// === Auto-generated preamble library stuff === //======================================== // Runtime code shared with compiler //======================================== {{RUNTIME}} #if ASM_JS #if RESERVED_FUNCTION_POINTERS function jsCall() { var args = Array.prototype.slice.call(arguments); return Runtime.functionPointers[args[0]].apply(null, args.slice(1)); } #endif #endif #if BENCHMARK Module.realPrint = Module.print; Module.print = Module.printErr = function(){}; #endif #if SAFE_HEAP //======================================== // Debugging tools - Heap //======================================== var HEAP_WATCHED = []; var HEAP_HISTORY = []; function SAFE_HEAP_CLEAR(dest) { #if SAFE_HEAP_LOG Module.print('SAFE_HEAP clear: ' + dest); #endif HEAP_HISTORY[dest] = undefined; } var SAFE_HEAP_ERRORS = 0; var ACCEPTABLE_SAFE_HEAP_ERRORS = 0; function SAFE_HEAP_ACCESS(dest, type, store, ignore, storeValue) { //if (dest === A_NUMBER) Module.print ([dest, type, store, ignore, storeValue] + ' ' + new Error().stack); // Something like this may be useful, in debugging assert(dest > 0, 'segmentation fault'); #if USE_TYPED_ARRAYS // When using typed arrays, reads over the top of TOTAL_MEMORY will fail silently, so we must // correct that by growing TOTAL_MEMORY as needed. Without typed arrays, memory is a normal // JS array so it will work (potentially slowly, depending on the engine). assert(ignore || dest < Math.max(DYNAMICTOP, STATICTOP)); assert(ignore || DYNAMICTOP <= TOTAL_MEMORY); #endif #if USE_TYPED_ARRAYS == 2 return; // It is legitimate to violate the load-store assumption in this case #endif if (type && type.charAt(type.length-1) == '*') type = 'i32'; // pointers are ints, for our purposes here // Note that this will pass even with unions: You can store X, load X, then store Y and load Y. // You cannot, however, do the nonportable act of store X and load Y! if (store) { HEAP_HISTORY[dest] = ignore ? null : type; } else { #if USE_TYPED_ARRAYS == 0 if (!HEAP[dest] && HEAP[dest] !== 0 && HEAP[dest] !== false && !ignore) { // false can be the result of a mathop comparator var error = true; try { if (HEAP[dest].toString() === 'NaN') error = false; // NaN is acceptable, as a double value } catch(e){} if (error) throw('Warning: Reading an invalid value at ' + dest + ' :: ' + new Error().stack + '\n'); } #endif if (type === null) return; var history = HEAP_HISTORY[dest]; if (history === null) return; if (!ignore) assert(history, 'Must have a history for a safe heap load! ' + dest + ':' + type); // Warning - bit fields in C structs cause loads+stores for each store, so // they will show up here... // assert((history && history[0]) /* || HEAP[dest] === 0 */, "Loading from where there was no store! " + dest + ',' + HEAP[dest] + ',' + type + ', \n\n' + new Error().stack + '\n'); // if (history[0].type !== type) { if (history !== type && !ignore) { Module.print('Load-store consistency assumption failure! ' + dest); Module.print('\n'); Module.print(JSON.stringify(history)); Module.print('\n'); Module.print('LOAD: ' + type + ', ' + new Error().stack); Module.print('\n'); SAFE_HEAP_ERRORS++; assert(SAFE_HEAP_ERRORS <= ACCEPTABLE_SAFE_HEAP_ERRORS, 'Load-store consistency assumption failure!'); } } } function SAFE_HEAP_STORE(dest, value, type, ignore) { #if SAFE_HEAP_LOG Module.print('SAFE_HEAP store: ' + [dest, type, value, ignore]); #endif if (!ignore && !value && (value === null || value === undefined)) { throw('Warning: Writing an invalid value of ' + JSON.stringify(value) + ' at ' + dest + ' :: ' + new Error().stack + '\n'); } //if (!ignore && (value === Infinity || value === -Infinity || isNaN(value))) throw [value, typeof value, new Error().stack]; SAFE_HEAP_ACCESS(dest, type, true, ignore, value); if (dest in HEAP_WATCHED) { Module.print((new Error()).stack); throw "Bad store!" + dest; } #if USE_TYPED_ARRAYS == 2 // Check alignment switch(type) { case 'i16': assert(dest % 2 == 0); break; case 'i32': assert(dest % 4 == 0); break; case 'i64': assert(dest % 8 == 0); break; case 'float': assert(dest % 4 == 0); break; #if DOUBLE_MODE == 1 case 'double': assert(dest % 4 == 0); break; #else case 'double': assert(dest % 4 == 0); break; #endif } #endif setValue(dest, value, type, 1); } function SAFE_HEAP_LOAD(dest, type, unsigned, ignore) { SAFE_HEAP_ACCESS(dest, type, false, ignore); #if SAFE_HEAP_LOG Module.print('SAFE_HEAP load: ' + [dest, type, getValue(dest, type, 1), ignore]); #endif #if USE_TYPED_ARRAYS == 2 // Check alignment switch(type) { case 'i16': assert(dest % 2 == 0); break; case 'i32': assert(dest % 4 == 0); break; case 'i64': assert(dest % 8 == 0); break; case 'float': assert(dest % 4 == 0); break; #if DOUBLE_MODE == 1 case 'double': assert(dest % 4 == 0); break; #else case 'double': assert(dest % 4 == 0); break; #endif } #endif var ret = getValue(dest, type, 1); if (unsigned) ret = unSign(ret, parseInt(type.substr(1)), 1); return ret; } function SAFE_HEAP_COPY_HISTORY(dest, src) { #if SAFE_HEAP_LOG Module.print('SAFE_HEAP copy: ' + [dest, src]); #endif HEAP_HISTORY[dest] = HEAP_HISTORY[src]; SAFE_HEAP_ACCESS(dest, HEAP_HISTORY[dest] || null, true, false); } //========================================== #endif #if CHECK_HEAP_ALIGN //======================================== // Debugging tools - alignment check //======================================== function CHECK_ALIGN_8(addr) { assert((addr & 7) == 0, "address must be 8-byte aligned, is " + addr + "!"); return addr; } function CHECK_ALIGN_4(addr) { assert((addr & 3) == 0, "address must be 4-byte aligned, is " + addr + "!"); return addr; } function CHECK_ALIGN_2(addr) { assert((addr & 1) == 0, "address must be 2-byte aligned!"); return addr; } #endif #if CHECK_OVERFLOWS //======================================== // Debugging tools - Mathop overflows //======================================== function CHECK_OVERFLOW(value, bits, ignore, sig) { if (ignore) return value; var twopbits = Math.pow(2, bits); var twopbits1 = Math.pow(2, bits-1); // For signedness issue here, see settings.js, CHECK_SIGNED_OVERFLOWS #if CHECK_SIGNED_OVERFLOWS if (value === Infinity || value === -Infinity || value >= twopbits1 || value < -twopbits1) { throw 'SignedOverflow'; if (value === Infinity || value === -Infinity || Math.abs(value) >= twopbits) throw 'Overflow'; } #else if (value === Infinity || value === -Infinity || Math.abs(value) >= twopbits) { throw 'Overflow'; } #endif #if CORRECT_OVERFLOWS // Fail on >32 bits - we warned at compile time if (bits <= 32) { value = value & (twopbits - 1); } #endif return value; } #endif #if LABEL_DEBUG //======================================== // Debugging tools - Code flow progress //======================================== var INDENT = ''; #endif #if EXECUTION_TIMEOUT //======================================== // Debugging tools - Execution timeout //======================================== var START_TIME = Date.now(); #endif //======================================== // Runtime essentials //======================================== var __THREW__ = 0; // Used in checking for thrown exceptions. #if ASM_JS == 0 var setjmpId = 1; // Used in setjmp/longjmp var setjmpLabels = {}; #endif var ABORT = false; // whether we are quitting the application. no code should run after this. set in exit() and abort() var undef = 0; // tempInt is used for 32-bit signed values or smaller. tempBigInt is used // for 32-bit unsigned values or more than 32 bits. TODO: audit all uses of tempInt var tempValue, tempInt, tempBigInt, tempInt2, tempBigInt2, tempPair, tempBigIntI, tempBigIntR, tempBigIntS, tempBigIntP, tempBigIntD; #if USE_TYPED_ARRAYS == 2 var tempI64, tempI64b; var tempRet0, tempRet1, tempRet2, tempRet3, tempRet4, tempRet5, tempRet6, tempRet7, tempRet8, tempRet9; #endif function abort(text) { Module.print(text + ':\n' + (new Error).stack); ABORT = true; throw "Assertion: " + text; } function assert(condition, text) { if (!condition) { abort('Assertion failed: ' + text); } } var globalScope = this; // C calling interface. A convenient way to call C functions (in C files, or // defined with extern "C"). // // Note: LLVM optimizations can inline and remove functions, after which you will not be // able to call them. Closure can also do so. To avoid that, add your function to // the exports using something like // // -s EXPORTED_FUNCTIONS='["_main", "_myfunc"]' // // @param ident The name of the C function (note that C++ functions will be name-mangled - use extern "C") // @param returnType The return type of the function, one of the JS types 'number', 'string' or 'array' (use 'number' for any C pointer, and // 'array' for JavaScript arrays and typed arrays). // @param argTypes An array of the types of arguments for the function (if there are no arguments, this can be ommitted). Types are as in returnType, // except that 'array' is not possible (there is no way for us to know the length of the array) // @param args An array of the arguments to the function, as native JS values (as in returnType) // Note that string arguments will be stored on the stack (the JS string will become a C string on the stack). // @return The return value, as a native JS value (as in returnType) function ccall(ident, returnType, argTypes, args) { return ccallFunc(getCFunc(ident), returnType, argTypes, args); } Module["ccall"] = ccall; // Returns the C function with a specified identifier (for C++, you need to do manual name mangling) function getCFunc(ident) { try { var func = globalScope['Module']['_' + ident]; // closure exported function if (!func) func = eval('_' + ident); // explicit lookup } catch(e) { } assert(func, 'Cannot call unknown function ' + ident + ' (perhaps LLVM optimizations or closure removed it?)'); return func; } // Internal function that does a C call using a function, not an identifier function ccallFunc(func, returnType, argTypes, args) { var stack = 0; function toC(value, type) { if (type == 'string') { if (value === null || value === undefined || value === 0) return 0; // null string if (!stack) stack = Runtime.stackSave(); var ret = Runtime.stackAlloc(value.length+1); writeStringToMemory(value, ret); return ret; } else if (type == 'array') { if (!stack) stack = Runtime.stackSave(); var ret = Runtime.stackAlloc(value.length); writeArrayToMemory(value, ret); return ret; } return value; } function fromC(value, type) { if (type == 'string') { return Pointer_stringify(value); } assert(type != 'array'); return value; } var i = 0; var cArgs = args ? args.map(function(arg) { return toC(arg, argTypes[i++]); }) : []; var ret = fromC(func.apply(null, cArgs), returnType); if (stack) Runtime.stackRestore(stack); return ret; } // Returns a native JS wrapper for a C function. This is similar to ccall, but // returns a function you can call repeatedly in a normal way. For example: // // var my_function = cwrap('my_c_function', 'number', ['number', 'number']); // alert(my_function(5, 22)); // alert(my_function(99, 12)); // function cwrap(ident, returnType, argTypes) { var func = getCFunc(ident); return function() { return ccallFunc(func, returnType, argTypes, Array.prototype.slice.call(arguments)); } } Module["cwrap"] = cwrap; // Sets a value in memory in a dynamic way at run-time. Uses the // type data. This is the same as makeSetValue, except that // makeSetValue is done at compile-time and generates the needed // code then, whereas this function picks the right code at // run-time. // Note that setValue and getValue only do *aligned* writes and reads! // Note that ccall uses JS types as for defining types, while setValue and // getValue need LLVM types ('i8', 'i32') - this is a lower-level operation function setValue(ptr, value, type, noSafe) { type = type || 'i8'; if (type.charAt(type.length-1) === '*') type = 'i32'; // pointers are 32-bit #if SAFE_HEAP if (noSafe) { switch(type) { case 'i1': {{{ makeSetValue('ptr', '0', 'value', 'i1', undefined, undefined, undefined, '1') }}}; break; case 'i8': {{{ makeSetValue('ptr', '0', 'value', 'i8', undefined, undefined, undefined, '1') }}}; break; case 'i16': {{{ makeSetValue('ptr', '0', 'value', 'i16', undefined, undefined, undefined, '1') }}}; break; case 'i32': {{{ makeSetValue('ptr', '0', 'value', 'i32', undefined, undefined, undefined, '1') }}}; break; case 'i64': {{{ makeSetValue('ptr', '0', 'value', 'i64', undefined, undefined, undefined, '1') }}}; break; case 'float': {{{ makeSetValue('ptr', '0', 'value', 'float', undefined, undefined, undefined, '1') }}}; break; case 'double': {{{ makeSetValue('ptr', '0', 'value', 'double', undefined, undefined, undefined, '1') }}}; break; default: abort('invalid type for setValue: ' + type); } } else { #endif switch(type) { case 'i1': {{{ makeSetValue('ptr', '0', 'value', 'i1') }}}; break; case 'i8': {{{ makeSetValue('ptr', '0', 'value', 'i8') }}}; break; case 'i16': {{{ makeSetValue('ptr', '0', 'value', 'i16') }}}; break; case 'i32': {{{ makeSetValue('ptr', '0', 'value', 'i32') }}}; break; case 'i64': {{{ makeSetValue('ptr', '0', 'value', 'i64') }}}; break; case 'float': {{{ makeSetValue('ptr', '0', 'value', 'float') }}}; break; case 'double': {{{ makeSetValue('ptr', '0', 'value', 'double') }}}; break; default: abort('invalid type for setValue: ' + type); } #if SAFE_HEAP } #endif } Module['setValue'] = setValue; // Parallel to setValue. function getValue(ptr, type, noSafe) { type = type || 'i8'; if (type.charAt(type.length-1) === '*') type = 'i32'; // pointers are 32-bit #if SAFE_HEAP if (noSafe) { switch(type) { case 'i1': return {{{ makeGetValue('ptr', '0', 'i1', undefined, undefined, undefined, undefined, '1') }}}; case 'i8': return {{{ makeGetValue('ptr', '0', 'i8', undefined, undefined, undefined, undefined, '1') }}}; case 'i16': return {{{ makeGetValue('ptr', '0', 'i16', undefined, undefined, undefined, undefined, '1') }}}; case 'i32': return {{{ makeGetValue('ptr', '0', 'i32', undefined, undefined, undefined, undefined, '1') }}}; case 'i64': return {{{ makeGetValue('ptr', '0', 'i64', undefined, undefined, undefined, undefined, '1') }}}; case 'float': return {{{ makeGetValue('ptr', '0', 'float', undefined, undefined, undefined, undefined, '1') }}}; case 'double': return {{{ makeGetValue('ptr', '0', 'double', undefined, undefined, undefined, undefined, '1') }}}; default: abort('invalid type for setValue: ' + type); } } else { #endif switch(type) { case 'i1': return {{{ makeGetValue('ptr', '0', 'i1') }}}; case 'i8': return {{{ makeGetValue('ptr', '0', 'i8') }}}; case 'i16': return {{{ makeGetValue('ptr', '0', 'i16') }}}; case 'i32': return {{{ makeGetValue('ptr', '0', 'i32') }}}; case 'i64': return {{{ makeGetValue('ptr', '0', 'i64') }}}; case 'float': return {{{ makeGetValue('ptr', '0', 'float') }}}; case 'double': return {{{ makeGetValue('ptr', '0', 'double') }}}; default: abort('invalid type for setValue: ' + type); } #if SAFE_HEAP } #endif return null; } Module['getValue'] = getValue; var ALLOC_NORMAL = 0; // Tries to use _malloc() var ALLOC_STACK = 1; // Lives for the duration of the current function call var ALLOC_STATIC = 2; // Cannot be freed var ALLOC_DYNAMIC = 3; // Cannot be freed except through sbrk var ALLOC_NONE = 4; // Do not allocate Module['ALLOC_NORMAL'] = ALLOC_NORMAL; Module['ALLOC_STACK'] = ALLOC_STACK; Module['ALLOC_STATIC'] = ALLOC_STATIC; Module['ALLOC_DYNAMIC'] = ALLOC_DYNAMIC; Module['ALLOC_NONE'] = ALLOC_NONE; // allocate(): This is for internal use. You can use it yourself as well, but the interface // is a little tricky (see docs right below). The reason is that it is optimized // for multiple syntaxes to save space in generated code. So you should // normally not use allocate(), and instead allocate memory using _malloc(), // initialize it with setValue(), and so forth. // @slab: An array of data, or a number. If a number, then the size of the block to allocate, // in *bytes* (note that this is sometimes confusing: the next parameter does not // affect this!) // @types: Either an array of types, one for each byte (or 0 if no type at that position), // or a single type which is used for the entire block. This only matters if there // is initial data - if @slab is a number, then this does not matter at all and is // ignored. // @allocator: How to allocate memory, see ALLOC_* function allocate(slab, types, allocator, ptr) { var zeroinit, size; if (typeof slab === 'number') { zeroinit = true; size = slab; } else { zeroinit = false; size = slab.length; } var singleType = typeof types === 'string' ? types : null; var ret; if (allocator == ALLOC_NONE) { ret = ptr; } else { ret = [_malloc, Runtime.stackAlloc, Runtime.staticAlloc, Runtime.dynamicAlloc][allocator === undefined ? ALLOC_STATIC : allocator](Math.max(size, singleType ? 1 : types.length)); } if (zeroinit) { var ptr = ret, stop; #if USE_TYPED_ARRAYS == 2 assert((ret & 3) == 0); stop = ret + (size & ~3); for (; ptr < stop; ptr += 4) { {{{ makeSetValue('ptr', '0', '0', 'i32', null, true) }}}; } #endif stop = ret + size; while (ptr < stop) { {{{ makeSetValue('ptr++', '0', '0', 'i8', null, true) }}}; } return ret; } #if USE_TYPED_ARRAYS == 2 if (singleType === 'i8') { if (slab.subarray || slab.slice) { HEAPU8.set(slab, ret); } else { HEAPU8.set(new Uint8Array(slab), ret); } return ret; } #endif var i = 0, type, typeSize, previousType; while (i < size) { var curr = slab[i]; if (typeof curr === 'function') { curr = Runtime.getFunctionIndex(curr); } type = singleType || types[i]; if (type === 0) { i++; continue; } #if ASSERTIONS assert(type, 'Must know what type to store in allocate!'); #endif #if USE_TYPED_ARRAYS == 2 if (type == 'i64') type = 'i32'; // special case: we have one i32 here, and one i32 later #endif setValue(ret+i, curr, type); // no need to look up size unless type changes, so cache it if (previousType !== type) { typeSize = Runtime.getNativeTypeSize(type); previousType = type; } i += typeSize; } return ret; } Module['allocate'] = allocate; function Pointer_stringify(ptr, /* optional */ length) { // Find the length, and check for UTF while doing so var hasUtf = false; var t; var i = 0; while (1) { t = {{{ makeGetValue('ptr', 'i', 'i8', 0, 1) }}}; if (t >= 128) hasUtf = true; else if (t == 0 && !length) break; i++; if (length && i == length) break; } if (!length) length = i; var ret = ''; #if USE_TYPED_ARRAYS == 2 if (!hasUtf) { var MAX_CHUNK = 1024; // split up into chunks, because .apply on a huge string can overflow the stack var curr; while (length > 0) { curr = String.fromCharCode.apply(String, HEAPU8.subarray(ptr, ptr + Math.min(length, MAX_CHUNK))); ret = ret ? ret + curr : curr; ptr += MAX_CHUNK; length -= MAX_CHUNK; } return ret; } #endif var utf8 = new Runtime.UTF8Processor(); for (i = 0; i < length; i++) { #if ASSERTIONS assert(ptr + i < TOTAL_MEMORY); #endif t = {{{ makeGetValue('ptr', 'i', 'i8', 0, 1) }}}; ret += utf8.processCChar(t); } return ret; } Module['Pointer_stringify'] = Pointer_stringify; // Memory management var PAGE_SIZE = 4096; function alignMemoryPage(x) { return ((x+4095)>>12)<<12; } var HEAP; #if USE_TYPED_ARRAYS == 1 var IHEAP, IHEAPU; #if USE_FHEAP var FHEAP; #endif #endif #if USE_TYPED_ARRAYS == 2 var HEAP8, HEAPU8, HEAP16, HEAPU16, HEAP32, HEAPU32, HEAPF32, HEAPF64; #endif var STATIC_BASE = 0, STATICTOP = 0, staticSealed = false; // static area var STACK_BASE = 0, STACKTOP = 0, STACK_MAX = 0; // stack area var DYNAMIC_BASE = 0, DYNAMICTOP = 0; // dynamic area handled by sbrk #if USE_TYPED_ARRAYS function enlargeMemory() { #if ALLOW_MEMORY_GROWTH == 0 #if ASM_JS == 0 abort('Cannot enlarge memory arrays. Either (1) compile with -s TOTAL_MEMORY=X with X higher than the current value, (2) compile with ALLOW_MEMORY_GROWTH which adjusts the size at runtime but prevents some optimizations, or (3) set Module.TOTAL_MEMORY before the program runs.'); #else abort('Cannot enlarge memory arrays in asm.js. Either (1) compile with -s TOTAL_MEMORY=X with X higher than the current value, or (2) set Module.TOTAL_MEMORY before the program runs.'); #endif #else // TOTAL_MEMORY is the current size of the actual array, and DYNAMICTOP is the new top. #if ASSERTIONS Module.printErr('Warning: Enlarging memory arrays, this is not fast, and ALLOW_MEMORY_GROWTH is not fully tested with all optimizations on! ' + [DYNAMICTOP, TOTAL_MEMORY]); // We perform safe elimination instead of elimination in this mode, but if you see this error, try to disable it and other optimizations entirely assert(DYNAMICTOP >= TOTAL_MEMORY); assert(TOTAL_MEMORY > 4); // So the loop below will not be infinite #endif while (TOTAL_MEMORY <= DYNAMICTOP) { // Simple heuristic. Override enlargeMemory() if your program has something more optimal for it TOTAL_MEMORY = alignMemoryPage(2*TOTAL_MEMORY); } assert(TOTAL_MEMORY <= Math.pow(2, 30)); // 2^30==1GB is a practical maximum - 2^31 is already close to possible negative numbers etc. #if USE_TYPED_ARRAYS == 1 var oldIHEAP = IHEAP; Module['HEAP'] = Module['IHEAP'] = HEAP = IHEAP = new Int32Array(TOTAL_MEMORY); IHEAP.set(oldIHEAP); IHEAPU = new Uint32Array(IHEAP.buffer); #if USE_FHEAP var oldFHEAP = FHEAP; Module['FHEAP'] = FHEAP = new Float64Array(TOTAL_MEMORY); FHEAP.set(oldFHEAP); #endif #endif #if USE_TYPED_ARRAYS == 2 var oldHEAP8 = HEAP8; var buffer = new ArrayBuffer(TOTAL_MEMORY); Module['HEAP8'] = HEAP8 = new Int8Array(buffer); Module['HEAP16'] = HEAP16 = new Int16Array(buffer); Module['HEAP32'] = HEAP32 = new Int32Array(buffer); Module['HEAPU8'] = HEAPU8 = new Uint8Array(buffer); Module['HEAPU16'] = HEAPU16 = new Uint16Array(buffer); Module['HEAPU32'] = HEAPU32 = new Uint32Array(buffer); Module['HEAPF32'] = HEAPF32 = new Float32Array(buffer); Module['HEAPF64'] = HEAPF64 = new Float64Array(buffer); HEAP8.set(oldHEAP8); #endif #endif } #endif var TOTAL_STACK = Module['TOTAL_STACK'] || {{{ TOTAL_STACK }}}; var TOTAL_MEMORY = Module['TOTAL_MEMORY'] || {{{ TOTAL_MEMORY }}}; var FAST_MEMORY = Module['FAST_MEMORY'] || {{{ FAST_MEMORY }}}; // Initialize the runtime's memory #if USE_TYPED_ARRAYS // check for full engine support (use string 'subarray' to avoid closure compiler confusion) assert(!!Int32Array && !!Float64Array && !!(new Int32Array(1)['subarray']) && !!(new Int32Array(1)['set']), 'Cannot fallback to non-typed array case: Code is too specialized'); #if USE_TYPED_ARRAYS == 1 HEAP = IHEAP = new Int32Array(TOTAL_MEMORY); IHEAPU = new Uint32Array(IHEAP.buffer); #if USE_FHEAP FHEAP = new Float64Array(TOTAL_MEMORY); #endif #endif #if USE_TYPED_ARRAYS == 2 var buffer = new ArrayBuffer(TOTAL_MEMORY); HEAP8 = new Int8Array(buffer); HEAP16 = new Int16Array(buffer); HEAP32 = new Int32Array(buffer); HEAPU8 = new Uint8Array(buffer); HEAPU16 = new Uint16Array(buffer); HEAPU32 = new Uint32Array(buffer); HEAPF32 = new Float32Array(buffer); HEAPF64 = new Float64Array(buffer); // Endianness check (note: assumes compiler arch was little-endian) HEAP32[0] = 255; assert(HEAPU8[0] === 255 && HEAPU8[3] === 0, 'Typed arrays 2 must be run on a little-endian system'); #endif #else // Make sure that our HEAP is implemented as a flat array. HEAP = []; // Hinting at the size with |new Array(TOTAL_MEMORY)| should help in theory but makes v8 much slower for (var i = 0; i < FAST_MEMORY; i++) { HEAP[i] = 0; // XXX We do *not* use {{| makeSetValue(0, 'i', 0, 'null') |}} here, since this is done just to optimize runtime speed } #endif Module['HEAP'] = HEAP; #if USE_TYPED_ARRAYS == 1 Module['IHEAP'] = IHEAP; #if USE_FHEAP Module['FHEAP'] = FHEAP; #endif #endif #if USE_TYPED_ARRAYS == 2 Module['HEAP8'] = HEAP8; Module['HEAP16'] = HEAP16; Module['HEAP32'] = HEAP32; Module['HEAPU8'] = HEAPU8; Module['HEAPU16'] = HEAPU16; Module['HEAPU32'] = HEAPU32; Module['HEAPF32'] = HEAPF32; Module['HEAPF64'] = HEAPF64; #endif function callRuntimeCallbacks(callbacks) { while(callbacks.length > 0) { var callback = callbacks.shift(); if (typeof callback == 'function') { callback(); continue; } var func = callback.func; if (typeof func === 'number') { if (callback.arg === undefined) { Runtime.dynCall('v', func); } else { Runtime.dynCall('vi', func, [callback.arg]); } } else { func(callback.arg === undefined ? null : callback.arg); } } } var __ATINIT__ = []; // functions called during startup var __ATMAIN__ = []; // functions called when main() is to be run var __ATEXIT__ = []; // functions called during shutdown var runtimeInitialized = false; function ensureInitRuntime() { if (runtimeInitialized) return; runtimeInitialized = true; callRuntimeCallbacks(__ATINIT__); } function preMain() { callRuntimeCallbacks(__ATMAIN__); } function exitRuntime() { callRuntimeCallbacks(__ATEXIT__); } // Tools // This processes a JS string into a C-line array of numbers, 0-terminated. // For LLVM-originating strings, see parser.js:parseLLVMString function function intArrayFromString(stringy, dontAddNull, length /* optional */) { var ret = (new Runtime.UTF8Processor()).processJSString(stringy); if (length) { ret.length = length; } if (!dontAddNull) { ret.push(0); } return ret; } Module['intArrayFromString'] = intArrayFromString; function intArrayToString(array) { var ret = []; for (var i = 0; i < array.length; i++) { var chr = array[i]; if (chr > 0xFF) { #if ASSERTIONS assert(false, 'Character code ' + chr + ' (' + String.fromCharCode(chr) + ') at offset ' + i + ' not in 0x00-0xFF.'); #endif chr &= 0xFF; } ret.push(String.fromCharCode(chr)); } return ret.join(''); } Module['intArrayToString'] = intArrayToString; // Write a Javascript array to somewhere in the heap function writeStringToMemory(string, buffer, dontAddNull) { var array = intArrayFromString(string, dontAddNull); var i = 0; while (i < array.length) { var chr = array[i]; {{{ makeSetValue('buffer', 'i', 'chr', 'i8') }}} i = i + 1; } } Module['writeStringToMemory'] = writeStringToMemory; function writeArrayToMemory(array, buffer) { for (var i = 0; i < array.length; i++) { {{{ makeSetValue('buffer', 'i', 'array[i]', 'i8') }}}; } } Module['writeArrayToMemory'] = writeArrayToMemory; {{{ unSign }}} {{{ reSign }}} #if PRECISE_I32_MUL if (!Math['imul']) Math['imul'] = function(a, b) { var ah = a >>> 16; var al = a & 0xffff; var bh = b >>> 16; var bl = b & 0xffff; return (al*bl + ((ah*bl + al*bh) << 16))|0; }; #else Math['imul'] = function(a, b) { return (a*b)|0; // fast but imprecise }; #endif // A counter of dependencies for calling run(). If we need to // do asynchronous work before running, increment this and // decrement it. Incrementing must happen in a place like // PRE_RUN_ADDITIONS (used by emcc to add file preloading). // Note that you can add dependencies in preRun, even though // it happens right before run - run will be postponed until // the dependencies are met. var runDependencies = 0; var runDependencyTracking = {}; var calledInit = false, calledRun = false; var runDependencyWatcher = null; function addRunDependency(id) { runDependencies++; if (Module['monitorRunDependencies']) { Module['monitorRunDependencies'](runDependencies); } if (id) { assert(!runDependencyTracking[id]); runDependencyTracking[id] = 1; #if ASSERTIONS if (runDependencyWatcher === null && typeof setInterval !== 'undefined') { // Check for missing dependencies every few seconds runDependencyWatcher = setInterval(function() { var shown = false; for (var dep in runDependencyTracking) { if (!shown) { shown = true; Module.printErr('still waiting on run dependencies:'); } Module.printErr('dependency: ' + dep); } if (shown) { Module.printErr('(end of list)'); } }, 10000); } #endif } else { Module.printErr('warning: run dependency added without ID'); } } Module['addRunDependency'] = addRunDependency; function removeRunDependency(id) { runDependencies--; if (Module['monitorRunDependencies']) { Module['monitorRunDependencies'](runDependencies); } if (id) { assert(runDependencyTracking[id]); delete runDependencyTracking[id]; } else { Module.printErr('warning: run dependency removed without ID'); } if (runDependencies == 0) { if (runDependencyWatcher !== null) { clearInterval(runDependencyWatcher); runDependencyWatcher = null; } // If run has never been called, and we should call run (INVOKE_RUN is true, and Module.noInitialRun is not false) if (!calledRun && shouldRunNow) run(); } } Module['removeRunDependency'] = removeRunDependency; Module["preloadedImages"] = {}; // maps url to image data Module["preloadedAudios"] = {}; // maps url to audio data function addPreRun(func) { if (!Module['preRun']) Module['preRun'] = []; else if (typeof Module['preRun'] == 'function') Module['preRun'] = [Module['preRun']]; Module['preRun'].push(func); } #if PGO var PGOMonitor = { called: {}, dump: function() { var dead = []; for (var i = 0; i < this.allGenerated.length; i++) { var func = this.allGenerated[i]; if (!this.called[func]) dead.push(func); } Module.print('-s DEAD_FUNCTIONS=\'' + JSON.stringify(dead) + '\'\n'); } }; Module['PGOMonitor'] = PGOMonitor; __ATEXIT__.push({ func: function() { PGOMonitor.dump() } }); addPreRun(function() { addRunDependency('pgo') }); #endif var awaitingMemoryInitializer = false; function loadMemoryInitializer(filename) { function applyData(data) { #if USE_TYPED_ARRAYS == 2 HEAPU8.set(data, STATIC_BASE); #else allocate(data, 'i8', ALLOC_NONE, STATIC_BASE); #endif runPostSets(); } // always do this asynchronously, to keep shell and web as similar as possible addPreRun(function() { if (ENVIRONMENT_IS_NODE || ENVIRONMENT_IS_SHELL) { applyData(Module['readBinary'](filename)); } else { Browser.asyncLoad(filename, function(data) { applyData(data); }, function(data) { throw 'could not load memory initializer ' + filename; }); } }); awaitingMemoryInitializer = false; } // === Body ===