654
655	1.29k	_ (ParseModuleSection (module, section, pos, sectionLength));
656
657	1.21k	pos += sectionLength;
658	1.21k	}
659
660	184	} _catch:
661
662	184	if (result)
663	88	{
664	88	m3_FreeModule (module);
665	88	module = NULL;
666	88	}
667
668	184	* o_module = module;
669
670	184	return result;
671	180	}