Skip to content

/emscripten

Permalink
Tag: llvm-opts
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
1234 lines (1132 sloc) 50.5 KB
Raw Blame History
//"use strict";
// Convert analyzed data to javascript. Everything has already been calculated
// before this stage, which just does the final conversion to JavaScript.
// Handy sets
var STRUCT_LIST = set('struct', 'list');
var UNDERSCORE_OPENPARENS = set('_', '(');
// JSifier
function JSify(data, functionsOnly, givenFunctions) {
var mainPass = !functionsOnly;
if (mainPass) {
// We will start to print out the data, but must do so carefully - we are
// dealing with potentially *huge* strings. Convenient replacements and
// manipulations may create in-memory copies, and we may OOM.
//
// Final shape that will be created:
// shell
// (body)
// preamble
// runtime
// generated code
// postamble
// global_vars
//
// First, we print out everything until the generated code. Then the
// functions will print themselves out as they are parsed. Finally, we
// will call finalCombiner in the main pass, to print out everything
// else. This lets us not hold any strings in memory, we simply print
// things out as they are ready.
var shellFile = BUILD_AS_SHARED_LIB ? 'shell_sharedlib.js' : 'shell.js';
var shellParts = read(shellFile).split('{{BODY}}');
print(shellParts[0]);
var preFile = BUILD_AS_SHARED_LIB ? 'preamble_sharedlib.js' : 'preamble.js';
var pre = processMacros(preprocess(read(preFile).replace('{{RUNTIME}}', getRuntime())));
print(pre);
Functions.implementedFunctions = set(data.unparsedFunctions.map(function(func) { return func.ident }));
}
// Does simple 'macro' substitution, using Django-like syntax,
// {{{ code }}} will be replaced with |eval(code)|.
function processMacros(text) {
return text.replace(/{{{[^}]+}}}/g, function(str) {
str = str.substr(3, str.length-6);
return eval(str).toString();
});
}
var substrate = new Substrate('JSifyer');
if (mainPass) {
// Handle unparsed types TODO: Batch them
analyzer(intertyper(data.unparsedTypess[0].lines, true), true);
data.unparsedTypess = null;
// Add additional necessary items for the main pass. We can now do this since types are parsed (types can be used through
// generateStructInfo in library.js)
LibraryManager.load();
var libFuncsToInclude;
if (INCLUDE_FULL_LIBRARY) {
assert(!BUILD_AS_SHARED_LIB, 'Cannot have both INCLUDE_FULL_LIBRARY and BUILD_AS_SHARED_LIB set.')
libFuncsToInclude = [];
for (var key in LibraryManager.library) {
if (!key.match(/__(deps|postset)$/)) {
libFuncsToInclude.push(key);
}
}
} else {
libFuncsToInclude = ['memcpy', 'memset', 'malloc', 'free'];
}
libFuncsToInclude.forEach(function(ident) {
data.functionStubs.push({
intertype: 'functionStub',
ident: '_' + ident
});
});
}
// Functions
Functions.currFunctions = !mainPass ? givenFunctions.currFunctions : {};
Functions.currExternalFunctions = !mainPass ? givenFunctions.currExternalFunctions : {};
// Now that first-pass analysis has completed (so we have basic types, etc.), we can get around to handling unparsedFunctions
(!mainPass ? data.functions : data.unparsedFunctions.concat(data.functions)).forEach(function(func) {
// Save just what we need, to save memory
Functions.currFunctions[func.ident] = {
hasVarArgs: func.hasVarArgs,
numParams: func.params.length,
labelIds: func.labelIds // TODO: We need this for globals, but perhaps we can calculate them early and free this
};
});
data.functionStubs.forEach(function(func) {
// Don't overwrite stubs that have more info.
if (!Functions.currExternalFunctions.hasOwnProperty(func.ident) ||
!Functions.currExternalFunctions[func.ident].numParams === undefined) {
Functions.currExternalFunctions[func.ident] = {
hasVarArgs: func.hasVarArgs,
numParams: func.params && func.params.length
};
}
});
var MAX_BATCH_FUNC_LINES = 1000;
while (data.unparsedFunctions.length > 0) {
var currFuncLines = [];
var currBaseLineNums = [];
while (currFuncLines.length == 0 ||
(data.unparsedFunctions.length > 0 && currFuncLines.length + data.unparsedFunctions[0].lines.length <= MAX_BATCH_FUNC_LINES)) {
currBaseLineNums.push([currFuncLines.length, data.unparsedFunctions[0].lineNum-1]);
currFuncLines = currFuncLines.concat(data.unparsedFunctions[0].lines); // for first one, assign, do not concat?
data.unparsedFunctions.shift();
}
dprint('unparsedFunctions','====================\n// Processing function batch of ' + currBaseLineNums.length +
' functions, ' + currFuncLines.length + ' lines, functions left: ' + data.unparsedFunctions.length);
if (DEBUG_MEMORY) MemoryDebugger.tick('pre-func');
JSify(analyzer(intertyper(currFuncLines, true, currBaseLineNums), true), true, Functions);
if (DEBUG_MEMORY) MemoryDebugger.tick('post-func');
}
currFuncLines = currBaseLineNums = null; // Do not hold on to anything from inside that loop (JS function scoping..)
data.unparsedFunctions = null;
// Actors
// type
// FIXME: This is no longer used, we do not actually need to JSify on types. TODO: Remove this and related code
substrate.addActor('Type', {
processItem: function(item) {
var type = Types.types[item.name_];
var niceName = toNiceIdent(item.name_);
// We might export all of Types.types, cleaner that way, but do not want slowdowns in accessing flatteners
item.JS = 'var ' + niceName + '___SIZE = ' + Types.types[item.name_].flatSize + '; // ' + item.name_ + '\n';
if (type.needsFlattening && !type.flatFactor) {
item.JS += 'var ' + niceName + '___FLATTENER = ' + JSON.stringify(Types.types[item.name_].flatIndexes) + ';';
}
return [item];
}
});
function makeEmptyStruct(type) {
var ret = [];
var typeData = Types.types[type];
assertTrue(typeData);
for (var i = 0; i < typeData.flatSize; i++) {
ret.push(0);
}
return ret;
}
function alignStruct(values, type) {
var typeData = Types.types[type];
assertTrue(typeData);
var ret = new Array(typeData.flatSize);
var index = 0;
var i = 0;
while (i < values.length) {
// Pad until the right place
var padded = typeData.flatFactor ? typeData.flatFactor*i : typeData.flatIndexes[i];
while (index < padded) {
ret[index++] = 0;
}
// Add current value(s)
var currValue = flatten(values[i]);
if (I64_MODE == 1 && typeData.fields[i] == 'i64') {
// 'flatten' out the 64-bit value into two 32-bit halves
ret[index++] = currValue>>>0;
ret[index++] = 0;
ret[index++] = 0;
ret[index++] = 0;
ret[index++] = Math.floor(currValue/4294967296);
ret[index++] = 0;
ret[index++] = 0;
ret[index++] = 0;
} else if (typeof currValue == 'object') {
for (var j = 0; j < currValue.length; j++) {
ret[index++] = currValue[j];
}
} else {
ret[index++] = currValue;
}
i += 1;
}
while (index < typeData.flatSize) {
ret[index++] = 0;
}
return ret;
}
// Gets an entire constant expression
function makeConst(value, type, ident) {
// Gets an array of constant items, separated by ',' tokens
function handleSegments(tokens) {
// Handle a single segment (after comma separation)
function handleSegment(segment) {
var ret;
if (segment.intertype === 'value') {
ret = segment.ident;
} else if (segment.intertype === 'emptystruct') {
ret = makeEmptyStruct(segment.type);
} else if (segment.intertype in PARSABLE_LLVM_FUNCTIONS) {
ret = finalizeLLVMFunctionCall(segment);
} else if (segment.intertype in STRUCT_LIST) {
ret = alignStruct(handleSegments(segment.contents), segment.type);
} else if (segment.intertype === 'string') {
ret = parseLLVMString(segment.text); // + ' /* ' + text + '*/';
} else if (segment.intertype === 'blockaddress') {
ret = finalizeBlockAddress(segment);
} else {
throw 'Invalid segment: ' + dump(segment);
}
assert(segment.type, 'Missing type for constant segment!');
return indexizeFunctions(ret, segment.type);
};
return tokens.map(handleSegment)
}
//dprint('jsifier const: ' + JSON.stringify(value) + ',' + type + '\n');
if (value.intertype in PARSABLE_LLVM_FUNCTIONS) {
return [finalizeLLVMFunctionCall(value)];
} else if (Runtime.isNumberType(type) || pointingLevels(type) >= 1) {
return indexizeFunctions(parseNumerical(value.value), type);
} else if (value.intertype === 'emptystruct') {
return makeEmptyStruct(type);
} else if (value.intertype === 'string') {
return JSON.stringify(parseLLVMString(value.text)) +
' /* ' + value.text.substr(0, 20).replace(/\*/g, '_') + ' */'; // make string safe for inclusion in comment
} else {
return alignStruct(handleSegments(value.contents), type);
}
}
function parseConst(value, type, ident) {
var constant = makeConst(value, type);
if (typeof constant === 'object') {
constant = flatten(constant).map(function(x) { return parseNumerical(x) })
}
return constant;
}
// globalVariable
substrate.addActor('GlobalVariable', {
processItem: function(item) {
function needsPostSet(value) {
return value[0] in UNDERSCORE_OPENPARENS || value.substr(0, 14) === 'CHECK_OVERFLOW'
|| value.substr(0, 13) === 'STRING_TABLE.';
}
item.intertype = 'GlobalVariableStub';
assert(!item.lines); // FIXME remove this, after we are sure it isn't needed
var ret = [item];
if (item.ident == '_llvm_global_ctors') {
item.JS = '\n__ATINIT__ = __ATINIT__.concat([\n' +
item.ctors.map(function(ctor) { return ' { func: ' + toNiceIdent(ctor) + ' }' }).join(',\n') +
'\n]);\n';
return ret;
} else {
if (item.external && BUILD_AS_SHARED_LIB) {
// External variables in shared libraries should not be declared as
// they would shadow similarly-named globals in the parent.
item.JS = '';
} else {
if (!(item.ident in Variables.globals) || !Variables.globals[item.ident].isString) {
item.JS = 'var ' + item.ident + ';';
}
}
var constant = null;
if (item.external) {
// Import external global variables from the library if available.
var shortident = item.ident.slice(1);
if (LibraryManager.library[shortident] &&
LibraryManager.library[shortident].length &&
!BUILD_AS_SHARED_LIB) {
var val = LibraryManager.library[shortident];
var padding;
if (Runtime.isNumberType(item.type) || isPointerType(item.type)) {
padding = [item.type].concat(zeros(Runtime.getNativeFieldSize(item.type)));
} else {
padding = makeEmptyStruct(item.type);
}
var padded = val.concat(padding.slice(val.length));
var js = item.ident + '=' + makePointer(JSON.stringify(padded), null, 'ALLOC_STATIC', item.type) + ';'
if (LibraryManager.library[shortident + '__postset']) {
js += '\n' + LibraryManager.library[shortident + '__postset'];
}
ret.push({
intertype: 'GlobalVariablePostSet',
JS: js
});
}
return ret;
} else {
constant = parseConst(item.value, item.type, item.ident);
if (typeof constant === 'string' && constant[0] != '[') {
constant = [constant]; // A single item. We may need a postset for it.
}
if (typeof constant === 'object') {
// This is a flattened object. We need to find its idents, so they can be assigned to later
constant.forEach(function(value, i) {
if (needsPostSet(value)) { // ident, or expression containing an ident
ret.push({
intertype: 'GlobalVariablePostSet',
JS: makeSetValue(item.ident, i, value, 'i32', false, true) + ';' // ignore=true, since e.g. rtti and statics cause lots of safe_heap errors
});
constant[i] = '0';
}
});
constant = '[' + constant.join(', ') + ']';
}
// NOTE: This is the only place that could potentially create static
// allocations in a shared library.
constant = makePointer(constant, null, BUILD_AS_SHARED_LIB ? 'ALLOC_NORMAL' : 'ALLOC_STATIC', item.type);
var js;
// Strings are held in STRING_TABLE, to not clutter up the main namespace (in some cases we have
// many many strings, possibly exceeding the js engine limit on global vars).
if (Variables.globals[item.ident].isString) {
js = 'STRING_TABLE.' + item.ident + '=' + constant + ';';
} else {
js = item.ident + '=' + constant + ';';
}
// Special case: class vtables. We make sure they are null-terminated, to allow easy runtime operations
if (item.ident.substr(0, 5) == '__ZTV') {
js += '\n' + makePointer('[0]', null, BUILD_AS_SHARED_LIB ? 'ALLOC_NORMAL' : 'ALLOC_STATIC', ['void*']) + ';';
}
if (item.ident in EXPORTED_GLOBALS) {
js += '\nModule["' + item.ident + '"] = ' + item.ident + ';';
}
if (BUILD_AS_SHARED_LIB == 2 && !item.private_) {
// TODO: make the assert conditional on ASSERTIONS
js += 'if (globalScope) { assert(!globalScope["' + item.ident + '"]); globalScope["' + item.ident + '"] = ' + item.ident + ' }';
}
return ret.concat({
intertype: 'GlobalVariable',
JS: js,
});
}
}
}
});
// alias
substrate.addActor('Alias', {
processItem: function(item) {
item.intertype = 'GlobalVariableStub';
var ret = [item];
item.JS = 'var ' + item.ident + ';';
// Set the actual value in a postset, since it may be a global variable. We also order by dependencies there
var value = Variables.globals[item.ident].resolvedAlias = finalizeLLVMParameter(item.value);
ret.push({
intertype: 'GlobalVariablePostSet',
ident: item.ident,
dependencies: set([value]),
JS: item.ident + ' = ' + value + ';'
});
return ret;
}
});
var addedLibraryItems = {};
// functionStub
substrate.addActor('FunctionStub', {
processItem: function(item) {
function addFromLibrary(ident) {
if (ident in addedLibraryItems) return '';
// Don't replace implemented functions with library ones (which can happen when we add dependencies).
// Note: We don't return the dependencies here. Be careful not to end up where this matters
if (('_' + ident) in Functions.implementedFunctions) return '';
addedLibraryItems[ident] = true;
var snippet = LibraryManager.library[ident];
var redirectedIdent = null;
var deps = LibraryManager.library[ident + '__deps'] || [];
var isFunction = false;
if (typeof snippet === 'string') {
if (LibraryManager.library[snippet]) {
// Redirection for aliases. We include the parent, and at runtime make ourselves equal to it.
// This avoid having duplicate functions with identical content.
redirectedIdent = snippet;
deps.push(snippet);
snippet = '_' + snippet;
}
} else if (typeof snippet === 'object') {
snippet = stringifyWithFunctions(snippet);
} else if (typeof snippet === 'function') {
isFunction = true;
snippet = snippet.toString();
assert(snippet.indexOf('XXX missing C define') == -1,
'Trying to include a library function with missing C defines: ' + ident + ' | ' + snippet);
// name the function; overwrite if it's already named
snippet = snippet.replace(/function(?:\s+([^(]+))?\s*\(/, 'function _' + ident + '(');
if (LIBRARY_DEBUG) {
snippet = snippet.replace('{', '{ print("[library call:' + ident + ']"); ');
}
}
var postsetId = ident + '__postset';
var postset = LibraryManager.library[postsetId];
if (postset && !addedLibraryItems[postsetId]) {
addedLibraryItems[postsetId] = true;
ret.push({
intertype: 'GlobalVariablePostSet',
JS: postset
});
}
if (redirectedIdent) {
deps = deps.concat(LibraryManager.library[redirectedIdent + '__deps'] || []);
}
// $ident's are special, we do not prefix them with a '_'.
if (ident[0] === '$') {
ident = ident.substr(1);
} else {
ident = '_' + ident;
}
var text = (deps ? '\n' + deps.map(addFromLibrary).join('\n') : '');
text += isFunction ? snippet : 'var ' + ident + '=' + snippet + ';';
if (ident in EXPORTED_FUNCTIONS) {
text += '\nModule["' + ident + '"] = ' + ident + ';';
}
return text;
}
var ret = [item];
if (IGNORED_FUNCTIONS.indexOf(item.ident) >= 0) return null;
var shortident = item.ident.substr(1);
if (BUILD_AS_SHARED_LIB) {
// Shared libraries reuse the runtime of their parents.
item.JS = '';
} else if (LibraryManager.library.hasOwnProperty(shortident)) {
item.JS = addFromLibrary(shortident);
} else {
item.JS = 'var ' + item.ident + '; // stub for ' + item.ident;
}
return ret;
}
});
// function splitter
substrate.addActor('FunctionSplitter', {
processItem: function(item) {
var ret = [item];
item.splitItems = 0;
item.labels.forEach(function(label) {
label.lines.forEach(function(line) {
line.func = item.ident;
line.funcData = item; // TODO: remove all these, access it globally
line.parentLabel = label.ident;
ret.push(line);
item.splitItems ++;
});
});
this.forwardItems(ret, 'FuncLineTriager');
}
});
// function reconstructor & post-JS optimizer
substrate.addActor('FunctionReconstructor', {
funcs: {},
seen: {},
processItem: function(item) {
if (this.seen[item.__uid__]) return null;
if (item.intertype == 'function') {
this.funcs[item.ident] = item;
item.relines = {};
this.seen[item.__uid__] = true;
return null;
}
var line = item;
var func = this.funcs[line.func];
if (!func) return null;
// Re-insert our line
this.seen[item.__uid__] = true;
var label = func.labels.filter(function(label) { return label.ident == line.parentLabel })[0];
label.lines = label.lines.map(function(line2) {
return (line2.lineNum !== line.lineNum) ? line2 : line;
});
func.splitItems --;
// OLD delete line.funcData; // clean up
if (func.splitItems > 0) return null;
// We have this function all reconstructed, go and finalize it's JS!
if (IGNORED_FUNCTIONS.indexOf(func.ident) >= 0) return null;
func.JS = '\n';
if (CLOSURE_ANNOTATIONS) {
func.JS += '/**\n';
func.paramIdents.forEach(function(param) {
func.JS += ' * @param {number} ' + param + '\n';
});
func.JS += ' * @return {number}\n'
func.JS += ' */\n';
}
func.JS += 'function ' + func.ident + '(' + func.paramIdents.join(', ') + ') {\n';
if (PROFILE) {
func.JS += ' if (PROFILING) { '
+ 'var __parentProfilingNode__ = PROFILING_NODE; PROFILING_NODE = PROFILING_NODE.children["' + func.ident + '"]; '
+ 'if (!PROFILING_NODE) __parentProfilingNode__.children["' + func.ident + '"] = PROFILING_NODE = { time: 0, children: {}, calls: 0 };'
+ 'PROFILING_NODE.calls++; '
+ 'var __profilingStartTime__ = Date.now() '
+ '}\n';
}
// Prepare the stack, if we need one. If we have other stack allocations, force the stack to be set up.
func.JS += ' ' + RuntimeGenerator.stackEnter(func.initialStack, func.otherStackAllocations) + ';\n';
// Make copies of by-value params
// XXX It is not clear we actually need this. While without this we fail, it does look like
// Clang normally does the copy itself, in the calling function. We only need this code
// when Clang optimizes the code and passes the original, not the copy, to the other
// function. But Clang still copies, the copy is just unused! Need to figure out if that
// is caused by our running just some optimizations (the safe ones), or if its a bug
// in Clang, or a bug in our understanding of the IR.
func.params.forEach(function(param) {
if (param.byVal) {
var type = removePointing(param.type);
var typeInfo = Types.types[type];
func.JS += ' var tempParam = ' + param.ident + '; ' + param.ident + ' = ' + RuntimeGenerator.stackAlloc(typeInfo.flatSize) + ';' +
makeCopyValues(param.ident, 'tempParam', typeInfo.flatSize, 'null', null, param.byVal) + ';\n';
}
});
if (LABEL_DEBUG) func.JS += " print(INDENT + ' Entering: " + func.ident + "'); INDENT += ' ';\n";
if (true) { // TODO: optimize away when not needed
if (CLOSURE_ANNOTATIONS) func.JS += '/** @type {number} */';
func.JS += ' var __label__;\n';
}
if (func.needsLastLabel) {
func.JS += ' var __lastLabel__ = null;\n';
}
// Walk function blocks and generate JS
function walkBlock(block, indent) {
if (!block) return '';
dprint('relooping', 'walking block: ' + block.type + ',' + block.entries + ' : ' + block.labels.length);
function getLabelLines(label, indent) {
if (!label) return '';
var ret = '';
if (LABEL_DEBUG) {
ret += indent + "print(INDENT + '" + func.ident + ":" + label.ident + "');\n";
}
if (EXECUTION_TIMEOUT > 0) {
ret += indent + 'if (Date.now() - START_TIME >= ' + (EXECUTION_TIMEOUT*1000) + ') throw "Timed out!" + (new Error().stack);\n';
}
// for special labels we care about (for phi), mark that we visited them
return ret + label.lines.map(function(line) { return line.JS + (Debugging.on ? Debugging.getComment(line.lineNum) : '') })
.join('\n')
.split('\n') // some lines include line breaks
.map(function(line) { return indent + line })
.join('\n');
}
var ret = '';
if (block.type == 'emulated') {
if (block.labels.length > 1) {
if (block.entries.length == 1) {
ret += indent + '__label__ = ' + getLabelId(block.entries[0]) + '; ' + (SHOW_LABELS ? '/* ' + block.entries[0] + ' */' : '') + '\n';
} // otherwise, should have been set before!
ret += indent + 'while(1) switch(__label__) {\n';
ret += block.labels.map(function(label) {
return indent + ' case ' + getLabelId(label.ident) + ': // ' + label.ident + '\n'
+ getLabelLines(label, indent + ' ');
}).join('\n');
ret += '\n' + indent + ' default: assert(0, "bad label: " + __label__);\n' + indent + '}';
} else {
ret += (SHOW_LABELS ? indent + '/* ' + block.entries[0] + ' */' : '') + '\n' + getLabelLines(block.labels[0], indent);
}
ret += '\n';
} else if (block.type == 'reloop') {
ret += indent + block.id + ': while(1) { ' + (SHOW_LABELS ? ' /* ' + block.entries + + ' */' : '') + '\n';
ret += walkBlock(block.inner, indent + ' ');
ret += indent + '}\n';
} else if (block.type == 'multiple') {
var first = true;
var multipleIdent = '';
ret += indent + block.id + ': do { \n';
multipleIdent = ' ';
// TODO: Find out cases where the final if/case is not needed - where we know we must be in a specific label at that point
var SWITCH_IN_MULTIPLE = 0; // This appears to never be worth it, for no amount of labels
if (SWITCH_IN_MULTIPLE && block.entryLabels.length >= 2) {
ret += indent + multipleIdent + 'switch(__label__) {\n';
block.entryLabels.forEach(function(entryLabel) {
ret += indent + multipleIdent + ' case ' + getLabelId(entryLabel.ident) + ': {\n';
ret += walkBlock(entryLabel.block, indent + ' ' + multipleIdent);
ret += indent + multipleIdent + ' } break;\n';
});
ret += indent + multipleIdent + '}\n';
} else {
block.entryLabels.forEach(function(entryLabel) {
ret += indent + multipleIdent + (first ? '' : 'else ') + 'if (__label__ == ' + getLabelId(entryLabel.ident) + ') {\n';
ret += walkBlock(entryLabel.block, indent + ' ' + multipleIdent);
ret += indent + multipleIdent + '}\n';
first = false;
});
}
ret += indent + '} while(0);\n';
} else {
throw "Walked into an invalid block type: " + block.type;
}
return ret + walkBlock(block.next, indent);
}
func.JS += walkBlock(func.block, ' ');
// Finalize function
if (LABEL_DEBUG) func.JS += " INDENT = INDENT.substr(0, INDENT.length-2);\n";
// Add an unneeded return, needed for strict mode to not throw warnings in some cases.
// If we are not relooping, then switches make it unimportant to have this (and, we lack hasReturn anyhow)
if (RELOOP && func.lines.length > 0 && func.labels.filter(function(label) { return label.hasReturn }).length > 0) {
func.JS += ' return' + (func.returnType !== 'void' ? ' null' : '') + ';\n';
}
func.JS += '}\n';
if (func.ident in EXPORTED_FUNCTIONS) {
func.JS += 'Module["' + func.ident + '"] = ' + func.ident + ';';
}
if (func.lines.length >= CLOSURE_INLINE_PREVENTION_LINES) {
func.JS += func.ident + '["X"]=1;';
}
if (BUILD_AS_SHARED_LIB == 2) {
// TODO: make the assert conditional on ASSERTIONS
func.JS += 'if (globalScope) { assert(!globalScope["' + func.ident + '"]); globalScope["' + func.ident + '"] = ' + func.ident + ' }';
}
return func;
}
});
function getVarData(funcData, ident) {
var local = funcData.variables[ident];
if (local) return local;
var global = Variables.globals[ident];
return global || null;
}
function getVarImpl(funcData, ident) {
if (ident === 'null' || isNumber(ident)) return VAR_NATIVIZED; // like nativized, in that we have the actual value right here
var data = getVarData(funcData, ident);
assert(data, 'What variable is this? |' + ident + '|');
return data.impl;
}
substrate.addActor('FuncLineTriager', {
processItem: function(item) {
if (item.intertype == 'function') {
this.forwardItem(item, 'FunctionReconstructor');
} else if (item.JS) {
if (item.parentLineNum) {
this.forwardItem(item, 'AssignReintegrator');
} else {
this.forwardItem(item, 'FunctionReconstructor');
}
} else {
this.forwardItem(item, 'Intertype:' + item.intertype);
}
}
});
// assignment
substrate.addActor('Intertype:assign', {
processItem: function(item) {
var pair = splitItem(item, 'value', ['funcData']);
this.forwardItem(pair.parent, 'AssignReintegrator');
this.forwardItem(pair.child, 'FuncLineTriager');
}
});
substrate.addActor('AssignReintegrator', makeReintegrator(function(item, child) {
// 'var', since this is SSA - first assignment is the only assignment, and where it is defined
item.JS = '';
if (CLOSURE_ANNOTATIONS) item.JS += '/** @type {number} */ ';
item.JS += (item.overrideSSA ? '' : 'var ') + toNiceIdent(item.ident);
var type = item.value.type;
var value = parseNumerical(item.value.JS);
var impl = getVarImpl(item.funcData, item.ident);
switch (impl) {
case VAR_NATIVE: {
break;
}
case VAR_NATIVIZED: {
// SSA, so this must be the alloca. No need for a value
if (!item.overrideSSA) value = '';
break;
}
case VAR_EMULATED: {
break;
}
default: throw 'zz unknown impl: ' + impl;
}
if (value)
item.JS += '=' + value;
item.JS += ';';
this.forwardItem(item, 'FunctionReconstructor');
}));
// Function lines
function makeFuncLineActor(intertype, func) {
return substrate.addActor('Intertype:' + intertype, {
processItem: function(item) {
item.JS = func(item);
if (!item.JS) throw "No JS generated for " + dump(item);
this.forwardItem(item, 'FuncLineTriager');
}
});
}
makeFuncLineActor('value', function(item) {
return item.ident;
});
makeFuncLineActor('noop', function(item) {
return ';';
});
makeFuncLineActor('var', function(item) { // assigns into phis become simple vars when MICRO_OPTS
return 'var ' + item.ident + ';';
});
makeFuncLineActor('store', function(item) {
var value = finalizeLLVMParameter(item.value);
if (pointingLevels(item.pointerType) == 1) {
value = parseNumerical(value, item.valueType);
}
var impl = VAR_EMULATED;
if (item.pointer.intertype == 'value') {
impl = getVarImpl(item.funcData, item.ident);
}
switch (impl) {
case VAR_NATIVIZED:
return item.ident + '=' + value + ';'; // We have the actual value here
break;
case VAR_EMULATED:
if (item.pointer.intertype == 'value') {
return makeSetValue(item.ident, 0, value, item.valueType, 0, 0, item.align) + ';';
} else {
return makeSetValue(0, finalizeLLVMParameter(item.pointer), value, item.valueType, 0, 0, item.align) + ';';
}
break;
default:
throw 'unknown [store] impl: ' + impl;
}
return null;
});
makeFuncLineActor('deleted', function(item) { return ';' });
function getLabelId(label) {
var funcData = Framework.currItem.funcData;
var labelIds = funcData.labelIds;
if (labelIds[label] !== undefined) return labelIds[label];
return labelIds[label] = funcData.labelIdCounter++;
}
function makeBranch(label, lastLabel, labelIsVariable) {
var pre = '';
if (!MICRO_OPTS && lastLabel) {
pre = '__lastLabel__ = ' + getLabelId(lastLabel) + '; ';
}
if (label[0] == 'B') {
assert(!labelIsVariable, 'Cannot handle branches to variables with special branching options');
var parts = label.split('|');
var trueLabel = parts[1] || '';
var oldLabel = parts[2] || '';
var labelSetting = oldLabel ? '__label__ = ' + getLabelId(oldLabel) + ';' +
(SHOW_LABELS ? ' /* to: ' + cleanLabel(oldLabel) + ' */' : '') : ''; // TODO: optimize away the setting
if (label[1] == 'R') {
if (label[2] == 'N') { // BRNOL: break, no label setting
labelSetting = '';
}
return pre + labelSetting + 'break ' + trueLabel + ';';
} else if (label[1] == 'C') { // CONT
if (label[2] == 'N') { // BCNOL: continue, no label setting
labelSetting = '';
}
return pre + labelSetting + 'continue ' + trueLabel + ';';
} else if (label[1] == 'N') { // NOPP
return pre + ';'; // Returning no text might confuse this parser
} else if (label[1] == 'J') { // JSET
return pre + labelSetting + ';';
} else {
throw 'Invalid B-op in branch: ' + trueLabel + ',' + oldLabel;
}
} else {
if (!labelIsVariable) label = getLabelId(label);
return pre + '__label__ = ' + label + ';' + (SHOW_LABELS ? ' /* to: ' + cleanLabel(label) + ' */' : '') + ' break;';
}
}
function calcPhiSets(item) {
if (!item.phi) return null;
var phiSets = {};
item.dependent.params.forEach(function(param) {
if (!phiSets[param.targetLabel]) phiSets[param.targetLabel] = [];
phiSets[param.targetLabel].push(param);
param.valueJS = finalizeLLVMParameter(param.value);
});
return phiSets;
}
function getPhiSetsForLabel(phiSets, label) {
if (!phiSets) return '';
label = getOldLabel(label);
if (!phiSets[label]) return '';
var labelSets = phiSets[label];
// FIXME: Many of the |var |s here are not needed, but without them we get slowdowns with closure compiler. TODO: remove this workaround.
if (labelSets.length == 1) {
return 'var ' + labelSets[0].ident + ' = ' + labelSets[0].valueJS + ';';
}
// TODO: eliminate unneeded sets (to undefined etc.)
var deps = {}; // for each ident we will set, which others it depends on
var valueJSes = {};
labelSets.forEach(function(labelSet) {
deps[labelSet.ident] = {};
valueJSes[labelSet.ident] = labelSet.valueJS;
});
labelSets.forEach(function(labelSet) {
walkInterdata(labelSet.value, function mark(item) {
if (item.intertype == 'value' && item.ident in deps && labelSet.ident != item.ident) {
deps[labelSet.ident][item.ident] = true;
}
});
});
var pre = '', post = '', idents;
mainLoop: while ((idents = keys(deps)).length > 0) {
var remove = function(ident) {
for (var i = 0; i < idents.length; i++) {
delete deps[idents[i]][ident];
}
delete deps[ident];
}
for (var i = 0; i < idents.length; i++) {
if (keys(deps[idents[i]]).length == 0) {
pre = 'var ' + idents[i] + ' = ' + valueJSes[idents[i]] + ';' + pre;
remove(idents[i]);
continue mainLoop;
}
}
// If we got here, we have circular dependencies, and must break at least one.
pre = 'var ' + idents[0] + '$phi = ' + valueJSes[idents[0]] + ';' + pre;
post += 'var ' + idents[0] + ' = ' + idents[0] + '$phi;';
remove(idents[0]);
}
return pre + post;
/* // Safe, unoptimized copying
var ret = '';
for (var i = 0; i < labelSets.length-1; i++) {
ret += 'var ' + labelSets[i].ident + '$phi = ' + labelSets[i].valueJS + ';';
}
ret += labelSets[labelSets.length-1].ident + ' = ' + labelSets[labelSets.length-1].valueJS + ';';
for (var i = 0; i < labelSets.length-1; i++) {
ret += labelSets[i].ident + ' = ' + labelSets[i].ident + '$phi;';
}
return ret;
*/
}
makeFuncLineActor('branch', function(item) {
var phiSets = calcPhiSets(item);
if (!item.value) {
return getPhiSetsForLabel(phiSets, item.label) + makeBranch(item.label, item.currLabelId);
} else {
var condition = finalizeLLVMParameter(item.value);
var labelTrue = getPhiSetsForLabel(phiSets, item.labelTrue) + makeBranch(item.labelTrue, item.currLabelId);
var labelFalse = getPhiSetsForLabel(phiSets, item.labelFalse) + makeBranch(item.labelFalse, item.currLabelId);
if (labelTrue == ';' && labelFalse == ';') return ';';
var head = 'if (' + condition + ') { ';
var head2 = 'if (!(' + condition + ')) { ';
var else_ = ' } else { ';
var tail = ' }';
if (labelTrue == ';') {
return head2 + labelFalse + tail;
} else if (labelFalse == ';') {
return head + labelTrue + tail;
} else {
return head + labelTrue + else_ + labelFalse + tail;
}
}
});
makeFuncLineActor('switch', function(item) {
// TODO: Find a case where switch is important, and benchmark that. var SWITCH_IN_SWITCH = 1;
var phiSets = calcPhiSets(item);
// Consolidate checks that go to the same label. This is important because it makes the
// js optimizer hoistMultiples much easier to implement (we hoist into one place, not
// many).
var targetLabels = {}; // for each target label, the list of values going to it
item.switchLabels.forEach(function(switchLabel) {
if (!targetLabels[switchLabel.label]) {
targetLabels[switchLabel.label] = [];
}
targetLabels[switchLabel.label].push(switchLabel.value);
});
var ret = '';
var first = true;
for (var targetLabel in targetLabels) {
if (!first) {
ret += 'else ';
} else {
first = false;
}
ret += 'if (' + targetLabels[targetLabel].map(function(value) { return item.ident + ' == ' + value }).join(' || ') + ') {\n';
ret += ' ' + getPhiSetsForLabel(phiSets, targetLabel) + makeBranch(targetLabel, item.currLabelId || null) + '\n';
ret += '}\n';
}
if (item.switchLabels.length > 0) ret += 'else {\n';
ret += getPhiSetsForLabel(phiSets, item.defaultLabel) + makeBranch(item.defaultLabel, item.currLabelId) + '\n';
if (item.switchLabels.length > 0) ret += '}\n';
if (item.value) {
ret += ' ' + toNiceIdent(item.value);
}
return ret;
});
makeFuncLineActor('return', function(item) {
var ret = RuntimeGenerator.stackExit(item.funcData.initialStack, item.funcData.otherStackAllocations) + ';\n';
if (PROFILE) {
ret += 'if (PROFILING) { '
+ 'PROFILING_NODE.time += Date.now() - __profilingStartTime__; '
+ 'PROFILING_NODE = __parentProfilingNode__ '
+ '}\n';
}
if (LABEL_DEBUG) {
ret += "print(INDENT + 'Exiting: " + item.funcData.ident + "');\n"
+ "INDENT = INDENT.substr(0, INDENT.length-2);\n";
}
ret += 'return';
if (item.value) {
ret += ' ' + finalizeLLVMParameter(item.value);
}
return ret + ';';
});
makeFuncLineActor('resume', function(item) {
return (EXCEPTION_DEBUG ? 'print("Resuming exception");' : '') +
'throw ' + makeGetValue('_llvm_eh_exception.buf', '0', 'void*') + ';';
});
makeFuncLineActor('invoke', function(item) {
// Wrapping in a function lets us easily return values if we are
// in an assignment
var phiSets = calcPhiSets(item);
var call_ = makeFunctionCall(item.ident, item.params, item.funcData, item.type);
var ret = '(function() { try { __THREW__ = false; return '
+ call_ + ' '
+ '} catch(e) { '
+ 'if (typeof e != "number") throw e; '
+ 'if (ABORT) throw e; __THREW__ = true; '
+ (EXCEPTION_DEBUG ? 'print("Exception: " + e + ", currently at: " + (new Error().stack)); ' : '')
+ 'return null } })(); if (!__THREW__) { ' + getPhiSetsForLabel(phiSets, item.toLabel) + makeBranch(item.toLabel, item.currLabelId)
+ ' } else { ' + getPhiSetsForLabel(phiSets, item.unwindLabel) + makeBranch(item.unwindLabel, item.currLabelId) + ' }';
return ret;
});
makeFuncLineActor('atomic', function(item) {
var type = item.params[0].type;
var param1 = finalizeLLVMParameter(item.params[0]);
var param2 = finalizeLLVMParameter(item.params[1]);
switch (item.op) {
case 'add': return '(tempValue=' + makeGetValue(param1, 0, type) + ',' + makeSetValue(param1, 0, 'tempValue+' + param2, type) + ',tempValue)';
case 'xchg': return '(tempValue=' + makeGetValue(param1, 0, type) + ',' + makeSetValue(param1, 0, param2, type) + ',tempValue)';
case 'cmpxchg': {
var param3 = finalizeLLVMParameter(item.params[2]);
return '(tempValue=' + makeGetValue(param1, 0, type) + ',(' + makeGetValue(param1, 0, type) + '==' + param2 + ' && (' + makeSetValue(param1, 0, param3, type) + ')),tempValue)';
}
default: throw 'unhandled atomic op: ' + item.op;
}
});
makeFuncLineActor('landingpad', function(item) {
// Just a stub
return '{ f0: ' + makeGetValue('_llvm_eh_exception.buf', '0', 'void*') +
', f1:' + makeGetValue('_llvm_eh_exception.buf', QUANTUM_SIZE, 'void*') + ' }';
});
makeFuncLineActor('load', function(item) {
var value = finalizeLLVMParameter(item.pointer);
var impl = item.ident ? getVarImpl(item.funcData, item.ident) : VAR_EMULATED;
switch (impl) {
case VAR_NATIVIZED: {
return value; // We have the actual value here
}
case VAR_EMULATED: return makeGetValue(value, 0, item.type, 0, item.unsigned, 0, item.align);
default: throw "unknown [load] impl: " + impl;
}
});
makeFuncLineActor('extractvalue', function(item) {
assert(item.indexes.length == 1); // TODO: use getelementptr parsing stuff, for depth. For now, we assume that LLVM aggregates are flat,
// and we emulate them using simple JS objects { f1: , f2: , } etc., for speed
return item.ident + '.f' + item.indexes[0][0].text;
});
makeFuncLineActor('insertvalue', function(item) {
assert(item.indexes.length == 1); // TODO: see extractvalue
var ret = '(', ident;
if (item.ident === 'undef') {
item.ident = 'tempValue';
ret += item.ident + ' = [' + makeEmptyStruct(item.type) + '], ';
}
return ret + item.ident + '.f' + item.indexes[0][0].text + ' = ' + finalizeLLVMParameter(item.value) + ', ' + item.ident + ')';
});
makeFuncLineActor('indirectbr', function(item) {
return makeBranch(finalizeLLVMParameter(item.value), item.currLabelId, true);
});
makeFuncLineActor('alloca', function(item) {
if (typeof item.allocatedIndex === 'number') {
if (item.allocatedSize === 0) return ''; // This will not actually be shown - it's nativized
return getFastValue('__stackBase__', '+', item.allocatedIndex.toString());
} else {
return RuntimeGenerator.stackAlloc(getFastValue(calcAllocatedSize(item.allocatedType), '*', item.allocatedNum));
}
});
makeFuncLineActor('phi', function(item) {
var params = item.params;
assert(!MICRO_OPTS);
function makeOne(i) {
if (i === params.length-1) {
return finalizeLLVMParameter(params[i].value);
}
return '__lastLabel__ == ' + getLabelId(params[i].label) + ' ? ' +
finalizeLLVMParameter(params[i].value) + ' : (' + makeOne(i+1) + ')';
}
var ret = makeOne(0);
if (item.postSet) ret += item.postSet;
return ret;
});
makeFuncLineActor('mathop', processMathop);
makeFuncLineActor('bitcast', function(item) {
return processMathop({
op: 'bitcast', variant: null, type: item.type,
param1: item.params[0]
});
});
function makeFunctionCall(ident, params, funcData, type) {
// We cannot compile assembly. See comment in intertyper.js:'Call'
assert(ident != 'asm', 'Inline assembly cannot be compiled to JavaScript!');
var shortident = LibraryManager.getRootIdent(ident.slice(1)) || ident.slice(1); // ident may not be in library, if all there is is ident__inline
var args = [];
var argsTypes = [];
var varargs = [];
var varargsTypes = [];
var ignoreFunctionIndexizing = [];
var useJSArgs = (shortident + '__jsargs') in LibraryManager.library;
var hasVarArgs = isVarArgsFunctionType(type);
var normalArgs = (hasVarArgs && !useJSArgs) ? countNormalArgs(type) : -1;
if (I64_MODE == 1 && ident in LLVM.INTRINSICS_32) {
// Some LLVM intrinsics use i64 where it is not needed, and would cause much overhead
params.forEach(function(param) { if (param.type == 'i64') param.type = 'i32' });
}
params.forEach(function(param, i) {
var val = finalizeParam(param);
if (!hasVarArgs || useJSArgs || i < normalArgs) {
if (param.type == 'i64' && I64_MODE == 1) {
val = makeCopyI64(val); // Must copy [low, high] i64s, so they don't end up modified in the caller
}
args.push(val);
argsTypes.push(param.type);
} else {
if (!(param.type == 'i64' && I64_MODE == 1)) {
varargs.push(val);
varargs = varargs.concat(zeros(Runtime.getNativeFieldSize(param.type)-1));
varargsTypes.push(param.type);
varargsTypes = varargsTypes.concat(zeros(Runtime.getNativeFieldSize(param.type)-1));
} else {
// i64 mode 1. Write one i32 with type i64, and one i32 with type i32
varargs.push(val + '[0]');
varargs = varargs.concat(zeros(Runtime.getNativeFieldSize('i32')-1));
ignoreFunctionIndexizing.push(varargs.length); // We will have a value there, but no type (the type is i64, but we write two i32s)
varargs.push(val + '[1]');
varargs = varargs.concat(zeros(Runtime.getNativeFieldSize('i32')-1));
varargsTypes.push('i64');
varargsTypes = varargsTypes.concat(zeros(Runtime.getNativeFieldSize('i32')-1));
varargsTypes.push('i32');
varargsTypes = varargsTypes.concat(zeros(Runtime.getNativeFieldSize('i32')-1));
}
}
});
args = args.map(function(arg, i) { return indexizeFunctions(arg, argsTypes[i]) });
varargs = varargs.map(function(vararg, i) {
if (ignoreFunctionIndexizing.indexOf(i) >= 0) return vararg;
return vararg === 0 ? 0 : indexizeFunctions(vararg, varargsTypes[i])
});
if (hasVarArgs && !useJSArgs) {
if (varargs.length === 0) {
varargs = [0];
varargsTypes = ['i32'];
}
var offset = 0;
varargs = '(tempInt=' + RuntimeGenerator.stackAlloc(varargs.length, ',') + ',' +
varargs.map(function(arg, i) {
var type = varargsTypes[i];
if (type == 0) return null;
if (I64_MODE == 1 && type == 'i64') type = 'i32'; // We have [i64, 0, 0, 0, i32, 0, 0, 0] in the layout at this point
var ret = makeSetValue(getFastValue('tempInt', '+', offset), 0, arg, type, null, null, QUANTUM_SIZE);
offset += Runtime.getNativeFieldSize(type);
return ret;
}).filter(function(arg) {
return arg !== null;
}).join(',') + ',tempInt)';
}
args = args.concat(varargs);
var argsText = args.join(', ');
// Inline if either we inline whenever we can (and we can), or if there is no noninlined version
var inline = LibraryManager.library[shortident + '__inline'];
var nonInlined = shortident in LibraryManager.library;
if (inline && (INLINE_LIBRARY_FUNCS || !nonInlined)) {
return inline.apply(null, args); // Warning: inlining does not prevent recalculation of the arguments. They should be simple identifiers
}
if (getVarData(funcData, ident)) {
ident = 'FUNCTION_TABLE[' + ident + ']';
}
return ident + '(' + args.join(', ') + ')';
}
makeFuncLineActor('getelementptr', function(item) { return finalizeLLVMFunctionCall(item) });
makeFuncLineActor('call', function(item) {
if (item.standalone && LibraryManager.isStubFunction(item.ident)) return ';';
return makeFunctionCall(item.ident, item.params, item.funcData, item.type) + (item.standalone ? ';' : '');
});
makeFuncLineActor('unreachable', function(item) { return 'throw "Reached an unreachable!"' }); // Original .ll line: ' + item.lineNum + '";' });
// Final combiner
function finalCombiner(items) {
dprint('unparsedFunctions', 'Starting finalCombiner');
var itemsDict = { type: [], GlobalVariableStub: [], functionStub: [], function: [], GlobalVariable: [], GlobalVariablePostSet: [] };
items.forEach(function(item) {
item.lines = null;
var small = { intertype: item.intertype, JS: item.JS, ident: item.ident, dependencies: item.dependencies }; // Release memory
itemsDict[small.intertype].push(small);
});
items = null;
var splitPostSets = splitter(itemsDict.GlobalVariablePostSet, function(x) { return x.ident && x.dependencies });
itemsDict.GlobalVariablePostSet = splitPostSets.leftIn;
var orderedPostSets = splitPostSets.splitOut;
var limit = orderedPostSets.length * orderedPostSets.length;
for (var i = 0; i < orderedPostSets.length; i++) {
for (var j = i+1; j < orderedPostSets.length; j++) {
if (orderedPostSets[j].ident in orderedPostSets[i].dependencies) {
var temp = orderedPostSets[i];
orderedPostSets[i] = orderedPostSets[j];
orderedPostSets[j] = temp;
i--;
limit--;
assert(limit > 0, 'Could not sort postsets!');
break;
}
}
}
itemsDict.GlobalVariablePostSet = itemsDict.GlobalVariablePostSet.concat(orderedPostSets);
//
if (!mainPass) {
var generated = itemsDict.function.concat(itemsDict.type).concat(itemsDict.GlobalVariableStub).concat(itemsDict.GlobalVariable).concat(itemsDict.GlobalVariablePostSet);
Functions.allIdents = Functions.allIdents.concat(itemsDict.function.map(function(func) {
return func.ident;
}).filter(function(func) {
return IGNORED_FUNCTIONS.indexOf(func.ident) < 0;
}));
if (!DEBUG_MEMORY) print(generated.map(function(item) { return item.JS }).join('\n'));
return;
}
// This is the main pass. Print out the generated code that we have here, together with the
// rest of the output that we started to print out earlier (see comment on the
// "Final shape that will be created").
var generated = itemsDict.functionStub.concat(itemsDict.GlobalVariablePostSet);
generated.forEach(function(item) { print(indentify(item.JS || '', 2)); });
if (RUNTIME_TYPE_INFO) {
Types.cleanForRuntime();
print('Runtime.typeInfo = ' + JSON.stringify(Types.types));
print('Runtime.structMetadata = ' + JSON.stringify(Types.structMetadata));
}
var postFile = BUILD_AS_SHARED_LIB ? 'postamble_sharedlib.js' : 'postamble.js';
var postParts = processMacros(preprocess(read(postFile))).split('{{GLOBAL_VARS}}');
print(postParts[0]);
// Print out global variables and postsets TODO: batching
JSify(analyzer(intertyper(data.unparsedGlobalss[0].lines, true), true), true, Functions);
data.unparsedGlobalss = null;
print(Functions.generateIndexing()); // done last, as it may rely on aliases set in postsets
// Load runtime-linked libraries
RUNTIME_LINKED_LIBS.forEach(function(lib) {
print('eval(read("' + lib + '"))(FUNCTION_TABLE.length, this);');
});
print(postParts[1]);
print(shellParts[1]);
// Print out some useful metadata (for additional optimizations later, like the eliminator)
print('// EMSCRIPTEN_GENERATED_FUNCTIONS: ' + JSON.stringify(Functions.allIdents) + '\n');
return null;
}
// Data
if (mainPass) {
substrate.addItems(data.functionStubs, 'FunctionStub');
assert(data.functions.length == 0);
} else {
substrate.addItems(values(data.globalVariables), 'GlobalVariable');
substrate.addItems(data.aliass, 'Alias');
substrate.addItems(data.functions, 'FunctionSplitter');
}
finalCombiner(substrate.solve());
dprint('framework', 'Big picture: Finishing JSifier, main pass=' + mainPass);
}
You can’t perform that action at this time.